Linguistic and Communicative Diversity in Inclusive Settings—Effects, Challenges, and Opportunities

Abstract

Diversity is understood as a significant advantage for inclusive learning. However, the questions of whether linguistic and communicative diversity in the classroom is truly beneficial and whether mutually linguistic support is effective between students with and without language impairments have not yet been answered comprehensively. Therefore, within the research project ‘LINUS—Learners in Inclusive Settings’, interactions between children (n = 48; aged 6 y 7 m to 7 y 7 m) in dyads in primary school were videotaped and micro-sequentially analyzed. The pairs of children were engaged in a 15 min playful framework within a quasi-experimental design. Based on quantitative language corpus analyses, the occurrence of linguistic elements was compared between pairs of children with (a) typical age-appropriate language acquisition, (b) developmental language disorders, and (c) heterogeneous settings with respect to backgrounds of language acquisition. Additionally, quantitative coding was used to analyze how often linguistic elements were productively imitated from one child to the other in different peer settings. The results show that there is a higher potential for mutually motivated communication and more resources for linguistic support in mixed dyads. From this study, specific principles for teachers and students can be drawn to foster linguistic support between children in the context of diversity.

1. Introduction

In inclusive linguistic-communicative educational processes, it is assumed that diversity proves to be an advantage. Based on this assumption, this article first contextualizes diversity as a construct in the context of teaching and learning and then presents theoretical and empirical references.

1.1. Classification

There is a hypothesis that children and adolescents continuously engage in interactive cooperative alignment, providing each other with learning stimuli from different experiential backgrounds (Hinz & Boban, 2022). They mutually relate to each other with their individual intentions, goals, and desires in the sense of intersubjective learning (Lailach-Hennrich, 2011). Learning in groups with students of different starting points has gradually developed through various convictions (Sliwka, 2012; Wacker, 2020). While differences were perceived as disturbances in the paradigm of teaching and learning in homogeneous groups, and supposed uniformity was considered effective, the perspective of heterogeneity already acknowledged the differences among students and the corresponding stimulating diversity (Sliwka, 2012; Wacker, 2020). However, it is only with the appreciation and inclusion of differences as advantageous resources for joint teaching and learning that the characteristics of otherness, strangeness, and impairment can be utilized in the sense of diversity (Gramelt, 2020; Stitzinger, 2021, 2019; Walgenbach, 2021). The notion of diversity emphasizes “the multiplicity, overlapping and crossing between sources of human variability” (Dietz, 2009, p. 58). Therefore, the research project ‘LINUS—Learners in Inclusive Settings’ at Leibniz University Hannover aimed to investigate the extent to which the hypothesis of beneficial diversity can be applied to the interaction behavior of students with heterogeneous linguistic-communicative backgrounds (Stitzinger, 2024, 2023, 2021).

1.2. Theoretical Background

The guiding principles of diversity in inclusive education are based on concepts of relational language acquisition theories. The dyad, characterized by the reciprocal interactive relationship between subjects (Lüdtke, 2016), is viewed in relational theories as a developmental space that is particularly conducive to cultural learning and, importantly, to the acquisition of verbal language (Lüdtke, 2012a). Within the psychobiological dyad, with its child anticipation and the dualism of mirror mechanisms, the child is perpetually dependent on an emotionally affirming, communicative other for the maturation of its neural and linguistic development (Lüdtke, 2012a). Therefore, the acquisition of linguistic meaning requires an emotionally supportive, didactic context (Lüdtke, 2012b). Hence, peer interactions where children are in emotionally meaningful relationships play a significant role in language acquisition (Stitzinger, 2021).

Moreover, in reference to interactionist language acquisition theories, linguistic-communicative coordination between subjects also occurs with shared attention to the context of action (joint attention) (Papoušek, 2008; Tomasello, 2020). Accordingly, the child needs a communicative counterpart within the context of meaningful objects (Lüdtke & Stitzinger, 2015). Triangulation between interaction partners and real objects assigns an important language-constitutive to these reference objects (Tomasello, 2020). In the context of diversity-sensitive language didactic considerations, the focus is on joint interactive language use to break down linguistic-communicative learning barriers and build competence-enhancing language experiences (Lüdtke & Stitzinger, 2017).

Additionally, processes of peer interaction can be broadly explained by constructivist theories. The individual perceives and processes their surrounding environment by actively engaging with their world of experience and life (Kricke & Reich, 2015; Reich, 2010). Interactionist constructivism points to the aspects of construction, reconstruction, and deconstruction (Reich, 2012). In the linguistic-communicative area, construction means that children constantly experiment with and modify their language actions within their own creative and innovative possibilities. Reconstruction of linguistic-communicative processes occurs as children discover, adopt, apply, and repeat linguistic-communicative structures, restructuring these as their own constructions. Finally, the process leads to deconstruction, as children identify and reorder incomplete and unforeseen structures and patterns in interaction with other children. Through critical observations, they reveal divergent phenomena (Stitzinger, 2021, 2018).

1.3. Empirical Background

In the context of linguistically heterogeneous group constellations, effects on language acquisition and language use among children have primarily been investigated in early childhood education. Various international research findings highlight the effectiveness of peer-supported learning in the context of preventive language education for children with different linguistic-communicative backgrounds (Justice et al., 2014). For example, in Germany, peer interactions between children growing up with Turkish as their first language and German as a second language were observed, and the children’s language skills in German were compared in a pre-post control group design with a follow-up phase (Licandro, 2016). The child pairs were selected such that one child had slightly more advanced German proficiency. The results demonstrated an increase in the storytelling abilities of the multilingual children. Interestingly, it was not only the children with more challenging language access who developed, but also those who served as language models (Licandro, 2016). However, other studies have identified disruptions in peer interaction due to the children’s difficulties with linguistic-communicative expression (Albers, 2013). Additionally, it has been found that children who have complex communication needs (CCN) tend to prefer adults as communication partners and are sometimes ignored by their peers (Biggs & Snodgrass, 2020). Teachers, speech-language therapists, and children with CCN face particular challenges, as developing positive peer relationships in inclusive classroom settings can be difficult. Research has shown that with the deployment and training of paraeducators, the interaction of students with CCN with their peers can be successfully supported (McCarty & Light, 2022).

In Germany, studies on learning in heterogeneous groups in schools under conditions of language and communication impairments mainly focus on the learning development and environment of the students (Kocaj et al., 2014). Longitudinal studies from 1st to 9th grade have contrasted academic performance during transitions from special schools for speech, language, and communication needs to regular schools and vice versa (Spreer et al., 2018; Theisel et al., 2021). It was found that primarily children with high cognitive abilities benefited from the transition to inclusive schools. Conversely, children who remained in or moved to special needs schools had problems adequately improving their literacy and math performance in an age-appropriate manner. The inclusive system was beneficial for students with language and communication impairments only if they had sufficient cognitive processing capabilities. However, regardless of the type of school, students felt well integrated and experienced little impairment in their verbal communication by the end of their school education (Theisel et al., 2021). Similarly, the results of an evaluation study at 24 schools in Germany from 1st to 9th grade showed that inclusive schooling did not have a detrimental effect on the literacy development of children with language and communication needs. However, math performance and emotional aspects need to be more supportively focused on in the inclusive setting (Mahlau, 2023).

In the studies exemplified, positive effects within the framework of prevention for children with language acquisition difficulties and intervention for children with developmental language disorders (DLD) were predominantly found in early childhood education. Although there are research findings on the effectiveness of inclusive settings concerning language and communication needs in the school sector, the potential of peer learning has rarely been investigated in this context. Therefore, for the first time in the context of foundational research, the present study demonstrates, through language corpus analyses, that the utterances of children with different linguistic-communicative backgrounds vary depending on group constellations. This provides a theoretical clarification that inclusive group settings offer an advantage over homogeneously segregated groups of schoolchildren with linguistic-communicative impairments. From this, insights for further research as well as practical implementations can be derived.

2. Materials and Methods

In the LINUS research project, the aim was to further investigate theoretical assumptions and empirical evidence. For this purpose, a quasi-experimental design with sequence analysis was developed. The sample was selected through targeted case selection based on specific criteria (Stitzinger, 2021, 2023, 2024).

2.1. Objectives

Within the framework of the research project, the linguistic utterances between child pairs in homogeneous segregated and inclusive settings were investigated. The following scientific objectives were the focus (Stitzinger, 2023, 2024):

• Which language levels do children with typical age-appropriate language development (TLD) and children with developmental language disorder (DLD) have difficulties with in their linguistic utterances across different pair constellations?
• To what extent do children with TLD and DLD produce phones, words, and sentences across different pair constellations?
• How is the dialog behavior characterized in terms of the frequency of turn taking, and how often do children with DLD imitate linguistic markers from their conversation partners’ models across different pair constellations?

Based on these questions, the comparison of homogeneous and mixed pair constellations aims to identify beneficial effects and potentials. There is the hypothesis that children with DLD in inclusive settings produce more phones, words per sentence, and linguistic model adoptions compared to children with DLD in settings where all participants have DLD.

2.2. Data Collection and Analysis

To methodologically control the comparison of the different groups, the team established similar conditions in all groups and focused on linguistic markers. The data collection occurred in comparable 15 min playful settings within a thematic material field involving primary students (Stitzinger, 2021, 2023, 2024). In line with a quasi-field experiment (Brosius et al., 2016; Hertel et al., 2010; Reichardt, 2019; Singh, 2021), all selected child pairs were given a joint planning and construction task. The students received a child-appropriate task via an audio device, which encouraged linguistic-communicative collaboration and the creation of a common product. The task, set within the context of a pirate story, involved the children crafting a pirate ship from various everyday materials. Each child pair worked on the task in a separate and familiar group room. During this time, the children’s interactions and dialogs were videotaped without the presence of any additional persons (Stitzinger, 2021, 2023, 2024).

The linguistic components of the audiovisual recordings were later transcribed linguistically. The team applied the Conversation Analytic Transcription System 2 (GAT 2) in its basic form of minimal transcription, with slight modifications (Selting et al., 2009). Based on the transcriptions, the children’s linguistic utterances were marked according to selected linguistic criteria and examined through conversational analysis (Stitzinger, 2021, 2023, 2024). The analysis focused on linguistic aspects including phonetics, phonology, lexicon, semantics, morphology, and syntax (See Appendix A for more details). Through quantitative language corpus analyses (Tognini-Bonelli, 2001; Tonkin, 2016), correct and incorrect linguistic markers per child were filtered. Additionally, the team analyzed how much a child spoke by counting the number of phones (the smallest distinguishable phonetic units in the speech continuum). The team also counted the spoken words (lexical units which were given a meaning) and sentences (syntactic units comprising one or more words). This allowed for the correlation of the number of words with the formed sentences. The dialog behavior of the child pairs was analyzed based on turns (the alternation of speaking and listening through turn taking) (Levinson, 2016). Furthermore, the effects of mutual modeling and potential imitation by conversation partners in the children’s dialogs were assessed. This involved identifying occurrences in the data where the child had previously marked linguistic markers incorrectly, but the conversation partner offered them correctly as a model (See Appendix A for more details.). Both target language-appropriate and non-target language-appropriate linguistic model adoptions by the conversation partner were evaluated (Stitzinger, 2021, 2023, 2024).

It was expected that the comparison of linguistic abilities, number of spoken phones and words per sentence, turns conducted in the dialogs, and the adoption of linguistic models could provide initial theoretical insights into the advantages of inclusive group constellations. Comparisons of other types of utterances in the area of paraverbal or nonverbal communication were not made here, as these do not generally occur prominently.

2.3. Case Selection

The case selection was conducted at a primary school in Germany with integrated classes for students with language and communication needs (Stitzinger, 2021, 2023, 2024). The sample size was intentionally kept small to allow for both a quantitative language corpus analysis presented here and a qualitative conversation analysis, which is not the subject of this article. The children (n = 48; aged 6 y 7 m to 7 y 7 m; 39.4% female, 60.4% male) who participated in the study were familiar with one another from various pedagogical settings in the first grade, either in an inclusive classroom or a classic pull-out model. The child pairs were selected by the teachers based on the following criteria:

• 8 child pairs (9 female, 7 male), both having typical age-appropriate language development;
• 8 child pairs (4 female, 12 male), both having linguistic impairments in multiple language components as a genuine developmental language disorder (primary developmental language disorder, not a consequence of other impairments);
• 8 child pairs (6 female, 10 male), in heterogeneous constellations concerning typical age-appropriate language development and genuine developmental language disorder (Stitzinger, 2021, 2023, 2024).

To select the child pairs, it was required that the children were familiar with each other from shared classroom settings. Moreover, constellations exhibiting traits of extraordinary linguistic dominance or significantly suppressed language behavior (e.g., speech inhibition, speech anxiety, mutism) were not included in the sample. The biographical data of the students were collected through a questionnaire addressed to the parents, pseudonymized and assigned to the transcriptions (Stitzinger, 2021).

3. Results

The selected results of the research project presented in this article will be shown according to three analytical focal points. First, the configurations of linguistic difficulties within the respective child groups are exemplarily presented (Section 3.1). Then, the language corpus of the children in the different pairs is examined concerning phones, words, and words per sentence (Section 3.2). Finally, the dialog behavior in the dyads is analyzed by illustrating how frequently turn taking occurred in the dialogs and to what extent the children in the different settings imitated language models from one another (Section 3.3).

3.1. Linguistic Difficulties

It is not surprising that the language use of students with TLD predominantly conformed to linguistic norms. In the analyses of the dyads with children with DLD, however, there were more linguistic aspects that were marked as not target-language appropriate in the transcripts. Concerning the different language components, the children with TLD and the children with DLD exhibited different language profiles (Figure 1).

For children with TLD in pairs, both having typical language development, almost no linguistic difficulties were observed in their utterances overall (M = 0.0–0.8). Similarly, only minor deviations from the norm were found in the language productions of children with TLD in the inclusive setting in the components of phonetics, phonology, and semantics (M = 0.0–0.5). However, in lexical, morphological, and syntactic components, the utterances deviated slightly more from the expected norm (M = 2.5–11.8). In contrast, children with DLD in pairs, both having developmental language disorders, showed linguistic difficulties across all language components (M = 2.1–20.6). A similar tendency was observed for children with DLD in the inclusive setting in phonetic, lexical, semantic, morphological, and syntactic components, although the deviations from the norm were marginally lower (M = 1.1–15.2). However, a significantly higher mean value was noted on the phonological level (M = 47.8).

3.2. Phones, Words, and Sentences

The linguistic utterance in terms how much a child spoke was measured by the phones, produced during the joint task. When comparing the number of phones in the different settings, clear differences and similarities were observed (Figure 2). The statistical dispersion (see boxes in the chart) of the number of phones was similar within the constellation of child groups, the homogeneously formed groups (TLD, DLD) and the mixed pair constellation (TLD Inclusive/DLD Inclusive). Larger interquartile ranges (see boxes in the chart) were observed in the mixed child pairs compared to the homogeneous constellations.

The detailed count of phones showed a similar range (see whiskers in the chart) with almost identical maximum values for children with TLD in pairs, both having typical language development, and children with DLD in mixed pairs (TLD: max = 3449; DLD Inclusive: max = 3421). On the other hand, the maximum value was slightly lower for the child group with TLD in the mixed setting (TLD Inclusive: max = 2896), but still higher than for the children in pairs, both having DLD. In this child group, the lowest maximum value of the recorded phones was present (DLD: max = 2340).

In comparing the values of phones with the values of spoken words, correspondences became apparent (Figure 3). Both the interquartile ranges (see boxes in the chart) and the ranges (see whiskers in the chart) showed similar tendencies between the number of phones and words. The maximum values of spoken words by children with TLD and DLD in the different constellations exhibited a comparable gradation to the graph of phones (TLD: max = 885; DLD: max = 647; TLD Inclusive: max = 732; DLD Inclusive: max = 934).

By correlating the spoken words with the sentences formulated during the conversation, the length of the sentences could be determined. When calculating the words per sentence (Figure 4), child pairs, both having DLD, on average, produced the fewest words within sentence structures compared to the other pair constellations (TLD: M = 4.59; DLD: M = 3.85; TLD Inclusive: M = 4.58; DLD Inclusive: M = 4.02). This means that children with DLD overall formed shorter sentences than children with TLD. However, children with DLD were capable of slightly longer sentences when they were paired with conversation partners with TLD.

3.3. Turn Taking and Effects of Modeling

In the context of intersubjective linguistic-communicative actions, results emerged that provided insights into the unique situation of child pairs, both having DLD. The turn taking (alternation of speaking and listening) within the dialogs of the child pairs showed the highest frequency of turns (Figure 5) in the dyads where both children had DLD (DLD: M = 135.9). In contrast, lower average frequencies in turn taking were observed in the other pair constellations (TLD: M = 127.9; TLD Inclusive/DLD Inclusive: M = 125.4).

Considering the partially higher speech output (see number of phones, words, and words per sentence) of children with TLD and children with DLD in the inclusive setting in correlation with the lower turn taking frequency, it can be inferred that in these cases, mutual communication was conducted predominantly with longer utterances. Conversely, the higher turn taking frequency observed in child pairs, both having DLD, suggests that their interactions were characterized by shorter and more fragmentary linguistic utterances.

The analysis results indicate that reciprocal language modeling, both correct and incorrect, occurred in all child pair constellations. For example, children with TLD, alongside age-appropriate correct language use, also surprisingly conveyed unusual, colloquial expressions, which were imitated by their conversation partners.

In the context of linguistic-communicative diversity, the effects of modeling in children with DLD are particularly noteworthy. Correct models (Figure 6) were imitated to a significantly lesser extent when both children in the dyads had DLD. This effect was observed only in the phonological area (DLD: M = 0.19). In contrast, children with DLD in mixed constellations more frequently imitated correct models from their conversation partners with TLD. The correct imitations occurred on the phonological (DLD Inclusive: M = 0.38) and lexical (DLD Inclusive: M = 0.13) level.

In addition to correct language models, incorrect models that were imitated by children with DLD also emerged in the dyads (Figure 7). It is notable that children in pairs, both having DLD, particularly imitated incorrect models, especially on the level of phonology (DLD: M = 0.44), but also phonetics (DLD: M = 0.06), lexicon (DLD: M = 0.19), semantics (DLD: M = 0.13), and morphology (DLD: M = 0.13). In the mixed settings, children with DLD imitated significantly fewer not correct language models, in these cases only on the lexical level (DLD Inclusive: M = 0.13).

Overall, the mean values of language model imitation appear to be very low. However, these must be considered in the context of the relatively short interaction time of 15 min within the quasi-experimental design.

4. Discussion

The study is to be understood as foundational research aimed at developing new theories, in which existing theories were experimentally tested. In this context, the limitations and research desiderata will be discussed.

4.1. Addressing the Research Questions

The results from the analyses of the linguistic competencies of the children who participated in the study revealed characteristics of linguistic features in both typical age-appropriate and impaired language acquisition. The differences between children with TLD and DLD became clear, although there was a slight tendency towards alignment between children with TLD and DLD in the mixed setting. In the inclusive setting, children with TLD exhibited slightly more linguistic anomalies than in homogeneous constellations, and children with DLD showed somewhat fewer linguistic difficulties than children in pairs both having DLD. Indicators of language impairments were primarily found in the components of phonology, lexicon, morphology, and syntax. Overall, the results broadly confirm theories of co-constructive language learning with shared attention on the learning object (Tomasello, 2020; Kricke & Reich, 2015; Reich, 2010, 2012).

The language corpus analyses also showed that in the mixed setting, children with DLD had a slight advantage in terms of the production of phones, words, and sentences compared to children with DLD in homogeneous constellations. Here, too, a tendency towards positive intersubjective coordination emerged (Stitzinger, 2021; Lüdtke, 2012a, 2012b, 2016), although children with TLD in the mixed setting, despite high empathy towards their conversational partners, occasionally held back with their linguistic-communicative abilities, in contrast to other studies (Licandro, 2016).

The effects of linguistic modeling, in particular on the phonology and lexical level of children with TLD and successful imitations by children with DLD in the mixed pairs, appear noteworthy given the short experimental setup, as language acquisition ultimately occurs through a gradual, long-term process. It is also worth noting the negative transfer of incorrect language models in homogeneous constellations of children with DLD. Frequently occurring linguistic difficulties can thus lead to disruptions in linguistic peer learning (Albers, 2013).

Overall, the study’s results encourage viewing inclusive settings as an opportunity for beneficial linguistic-communicative learning experiences. Mixed groups of children, where both age-appropriate and impaired linguistic accessibilities are present show a higher potential for mutually motivated communication and language model resources compared to homogeneous groups.

4.2. Limitations and Suggestions for Further Research

The sometimes large ranges and variabilities in the measurements indicate that not all children were able to benefit from the advantageous constellations. Thus, the present study is not robust proof for the generalization of the results to the overall population of students with and without linguistic-communicative impairments, nor to real, non-experimental classroom settings. This also explains the currently inconclusive research status in the context of intervention studies (Spreer et al., 2018; Theisel et al., 2021; Mahlau, 2023). It must be emphasized that the results are specific to the selected samples and are not global. For instance, different results may be obtained in varying educational or socio-economic contexts. Similarly, the results cannot be readily transferred to or generalized for non-German languages.

The extent to which constellations of inclusive settings can indeed be utilized to promote language and learning will depend multifactorially on other influencing conditions. To explore this, further targeted field studies under real classroom conditions should be conducted. To identify possible influencing factors, various task formats, school subjects, and group sizes should be considered. Additionally, alongside identifying abilities and modeling on linguistic levels, comparisons in the area of paraverbal and nonverbal utterances of children in different group settings, such as voice modulation, eye contact or facial expressions, are also possible. This could incorporate the holistic nature of language and communication. However, incorporating holistic linguistic and communicative variabilities is significantly more challenging to control in a comparative study.

5. Conclusions

To successfully address the challenges of inclusive education systems for students and teachers, everyday school life must be supported by professional and specific guidance (Lüdtke & Stitzinger, 2017; Stitzinger, 2021, 2023, 2024). Teachers must primarily support children with linguistic-communicative impairments by first building the children’s self-confidence in adopting correct language models from other children. At the same time, children with strong language skills should be made aware that they can have a positive influence as language models for other children (Lüdtke & Stitzinger, 2017; Stitzinger, 2021, 2023, 2024).

In inclusive education, advantageous peer relationships must fundamentally be considered, and suitable situations for peer interactions must be utilized in order to truly harness the potential of linguistic-communicative diversity and variety. It is crucial that through the balance and complement of different children’s competencies, mutual benefits for all can unfold (Lüdtke & Stitzinger, 2017; Stitzinger, 2021, 2023, 2024).