J Nonverbal Behav (2007) 31:169–184 DOI 10.

1007/s10919-007-0030-x

The Nonverbal Decoding Ability of Children Exposed to Family Violence or Maltreatment: Prospective Evidence from a British Cohort
Erica Bowen Æ Stephen Nowicki

Published online: 27 April 2007 Ó Springer Science+Business Media, LLC 2007

Abstract The prospective association between exposure to family violence or maltreatment in the first four years of life and nonverbal decoding ability at age 8.5 years was examined in a British birth cohort. Overall differences were very minor, except it was found that children exposed to maternal victimization were less accurate in decoding low intensity expressions of fear, than were children from nonviolent homes. Children from violent homes were also more likely to identify expressions of anger, fear, and sadness as happiness. Maltreated children showed a global deficit in decoding emotions, but only when presented as high intensity expressions. Whilst these results point to longitudinal associations between these variables, they do not discount the potential impact of third variables. Results are discussed in relation to current theoretical positions, and the need for further research. Keywords Cohort Á Emotion recognition Á Longitudinal Á Maltreatment Á Nonverbal decoding Á Spouse abuse Á ALSPAC

The ability to accurately process, that is recognize (decode) and express (encode) nonverbal cues such as facially expressed emotions is important for normal emotional development (Herba & Phillips, 2004). Emotions are important regulators of children’s intra-and interpersonal behavior (Barrett & Campos, 1987) and the ability to identify and understand emotions is viewed by some as a powerful tool in managing feelings and in dealing with interpersonal conflict (Kopp, 1989). Despite much evidence for the importance of nonverbal abilities for socio-emotional development, little is known about how such abilities develop (Camras et al., 1988; McClure, 2000). Currently two perspectives

E. Bowen (&) Department of Psychology, Coventry University, James Starley Building, Coventry, CV1 5FB, UK e-mail: aa0522@coventry.ac.uk S. Nowicki Emory University, Atlanta, USA

123

1997). Pollack et al.. & MacDonald. First.. 1992. 2000). locus of control and academic achievement (Nowicki & Duke. Camras et al. Both of these explanations account for the observation that emotion recognition skill is gradually refined throughout childhood from early infancy and improves with age (Fogel et al. it is probable that this maturational pattern reflects both underlying neuro-behavioral and socialization mechanisms (McClure. However. 1992). 2000)... Camras et al. 1990) and are more prone to peer rejection (Boyatzis & Satyaprasad. 1986. Deficits in nonverbal decoding ability have also been found to have negative consequences for adults. Halberstadt. 1983). marital adjustment (Sabatelli.170 J Nonverbal Behav (2007) 31:169–184 offer potential explanations.. Current explanations emphasize both neurological and socialization processes.. Camras. 1994. Kessler. Individuals with deficits in nonverbal decoding skills are socially disadvantaged (Hodgins & Belch. & Pape. Such deficits have also been linked to children’s popularity. Finally. For example. It is argued that the observation and imitation of adult interactions enables the child to acquire and develop increasingly sophisticated facial emotion processing skills.. Conger.. McClure. 1996) emphasize the role of social interaction as the method through which emotion decoding skill develops.g. Pollack et al. Klinnert et al. 1977) suggests that decoding skills develop from the need to cope with an emotionally difficult environment. 1984).. 1988.. identified as such on the basis of either child maltreatment or neglect (Camras et al. 1983. Nelson & de Haan. 1986) and satisfaction (Carton. Noller & Feeney. The present study draws upon this second approach and extends the current evidence base by examining the prospective longitudinal association between exposure to either inter-parental violence or maltreatment during the first four years of life and nonverbal decoding abilities of children aged 8. it has been observed that children who are unable to recognize nonverbal cues show low levels of social competence (Philippot & Feldman. These include poor social competence (Firth. Proponents of the neuro-behavioral maturation explanation (e.g. 1985. the first of which emphasizes the role of neurological substrates (Nelson & DeHaan. Secondly. as well as feelings of depression and poor perceived competence (Nowicki & Carton. suggests that a favorable family environment promotes the development of decoding skill.5 years. the positive relatedness hypothesis. 1988. 1986) offers that decoding skills develop because of emotional inhibition within the family. 2000). & Dorcey. 2000). 2000) and can be interpreted within these three hypothesized mechanisms. In addition. (1988) examined the quality of facial expression posing and recognition in a 123 . they do provide evidence of the effect of family environment on the development of these skills (e. 1999. Hodgins and Belch (2000) suggest three related hypotheses for the development of nonverbal skills within a social constructivist perspective. Camras et al. posited by Hodgins and Belch themselves. 1994). 1986). Kuhlenschmidt. & Kenny. the socialization hypothesis (Halberstadt. Edwards. mothers at high risk for physical child abuse have been found to have decoding deficits (Balge & Milner. Manstead.g. 2000) or inter-parental violence (Hodgins & Belch. Buck. 1997). 1997) argue that early developments in facial emotion processing rely on maturational changes in a neurological subsystem specialized for recognition of complex patterns such as faces and selected facial features including expressions. l990. Meadows. 1985. Specifically. Proponents of the alternative social constructivist model of nonverbal processing ability (e. The second emphasizes the contribution of socialization (Camras. the defensogenic origins hypothesis (Cunningham. A small corpus of studies has examined the facial emotion recognition abilities of children raised in adverse family environments. When the findings of studies in which the facial emotion encoding and decoding abilities of maltreated or neglected children and controls are compared. 2000).

It was hypothesized that children exposed to acts of omission (i. This is consistent with the positive relatedness hypothesis (Hodgins & Belch. students from violent homes would be more accurate in decoding anger and fear due to the hypothesized emotional environment within the family that would require their ability to anticipate and cope with impending aggression. fear) and less adept at decoding other emotions. students from violent homes decoded anger. 1986) socialization hypothesis. than controls. the main difference being that well functioning families resolve negative emotional environments non-violently. In addition.25 years to 5. This hypothesis is consistent with a defensogenic explanation through which it is expected that individuals become more accurate at decoding emotions that would lead to negative outcomes (e. Pollack et al.e. In addition. anger leading to abuse).J Nonverbal Behav (2007) 31:169–184 171 cross-sectional study of twenty abused and non-abused mother-child dyads. neglect) would show global deficits in decoding emotions consistent with a positive relatedness explanation. physically abused children were no more accurate in identifying anger than were controls. children who had been physically abused were significantly less accurate in recognizing sadness than controls. More specifically. Children who were physically abused were expected to be more adept at the detection of threat-related information (anger. Contrary to expectations. Hodgins and Belch (2000) compared the nonverbal decoding ability of 58 undergraduate students who retrospectively reported experiencing inter-parental violence during childhood with controls. In addition. Participants consisted of 16 physically neglected. The findings of the study are taken to suggest that students from violent homes may have less exposure to positive emotion and therefore may have less opportunity and need to recognize it. 17 physically abused. this position is untenable given that students from violent homes were no more accurate at this than those from non-violent homes. Initial analyses focused on whether students from violent homes would show greater skill at reading expressions of fear and anger relative to happiness when compared to students from non-violent homes. fear. and sadness as well as students from nonviolent homes. (2000) examined children’s ability to match a facially expressed emotion with an emotional context described in a vignette. Consistent with Halberstadt’s (1983. Hodgins and Belch (2000) suggest that the best explanation of their findings results from combining both defensogenic and positive relatedness positions into one adaptation explanation. abused children had poorer recognition abilities than did non-abused children. Further. 2000). As a 123 . increasing decoding accuracy for negative emotions..5 years old. However. suggesting that exposure to interparental violence may interfere with the ability to decode positive emotions rather than as hypothesized. Consistent with the positive relatedness explanation it was found that neglected and physically abused children made more errors than did non-maltreated children. suggesting that a favorable family environment promotes the development of decoding skill. children who had been neglected were less accurate in identifying anger and in their ability to discriminate between emotions than both the physically abused and control group. students from violent homes had decoding deficits for positive emotions specifically for happiness. Contrary to the defensogenic hypothesis. it was found that experiencing inter-parental violence during childhood was associated with problems in recognizing happiness.g. abused children and their mothers posed less recognizable expressions than did their non-abused counterparts.. In order for a defensogenic position to be upheld. Therefore it may be that people learn to decode the nonverbal cues that they are exposed to and that are useful to them. In contrast. students from non-violent homes are no less likely to be exposed to negative emotions. and 15 non-maltreated children matched for cognitive ability ranging in age from 3.

The focus of the present study is the 6637 children (3300 49.541 mothers enrolled in the study. This study received ethical approval from the Ethical Boards of the Bristol and Weston Health Authority. In total. with expected dates of delivery between 1. 2001). at 5% of children this is lower than the 7. Although the assessment sessions were scheduled for when the children were eight years old.9% for the British general population (ONS.062 live births delivered on or after 20 weeks. The ALSPAC study area has a population of approximately 1 million and includes the city of Bristol (population 500. a mixture of inner city deprivation (7% of Avon children live in poor urban areas). Southmead Health Authority. despite there existing considerable evidence of sex differences in emotion recognition ability (McClure.12. anger. cross sectional designs in which the decoding abilities of groups of participants who are either currently abused or maltreated. leafy suburbs and moderate sized towns (Golding et al. and fear. Pregnant mothers resided within the former county of Avon..5 years (89 months) and 10. Extant research has relied upon small sample.3% females) who completed the Diagnostic Assessment of Non Verbal Accuracy (DANVA) assessment at age 8. The study focused on the primary expressions of happiness.5 years (127 months). 14. no study has explicitly examined such differences in relation to family violence exposure. 2000).5 years (103 months). Individual differences in the ability and willingness to attend scheduled appointments resulted in the age range of children attending this session being between 7. the average age of the children at this assessment was 8. and 13. Participation in the study was entirely voluntary. The present study aimed to fill this gap by examining the nonverbal decoding accuracy of children from a large British birth cohort.91 and 31. Method Participants The Avon Longitudinal Study of Parents and Children (ALSPAC) is a large multi-wave community-based longitudinal study of children and their families. There were 14.92 were recruited to the study. To date.5 years. In accordance with previous research it was expected that children who were maltreated or who were exposed to interparental violence would show deficits in nonverbal decoding ability. in conjunction with considerable local publicity and direct contact with nonenrolled mothers. 3367 50. rural areas (15%). 123 . or who retrospectively report interparental violence during childhood. Enrollment was primarily through midwives. no large-scale prospective study has been conducted to examine the relationship between exposure to adverse family environments and decoding accuracy. The families in the ALSPAC study are characteristic of those in Britain as a whole with a slight under-representation of minority groups. representing an estimated 85%–90% of the eligible population.7% males.000). were reported either to be exposed to inter-parental violence or maltreatment. 69 cases were lost to follow-up.4. 2001). Moreover.971 infants alive after 12 months. and Frenchay Health Authority. Of these. are compared to those of inadequately matched controls. 551 pregnancies ended before 20 weeks gestation. sadness.172 J Nonverbal Behav (2007) 31:169–184 result. who during the first four years of life. students from non-violent homes may be exposed to a wider spectrum of emotions that they then need to be able to cope with. which is achieved through the development of accurate decoding skill.

J Nonverbal Behav (2007) 31:169–184 173 Measures Maternal Victimization At 8. Of interest to the present study were the items ‘Your partner has been physically cruel to you’ and ‘Your partner has been emotionally cruel to you’. & Rust. 1994). the binary response categories at each time point were used. A short form of the measure was employed in which alternative items (starting with item number 1 in the standard form) were used for all sub-tests with the exception of the coding subtest which was administered in its full form (c. Child Maltreatment A cumulative index of child maltreatment was devised based on previous research by Caspi et al. comprehension sub-tests) and the second part assesses visuo-spatial skills (picture completion. half of the presented stimuli were of high intensity and the other half were of low intensity in that each emotion was more or less obviously modeled. 1992) was used to assess cognitive function. Hanna. and harsh discipline across the first four years of life were ascertained. (2002). 33.f. For each emotion. Whilst it is acknowledged that this conceptualization of maternal victimization does not adequately reflect the complex nature of domestic violence. The WISC-III (Weschler. or fear. sad. 21. Intelligence It has been found that emotion recognition skill is associated with general cognitive ability (Moore. and the child had to respond as to whether the person in the photograph is happy.f. sexual abuse. picture arrangement. coding. block design. 2001). The WISC-III consists of two parts. neglect of hygiene). children’s experiences of neglectful parenting (poor interaction. angry. Finch & Childress. Therefore intelligence was used as a covariate in this study. On the basis of these experiences children were categorized as experiencing ‘probable’ maltreatment if they had experienced one of these exposures. For the purpose of this study. Golombok. inconsistent caregivers. Facial Emotion Recognition The ability of children to identify specific emotions expressed facially was assessed at age 8.5 years using the child faces subtest of the DANVA (Nowicki & Duke. A variable that represented ‘any maternal victimization’ was derived from combining all reports of physical and emotional cruelty experienced during the 47month period. A full description of the development of this index is available in the appendix. or afraid. 2002). anger. This consisted of 24 photographs of child faces. with each face showing one of four emotions: happiness.. sadness. these variables have been shown to be valid indicators of victimization (Bowen et al. mothers were sent postal questionnaires to complete within which was a 42–item life events inventory based on those developed by Barnett. 1975). Each photograph was displayed on a computer screen for two seconds. 2005). 123 . arithmetic. In this index. and Parker (1983) and Brown and Harris (1978). the first assesses verbal skills (knowledge. vocabulary. and 47 months. and ‘severe’ maltreatment if they had experienced two or more such exposures (c. similarities. object assembly sub-tests). which is documented to include sexual and financial abuse. Caspi et al. parental cruelty to the child..

The majority of children (4166/64. unskilled. The listening comprehension subtest of the WOLD is divided into two parts.001). The second part was used.7%) had not experienced maternal victimization or maltreatment. IV and V and AF = Partly skilled. housing tenure.6%) were categorized as having experienced some form of parental maltreatment.5 years.5 years. had mothers with educational qualifications of at least O’level (examinations taken at the end of compulsory education at age 16) (v2 (2) = 522.5 were more likely to come from families in which parents worked in a professional/managerial occupation (v2 (2) = 152. Of these. Results Preliminary Analyses Sample bias In order to examine the possibility of bias within the sample. Significant associations were found between the provision of DANVA data and all of the socio-economic characteristics examined. 6441 (97%) cases had data relating to both maternal victimization and maltreatment experiences. p < .001). and lived in houses that were more likely to be owned (v2 (2) = 609. 123 . A further 240 children were categorized as experiencing both forms of family violence. and armed forces). Taken together these data indicate that the sample on which the present study is focused is not wholly representative of the original ALSPAC cohort.174 J Nonverbal Behav (2007) 31:169–184 Language Comprehension Language comprehension was assessed using the listening comprehension scale of the Weschler Objective Language Dimensions (WOLD. Children who provided DANVA data at age 8. A total score was derived which reflects the number of questions (out of 16) answered correctly. 984 (15. Alternate items from the standard test were used except in cases where the item had American cultural loading. In this part children are read a paragraph describing a scene from which they are to make inferences through a series of questions. Prevalence of Maternal Victimization and Maltreatment There were 6637 children (3216/50% male) who provided DANVA data at age 8.001). IIINM and IIIM = nonmanual and manual skilled occupations. 1068 (16. Rust.89. this was not used. those children for whom DANVA data were available (approximately 50% of the original cohort) were compared to the remainder of the original cohort on a variety of socio-economic indicators. p < .61. 1996) administered at age 8.3%) had mothers who reported some form of victimization by a partner. mothers’ highest educational achievement. in which case the next item was used.44. For the purpose of this study this group was not included in the analysis in order to focus more specifically on the impact of each different exposure. These included social class (three groups based on the UK Registrar General’s Classification: I and II = professional and managerial/technical occupations. and parity. p < . As the first assesses receptive vocabulary skills and is therefore very similar to the vocabulary subtest of the WISC-III.

123 0. intelligence. In addition. 4705) = 2. 4607) = 0. Bonferroni corrections were employed to account for multiple testing and reduce type I error.139 À0. The Impact of Early Experiences on Facial Emotion Recognition Accuracy The independent variable in this study was the type of family violence that children had been reported to be exposed to À either maternal victimization (MV).145 À0. 4607) = 2.129 À0.538 À0.269 0. and language comprehension.164 À0. 4607) = 2.63.J Nonverbal Behav (2007) 31:169–184 175 Associations between Confounder and Outcome Variables Bivariate Pearson’s correlations were conducted to determine the association between the potential confounder variables and outcomes. fearful).g. In addition.507 2 0. 4607) = 0.117 0.001) 123 .19.05).045 Note: Numbers above the diagonal represent the Pearsons correlations for boys.246 À0. Significant interactions were explored further through post-hoc simple contrasts in the form of Univariate Analyses of Covariance (ANCOVA) controlling for age.437 0. all happy. Age. These are presented in Table 1 below. low intensity happy faces).083 À0. and language comprehension were statistically controlled.900 À0.112 À0. A similar pattern of multivariate results were identified when considering the impact of maltreatment.085 À0.05).56. Gender was included as an interaction term in each model.162 À0.187 À0.05) was found. those below the diagonal are the correlation coefficients for girls. or maltreatment (MAL). p > .05. p > . angry.116 À0. In each model the composite variable and each related intensity variable were included (e.184 À0.05).511 0.02. p > .092 0.103 À0.053 À0. and therefore their inclusion in multivariate models is justified. Table 1 shows that in all cases the confounder variables were significantly associated with all outcome variables. Each MANCOVA conducted examined accuracy in relation to a specific set of stimuli (all faces. The emotion recognition accuracy (number of errors) of exposed and non-exposed (controls) children were compared through a series of a series of Multivariate Analyses of Covariance (MANCOVA). No significant effect of maltreatment on emotion recognition accuracy in general was found (F(2.060 À0.459 0.172 À0.199 À0.185 À0. or fear (F(2.081 À0. All correlations were significant (p < .46.479 0.889 0. sad. high intensity happy. p > . Multivariate Effects Multivariate analyses revealed no significant effect of maternal victimization on emotion recognition accuracy in general (F(2. p > .05). p > .855 3 4 5 6 7 0. verbal and spatial intelligence.104 À0. happy. sadness (F(2.05).845 0.153 À0.085 À0.162 À0.063 0.144 À0.179 À0. no effect of maltreatment on Table 1 Pearson’s Correlation Coefficients between All Confounding Variables 1 1 Total errors all stimuli 2 Total errors for high intensity stimuli 3 Total errors for low intensity stimuli 4 Age 5 WISC: Verbal IQ 6 WISC: Performance IQ 7 WOLD: Comprehension À 0. anger (F(2.101 À0. no significant effect of maternal victimization on recognition accuracy for happiness (F(2. 4607) = 1.

In both instances however. p > . 95% CI = 1.68. After adjusting for cognitive abilities and age it was found that in contrast to children from non-violent homes. p < .02 – 2.83. the findings of this study indicate that in contrast to children from nonviolent families. or fear (F(2.58. those raised in families in which their mothers were reported to be victimized were less accurate in identifying low intensity expressions of fear. sadness. When presented with expressions of anger. In both sets of analyses.09–1. those children who were maltreated were significantly less accurate in recognizing emotions when presented at high intensity. children within these families were more likely to make specific misattributions when inaccurate. Finally.001). expressions of anger (OR = 1. p < .001). Misattributions The final phase of analysis concerned identifying specific misattributions resulting from exposure to maternal victimization or maltreatment. Maternal Victimization (MV) Group Table 2 shows that after controlling for cognitive factors only a significant univariate effect on the accuracy of identifying low intensity fearful faces remained. In general boys were less accurate in their decoding ability than were girls with global differences found for both high and low intensity stimuli. happy faces presented at low intensity.01). 95% CI = 0. In addition it was specifically found that boys were less accurate in their ability to recognize happy faces. and specifically for happiness (F(2. and fearful expressions (OR = 1.18. sadness (F(2.05).63.14 – 1.71. p < . but only for high intensity emotional stimuli.176 J Nonverbal Behav (2007) 31:169–184 the recognition accuracy of happiness (F(2. There were no significant Group X Gender interactions. p < .39. expressions of sadness. A series of chi-square tests were conducted in the first instance (data not presented). p > .99. and on the basis of these results logistic regression analysis were conducted to determine whether group membership predicted such misattributions after controlling for the previously identified confounding variables. p > . and fear these children 123 . and all angry faces compared to girls. 4705) = 0. 4705) = 1.05). p < . 95% CI: 1. p < .41. p < . In summary. children from violent homes were more likely to misattribute as expressions of happiness.37.10.69. a series of significant gender differences were identified.01). anger (F(2. 4705) = 19. 4705) = 13.49. 4705) = 1.05).85.001) and anger (F(2.05).05) was found. 4705) = 1. The results for each group will be discussed separately.69 – 0. Inspection of group means indicate that children whose mothers reported victimization were significantly less accurate in identifying fearful faces when presented in the low intensity stimuli condition. 4705) = 9. p > . 95% CI: 1. No significant Group X Gender interactions were identified in either set of analyses.05). significant gender differences were found in the recognition of emotions in general (F(2. Maltreated (MAL) Group Table 3 shows that after cognitive factors were controlled. it was found that children from violent homes were less likely than those from non-violent homes to misattribute expressions of anger as expressions of fear (OR = 0. In addition. The adjusted sample descriptive statistics and F ratios for group and gender are presented in Tables 2 and 3. Inspection of group means indicates that in contrast to controls. a global decoding deficit remained significant.38. (OR = 1.

69 ( p < .31 2.20 (.85 (.93) High intensity .74) 17.75*** 0.26 (.72) 17.35 (.50) 1.39 0.84) .72 (.75 0.76) .78) 4.72 (.45) 1.91) .35 (. 4608) Emotion Variable Male M (SD) Female M (SD) Adjusted F(1.95*** 1.08 (.75) All low intensity faces 3.55) 1.74 2.22 (.00 0.52) .16 0. 4608) MV M (SD) Control M(SD) Adjusted 0.34 1.31) .50 .30 (.17 (.20 (1.73) 177 123 .06 (1.30 (.31) Sad All .36) Low intensity 1.08 (1.33 (.91) .74) 1.02 0.16*** 39.40 (.83) 4.33 (.64) .75) 3.92) 1.38 (.54 (1.52 1.76 (.97* 1.65) .91) .70 (.98*** J Nonverbal Behav (2007) 31:169–184 General All faces 4.21 (.73) 2.33 (.53 (1.29 .09 0.60*** 4.75 3.09 (.37) Low intensity 1.39 (1.39 F(1.89) .79) 1.53) .73 (.79 2.91) Low intensity .05 .44) .87) 2.49) .52 (1.59) 28.09 (.33 (.21 (.38 (.83 0.48) High intensity .69) All high intensity faces 1.66) 12.37 1.25 (.45) 14.03 0.30) .36 (.05 0.16*** 1.62 (1.58 (2.32 (.66) .49) 1.62) Low intensity .84 (2.33 (.63) .86) High intensity .50 .39) .24 38.36) 2.86) 1.30 (.78) Fearful All 1.20 .31) .64) High intensity .39) 1.00 0.35 (.82) .68*** 3.76) .51 .98 (1.22 (1.37 (2.38 (.55) Angry All 2.65 (.31) .69 (.15 2.06) 3.99 1.54) .73 (2.32 (.45) Happy All .27 (.41) 1.25 (. 4608) Group Group x Gender interaction Adjusted .00 0.65 (1.61) .01 0.07 1.09 (.21 0.22 (.50 .90 (1.Table 2 Summary of Descriptive Statistics and Adjusted Univariate ANOVA Comparing the Number of Emotion Recognition Errors of Children Whose Mothers were Victimized (MV) Gender F(1.55 (1.22 (1.33*** 37.06 1.05 (1.90) .50 (1.32 .11 0.

17 (.18 (.11 1.32 (.79) 1.09 (.38 (.35 (.36) 17.38 1.21 (.42 .21 (.06) 39. 4706) Emotion Variable Male M (SD) Female M (SD) Adjusted 4.30 (.33 (.05*** 0.43 2.09 23.30) .88 0.31 (2.44) 1.54) Angry All 2.19 Adjusted 1.88) 1.08 (.36 (1.19 (1.28 0.33 (.06 (1.47) High intensity .66 (.98) .72 (. 4706) General All faces 4.69) 16.77 (2.18 (.17 0.28 (.89) .68) All low intensity faces 3.12 0.85 ( p < .64 0.50 (1.37 (2.74*** Group MAL M (SD) Control M(SD) Adjusted 2.52 ( p < .65) High intensity .04*** 1.39) 1.45) 1.89) .73) .06 Low intensity 1.71) All high intensity faces 1.58 (2.25 (.00 3.06 1.45) .73 (.62) Low intensity .70) .89) J Nonverbal Behav (2007) 31:169–184 High intensity .30) .40 (.78) Fearful All 1.90) .15 0.87) .30*** 3.29 (.91 (1.37 (.74*** 2.54 2.53) 2.51) .18 0.03 F(1.66) .31) Low intensity 1.75) 4.38) 1.07) 0.88 1.34) 1.67 3.91) Low intensity .33 (.01 (1.47 4.18 (1.57) .48 (1.76) .95 2.35) .95 (1.86) High intensity .69) 3.34 (.74) 1.38 ( p < .64) .36 (1.85) .13 0.23 0.75) .47) .34) 2.08) 2.70 (.89* 0.36) 3.01 (1. 4706) Table 3 Summary of Descriptive Statistics and Adjusted Univariate ANOVA Comparing the Number of Emotion Recognition Errors of Children Who Were Maltreated (MAL) Group x Gender Interaction F(1.71*** 27.29 (.55 1.04 0.89) 1.36) Happy All .07 (.78 0.28 (.31) 1.50 (1.29 ( p < .82 0.62) .37) 1.91) .55 0.52 (1.12 0.08 (1.31) Sad All .58) 22.00 0.32 (.42 0.54) 1.41 (1.76) 3.62) .30 (.66 (.07) 2.92) .44 0.19 (.34 (.76 (.178 123 Gender F(1.25 (.85 (.72) 1.68) 8.13 (.25 (.94** 4.52) .09 (.52 (1.30 (.67 (.

In particular when children from violent homes incorrectly identified expressions of sadness. The results of this study question whether maternal victimization or maltreatment during the first four years of life has enduring effects on the emotion processing ability of children four years later. In addition. Maternal Victimization For one thing children whose mothers reported being victimized by an intimate partner were significantly less accurate in decoding expressions of fear when presented with low intensity expressions than were children from non-violent homes. and fear. Although not strictly legitimate due to non-significant multivariate effects. The disadvantage for decoding fearful expressions however is consistent with predictions based on the socialization hypothesis. 1986) socialization hypothesis in which nonverbal sensitivity is assumed to arise from a lack of nonverbal information. Children from violent homes were also less likely than children from non-violent homes to identify anger as fear. they were more likely than children from non-violent homes. Children who were reported to be maltreated in contrast. when they made errors they formed consistent patterns of emotion-specific misattributions. were less accurate when decoding high intensity emotions in general. Children within violent families would be expected to show advantages for decoding happiness (which would be comparatively rarely exhibited) and disadvantages for recognizing anger and fear expressions (which would be openly expressed). possibly reflecting the higher incidence of anger and fear cues within their family environment. In contrast to expectations. This might conceivably reflect a climate of fear within the family perpetuated by psychological rather than physical abuse. anger.J Nonverbal Behav (2007) 31:169–184 179 were more likely to identify these emotions as happy. they showed no specific patterns of misattributions when incorrectly identifying happiness. This suggests that children exposed to maternal victimization are very good decoders of negative emotions and the differences between them. Discussion This is the first prospective longitudinal study to examine the relationships between early exposure to aversive family environments and nonverbal decoding abilities. While children from violent homes were just as accurate as children from non-violent homes in decoding facial expressions of anger and sadness. This association is consistent with Halberstadt’s (1983. Perhaps this reflects a ceiling effect because there were fewer errors for decoding happiness than any other emotion. Further empirical investigation of this possibility is warranted. but their misattribution errors were not systematic. However. to identify these emotions as happiness. as the majority of women who reported victimization reported experiencing emotional rather than physical cruelty. It may be that children in violent families are more often exposed to very subtle (low intensity) expressions of fear which are not as behaviorally meaningful as high intensity fear expressions would be. This 123 . inspection of univariate analyses based on more specific predictions did yield significant group effects. no advantage in decoding happiness was found. no disadvantage for recognizing anger was identified. Tentatively the results suggest that having a mother who reports being victimized during this period is associated with enduring patterns of misattributions in cases of inaccuracy.

No significant interactions between adverse family conditions and gender were found. First. after controlling for cognitive factors. thereby implicating early neurological maturation in the development of these skills. It may be that seeing high intensity emotion generates anxiety in maltreated children that in turn interferes with the child’s ability to accurately decode it. A second explanation concerns neurological bases of emotion recognition. 1993). 2000). 2000) boys were less accurate than girls in decoding nonverbal cues. McClure. Children who are maltreated are not exposed to a favorable family environment. that early neuronal maturation initiates sex differences while later socialization perpetuates sex differences in emotion recognition ability (McClure. Cicchetti. McClure (2000) found evidence that girls’ relative advantage in identifying facially expressed emotion is present during infancy. 1991). girls are exposed to more emotional expressions and are socialized to attend to and respond to emotional cues (Fivush. Limitations The findings of the current study should be interpreted within the context of its limitations.g. and as a result develop global decoding deficits. emotional. Typically. and to determine the developmental significance of such misattributions. There are at least three possible explanations for these results. a possible reason why associations between facial nonverbal decoding skill and having a mother who was victimized were not stronger may be due to the way the groups were established. The combined rate of victimization reported in the sample across the first four years 123 . The data are consistent with the positive relatedness hypothesis that views nonverbal decoding skill as a feature of social competence more broadly. boys and girls are socialized differently with regard to being educated about emotional information. Finally. 2004). these variables may inadequately reflect the reality of abuse experienced by women.. This possibility requires further empirical examination. financial.. that can include physical.180 J Nonverbal Behav (2007) 31:169–184 in itself lends support to the notion of a defensogenic origin of negative emotion decoding skill (Cunningham. boys receive less parental guidance and direction with regard to emotion (Fivush. The relative inaccuracy of boys to identify facially expressed emotions was not related to negative socialization experiences during the first four years of life. Gender Differences Consistent with previous research (e. More research is required in order to validate these results. and sexual abuse (Walby & Allen. Maltreated Children The pattern of univariate results obtained for children who were maltreated were different from those of children from violent homes. and persists throughout development. it is possible. 1977). Mothers were identified as being victimized on the basis of their responses to two dichotomous variables ‘my partner was emotionally cruel to me’ and ‘my partner was physically cruel to me’. Consistent with Pollak. Hornung. First. arising from favorable family environments. children who were maltreated showed a global deficit in identifying high intensity emotional expressions. 2005). As has been noted elsewhere (Bowen et al. & Reed (2000). In contrast.

The way that maltreated children were identified also may have prevented stronger associations to be found between maltreated groups. included only data provided by the mothers. 1988. In addition. No longitudinal studies examining the potential mediation of the relationship between family emotional environment and children’s decoding skills have been conducted and more research is required to identify potential protective mechanisms. Peer relationships. research should attempt to quantify children’s perceptions of emotion expression within violent and non-violent families in order to validate the assumption that there are 123 . Other socializing factors. 2002). For example.J Nonverbal Behav (2007) 31:169–184 181 post-partum was 17. and of lower risk than were those samples used in previous studies. To this end. it is likely that the maltreated group identified in the present study was qualitatively different. Pollak et al. groups of maltreated children were identified on the basis of a less objective maltreatment index. despite much evidence that the experience of violence within intimate relationships is not necessarily confined to violence against women (Walby & Allen. In previous studies (e.. Contributing to possible attenuation is that fact it is not known whether the children had actually witnessed the reported victimization of their mothers. provide alternative methods through which nonverbal skills may be refined (Denham. Zoller. below the 21% lifetime prevalence of non-sexual victimization reported in a recent British Crime Survey of domestic violence (Walby & Allen.g. & Couchoud. the present study only identified children who came from families within which their mothers reported being victimized. both positive and negative. had actually experienced some form of victimization but had not reported it.. additional relevant data are being collected within the procedures of ALSPAC and may shed light on this important question. Alternatively. No information about the possible use of violence by mothers against their partners was used in the present study. 2004). thus attenuating any expected group effects. continued exposure to either maternal victimization or maltreatment may also be reflected in these results... As a consequence. Children who were not included in this study due to missing DANVA data were more likely to be characterized by socio-economic indicators that are well known to also be associated with both domestic violence and child maltreatment (e. the results presented should be interpreted with caution pending further empirical investigation including families across a broad spectrum of socio-economic strata. Camras et al.. In the present study. for example. and may have influenced the possible long term impact of adverse family environments upon the nonverbal decoding skills of children. It is possible that a proportion of women who were identified in the non-exposed group. 2000) abused children were drawn from samples of families for which a statutory record of reported abuse existed. 1994).8%. Future Research Findings have highlighted the need for further research examining the processes underlying the association between adverse family environments and nonverbal decoding ability in children including the potential role of third variables within any such associations. A final consideration is the impact of attrition within the study. Holtzworth-Munroe et al. that while based on a published methodology (Caspi et al. A further limitation concerns the possible impact of intervening variables during the four and a half year lapse between the last assessment of maternal and child victimization and decoding skill.. may have played an important role in the nonverbal decoding skills assessed in this study. 2004).g. Therefore it is likely that those not included would be more likely to have been exposed to either maternal victimization or maltreatment. 1997). Therefore.

Questions concerned the frequency with which the mother participated in several activities with the child (e. Each item was responded to using a 3-point Likert-type scale (0 = hardly ever. Acknowledgements We are extremely grateful to all the families who took part in this study.182 J Nonverbal Behav (2007) 31:169–184 systematic differences associated with decoding ability. playing. it would be important to see if higher levels of anxiety are experienced by maltreated and non-maltreated children when presented with high intensity emotions and if it interferes with decoding accuracy. Those mothers who reported bathing their child either ‘once a week’ or ‘hardly ever’ at each time point were identified as having neglected their child’s hygiene. receptionists and nurses. 18. 24.6% of the available sample) mothers were categorized as neglecting their children’s hygiene at any of the time points and as a result were coded as ‘hygiene neglecting’ in the maltreatment index. 30. 2 = often) and scale scores were derived from summing items. excessive changes of caregiver. Neglect of child hygiene was determined from one question at 6. only 21 (0. In keeping with the methodology employed by Caspi et al. those mothers whose parenting scores approximated the lowest ten percent were identified as exhibiting neglectful parenting. Mothers were asked at 18. Within this context. 12 cases.). those children who had experienced 2 or more caregiver changes were integrated into the final index. obtained from questionnaires administered throughout this period.g. The UK Medical Research Council. The quality of mother-child interactions was determined from a series of postal questionnaires administered when the child was 6. the Wellcome Trust. (2002). computer and laboratory technicians. Mothers were asked at 18. and 42 months whether the child’s primary caregiver had recently changed. 38. research scientists. 30. a cumulative neglectful parenting index was developed that indicated how many times across the 5 questionnaires mothers fell into this category.1%) of the sample. From this. and the University of Bristol provide core support for ALSPAC.2%) mothers reported the sexual abuse of a child. at 30 months. Finally. and harsh discipline. and 42 months old. and the whole ALSPAC team. the midwives for their help in recruiting them. etc. Those who were present in this category at least two times (14%) were labelled neglectful. both cross-sectional and longitudinal studies should be undertaken to identify potential protective factors that prevent children who live within violent or maltreating families from developing decoding deficits. This publication is the work of the authors and Dr Erica Bowen will serve as a guarantor for the contents of this paper. and 38 months in which mothers were asked how frequently they bathed their child. which includes interviewers. 123 . neglect of child hygiene. 4 cases. The authors would also like to thank two reviewers for their comments on an earlier draft of the paper. singing. volunteers. Overall. This related to (3. and 42 months whether the child had been sexually abused. sexual abuse. 5 cases. managers. and at 42 months. A cumulative hygiene neglect score was derived from summing the instances in which mothers had been categorized as neglecting the hygiene of their child across the three time points. 24. parental cruelty to child. At each time point. clerical workers. Very few mothers reported that their child had been sexually abuse at any of the time points: at 18 months. 1300 (10.. Appendix Maltreatment Index Evidence of childhood maltreatment during the first four years of life was ascertained using parental reports of mother-child interaction.

L. Lewis. British Journal of Obstetrics and Gynecology. (2005).. Spacarelli. S. & Poulton. This resulted in 1232 cases (10. Maternal facial behavior and the recognition and production of emotional expression by maltreated and nonmaltreated children.. Socialization of affect communication. Zoller.. Journal of Psychosomatic Research. 23.. Maternal facial behavior and the recognition and production of emotional expression by maltreated and nonmaltreated children.... Taylor...040–1.. W. Caspi A. Developmental Psychology. Socialization of preschoolers’ emotion understanding. a dichotomous variable was derived based on those mothers who were categorized as having used harsh discipline at any time versus never. Brown.. S.. 313–320. 304–312. Boyatzis. J. Domestic violence risk during and after pregnancy: Evidence from a British longitudinal study. In the present study. & the ALSPAC study team. C. (1985).. 30. Hill.. (1988). J.. T. 26. A. Martino. Craig. 970 (8. S. (1994)... Camras. how often they smacked their child.. Overall. Personality and the structure of the nonverbal communication of emotion. Cunningham. C.. Journal of Nonverbal Behavior. L. Emotion recognition ability in mothers at high and low risk for child physical abuse. G. T. Waylen. Sachs. Edwards. R.2%) of children were exposed to some form of parental cruelty within the first 4 years of life. K. J. R. In M.014). J. D. Camras. mothers were asked how often they slapped their child. K. and 47 months to report whether they themselves or their partner had been ‘emotionally’ and ‘physically’ cruel to their child. (1984). From this. B. Martin. (1990). S. E. the number of time points (0–3) in which a child was deemed to have experienced harsh discipline (physical punishment reported frequently at each time point). J. Manstead. J. S.. Ribordy.414. New York: Plenum Press. A. W. & Harris. Camras. 21.). 928–936.. & MacDonald. Fifth. S. L. London: Tavistock Press. (1977). Social origins of depression: A study of psychiatric disorder in women. 1289–1298. R. 14. Wolke. C. Saarni (Eds. Life events for obstetric groups. J. 564–584. G.5 years by 1. (1999). R.. E. Role of genotype in the cycle of violence in maltreated children. S. I.. 26. 297. European Journal of Social Psychology. S.21 times (95% CI: 1.. 33.. J. was calculated. Carton. 112. Kessler.8%) being categorized as having experienced harsh discipline during the first 4 years of life. Science. Hill.. three questions asked of mothers during the relevant period related to the use of harsh discipline. Child Abuse and Neglect. A. Journal of Personality. 555–578). (1987). & C. Developmental Psychology. Developmental Psychology. A. Perspectives on emotional development II: A functionalist approach to emotions. J. 91–100. The socialization of emotions (pp. & Parker. mothers were asked how frequently they smacked their child in order to control temper tantrums. Moffitt. C. McClay. & Campos. Barnett. It was found that experiencing some form of maltreatment during the first four years of life increased the likelihood of self-reported participation in some form of antisocial behavior by age 8. 851–854. (2000). Journal of Nonverbal Behavior. 235–238. In J. & Satyaprasad. S. 27. Spacarelli. & Stefani. J. 24. 123 . S. A. J. Hanna.. Children’s facial and gestural decoding and encoding: Relations between skills and with popularity. Osofsky (Ed. 18.. & Milner. Denham. S. 4313 (33%) of the cohort had experienced some form of maltreatment. (1983). E. A... Barrett. 304–312.. E. A maltreatment index was derived from summing the number of maltreatment experiences during this 47-month period. References Balge. and at 42 months. At 18 months. W. & Couchoud. C.J Nonverbal Behav (2007) 31:169–184 183 Mothers were asked at 8. Bowen. V. 45. Martino. In order to derive an index of harsh discipline.). Handbook of infant development (2nd ed. 37–55. At 24 months. (2002). & Paper.. 1083–1089. The relationship between children’s sociometric status and ability to recognize facial expressions of emotion. 141–160). L. D. (1994). Nonverbal decoding skills and relationship well-being in adults. Heron.. Ribordy.. (1978). A. J. Mill. M. p = . pp. Oxford: Wiley.. & Stefani. R. A.

385–393. 122–142. Kopp.. Pembrey. G. children and adolescents. (1986). P. Mental Retardation.... S. J.. B. D. Nowicki.. J. Russell. The Journal of Genetic Psychology. Jr. S. Family socialization of emotion expression and nonverbal communication styles and skills. UK: The Psychological Corporation. (2004). E. & J. & Feeney. Holtzworth-Munroe... Journal of Nonverbal Behavior.(1993). (1989). Journal of Nonverbal Behavior. & Reed. M. (1994). (1997).. Developmental Psychology. S. Memory and affect in development: The Minnesota Symposia on Child Psychology (Vol. 2. Nowicki. M. Annotation: Development of facial expression recognition from childhood to adolescence: Behavioural and neurological perspectives. The association of children’s nonverbal decoding abilities with their popularity. 1185–1199. 45. A. & de Haan.. P. A. 276. & Carton. (2000). Journal of Nonverbal Behavior. N. 85–100. & Capos. Family expressiveness styles and nonverbal communication skills. P. (1996).. 9–35. Social Development. Individual differences in the nonverbal communication of affect: The Diagnostic Analysis of Nonverbal Accuracy Scale.. WOLD Wechsler Objective Language Dimensions Manual. S. ONS (2001). Social process theory of emotion: A dynamic systems approach. 395–412. 51. Stuart. Parenting behaviour and children’s cognitive development. Home Office.. American Journal on Mental Retardation. K. 1. (2001). A. Developmental Psychology. Hove. M. M. Wechsler.. A. 24. Hornung. Jr. M. J. A. Recognizing emotion in faces: Developmental effects of child abuse and neglect. Sandin.. Nowicki. Developmental Psychology. 126. W. C.). Jones. The relation of nonverbal processing ability of faces and voices and children’s feelings of depression and competence. E. Halberstadt. Nwokah. 25. J. Office for National Statistics. 357–363. S.. Butterfield. (1991).. A neurobehavioral approach to the recognition of facial expressions in infancy. Y. London. & Duke.184 J Nonverbal Behav (2007) 31:169–184 Finch. (2001). Messinger. E. R. & Phillips. Philippot. Emde. 427–432. WISC-IIIUK Wechsler Intelligence Scale for Children Third Edition UK Manual. D. Meadows. A. Kuhlenschmidt. Fogel. Journal of Social and Clinical Psychology. & Childress. 29. E. Journal of Genetic Psychology. (1992). & Holt. Home Office Research Study No.. J. J. locus of control. R. London. attachment and nonverbal accuracy in early marriage. Emotional content of parent-child conversations about the past. 325–341. Regulation of distress and negative emotions: A developmental view.. (1997). S. A. P... 679–688. England: Psychology Press. & Dorcey.. 14–26. (1997). 4. (Eds. Fivush. L. UK: The Psychological Corporation. 4.. (1986). C. Pollak. . (1983). Nelson. G. 481–502. Fivush. A comparison of WISC selected subtest short forms with. C. W. Age and social competence in preschoolers’ decoding of facial expression. In J. C. T.). 22. Interparental violence and nonverbal abilities.. Journal of Nonverbal Behavior. M. mentally retarded children. S. Reassessing emotion recognition performance in people with mental retardation: A review. Sidcup. K. Paediatric and Perinatal Epidemiology. 153. Nelson (Ed. E. & Allen. Matusou. P. Dickinson. Hillsdale. & McLaughlin. A. 123 . 36.. and academic achievement. (2004). (1992). A. Personal Relationships. (1990). L. Relationship satisfaction. (1996). A meta-analytic review of sex differences in facial expression processing and their development in infants. 827–836. 343–354. D. The psychology of facial expression (pp. In C. New York: Cambridge University Press. et al. M. A. Dedo. Comparing the social support behaviors of violent and nonviolent husbands during discussions of wife personal problems. S. D. J. G. ALSPAC – The Avon Longitudinal Study of Parents and Children.First results on population for England & Wales. Smultzer. P. pp. 106. Hodgins. Social competence and social perceptivity. R. Golding. British Journal of Social Psychology. Jr. 18. R. Golombok. 176–204). J.. sexual assault and stalking: Findings from the British Crime Survey. B... McClure. Firth. 424–453.. 3–24. 20–21.. 39– 77).. 199–221. 13.. 15. 8. S. B. (2000). J. 4. Psychological Bulletin. & Duke. Conger. Gender and emotion in mother-child conversations about the past. 43–54. N. S. & Feldman. Noller.. D. Domestic violence. R. (1994). 18. Walby. Study methodology. Rust. (1986). Klinnert.. C. Halberstadt. (1975). Social referencing: The infants use of emotional signals from a friendly adult with mother present. S.. & Belch. Journal of Personality and Social Psychology. Herba. Moore. 158. J. H. A. 2001 Census . Journal of Child Psychology and Psychiatry. 26. (1992). Journal of Narrative and Life History. (2000).. Cicchetti.. Jr. 74–87. & Rust.. NJ: Erlbaum. E. S.

Reproduced with permission of the copyright owner. . Further reproduction prohibited without permission.

Sign up to vote on this title
UsefulNot useful