INT J LANG COMMUN DISORD, MAY–JUNE VOL. 50, NO. 3, 358–373

2015,

Research Report Early language impairments and developmental pathways of emotional problems across childhood Shaun Goh Kok Yew and Richard O’Kearney∗ Research School of Psychology, Australian National University, Canberra, ACT, Australia

(Received February 2014; accepted September 2014) Abstract Background: Language impairments are associated with an increased likelihood of emotional difficulties later in childhood or adolescence, but little is known about the impact of LI on the growth of emotional problems. Aims: To examine the link between early language status (language impaired (LI), typical language (TL)) and the pattern and predictors of growth in emotional difficulties from school entry to the start of high school in a large cohort of Australian children. Methods & Procedures: Unconditional latent growth curves of emotional difficulties were modelled across four waves (ages 4–5, 6–7, 8–9 and 10–11) using data from 1627 boys (280 LI, 1347 TL) and 1609 girls (159 LI, 1450 TL). Conditional latent growth curves estimated the main effects of LI on the severity and slope of growth in emotional problems. Simultaneous multiple regression tested the interaction between language status and the other predictors of the development of emotional symptoms. Outcomes & Results: LI predicted a significant persistent elevation in severity of emotional difficulties across childhood among boys (d = 0.33–0.57) and girls (d = 0.25–0.39) but was not associated with their growth. LI moderated the association between hostile parenting and the severity of emotional symptoms for boys and the effect of socioeconomic status (SES) and temperamental sociability on the linear and quadratic growth of emotional problems for girls but had no impact on the influence of other predictors. Conclusions & Implications: There is no effect of LI on the characteristic rate and shape of growth in emotional symptoms across childhood although LI children maintain elevated severities of emotional difficulties. The associations between child reactivity, peer problems, prosocial behaviours, maternal distress and parental warmth and the development of emotional difficulties were the same for LI and TL children. LI enhanced the influence of hostile parenting on a higher severity of emotional symptoms for boys and of lower SES on a faster rate of development of emotional symptoms for girls. LI offset the usual protective effect of higher sociability and the usual vulnerability of higher social avoidance to a faster increase in emotional symptoms with age. Keywords: language difficulties, emotional problems, developmental trajectories.

What this paper adds? What is already known on this subject? Children with early language impairment are more likely to have emotional difficulties later in childhood or adolescence. However, there is little evidence about the impact of language impairment on the trajectory of emotional problems or about what factors contribute to the development of these difficulties in children with language impairment.



Address correspondence to: Richard O’Kearney, Research School of Psychology, Australian National University, Canberra, ACT 0200, Australia; e-mail: [email protected] International Journal of Language & Communication Disorders C 2014 Royal College of Speech and Language Therapists ISSN 1368-2822 print/ISSN 1460-6984 online  DOI: 10.1111/1460-6984.12142

Early language impairments and developmental pathways of emotional problems across childhood

359

What this study adds? This longitudinal cohort study found that despite elevated levels of emotional difficulties across childhood the growth of emotional problems in children with language impairment did not differ from that of typical children. While in general the association of established child, family, peer and social factors and the development of emotional difficulties were similar across both groups, language impairment intensified the influence of hostile parenting for boys and of SES for girls, and neutralized the effect of child sociability for girls.

Introduction Children with early language impairments show higher levels of emotional problems such as depressed mood and anxiety and a higher risk for internalizing disorders later in childhood and in adolescence than children with typical language (Beitchman et al. 2001, Redmond and Rice 2002, Snowling et al. 2006, Wadman et al. 2011). A recent meta-analysis of controlled prospective studies (Yew and O’Kearney 2013) found that 3–8 years olds with specific language impairments were almost twice as likely as their typical language peers to show later clinical levels of emotional problems (RR = 1.84; 95% CI = 1.04–3.25) and that language-impaired children were on average at the 72%ile (95% CI = 65–79%ile) on severity of emotional symptoms at follow-up. Taken together with the high rates of co-occurrence of childhood language impairments and depressive and anxiety disorders in clinical settings (Benner et al. 2002), such estimates indicate a clinically important connection between childhood language impairments and the development of internalizing difficulties like withdrawal, fearfulness, anxiety and depression. It is not clear whether this association represents a direct contribution of early language impairments to the development of internalizing problems or the interaction of language impairments with other predictors of emotional difficulties during development. Some longitudinal studies which consider the initial level of emotional symptoms have reported increased relative risk to later emotional disorders for language-impaired children (Beitchman et al. 2001) while others show a decline in the levels of anxiety and internalizing symptoms over time in language-impaired children compared with typical language peers (Redmond and Rice 2002). Such inconsistencies may reflect study differences in children’s ages at the initial and outcome assessments. They also highlight limitations to conclusions because existing studies examine early language difficulties as a predictor of variability in emotional symptoms at single endpoint ages rather than ask how language impairments influence the development of emotional symptoms. The development of emotional problems A shift towards understanding how early child and environmental factors impact the developmental course of

symptoms across childhood and into the transition to adolescence is particularly important in the case of internalizing disorders because anxiety disorders and major depressive episodes markedly increase in frequency in adolescence (Angold et al. 1998). In addition, other serious maladaptive behaviours, such as substance misuse and non-suicidal self-injurious behaviours, which are linked to emotional dysregulation, often first occur and increase rapidly in adolescence (Jacobson and Gould 2007). Despite the evidence of increased risk to internalizing problems conferred by early language problems and the fundamental role of communication competencies in socio-emotional development relatively little is known about how early language difficulties contribute as a direct or indirect influence onto the course of the emotional difficulties during childhood. The current study aims to address this shortfall in our knowledge. Because children with language problems can be reliably identified at an early age and many are responsive to intervention, such knowledge could contribute to bettertargeted programmes to enhance the modest benefits of current early intervention and prevention strategies for adolescent emotional disorders (Merry et al. 2011). The trajectory of internalizing symptoms across the primary and middle school years and into adolescence identified at the population level typically shows a modest linear increase in withdrawn, anxious and depressive symptoms especially for girls (Bongers et al. 2003, Leve et al. 2005). This sex dependent variability in the development of emotional difficulties is a consistent finding (Hale et al. 2008) and strongly suggests the influences on the development of emotional problems vary between girls and boys. Nevertheless, independent of gender, the trajectory of emotional difficulties prior to adolescence has significant clinical and psychosocial implications. Dekker et al. (2007) found that children with increasing levels of internalizing symptoms had the highest rates of depressive and other mental health problems in adolescence and young adulthood as well as lower educational outcomes and higher rates of referral for mental health care. Similarly, Feng et al. (2008) and Sterba et al. (2007) reported that having a more rapidly increasing trajectory of internalizing symptoms from 2 to 10 years of age predicted both anxiety and depressive disorders and co-morbid anxiety and depressive disorder in the transition to adolescence.

360

Shaun Goh Kok Yew and Richard O’Kearney Language difficulties and the development of emotional problems

Early language and communication difficulties can affect the trajectories of emotional problems principally in two ways. First, language difficulties in early childhood may maintain or aggravate problems in emotional regulation which are related to intrinsic child factors such as behavioural inhibition or emotional reactivity, or which arise due to the influence of family or other socialization contexts. Delays or inhibitions in the use of language to communicate about and conceptualize emotional situations (Zadeh et al. 2007) may result in a deficit in the use of language for self-instruction and self-regulation, and promote the persistence of problematic and immature behavioural strategies to manage fear and distress such as avoidance or acting out (Fujiki et al. 2002). Secondly, language difficulties may have progressive influence on increasingly poorer emotional functioning after entry into school years via the accumulating influences of peer relationships disrupted by communication difficulties and/or academic underachievement as peer relationships and academic performance become more important to self-esteem and wellbeing (Durkin and Conti-Ramsden 2010, Harrison et al. 2009). At present, while there is little empirical evidence from a developmental prospective to directly support either of these positions, there is evidence that language is related to established child, family and peer influences on the development of adolescent internalizing problems. Predictors of emotional problems A shy, behavioural inhibited and fearful temperament, and a negative quality to peer interactions, particularly peer rejection, have been shown to be associated with persistently high or increasing trajectories of internalizing problems across childhood (Leve et al. 2005). Frequent angry and low warmth responses from parents and maternal internalizing symptoms, particularly depression, predict higher levels of internalizing problems across childhood (Leve et al. 2005). Each of these predictive factors has been shown to have connections to language abilities. Maternal depression (Sohr-Preston and Scaramella 2006), the quality of child directed interactions from parents (Hart and Risley 1995) and child temperament (Hart et al. 2004) all impact on the development of the children’s language competencies. In particular, higher levels of shyness and social withdrawal and lower levels of pro-social behaviours have been reported for children with language difficulties (Hart et al. 2004). Children with language difficulties also experience significantly more negative peer interactions including bullying and describe a lower quality of friendships than children with typical language (Durkin

and Conti-Ramsden 2010). In addition to the associations with these specific vulnerabilities, early language abilities are also related to broader factors of gender and social economic status shown to be associated with the development of emotional problems (Hart and Risley 1995). Notwithstanding these interconnections, there are no available data addressing whether early language difficulties restrain or augment the influence that these child, family, peer and social factors have on the developmental course of emotional problems from childhood into adolescence. The current study Broadly, the aim of the current study is to provide evidence about the link between early language problems and the development of emotional problems across the primary and middle school years leading into adolescence and how language impairment interacts with established predictors of the development of emotional difficulties. The study uses data from the prospective population based cohort study of Australian children (The Longitudinal Study of Australian Children— LSAC; Soloff et al. 2005). Using a large community database is appropriate in the present context because it facilitates the estimation of valid population-level trajectories of emotional problems and allows consideration of the interaction of language difficulties and other putative predictors of these trajectories. The LSAC sampling procedure also provides more representativeness than previous prospective studies of language impairment and emotional problems and overcomes potential biases arising out of selection of comparison children. In addition, because LSAC attempts a comprehensive assessment of child, family, social and environmental factors it enables inclusion of the main established predictors of the course of emotional symptoms. The present study uses LSAC data to address two specific questions. Are the magnitude and trajectory of emotional problems across childhood from age 4 to 11 related to early language difficulties? Do early language difficulties affect the influence that child’s temperament, parenting practices, maternal depression, peer relationships and social economic status have on the extent and course of emotional problems across childhood? Methods Participants LSAC is a national cohort study that examines the physical, cognitive, social and emotional development of Australian children and the context of this development at regular intervals across infancy, early, middle and later childhood. Recruitment of the LSAC sample

Early language impairments and developmental pathways of emotional problems across childhood was facilitated by the Australian government using the national Medicare database. The study uses a crosssequential design that follows two cohorts of children. One cohort sampled from children born between March 2003 and February 2004 (B cohort) and the second from children born between March 1999 and February 2000 (K cohort). Data from the K cohort are analysed in the current study. The recruitment process used a two-stage clustered design and stratification to ensure proportional geographic representation for Australian states and territories and urban, rural and remote areas. Response rate for the K cohort was 59.4%. Comprehensive details on the design and sample are available elsewhere (Soloff et al. 2005). At wave 1, a total of 4983 children (2537 boys, 2446 girls) and their parents participated in the K cohort. At wave 1 children ranged in age from 4;3 (years;months) to 5;7, with 80% being within the 4;6–5;0 age span. The mean age was 56.91 months (SD = 2.64).This wave coincides with the pre-school or first year of schooling for Australian children. A similar proportion of boys (50.9%) and girls (49.1%) were included, and boys and girls did not differ by age. The families were a close match to the Australian population of families with a 4–5-year-old child on key characteristics including ethnicity, indigenous status, country of birth, education and income, and whether a language other than English was spoken at home. There have been three subsequent assessments every two years with average retention of families of 88%. The mean age for wave 4 children is 10.9 years. Children retained in the study to wave 4 were more likely to have higher receptive language abilities, existing emotional problems and live in a household with a smaller number of siblings with parents who speak only English at home (Daraganova and Sipthorp 2011 describe full characteristics of families and children who completed wave 4 data).

361

Test—III (PPVT-III) and (5) obtain consent to contact the child’s teacher.

Measures Child’s emotional symptoms The emotional problems subscale of The Strength and Difficulties Questionnaire (SDQ) (Goodman 1997) is used as the measure of internalizing problems at each wave. The SDQ is an informant report measure that identifies behavioural and emotional problems in children and adolescents. It has parent and teacher report forms for children 3–16 and a self-report form for 11–16 year olds. The SDQ provides scores on three clinical subscales (conduct problems, hyperactivity, emotional problems) and two interpersonal subscales (peer problems, pro-social behaviour) as well as a total difficulties score. The emotional problems subscale measures internalizing symptoms such as ‘often seems worried’ and ‘often unhappy, depressed and tearful’ which are endorsed as 0 (not true), 1 (somewhat true) or 2 (certainly true). The emotional subscale scores ranges from 0 to 10, with 10 indicting highest level of symptoms. The questionnaire has good reliability and the individual clinical subscales show sound external validity in predicting risk to a related mental disorder and clinician rated severity of the disorder based on structured interview (Hawes and Dadds 2004) including for an internalizing disorder based on elevated scores on the emotional problems subscale (Hawes and Dadds 2004). The emotional subscale score from the parent report at each of the four waves was used to estimate the trajectory of emotional symptoms from 4–5 to 10–11 years of age.

Child’s Language Procedure Families were visited every 2 years and data collected during an interview and via two questionnaires. Interviewers completed the following tasks: (1) obtain detailed information about the child, including his/her health and aspects of social, cognitive, behavioural and emotional development from the primary carer (the child’s mother in over 97% of cases); (2) obtain sociodemographic information on the family (e.g., household structure and parental labour force status, educational attainment and income); (3) provide a questionnaire for the primary carer covering aspects of the child’s personality and behaviour as well as other areas of family functioning, health and support, to be completed while the interviewer was in the home or to be mailed back; (4) administer the adapted Peabody Picture Vocabulary

Parent’s Evaluation of Developmental Status interview (PEDS) (Glascoe 2000).PEDS was used to assess the child’s language status at wave 1. Two PEDS questions were about the child’s speech and language: Do you have any concerns about how your child talks and makes speech sounds? Do you have any concerns about how your child understands what you say to him/her? Items are scored 0, 1 or 2, as ‘No’, ‘A little’ and ‘Yes’. These questions have shown high accuracy, 83% specificity and 72% sensitivity in detecting language difficulties and have good concurrent validity (> 0.7) with the Test of Language Ability, Expressive language subtest of the Child Development Inventory and the Communication Scale of the Vineland Adaptive Behaviour Scales (Glascoe 2000). Glascoe reported that for children who were identified with parental concerns about speech

362 and language, there was a 79% chance of a referral to speech and language pathology (SLP) services. Peabody Picture Vocabulary Test—III (PPVT-III). The children’s receptive vocabulary at wave 1 was assessed by the interviewer with a shortened version of the PPVT-III. For each word presented, the child was shown a card containing four pictures and was asked to point to the picture corresponding to the word (e.g., ‘Show me wrapping’). The K cohort had a mean of 64.09 and SD of 6.16. Exclusion criteria Non-verbal intelligence (WISC-IV Matrix reasoning). This wave 2 measure provides an estimate of non-verbal intelligence and requires children to look at and complete incomplete visual patterns or matrices by choosing the best option from a range of choices. The K cohort had a mean of 10.35 and SD of 2.98. A cut-off of z = –2 based on the K-cohort distribution was used as an exclusion based on low IQ. Other impairments.The PEDS wave 1 responses were used to assess hearing difficulties (Which medical condition does this child have—loss of hearing—where communication is restricted, or an aid to assist with, or substitute for hearing, is used?); neurological impairment (Which restriction does this child have—long term effects as a result of a head injury, stroke or other brain damage?); epilepsy or autism spectrum disorders (Does the study child have any of these ongoing conditions that do not have to be diagnosed by a doctor: epilepsy or seizure disorder; autism, autism spectrum disorder or Asperger’s?). Language status classification We classified the children into a language-impaired (LI) or typical language (TL) group or excluded them for the analyses based on wave 1 assessment and the WISC-IV Matrix reasoning score. To be included in the languageimpaired group the parent needed to report yes to the PEDS language concerns item: Do you have any concerns about how your child understands what you say to him/her? or the child scored below the 13%ile for this cohort on the PPVT, and the parent report no to all exclusory items (hearing impairment, a head injury, autism and neurological disorder or epilepsy), and the child have a non-verbal IQ above z = 2 for this cohort. Validation of language status was assessed against two additional indices. Parents were asked to report their use of or need for SLP services for their children at each wave. The items were (1) In the past 12 months, have you used speech–language pathology services for

Shaun Goh Kok Yew and Richard O’Kearney your child? (2) In the last 12 months, have you needed speech or language pathology but could not get this service? Parents also completed scales A–D of the Child Communication Checklist 2 (CCC2 Bishop 2003) at wave 2. Each of these scales has seven items addressing difficulties and strengths in articulation, language structure, vocabulary and discourse. Predictors Temperament Wave 1 scores on the reactivity (emotionality) and approach sociability (shy/social inhibition) subscales of the parent reported Short Temperament Scale for Children (Sanson et al. 1994) were used. Both subscales consist of four items, scored 1–6, 1 = Almost never and 6 = Almost always, with summary scores calculated as the average across items. Sanson et al. (1994) reported Cronbach’s alpha > 0.82 for each subscale. Quality of peer interactions Parent’s wave 1 responses on the prosocial behaviour (five items; score 0 (least) to 10 (most pro-social)) and peer problems (five items; score 0 (least) to 10 (most peer problems)) subscales of the SDQ were used. Maternal depression The Kessler-6 is a six-item measure of non-specific psychological distress among adults. Items were scored 1–5, 1 = None of the time to 5 = All of the time, with scores calculated as the average of all six items. It has good sensitivity and specificity (Furukawa et al. 2003) with regards to identifying DSM-IV diagnoses of major depressive disorder and anxiety disorders among Australian adults. Scores at wave 1 were used. Parenting practices Parenting warmth is a six-item subscale from the Child Rearing Questionnaire while the hostile parenting scale is five items taken from the Canadian National longitudinal Survey of Children and Youth (Sanson et al. 2010). For both scales items were scored 1–5, 1 = Never to 5 = Always, with scores calculated as the average of all items. Wave 1 scores were used. Adequate to good internal reliability (alpha = 0.83, 0.68) was calculated for parenting warmth and hostile parenting scales respectively as well as good concurrent validity between these subscales, child psychological adjustment and parent factors.

Early language impairments and developmental pathways of emotional problems across childhood Socioeconomic status (SES) The occupation and education subscale of the Social Economic Indexes for Areas (Blackmore et al. 2006) measured at wave 1 was utilized as a measure of SES. This composite of parental income, parental occupational prestige and parental education was calculated and standardized to a mean of 0 and an SD of 1 (Blackmore et al. 2006). This measure of SES correlates highly with income (r = 0.67), education (r = 0.76) and occupational prestige (r = 0.75) and shows good concurrent validity (r = 0.50–0.83) with other measures of SES such as joblessness or economic hardship (Blackmore et al. 2006). Data analyses Descriptive statistics for all variables were calculated using SPSS Complex Samples add-on, to take into account design effects of stratification and clustering. We first sought to establish the validity of unconditional latent growth curve models (LGC) of emotional difficulties separately for boys and girls across the four waves in Mplus v7.0. Latent growth analysis uses emotional problem data for each child at each wave to estimate the trajectory which best represents how emotional symptoms change on average from waves 1 to 4 for this cohort of children. The estimated latent growth curve is defined by its intercept and shape parameters. To account for the stratification and clustering in the selection of participants in LSAC estimates of standard errors were corrected using the TYPE = COMPLEX command in Mplus. A maximum likelihood robust (MLR) estimator, which yields unbiased parameters and standard errors under conditions of non-normality was utilized. Missing data was accounted for by full information maximum likelihood (FIML), which is robust to data missing at random. The intercept parameter was centred at wave 2 to allow predictive inferences from wave 1 measures of language status and child, family and peer predictors to a later emotional symptom severity outcome. Within each gender, unconditional LGCs that estimated intercept, linear, or quadratic slope parameters were compared on fit indices, with the best-fitting models for boys and girls chosen. We then examined the contribution that language status (LI, TL) had on the curve parameters (intercept; slope factors) through the modelling of conditional LGCs with language status as a covariate (LI = 1; TL = 0) in the estimation of growth parameters. Following this analysis a multivariate regression analyses was carried out in Mplus separately for boys and girls. Language status (LI, TL) was entered simultaneously with child temperament (reactivity, sociability), peer interactions (peer problems, pro-social behaviour), maternal distress, parenting (warmth, hostility) and

363

SES and regressed onto the LGCs parameters (intercept; slope factors). All predictors were measured at wave 1 and were specified as time-invariant and mean centred. These regression analyses tested how well the wave 1 predictors accounted for variability in the intercept (severity at wave 2) and growth of emotional problems in the LSAC children. The regression models also tested the 2 way interactions of each predictor with language status on each LGC parameter. Missing values on predictors were accounted for by person mean imputation although 181 boys and 187 girls had to be dropped due to having missing data not recoverable on at least one predictor. For significant two-way interactions (language status × predictor) a multi-group analysis was utilized to estimate the strength of the association (effect size) between the predictor and LGC parameter separately for LI and TL children. The effect sizes for the predictor estimated from this regression were then presented for the LI and TL groups in figures to clarify the nature of the significant interactions with language status. Results There were 3687 (74.0%) participants with sufficient information to have their language status classified based on parental endorsement of the language measures and exclusionary items. The 1296 (27.0%) children whose language or exclusionary status could not be determined were dropped from subsequent classification. Of the remaining children, 152 had WISC Matrix score > 2 SDs below the mean, 99 children had a hearing impairment, 160 a speech-only impairment, eight a head injury, 21 autism and 92 epilepsy. Of the remaining 3236 (64.9%) children 439 (280 boys) were classified as having a language impairment and 2797 children (1347 boys) were identified as having typical language ability. Accessing speech and language therapy across waves 1–4 was significantly associated with language status with 51.4% of the boys and 37.7% of the girls with LI compared with 11.7% of TL boys and 7.0% of TL girls making use of these services, χ 2 (1,270) = 243.21, p = 0.000 for boys; χ 2 (1,270) = 149.2, p = 0.000 for girls. At wave 2 parents rated LI boys’ language as significantly poorer than typical language boys’ in terms of syntax, F(1,254) = 111.66, p < 0.001, semantics, F(1,254) = 118.97, p < 0.001 and coherence, F(1,254) = 121.59, p < 0.001. LI girls’ language was also rated as significantly poorer than TL girls’ language for syntax, F(1,261) = 38.63, p < 0.01, semantics, F(1,261) = 15.51, p < 0.01 and coherence, F(1, 261) = 42.33, p < 0.01. Compared with the included children the 1296 children with missing information were older (Wald F(1,270) = 7.26, p = 0.01), came from families with lower SES (Wald F(1, 270) = 112.36, p < 0.001), were

364

Shaun Goh Kok Yew and Richard O’Kearney

temperamentally more reactive (Wald F(1,270) = 6.09, p = 0.01) had mothers who reported higher psychological distress (Wald F(1,270) = 12.22, p < 0.01) performed more poorly on the PPVT (Wald F (1,270) = 49.42, p < 0.001) and were rated by their parents as having higher levels of emotional problems at waves 1 (Wald F(1,270) = 5.52, p = 0.02) and 2 (Wald F(1,270) = 3.80, p = 0.05) but they were not different on the other child factors, the parenting factors or emotional problems at wave 3 and 4. Table 1 presents the descriptive information for the demographic, predictor and emotional outcomes for the language-impaired and typical language boys and girls and the test of the difference between LI and TL on these variables. LI boys are rated by their parents as having more reactive temperaments and at each wave were rated more highly on severity of emotional difficulties then TL boys. LI girls are rated by their parents as having more reactive temperaments and more peer problems. At waves 1, 2 and 4 LI girls were rated more highly on severity of emotional difficulties then TL girls. The difference was not significant at wave 3. Mothers of LI boys and girls report a higher level of psychological distress than mothers of TL boys and girls. Unconditional LGCs The estimated fit statistics and the descriptives for the parameters (intercept, slope) of the unconditional latent growth curve model are presented in table 2. For boys the chi-square test of model fit was non-significant, while the approximate fit indices root mean square of approximation (RMSEA) < 0.08, comparative fit index (CFI) > 0.95, standardized root mean square residual (SRMR) < 0.08 were indicative of good fit. The unconditional LGC for girls showed good fit on SRMR and CFI, though the chi-square test was significant and the RMSEA was above the recommended cut-off. It is likely that the chi-square test was oversensitive to rejection of valid models due to the large sample size, while a probability of close fit test indicated that RMSEA was not significantly different (p = 0.058) from 0.05. It is likely that the estimated RMSEA was inflated due to the small degrees of freedom (d.f. = 1). The best estimate of the latent growth curve of emotional symptoms has three significant growth factors (intercept, linear slope and quadratic slope) with the positive quadratic slope suggesting that the overall trajectory of emotional difficulties for boys and girls increases over time and that this increase accelerates as the children become older. The significant correlations between the intercept (level) and the linear slope for both genders indicate that higher emotional difficulties scores at wave 2 are associated with steeper increases over time. The negative correlations between linear and quadratic

slopes for girls indicate a ‘braking’ effect, whereby steeper trajectories of emotional difficulties among girls are associated with slower acceleration in the rate of increase in emotional difficulties closer to adolescence. Conditional LGCs: language status analysis Table 3 presents the estimated beta weights and standard errors for the regression of language status (LI, TL) onto the LGC parameters estimates. Language status does not predict the linear or quadratic slope, but being language impaired at wave 1 does predicts a higher wave 2 intercept of emotional symptoms for both boys and girls. Re-estimation of the models with the intercept centred at waves 3 and 4 indicates that the elevation of emotional difficulties associated with LI was significant across all time points for both boys and girls. Figure 1 plots the estimated latent growth curve from the model conditional on language status with separate curves for LI and TL boys and girls. The plots demonstrate the overall persistent relative elevation in the severity of emotional symptoms in the language-impaired group as well as the similarity in the shapes of growth of emotional difficulties for LI children and TL children. We also evaluated the equivalence in shape of the LI and unconditional models with sensitivity testing comparing the growth curve parameter estimates with a test of configural invariance. These results indicate that two trajectories of emotional difficulties for boys and girls follow equivalent linear and quadratic curves and that modelling language impairment status as a covariate did not artificially constrain the growth trajectories to a similar form. Conditional LGCs: multivariate analysis The results of the multiple regressions of the predictors and their interactions with language status onto the three latent growth curve parameters (intercept, linear and quadratic slope) for boys and girls are presented in table 4. Predictors (language status, child reactivity, child sociability, peers problems, prosocial behaviours, maternal distress, parental warmth, parent hostility and SES) and the two-way interactions terms (language status by each predictor) were entered as a block. For boys, there were significant main effects of temperamental reactivity and sociability, peer problems and prosocial behaviours, maternal distress, parental hostility and SES on the intercept of emotional problems. The direction of these effect were as expected with higher temperamental reactivity, more peer problems, higher maternal distress, higher parental hostility and lower SES associated with higher severity of subsequent emotional problems and higher temperamental sociability associated with lower severity. For boys higher temperamental sociability

56.75 (0.08) 0.10 (0.08) 1.84 (0.03) 1.44 (0.03) 2.67 (0.03) 3.97 (0.03) 1.30 (0.14) 7.39 (0.14) 1.63 (0.02) 2.40 (0.02) 4.05 (0.20) 1.46 (0.04) 1.33 (0.04) 1.37 (0.05) 1.61 (0.05)

56.72 (0.15) −0.86 (0.50) 1.98 (0.06) 1.55 (0.06) 2.95 (0.06) 3.87 (0.07) 1.67 (0.37) 7.07 (0.38) 1.84 (0.05) 2.51 (0.04) 4.05 (0.37) 2.14 (0.13) 1.84 (0.12) 1.80 (0.13) 2.12 (0.12)

TL (N = 1347)

Notes: a Estimate based on 235 LI and 1219 TL boys, 133 LI and 1295 TL girls. b Estimate based on 236 LI and 1221 T boys, 135 LI and 1296 TL girls. c Estimate based on 279 LI and 1344 TL boys. d Estimate based on 270 LI and 1326 TL boys, 151 LI and 1417 TL girls. e Estimate based on 238 LI and 1197 TL boys, 125 LI and 1289 TL girls. f Estimate based on 275 LI and 1335 TL boys, 153 LI and 1442 TL girls.

Demographics Age Socioeconomic status Birth order Number of siblings Child factors Temperament—reactivitya Temperament—sociabilitya Peer problems Prosocial behaviour Parent factors Psychological distressb Parenting hostility Parenting warmth Outcomes Wave 1—emotional difficultiesc Wave 2—emotional difficultiesd Wave 3—emotional difficultiese Wave 4—emotional difficultiesf

LI (N = 280)

Boys

1, 261 1, 260 1, 258 1, 261

1, 257 1, 261 1, 261

1, 257 1, 257 1, 261 1, 261

1, 261 1, 261 1, 261 1, 261

d.f.

28.26 18.11 9.89 14.75

15.20 6.54 0.00

17.47 1.68 0.87 0.57

0.04 3.64 5.05 2.87

Wald F

0.00 0.00 0.00 0.00

0.00 0.01 0.99

0.00 0.20 0.35 0.45

0.85 0.06 0.03 0.09

p-value

2.09 (0.14) 1.91 (0.13) 1.71 (0.15) 2.27 (0.15)

1.78 (0.05) 2.39 (0.03) 4.40 (0.03)

2.79 (0.07) 3.67 (0.10) 1.86 (0.13) 7.81 (0.14)

56.36 (0.21) −1.43 (0.88) 2.04 (0.09) 1.69 (0.09)

LI (N = 159)

1.57 (0.04) 1.54 (0.04) 1.49 (0.05) 1.89 (0.05)

1.61 (0.02) 2.31 (0.01) 4.44 (0.01)

2.54 (0.02) 3.75 (0.03) 1.39 (0.04) 8.06 (0.05)

56.92 (0.08) 0.03(0.11) 1.88 (0.03) 1.44 (0.03)

TL (N = 1450)

Girls

1, 266 1, 264 1, 262 1, 266

1, 263 1, 266 1, 266

1, 264 1, 264 1, 266 1, 266

1, 266 1, 266 1, 266 1, 266

d.f.

Table 1. Means and standard errors of demographic factors, child factors, parent factors and outcomes at each wave by LI status and gender

11.92 6.61 2.06 6.15

9.76 3.89 1.72

11.25 0.55 12.51 3.10

6.37 2.71 3.44 6.90

Wald F

0.00 0.01 0.15 0.01

0.00 0.05 0.19

0.00 0.46 0.00 0.08

0.01 0.10 0.07 0.01

p-value

Early language impairments and developmental pathways of emotional problems across childhood 365

366

Shaun Goh Kok Yew and Richard O’Kearney Table 2. Fit indices and model parameters of unconditional LGCs for boys and girls Boys (n = 1627)

Fit indices χ 2 /d.f. RMSEA (90% CI) CFI SRMR Means (SE) Intercept at W2 Linear slope Quadratic slope Variances (SE) Intercept at W2 Linear slope Quadratic slope Covariances (SE) Linear slope—intercept Linear slope—quadratic slope Quadratic slope—intercept

Girls (n = 1609) 12.77/1∗∗∗ 0.08 (0.048–0.130) 0.989 0.016

0.00/1 0.00 (0.00–0.00) 1.00 0.00 1.42 (0.04)∗∗∗ −0.06 (0.02)∗∗ 0.10 (0.02)∗∗∗

1.52 (0.04)∗∗∗ −0.01 (0.03) 0.10 (0.02)∗∗∗

1.34 (0.09)∗∗∗ 0.25 (0.08)∗∗ 0.06 (0.02)∗

1.39 (0.09)∗∗∗ 0.41 (0.09)∗∗∗ 0.08 (0.03)∗∗

0.13 (0.05)∗∗ −0.06 (0.04) −0.05 (0.03)

0.15 (0.05)∗∗ −0.13 (0.04)∗∗ −0.06 (0.04)

Note: W2 = Wave 2; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.

Table 3. Estimated unstandardized and standardized (Cohen’s d) beta-weights and standard errors for the regression of language status (LI = 1; TL = 0) on latent growth factors for boys and girls Boys (n = 1627) Intercept B (SE)

β (SE)

Intercept (Wave 1 centred) 0.68∗∗∗ 0.57∗∗∗ (0.13) (0.11) Intercept (Wave 2 centred) 0.49∗∗∗ 0.42∗∗∗ (0.11) (0.09) Intercept (Wave 3 centred) 0.34∗∗∗ 0.44∗∗∗ (0.11) (0.09) Intercept (Wave 4 centred) 0.51∗∗∗ 0.33∗∗∗ (0.13) (0.09)

Linear

Girls (n = 1609) Quadratic slope

Intercept

Linear

Quadratic slope

B (SE)

β (SE)

B (SE)

β (SE)

B (SE)

β (SE)

B (SE)

β (SE)

B (SE)

β (SE)

−0.25 (0.14)

−0.30 (0.18)

0.07 (0.04)

0.28 (0.20)

0.52∗∗∗ (0.15)

0.39∗∗∗ (0.12)

−0.21 (0.19)

−0.19 (0.18)

0.06 (0.06)

0.22 (0.23)

−0.12 (0.07)

−0.24 (0.13)

0.07 (0.04)

0.28 (0.20)

0.37∗∗ (0.13)

0.31∗∗ (0.11)

−0.09 (0.09)

−0.14 (0.14)

0.06 (0.06)

0.22 (0.23)

0.01 (0.06)

0.02 (0.13)

0.07 (0.04)

0.28 (0.20)

0.34∗ (0.13)

0.25∗ (0.10)

0.03 (0.08)

0.08 (0.19)

0.06 (0.06)

0.22 (0.23)

0.14 (0.14)

0.17 (0.18)

0.07 (0.04)

0.28 (0.20)

0.43∗∗ (0.15)

0.29∗∗ (0.10)

0.15 (0.18)

0.20 (0.26)

0.06 (0.06)

0.22 (0.23)

Note: ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.

predicted a faster increase in emotional problems overall but a slower acceleration in this increase closer to adolescence while more peer problems predicted an overall slower rate of increase but a faster rate of increase closer to adolescence. There were two unexpected main effects for boys. Higher prosocial behaviour was associated with higher severity of subsequent emotional problems and higher parental warmth was associated with an overall steeper increase in emotional problems. For girls, lower temperamental sociability, more peer problems, higher maternal distress and higher parental hostility had the expected augmenting effects on the subsequent severity of emotional problems while temperamental sociability, peer problems and maternal distress influenced the rate of growth in emotional problems. Girls with higher temperamental sociability showed a

faster increase in emotional problems overall but a slower acceleration in this increase closer to adolescence while girls with more peer problems had an overall slower rate of increase in emotional problems but a faster rate of increase closer to adolescence. Higher maternal distress predicted an overall faster rate of increase in the severity of girls’ emotional symptoms. The multiple regressions identified three significant interaction effects. For boys the interaction of parental hostility and language status predicted the severity of subsequent emotional problems while for girls the interaction of SES with language status predicted the overall rate of growth in emotional problems and the interaction of child sociability with language status predicted the rate of increase in the growth of emotional problems closer to adolescence. The strength (regression

Early language impairments and developmental pathways of emotional problems across childhood

367

SDQ Emotional Difficulties

3

2

1

Age

0 4

6

8 LI Boys

10

TL Boys

SDQ Emotional Difficulties

3

2

1

Age

0 4

6 LI Girls

8 TL Girls

10

Figure 1. Model estimates of trajectories of emotional difficulties with 95% confidence intervals by language status for boys and girls.

coefficients) and direction of the associations for each significant interaction for LI and TL children are displayed in figure 2. For boys with language impairment hostile parenting was strongly associated with a higher severity of emotional symptoms while for boys with typical language hostile parenting practices showed no association with emotional symptom severity. SES was associated with the rate of growth in emotional symptoms for girls with early language impairment but not for girls with typical language. For language-impaired girls lower SES related to steeper growth and higher SES with slower growth. For girls with typical language higher social inhibition was associated with a faster acceleration in the increase of emotional symptoms with age while higher social approach dampened this acceleration. For language-impaired girls temperamental sociability was not a predictor of the growth of emotional symptoms.

Discussion Using data from LSAC we examined for boys and girls separately whether language difficulties at ages 4–5 are associated with the developmental pathway of emotional problems across childhood and into the transition in to adolescence. We also asked whether early language difficulties affect the influence that other child, family and environmental factors have in the development of emotional problems. To our knowledge, this is the first investigation of the direct and indirect associations between early language difficulties and the trajectory of internalizing symptoms across childhood. We found for both boys and girls overall that emotional symptoms exhibited significant growth between ages 4 and 11 with a modest, overall increase in withdrawn, anxious and depressive symptoms and

368

Shaun Goh Kok Yew and Richard O’Kearney A SDQ Emotional Difficulties Intercept

4 3 2 1 Parenting Hostility

0 1

SDQ Emotional Difficulties Linear Slope

B

2 3 LI Boys (B = 0.67, β= 0.35)*** 1

0 3

-2

-1

0

1

2

SES 3

-1 LI Girls (B = -0.19 , β = -0.22)*

C

4 5 TL Boys (B = 0.07, β = 0.04)

TL Girls (B = -0.01, β = -0.01)

SDQ Emotional Difficulties Quadratic Slope

0.4

0.2 Tempermental Sociability

0 1

2

3

4

5

6

-0.2 LI Girls (B = 0.07, β = 0.24)

TL Girls (B = -0.04, β = -0.15)*

Figure 2. Significant two-way interactions between LI status and predictors of emotional difficulties: (A) LI × parenting hostility on intercept for boys; (B) LI × SES on linear slope growth factor for girls; and (C) LI × temperamental sociability on quadratic slope growth factor for girls.

a faster rate of increase with age. This trajectory of internalizing symptoms is consistent with those found in other studies which estimate the pathways of emotional problems across the primary and middle school years into adolescence (Bongers et al. 2003, Leve et al. 2005). The shape and rate of growth in emotional symptoms across childhood for boys and girls with language difficulties was not different from the pattern of growth for children with typical language. Early language difficulties did, however, predict a persistently higher level of emotional symptoms. These findings suggest that on average children with language difficulties when they begin school are no more vulnerable to the effects of peer problems and poor academic performance

on their emotional well-being over the next 6 years as children with typical language abilities despite the higher incidence of these difficulties in children with language impairments (Durkin and Conti-Ramsden 2010). Our data also suggest that on average early language-impaired and typical language children are on the same trajectory of emotional problems as they enter high school. Whether adolescence has distinguishing effects on emotional symptoms for the two groups will become clearer as more waves of data accumulate. The findings from this large, representative cohort study of an early and persistent relative elevation in depressed, anxious and withdraw symptoms in languageimpaired children adds weight to the conclusion from

Early language impairments and developmental pathways of emotional problems across childhood smaller prospective studies that these children are at increased risk for emotional difficulties through childhood and adolescence (Yew and O’Kearney 2013). Although on average children with early language impairment were not showing clinical severity of emotional symptoms their level of symptoms was significantly higher than typical children at each of the four ages. We also found that higher mean levels of symptoms were associated with an overall steeper increase in symptoms from 4 to 11 in boys and girls regardless of their language status. As children with steeper trajectories of internalizing symptoms have higher rates of depression, anxiety disorders and other significant mental health problems in adolescence (Dekker et al. 2007, Feng et al. 2008, Sterba et al. 2007), the early elevation in symptoms put language-impaired children at higher risk of these clinical outcomes as they get older. In addition to the direct effects of language impairment on the development of emotional problems, we also examined its interactive relationship with established or assumed predictors. We found that all the predictors used were significantly associated with the level and/or growth of emotional symptoms for boys or girls and that these associations were generally in the expected directions. The association between higher prosocial behaviour at 4 years and higher level of emotional problem later for boys may reflect a heightened role for interpersonal sensitivity in self-esteem, while the unexpected link between higher maternal warmth at wave 1 and a faster rate of increase in emotional symptoms over childhood for boys challenges the assumption of a protective effect of parental warmth when considering change in symptoms with age. More importantly, early language impairment moderated the influences of only three of the predictors. In other words, most child, parent and environmental factors considered to be important in the development of emotional problems have equivalent influences on their severity and course for boys and girls regardless of the presence or absence of early language impairment. Nevertheless, our findings suggest that in boys early language impairment significantly increases the risk associated with hostile parenting towards a higher severity of emotional problems while for girls, language impairment significantly increases the risk and protective influences of SES to the rate of increase in emotional symptoms over the primary school years. The relationship between temperamental sociability and the increase in the rate of growth as the child gets older was significant for typical language girls but not for the language-impaired group who showed a trend in opposite directions. Social approach provides the usual protection against and social inhibition the expected vulnerability towards increasingly steeper growth in emotional problems with age for girls with typical language but does not operate in the same way for girls

369

with early language impairment for whom social approach may be a vulnerability and social inhibition protective. Possible explanations for this atypical pattern in language-impaired children may include differences in the responses of peers to attempts at social connection or in the relative importance of peer acceptance as a source of self-esteem. Taken together, the findings of a persistent elevation in emotional symptoms and the increasing shape of the trajectories of these problems across childhood suggest that language impairment at school entry identifies one group of young children where early intervention for or prevention of future emotional difficulties is indicated. Given that, in general, the influences on the growth of emotional problems are the same for language-impaired children as for children with typical language empirically supported programmes which enhance resilience, emotional competencies and problem focused coping in typical four year olds (Rapee 2013) could be useful for language-impaired children. The clinical challenge is how best to adapt and deliver these programmes for children whose language and communication skills are compromised. Because this study used an existing cohort database there are limitations that arise from the choice of measures in LSAC. Most importantly, the language status classification was based on parents’ report of language concerns or the child’s score on the Peabody, and the exclusion criteria rather than a formal language assessment. There is a risk that this method of classification may have included a higher proportion of children with less severe and more transient language problems in the language-impaired group than reported in comparable studies of children with language impairment or specific language impairment. We considered cross wave criteria for language impairment status but this would have both increased missing data and reduced our ability to infer that early language difficulties at around school entry influenced later severity and growth of emotional difficulties from the findings. We also acknowledge that the selected group may have included fewer children with more severe language problems because children whose language status could not be determined because of missing information had poorer receptive and expressive abilities. Nevertheless, PEDS language concerns items are sensitive to referral for speech and language services generally (Glascoe 2000) and in our study, and the children classified as language impaired at age 4 in the study scored significantly lower on the CCC2 at age 6–7 years. In addition, we did find elevations in emotional difficulties related to early language impairment and the size of our language-impaired and typical language groups allowed identification of moderating effects of language status on predictors of the development of emotional problems which have not been investigated before. The

SES

Parental hostility

Parental warmth

Parent factors Psychological distress

Sociability

Reactivity

Prosocial behaviour

Peer problems

Main effects Child factors Language status

a

0.48∗∗∗ (0.06) −0.06 (0.09) 0.17∗ (0.08) −0.10∗∗ (0.04)

0.14 (0.09) 0.15∗∗∗ (0.02) 0.06∗∗ (0.02) 0.20∗∗∗ (0.04) −0.17∗∗∗ (0.03)

B (SE)

0.26∗∗∗ (0.03) −0.03 (0.04) 0.09∗ (0.04) −0.09∗∗ (0.03)

0.05 (0.03) 0.19∗∗∗ (0.03) 0.09∗∗ (0.04) 0.16∗∗∗ (0.04) −0.18∗∗∗ (0.03)

β (SE)

Intercept

0.05 (0.06) 0.11∗ (0.05) −0.03 (0.05) −0.02 (0.03)

−0.03 (0.07) −0.07∗∗∗ (0.02) −0.03 (0.02) 0.01 (0.03) 0.12∗∗∗ (0.02)

B (SE)

0.06 (0.07) 0.10∗ (0.05) −0.04 (0.06) −0.04 (0.05)

−0.03 (0.05) −0.22∗∗∗ (0.06) −0.10 (0.06) 0.01 (0.05) 0.29∗∗∗ (0.07)

β (SE)

Linear slope

Boys (n = 1446)

0.00 (0.03) 0.00 (0.04) 0.00 (0.04) 0.01 (0.02)

0.01 (0.04) 0.05∗∗∗ (0.01) 0.00 (0.01) 0.02 (0.02) −0.04∗ (0.02)

B (SE)

−0.01 (0.09) −0.01 (0.07) 0.01 (0.09) 0.04 (0.07)

0.01 (0.07) 0.30∗∗∗ (0.09) −0.02 (0.08) 0.07 (0.08) −0.19∗ (0.10)

β (SE)

Quadratic slope

0.49∗∗∗ (0.08) 0.07 (0.09) 0.30∗∗∗ (0.07) −0.05 (0.03)

0.19 (0.12) 0.16∗∗∗ (0.03) 0.03 (0.02) 0.08 (0.05) −0.22∗∗∗ (0.03)

B (SE)

0.24∗∗∗ (0.04) 0.03 (0.03) 0.14∗∗∗ (0.03) −0.05 (0.03)

0.05 (0.03) 0.19∗∗∗ (0.04) 0.04 (0.03) 0.06 (0.04) −0.23∗∗∗ (0.03)

β (SE)

Intercept

0.16∗∗ (0.05) 0.05 (0.06) −0.03 (0.06) −0.03 (0.02)

−0.04 (0.09) −0.10∗∗∗ (0.02) −0.03 (0.02) −0.02 (0.03) 0.12∗∗∗ (0.02)

B (SE)

0.13∗∗ (0.04) 0.03 (0.04) −0.03 (0.04) −0.04 (0.04)

−0.02 (0.04) −0.20∗∗∗ (0.04) −0.07 (0.04) −0.02 (0.04) 0.21∗∗∗ (0.03)

β (SE)

Linear slope

Girls (n = 1422)

−0.05 (0.04) 0.01 (0.04) 0.06 (0.04) −0.01 (0.02)

0.02 (0.06) 0.03∗ (0.02) 0.01 (0.01) 0.03 (0.02) −0.03∗ (0.01)

B (SE)

Continued

−0.09 (0.07) 0.01 (0.06) 0.10 (0.07) −0.02 (0.05)

0.01 (0.05) 0.15∗ (0.07) 0.06 (0.06) 0.09 (0.06) −0.11 (0.06)

β (SE)

Quadratic slope

Table 4. Estimated beta-weights and standard errors for the contribution of predictors and language status by predictor interactions on latent growth factors for boys and girls

370 Shaun Goh Kok Yew and Richard O’Kearney

0.21 (0.16) 0.36 (0.22) 0.60∗∗ (0.21) −0.13 (0.10)

0.12 (0.07) 0.00 (0.07) −0.10 (0.14) 0.03 (0.09)

B (SE)

−0.05 (0.04) 0.05 (0.03) 0.12∗∗ (0.04) −0.04 (0.03)

0.06 (0.04) 0.00 (0.04) −0.03 (0.04) 0.01 (0.03)

β (SE)

−0.15 (0.13) −0.10 (0.16) −0.03 (0.15) 0.01 (0.08)

−0.06 (0.05) −0.01 (0.04) −0.07 (0.10) 0.08 (0.07)

B (SE)

−0.08 (0.07) −0.03 (0.05) −0.01 (0.07) 0.01 (0.05)

−0.07 (0.06) −0.02 (0.05) −0.05 (0.07) 0.07 (0.06)

β (SE)

Linear slope

0.05 (0.08) 0.02 (0.10) −0.09 (0.09) 0.08 (0.05)

0.01 (0.03) −0.02 (0.03) −0.01 (0.06) −0.05 (0.05)

B (SE)

0.06 (0.09) 0.01 (0.07) −0.09 (0.10) 0.12 (0.07)

0.03 (0.09) −0.05 (0.08) −0.01 (0.09) −0.09 (0.09)

β (SE)

Quadratic slope

0.14 (0.27) −0.14 (0.31) −0.22 (0.25) −0.14 (0.15)

0.00 (0.10) −0.06 (0.07) 0.04 (0.16) −0.02 (0.10)

B (SE)

0.02 (0.04) −0.02 (0.03) −0.03 (0.03) −0.04 (0.04)

0.00 (0.04) −0.03 (0.03) 0.01 (0.04) −0.01 (0.03)

β (SE)

Intercept

0.00 (0.20) −0.24 (0.20) 0.30 (0.18) −0.18∗ (0.09)

−0.07 (0.08) 0.01 (0.06) 0.02 (0.10) −0.12 (0.06)

B (SE)

0.00 (0.05) −0.04 (0.04) 0.07 (0.04) −0.08∗ (0.04)

−0.05 (0.05) 0.01 (0.04) 0.01 (0.04) −0.06 (0.04)

β (SE)

Linear slope

Girls (n = 1422)

Notes: a :Main effects for all predictors except LI status are conditional on mean−centred language status = 0, while main effects for LI status is conditional on all other mean-centred predictors = 0. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p< 0.001.

LS × SES

LS × parental hostility

LS × parental warmth

Parent factors LS × psychological distress

LS × sociability

LS × reactivity

LS × prosocial behaviour

Interactions Child factors LS × peer problems

Intercept

Boys (n = 1446)

Table 4. Continued

0.07 (0.13) 0.09 (0.12) −0.01 (0.11) 0.11 (0.06)

0.06 (0.04) −0.02 (0.04) −0.03 (0.06) 0.11∗ (0.04)

B (SE)

0.04 (0.07) 0.04 (0.05) −0.01 (0.06) 0.10 (0.06)

0.09 (0.06) −0.02 (0.06) −0.02 (0.06) 0.13∗ (0.05)

β (SE)

Quadratic slope

Early language impairments and developmental pathways of emotional problems across childhood 371

372 other major limitation to inferences from the study arises from shared method variance because outcome and predictor indicators come from the same informant. We chose to use the parent’s reports across measures because they had the fewest missing values for the predictors and because multivariate regression within the LGC modelling approach requires good sample sizes and removes cases for any missing value. There was too much missing information from the teacher measures to include both informant scores (parent, teacher) in the regressions and LGC models. There were two other important decisions in the analytic approach that need to be considered when interpreting the current results. First, the study estimates the trajectories of emotional problems and the influences on this trajectory from a variable centred approach which assumes a single, ‘typical’ shape of growth in emotional symptoms from 4 to 11 years and focuses on variability in the parameters of this growth curve. We present these results as a first step in examining the relationship between early language impairment and the development of childhood and adolescence emotional difficulties and recognize that there may be clinically important differences between groups of children in the nature of growth of emotional problems across childhood and that these different patterns of growth may be associated with early language difficulties. For example, the present results suggest that early language problems are more likely to occur in children who begin with high levels of emotional difficulties which continue at these high levels across childhood than in children whose emotional problems start low and become problematic later in childhood or in those children whose high levels of early emotional problems decline with age. We plan to examine such predictions in subsequent studies using person centred approaches. Second, we choose to measure all the predictors at wave 1 therefore disregarding growth or change in some of these measures and how any change over time may be related to the trajectory of emotional difficulties. This allowed clearer and more straightforward interpretations of any moderating influence of language impairment on the associations between the predictors and the severity and growth in emotional symptoms. As more information accumulates about the nature of the relationships between growth in child and parent factors as predictors and growth in emotional problems, future work may need to consider the impact of children’s language impairment as a changing ability on these more complex relationships. Despite these limitations, the results of this study extend knowledge about the association between language impairment and later emotional outcomes by providing new evidence about early language difficulties and the developmental pathway of emotional problems across childhood. The results indicate that when compared

Shaun Goh Kok Yew and Richard O’Kearney with children with typical language both boys and girls with early language difficulties show, on average, persistently higher level of emotional symptoms across childhood. The characteristic rate and shape of growth in emotional symptoms across childhood for boys and girls with language difficulties is not different from the pattern of growth of typical children. We also found that in general the contributions of child, parent and environmental predictors considered to be important in the development of emotional problems across childhood were the same regardless of the presence or absence of early language impairment. The influence of hostile parenting on the severity of emotional symptom development for boys and of SES on the rate of development of emotional symptoms for girls was evident for children with language impairment but not for other children. Language impairment offset the protective effect of higher sociability and the vulnerability effect of higher social avoidance to a faster increase in emotional symptoms with age observed for girls with typical language. Acknowledgements The authors acknowledge the Growing Up in Australia: The Longitudinal Study of Australian Children consortium. LSAC is conducted in a partnership between the Australian Government Department of Social Services, the Australian Institute of Family Studies (AIFS) and the Australian Bureau of Statistics (ABS). Mr Goh Kok Yew was assisted by the Ministry of Health, Singapore.

References ANGOLD, A., COSTELLO, E. J. and WORTHMAN, C. M., 1998, Puberty and depression: the roles of age, pubertal status and pubertal timing. Psychological Medicine, 28, 51–61. BEITCHMAN, J. H., WILSON, B., JOHNSON, C. J., ATKINSON, L., YOUNG, A., ADLAF, E., ESCOBAR, M. and DOUGLAS, L., 2001, Fourteen-year follow-up of speech/language-impaired and control children: psychiatric outcome. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 75–82. BENNER, G. J., NELSON, J. and EPSTEIN, M. H., 2002, Language skills of children with EBD: a literature review. Journal of Emotional and Behavioral Disorders, 10, 43–59. BISHOP, D. V. M., 2003, The Children’s Communication Checklist—2 (London: Psychological Corporation). BLACKMORE, T., GIBBINGS, J. and STRAZDINS, L., 2006, Measuring the socio-economic position of families in HILDA & LSAC. Paper presented at the ACSPRI Social Science Methodology Conference, University of Sydney, Sydney, NSW, Australia. BONGERS, I. L., KOOT, H. M., VAN DER ENDE, J. and VERHULST, F. C., 2003, The normative development of child and adolescent problem behavior. Journal of Abnormal Psychology, 112, 179– 192. DARAGANOVA, G. and SIPTHORP, M., 2011, The Longitudinal Study of Australian Children: LSAC Technical Paper No. 9: Wave 4 Weights (Australian Government, Australian Institute of Family Studies), Melbourne. DEKKER, M. C., FERDINAND, R. F., VAN LANG, N. D. J., BONGERS, I. L., VAN DER ENDE, J. and VERHULST, F. C., 2007, Developmental trajectories of depressive symptoms from early childhood to late adolescence: gender differences and adult

Early language impairments and developmental pathways of emotional problems across childhood outcome. Journal of Child Psychology and Psychiatry, 48, 657– 666. DURKIN, K. and CONTI-RAMSDEN, G., 2010, Young people with specific language impairment: a review of social and emotional functioning in adolescence. Child Language Teaching and Therapy, 26, 107–123. FENG, X., SHAW, D. S. and SILK, J. S., 2008, Developmental trajectories of anxiety symptoms among boys across early and middle childhood. Journal of Abnormal Psychology, 117, 32–47. FUJIKI, M., BRINTON, B. and CLARKE, D., 2002, Emotion regulation in children with specific language impairment. Language Speech and Hearing Services in Schools, 33, 102–111. FURUKAWA, T. A., KESSLER, R. C., SLADE, T. and ANDREWS, G., 2003, The performance of the K6 and K10 screening scales for psychological distress in the Australian National Survey of Mental Health and Well-Being. Psychological Medicine, 33, 357–362. GLASCOE, F. P., 2000, Parents’ Evaluation of Developmental Status: Authorized Australian Version (Parkville, VIC: Australia: Center for Community Child Health). GOODMAN, R., 1997, The strengths and difficulties questionnaire: a research note. Journal of Child Psychology and Psychiatry and Allied Disciplines, 38, 581–586. HALE, W. W., RAAIJMAKERS, Q., MURIS, P., VAN HOOF, A. and MEEUS, W., 2008, Developmental trajectories of adolescent anxiety disorder symptoms: a 5-year prospective community study. Journal of the American Academy of Child and Adolescent Psychiatry, 47, 556–564. HARRISON, L. J., MCLEOD, S., BERTHELSEN, D. and WALKER, S., 2009, Literacy, numeracy, and learning in school-aged children identified as having speech and language impairment in early childhood. International Journal of Speech–Language Pathology, 11, 392–403. Hart, B. and Risley, T. R. (eds), 1995, Meaningful Differences in the Everyday Experience of Young American Children (Paul H. Brookes), Baltimore. HART, K. I., FUJIKI, M., BRINTON, B. and HART, C. H., 2004, The relationship between social behavior and severity of language impairment. Journal of Speech Language and Hearing Research, 47, 647–662. HAWES, D. J. and DADDS, M. R., 2004, Australian data and psychometric properties of the Strengths and Difficulties Questionnaire. Australian and New Zealand Journal of Psychiatry, 38, 644–651. JACOBSON, C. M. and GOULD, M., 2007, The epidemiology and phenomenology of non-suicidal self-injurious behavior among adolescents: a critical review of the literature. Archives of Suicide Research, 11, 129–147. LEVE, L. D., KIM, H. K. and PEARS, K. C., 2005, Childhood temperament and family environment as predictors of internalizing and externalizing trajectories from ages 5 to 17. Journal of Abnormal Child Psychology, 33, 505–520.

373

MERRY, S. N., HETRICK, S. E., COX, G. R., BRUDEVOLD-IVERSEN, T., BIR, J. J. and MCDOWELL, H., 2011, Psychological and educational interventions for preventing depression in children and adolescents. Cochrane Database of Systematic Reviews, 12. RAPEE, R. M., 2013, The preventative effects of a brief, early intervention for preschool-aged children at risk for internalising: follow-up into middle adolescence. Journal of Child Psychology and Psychiatry, 54, 780–788. REDMOND, S. M. and RICE, M. L., 2002, Stability of behavioral ratings of children with SLI. Journal of Speech Language and Hearing Research, 45, 190–201. SANSON, A., HAWKINS, M. T., MISSON, S. and CONSORTIUM, L. R., 2010, The development and validation of Australian indices of child development—Part II: Validity support. Child Indicators Research, 3, 293–312. SANSON, A., SMART, D. F., PRIOR, M., OBERKLAID, F. and PEDLOW, R., 1994, The structure of temperament from age 3 to 7 years—age, sex, and sociodemographic influences. MerrillPalmer Quarterly Journal of Developmental Psychology, 40, 233–252. SNOWLING, M. J., BISHOP, D. V. M., STOTHARD, S. E., CHIPCHASE, B. and KAPLAN, C., 2006, Psychosocial outcomes at 15 years of children with a preschool history of speech–language impairment. Journal of Child Psychology and Psychiatry, 47, 759– 765. SOHR-PRESTON, S. L. and SCARAMELLA, L. V., 2006, Implications of timing of maternal depressive symptoms for early cognitive and language development. Clinical Child and Family Psychology Review, 9, 65–83. SOLOFF, C., LAWRENCE, D. and JOHNSTONE, R., 2005, The Longitudinal Study of Australian Children: An Australian Government Initiative. LSAC Technical Paper No. 1. Sample Design Institute of Family Studies, Melbourne. STERBA, S. K., PRINSTEIN, M. J. and COX, M. J., 2007, Trajectories of internalizing problems across childhood: heterogeneity, external validity, and gender differences. Development and Psychopathology, 19, 345–366. WADMAN, R., BOTTING, N., DURKIN, K. and CONTI-RAMSDEN, G., 2011, Changes in emotional health symptoms in adolescents with specific language impairment. International Journal of Language and Communication Disorders, 46, 641– 656. YEW, S. G. and O’KEARNEY, R., 2013, Emotional and behavioural outcomes later in childhood and adolescence for children with specific language impairments: meta-analyses of controlled prospective studies. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 54, 516–524. ZADEH, Z. Y., IM-BOLTER, N. and COHEN, N. J., 2007, Social cognition and externalizing psychopathology: an investigation of the mediating role of language. Journal of Abnormal Child Psychology, 35, 141–152.

Copyright of International Journal of Language & Communication Disorders is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Early language impairments and developmental pathways of emotional problems across childhood.

Language impairments are associated with an increased likelihood of emotional difficulties later in childhood or adolescence, but little is known abou...
310KB Sizes 0 Downloads 5 Views