Talking to Children Matters: Early Language Experience PDF

Psychological Science http://pss.sagepub.com/ Talking to Children Matters: Early Language Experience Strengthens Processing and Builds Vocabulary Adriana Weisleder and Anne Fernald Psychological Science 2013 24: 2143 originally publish...

Psychological Science http://pss.sagepub.com/ Talking to Children Matters: Early Language Experience Strengthens Processing and Builds Vocabulary Adriana Weisleder and Anne Fernald Psychological Science 2013 24: 2143 originally published online 10 September 2013 DOI: 10.1177/0956797613488145 The online version of this article can be found at: http://pss.sagepub.com/content/24/11/2143 Published by: http://www.sagepublications.com On behalf of: Association for Psychological Science Additional services and information for Psychological Science can be found at: Email Alerts: http://pss.sagepub.com/cgi/alerts Subscriptions: http://pss.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav >> Version of Record - Nov 8, 2013 OnlineFirst Version of Record - Sep 10, 2013 What is This? Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 488145 research-article2013 PSSXXX10.1177/0956797613488145Talking to Children MattersWeisleder, Fernald Research Article Psychological Science Talking to Children Matters: Early 24(11) 2143­–2152 © The Author(s) 2013 Reprints and permissions: Language Experience Strengthens sagepub.com/journalsPermissions.nav DOI: 10.1177/0956797613488145 Processing and Builds Vocabulary pss.sagepub.com Adriana Weisleder and Anne Fernald Department of Psychology, Stanford University Abstract Infants differ substantially in their rates of language growth, and slow growth predicts later academic difficulties. In this study, we explored how the amount of speech directed to infants in Spanish-speaking families low in socioeconomic status influenced the development of children’s skill in real-time language processing and vocabulary learning. All- day recordings of parent-infant interactions at home revealed striking variability among families in how much speech caregivers addressed to their child. Infants who experienced more child-directed speech became more efficient in processing familiar words in real time and had larger expressive vocabularies by the age of 24 months, although speech simply overheard by the child was unrelated to vocabulary outcomes. Mediation analyses showed that the effect of child-directed speech on expressive vocabulary was explained by infants’ language-processing efficiency, which suggests that richer language experience strengthens processing skills that facilitate language growth. Keywords language development, poverty, environmental effects, individual differences, cognitive processes Received 1/23/13; Revision accepted 3/27/13 At any given age, children show wide variability in their gesture from caregivers. Some parents talk more and use levels of language proficiency (Fenson et al., 1994). richer vocabulary and gestures in interactions with infants Although differences in verbal abilities among individuals than do others, and such differences in the quantity and are influenced to some extent by genetic factors (Oliver & quality of language input account in part for later dispari- Plomin, 2007), the contributions of early experience to ties among children in lexical and grammatical develop- such differences are also substantial. Factors associated ment, both within and between SES groups (Hart & Risley, with socioeconomic status (SES) are strongly related to 1995; Hoff, 2003b; Huttenlocher, Waterfall, Vasilyeva, variation in language outcomes. By the time they enter Vevea, & Hedges, 2010; Pan, Rowe, Singer, & Snow, 2005; kindergarten, children from disadvantaged backgrounds Rowe & Goldin-Meadow, 2009). A second source of vari- differ significantly from their more advantaged peers in ability in language learning relates to infants’ speech-pro- verbal and other cognitive abilities (Ramey & Ramey, cessing skills. Differences among infants in phonological 2004), and these disparities are predictive of later aca- discrimination (Tsao, Liu, & Kuhl, 2004) and spoken- demic success or failure (Hart & Risley, 1995; Lee & word recognition (Fernald, Perfors, & Marchman, 2006; Burkam, 2002). Identifying the environmental factors that Singh, Reznick, & Xuehua, 2012) predict early vocabulary shape these consequential differences in early language growth. In experimental studies in which infants look at proficiency is critical for remediating the growing achieve- pictures of familiar objects as one object is named, the ment gaps between children from impoverished and affluent families (Duncan & Murnane, 2011). Corresponding Author: What accounts for differences among children in early Adriana Weisleder, Department of Pediatrics, New York University language growth? One source of variability in rates of lan- School of Medicine, 550 First Ave., OBV A529, New York, NY 10016 guage learning is differential access to language and E-mail: [email protected] Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 2144 Weisleder, Fernald infants’ speed and accuracy in recognizing the object speech with an observer present (Hurtado et al., 2008; name and identifying the correct picture in real time pre- Pan et al., 2005), we collected more extensive and repre- dicts both early vocabulary development and later lan- sentative recordings of infants’ interactions with family guage and cognitive skills (Fernald et al., 2006; Fernald & members during a typical day at home. We examined Marchman, 2012; Marchman & Fernald, 2008). how these naturalistic measures of caregiver speech These studies have shown that children’s language related to experimental measures of language processing outcomes are linked both to early experience with lan- and to parent reports of expressive vocabulary. guage and to speech-processing skills in infancy, but it is not well understood how these two influences work Participants together during development to promote vocabulary growth. In the research reported here, we investigated Participants were 29 Spanish-learning infants (19 females, two alternative possibilities. One is that language experi- 10 males) tested at the ages of 19 and 24 months. Parents ence and language-processing skill are separate factors reported that all infants were full term and typically that contribute independently to lexical development. developing. An additional 6 children were excluded from That is, variation in children’s vocabulary growth could the sample because the home recordings were not con- result from differences in children’s exposure to speech— ducted properly (n = 3), the computer malfunctioned and, thus, in their opportunities to learn new words—as during testing (n = 2), or the infant received a diagnosis well as from preexisting differences in children’s ability of developmental delay during the course of the study to process speech efficiently, whereby some children are (n = 1). Family income ranged from less than $25,000 to better able to take advantage of the learning opportuni- $75,000 per year, with 79% of families reporting a yearly ties available to them. income below the federal poverty line. Although parents Another possibility is that early experience with lan- varied in years of education, most had not completed guage influences the development of efficiency in real- high school. Maternal education ranged from 4 to 16 time language processing. That is, experience in hearing years (M = 10, SD = 3) and was used as the primary index language from caregivers may sharpen infants’ skill in of SES, controlled in all analyses. All parents were native processing speech and, hence, improve their ability to Spanish speakers, and Spanish was the primary language learn from future language input. Our recent study com- in the homes of all of the children, with English constitut- paring infants from higher- and lower-SES families ing less than 25% of the language spoken in the home. showed that significant disparities in language-process- ing efficiency were already present when children were Measures of the home language 18 months of age (Fernald, Marchman, & Weisleder, 2013), which suggests that experiential factors associated environment with SES may contribute to differences in processing To measure adult speech accessible to infants in different skill. In addition, one previous study showed that infants families, we made audio recordings during a typical day exposed to richer language input were more efficient in at home when the child was 19 months old. A digital language processing (Hurtado, Marchman, & Fernald, recorder in the chest pocket of specialized clothing worn 2008). However, in this latter study, the relation between by the child enabled unobtrusive recording of both child- language experience and processing efficiency could be directed and overheard speech in daily interactions explained by children’s vocabulary size. To address this among family members (Ford, Baer, Xu, Yapanel, & Gray, gap, we asked the following questions: Is early experi- 2009). Parents were asked to record their child during a ence with language linked to the development of effi- typical day in the home and to keep a log of the locations ciency in language processing and, if so, do differences in which the recording was conducted, who was present, in processing efficiency mediate the well-established the main activities the child was engaged in, and whether relation between early language experience and later anything atypical occurred. vocabulary knowledge? Answers to these questions will Families were recorded for an average of 11 hr further the understanding of the developmental path- (range = 4–16) over the course of 1 to 6 days. Using infor- ways linking early language experience, speech-process- mation recorded in parents’ logs, we selected for each ing efficiency, and vocabulary growth. family the longest available recording that represented a typical day. Estimates of adult word counts based on these recordings were highly correlated with adult word Method counts aggregated over all days of recording (r =.84, p < We focused on infants from low-SES Latino families, a.001). After we eliminated nap times, the final sample of rapidly growing population of children in the United recordings had an average duration of 7 hr (range = States at risk for academic difficulties (Reardon & Galindo, 3–13). Differences in the length of these recordings were 2009). Rather than relying on short samples of mothers’ controlled for in all analyses. Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 Talking to Children Matters 2145 The home recordings were analyzed using LENA anal- pairs of images (e.g., of a dog and a baby) while hearing ysis software (Xu, Yapanel, & Gray, 2009). This software sentences naming one of the pictures. Children were was used to process the audio files and yield estimates of tested on words that are frequent in child-directed speech different components of the infant’s language environ- and are familiar to most children in the participants’ age ment, including the number of adult word tokens and the range, based on the MCDI lexical norms (Dale & Fenson, number of child vocalizations. The accuracy of these esti- 1996). When children were 19 months old, the eight tar- mates for English-language recordings has been estab- get nouns were el perro (dog), el libro (book), el jugo lished in previous studies (Oetting, Hartfield, & Pruitt, (juice), el globo (balloon), el zapato (shoe), el plátano 2009; Oller et al., 2010; Xu et al., 2009). To assess the (banana), la pelota (ball), and la galleta (cookie). When accuracy of the adult word estimates in Spanish-language children were 24 months old, four additional familiar environments, we asked native Spanish speakers other- words were included: el caballo (horse), el pájaro (bird), wise uninvolved in this research to transcribe 60-min la cuchara (spoon), and la manzana (apple). All of the samples from 10 of the home recordings. Our analysis of words were presented in simple sentence frames ending these transcriptions revealed a high correlation between with the target noun, for example, “Mira el perro” (“Look automated estimates of adult words and transcriber- at the dog”). based word counts (r =.80), which confirmed that the The speech stimuli were recorded by a native Spanish- LENA system provided reliable estimates of adult words speaking adult female and edited for prosodic compara- in Spanish-language environments (further details can be bility. Visual stimuli consisted of digital pictures of objects found in Supporting Methods in the Supplemental presented in yoked pairs. The pairs were matched for Material available online). visual salience, the grammatical gender of the object To differentiate between speech directed to the child name, and lexical familiarity on the basis of MCDI lexical and speech overheard by the child, we had native norms (Dale & Fenson, 1996). Each object was presented Spanish-speaking coders listen to each of the home an equal number of times as a target and as a distracter. recordings and classify each 5-min segment as containing Table S1 in the Supplemental Material lists the word pairs speech that was predominantly child directed or over- as presented in the experiments at 19 and 24 months. heard. The number of adult word tokens in segments Children sat on their parent’s lap approximately 60 cm classified as child directed, divided by the duration of the from the screen, and parents wore opaque sunglasses to recording, served as our measure of child-directed block their view of the images. On each trial, two pic- speech; the number of adult word tokens in segments tures were presented in silence for 2 s, followed by an classified as overheard, divided by the duration of the approximately 3-s speech stimulus and a 1-s silent period recording, served as our measure of overheard speech; during which the pictures remained on-screen. When and the number of speechlike vocalizations produced by children were 19 months old, the 8 target nouns were the target child in segments classified as child directed, presented four times each for a total of 32 test trials; divided by the duration of the recording, served as our when children were 24 months old, the 12 target nouns measure of child vocalizations (see Supporting Methods were presented three times each for a total of 36 test tri- in the Supplemental Material for further details). From als. Side of target presentation was counterbalanced these measures, we estimated the number of words or across trials, and trial order was counterbalanced across vocalizations per hour and in a 10-hr waking day. participants. The entire test session lasted 4 to 5 min. Children’s looking patterns were video recorded. Measures of expressive vocabulary Subsequently, highly trained coders blind to target loca- tion coded each child’s gaze patterns. For each frame, When the children were 24 months old, parents com- coders noted whether the child was fixating the left or pleted the MacArthur-Bates Inventario del Desarrollo right picture, in transition between the two pictures, or de Habilidades Comunicativas: Palabras y Enunciados looking away from both. A second coder independently (Inventario II; Jackson-Maldonado et al., 2003), the recoded all trials for 28% of the participants at each age. Spanish-language version of the MacArthur-Bates Com­ The proportion of frames on which observers agreed municative Development Inventories (MCDI). Productive- was 99%. vocabulary scores were based on the number of words Speech-processing efficiency was calculated as the parents reported that their child understood and said proportion of time the infant spent fixating the target (“comprende y dice”). picture out of total time spent looking at either the target or the distracter picture, within 300 to 1,800 ms of target- Measures of language-processing word onset (Fernald et al., 2008). Only those trials on which the child was looking at either the target or the efficiency distracter picture at the onset of the target noun were In the looking-while-listening task (Fernald, Zangl, included in these analyses. This measure of efficiency Portillo, & Marchman, 2008), infants were presented with captured children’s tendency to shift rapidly toward the Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 2146 Weisleder, Fernald target picture after initially looking at the distracter, as Links between language experience well as their tendency to maintain attention to the target and vocabulary when they were already looking at it. See Videos S1 and S2 and Legends for Supplemental Videos in the We next asked whether differences among families in Supplemental Material to view videos of children partici- amount of speech available to infants predicted children’s pating in the looking-while-listening task. vocabulary 6 months later. Those children who heard more child-directed speech at 19 months had larger vocabularies at 24 months (r =.57, p <.01), a result con- Results sistent with previous findings (Hoff, 2003b; Hurtado Among these low-SES families, there was striking vari- et al., 2008). In contrast, differences in exposure to over- ability in the total amount of adult speech accessible to heard speech directed to other adults and children were the infant, which ranged from almost 29,000 adult words not related to infants’ vocabulary size (r =.25, p =.2), to fewer than 2,000 words over the course of 10 hr (see which suggests that language spoken directly to infants is Fig. 1 for variability across the 29 participating families). more supportive of early lexical development than is When only talk addressed directly to the child was con- speech simply overheard by infants. sidered, these differences were even more extreme: In One alternative possibility is that infants with more one family, caregivers spoke more than 12,000 words to precocious language skills tend to vocalize more often, the infant, whereas in another family, the infant heard eliciting more speech from their caregivers. If this is true, only 670 words of child-directed speech during an entire and if infants who produce more speech early on have day—an 18-fold difference in the amount of child- larger productive vocabularies at 24 months, this might directed speech available to these two children. These account for the relation between child-directed speech differences in parental engagement were uncorrelated and later vocabulary (Newport, Gleitman, & Gleitman, with maternal education (r =.29, p =.13). In addition, 1977). To examine this possibility, we first analyzed the amount of child-directed speech was not correlated relation between infant vocalizations and child-directed with amount of overheard speech (r =.17, p =.38), which speech at 19 months. Infants who vocalized more often suggests that the observed differences in speech directed did hear more child-directed speech (r =.41, p <.05), to children were not due to overall differences in talk- which suggests some degree of concordance between ativeness among families but, rather, to caregivers’ degree infants’ and caregivers’ vocalizations. However, even after of verbal engagement with their infants. controlling for infant vocalizations at 19 months, we Overheard Speech Child-Directed Speech 30,000 25,000 Adult Words in a 10-Hr Day 20,000 15,000 10,000 5,000 0 1 8 15 22 29 Families With 19-Month-Old Infants Fig. 1. Mean number of words that infants heard adults speak in a typical day at home for each family and each type of speech. Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 Talking to Children Matters 2147 found that the relation between child-directed speech speech explained gains in processing efficiency from 19 and infants’ vocabulary at 24 months remained robust to 24 months. A particularly important finding was that (r =.51, p <.01). This result suggests that, over and above the relation between child-directed speech and process- differences in infants’ expressive language skill early on, ing efficiency at 24 months remained significant when exposure to child-directed speech predicted later vocab- controlling for differences in vocabulary size at 24 months ulary size. (r =.39, p <.05). This result indicates that over and above differences in vocabulary knowledge, children who were exposed to more child-directed speech were better able Links between language experience to identify familiar words during real-time language and language processing processing. These results support previous findings showing that early language experience predicts later vocabulary Can differences in processing explain knowledge. But are children who hear more child- the link between language experience directed speech also more efficient in processing familiar and vocabulary? words in real time? Amount of exposure to child-directed speech was reliably correlated with children’s processing Next, we asked whether the effect of language experi- efficiency at 19 months (r =.44, p <.05) and at 24 months ence on processing efficiency helps explain the well- (r =.51, p <.01; see Figs. 2 and 3b for illustrations of established relation between child-directed speech and these relations). Moreover, controlling for differences in vocabulary. We used mediation analysis to examine processing at 19 months, we found that children who whether processing skill at 19 months accounted for heard more child-directed speech were more efficient in the link between child-directed speech and 24-month language processing at 24 months than were those who vocabulary (while controlling for maternal education, heard less child-directed speech (r =.47, p <.05). This recording length, and infant vocalizations at 19 months). result indicates that amount of exposure to child-directed The scatter plots in Figure 3 illustrate the first three steps Infants Who Heard More Child-Directed Speech at 19 Months Infants Who Heard Less Child-Directed Speech at 19 Months 1.00 Mira el P E R R O Mean Proportion of Time Spent Looking to Target Picture at 24 Months.75.50.25 0 300 600 900 1,200 1,500 1,800 Time (ms) From Target-Noun Onset Fig. 2. Mean proportion of trials on which 24-month-old children looked to the target picture, measured from the onset of the target noun. Infants who heard more child-directed speech at 19 months and infants who heard less child-directed speech at 19 months were grouped on the basis of a median split. The dashed vertical line represents target-noun offset; error bars represent standard errors across participants. Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 2148 Weisleder, Fernald a b c 600 80 600 r 2 =.27 r 2 =.25 r 2 =.30 Processing Efficiency at 19 Months 70 Vocabulary at 24 Months Vocabulary at 24 Months 400 400 60 200 200 50 0 40 0 0 300 600 900 1,200 1,500 0 300 600 900 1,200 1,500 40 50 60 70 80 Child-Directed Speech: Child-Directed Speech: Processing Efficiency Number of Adult Words Per Hour Number of Adult Words Per Hour at 19 Months d Processing Efficiency (19 Months) β = 0.76* β = 6.85* β = 12.61** Vocabulary Child-Directed Speech Size (19 Months) β = 7.41 (24 Months) Fig. 3. Results. The three scatter plots (with best-fitting regression lines) show zero-order correlations between (a) vocabulary size (number of words) at 24 months and child-directed speech at home, (b) processing efficiency (mean percentage of time spent looking to the target picture) at 19 months and child-directed speech, and (c) vocabulary size at 24 months and processing efficiency at 19 months. Vocabulary size was measured as the number of words produced on the MacArthur-Bates Inventario del Desarrollo de Habilidades Comunicativas: Palabras y Enunciados (Jackson- Maldonado et al., 2003). The mediation model (d) shows the effect of child-directed speech at 19 months on vocabulary size at 24 months, as medi- ated by processing efficiency at 19 months. Along the lower path, the solid and dashed arrows show results when the mediator was not included and was included in the model, respectively. Asterisks indicate significant paths (*p <.05, **p <.01). of the mediation analysis: Exposure to child-directed be significantly reduced when the mediator variable speech at 19 months predicted vocabulary at 24 months (processing efficiency) is included in the model. As (Fig. 3a); exposure to child-directed speech also pre- shown in Figure 3d, the parameter estimate for the effect dicted processing efficiency at 19 months (Fig. 3b); and of child-directed speech on vocabulary was reduced processing efficiency at 19 months predicted vocabulary from 12.61 to 7.41 when processing efficiency was at 24 months (Fig. 3c), even when we controlled for included in the model. A bootstrap analysis (Preacher & child-directed speech. Hayes, 2004) of the significance of the indirect effect Finally, a critical condition for mediation is that the yielded a 95% confidence interval (corrected for bias) of path coefficient between the predictor variable (child- 0.44 to 13.61. This result confirmed that the mediation directed speech) and the outcome variable (vocabulary) was significant and suggests that language experience Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 Talking to Children Matters 2149 promotes vocabulary development at least in part via its variability in a cognitive skill that facilitates further lan- influence on processing efficiency. The final model guage learning. explained 47% of the variance in children’s vocabularies at 24 months. Discussion Are differences in processing This research yielded three main results. First, we found efficiency explained by infants’ that variation in experience with child-directed speech in low-SES Spanish-speaking families predicted children’s knowledge of the target words? later vocabulary. This result replicates findings from other One potential concern is that some children may have studies linking caregiver speech and vocabulary develop- been unfamiliar with certain target words used in the ment in low-SES children (Hurtado et al., 2008; Pan et al., study, in which case variability in processing efficiency 2005), but our study went beyond earlier research by might simply reflect differences in children’s knowledge using all-day recordings of daily interactions in the home of these words. To control for this possibility, we col- to sample children’s early language environments. Thus, lected an independent measure of each participant’s our measures of child-directed speech minimized poten- familiarity with the target words. Using a list of only the tial artifacts introduced by the presence of an observer or words used in the study, we asked parents whether their by parents’ reactions to a laboratory setting. Second, by child understood each target word. According to parents’ recording interactions with multiple family members and reports, all of the target words were understood by 66% identifying different sources of adult speech accessible to of the children at 19 months and by 72% of the children the child, we found that only speech addressed directly at 24 months. For each child, we removed those trials to the infant, and not speech in adult conversations over- containing target words that the child was reported not to heard by the child, facilitated vocabulary learning at this understand and then recomputed the processing effi- age, a result consistent with recent findings from studies ciency measures. After rerunning the mediation model of children in middle-class English-speaking families in reported earlier, we found that the pattern of results the United States (Shneidman, Arroyo, Levine, & Goldin- remained the same: Child-directed speech was related to Meadow, 2013) and in Yucatec Mayan families (Shneidman processing efficiency at 19 months (r =.40, p <.05), and & Goldin-Meadow, 2012). processing efficiency at 19 months predicted vocabulary Third, and most important, speech-processing effi- at 24 months (r =.53, p <.01), even when controlling for ciency mediated the relation between child-directed child-directed speech (r =.41, p <.05). Finally, the param- speech and vocabulary. This result shows that a critical eter estimate for the effect of child-directed speech on step in the path from early language experience to later vocabulary was significantly reduced from 12.61 to 8.75 vocabulary knowledge is the influence of language when processing efficiency was included in the model, exposure on infants’ speech-processing skill. In previous which indicates that processing efficiency mediated the studies, one explanation proposed for the association link between child-directed speech and vocabulary. between exposure to more child-directed speech and In a final analysis, we included only those children faster vocabulary growth has been that more diverse lan- whose mean level of accuracy was greater than 50% at 19 guage from caregivers provides children with more mod- months (n = 22), thus excluding those whose overall els to learn from as they begin to build a lexicon (e.g., level of performance was at or below chance level. This Hoff, 2003b; Rowe & Goldin-Meadow, 2009). Our find- analysis revealed even stronger correlations between ings reveal an additional mechanism by which differ- child-directed speech and processing efficiency (r =.58, ences in early language experience lead to differences in p <.01) and between processing efficiency and later vocabulary size: Infants who hear more talk have more vocabulary (r =.62, p <.01). Moreover, even in this opportunities to interpret language and to exercise skills smaller sample, processing efficiency mediated the link that are vital to word learning, such as segmenting speech between child-directed speech and vocabulary (i.e., the and accessing lexical representations (Gershkoff-Stowe, parameter estimate for the effect of child-directed speech 2002; Saffran, Newport, & Aslin, 1996). As a result, infants on vocabulary was significantly reduced from 15.61 to with more exposure to child-directed speech orient to 8.86 when processing efficiency was included in the familiar words more quickly and accurately when inter- model). These results provide further evidence that dif- preting speech in real time, which enables them to learn ferences in processing efficiency do not simply reflect new words and facilitates rapid vocabulary growth. variability in children’s all-or-nothing knowledge of the Our results also give rise to a challenging question: target words. Instead, differences in how quickly and reli- What factors explain the striking disparities observed ably children interpret familiar words in real time reflect between families in the amount of verbal stimulation Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 2150 Weisleder, Fernald provided to infants? Studies comparing advantaged and engage infants in language-rich interactions, given that disadvantaged families have shown that SES differences these beliefs may be more malleable than other influen- are linked to variability both in speech and gesture tial factors. directed to children and in children’s language outcomes Our results reveal that caregiver talk has direct as well (Hoff, 2003b; Huttenlocher et al., 2010; Rowe & Goldin- as indirect influences on lexical development. More Meadow, 2009). However, in such between-group com- exposure to child-directed speech not only provides parisons, differences in caregiver input are confounded more models for learning words but also sharpens infants’ with many other factors associated with SES that could emerging lexical processing skills, with cascading bene- also lead to variability in language learning—such as fits for vocabulary learning. If increased opportunities for parental education, access to resources, crowded living verbal interaction can strengthen critical processing skills conditions, and family stress levels (Evans, 2004). By that enable more efficient learning, then interventions focusing on differences within a homogeneous group of aimed at increasing parents’ verbal engagement with disadvantaged families, rather than on differences their infants have the potential to change the course of between SES groups, we reduced variability in these con- vocabulary growth and, in turn, to improve later out- founding factors. comes for disadvantaged children. Given this narrower focus, it was surprising to dis- cover differences in the amount of child-directed speech Author Contributions between families that were almost as large as those dif- A. Weisleder and A. Fernald developed the study concept and ferences reported in the landmark study by Hart and designed the study. A. Weisleder performed the research and Risley (1995), whose sample spanned a broad demo- analyzed the data. A. Weisleder and A. Fernald drafted the graphic range from poverty-level to professional families. manuscript. Although Hart and Risley found significant differences between SES groups, with a 20-fold difference in verbal Acknowledgments stimulation between parents who were the most and the We give special thanks to V. A. Marchman, R. Hoffmann Bion, least verbally engaged with their infants, our findings R. D. Fernald, C. M. Fausey, and three anonymous reviewers revealed an 18-fold difference in caregiver talk to infants for comments on earlier versions of the manuscript and to within a more demographically homogeneous group of N. Hurtado, L. Rodriguez Mata, C. Coon, M. Barraza, J. Villanueva, A. Arroyo, N. Otero, V. Limón, L. Martinez, and the disadvantaged families. Moreover, the differences in staff of the Center for Infant Studies at Stanford University for parental engagement observed within this low-SES sam- help with data collection and coding. We are grateful to the ple were not correlated with maternal education. An children and parents who participated. important implication of these findings is that although variability in parenting behaviors is consistently linked to Declaration of Conflicting Interests factors related to SES, there is also considerable variabil- The authors declared that they had no conflicts of interest with ity in parental verbal engagement that is independent of respect to their authorship or the publication of this article. social class. In ongoing research, we are exploring other factors Funding that could explain observed differences in children’s lan- This research was supported by a grant from the National guage environments. Previous studies have discussed Institutes of Health (R01 DC008838) to A. Fernald. several such factors, including variability in parents’ own verbal abilities or conversational style (Hoff-Ginsberg, Supplemental Material 1991), in the activities that parents tend to engage in with Additional supporting information may be found at http://pss their children (Hoff, 2003a), and in parental stress and.sagepub.com/content/by/supplemental-data emotional well-being (Conger, McCarty, Yang, Lahey, & Kropp, 1984). In addition, some studies have found that References parents from different sociocultural groups have different Conger, R. D., McCarty, J. A., Yang, R. K., Lahey, B. B., & beliefs about the role they play in children’s communica- Kropp, J. P. (1984). Perception of child, child-rearing val- tive development (Heath, 1983), and Rowe (2008) found ues, and emotional distress as mediating links between that parents’ knowledge and beliefs about child develop- environmental stressors and observed maternal behavior. ment mediated the relation between SES and caregiver Child Development, 55, 2234–2247. speech to children. Although not assessed in the current Dale, P. S., & Fenson, L. (1996). Lexical development norms for study, parental beliefs are one important factor to con- young children. Behavior Research Methods, Instruments, & sider in explaining differences in caregivers’ tendency to Computers, 28, 125–127. Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 Talking to Children Matters 2151 Duncan, G. J., & Murnane, R. J. (2011). Whither opportunity? Jackson-Maldonado, D., Thal, D. J., Marchman, V. A., Newton, T., Rising inequality, schools, and children’s life chances. New Fenson, L., & Conboy, B. T. (2003). MacArthur Inventarios York, NY: Russell Sage Foundation. del Desarrollo de Habilidades Comunicativas: User’s guide Evans, G. W. (2004). The environment of childhood poverty. and technical manual. Baltimore, MD: Brookes. American Psychologist, 59, 77–92. Lee, V. E., & Burkam, D. T. (2002). Inequality at the starting Fenson, L., Dale, P. S., Reznick, J. S., Bates, E., Thal, D. J., gate: Social background differences in achievement as & Pethick, S. J. (1994). Variability in early communicative children begin school. Washington, DC: Economic Policy development. Monographs of the Society for Research in Institute. Child Development, 59(5, Serial No. 242). Marchman, V. A., & Fernald, A. (2008). Speed of word recogni- Fernald, A., & Marchman, V. A. (2012). Individual differences in tion and vocabulary knowledge in infancy predict cognitive lexical processing at 18 months predict vocabulary growth and language outcomes in later childhood. Developmental in typically developing and late-talking toddlers. Child Science, 11, F9–F16. Development, 83, 203–222. Newport, E. L., Gleitman, L. R., & Gleitman, H. (1977). Mother, Fernald, A., Marchman, V. A., & Weisleder, A. (2013). SES dif- I’d rather do it myself: Some effects and non-effects of ferences in language processing skill and vocabulary are maternal speech style. In C. E. Snow & C. A. Ferguson evident at 18 months. Developmental Science, 16, 234–248. (Eds.), Talking to children: Language input and acquisition Fernald, A., Perfors, A., & Marchman, V. A. (2006). Picking up (pp. 109–149). Cambridge, England: Cambridge University speed in understanding: Speech processing efficiency and Press. vocabulary growth across the 2nd year. Developmental Oetting, J. B., Hartfield, L., & Pruitt, S. (2009). Exploring LENA Psychology, 42, 98–116. as a tool for researchers and clinicians. The ASHA Leader, Fernald, A., Zangl, R., Portillo, A. L., & Marchman, V. A. (2008). 14, 20–22. Looking while listening: Using eye movements to moni- Oliver, B. R., & Plomin, R. (2007). Twins’ Early Development tor spoken language comprehension by infants and young Study (TEDS): A multivariate, longitudinal genetic inves- children. In I. A. Sekerina, E. M. Fernandez, & H. Clahsen tigation of language, cognition and behavior problems (Eds.), Developmental psycholinguistics: On-line methods in from childhood through adolescence. Twin Research and children’s language processing (pp. 97–135). Amsterdam, Human Genetics, 10, 96–105. The Netherlands: John Benjamins. Oller, D. K., Niyogi, P., Gray, S., Richards, J. A., Gilkerson, J., Ford, M., Baer, C. T., Xu, D., Yapanel, U., & Gray, S. (2009). Xu, D.,... & Warren, S. F. (2010). Automated vocal anal- The LENA™ language environment analysis system: ysis of naturalistic recordings from children with autism, Audio specifications of the DLP-0121. Retrieved from language delay, and typical development. Proceedings of http://www.lenafoundation.org/TechReport.aspx/Audio_ the National Academy of Sciences, USA, 107, 13354–13359. Specifications/LTR-03-2 Pan, B. A., Rowe, M. L., Singer, J. D., & Snow, C. E. (2005). Gershkoff-Stowe, L. (2002). Object naming, vocabulary growth, Maternal correlates of growth in toddler vocabulary pro- and the development of word retrieval abilities. Journal of duction in low-income families. Child Development, 76, Memory and Language, 46, 665–687. 763–782. Hart, B. M., & Risley, T. R. (1995). Meaningful differences in Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures the everyday experience of young American children. for estimating indirect effects in simple mediation models. Baltimore, MD: Brookes. Behavior Research Methods, Instruments, & Computers, 36, Heath, S. B. (1983). Ways with words: Language, life, and work 717–731. in communities and classrooms. Cambridge, England: Ramey, C. T., & Ramey, S. L. (2004). Early learning and school Cambridge University Press. readiness: Can early intervention make a difference? Merrill- Hoff, E. (2003a). Causes and consequences of SES-related dif- Palmer Quarterly, 50, 471–491. ferences in parent-to-child speech. In M. H. Bornstein & Reardon, S. F., & Galindo, C. (2009). The Hispanic-White R. H. Bradley (Eds.), Socioeconomic status, parenting, and achievement gap in math and reading in the elementary child development (pp. 147–160). Mahwah, NJ: Erlbaum. grades. American Educational Research Journal, 46, 853– Hoff, E. (2003b). The specificity of environmental influence: 891. Socioeconomic status affects early vocabulary development Rowe, M. L. (2008). Child-directed speech: Relation to socioeco- via maternal speech. Child Development, 74, 1368–1378. nomic status, knowledge of child development and child Hoff-Ginsberg, E. (1991). Mother-child conversation in dif- vocabulary skill. Journal of Child Language, 35, 185–205. ferent social classes and communicative settings. Child Rowe, M. L., & Goldin-Meadow, S. (2009). Differences in early Development, 62, 782–796. gesture explain SES disparities in child vocabulary size at Hurtado, N., Marchman, V. A., & Fernald, A. (2008). Does input school entry. Science, 323, 951–953. influence uptake? Links between maternal talk, processing Saffran, J. R., Newport, E. L., & Aslin, R. N. (1996). Word speed and vocabulary size in Spanish-learning children. segmentation: The role of distributional cues. Journal of Developmental Science, 11, F31–F39. Memory and Language, 35, 606–621. Huttenlocher, J., Waterfall, H., Vasilyeva, M., Vevea, J., & Shneidman, L. A., Arroyo, M. E., Levine, S., & Goldin-Meadow, Hedges, L. V. (2010). Sources of variability in children’s S. (2013). What counts as effective input for word learning? language growth. Cognitive Psychology, 61, 343–365. Journal of Child Language, 40, 672–686. Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013 2152 Weisleder, Fernald Shneidman, L. A., & Goldin-Meadow, S. (2012). Language Tsao, F.-M., Liu, H.-M., & Kuhl, P. K. (2004). Speech perception in input and acquisition in a Mayan village: How important is infancy predicts language development in the second year of directed speech? Developmental Science, 15, 659–673. life: A longitudinal study. Child Development, 75, 1067–1084. Singh, L., Reznick, J. S., & Xuehua, L. (2012). Infant word Xu, D., Yapanel, U., & Gray, S. (2009). Reliability of the LENA™ segmentation and childhood vocabulary development: language environment analysis system in young children’s A longitudinal analysis. Developmental Science, 15, 482– natural home environment. Retrieved from http://www 495..lenafoundation.org/TechReport.aspx/Reliability/LTR-05-2 Downloaded from pss.sagepub.com at Stanford University Libraries on December 9, 2013