What Are the Peer Interaction Strengths and Difficulties in Children with Developmental Language Disorder? A Systematic Review

The current review gathers together research investigating peer interaction skills in children with Developmental Language Disorder (DLD) to give an overview of the strengths and challenges experienced by these children when interacting with other children. A systematic review was conducted to summarise the literature on peer interaction strengths and difficulties in children with DLD. No restrictions on time-period were made and the selection criteria accounted for many of the diagnostic labels previously used to refer to DLD. Studies included in this review involve English-speaking children of UK primary school age (4–11 years). A systematic search of databases identified 28 papers that met the inclusion criteria. Children with DLD are found to experience many challenges when interacting with peers. Difficulties have been found in studies exploring discourse characteristics such as turn-taking and in behaviours during play, such as access behaviours. Heterogeneity was however notable and peer interaction strengths are found in terms of the children’s abilities to make friends, use verbal and non-verbal behaviour to make joint decisions with peers, and abilities to engage with peers in social pretend play. While it is encouraging to find research exploring many different areas of peer interaction competence in children with DLD, the research is highly disparate and there are many research findings awaiting replication. The current evidence base is unable to comprehensively define the characteristics of peer interactions of children with DLD.


Introduction
Developmental Language Disorder (DLD) affects approximately 7.6% of 4-5 year olds in the UK [1]. It is diagnosed when children have significantly impaired expressive language and/or receptive language skills in the absence of any hearing or other neurodevelopmental disorder [2]. Children with DLD maintain lower than average language levels throughout their development, and so continue to lag behind their peers throughout childhood and beyond [3,4]. While there is typical development aside from a primary problem with language [5], it is common for children with DLD to show some level of difficulty in areas of attention, motor skills, and social skills [2]. The present study aims to establish the nature of the strengths and difficulties children with DLD display when interacting with peers by systematically reviewing the research in this area. It is difficult to devise strategies to support children with DLD in their social development without a deep insight into possible mechanisms underlying their social difficulties. Establishing the specific nature of peer interaction skills in children with DLD is a critical first step.
Longitudinal studies of large population cohorts have established evidence of peer problems amongst children with DLD [6,7]. Cross-sectional studies similarly find children with DLD to be rated by teachers as having lower social skills, and they are also found to have fewer peer relationships [8], findings which suggest an association between peer problems and longer-term negative consequences such as the inability to maintain friendships.
One characteristic social behaviour often cited as being common in children with DLD is social withdrawal [9][10][11][12]. Studies, again using teacher-rated questionnaires, find children with DLD to have significantly higher levels of social withdrawal compared to children without DLD [13,14]. While the severity of the children's language difficulties alone do not predict levels of reticence, emotion regulation skills together with language skills can predict reticence in children with DLD [15]. This finding has led researchers to assume children with DLD are fearful of social situations, not just due to having a communication difficulty, but due to difficulties in emotional development [13]. This idea ties in with a previously proposed theory termed The Social Deviance Model [9]. This theory proposes that children with DLD might have inherent difficulties with their socioemotional development, independent of their language difficulties, and it is these socioemotional issues which result in challenges with socialising with other people [9]. The available research into social withdrawal in children with DLD provides a promising, albeit tentative, insight into the underlying reasons for some of the social characteristics observed in children with DLD. However, it is not clear whether social withdrawal is the key behaviour contributing to the peer problems experienced by children with DLD. For example, it is not known whether social withdrawal affects all children with DLD and if not, whether these other children still experience peer problems.
A systematic review and meta-analysis has shown children with low expressive language skills and low receptive language skills have higher levels of behavioural problems, pointing to the possibility that underlying reasons for peer problems in children with DLD might extend beyond simply social withdrawal [16]. Many studies investigating the behaviour of children with disordered language find evidence of externalising problems, which includes angry, oppositional and aggressive behaviour [17][18][19]. More research is necessary to understand whether these externalising behaviours are displayed during social interactions, thus accounting for some of their "peer problems". It could be that children who lack the expressive language skills necessary for a clear expression of their needs use aggressive behaviour instead of using language. Indeed one study observed behaviour in minimally verbal autistic children and found those who displayed challenging behaviours did so in place of requests, or rejections [20]. Alternatively, disordered receptive language skills might impair the ability to understand social situations and so internal thinking could be relatively immature and involve inappropriate thoughts about other's intentions, and this could result in an inappropriate response to the situation, such as using physical aggression [19]. It will therefore be a useful endeavour to establish any consistency across research findings about the nature of externalising behaviours and their relation to peer interactions in children with DLD.
Despite evidence that children with language disorder risk developing peer problems, few interventions exist to support their social development. A systematic review conducted in 2012 found only eight studies assessing interventions to support social communication in children with disordered language skills [21]. All of these were exploratory studies, involving samples of less than 20 children, to test the feasibility of interventions, suggesting research in this area is still in its infancy. All eight studies focused on improving the children's discourse skills, thereby assuming improved discourse skills will lead to enhanced social interaction skills. For example, some focused on improving comprehension skills, in terms of repairing communication breakdowns with their communication partner or monitoring their understanding during conversations [22,23]. In this way, current social skills interventions for DLD build on the premise that their peer interaction difficulties are a direct result of their language difficulties.
To our knowledge, no systematic review of research investigating the social interactions of children with DLD has yet been conducted. In the current study, ten databases are searched using broad search terms with no restrictions on year of publication to capture as many studies as possible exploring this research area. The aim of the current study is to systematically review the findings of studies of peer interaction involving children with DLD in order to establish any consistency as to the specific nature of their strengths and difficulties when socialising with peers. This may enhance our understanding of the potential mechanisms underlying the peer difficulties shown in this population and assist the development of effective, tailored interventions.

Materials and Methods
The current review includes a systematic search of the literature and narrative synthesis of the research on peer interaction skills in children with DLD.

Search Strategy
Many different terms have been used to define DLD. The current review aimed to capture as many studies of peer interaction characteristics in children with DLD as possible, regardless of how DLD had been defined in the past. To this end the following terms were included as acceptable terms to define language impairment: Developmental Language Disorder (DLD), Specific Language Impairment (SLI), expressive language disorder, mixed expressive-receptive language disorder, previously identified language impairment, language delay, language learning impairment, language disability, language problems, developmental aphasia and developmental dysphasia. These terms were based on a review article [24], and the indexed terms recorded in the included databases. No restrictions were placed on the year of publication, or on the study design.
Searches were conducted from March 2018 to May 2018 using the following databases; PubMed, Embase, Web of Science Core Collection, Web of Science BIOSIS Citation Index and SciELO Citation Index, PsycNET (PsycINFO), ERIC, Proquest International Bibliography of the Social Sciences, Dissertations and Theses A&I, and Ovid (Social Policy and Practice). The search terms included variations of the words "children", "interaction" and "Developmental Language Disorder", and the Boolean operators AND and OR were used. Search terms were grouped into three searches which were inserted in the same way in every database (see Appendix A).

Inclusion Criteria
To be included in the review, studies were required to meet the following criteria:

•
Children must be 4-11 years old as this is the age at which children attend primary school in the UK. Furthermore, children who were 3 years old or younger were not included because language abilities at this age tend to be too unstable to make an accurate diagnosis of DLD [3].

•
The language impairment must be identified as meeting clinical cut-off scores on a standardised language assessment by a researcher or through formal diagnosis by a speech and language practitioner. • Even if the child has a comorbid condition, such as emotional and/or behavioural difficulties or Attention Deficit Hyperactivity Disorder (ADHD), the child has been selected for the study because language is their primary area of need.

•
It must be an empirical study; Intervention studies could be included so long as they included baseline measures of peer interactions.

•
The study must be available in English.

•
The study must include a measure of peer interactions. There are no restrictions on time period or study design in the review, thus giving an outline of the full breadth of research in this area.

Exclusion Criteria
Studies of children who did not speak English, or who had a first language other than English, were not included in the present review. Studies which only included children with language problems who had Autism Spectrum Disorder (ASD), hearing loss, otitis media, or an identified genetic condition known to cause language problems (for example Downs Syndrome, or Fragile X Syndrome), were not included, in line with guidelines for a diagnosis of DLD [2].

Quality Assessment Procedure
The scientific quality of the studies included in this review was assessed using a framework which appraises the quality of both quantitative and qualitative literature [25]. Nine criteria were used to assess quantitative studies and eleven criteria were used to assess qualitative studies (see Appendix B). For each criterion a rating of 1 (low) to 3 (high) was assigned depending on how well the study fulfilled the criterion guidelines. A criterion guide, giving a description of the evidence required to meet a score of 1, 2, or 3 was created by the first author, using the framework provided by Alderfer et al. (2010) [25]. The average of these criteria divided by the number of criteria assessed gave the final quality score.

Results
The initial search identified 29,686 records. A further 6 items which were not captured by the initial database searching, but were part of a relevant review [21], were also identified. After duplicates were removed 14,213 unique titles and abstracts were screened for eligibility. Any items which were irrelevant to the subject of peer interactions of children with DLD were excluded at this stage. This resulted in 616 studies being included in the abstract screening stage. Initially, 398 studies were excluded from the review at the abstract screening stage. The main reasons being that children did not meet inclusion criteria for age or native language, or the study was not empirical, or the study did not measure peer interactions, for example mother-child interactions were measured instead. A second-rater screened 10% of these studies (interrater reliability 79.4%). All disagreements were discussed until a mutual understanding was met. Consequently, the primary investigator revisited all items excluded from the initial abstract screening. A decision was made to include a further 48 items which had abstracts providing too little information regarding the subjects or methods to justify an exclusion at this stage. In total, 266 studies were included in the full text screening. Ten percent of the studies included in the full text screening were screened by a second-rater (interrater reliability 91.3%). In total, 28 studies were deemed eligible for inclusion in the systematic review (see Figure 1 and Table 1). Figure 1 has been adapted from the PRISMA flowchart made freely available online [26,27]. See Table 2 for a summary of findings from the 28 included studies.

Reason Excluded Total Items Excluded
Did not measure peer interactions 63 Children did not meet criteria for having DLD in the absence of other diagnoses known to impact on language skills 24 Children were not native-English speakers 55 Children were not within the age range 4-11 years 72 Not an empirical study 9 Same sample was described in a later study 14 Search had to be terminated because the item was a thesis and the university holding the thesis was unable to send or lend the item 1 Figure 1. Prisma flowchart to show study selection process [27]. Strengths: Many children with DLD (60%) did not have peer problems and many (56%) had typical levels of prosocial behaviour. * Difficulties: Some children with DLD (40%) had peer problems at age 10-11. Some children with DLD (44%) had few prosocial behaviours. * Children with DLD had a higher risk of being bullied at school than TD children (p < 0.005). Some children with DLD (35%) aged 10-11 experienced at least 3 different types of peer victimisation at least once a week. Observation during 20-min play sessions. Investigators devised their own coding scheme.
Strengths: All children with DLD (N = 3) engaged in social pretend play with TD and DLD peers. Children with DLD showed a higher proportion of successful conversational turns during social pretend play compared to non-social pretend play episodes. * Difficulties: Turn-taking was less frequent in DLD-dyads than in DLD-TD dyads and turn-taking in DLD-TD dyads was less frequent than in TD-dyads.*  Strengths: Child with DLD produced more utterances and higher proportion of assertions than TD child. Child with DLD makes equal number of turn initiations as TD child. * Difficulties: DLD child responsible for high proportion of noncollaborative utterances because frequently makes non partner-directed utterances. * Difficulties: Children with DLD who did not form reciprocal friendships observed peers significantly more often (p < 0.05) and sought the attention of their peers significantly more frequently (p < 0.001) than children with DLD who did form reciprocal friendships. SDQ (Goodman, 1997): Peer problems, Prosocial subscales. Parent-report.

Strengths:
No difference between children with and without DLD in peer problem scores at age 7, or in prosocial scores at 5 and 7, after adjusting statistical models for potential confounders.
Difficulties: Peer problem scores significantly higher in children with DLD than children without at age 4 (p < 0.001) and 5 (p = 0.01), after adjusting statistical models for potential confounders. Prosocial scores significantly lower in children with DLD than children without at age 4 (p < 0.001). Cross-sectional between subjects. Triadic interactions with same age TD children (13N).
Behaviour during play opportunities: Type of play behaviour, Behaviour and discourse during play relating to peer access, and whether access was successfully achieved.
Observation during 10-min play sessions. Investigators devised their own coding scheme.
Strengths: Six children with DLD did access play with peers.
Difficulties: Four children with DLD did not achieve access to play with peers. Children with DLD took longer than TD peers to achieve access play (p = 0.024).

Overview of Studies
Most of the studies included in this review (60.7%) took place in the United States of America, while others took place in the UK (25.0%), Canada (10.7%) and Australia (3.6%). Some of the included studies come from grey literature (14.3%), specifically PhD theses. While some studies (28.6%) have been published within the last 10 years, a sizeable proportion of studies (35.7%) were conducted before the year 2000, with half of these conducted before 1990.
The studies included in this review involve a total of 856 children with DLD. Twenty-three studies were cross-sectional and five were longitudinal. Of these longitudinal studies three reported attrition rates during data collection at the age range included in this review [7,10,31]. Attrition ranged from 17.4-22.0% and resulted from the researchers being unable to locate the participants or from participants failing to return questionnaires. One study, containing eight children, did not report the gender of the sample [35]. Of the remaining 848 children, 227 are female and 621 are male. None of the children included in this review were reported to have co-occurring conditions. Over half of studies (57.1%) did not identify any co-occurring conditions, while the remaining studies (42.9%) did not report whether children with co-occurring conditions were included. Different age groups are reasonably well represented. The age range of the children with DLD is wider in some studies than others. By looking at each year group separately we find studies include children who are within the age range of 4-5 years (42.9%), 6-7 years (67.9%), 8-9 years (50.0%) and 10-11 years (32.1%). See Table 3 for an overview of the type of school children with DLD were enrolled in across all included studies.

Quality Appraisal
Alderfer et al.'s (2010) quality appraisal framework [25] was used to assess the quality of all the included studies. Studies using quantitative measures were assessed on parameters which included statistical power and appropriate methods. These were felt to be particularly important for the current review of the strengths and difficulties in the domain of peer-based social skills reported by studies of primary school children with DLD. Studies with high statistical power with a high level of internal validity are valuable for this review because this ensures the results of the study are generalisable to other primary school children with DLD.
A second-rater independently appraised the quality of 36% of the papers. Using item-by-item agreement, the two raters were found to agree on 64.4% of quality scores. Inter-rater agreement was low for some of the papers. In particular there was disagreement on the scoring criteria used for categorising papers as low, medium or high quality for the reliable measurement of their variables and whether their statistical power was sufficient. As a result, the two raters engaged in detailed conversations about what criteria would be necessary to achieve the highest quality score on these measures. Previously, for example, there was disagreement over the quality scores given for level of statistical power if the study included fewer than ten children. Following discussions, the raters agreed that medium quality was the maximum score studies including fewer than ten children could achieve for sufficient statistical power. The raters jointly analysed each of the 36% of papers and any disagreements were discussed until 100% agreement was reached on all of the scores. These jointly agreed scores were used for the final quality scores (see Appendix C).
The studies included in this review achieved total quality scores in the range of 1.33 and 3.00 (mean = 2.47, SD = 0.36). In line with previous literature [25] studies with a total quality score at least one standard deviation below the mean (rating < 2.12) were treated as having low scientific quality (N = 3, 11%). In this review, studies with a total quality score at least one standard deviation above the mean (rating > 2.82) were deemed to have high scientific quality (N = 5, 18%). Low quality papers had small sample sizes and therefore scored low on statistical power and low or medium on external validity. Low quality papers also used their own scoring schemes to measure observed peer interactions and therefore scored medium on the criteria for appropriate methods because while their methods were appropriate, they did not provide enough detail to allow for replication.
High quality papers scored high on criteria for appropriate methods because they used questionnaires such as the Strengths and Difficulties Questionnaire which is easily replicated (SDQ, [53]). Four of the five high quality papers scored high on statistical power and external validity too because their samples included at least 30 children. Those with high scientific quality are highlighted in the text and are useful for interpreting the current evidence base. Those with low scientific quality are retained for this review and are given merit because despite the low generalisability of their findings due to their small samples, their observations provide an important insight into the possible strengths and difficulties children with DLD have when interacting with peers.

Peer-Based Social Domains Measured
We have identified five different skill areas which are investigated in the included studies: overall peer competence, behaviour during play opportunities, discourse characteristics, cooperative behaviour and victimisation. Table 4 provides an overview of the ways in which these skills domains have been explored. It is possible some of these skill areas overlap with respect to the constructs the researchers were aiming to assess. For example, the rationale for investigating discourse characteristics may have been to gain an insight into the children's play with peers. We categorise the research in this way to allow for greater ease of interpretation. Note that studies are not mutually exclusive; Some studies include measures on more than one relevant peer interaction variable and some studies include measures spanning more than one skill area.
A range of informants were used in different studies measuring overall peer competence including teacher-reports (46.1%), teacher-reports in combination with self-reports (7.7%), teacher-reports in combination with peer-reports (7.7%) parent-reports (15.4%), a combination of teacher and parent reports (15.4%), or direct observation (7.7%). No studies used questionnaire methods in combination with direct observation to measure overall peer competence. All studies measuring behaviour during play opportunities or discourse characteristics used direct observation. Studies which measured cooperative behaviour did so using self-report (33.3%) or direct observation (66.7%). Studies measuring victimisation did so using direct observation (25%), self-report (50%) or teacher-report (25%).
The following narrative synthesis will summarise the available literature on peer interaction strengths and difficulties in primary school children with DLD. It should be noted that 15 (53.6%) of the 28 included studies have not conducted statistical analyses on at least one of their outcome variables which are relevant to peer interaction. This is often due to the small size of the included samples. The results of these studies are therefore largely descriptive which can make it difficult to draw firm conclusions. Nevertheless, the studies summarised in the following narrative synthesis provide a useful foundation on which to build new lines of research.

Overall Peer Competence
Most studies exploring overall peer competence found higher levels of peer problems in children with DLD compared to children without DLD [10,28,31,44,45,48]. Some studies found lower levels of prosocial skills in primary school children with DLD compared to children without DLD [28,31]. There is some discrepancy in the age that peer problems are most pronounced, according to available longitudinal data on primary school children. One recent study finds elevated levels of peer problems at 4 and 5 years, which appear to subside by 7 years [44], yet another study finds elevated peer problems at ages 8 and 10 [45]. Sociometric measures found children with DLD to be less well liked and accepted by their peers compared to children without DLD [38].
There are individual differences in overall peer competence that should be highlighted. One study finds that although some children with DLD have high levels of aggression and some have high levels of withdrawal, other children with DLD have a typical social profile [29]. Similarly, another study found 20% of children with DLD in their sample did not have peer problems [7]. Peer interaction strengths were noted in an earlier study where children with DLD were able to form reciprocal friendships to the same extent as children without DLD [42]. Additionally, some found no differences in the level of prosocial behaviour displayed by children with and without DLD [36]. It seems peer problems are not inevitable for children with DLD because some children have relatively good social skills.

Behaviour during Play Opportunity
Children with DLD were found to interact with their peers on the playground less frequently than children without DLD [11]. There are no conclusive results regarding the type of peer interaction children with DLD are most likely to engage in. In some studies children with DLD are found to engage in non-play more often than play when given the opportunity to interact with peers [51] and high levels of active withdrawal from peers are observed [11]. On the other hand, others find non-play, onlooker behaviour to be rare, with children with DLD showing high levels of interaction with peers on the playground [49]. Furthermore, children with DLD are able to engage in pretend play with peers, with some showing more sophisticated levels of pretend play than children with typical language development [34]. Interestingly, one study found children with DLD mainly engage in social-conversation with peers when they are interacting with peers during free-play, and rough-and-tumble play is rare, suggesting children with DLD do not use more physical forms of play to overcome their language difficulties [11].
Each of the studies measuring access behaviours find children with DLD generally display difficulties accessing play [33,46,47]. Children with DLD use passive social entry patterns [47], they rarely approach peers [33], and they wait longer for an invitation by their peers [46].

Discourse Characteristics
The results from studies measuring discourse characteristics during peer interactions by children with DLD are highly mixed. Studies find children with DLD experience difficulties maintaining conversation with their peers [37]. Children with DLD direct fewer requests to their peers compared to adults [52]. They also ask fewer internal state questions when they are paired with same-age peers compared to when they are paired with younger children [37]. Additionally, children with DLD are more likely than children without DLD to reintroduce topics that have already been introduced, suggesting they find it harder to introduce new topics of conversation [35]. These studies therefore demonstrate that children with DLD experience difficulties during talk with peers. However, a word of caution is needed, because two of the studies [37,52] included fewer than ten children and therefore these results require further replication.
On the other hand, many studies found surprising strengths in the discourse characteristics of children with DLD. They use conversation as a way of seeking information from their partner no less often than children without DLD [49] and are able to talk about rules and plans to engage in pretend play with their peers [43]. Indeed, it seems that play might facilitate discourse in children with DLD. Peers make other-directed turns more frequently during play than between play intervals, and this seems to help children with DLD maintain conversations because their partner can create a shared referent for them to build on and use to make requests [32]. During pretend play, children with DLD are also able to share scripts using non-verbal behaviour [43]. Again, however, these findings come from small sample sizes; Craig and Gallagher's (1986) [32] findings come from a single case study. These positive findings are therefore in need of replication before they can be trusted.

Cooperative Behaviour
Cooperative behaviour has either been measured in primary school children with DLD using tasks which require children to work together to complete a group activity and their behaviour or discourse is observed, or by presenting children with hypothetical situations and asking how they would behave. Numerous studies find evidence of poor conflict resolution skills in children with DLD [28,30,48]. Unlike typically developing children who ask for clarification from their peers to understand their motives for their actions leading to the conflict event, children with DLD select less sophisticated conflict resolution strategies, such as involving an adult or physical retaliation. It has been suggested that children with DLD have a less nuanced understanding of peer conflict situations than children without DLD because they provide less precise judgements about conflict resolution strategies compared to children without DLD using a visual analogue scale task [30]. There are mixed results regarding the ability of children with DLD to collaborate with their peers. Children with DLD are found to produce fewer validating comments during cooperative tasks than children without DLD [38], and children with DLD who show high levels of withdrawal or high levels of aggression are found to perform poorly on cooperative tasks [29]. Furthermore, children with DLD paired with other DLD children are found to take longer to reach group decisions compared to typically developing dyads [41]. It therefore seems that children with DLD find it difficult to work with their peers to achieve shared goals.
On the other hand, other children with DLD who are scored as having a typical social profile, display fair levels of cooperative skills that are no worse than their typically developing peers [29]. Furthermore, when children with DLD make group decisions they successfully use non-verbal behaviours to do so, in addition to using verbal utterances at the same rate as children without DLD [41]. This therefore suggests poor cooperation skills affect only some children with DLD and, as with other areas of social competence in children with DLD, there are individual differences here.

Victimisation
Only four studies included measures of victimisation [11,29,31,50]. There is evidence showing children with DLD have an elevated risk of being victimised by their peers at age 7-8 [50] and at age 11 [31]. The study by   [11], however, finds no difference in levels of victimisation between children with and without DLD at age 6-10. The study by Conti-Ramsden and Botting (2004) [31] includes a large sample size of two-hundred children with DLD. These more reliable findings indicate there is an increased risk of victimisation among children with DLD. The discrepancy in the present findings on victimisation in children with DLD may result from differences in measurement techniques, with self-report measures being using in some studies [31,50] and direct observation being used in others [11]. The peers of children with DLD are possibly less likely to engage in bullying while they are being observed by an adult, and therefore self-report measures might more accurately portray levels of victimisation of children with DLD.

Discussion
This systematic review with narrative synthesis sought to refine our understanding of the nature of peer interaction strengths and difficulties shown by children with DLD. Of the studies reviewed, the heterogeneity of skills domains studied, and range of measures used means synthesising findings across a disparate literature is complex. The quality appraisal found most studies (71%) had medium scientific quality. Few studies (18%) had high scientific quality. In this review, the studies with the highest scientific quality are mostly those using questionnaires which measure overall peer competence, because these methods are described in enough detail to be replicated. However, findings from these types of studies provide minimal details into the specific nature of the peer interaction strengths and difficulties experienced by children with DLD. This review has revealed a dearth of research with high scientific quality investigating specific aspects of peer interaction in children with DLD. Nevertheless, those lower quality studies measuring distinct peer interaction characteristics are valuable in providing a basis from which to conduct further focused research on the social development of children with DLD. The available evidence points to the idea that primary school children with DLD struggle to access play and have poor conflict resolution skills. The outcomes of research into the play behaviour and discourse characteristics of these children are less clear but they raise some interesting questions and have the potential to form the foundation for a fascinating new area of research.
Children with DLD are a heterogenous group in terms of the nature and severity of their language difficulties [54]. It is clear from the available evidence that children with DLD are also heterogeneous in their social skill levels. While this review has found studies are consistent in reporting elevated levels of peer problems in primary school children with DLD [10,28,31,44,45,48], not all children with DLD seem to experience difficulties interacting with peers or making friends [7,29,42]. The reasons some children with DLD demonstrate more competent peer interaction skills than others are unclear. In the study by   [29] there was not a direct relationship between language disorder severity and language profile, suggesting there are additional factors which contribute to overall social skills when children have DLD. Indeed, there are a wide range of variables which might influence the children's social skills. For example, their level of pragmatic skill, such as turn-taking ability during conversation, and the characteristics of the children in their peer group, such as whether the other children in their class also have language difficulties. The study by   [7] finds evidence indicating children's level of prosocial behaviour and emotional symptoms influences their competence socialising with peers. Future research should further unpick the underlying factors protecting some children with DLD from experiencing peer interaction difficulties. If the underlying factors are malleable, it might be possible to develop targeted interventions to support children with DLD, who are experiencing peer problems, in developing better social skills.
In the current literature exploring the peer interactions of primary school children, females with DLD are underrepresented. A higher proportion of boys than girls are clinically diagnosed with DLD [55]. However, an epidemiological study has found the prevalence and severity of language disorder to be very similar across the sexes, with a sex ratio (male:female) of 1. 22:1 [1]. The large discrepancy in numbers of girls and boys included in current studies cannot therefore be justified by prevalence rates across sexes. Caution should therefore be taken when generalising current findings relating to the peer interaction skills of primary school children with DLD to girls because the available evidence is more representative of boys than girls.
Almost half of the studies included in the present review made no mention of whether the children in their sample experienced any co-occurring conditions. The studies included in this review were of children who had been selected for the research due to their language difficulties. However, DLD is a highly heterogenous condition [54] with many children having comorbid diagnoses with attention deficit hyperactivity disorder (ADHD) [56] and emotional behavioural difficulties (EBDs) [57]. Research has shown children with ADHD and EBDs are at risk of experiencing social skill difficulties [58,59]. It is therefore important to take comorbid conditions into account when exploring peer interaction in children with DLD. Going forward, it will be useful for future researchers to acknowledge this comorbidity and include more detail about the children represented in their samples.
The studies reviewed here have investigated the peer interaction strengths and difficulties experienced by children with DLD, but there has been more of a focus on difficulties. One reason we currently have little knowledge of the peer interaction strengths experienced by children with DLD may be that the focus of the available studies has not been on the children's own perceptions of their peer interactions or friendships. While two of the included studies use self-reports to investigate victimisation, the questionnaires employed do not offer children the opportunity to comment on positive features of social interactions [31,50]. One of the included studies used sociometric ratings to measure friendships in children with DLD [38]. The children with DLD and their peers were asked to list their three best friends which gave a measure of the number of reciprocal friendships the children have. It is encouraging to see child-centric methods employed. However, the current findings offer little insight into the perception children with DLD have of their relationships with their peers. Future research could benefit from the use of self-reports that allow children with DLD to report their strengths with regards to peer interactions. One way to do this is to employ qualitative methods. This will come with unique challenges, given the difficulties children with DLD will have in engaging with a verbal discussion. However, arts-based qualitative methods, such as Photovoice may prove feasible for this type of investigation [60]. Furthermore, guidelines on conducting qualitative research with those who have low language abilities, such as those with aphasia, can guide future researchers in this area [61,62]. A more rounded understanding of the peer interaction skills of children with DLD will better enable teachers and Speech and Language Therapists to build on the children's pre-existing strengths to support their social development.
It can be seen from this review that the peer interactions of children with DLD have been studied both at a macro and micro level and a wide range of different skills domains have been explored. Studies which look at the social interactions of children with DLD in close detail, such as DeKroon et al.'s (2002) [34] study of pretend play and Grove et al.'s (1993) [41] study of joint decision-making, provide exciting avenues of new research because they lay the groundworks from which to base future research projects. Overall, however, research investigating the peer interactions of children with DLD is sparse. Much more work needs to be done to uncover the underlying reasons for the peer problems so far observed in children with DLD [10,28,31,44,45,48].
The most reasonable next step for research in this area might be to combine different skills domains. Play behaviour among children with DLD, for example, remains to be fully explored. While there is some evidence showing children with DLD frequently engage in non-play during playtime [51] and show high levels of social withdrawal [11], the underlying reasons for such withdrawal behaviour are yet unknown and this limits our understanding of the underlying reasons for peer problems in children with DLD. It could be that children with DLD withdraw from social situations because they are unsure how to access peer play [33,46,47]. Studies linking access behaviour to questions regarding social withdrawal have not yet been conducted in primary school children with DLD. Overall, current literature does not provide a coherent picture of the peer interaction skills of children with DLD. Many studies, particularly older studies using observational techniques, investigate very specific aspects of these children's peer interaction skills. There is little unity between these studies and a lack of replication of findings because each pose different research questions. In contrast, multiple studies from the past two decades investigate peer interaction skills more broadly and present replicable findings because they use standardised questionnaires. However, a limitation of studies using these questionnaire measures is that they provide minimal detail into underlying reasons for the children's peer problems. Moving forward, research in this area should now investigate the relationships between multiple skills domains to answer specific research questions using replicable research methods. This has started to happen in studies exploring conflict resolution [28,38,48] or victimisation [11,29,50] in combination with overall peer interaction skills. Future research could bridge other skills domains. For example, the same study could measure access behaviour in combination with discourse characteristics, overall peer competence and victimisation.
Another approach to understanding the underlying reasons for peer problems in children with DLD might be to build on methods used in previous studies which explore discourse in children with DLD [32,35,37,39,43,49,52] to find out more about the peer interactions taking place when children with DLD are engaged in play. Studies find children with DLD are able to engage in social pretend play [34] and are able to share referents with their peers [32], which begs the question of why children with DLD are being rated by their teachers and parents as having peer problems [10,28,31,44,45,48]. No studies investigating the discourse of children with DLD have been conducted within the last 10 years. With new technologies, such as the LENA system [63], it is now possible to record the speech of children with DLD in a non-intrusive way while they are engaged in play in naturalistic settings, such as the school playground. These technologies could provide new opportunities to further explore the strengths and difficulties children with DLD experience while they play with their peers.
One area where children with DLD may need support in order to develop good social skills is understanding and managing situations of peer conflict. Studies conducted so far find children with DLD have worse conflict resolution skills than children without DLD [28,30,48]. The findings from the studies included in this review show children with DLD more frequently select conflict management styles which are considered "low-level" including no response, physical retaliation, and submission [28,48]. These strategies place less reliance on verbal skills than so-called higher-level responses, such as asking for clarification. Some have suggested children with DLD rely more heavily on non-verbal strategies to overcome conflict situations as a coping strategy to avoid negotiation [48]. Therefore, poor conflict resolution skills in children with DLD could be a direct result of their low language abilities. However, it is unclear whether the difficulties children with DLD have in managing conflicts result entirely from their expressive and receptive language difficulties or whether they also have a difficulty understanding social situations involving conflicts. Campbell and Skarakis-Doyle (2011) have suggested children with DLD have a less nuanced understanding of conflict resolution strategies and goals than children without DLD [30]. It will be important for future research to address this in order to develop appropriate interventions to support children with DLD. If children with DLD lack a complete understanding of social situations involving conflicts, it will be important for clinicians to address their social awareness and not merely their language skills.
There is evidence to suggest that children with DLD have an elevated risk of being victimised by their peers [31,50]. If this is the case, it will be important for future studies to explore the reasons children with DLD are victimised. One study found poor expressive language skills consistently correlated with victimisation scores, which suggests the skill with which children can communicate their ideas to their peers might influence their risk of being bullied [31]. Another possible risk factor, as shown by a study of victimisation in older children with DLD, is competence in understanding one's own basic emotions [64]. Possibly, the ability to understand one's own emotions enables one to more competently mask their negative feelings from a bullying peer, thus preventing the bully from gaining a sense of power, which reduces the risk of further victimisation [64,65]. While a small number of studies investigated the conflict resolution skills of primary school children with DLD, no studies have yet looked at victimisation in tandem with conflict resolution skills. This may therefore open another promising new area of research. It is possible that children with DLD with more advanced conflict resolution abilities are less likely to be victimised compared to other children with DLD. Research shows children who are bullied during childhood have an elevated risk of developing psychiatric problems, such as generalised anxiety and panic disorder, in later life [66]. Indeed, new research shows adolescents with a history of DLD who have experienced bullying have an increased risk of developing internalising symptoms [67]. Protecting children from bullying is therefore important. By understanding why language difficulties increase the risk a child will be victimised, it might be possible to develop appropriate strategies which schools could use to minimise bullying.
While conducting this systematic review, there were certain papers which could not be included due to restraints on age of participating children. Only papers assessing the social interaction skills of children aged 4-11 could be included. Some papers assessed the social interaction skills of children older than, but also including, this age range. If a separate analysis was not provided for children above the age of 11, these papers could not be included, thus certain papers with potentially interesting findings were not included in this review. While a review of literature relating to social development in adolescents with DLD exists [68], to our knowledge, no review has yet focused specifically on the primary school years. It was therefore essential to place restrictions on the inclusionary criteria in this way in order to review what is currently known about the peer interaction skills of children with DLD between the ages of 4 and 11.
It is difficult to conduct a comprehensive review relating to DLD, since many different terms have previously been used to describe the condition [69]. The inclusionary criteria relating to the term DLD was kept broad, to try to include as many variations of the term as possible. Papers could be included if disordered language was identified either through formal diagnosis by a speech and language practitioner, or through language assessment by a researcher. Papers could be included if the sample of children fulfilled all exclusionary criteria and were given a diagnosis of a wide range of possible terms (detailed in our methods section), to reflect historic changes in terminology [69]. Despite these efforts, it is possible that certain papers investigating the social interaction skills of children who would now be considered to have DLD were missed.

Conclusions
This systematic review has shown that research exploring the peer interactions of children with DLD does exist, but the available literature is disparate in terms of the skills domains being explored. Studies using questionnaire methods generally find children with DLD have a higher prevalence of peer problems than children without DLD. Studies using direct observation, such as those measuring children's behaviour on the playground or discourse during peer interactions in the lab, provide tentative clues to the underlying reasons for these peer problems, although the results from these studies are highly varied and the relationship between DLD and social competence with peers appears to be highly complex. There is a need for the replication of the findings from these observational studies which tend to use small sample sizes. Future studies could take a more holistic approach by linking together different skills domains within the broad construct of peer interactions. Data is based on more than one situation and has been studied fairly systematically but not there is room for improvement

Appendix B
Data is based on a suitable range of informants and situations and/or the topic has been studied systematically and comprehensively within the specified population or situation