Children with mixed developmental language disorder have more insecure patterns of attachment

Developmental Language Disorders (DLD) are one of the most frequent causes of consultation in child psychiatry.

Their prevalence was estimated to be 7.56% in a recent survey on 12,398 children aged 4 to 5 in the United Kingdom, making them among the most common disorders in early childhood. They impact both the development and the affective life of children, and therefore are a major challenge for public health [1]. Indeed, adolescents with a preschool history of speech impairment have good psychiatric outcomes if their language delay had been resolved by age 5, whereas they have significant attention and social difficulties in adolescence if they still have language difficulties [2]. Different terminologies have been used to describe language impairment in children, focusing on different aspects of these disorders. Although the term Specific Language Impairment (SLI) has been the most frequently used in the scientific literature so far, terminology has been the subject of recent debates [3], leading to a change in both definition and terminology in the Diagnosis Statistical Manual (DSM 5) [4].

In the International Classification of Diseases (ICD10) as well as in the DSM IV-R, the definition of “Specific Disorder of Language Acquisition” focuses on the specific nature of the disorder, and a distinction is made between expressive (ELD) and mixed expressive-receptive (MLD) types of language impairment. (APA, 1994; WHO, 1992) [5].

In the DSM-5 [4], “Language Disorders” are included in the neurodevelopmental disorders category. The distinction between expressive and mixed types of language impairment has been removed, as has the difference between verbal and nonverbal intellectual skills; in addition, language disorders can be associated with other diagnoses, such as autism spectrum disorders. In both definitions, the diagnosis comes with certain exclusion criteria, such as neurological disorders, hearing impairment, or intellectual disability, and language disorder has a significant impact on the child’s global functioning.

Despite changes in definition and terminology, the clinical questions raised on the subject of children with language difficulties remain the same. How do children with major language difficulties develop their thought processes, and how do they learn and interact with others? Since children’s language develops in interaction with their parents, caregivers and peers, language disorders cannot be studied without considering the processes at play in language development. Geller and Foley [6] therefore underlined the need to incorporate mental health constructs such as the attachment theory into the study of communication disorders, and to work from a relationship-based perspective with children who are language-impaired.

Attachment theory was first developed by John Bowlby [7]. He defined attachment as an enduring emotional bond that an individual form with another person (1977). He developed the concept of working models as generalized expectations and beliefs about oneself, the world and relationships to others, based on the early experiences babies share with their caregivers. He described two main categories of attachment, secure and insecure. The insecure category includes three different subcategories: insecure-avoidant, insecure ambivalent and insecure disorganized [8]. Whereas secure attachment is associated with better emotional and cognitive development, insecure-disorganized attachment can later be associated with externalizing and internalizing symptoms and disrupt different areas of development [9].

What are the interactions between attachment and language development in children? According to Van IJzendoorn and collaborators [10], language development is stimulated in the context of a secure attachment relationship. This was confirmed by Murray and Yingling [11], who demonstrated the additive effects of attachment and home stimulation on language competence, in 24-months-old typically developing children, especially for receptive abilities.

In children with DLD, the links between attachment and language development have not been studied. Although DLD has been demonstrated to be a markedly genetic disorder, genetic effects are complex, and involve strong links between genetic factors and the environment [12]. Onnis [13] suggested new directions for research, in order to study how early verbal and non-verbal attachment practices on the part of caregivers may mediate the expression of human systems of language. The link between language delay and early interactions was studied by Holditch-Davis et al., in prematurely born children [14]. They showed that mothers of language-delayed prematurely born children provided less interactive stimulation than mothers of children with typical language skills, suggesting that their child’s poor comprehension discourages maternal involvement. Negative feedback of this sort could also be present between children with DLD and their mothers, and interfere with the construction of a secure attachment, resulting in the potentialization of genetic factors influencing the development of language. The relation between language development and attachment is not linear, but is rather part of a circular process that takes place in the early interaction between the child and his caregivers.

Both language difficulties and insecure attachment patterns could then contribute to the high prevalence of psychiatric disorders observed later in life [15]. Indeed, Snowling and collaborators [2], in 71 adolescents from 15 to 16 years old with a preschool history of speech impairment, showed that those who still had specific expressive difficulties exhibited significant attention problems, and those with receptive and expressive difficulties had significant social difficulties in adolescence, whereas children whose language delay had been resolved by age 5 and a half had good psychiatric outcomes. The frequent comorbidities between psychiatric symptoms and language impairment point to the need to place the entire spectrum of language disorders in an integrated framework [16], and to adapt therapeutic approaches to the needs of each child early enough to prevent adverse language or social outcomes.

Therefore, the aim of our study was to study the patterns of attachment in children with DLD. Our hypothesis was that insecure patterns of attachment are more frequent in children with language disorders and contribute to their high rates of psychiatric disorders and poor social outcome in later adolescence. In order to study the construction of attachment in children with language disorders, we chose the Attachment Stories Completed Task developed by Bretherton [17]. This test can be used in young children from 3 years of age, because they can complement their narratives with actions, therefore limiting the impact of language on the construction of the stories. We used the Q-sort scoring validated by Milkovitch [18] on a French-speaking control group. This measure provides a dimensional analysis of attachment and enables a quantitative as well as a qualitative approach. We investigated the attachment profiles of 46 children with expressive or mixed expressive-receptive language disorders in comparison with 23 normally developing children. Our hypothesis was that children with language disorders were more likely to exhibit insecure attachment patterns than normally developing children, especially when they had a mixed expressive-receptive language disorder.

To our knowledge, this is the first time that quality of attachment has been assessed in children with language disorders or specific language impairment using the ASCT. Results showed that the attachment style in children with mixed language disorders (MLD) was less secure and more disorganized than in normally developing children.

The children included in this study had severe and persistent language disorders despite speech remediation. Our results show that it is possible to assess their attachment patterns with the ASCT despite their language impairment. The children were able to continue the story initiated by the investigator using dolls, acting and language. Moreover, we showed that the results we obtained on the patterns of attachment were not influenced by the children’s difficulties in expression or understanding, since attachment scores in the four categories were not correlated to the expressive, receptive, and global index severity scores.

Our results show that children with mixed language disorders have significantly lower scores on the secure dimension and higher scores on the disorganized dimension, than children in the control group. This is not the case for children with expressive language disorders. The children in the three groups were able to perceive theme, but the children of the MLD group experienced greater difficulty in expressing their affects and in elaborating coherent stories.

Qualitatively, disorganization was manifested through several aspects: children lost their symbolic distance by acting themselves instead of acting through the dolls, they denied separation by erasing the beginning of the story, they launched into catastrophic never-ending scenarios, with very little cooperation between the different dolls and poor support from the parental figures. This disorganization was clearly revealed by the themes of separation and conflict contained in the ASCT, since the same children developed very restrictive scenarios in their free play (scenotest).

Higher disorganization scores among children of the MLD group could reflect the impact of the child’s poor comprehension on his caregiver’s involvement, which would then interfere with the construction of a secure attachment, as shown in prematurely born children with language delay [13]. Indeed, the understanding of language precedes its expression, and is stimulated in the context of secure attachment in normally developing children [10]. In response to their child’s poor understanding, parents may provide less verbal and non-verbal stimulation, and anticipate their children’s needs. The need for the parents to adapt to their child’s speech difficulties in turn increases the child’s linguistic and affective dependence [23]. This dependence is illustrated in the ASCT task by the fact that children are very dependent on the reactions of the investigator. The experience of separation, which is necessary for language to develop [24], becomes more and more difficult to overcome, and the process of separation more difficult to complete. As an aggravating factor, the difficulty these children have in using language to express their feelings and to build relationships with others interferes with the “linguistic co-construction of internal coherence” [25].

When the understanding of language was not impaired, in the ELD group, the children were as secure as in the control group, suggesting the early, central role of understanding in the co-construction of a secure attachment. However, we must underline that not all the children in the MLD group were disorganized. This suggests that attachment disorganization is not the linear consequence of difficulties in understanding, but rather results from a circular process that takes place in the early interaction between the child and his caregivers.

The disorganized patterns of attachment observed in children with mixed DLD could be related to the high prevalence of psychiatric disorders and to the poor social prognosis described in these children. Indeed, Yew and O’Kearney [26] in their systematic review and meta-analysis reported a high prevalence of psychiatric comorbidities with a marked increase in the severity of diverse emotional, behavioural and ADHD symptoms in DLD children. Adolescents with a history of DLD report levels of peer problems that are 12 times higher than for those without problems, and they are less emotionally engaged in close relationships [27, 28]. Finally, children with mixed DLD have the poorest social prognosis [15, 29]. The relationship between insecure attachments and psychopathology has already been demonstrated [9, 30, 31]. We therefore think that it may be very important to investigate the attachment patterns of children with DLD early on, together with language and cognitive assessments. The need to investigate additional factors has already been underlined by the Catalise consortium (a multinational and multidisciplinary Delphi consensus study of problems of language development) who recently proposed a set of consensual statements aiming to refer and assess children with language disorders [32]. These factors need to be evaluated early on, in order to improve the developmental trajectory of these children, and to decrease the serious negative consequences of their disorder for their educational and social outcomes [33, 34].

We have shown here that the ASCT can be used to investigate the attachment representations of children with language disorders. The initiation of the stories by the investigator helps the children to construct their scenarios, and the use of dolls enables them to unroll their stories even if words or syntax are lacking. The playful dimension of the task removes the stress of the evaluation, both for the child and for his/her parents. The information included in the test can be explained to the parents and can facilitate their understanding of the psychological difficulties encountered by their child, and the need for a psychotherapeutic approach combined with speech remediation when needed. In addition, the Q-score coding system developed by Miljkovitch [18, 22] gives a description of the attachment profile of each child according to a continuum, in a dimensional rather than in a categorical perspective, and gives access to the content of the stories. It is also sensitive to the changes induced by therapeutic approaches [35].

There are several limitations to our study.

A first limitation is related to the fact that we have investigated the children’s patterns of attachment with the ASCT, a test using language, in a group of children with a language impairment. This is the reason why we have checked through the correlation tests that there was no correlation between the scores of language impairment and the results on the category of attachment. ASCT has been specifically developped to assess attachment in children aged 3 to 8, and it has already been used in children with cleft lip and/or palate in a longitudinal study [20]. Other instruments for the evaluation of children’s attachment through parental or professional reports have been developed, such as a questionnaire aimed to measure attachment of three to 6 year old children by observers in kindergarten, but the results obtained were not concordant with the other attachment measures, such as the strange situation for preschool children and the attachment story completion task [36]. Parent report on their child’s attachment profile have been developed only for very young children under 1 year [37]. Therefore, the ASCT seems to be the best way to evaluate attachment representation in our population, despite language impairment.

A second limitation is related to the fact that our sample size is small, impacting the statistical power of our analysis. Indeed we could not make a power analysis to calculate the sample size, because severe language disorders are not frequent. Therefore we were not able to include more participants through our inpatient unit for children with language disorders. However, we obtained statistically significant differences between groups. Moreover, we have already published research papers comparing the characteristics of smaller groups on patients with DLD on multiple tasks [38].

A third limitation is related to the fact that the control group was recruited by another team, in another French-speaking country, with a different gender ratio and a smaller group of children. However, we checked that the two groups did not differ in terms of socio-economic status, and that there was no difference in the distribution of the 4 attachment categories according to gender (ANOVA, p > =0.05). We had the same results when comparing the ASCT scores of the DLD groups to the theoretical mean. Indeed ASCT scores were previously normed and validated on a large sample of typically developing children. Children in the two groups were significantly different according to their VIQ (p = 0.0001) but not to their PIQ (p > 0,05), as a consequence of the language impairment among children with DLD. However, the fact that the control group was younger reduced the differences between the DLD and control groups in their raw intellectual performances. Further to this, a previous study by Miljkovitch [20] showed that attachment profiles were not correlated to IQ, and we have shown here that there was no correlation between the severity of the language impairment and the distribution of the attachment patterns. On the contrary, we found a weak correlation between the severity scores on the expressive and narrative scales symbolic distance, poor narrative skills and appropriate expression of affects in the MLD group. The quantitative results obtained for narrative abilities on these three scales thus need to be interpreted with caution, as is the case with young pre-schoolers. Indeed, for pre-schoolers, Miljkovitch et al. [20] stated that the most important aspect is how the children process the attachment themes presented in the stories and how they respond to themes of distress, suggesting the need to “consider secondarily how these reactions could influence children’s narrative ability”. We think that the same precautions need to be taken with children presenting a language impairment.

Our study is a first attempt to capture the vulnerability of children with DLD towards psychiatric disorders through the perspective of attachment. We found that the use of the ASCT was well suited to the characteristics of children with DLD, especially at younger ages, when therapeutic interventions are thought to be the most efficacious. Our results showed that children with MLD were more insecure than children in the general population, with a larger proportion of disorganized profiles. Since insecure attachment is associated with a higher risk of developing psychiatric disorders, we think that investigating the quality of attachment in children with DLD is useful, in order to adapt therapeutic interventions.

We need to confirm these results on a larger group of children, to see if the attachment profiles we have described are stable over time even if there is language improvement, and to ascertain whether insecure profiles are correlated with psychiatric disorders at later ages. The next step would be to investigate whether the combination of speech remediation and specific psychotherapeutic approaches has an impact on both the attachment patterns of these children, and their social and psychiatric outcomes.