Population
Forty-six children, 12 girls and 34 boys aged 4–9 years with developmental language disorders (DLD), 11 with ELD and 35 with MLD, were investigated. These participants were recruited among children referred to our in- and out-patient clinic for severe and persistent language impairments between January 2012 and January 2014. The children underwent a comprehensive diagnostic examination consisting of a review of developmental history and of psychiatric and school records, a neuropsychological examination, and a standardised language test. An ICD-10 diagnosis was established by consensus between a psychologist, a speech therapist, and a senior psychiatrist involved in the evaluation of the child. Participants were diagnosed with DLD if they met the relevant ICD-10 criteria after language, psychological and psychomotor evaluations.
Language evaluation consisted of standardized validated tests in French of expression and comprehension. Inclusion criteria were scores adjusted to two standard deviations below the mean on expressive language subtests for ELD, and scores adjusted to two standard deviations below the mean on both expressive and receptive subtests for MLD.
Psychological evaluation included cognitive and projective assessments. Intellectual functioning was investigated with the appropriate Wechsler Intelligence Scale WISC-IV or the WPPSI-III tests. Inclusion criteria were a significant difference between the “verbal” and “performance” subscale scores (above 1.5 SD) and a Performance Intellectual Quotient (PIQ) over 70. We used projective tests (CAT, scenotest) for the psychopathological assessment.
For the psychomotor evaluation, we used standardised validated tests (NP-MOT, see below) assessing global and fine motor skills and coordination (Batterie d’évaluation des fonctions neuro-psychomotrices de l’enfant, NP-MOT, Vaivre-Douret L, ECPA, Paris, 2006).
Exclusion criteria were children with autism spectrum disorders, intellectual disability, neurological disorders or hearing loss. They were excluded after clinical and paramedical assessments (psychiatric evaluation, electroencephalography, audiometry).
The control group included 23 children, 15 girls and 8 boys who were recruited from the general population during their first months of life. This control group was part of a longitudinal study by the Lausanne research group and was chosen because of the absence of any language impairment. Ethics approval (N°20,110,508) was provided by the Ile de France ethics committee “Comité de Protection des Personnes” CPP-IDF2 de France II and written informed consent was obtained from participating parents and from the children when possible. Concerning the control group, the university of Lausanne ethics committee approved the research protocol.
Language assessment
Different aspects of language were assessed using validated tasks in French from different language batteries (ELO, NEEL, see below) according to the possibilities and the age of the children: Receptive Vocabulary, Expressive Vocabulary, Word Repetition, phonology, Sentence Understanding, and Sentence Completion (assessment of oral language - Evaluation du Langage Oral - ELO, Khomsy, 2001; new tests for language assessment - Nouvelles Epreuves pour l’Examen du Langage - N-EEL, Chevrie-Muller C and Plaza, 2001).
These tests were validated on 900 and 540 French-speaking children respectively, aged from 3 to 11 and from 3.7 to 8.7 years. Results are presented as percentiles or standard deviations from the mean. For most participants, all tasks were administered in a 60-min session.
Since the scoring systems of these different tests differ, we adjusted the scoring system and determined severity levels, as previously described by Demouy et al. [19]. We first considered the means and standard deviations or percentiles for each task. To adjust the scoring systems to the different tests, for each participant we determined the corresponding age for each score, and then calculated the discrepancy between “verbal age” and chronological age. The difference was converted into a severity rating using a 5-point Likert-type scale, 0 standing for the expected level for chronological age, 1 for 1-year delay from the expected level for the chronological age, 2 for a 2-year delay, 3 for a 3-year delay, and 4 for more than 3 years’ delay. The expressive index was obtained by the summing of expressive vocabulary and sentence completion scores, and the receptive index by the summing of the receptive vocabulary and sentence understanding scores (Table 2). These three severity indexes were then used for the correlation analyses.
Attachment story completion task (ASCT)
The ASCT has been specifically developped to assess attachment in children aged 3 to 8 [17]. Findings obtained with the ASCT have been validated in several studies with 3-year-olds, older preschoolers and school age children in several countries, including France [20]. Correlations have been reported with maternal AAIs, children’s self-representations and social competence at school. The ASCT has also been used in clinical group of children, such as children with cleft lip and/or palate in a recent longitudinal study [21].
It is composed of stories where the themes are intended to trigger the children’s attachment system and assess their attachment patterns. To complete the stories initiated, the children are given a set of dolls, each initially introduced as a member of a family (mother, father, children, and grandmother).
Each story beginning is presented by the examiner in a staged manner and the children are then asked to show and say what happens next.
There are 5 stories:
-
Spilt juice: members of the family are together to celebrate the child’s birthday. Suddenly, the child spills some juice. What happens next?
-
Hurt knee: the family goes into the garden. The child wants to climb on the rocks but his mother is worried and tells him that she is anxious that he might fall and hurt his knee. What happens next?
-
Monster in the bedroom: The parents put their child to bed after dinner. The child plays in his room and hears a noise. The child says: “Oh no! There is a monster in my bedroom”. What happens next?
-
Going away: The parents tell their children that they will be away for the week-end, and that they must stay with their grand-mother. What happens next? The examiner then provokes the departure of the parent’s figures if the participant does not. What happens during the parents’ absence and when they return?
-
Reunion: The child wants to play with his dog Toby, with his mother’s agreement. However, Toby is not there. What happens next?
All the stories involve attachment-related issues. Indeed, the conflicts arising at the beginning of each story enable us to investigate how the children relate to parental figures.
Each assessment was filmed and then coded according to the Attachment Story Completion Task Q-sort (ASCT Q-sort) [18, 22].
The ASCT Q-Sort is composed of 65 items that describe the form and the content of the stories. This enables the quality of attachment of each participant to be described according to four categories: security, disorganization (disruption), deactivation (avoidance), and hyper activation (resistance-ambivalence).
- Secure strategies are characterized by the ability to solve different conflicts with the help of parental figures.
-Deactivated attachment strategies tend to avoid conflicts; in the stories, parental characters are neither reassuring nor punitive.
- Hyper-activated strategies tend to focus on negative information, without being able to find a constructive solution.
- Disorganized narratives are characterized by the absence of a coherent strategy. For instance, the child loses control or is completely inhibited during play. The deactivated, hyper-activated and disorganized categories are defined as insecure [18].
The result of the test gives a description of the child’s quality of attachment in a dimensional manner (score for each category). In the development of the scoring system, the scores were normalized (T scores: M = 50, SD =10) on a control group of 187 French-speaking normally developing children [18]. Each child has a score on each of the four attachment style dimensions. Scores are significantly different from the mean when they are below 45 or over 55. However, a global attachment category can be deduced using the dimension where the participant scored the highest, or over 55. The results also enable an analysis of content and narrative characteristics according to 7 different scales: collaboration, parental support, positive narrative, expression of affects, reaction to separation, symbolic distance and poor narrative skills.
Statistical analysis
Statistical analyses were performed on R software version 2.4.
We first investigated whether there was a correlation between the attachment patterns and the language severity index scores by calculating the Spearman correlation coefficients for the 4 attachment scores and the expressive and receptive severity indexes. We checked that the language scores within-groups did not correlate with attachment scores.
We used ANOVA to compare the characteristics of the children in the three groups (ELD, MLD, and control, p < =0.05).
A two by three contingency table with χ2 tests was used to compare attachment categories (secure versus insecure) and groups (MLD, ELD and control). We then performed multiple ANOVA followed by Tukey post hoc comparisons across the 4 attachment categories in the 3 groups, and across the 7 narrative scales for the three groups (p < =0.05).