Internalising and Externalising Behaviour of Dutch Preschoolers with Developmental Language Disorder: the role of type of language difficulty, age, gender, informant, and
expressive phonology.
E.M.A. (Lisa) Veldt
Master’s Thesis General Linguistics - Clinical Track, University of Amsterdam
Student number: Supervisor: Second reader: Date:
11177470
Prof. Dr. J.E. Rispens Prof. Dr. J.C. Schaeffer 22 June 2020
External project Company name: Company supervisor:
Royal Dutch Kentalis Dr. A. Scheper
Abstract
Children with DLD have an increased risk of having internalising and externalising behavioural problems, but the literature is unclear on what kind of behavioural problems are present in preschoolers with DLD. The inconsistency of previous research outcomes may be attributable to the children’s type of language difficulty, age, gender and the choice of informant. Previous research has indicated a relation between phonology and behaviour, and many preschoolers with DLD have difficulties with expressive phonology. This thesis therefore examines internalising and externalising behaviour of Dutch preschoolers with DLD and investigates whether the behaviour reported depends on type of language difficulty, age, gender, or informant. The Child Behavior Checklist (CBCL) and Caregiver-Teacher Report Form (C-TRF) were filled in for 449 preschoolers with DLD, mean age 3;6 (years;months). A Percentage Consonants Correct (PCC) was calculated for all children as a reflection of their expressive phonology abilities. Results indicate that preschoolers with DLD exhibited elevated levels of internalising and externalising behaviour and suggest that the behaviour that is reported depends on type of language difficulty, age, gender, and informant (parents or pedagogical therapist). Preschoolers with DLD appear to be most vulnerable to somatic complaints and attention problems. Contrary to expectations, a positive relation between expressive phonology and internalising behaviour was found, indicating that better expressive phonology was related to more internalising problems. The effect was bigger for children with expressive language difficulties than for children with expressive-receptive language difficulties.
Keywords: developmental language disorder, DLD-E, DLD-RE, behaviour, expressive phonology
Table of Contents
1 Introduction ... 5
2 Theoretical Background ... 6
2.1 Developmental Language Disorder ... 6
2.2 The Relation between Language Difficulties and Behavioural Problems ... 7
2.3 Factors Influencing the Behaviour Reported ... 9
3 Current Research ... 13 4 Method ... 15 4.1 Participants ... 15 4.2 Measures ... 16 4.2.1 Behaviour Questionnaires ... 16 4.2.2 Phonological Measure ... 17 4.3 Data Analysis ... 18 5 Results ... 19 5.1 Reported Behaviour ... 20
5.1.1 Comparison with General Population ... 21
5.1.2 Kind of Behaviour Reported ... 22
5.2 Linear models ... 23
5.2.1 Group, Age, and Gender Effects ... 26
5.2.2 Expressive Phonology Effects ... 27
5.3 Relation Between CBCL and C-TRF Ratings ... 28
6 Discussion ... 29
6.1 Behavioural Profile Preschoolers with DLD ... 29
6.2 Factors Relating to Behavioural Problems in Preschoolers with DLD ... 32
6.2.1 Expressive or Expressive-Receptive Language Difficulty, Age, and Gender ... 32
6.2.2 Relation Between Parent and Pedagogical Therapist Behaviour Ratings ... 33
6.3 Relation Between Expressive Phonology and Behaviour ... 34
7 Conclusion ... 37
References ... 39
Appendix A ... 50
Appendix B ... 51
1 Introduction
Developmental language disorder (DLD) is a common disorder with a prevalence of around 6% (Law et al., 2000) or 7% (Tomblin et al., 1997). The disorder is characterised by persistent difficulties in language acquisition compared to typically developing (TD) peers (World Health Organisation [WHO], 2018). An estimated 50-70% of children with language impairment have comorbid emotional and behavioural problems (Redmond & Rice, 1998). Emotional problems are commonly called internalising problems and encompass, for example, withdrawal, anxiety, and depression. Behavioural problems are called externalising problems, and encompass, for example, defiance, aggression, and attention problems (Rescorla, 2005). The umbrella term for internalising and externalising problems is “behavioural problems”. The literature on behavioural problems in children with DLD is incongruent; evidence has been found for exclusively elevated levels of internalising behaviour (Coster et al., 1999; McCabe & Meller, 2004; Redmond & Rice, 1998, 2002; Stanton-Chapman et al., 2007), or externalising behaviour (Tomblin et al., 2000), as well as evidence for both internalising and externalising problems (Keegstra et al., 2010; Maggio et al., 2014; St Clair et al., 2011; van Daal et al., 2007). A possible reason for these sometimes conflicting findings is that the behaviour reported for children with DLD may differ depending on type of language problems, age, gender, or informant (Helland et al., 2018; Tervo, 2007).
DLD is a heterogeneous condition, meaning that DLD encompasses different types of language deficit (Conti-Ramsden, 2008). Because of this heterogony, subtypes of DLD have been identified according to the type of language difficulty the children experience. A commonly used classification of types of DLD in research and clinical practice is the distinction between children with production (expressive language) difficulties with relatively good comprehension (receptive language) (DLD-E) and children who have both expressive and receptive language difficulties (DLD-RE). As with the broader category of DLD, no behavioural profile has yet been constructed for young children with different types of language problems (DLD-E and DLD-RE) and the current research aims to provide a basis for this. In addition to expressive and/or receptive language difficulties, children with DLD often have phonological delays or disorders (Bishop et al, 2017; Bliekendaal, 2019). Phonology has previously been associated with behavioural problems (Fröhlich et al., 2010; Girard & Girolametto, 2013; van Daal et al., 2007), but research is limited. The current research therefore aims to expand the scarce existing literature by examining the relation between expressive
phonology abilities and internalising and externalising behaviour in children with DLD in general, as well as in the DLD-E and DLD-RE subgroups.
The behavioural problems children with DLD exhibit are likely to persist into adolescence (Beitchman et al., 1996; Charman et al., 2015; Lindsay & Dockrell, 2012; Lindsay et al., 2007; Redmond & Rice 2002; Snowling et al. 2006; St Clair et al., 2011) and often have adverse consequences on these children’s future academic achievement (Conti-Ramsden, 2008) and economic wellbeing (Law et al., 2009), as well as on their quality of friendships (Conti-Ramsden & Botting, 2008). Children exhibit different behavioural problems at different ages (Helland et al., 2018), and research into internalising and externalising problems in preschoolers has yielded inconsistent results. As diagnosis of (suspected) DLD generally happens at the age of four (van den Bedem, 2020), research into the behaviour of preschoolers with DLD could be beneficial for both identification and intervention purposes.
In short, the aim of the current research is twofold: first, it will describe the internalising and externalising problems of Dutch pre-schoolers with DLD and examine whether the internalising and externalising behaviour reported for preschoolers with DLD depend on type of language difficulty, age, gender, or informant. Second, the relation between phonological abilities and internalising and externalising problems is examined for children with DLD, as well as for the subgroups of DLD-E and DLD-RE. The paper is constructed as follows. In Section 2, the theoretical background for this thesis will be discussed. First, DLD and the distinction between DLD-E and DLD-RE will be discussed. Second, theoretical account of the relation between language difficulties and behavioural problems are examined. Finally, factors on which the behavioural problems reported for preschoolers with DLD could depend are discussed. The research questions and hypotheses of the current research will be presented in Section 3. The methodology will be described in Section 4, followed by the results of the data analysis in Section 5. A discussion of the results can be found in Section 6. Lastly, the conclusion of this thesis will be presented in Section 7.
2 Theoretical Background 2.1 Developmental Language Disorder
Children can encounter various difficulties in their development, among which difficulties with speech, language, and communication. These problems can have identifiable causes such as physical impairments and limited exposure to the language. When language problems occur without an identifiable cause, children can be diagnosed with developmental language disorder
(DLD) (Bishop et al., 2017). There is considerable variation in the linguistic profiles of children with DLD (Conti-Ramsden, 2008) and this has led to the attempts classifying subtypes within the group of children with DLD. Friedmann and Novogrodsky (2007) proposed that classification of children with DLD can be made on the basis of four language modules: lexicon, syntax, phonology, and pragmatics. A classification as suggested by Friedmann and Novogrodsky is not unusual, as the study of aphasia also assumes different language modules and the possibility of selective impairment (Friedmann & Novogrodsky, 2007). Van Daal et al. (2004, 2007) and van Weerdenburg et al. (2006) likewise identified 4 language factors (syntax, semantics, phonology and speech), although they identified a speech factor instead of a pragmatic factor. The most frequently used classification is however the distinction between children with expressive language (production) difficulties with relatively good receptive language (comprehension) abilities (DLD-E) and children with both expressive and receptive language difficulties (DLD-RE). In a clinical context, the DSM-IV (American Psychiatric Association [APA], 2000), ICD-10 and ICD-11 (WHO, 1990, 2018a) likewise makes this distinction between expressive and receptive language abilities. In academic context, the DLD-E/RE distinction is also widely used in research (Gerrits et al., 2017; Law et al., 2004; Leonard, 2014; Rescorla et al., 2007; Vermeij et al., 2019; Willinger et al., 2003, among others). On these grounds, the DLD-E and DLD-E distinction is also used in the current research.
2.2 The Relation between Language Difficulties and Behavioural Problems
There is a well-documented comorbidity between language difficulties and behavioural problems (Beitchman et al., 1990; Beitchman et al., 1996; Beitchman et al., 2014; Blankenstijn & Scheper, 2003; Fujiki et al., 2002; Yew & O’Kearney, 2013). Estimates for the comorbidity of DLD and behavioural problems are around 50 to 70% (Redmond & Rice, 1998). Although the causal relationship between language difficulties and behavioural problems is not investigated in the current research, a discussion of this topic is relevant as we presume that the two influence each other. Several hypotheses have been proposed to explain the co-occurrence of language difficulties and behavioural problems. A first hypothesis is that language difficulties lead to behavioural problems. Language difficulties could result in an impaired psychosocial mechanism (Redmond & Rice, 1998), emotion regulation difficulties (Fujiki et al., 2002), or limited emotional competence (van den Bedem, 2020), which subsequently result in behavioural problems. That language difficulties lead to behavioural problems, and not the other way around, has been substantiated by longitudinal studies. Petersen et al. (2013) investigated 12.091 children aged four to thirteen from the general population, and found that
internalising and externalising problems were predicted by language ability, even after controlling for demographic characteristics, intellectual abilities and prior behavioural problems. In a sample of 224 children aged four to fourteen from the general population, Bornstein et al. (2013) discovered that language abilities in early childhood affected behaviour in early adolescence. A second hypothesis proposes that behavioural problems may lead to language difficulties. A child’s behavioural problems can negatively influence parent-child interaction (Carson et al., 2007) and consequently harm language development (Safwat & Sheikhany, 2014). This direction of the relation between language difficulties and behavioural problems has been found, albeit together with the opposite direction, suggesting a bidirectional relation (Helland et al., 2018; Petersen et al., 2013). A third proposed explanation of the comorbidity of language difficulties and behavioural problems is that there is an underlying factor causing problems in both areas. Different underlying factors have been proposed, such as neurodevelopmental delay, pervasive developmental disorder, IQ, maternal education, and socioeconomic status (Benasich et al., 1993; Bornstein et al., 2013; Clegg et al., 2015; Rescorla et al., 2007).
Language and behaviour are both multifaceted and complex concepts, and a single cause of the comorbidity of behavioural problems and language difficulties might be non-existent. It is likely that multiple factors influence the relation at any one time with the strength of the influence of those factors varying per individual at different ages throughout life. Some factors appear to be more relevant to language difficulties and behavioural problems than others. Socioeconomic status, for example, has been indicated as a factor influencing both language (Ginsborg, 2006) and behaviour (Hosokawa & Katsura, 2018), but language skills and behavioural problems have been found to be related over and above socioeconomic status (Bornstein et al., 2013, Petersen et al., 2013). With regard to language, these theories do not deliberate on whether certain language areas have a different relation to behavioural problems than others. Expressive and receptive language abilities have long been studied in relation to behavioural problems (see Yew & O’Kearney, 2013, for a meta-analysis of studies pre 2012). Different associations have been found for expressive and respective language abilities (Lindsay & Dockrell, 2000; Snowling et al., 2006; Stevenson et al., 1985; Toppelberg & Shapiro, 2000), although not consistently (Keegstra et al., 2010; Rescorla et al., 2007; Willinger et al., 2003). Little research has focused on the relation between phonology and behavioural problems (but see Girard & Girolametto, 2013; Fröhlich et al., 2010; van Daal et al., 2007), or on the impact of phonological problems on behaviour, alongside expressive or expressive and
receptive language difficulties. The current research therefore examines behavioural profiles of children with DLD with different expressive and receptive language abilities, as well as the relation between behavioural problems and phonology in the different subtypes (DLD-E and DLD-RE).
2.3 Factors Influencing the Behaviour Reported
Research into the behavioural problems of children with DLD have produced inconsistent results with regard to the type and amount of problems they exhibit. In this section, I will examine evidence that the behavioural problems exhibited by children with DLD may differ depending on age, gender, informant and type of language difficulty (Grietens et al., 2004; Helland et al., 2018; Lindsay et al., 2007; Redmond & Rice, 1998; Tervo, 2007; Toppelberg & Shapiro, 2000).
Age-related differences in behavioural problems in children with DLD have been observed in previous research (Helland et al., 2018). These age differences have been found between children and adolescents with DLD (Lindsay & Dockrell, 2012; St Clair et al., 2011), but also for preschoolers. A cross-sectional study of 114 young children with DLD by Maggio et al. (2014) showed differences in parent-reported internalising and externalising problems between preschoolers (two-to-six-year-olds) and school-aged children (six-to-eight-year-olds) with DLD. Internalising and externalising problems were more often reported for preschoolers than school-aged children. Both preschoolers and school-aged children exhibited more internalising than externalising problems, but withdrawal was most often reported for preschoolers while anxious/depressed behaviour was most often reported for school-aged children. There are several reasons for children to exhibit different behaviour at different ages, applying both to TD children and children with DLD. The ability to control emotions improves throughout childhood (Eisenberg et al., 2010). Lower effortful control, which involves behavioural inhibition (Rothbart et al., 1994), has been related to more internalising and externalising behaviour (Eisenberg et al., 2001). So, young children may not yet be able to control their emotions which might explain why Maggio et al. (2014) found more behavioural problems in preschoolers than in school-aged children. The kind of behavioural problems children exhibit can also depend on age. Children who go to school or language intervention groups might exhibit more anxious/depressed behaviour, such as separation anxiety, than younger children who stay at home.
Gender differences with regard to behavioural problems in children with DLD have been identified, although there are mixed results. Some studies report boys having more
behavioural problems overall (Lindsay & Dockrell, 2000), while other studies suggest that girls have more problems overall (Coster, 2001), or more internalising problems (Toppelberg & Shapiro, 2000). Other research found elevated risks for internalising problems for both boys and girls, but elevated risk on conduct problems only for boys (Yew & O’Kearney, 2013). Others detected no difference between boys and girls (Bornstein et al., 2013; Conti-Ramsden & Botting, 2008; Snowling et al., 2006). Gender-related differences in behavioural problems have been found in sample of children referred for behavioural problems (van der Sluis et al., 2017). Internalising behaviour, but not withdrawal, was more often seen in girls, while externalising behaviour was more often displayed by boys. Girls develop social skills earlier than boys (Keenan & Shaw, 1997), which might mitigate withdrawn behaviour. In children with language difficulties, gender-related differences would also be expected. As discussed in Section 2.2., better language skills are generally related to less behavioural problems. As girls generally have further developed language skills than boys (Eriksson et al., 2012), girls might show less behavioural problems than boys. Additionally, language is more strongly related to behavioural problems in boys than in girls (Stowe et al., 1999), likewise implying that boys with DLD might exhibit more behavioural problems than girls with DLD. The scarce research into gender differences in preschoolers with DLD is divided, with some reporting more externalising problems for boys (Henrichs et al., 2013) and others reporting no difference (Keegstra et al., 2010; Maggio et al., 2014). It is possible that sample characteristics obscure gender differences. Maggio et al. (2014) report no difference but the girl/boy ratio in their study was 1:6. Henrichs et al. (2013) do report a difference and they had a 1:1 ratio. High ratios reduce the power to identify possible gender differences (Conti-Ramsden, 2013), and this could be why not all studies report gender differences.
Another factor that might have an effect on the behaviour that is reported for children with DLD is the choice of informant (Lindsay et al., 2007; Redmond & Rice, 1998). Low to modest correlations (below .50) are found among reports by different informants (Achenbach et al., 1987; Rescorla et al., 2012), indicating that parents and teachers rate the same child differently. Low correlations (below .30) between parents and preschool teachers were also found by Vermeij et al. (2019) in a study on behavioural problems and parental stress in three-year-olds with DLD. The low agreement between parents and teachers is likely to be the result of actual differences in the child’s behaviour and of variance due to informant differences. There are different (communicative) demands at home and at school or at an intervention group and, as a result, children exhibit different behaviour in different settings (Lindsay et al., 2007).
On the other hand, parents and teachers (or pedagogical therapists) judge behaviour differently, namely from a parental or pedagogical point of view. For example, parents are more likely to detect emotional behaviour, something that is less easy to detect for teachers or pedagogical therapists that spend less time with the child (Lindsay & Dockrell, 2004). Teachers and pedagogical therapists, however, usually have more knowledge about the development of children and what is considered normal behaviour for their age. Comparative data from more than one informant is considered essential for comprehensive assessment (Achenbach, 2006). Using multiple informants can give valuable information about behaviour in different contexts, and therefore both ratings by parents, with the Child Behavior Checklist (CBCL), and ratings by pedagogical therapists, with the Caregiver-Teacher Report Form (C-TRF) (Achenbach & Rescorla, 2000), are considered in the current research.
Aside from the three non-linguistic factors described above, behaviour in children with DLD may also depend on the type of language difficulty they have (Helland et al., 2018; Tervo, 2007). Expressive and receptive language difficulties are areas that have been widely studied (Yew & O’Kearney, 2013). Various studies have shown that children with DLD-RE profiles display internalising and externalising problems more often than children with DLD-E profiles (Lindsay & Dockrell, 2000; Snowling et al., 2006; Stevenson et al., 1985; Toppelberg & Shapiro, 2000; Yew and O’Kearney, 2013). Both types of DLD involve production problems that can cause, or be the result of (see Section 2.2), behavioural problems. Children with DLD-RE, however, have an additional burden of comprehension difficulties. Intuitively, this would justify a difference in behaviour between children with DLD-E and DLD-RE. Differences have however been discerned less convincingly in research with preschoolers with DLD. Tervo (2007) found more parent-reported withdrawal for children with DLD-RE than for children with DLD-E, but children with DLD-RE were overrepresented (ratio 1:8). Others have detected no differences in parent-reported behaviour for children with DLD-E and DLD-RE (Keegstra et al., 2010; Rescorla et al., 2007; Willinger et al., 2003). Vermeij et al. (2019) found no differences between DLD-E and DLD-RE on parent-reported behaviour problems, but they did find more teacher-reported internalising and externalising problems for children with DLD-RE. It is possible that sample sizes were not large enough to detect differences. Keegstra et al. (2010), Rescorla et al. (2007), and Willinger et al. (2003) included less than 100 preschoolers with DLD. Still, Vermeij et al. (2019) had a larger sample size (n = 185) and also did not detect a difference in parent-reported behaviour. It is therefore plausible that behavioural differences between children with DLD-E and DLD-RE are not yet so pronounced in a home setting at
preschool age. A school or intervention group setting may evoke noticeable differences in behaviour of children with DLD-E and DLD-RE due to social and educational demands that are different in a school or intervention group setting than at home.
Few studies have looked into associations between the behaviour of preschoolers and language areas other than expressive-receptive language ability. One area in particular that has been found to be related to behavioural problems is phonology. Girard and Girolametto (2013) examined TD four-year-olds using the Child Behavior Scale (CBS) and detected a relation between externalising behaviour and later phonological awareness abilities. Concurrent relations in TD preschoolers were investigated by Fröhlich et al. (2010) with the Strength and Difficulties Questionnaire (SDQ). Results showed that preschoolers with poorer phonological awareness had more internalising problems, as well as more hyperactivity and social problems. Van Daal et al. (2007), using the CBCL/1.5-5, found a negative relation between phonological awareness and both Internalising and Externalising scales and all but one subscale (see Section 4.2.1 for an overview of the CBCL subscales). Somatic complaints were not related to any of the language modules used in the research (phonology, semantics, syntax and speech). Semantics, syntax and speech were related only to one or five (of eight) subscales, and either the Internalising or Externalising scales, or neither. Phonological difficulties are more salient and conceivably more hindering than semantic or syntactic difficulties. Therefore, phonological difficulties may be responsible for more miscommunication and subsequent frustration than semantic or syntactic difficulties.
Children with DLD often have poorer expressive phonology abilities compared to TD children. The term expressive phonology refers to the oral or signed expression of phonology. In the current research the term refers to the production of speech sounds. Evidence for poorer expressive phonology in children with DLD has been found in various languages, including Hebrew (Owen et al., 2001), French (Maillart & Parisse, 2006), Spanish-Catalan (Aguilar-Mediavilla et al., 2002), and English and Italian (Bortolini & Leonard, 2000). Expressive phonology difficulties are also present in a large group of Dutch preschoolers with DLD (Bliekendaal, 2019). Different expressive phonology abilities have been reported for Dutch preschoolers with DLD-E and DLD-RE (Bliekendaal, 2019; Bron et al., in press) for a sample comparable to one in the present research. A higher Percentage Consonants Correct (PCC; Shriberg & Kwiatkowski, 1982) was found for children with DLD-RE than for children with DLD-E, indicating less severe speech production problems for children with DLD-RE. Bliekendaal’s (2019) results motivate the distinction between children with E and
DLD-RE when examining phonological abilities. Adequate expressive phonology relies on well-specified and accessible phonological representations (Sutherland & Gillon, 2006). A phonological representation is the storage of phonological information of a word in long-term memory (Stackhouse & Wells, 1997). Aside from inaccurate speech sound production, difficulties with the development and access of underlying phonological representations of wordforms is also reflected in poorer phonological awareness abilities (Bird et al., 1995; Rvachew et al., 2003; Swan & Goswami, 1997). Phonological awareness has been shown to be related to behaviour, both in TD preschoolers (Fröhlich et al., 2010; Girard & Girolametto, 2013), and in preschoolers with DLD (van Daal et al., 2007). In a similar fashion, the current research examines the relation between expressive phonology, as measured with PCC, and internalising and externalising behaviour.
In summary, behavioural problems reported for children with DLD may differ depending on age, gender, informant and type of language difficulty. The current research therefore takes these factors into account in examining and describing the behaviour reported for Dutch preschoolers with DLD. The present research looks at phonological difficulties in particular and examines if expressive phonology is related to behaviour in preschoolers with DLD.
3 Current Research
Children with DLD have an increased risk of having internalising and externalising problems and these problems are likely to persist into adolescence (Beitchman et al., 1996; Charman et al., 2015; Lindsay & Dockrell, 2012; Lindsay et al., 2007; Redmond & Rice 2002; Snowling et al., 2006; St Clair et al., 2011). There are inconsistent findings with regard to the kind of internalising and externalising behaviour displayed by preschoolers with DLD. The current research aims to form a behavioural profile of Dutch preschoolers with DLD in order to expand our knowledge of internalising and externalising behaviour in pre-schoolers with DLD. The possibility that the internalising and externalising behaviour reported for preschoolers with DLD depend on type of language difficulty (DLD-E or DLD-RE), age, gender, or informant is taken into account. The current research examines the relation between behaviour and phonological abilities, as phonology has been implicated as having a strong relation to behaviour, in both TD children (Girard & Girolametto, 2013; Fröhlich et al., 2010) and children with DLD (van Daal et al., 2007). Whereas previous research has concentrated on phonological awareness, the focus is presently on expressive phonology. Little research has been done into this relation in children with DLD or in children with DLD-E and DLD-RE. More severe
expressive phonology problems have however been found in children with DLD-E than in children with DLD-RE (Bliekendaal, 2019; Bron et al., in press). If children with DLD-E have more phonological difficulties, this suggests that behaviour that is related to phonological abilities may be more present in children with DLD-E than children with DLD-RE. Therefore, a second aim of the current research is to expand the scarce existing literature on the relation between phonological abilities and internalising and externalising behaviour in children with DLD, DLD-E and DLD-RE. This thesis will thus aim to answer the following research questions:
1. Are the CBCL and C-TRF scores of preschoolers with DLD comparable to those reported for the general population? And if not, what kind of behaviour is reported for preschoolers with DLD?
2. Does internalising and externalising behaviour of preschoolers with DLD depend on type of language difficulty, age, and gender? And what is the relation between behaviour ratings by parents and pedagogical therapists?
3. Is there a relation between internalising and externalising behaviour and expressive phonology abilities in preschoolers with DLD, and is there a difference between preschoolers with DLD-E and DLD-RE?
With regard to the first research question, I hypothesise elevated levels of (sub)clinical behaviour for our sample of preschoolers with DLD compared to the general population. Previous research has produced mixed results with regard to the kind of behavioural problems in children with DLD, this part of the first research question is therefore exploratory. The variance in behavioural problems for preschoolers reported in previous research may depend on type of language difficulty, age, gender, and informant. The effect of these factors on the behaviour observed in preschoolers with DLD is examined in the current research. Due to the inconsistencies in previous research with regard to the influence of these factors, most of this part of the second research question is exploratory. Based on previous research, however, it is expected that there will be a weak correlation between the ratings by parents and pedagogical therapists. With regard to the third research question, I hypothesise that better expressive phonology abilities (as reflected in higher PCC scores) will be associated with less internalising and externalising behaviour, in line with van Daal et al. (2007). As difference in PCC scores between children with DLD-E and DLD-RE was observed in previous research (Bliekendaal, 2019), I expect to observe a difference in the relation between expressive phonological abilities and behaviour for the children with DLD-E and DLD-RE in our sample. As no previous
research has examined this relation, no hypothesis is formed regarding the nature of this difference. This part of the third research question is thus exploratory.
4 Method 4.1 Participants
The current research uses data collected by the Royal Dutch Kentalis in light of research into prevalence, nature and treatment effects in young children with DLD with intelligibility disorders. Of the 1455 children available, 449 participants were included in the final sample with a mean age of 3;6 (years;months). More boys (n = 322) than girls (n = 127) were included, in line with the distributions previously reported (Conti-Ramsden & Botting, 1999; Tomblin et al., 1997). Participants were included when they had a non-verbal IQ of at least 70 (as measured by the SON-R; Snijders & Snijders-Oomen, 1975), and had available behavioural (Section 4.2.1), linguistic (Section 4.1), and phonological data (Section 4.2.2). Although often excluded from research samples, the current research included children with (suspected) Autism Spectrum Disorder (ASD) and multilingual children. Multilingualism is defined as using another language than Dutch for communication at home. The reason to include children with (suspected) ASD is that ASD is difficult to differentiate from DLD at a young age, as the language difficulties associated with both disorders overlap (Kjelgaard & Tager-Flusberg, 2001; Loucas et al., 2008; Tager-Flusberg & Joseph, 2003). Multilinguals are included because a substantial part (36%) of the children in the original sample (N = 1466) spoke another language besides Dutch. Retaining children with (suspected) ASD (n = 21) and multilingual children (n = 140) in the final sample is on these grounds considered to be a more truthful reflection of the clinical population. Importantly, however, as they were referred to Kentalis, the primary care request for these children was related to DLD and not to ASD or multilingualism. IQ, multilingualism and ASD diagnosis was not available for all participants, descriptive statistics (Section 5) for these variables are based on participants for whom this information was available.
Placement of participants in either the DLD-E or DLD-RE group was done based on expressive-receptive language abilities. Expressive abilities were measured with the Schlichting Test for Language Production (WQ and ZQ scores) (Schlichting & Spelberg 2010a). Receptive abilities were measured with the Schlichting Test for Language Comprehension (TBQ score) (Schlichting & Spelberg, 2010b) and the Peabody Picture Vocabulary Test (WBQ score) (Dunn & Dunn, 2005). Children with DLD were classified as having DLD-E when they scored at least 1 SD below the mean (i.e., <85) on an expressive
measure (WQ or ZQ) and scored average on the receptive measures (TBQ and WBQ). Children were classified as having DLD-RE when they scored 1 SD below the mean on both an expressive (WQ or ZQ) and a receptive measure (TBQ or WBQ). The expressive-receptive language and phonological measurements were carried out by a speech therapist within one to three months after the beginning of treatment at Kentalis. The behaviour checklists were filled in pre-treatment by the child’s parent(s) and/or a pedagogical therapist from the intervention group at Kentalis.
4.2 Measures
4.2.1 Behaviour Questionnaires
The Child Behavior Checklist for Ages 1.5-5 (CBCL/1.5-5) and the Caregiver-Teacher Report Form (C-TRF/1.5-5) are used to assess behavioural problems in children (Rescorla, 2005). The current research uses the Dutch versions of the CBCL/1.5-5 and the C-TRF /1.5-5 (Verhulst & van der Ende, 2000), henceforth referred to as CBCL and C-TRF. The CBCL was filled in by parents (father, mother, or both parents), the C-TRF was filled in by pedagogical therapists from the early language intervention groups at Kentalis. The questionnaires contain 100 items (statements) about the behaviour of the child, for example the statement “Constantly seeks help”. The CBCL and C-TRF contain mostly the same items, although some items are adapted for appropriateness. For example, the CBCL contains questions about eating habits whereas the C-TRF has questions about contact with other children. The CBCL also includes questions on sleep problems, the C-TRF does not. The focus is currently on behaviour falling under the Internalising and Externalising scales (see Table 1), therefore the subscale Sleep Problems was not analysed.
Informants (parents or pedagogical therapists) rate each item on a 3-point Likert scale: not at all (0), a little or sometimes (1), and yes or often (2). The scores on specific clusters of items make up the subscale scores. The subscales are grouped into a broader Internalising (emotional problems) and Externalising (behavioural problems) scales (Table 1) for which a total score is also calculated. The Internalising scale encompasses the Emotionally Reactive, Anxious/Depressed, Somatic Complaints, and Withdrawn subscales. The Externalising scale encompasses Attention Problems and Aggressive Behaviour. The Total Problems score is based on the scores on all subscale scales. Scores on each scale are converted to T-scores ranging from 0 to 100. The cutoff scores for the classification of problems into ‘typical’, ‘subclinical’, and ‘clinical’ for the subscales and Internalising/Externalising/Total Problems scales can be
found in Table 1. The Internalising and Externalising scales and the Total Problems scales have lower cutoff scores as these scales contain more items than the subscales.
Table 1
CBCL and C-TRF (sub)scales with T-score cutoff scores for typical, subclinical and clinical classification
Scale T-score cutoff
Total Problems T = <60 T = 60-63 T = >63 typical subclinical clinical Internalising Problems Externalising Problems
Subscales
Emotionally Reactive Attention Problems
T = <65 T = 65-69 T = >69 typical subclinical clinical Anxious/Depressed Aggressive Behaviour
Somatic Complaints
Withdrawn
4.2.2 Phonological Measure
The phonological measures used in the current research is the Percentage Consonants Correct (PCC; Shriberg & Kwiatkowski, 1982) derived from the Metaphon Screening (Howell & Dean, 1994; translated into Dutch by Leijdekker-Brinkman, 2002). The Metaphon Screening is a diagnostic screening instrument for phonological problems in children aged 3;6 to 5;0. The Dutch version of the Metaphon Screening tests 42 pictures of mono-, di- or multisyllabic words (Leijdekker-Brinkman, 2002). Based on a phonetic or orthographic transcription of the child’s productions in the Metaphon Screening, an indication of the severity of the phonological problems can be calculated. This is done by calculating the percentage of consonants correct (PCC; Shriberg & Kwiatkowski, 1982) of the words said during the Metaphon Screening:
PCC = number of consonants correct
number of correct plus incorrect consonants × 100
A phoneme is scored as correct when it is completely intelligible and has an unquestionable interpretation (Shriberg & Kwiatkowski, 1982). Four degrees of severity are distinguished based on the PCC score: mild (>85% correct), mild-moderate (65-85% correct), moderate-severe (50-65% correct), or moderate-severe (<50% correct). In the current research, PCC scores are taken as a reflection of expressive phonology ability so that higher PCC scores indicate better expressive phonology. Better expressive phonology means higher intelligibility.
4.3 Data Analysis
The first research question this thesis aimed to answer concerned the internalising and externalising behaviour of preschoolers with DLD. To see whether there are elevated levels of behavioural problems in our sample of preschoolers with DLD, our CBCL and C-TRF scores were compared to those presented in previous research. The CBCL mean scores were compared to Tick et al (2007), who examined Dutch preschoolers from the general population (N = 279). C-TRF mean scores of a Dutch general population sample was not available. The C-TRF mean scores from our sample were therefore compared to the omnicultural means presented by Rescorla et al. (2012). In order to identify what kind of problems preschoolers with DLD exhibit, the CBCL and C-TRF ratings of the subscales and Internalising/Externalising/Total Problems scales were examined and often reported (for more than 10% of the sample) subscales were highlighted. The percentages of children scoring certain behavioural problems as reported by parents were then compared to a general population sample (Tick et al., 2007) in order to identify problems that are possibly specifically present in children with DLD. Often reported behavioural problems by pedagogical therapists were not compared to a general population sample as this information was not available.
Regarding the second research question, 4 linear models were created to examine the effect of expressive or expressive-receptive language difficulty (group), age, and gender on the CBCL and C-TRF Internalising and Externalising scores. The function lm from R (version 4.0.0, R Core Team, 2020) was used on the data for fitting linear models of the Internalising and Externalising scale scores (CBCL Internalising, CBCL Externalising, C-TRF Internalising and C-TRF Externalising) as a function of the predictors "age" (continuous), "group" (binary: DLD-E/DLD-RE), and "gender" (binary: boy/girl). PCC score (continuous) was included as a predictor as well in order to answer the third research question. The levels of the binary predictors (group and gender) were coded into two orthogonal contrasts and the continuous predictors (age and PCC) were centred. Interactions were included to investigate possible moderator variables. Additionally, for the second research question, the relation between internalising and externalising ratings by different informants (parents with the CBCL and pedagogical therapists with the C-TRF) was investigated. Only the participants for whom both questionnaires were available were included in this analysis (n = 342). The CBCL and C-TRF scores were non-normally distributed (see Section 5.3). The correlations between the CBCL and C-TRF ratings were therefore tested with Spearman’s correlation test.
For the third research question, an unpaired Wilcoxon test was conducted to compare PCC scores of children with DLD-E and DLD-RE, as the two groups were independent and PCC scores were non-normally distributed (see Section 5.2.2). A difference in PCC scores would corroborate findings by Bliekendaal (2019) and motivate the examination of behavioural differences between the DLD subgroups. The relation between internalising and externalising behaviour and PCC scores for children with DLD, DLD-E, and DLD-RE was analysed with the linear models described above. Of interest for the third research question was the effect of PCC scores as well as the interaction between group and PCC score.
5 Results
Of the total sample of 449 preschoolers with DLD, 169 (38%) had been classified as having DLD-E and 280 (62%) as having DLD-RE, based on their WQ, ZQ, TBQ and WBQ scores. Table 2 presents the descriptive statistics for age, IQ, gender, multilingualism, and ASD diagnosis, for the total sample, as well as for the DLD-E and DLD-RE groups.
Table 2
Descriptive statistics age, IQ (non-verbal), gender, multilingualism, and ASD for total sample (N = 449), DLD-E (n = 169), and DLD-RE (n = 280)
mean SD range based on n = n =
DLD
Age (months) 42 4.6 30-56 n = 449 Gender 322M/127F
IQ 96 12.9 70-146 n = 298 Multilingual 140
ASD 20S/1D
DLD-E
Age (months) 40 4.4 30-54 n = 169 Gender 124M/45F
IQ 102 13.4 71-146 n = 102 Multilingual 18
ASD 6S/1D
DLD-RE
Age (months) 43 4.6 31-56 n = 280 Gender 198M/82F
IQ 92 11.5 70-131 n = 196 Multilingual 122
ASD 14S
Note. M = male; F = female; S = suspected ASD; D = diagnosed with ASD.
Descriptive statistic of the scores on the CBCL and C-TRF for our total sample can be found in Table 3 and Table 4, respectively. Descriptive statistics for the children with DLD-E and DLD-RE can be found in Appendices A and B, respectively.
Table 3
Descriptive statistics CBCL for preschoolers with DLD (n = 372)
mean SD range subclinical (n = …)
clinical (n = …) Emotionally Reactive INT 54 6.79
50-87
32 14
Anxious/Depressed INT 52 4.63 50-87
10 3
Somatic Complaints INT 55 6.53 50-76
27 20
Withdrawn INT 56 7.03
50-84
20 24
Attention Problems EXT 56 7.52 50-80
33 33
Aggressive Behaviour EXT 54 6.45 50-91 10 11 Internalising Problems 50 10.64 29-78 30 39 Externalising Problems 50 10.95 28-89 32 38 Total Problems 50 10.67 28-80 27 36
Note. Subclinical = subscales T = 65-69, Internalising/Externalising/Total Problems T = 60-63; Clinical
= subscales T >69, Internalising/Externalising/Total Problems T >63; INT = subscale falls under
Internalising scale; EXT = subscale falls under Externalising scale.
Table 4
Descriptive statistics C-TRF for preschoolers with DLD (n = 419)
mean SD range subclinical (n = …)
clinical (n = …) Emotionally Reactive INT 53 5.02
50-79
17 5
Anxious/Depressed INT 53 4.83
50-81
13 8
Somatic Complaints INT 52 3.89 50-70
4 1
Withdrawn INT 55 6.11
50-90
23 10
Attention Problems EXT 57 8.07 50-95
31 35
Aggressive Behaviour EXT 53 4.67
50-94 7 2 Internalising Problems 49 8.86 34-76 25 18 Externalising Problems 51 8.29 36-89 44 21 Total Problems 51 7.41 29-75 41 17
Note. Subclinical = subscales T = 65-69, Internalising/Externalising/Total problems T = 60-63; Clinical
= subscales T >69, Internalising/Externalising/Total Problems T >63; INT = subscale falls under
Internalising scale; EXT = subscale falls under Externalising scale.
5.1 Reported Behaviour
The first research question concerned the CBCL and C-TRF scores of Dutch preschoolers with DLD as compared to those reported for general population samples, and the kind of behavioural problems reported for our sample. The mean scores of the total sample on all the CBCL and the C-TRF subscales, and the Internalising, Externalising and Total Problems scales were in the
typical range (<65/60) (Table 3 and 4). This means that, on average, the preschoolers with DLD in our sample do not exhibit behavioural problems. More precisely, however, no problems were reported for 228 (61%) children with the CBCL and for 292 (70%) with the C-TRF. There is nonetheless a substantial portion of children (30%-39%) for whom behavioural problems at subclinical level or above (i.e., scoring more than 65 points on the subscales or more than 60 points on the Internalising, Externalising and Total Problems scales) were reported with the CBCL and/or with the C-TRF.
5.1.1 Comparison with General Population
To determine if our sample of preschoolers show elevated levels of behavioural problems compared to the general population, the CBCL and C-TRF mean scores were calculated for comparison to the mean scores found for general population samples. The raw scores of our sample are listed as raw scores were used in previous research. Table 5 shows the CBCL means from Tick et al. (2007) for a general population sample of Dutch preschoolers, the C-TRF omnicultural means based on general population samples from 15 societies, and the means from the current sample of Dutch preschoolers with DLD. Compared to the general population, our preschoolers with DLD have elevated parent-reported behaviour overall, with the exception of aggressive behaviour and externalising behaviour. Compared to the omnicultural means, our preschoolers with DLD score higher on therapist-reported Withdrawn and Attention Problems behaviour.
Table 5
CBCL means from Tick et al. (2007), C-TRF omnicultural means from Rescorla et al. (2012), and CBCL and C-TRF means from current research.
CBCL C-TRF Tick et al. (2007) mean Current research mean Rescorla et al. (2012) omnicultural mean Current research mean
Emotionally Reactive INT 2.15 2.5 1.6 1.07
Anxious/Depressed INT 1.59 1.81 2.4 1.35
Somatic Complaints INT 1.86 2.23 0.8 0.28
Withdrawn INT 1.19 2.13 2.5 3.02
Attention Problems EXT 2.41 3.13 3.5 4.96
Aggressive Behaviour EXT 10.84 9.72 6.9 4.61
Internalising Problems 6.8 8.67 7.3 5.72
Externalising Problems 13.25 12.85 10.3 9.58
Total Problems 30.52 33.35 24.1 21.41
Note. INT = subscale falls under Internalising scale; EXT = subscale falls under Externalising scale.
5.1.2 Kind of Behaviour Reported
Scores on the CBCL and C-TRF subscales and Internalising/Externalising/Total Problems scales were examined to identify what kind of behavioural problems are often reported for our sample of preschoolers with DLD. The most frequently occurring subscales (that is, with >10% incidence in the sample) by parents were the Emotionally Reactive, Somatic Complaints, Withdrawn and Attention Problems subscales (see Table 6). Internalising and Externalising Problems are reported for equally often. The percentage of children from a general population sample (Tick et al., 2007) scoring certain behavioural problems are listed in Table 6 as well, in order to identify problems that are possibly specific to DLD. Therapists most often reported Attention Problems subscales, the second most (but less than 10%) mentioned was Withdrawn subscale. Externalising problems were reported for more children than internalising problems were.
Table 6
Percentage children scoring in CBCL/C-TRF subclinical and clinical ranges from Tick et al. (2007) and the current research.
CBCL C-TRF
% Tick et al. (2007)
% current research % current research
Emotionally Reactive INT 12.5 12.4 5.3
Anxious/Depressed INT 11.6 3.5 5.0
Somatic Complaints INT 10.4 12.6 1.2
Withdrawn INT 14.9 11.8 7.9
Attention Problems EXT 12.9 17.7 15.8
Aggressive Behaviour EXT 7.7 5.6 2.1
Internalising Problems 15.3 18.5 10.3
Externalising Problems 18 18.8 15.5
Total Problems 14 16.9 13.8
Note. INT = subscale falls under Internalising scale; EXT = subscale falls under Externalising scale.
5.2 Linear models
Linear models with the CBCL and C-TRF Internalising and Externalising scales as outcome variables were run in order to answer the second and third research question (see Section 3). The outcomes of the linear models can be found in Table 7 (CBCL Internalising and Externalising scales) and Table 8 (C-TRF Internalising and Externalising scales).
Table 7
Outcomes linear model CBCL Internalising and Externalising scales
CBCL Internalising scale CBCL Externalising scale
Estimate Std. Error t-value p-value Estimate Std. Error t-value p-value (Intercept) 49.52075 0.68726 72.056 <2e-16 *** 48.53283 0.69807 69.524 <2e-16 ***
age -0.18010 0.15558 -1.158 0.248 -0.14235 0.15802 -0.901 0.368 group-E+RE -0.10773 1.37451 -0.078 0.938 1.40712 1.39615 1.008 0.314 gender-B+G -0.26226 1.37451 -0.191 0.849 -2.25417 1.39615 -1.615 0.107 PCC 9.78119 3.87334 2.525 0.012 * 6.27988 3.93431 1.596 0.111 age:group-E+RE 0.17314 0.31115 0.556 0.578 -0.09766 0.31605 -0.309 0.757 age:gender-B+G -0.09762 0.31115 -0.314 0.754 0.23725 0.31605 0.751 0.453 group-E+RE:gender-B+G 0.13815 2.74903 0.050 0.960 1.3004 2.7923 0.466 0.642 age:PCC -0.24603 0.76968 -0.320 0.749 1.04696 0.7818 1.339 0.181 group-E+RE:PCC -2.57666 7.74669 -0.333 0.740 6.73749 7.86862 0.856 0.392 gender-B+G:PCC 4.78130 7.74669 0.617 0.537 6.18243 7.86862 0.786 0.433 age:group-E+RE:gender-B+G -0.36814 0.62230 -0.592 0.555 -0.34867 0.6321 -0.552 0.582 age:group-E+RE:PCC 1.40282 1.53936 0.911 0.363 -1.22447 1.56359 -0.783 0.434 age:gender-B+G:PCC -2.18079 1.53936 -1.417 0.157 -1.98719 1.56359 -1.271 0.205 group-E+RE:gender-B+G:PCC -14.12465 15.49337 -0.912 0.363 -1.44056 15.73723 -0.092 0.927 age:group-E+RE:gender-B+G:PCC 0.46196 3.07872 0.150 0.881 -0.16472 3.12718 -0.053 0.958 * p <0.05, *** p <0.001
Table 8
Outcomes linear model C-TRF Internalising and Externalising scales
C-TRF Internalising scale C-TRF Externalising scale
Estimate Std. Error t-value p-value Estimate Std. Error t-value p-value (Intercept) 49.32874 0.539921 91.363 <2e-16 *** 50.43283 0.50218 100.428 <2e-16 ***
age 0.096624 0.121874 0.793 0.4283 -0.01946 0.11335 -0.172 0.86381 group-E+RE -0.31105 1.079842 -0.288 0.7735 3.53557 1.00436 3.520 0.00048 *** gender-B+G 0.251461 1.079842 0.233 0.816 0.14436 1.00436 0.144 0.88578 PCC 1.284488 3.168081 0.405 0.6854 0.74522 2.94662 0.253 0.80047 age:group-E+RE 0.494754 0.243748 2.03 0.0430 * 0.43303 0.22671 1.910 0.05684 . age:gender-B+G 0.336673 0.243748 1.381 0.168 0.07499 0.22671 0.331 0.74097 group-E+RE:gender-B+G -2.00816 2.159684 -0.93 0.353 1.4873 2.00871 0.740 0.45947 age:PCC 0.005228 0.655366 0.008 0.9936 0.88511 0.60955 1.452 0.14726 group-E+RE:PCC -15.5563 6.336162 -2.455 0.0145 * -10.68185 5.89324 -1.813 0.07064 . gender-B+G:PCC -5.8782 6.336162 -0.928 0.3541 -1.41215 5.89324 -0.240 0.81075 age:group-E+RE:gender-B+G 0.607456 0.487496 1.246 0.2135 0.91926 0.45342 2.027 0.04328 * age:group-E+RE:PCC -0.46347 1.310731 -0.354 0.7238 -0.03177 1.21911 -0.026 0.97922 age:gender-B+G:PCC 1.970226 1.310731 1.503 0.1336 0.83174 1.21911 0.682 0.49547 group-E+RE:gender-B+G:PCC -23.6526 12.67232 -1.866 0.0627 . -21.22821 11.78649 -1.801 0.07244 . age:group-E+RE:gender-B+G:PCC -2.34995 2.621462 -0.896 0.3706 -0.80388 2.43821 -0.330 0.74180 . p <0.1, * p <0.05, *** p <0.001
5.2.1 Group, Age, and Gender Effects
Group, age, and gender effects were examined in order to answer the second research question. The linear models indicated significant effects of age, group, and gender on the C-TRF Internalising and Externalising scales (Table 8). No significant effects of age, group, and gender were found on the CBCL Internalising and Externalising scales (Table 7). The linear model with the C-TRF Internalising scale as outcome variable revealed a significant interaction of age and group on the C-TRF Internalising scale (t[403] = 2.030, p = 0.043). The difference in the rate of increase of C-TRF Internalising problems depending on age is 0.49 points greater for children with DLD-RE than for children with DLD-E (95% confidence interval = 0.016 .. 0.97 points). This means that the increase of therapist-reported internalising behaviour as a function of age is bigger for children with DLD-RE than for children with DLD-E (Figure 1).
Figure 1
Two-way interaction age * group for the C-TRF Internalising scale
The linear model with the C-TRF Externalising scale as outcome variable revealed a significant group effect (t[403]= 3.520, p = 0.00048). Therapist ratings on the Externalising scale were higher for children with DLD-RE than for children with DLD-E (estimated difference = 3.54 points, 95% confidence interval = 1.56 .. 5.51 points). A significant interaction of age, group and gender was also found for the C-TRF Externalising scale (t[403] = -2.027, p = 0.043). This interaction suggests that girls with DLD-E score lower on the C-TRF Externalising scale as they get older but girls with DLD-RE score higher on this scale as they
get older (estimated difference = 0.92 points, 95% confidence interval = 0.028 .. 1.81 points) (Figure 2). This effect was not found for boys.
Figure 2
Three-way interaction age * group * gender for the C-TRF Externalising scale
5.2.2 Expressive Phonology Effects
PCC effects were examined in order to answer the third research question. To warrant analysis of the DLD subgroups, PCC scores of children with DLD-E and DLD-RE were compared. Descriptive statistics of PCC scores for children with DLD-E and DLD-RE can be found in Table 9.
Table 9
Descriptive statistics of PCC scores of children with DLD-E and DLD-RE
mean SD range
DLD-E
(n = 169) 53.43 18.49 7.5-91.67 DLD-RE
(n = 280) 64.26 18.47 7.8-100
The Shapiro-Wilk test, using the shapiro.test function from the dplyr package from R (Wickham et al., 2020), indicated that the PCC scores were non-normally distributed (p = 0.0081). The difference between PCC scores was therefore tested with an unpaired Wilcoxon test in R (R Core Team, 2020). The Wilcoxon test revealed that the difference in PCC score between
children with DLD-E and DLD-RE in the current sample was significant (p = 1.193e-08). This difference in PCC scores between children with DLD-E and DLD-RE motivates looking for a group effect on the CBCL and C-TRF Internalising and Externalising scales. The linear models indicated significant effects of PCC score on the CBCL and C-TRF Internalising scales (Table 7 and 8, respectively). No significant effects were found on the CBCL and C-TRF Externalising scales (Table 7 and 8, respectively). The linear model with the CBCL Internalising scale as outcome variable revealed a significant effect of PCC (t[356] = 2.525, p = 0.012), indicating that every point higher on the PCC is associated with a 9.78 point increase in CBCL Internalising score (95% confidence interval = 2.16 .. 17.40 points). The linear model with the C-TRF Internalising scale as outcome variable revealed a significant effect of the PCC and group interaction (t[356] = 2.455, p = 0.015). This indicated that the increase of the C-TRF Internalising score associated with a higher PCC score is 15.56 points greater for children with DLD-E than for children with DLD-RE (95% confidence interval = 3.10 .. 28.01 points) (Figure 3).
Figure 3
Two-way interaction PCC * group for the C-TRF Internalising scale
5.3 Relation Between CBCL and C-TRF Ratings
The second part of the second research question concerned the relation between internalising and externalising behaviour and informant (parents or pedagogical therapists). In order to investigate the relation between informant and internalising and externalising behaviour, the correlation between CBCL and C-TRF ratings was calculated for the children with both CBCL
and C-TRF ratings (n = 342). The Shapiro-Wilk test, using the shapiro.test function from the dplyr package from R (Wickham et al., 2020), showed that the CBCL and C-TRF Internalising and Externalising scales were non-normally distributed (p <.001). The correlations between the CBCL and C-TRF ratings were therefore tested with Spearman’s rank correlation test in R (R Core Team, 2020). Significant weak positive correlations between the CBCL and C-TRF ratings were found for both the Internalising and Externalising scales (Table 10).
Table 10
Spearman’s correlations between CBCL and C-TRF Internalising and Externalising scales C-TRF Internalising C-TRF Externalising
CBCL Internalising .11 *
CBCL Externalising .26 **
* p <0.05, ** p <0.01
6 Discussion 6.1 Behavioural Profile Preschoolers with DLD
The first research question this thesis aimed to answer was whether preschoolers with DLD exhibit levels of internalising and externalising behaviour comparable to the general population, and if not, what kind of behaviour was reported for preschoolers with DLD. Internalising and externalising behaviour occurs in preschoolers in general, it is not restricted to preschoolers with language difficulties. The parent and pedagogical therapist ratings of behavioural problems for our preschoolers were therefore compared to those reported for the general population to determine if Dutch preschoolers with DLD exhibit elevated levels of internalising and externalising behaviour. Comparison to the mean scores from a Dutch general population sample (Tick et al., 2007) suggests that our sample of preschoolers with DLD generally exhibit elevated levels of behavioural problems according to parents. Only the mean score for aggressive behaviour was lower than the means of the general population, resulting in a lower externalising problems mean. An absence of aggressive behaviour has been argued to be characteristic of children with DLD (Coster et al., 1999). This hypothesis was based on a sample of eight-to-twelve-year-old children with DLD, and has now been replicated with preschoolers. This finding is however not conclusive, as other studies have found elevated levels of aggressive behaviour in preschoolers with DLD (Maggio et al., 2014; Özcebe et al., 2019; van Daal et al., 2007; Willinger et al., 2003). The mean scores of therapist-reported behavioural problems for the current sample are mostly lower than the omnicultural means. This does not necessarily mean that pedagogical therapists report few behavioural problems for our
preschoolers with DLD, as the Netherlands appears to score on the lower side of the omnicultural means (Rescorla et al., 2012). More research regarding therapist-ratings of behavioural problems in Dutch preschoolers with and without DLD is needed to form a clearer picture of behavioural problems exhibited in a school or intervention group setting. The mean scores for withdrawal and attention problems are higher than the omnicultural means, suggesting that our preschoolers with DLD exhibit elevated levels of withdrawal and attention problems compared to the general population. The comparison to general population samples indicates that our preschoolers with DLD exhibit elevated levels of behavioural problems according to parents, with the exception of aggressive behaviour, and elevated levels of withdrawal and attention problems according to pedagogical therapists.
The most often parent-reported behaviour for preschoolers with DLD in the current study are emotionally reactive behaviour, somatic complaints, withdrawal, and attention problems. These exploratory results are consistent with previous research also finding elevated levels of these kinds of behaviour (Beitchman et al., 1989; Benasich et al., 1993; Maggio et al., 2014; Özcebe et al., 2019; Rescorla et al., 2007; Stanton-Chapman et al., 2007; van Daal et al., 2007; Willinger et al., 2003). However, these behavioural problems do not consistently co-occur in research and they appear in combination with various other kinds of behaviour, such as aggressive behaviour (Maggio et al., 2014; Özcebe et al., 2019; van Daal et al., 2007; Willinger et al., 2003), and anxious/depressed beahviour (Maggio et al., 2014; Özcebe et al., 2019; Stanton-Chapman et al., 2007), which are not particularly elevated for the current sample. Parents in the present research reported similar levels of internalising and externalising behaviour. This is consistent with findings for a similar Dutch samples (Keegstra et al., 2010; van Daal et al., 2007), although some non-Dutch studies have found more internalising problems (Maggio et al., 2014; Stanton-Chapman et al., 2007). A possible explanation for the different kinds of behaviour reported by parents for preschoolers across studies is that there could cultural differences with regard to the behaviour that is considered to be problematic. However, although Rescorla et al. (2011) found large differences among countries in amount of problems reported, these differences were not related to differences on specific subscales. Rather, variation could be attributed to the difference in the tendency to generally score high or low overall. Different results can also be observed between Dutch studies. Van Daal et al. (2007) mention often occurring aggressive behaviour, contrary to what was found in the current research. A more likely explanation is a difference in sample characteristics. The current research included both children with (suspected) ASD and multilingual children, whereas other
studies excluded these children (Maggio et al., 2014; Stanton-Chapman et al., 2007; van Daal et al., 2007). Different non-verbal IQ cutoff scores are also used, such as 75 (van Daal et al., 2007), 80 (Stanton-Chapman et al., 2007), and 85 (Keegstra et al., 2010). The current research used a cutoff score of 70. The rationale behind this is that low non-verbal IQ (70-85) does not exclude a possible DLD diagnosis (Bishop et al., 2016). There are also differences in sample size, from 38 participants in Özcebe et al. (2019) to 449 in the current sample. Small sample sizes are not as representative of the entire population as bigger samples, which could explain the varying outcomes in previous research.
Pedagogical therapists most frequently mention attention problems, which causes the predominance of externalising problems (aggressive behaviour was hardly mentioned). Second most often reported is withdrawal. Therapist-ratings of externalising behaviour are higher than the ratings for internalising behaviour. These results are consistent with the few studies that have examined teacher ratings of preschoolers with DLD. Vermeij et al. (2019) found a similar difference for a sample of 185 Dutch preschoolers with DLD, and Qi and Kaiser (2004) report a higher average for externalising than internalising behaviour as well for a sample of 60 American 3- and 4-year-olds with DLD. As more externalising behaviour is reported both by teachers and pedagogical therapists, it appears that children struggle with attention keeping both in school and in intervention groups. These results are not surprising, teachers and pedagogical therapists are likely to register deviant behaviour with respect to attention, as paying attention is important behaviour in a school and intervention group setting (Lindsay et al., 2007). Withdrawal is also likely to be registered as participation from children is expected. Other internalising behaviour, on the other hand, is less easily detected in a school setting (Berg-Nielsen et al., 2012).
Somatic complaints and attention problems are not only reported most often in the current research, they are also elevated compared to the general population (Section 6.1). This suggests that children with DLD may be especially vulnerable to problems in these areas. It has been argued that somatic complaints are not the result of language problems per se (Samson et al., 2020). Van Daal et al. (2007) also reported that somatic complaints were not associated with any of their language measures. Samson et al. (2020) argue that somatic complaints are the result of lower emotional competence. Emotional competence is described as the ability to understand and communicate emotions (Samson et al., 2020). However, the language difficulties children with DLD have hinder the development of emotional competence (Samson et al., 2020). Less emotional competence due to language difficulties may thus explain why
children with DLD are reported as having somatic complaints (van Daal et al., 2007; Özcebe et al., 2019). That children with DLD would exhibit attention problems in situations that are verbally demanding, such as in intervention groups, is not surprising. However, attention problems are rated highest and are reported often by both parents and pedagogical therapists. When attention problems are evident across multiple settings (such as at home and at the intervention group), this is a reason to suspect Attention Deficit Hyperactivity Disorder (ADHD; WHO, 2018b). A relation to ADHD has been demonstrated in previous research (Benasich et al., 1993; Beitchman et al., 1989; Yew & O’Kearney, 2013), although there appears to be a difference between the attention problems seen in DLD and those in ADHD (Leonard, 2014). Language, attention, and executive functioning are interrelated (Tomas & Vissers, 2019), which could be the reason why children with language problems, such as DLD, have high levels of attention problems.
6.2 Factors Relating to Behavioural Problems in Preschoolers with DLD 6.2.1 Expressive or Expressive-Receptive Language Difficulty, Age, and Gender
The second research question of this thesis concerned the relation between expressive or expressive-receptive language difficulty, age, and gender and the behaviour reported for Dutch preschoolers with DLD. Results indicated that there is a bigger increase of therapist-reported internalising behaviour for children with DLD-RE as they grow older, compared to the increase seen in children with DLD-E. In the current research, the therapist-reported internalising behaviour means of children with DLD-E and DLD-RE are equal. If the trend of a bigger increase of internalising behaviour for children with DLD-RE persists, this suggests that children with DLD-RE will have higher internalising behaviour means than children with DLD-E when they are a few years older. Evidence for elevated levels of internalising behaviour for children with DLD-RE compared to children with DLD-E has indeed been reported in previous research with school-aged children (Lindsay and Dockrell, 2000; Toppelberg & Shapiro, 2000; Yew and O’Kearney, 2013). However, this difference could also be present already at a young age, as Vermeij et al. (2019) found higher teacher-reported internalising scores for children with DLD-RE compared to DLD-E for a group of 3-year-old children with DLD. Future research is needed to confirm a bigger increase of internalising behaviour for children with DLD-RE than DLD-E. A difference between DLD-E and DLD-RE was also established for therapist-reported externalising behaviour. Pedagogical therapists report more externalising problems and higher externalising scores for children with DLD-RE than for children with DLD-E. These findings are consistent with results from Vermeij et al. (2019) and
