Introduction
Language development is a complex and multifaceted process that requires both person-centered and environmental factors (Hindman, Wasik & Snell, Reference Hindman, Wasik and Snell2016). Researchers have consistently shown that language exposure and the quality of the language environment in which the child develops are important predictors to children's later linguistic skills and cognitive development (e.g., Romeo, Leonard, Robinson, West, Mackey, Rowe & Gabrieli, Reference Romeo, Leonard, Robinson, West, Mackey, Rowe and Gabrieli2018). Additionally, maternal mental health is a known risk factor to child development. Particularly, maternal depressive symptoms have been related to poorer child language scores in the first few years of life (Paulson, Keefe & Leiferman, Reference Paulson, Keefe and Leiferman2009; Stein, Malmberg, Slyva, Barnes & Leach, Reference Stein, Malmberg, Slyva, Barnes and Leach2008). However, researches regarding maternal mental health and children's language development have generally relied on parental reports of children's language and have focused mainly on maternal depression. Thus, the current study seeks to extend the current literature by examining maternal mental health as it relates to young children's home language environment and indices of expressive language using in-home observational measures.
Fostering children's language development
The complex process of language development is heavily influenced by various child and environmental-level factors. According to the emergentist coalition model, child-level factors include various domain-general and domain-specific cognitive processes (Hollich, Hirsh-Pasek, Golinkoff, Brand, Brown, Chung, Hennon & Rocroi, Reference Hollich, Hirsh-Pasek, Golinkoff, Brand, Brown, Chung, Hennon and Rocroi2000). From initial exposure to mastery, these cognitive processes and strategies are used intermittently to foster progression in language learning. For example, a child initiates learning a specific word at the initial exposure of a novel object, utilizing domain-general processes (e.g., object representation, memory). Other domain-specific processes (e.g., whole-object bias, taxonomic bias) also work at various points of word learning, progressing towards mastery of the specific word. These processes are critical for language development; however, children can only learn words that are present in their environment.
In addition to these individual child-level processes, social-interactional cues are utilized more often as maturation of child-level processes progress (Hollich et al., Reference Hollich, Hirsh-Pasek, Golinkoff, Brand, Brown, Chung, Hennon and Rocroi2000). Thus, the emergentist coalition model posits that children need the guidance and help of others to foster language learning. Indeed, children are better able to utilize their cognitive processes when presented with initial and repeated exposures to new stimuli (Clay, Gill, Glynn, McNaughton & Salmon, Reference Clay, Gill, Glynn, McNaughton and Salmon2007), including language. Though contextual factors such as SES may place some children at a disadvantage for developing language, previous work has demonstrated the important role of caregivers in fostering language development (National Institute of Child Health and Human Development Early Child Care Research Network, 2005; Fernald, Marchman & Weisleder, Reference Fernald, Marchman and Weisleder2013). Further, in addition to the quantity of parental speech (Weisleder & Fernald, Reference Weisleder and Fernald2013), research has shown that the quality of speech is important when fostering children's language development (Song, Demuth & Morgan, Reference Song, Demuth and Morgan2018; Zimmerman, Gilkerson, Richards, Christakis, Xu, Gray & Yapanel, Reference Zimmerman, Gilkerson, Richards, Christakis, Xu, Gray and Yapanel2009). Moreover, research has discovered specific social-interactions that facilitate language learning. For example, children are better able to learn words when they are presented in a meaningful context rather than in isolation (Han, Moore, Vukelich & Buell, Reference Han, Moore, Vukelich and Buell2010). Next, children benefit when encouraged to connect word meanings to their own personal experience and background knowledge, thereby extending what they already know onto new knowledge (Harris, Golinkoff & Hirsh-Pasek, Reference Harris, Golinkoff, Hirsh-Pasek, Neuman and Dickinson2011). Lastly, adults can foster language development by participating in meaningful, elaborate, and repeated exchanges with children (Tamis-LeMonda, Kuchirko & Song, Reference Tamis-LeMonda, Kuchirko and Song2014). In sum, children's language development is the result of various cognitive processes and can be scaffolded and enhanced through positive social-interactional processes.
Maternal mental health, the home language environment, and language development
Given the importance of positive social interactions in children's language development, the mental health of mothers can be a risk factor for children's home language environment and language development (Cooper & Murray, Reference Cooper and Murray1998). Indeed, Goodman and Gotlib theorized multiple mechanisms through which mother's prenatal and postnatal mental health may transmit risk to children in developing various vulnerabilities and outcomes (Reference Goodman and Gotlib1999). One of these mechanisms is through mother's maladaptive cognitions, behaviors, and affect. These factors may make mothers less able to be an adaptive social partner by creating a rich, stimulating environment for their children (Wu, East, Delker, Blanco, Caballero, Delva & Gahagan, Reference Wu, East, Delker, Blanco, Caballero, Delva and Gahagan2018). Empirical work has supported this notion. For example, Lovejoy, Graczyk, O'Hare and Neuman (Reference Lovejoy, Graczyk, O'Hare and Neuman2000) conducted a meta-analysis and reported associations between maternal depression and negative/coercive behavior (e.g., intrusive, rough, unpleasant tone), disengagement (e.g., lack of gaze toward child, less responsivity, flatter affect), and fewer positive behaviors (e.g., playing, affectionate interactions, warmth). Hence, it appears that the maladaptive interactions that often accompany depressed mothers may place these children at risk for delayed language development by creating a less stimulating environment. Less work has examined relations between maternal anxiety and its impact on the home environment. Further, few studies (e.g., Rowe, Pan & Ayoub, Reference Rowe, Pan and Ayoub2005) have examined associations between aspects of maternal mental health and the quantitative and qualitative characteristics of their parental speech to their children.
Maternal depression
According to the U.S. National Institute of Mental Health, depression is especially prevalent among women, with 8.7% experiencing a major depressive episode in 2017 (NIMH, 2018). This estimate potentially excludes the millions of women who experience some level of depressive symptoms. Previous studies have suggested maternal depression to have negative effects on children's language development. For example, scholars in one study found a negative relation between postnatal maternal depression at 18 months and seven and eight-year-old children's verbal intelligence as measured by the Wechsler Intelligence Scale for Children (WISC-III; Barker, Jaffee, Uher & Maughan, Reference Barker, Jaffee, Uher and Maughan2011). Another study indicated that maternal depression at both 1.5 and 12 months was negatively related to 12-month-old childrens’ receptive and expressive communication scores on the Bayley Scales of Infant Development III (Bayley-III) (Quevedo, Silva, Godoy, Jansen, Matos, Tavares Pinheiro & Pinheiro, Reference Quevedo, Silva, Godoy, Jansen, Matos, Tavares Pinheiro and Pinheiro2012). Kawai, Takagai, Takei, Itoh, Kanayama and Tsuchiya (Reference Kawai, Takagai, Takei, Itoh, Kanayama and Tsuchiya2017) also found postpartum depression at 1 month (but not 3 months) to be negatively related to children's non-verbal communication skills at 14 months as measured by the Japanese version of the MacArthur-Bates Communicative Development Inventory (MCDI). Ahun, Geoffroy, Herba, Brendgen, Séguin, Sutter-Dallay and Côté (Reference Ahun, Geoffroy, Herba, Brendgen, Séguin, Sutter-Dallay and Côté2017) found relations between postpartum maternal depression and children's language. Specifically, mothers who were chronically depressed over five years postpartum had children who scored lower on the Peabody Picture Vocabulary Test-Revised (PPVT-R). However, in another study, no relations were found between postpartum depression at six weeks or six months and children's language composite scores on the Bayley-III at 18 months (Piteo, Yelland & Makrides, Reference Piteo, Yelland and Makrides2012). Similarly, no relations were found between postpartum depression at 14 months and children's receptive vocabulary at 5 years (Chazan-Cohen, Raikes, Brooks-Gunn, Ayoub, Pan, Kisker, Roggman, & Fuligni, Reference Chazan-Cohen, Raikes, Brooks-Gunn, Ayoub, Pan, Kisker and Fuligni2009). Interestingly, Keim, Daniels, Dole, Herring, Siega-Riz and Scheidt (Reference Keim, Daniels, Dole, Herring, Siega-Riz and Scheidt2011) found negative relations between postpartum depressive symptoms at four months and children's receptive language ability captured by the Mullen Scales of Early Learning at 12 months. However, when examining these relations non-linearly, children scored higher in receptive language at lower and higher levels of postpartum depression but lower in receptive language near the average scores of postpartum depression. In sum, previous studies of direct relations between postpartum depressive symptoms and children's language have shown divergent results that require further clarification.
Additionally, some scholars have investigated mediating mechanisms between parental depression and children's language outcomes. For example, in one longitudinal study, Stein, Malmberg, Slyva, Barnes and Leach (Reference Stein, Malmberg, Slyva, Barnes and Leach2008) found maternal caregiving to mediate the relations between maternal depression at ten months and children's language development (Reynell Developmental Language Scale) at 36 months. In another longitudinal study, parent-child reading activities were found to mediate the relations between parental depression at nine months and children's expressive vocabulary (MCDI) at 24 months (Paulson et al., Reference Paulson, Keefe and Leiferman2009). Lastly, Baydar, Küntay, Yagmurlu, Aydemir, Cankaya, Göksen, and Cemalcilar, (Reference Baydar, Küntay, Yagmurlu, Aydemir, Cankaya, Göksen and Cemalcilar2014) examined several protective and mediating factors that support children's vocabulary development and found varying results depending on the level of environmental risk experienced by the child (e.g., depression and economic distress). The studies reviewed here contribute to our understanding of the relations between postpartum depression and children's language outcomes. Further, they point to possible moderating and mediating factors in these relations. Taken together, additional research is needed to clarify some of the discrepant findings and explore potential moderating and mediating mechanisms in these relations.
Maternal anxiety
As another aspect of mental health, feelings of anxiety can be a normal reaction to stress or anticipation of future events. However, when these feelings of nervousness or fear become excessive and affect one's ability to function normally, these individuals are at risk for developing anxiety disorders. As much as 30% of adults are affected by anxiety disorders at some point in their lives and anxiety is more common among women compared to men (APA, 2019). Research has indicated maternal anxiety to be related to maternal depression (Walker, Timmerman, Kim & Sterling, Reference Walker, Timmerman, Kim and Sterling2002). However, compared to the work regarding maternal depression, there is substantially less empirical work suggesting relations between maternal anxiety and children's language environments and language development. For example, in one study, several indices of mental health were investigated and researchers found no significant associations with maternal anxiety at four months and receptive or expressive language at 12 months as measured by the Mullen Scales of Early Learning (Keim et al., Reference Keim, Daniels, Dole, Herring, Siega-Riz and Scheidt2011). However, some have noted that stress during prenatal development or in the home can negatively affect productive language abilities (e.g., Laplante, Barr, Brunet, Du Fort, Meaney, Saucier, Zelazo & King, Reference Laplante, Barr, Brunet, Du Fort, Meaney, Saucier, Zelazo and King2004).
Despite the lack of empirical research on the relations between maternal anxiety and children's home language environments and language development, these constructs may still be conceptually related. Indeed, similar to mothers experiencing depression, maternal anxiety is linked to less effective mother-child interactions, which can be a hindrance in creating a stimulating, rich environment. For example, research has found mothers experiencing anxiety to use less structure and to be less sensitive (Feldman, Granat, Pariente, Kanety, Kuint & Gilboa-Schechtman, Reference Feldman, Granat, Pariente, Kanety, Kuint and Gilboa-Schechtman2009), to have fewer instances of positive engagement (Reck, Tietz, Müller, Seibold & Tronick, Reference Reck, Tietz, Müller, Seibold and Tronick2018), and to exert higher levels of control or intrusiveness with their children (van der Bruggen, Stams & Bögels, Reference van der Bruggen, Stams and Bögels2008). Hence, if children of more anxious mothers tend to experience fewer of the positive interactions needed to foster positive language development, these children may be at risk for a less stimulating, rich home language environment and subsequent positive language development (Zimmerman et al., Reference Zimmerman, Gilkerson, Richards, Christakis, Xu, Gray and Yapanel2009).
The current study
Despite the findings of previous work regarding the relations between maternal mental health, children's home language environments, and language development, a number of gaps exist that necessitate further work. First, more research is needed to understand potential associations between maternal anxiety and children's home language environments and language development. Second, direct measures of the home language environment and expressive language are needed to provide more precise estimates of the relations between these constructs. Third, the majority of previous research focuses on one aspect of maternal mental health, rather than considering anxiety and depression together (e.g., Quevedo et al., Reference Quevedo, Silva, Godoy, Jansen, Matos, Tavares Pinheiro and Pinheiro2012). This is important given the associations between these constructs (Keim et al., Reference Keim, Daniels, Dole, Herring, Siega-Riz and Scheidt2011) and the unique impact anxiety and depression may have on children's home language environments and language development. Thus, the current study aims to address these gaps by (a) examining two aspects of mental health in relation to the home language environment and children's language development in the same model and (b) utilizing a more precise methodology to capture children's home language environments and their expressive language. To guide our examination, we generated two hypotheses.
Hypothesis 1: Maternal mental health (i.e., depression and anxiety) will be negatively related to indices of the children's home language environment (i.e., observed number of adult words spoken and observed number of conversational turns).
Hypothesis 2: Maternal mental health (i.e., depression and anxiety) will be negatively related to indices of children's language development (i.e., number of child vocalizations, observed language production scores, and parent-reported language production and comprehension).
Method
Participants
Upon receiving approval from the university's institutional review board (application number F16089; Project Media Effects on Development from Infancy to Adulthood [MEDIA]), 250 primary caregiver-infant (under 1 year) dyads completed in-home assessments and surveys in the Denver metro-area in the Spring/Summer of 2017. Participants were recruited through mailers sent to the participant home through the Colorado Office of Health and Vital Records, which identified anyone in the local area who had birthed a child in the past year (34.4% of final sample). Research assistants were then sent to potential participants’ homes to invite them to participate, which yielded a participant rate of 66% in Wave 1 of Project MEDIA.
Participants were also recruited using multiple methods, including flyers in pediatrician offices, free clinics, social services offices, businesses focused on entertainment for young children, public parks and play spaces, and referral from a friend who participated (17.8%). A large proportion of the sample (47.8%) was recruited through an external data collection company. Participants were contacted again in 2018, with a 98.9% retention rate. All measures from the current study are from Wave 2 except for the demographic information, which is from Wave 1. In 2018, 19 additional very low-income families (10 household income below $30,000, 8 household income below $50,000, 1 below $60,000) were recruited through a Colorado Office of Health and Vital Records mailer in order to increase the diversity of our sample in terms of household income.
Participants included 269 infants-primary caregiver dyads (infants 49.6% female and 50.4% male, Mage = 17.03 months, SD = 3.44 months. Range = 11 to 29 months). The majority of the participants were mothers as the primary caregiver (N = 260; fathers N = 5; 2 missing; 3.2% of the sample was under 20, 20.4% were between 21–25, 31.2% were between 26–30, 26.6% were between 31–35, 15.6% were between 36–39, and 4.8% were above 40). Observations with the father as the primary caregiver were excluded from the analyses.
The participants were ethnically and socioeconomically diverse. Twenty-one percent of the participants were Latino, 3% Asian, 10% Black, 59% White, 1% Mixed Race, and 6% “Other.” Many participants (31.6%) had a household income below $30,000, 25.1% had a household income over $30,000, but below $50,000, 19.4% between $50,000 and below $70,000, 19.3% between $70,000 and under $100,000, and 4.4% more than $100,000. Approximately 37.2% currently receive public assistance and 46.6% received public assistance in the last year. In regard to education, 20.1% of primary caregivers’ highest level of education completed was high school (or equivalent), 36.5% had completed some college, 26.7% had completed a bachelor's degree, and 16.5% had completed graduate education. All families recruited for the current study were proficient in English. Approximately 22% of primary caregivers reported that their child was spoken to in two or more languages and approximately 70% reported only speaking to their child in one language at home. Primary caregivers who could not complete the survey and the in-home assessments in English were excluded from the study. Families that did not have children within the target range were excluded from the study.
Measures
Home language environment
To measure the home language environment, we used the Language Environment Analysis System designed to provide clinicians and researchers detailed information regarding the language environment of infants and toddlers. Parents placed a small digital recording device (language processor) in the pocket of a custom-made shirt, and this device recorded everything the infant uttered and heard during a 24-hour period. The raw audio files were transferred to LENA through a secure cloud-based system and analyzed by the LENA language environment analysis software. LENA software uses advanced language related algorithms to produce core reports related to the infants’ home language environment (Xu, Yapanel & Gray, Reference Xu, Yapanel and Gray2009). The estimates in these core reports have been found to be valid and reliable in previous studies (e.g., Gilkerson, Richards, Warren, Montgomery, Greenwood, Kimbrough Oller, Hansen & Paul, Reference Gilkerson, Richards, Warren, Montgomery, Greenwood, Kimbrough Oller, Hansen and Paul2017; Romeo et al., Reference Romeo, Leonard, Robinson, West, Mackey, Rowe and Gabrieli2018). This methodology yielded two measures of the home language environment (i.e., number of adult words spoken and number of conversational turns). Number of adult words spoken is a count of adult words spoken to or overheard by the child. This measure accounts for clear and unclear speech that has communicative intent including: interjections, reductions, contractions, and vocal play. Number of conversational turns is a count of back and forth alterations between the child and adults. Conversational turns are responses that include speech, interruptions, and communicative vocalizations (e.g., whining) occurring within five seconds. These LENA home language environment measures were standardized by age for analysis
Infant expressive language
Infant's expressive language was captured using both LENA and parental reports. LENA software produced three indices of children's language production including: number of child vocalizations, vocal productivity score (an automated canonical syllable count per conversational turn score), and automatic vocalization assessment (AVA; a categorized and quantified score of children's protophones and phonemes). These LENA expressive language scores were standardized by age for analysis. Further discussions of each of these expressive language measures are presented as technical reports by the LENA foundation (e.g., Richards, Xu, Gilkerson, Yapanel, Gray & Paul, Reference Richards, Xu, Gilkerson, Yapanel, Gray and Paul2017).
Infant language production and comprehension
Parents completed the Words and Gestures Vocabulary Checklist short form of the MacArthur-Bates Communicative Development Inventory (MCDI) to capture language production and comprehension. Parents reported on 100 words and their infants’ ability to produce and comprehend these words. Examples included cookies, diaper, boots, bed, flower, and eat. Infant language development scores are then normed for age and gender for production and comprehension separately. This inventory has been found to be reliable and valid in English and several other languages (Fenson, Marchman, Thal, Dale & Bates, Reference Fenson, Marchman, Thal, Dale and Bates2007).
Maternal depression
Depression was measured using the ten-item Center for Epidemiological Studies Short Depression Scale (CES-10; Levine, Reference Levine2013). Mothers answered statements regarding their own feelings and behaviors in the last week. Mothers answered statements on a four point Likert scale from 1 (Rarely or none of the time [less than 1 day]) to 4 (All of the time [5–7 days]). Two items were reverse-coded and the items were summed. Higher scores are indicative of more depression symptoms. Scores of 10 or higher are considered sufficient for clinical depression. Example items include “I could not get going” and “I felt depressed.” Cronbach's alpha reliability was α = .81.
Maternal anxiety
Anxiety was measured using the 16-item Penn State Worry Questionnaire (PSWQ; Meyer, Miller, Metzger & Borkovec, Reference Meyer, Miller, Metzger and Borkovec1990). Mothers rated statements on how characteristic they are of themselves on a five point Likert scale from 1 (Not at all typical of me) to 5 (Very typical of me). Five of the items were reverse-coded and higher scores indicated more worry. The PSWQ has been found to be a reliable measure of generalized anxiety disorder (GAD), as worry is the dominate feature of GAD. Scores of 62 or higher are recommended for balancing specificity and sensitivity in detecting GAD in the general population (Behar, Alcaine, Zuellig & Borkovec, Reference Behar, Alcaine, Zuellig and Borkovec2003). Example items include “Many situations make me worry” and “When I am under pressure, I worry a lot.” Cronbach's alpha reliability was α = .93.
Control variables
Socio-economic status (SES) was measured using participants responses to questions about their family income “What is your yearly gross combined household income (including you and your partner)?” ranging from 1 (Less than $15,000) to 8 ($150,000 or more) and maternal education “What is the highest level of education you have completed?” ranging from 1 (No formal school) to 6 (Doctoral/Professional degree). Scores on these separate items were standardized and then mean averaged to create a SES variable. Parents also answered questions regarding their ethnicity “What is your ethnicity” (White, African American, U.S. Latino or Hispanic, Asian American, Other) and age “How old are you?”.
Procedure
All procedures were approved by the Institutional Review Board of (Brigham Young University application number F16089 titled Project MEDIA). All data were collected in the homes of primary caregiver-infant dyads. The data and measures used for the present study are part of a larger longitudinal study of children's development. Primary caregivers completed online surveys concerning their demographic information and answered questions about themselves and their children. A subset of the larger sample also participated in two in-home visits (1.5 hours and 45 minutes, three days apart) where parents and/or children completed tasks and assessments with research assistants. LENA records were completed during a 24-hour period the day before the second in-home assessment. Primary caregivers were unaware of the specific research questions of the study and were instructed to place the LENA device in the infant's shirt (and given a special shirt with a pocket for the device) for the next 24-hour period. During bath and sleep times the parents were instructed to leave the LENA device in the room with the infant and close enough to infant to capture the language spoken. When the LENA was picked up from the parent during the second in-home visit, parents were asked if the previous day was a typical day. If it was not (e.g., the child or parent was sick, family members were in town, etc.), the parents were asked to have the infant wear the LENA device for an additional 24-hour period that was reflective of a typical day. Primary caregivers provided informed consent, completed several play-based assessments with their infants, and were compensated $125 in either Visa or Amazon gift cards for participating in both the online and in-home portions of the study.
Analysis plan
First, we checked that our data met statistical assumptions by examining the descriptive statistics and distributions of each variable. Next, we examined the scatter plots and used correlational analyses to examine the relations between our variables. Finally, we used a path model to examine the relations between maternal mental health (i.e., depression and anxiety), indices of the home language environment (i.e., number of adult words and number of conversational turns), and indices of the children's language development (i.e., LENA reported number of child vocalizations, vocal productivity, AVA, parent-reported language production and comprehension percentile scores). To help account for demographic heterogeneity, we included our SES, parent ethnicity, and parent age variables as covariates in the path model. Descriptive statistics for each variable are displayed in Table 1. Correlational analyses revealed several relations between variables and are displayed in Table 2.
Note: ***p < .001; **p < .01; *p < .05; +p < .10
Results
Model specification
To test our hypotheses, we specified and estimated a path model using MLMV (i.e., full information maximum likelihood; Maydeu-Olivares, Reference Maydeu-Olivares2017) using STATA 15 MP to examine the relations between maternal mental health and indices of the home language environment and children's language development while accounting for our control variables. The exogenous independent variables included maternal depression and anxiety predicting our endogenous dependent variables (i.e., adult words, conversational turns, child vocalizations, vocal productivity, AVA, language production, and language comprehension percentile scores). The control variables (i.e., SES, ethnicity, and parental age) predicted the dependent variables and were covaried with the independent variables (i.e., maternal depression and anxiety) and each other. Finally, we covaried the errors of our endogenous variables. This specification yielded a saturated model with N = 265 observations. Thus, our analysis did not report model fit statistics.
Path model
Results revealed numerous relations between maternal mental health and indices of children's language environment and language development. Since all our dependent variables were standardized into percentiles, we used b's to interpret our coefficients. Maternal anxiety was found to negatively predict children's parent-reported language production percentile scores (b = −.41; p = .04). As an example of interpretation, a one-unit increase in maternal anxiety predicted at least a .41 percentile decrease in parent-reported language production scores while accounting for the other independent and control variables in the model. Further, maternal depression negatively predicted the percentile scores for the number of adult words spoken (b = −1.06; p = .03) and conversational turns (b = −1.07; p = .01) over a 24-hour period. Lastly, maternal depression negatively predicted the percentile scores for number of child vocalizations (b = −.93; p = .04) and AVA (b = −.91; p = .04). See Figure 1 for path model.
Discussion
The current study examined maternal mental health factors and their relations to a variety of home language environment and language indices. This study was unique in that it a) captured multiple aspects of maternal mental health, and b) utilized observational measures of the home language environment and expressive language in addition to parent-reported language comprehension and production. Each of these variables were then analyzed together in a single path model.
Our first hypothesis was partially supported – in that maternal depression (but not anxiety) was negatively related to indices of the home language environment (i.e., observed number of adult words spoken and observed number of conversational turns). Our second hypothesis was also partially supported – in that maternal depression was negatively related with indices of children's expressive language (i.e., number of child vocalizations, AVA scores) but not child vocal productivity scores or parent-reported language production or comprehension scores. Further, maternal anxiety was negatively related to parent-reported child language production scores. Previous researchers have shown that maternal depression is a risk factor to child development (Goodman, Rouse, Connell, Broth, Hall & Heyward, Reference Goodman, Rouse, Connell, Broth, Hall and Heyward2011; Goodman & Gotlib, Reference Goodman and Gotlib1999), and the current study lends support to this previous work. These results are in line with previous studies finding negative associations between maternal depression and assessments of children's language development (Ahun, et al., Reference Ahun, Geoffroy, Herba, Brendgen, Séguin, Sutter-Dallay and Côté2017; Kawai et al., Reference Kawai, Takagai, Takei, Itoh, Kanayama and Tsuchiya2017; Quevedo et al., Reference Quevedo, Silva, Godoy, Jansen, Matos, Tavares Pinheiro and Pinheiro2012; Stein et al., Reference Stein, Malmberg, Slyva, Barnes and Leach2008). Relatedly, our finding of a negative relation between maternal anxiety and parent-reported language production is novel: in that previous research has not found this association. Each of our findings is strengthened by the fact that we examined each of these constructs together in the same model. Another strength of the present study is the use of more precise observational measures of the home language environment and children's expressive language to examine each of these relations.
Relations between maternal mental health, the home language environment, and expressive language
These results were interpreted within the framework of the emergentist coalition model of children's language development (Hollich et al., Reference Hollich, Hirsh-Pasek, Golinkoff, Brand, Brown, Chung, Hennon and Rocroi2000). Language acquisition is a complex process that requires child-specific cognitive, environmental and social factors working in tandem to produce optimal conditions for language acquisition, comprehension, and production (Hollich et al., Reference Hollich, Hirsh-Pasek, Golinkoff, Brand, Brown, Chung, Hennon and Rocroi2000). Parents are a vital component of this process as they provide language exposure, support, scaffolding, and context which are all essential for language development (Clay et al., Reference Clay, Gill, Glynn, McNaughton and Salmon2007). Previous researchers have shown maternal depression to be related to less effective parenting (Hoffman, Crnic & Baker, Reference Hoffman, Crnic and Baker2006), harsher responses to children (Lovejoy et al., Reference Lovejoy, Graczyk, O'Hare and Neuman2000), and less maternal warmth and engagement (Campbell, Cohn & Meyers, Reference Campbell, Cohn and Meyers1995).
Maternal depression
The current study lends support to these previous studies and extends previous work by examining the effect of maternal depression on the quality of the home language environment. In line with past work, the results of the current study suggest maternal depression to modulate the quality of the home language environment and place children at increased risk for experiencing language delays (Ahun et al., Reference Ahun, Geoffroy, Herba, Brendgen, Séguin, Sutter-Dallay and Côté2017; Kawai et al., Reference Kawai, Takagai, Takei, Itoh, Kanayama and Tsuchiya2017; Quevedo et al., Reference Quevedo, Silva, Godoy, Jansen, Matos, Tavares Pinheiro and Pinheiro2012; Stein et al., Reference Stein, Malmberg, Slyva, Barnes and Leach2008).
Importantly, maternal depression was not related to concurrent parental reports of their child's language production or comprehension. There are three possible explanations for this finding. The first is that parents’ reports of language differ from observed measures of children's expressive language – in that the former is a parent's assessment of the words that children understand and/or have produced whereas the latter captures various measures of children's expressive language during one 24-hour period. The second explanation is that depressed mothers may not be accurate reporters of their own children's language development. Maternal depression is known to influence maternal reports of children's temperament (McGrath, Records & Rice, Reference McGrath, Records and Rice2008) and behavior (Briggs-Gowan, Carter & Schwab-Stone, Reference Briggs-Gowan, Carter and Schwab-Stone1996). However, less is known regarding the effects of maternal depression on mothers’ ability to accurately report on their child's language development. The fact that we found relations between maternal depression and observational measures of children's expressive language and not parental reports of children's language production provides some indication that mothers with depression may struggle to accurately report on their child's language development. Lastly, the relationship between maternal depression and children's parent-reported language production and comprehension may take more time to manifest (Goodman & Gotlib, Reference Goodman and Gotlib1999). For example, one study found relations between maternal depression and children's language only when maternal depression reached chronic levels over five years (Ahun et al., Reference Ahun, Geoffroy, Herba, Brendgen, Séguin, Sutter-Dallay and Côté2017). Further, some previous studies finding associations between maternal depression and language development occurred across years, allowing for differences between children to possibly become more exacerbated (Barker et al., Reference Barker, Jaffee, Uher and Maughan2011).
Maternal anxiety
In contrast, maternal anxiety was not related to observational measures of the home language environment (i.e., number of adult words and number of conversational turns) or children's observed expressive language (e.g., child vocalizations). Past research examining relations between maternal anxiety and child development has been inconclusive (e.g., Keim et al., Reference Keim, Daniels, Dole, Herring, Siega-Riz and Scheidt2011). However, we found a negative association between maternal anxiety and parental reports of children's language production. This finding highlights that maternal anxiety may influence the way in which mothers view their own children, their behavior, and their development. It is possible that anxious mothers may be at increased risk for underreporting their child's language, perhaps due to their own anxiety (Briggs-Gowan, Carter & Schwab-Stone, Reference Briggs-Gowan, Carter and Schwab-Stone1996). While previous researchers have found these discrepancies in children's behavior, with anxious parents overreporting on their child's behavioral and externalizing symptoms (Briggs-Gowan et al., Reference Briggs-Gowan, Carter and Schwab-Stone1996), it is unclear whether this may be the case for children's language.
We did find an association between maternal anxiety and one index of observed expressive language (i.e., AVA scores) when examined using bivariate correlations. However, this association did not hold after controlling for maternal depression and other variables in our path model. This suggests that maternal depression may be more problematic for children's home language environments and language development compared to maternal anxiety (Porritt, Zinser, Cachorowski & Kaplan, Reference Porritt, Zinser, Cachorowski and Kaplan2014).
Limitations and future directions
While the current study contributes important information to our knowledge of the relations between maternal mental health and children's home language environments and expressive language, it is not without limitations. First, the cross-sectional nature of the data prevents us from inferring causation. Future research should examine observational measures of the child's home language environment, language, and parental reports of children's language comprehension and production across time and development. By examining these relations longitudinally, a better understanding of sensitive periods for child susceptibility to maternal mental health issues can be captured. Further, longitudinal data is needed to examine mediating mechanisms in the relations between contextual (e.g., SES) and relational factors (e.g., parent-child interactions), maternal mental health (e.g., depression, anxiety, stress), the home language environment, and children's language development. By identifying mechanisms in these relations and the timing of these effects, scholars, interventionists, and practitioners will be better enabled to gain further knowledge on these topics and provide improved treatment to children and families. Additionally, we examined these relations using depressive and anxiety symptoms rather than clinical diagnoses of these mental disorders. Although elevated scores on our measures of depressive and anxiety symptoms are indicative of mental disorders, future work is needed using clinical and non-clinical samples when examining these relations. Lastly, a proportion (22%) of the parents in our sample reported their child being spoken to in two or more languages. Though in this study we only included families who could complete the in-home assessments and surveys in English, future research should examine the role of multiple languages being present in children's environments as this may relate to their home language environments and expressive language.
Conclusion
Despite its limitations, the current study was an important step in the literature connecting both maternal depression and anxiety to children's home language environments and expressive language. By examining each of these constructs simultaneously within the same model, the results provide a more comprehensive insight of these relations. Children acquire language through the constant interplay between child-centered cognitive characteristics and characteristics of their social world and environment (Hollich et al., Reference Hollich, Hirsh-Pasek, Golinkoff, Brand, Brown, Chung, Hennon and Rocroi2000). Parents are a vital source of language exposure, repetition, contextual support, and scaffolding; thus, when parents are less able to be involved in this process, children may struggle in their language development. The current study provided evidence that maternal mental health may modulate the quality of the home language environment and the expressive language of children. Lastly, the current study provided some indications that maternal depression and anxiety may have an impact on mother's ability to accurately report on their children's language development. These findings highlight the need to better understand how caregivers’ mental health shapes children's environments and language development over time.
Acknowledgements
We thank the School of Family Life, and the College of Family, Home, and Social Sciences at BYU, and recognize the generous support of the many private donors who provided assistance for this project. We also thank those families who were willing to spend valuable hours with our team in interviews and the many students who assisted in conducting the interviews.