Assessment of first language adds important information to the diagnosis of language disorders in multilingual children

Vienna is a melting pot of different languages and cultures. Approximately 60% of all Viennese children grow up with a first language (L1) other than German [1]. In multilingual settings, there are many factors which influence language proficiency like cross-language influence, or quantity and quality of language input in L1 and L2 (second language) [2]. Thus, even healthy children vary in their speed of L2 acquisition, that is, in their L2 proficiency at a given time. A lower L2 proficiency compared to monolingual children often leads professionals to consider a language impairment. Superficially, L2 speech patterns in early stages of acquisition may resemble patterns observed in impaired monolingual language development [3]. Lower L2 proficiency in multilingual children may hence be a normal phenomenon in the framework of L2 acquisition. However, for some children, it might indeed be traced back to a language disorder (LD [4]).¹ Because of the strong variation between children (e.g., L1, number of languages regularly spoken, [socio-]psychological factors) and the resultant complexity in the evaluation of the children’s language development, a comprehensive analysis including a language assessment in the children’s L1 can add important information (see also [5]).

An LD might occur in otherwise healthy children, or in children with primary disease. Depending on the type of disease, a comorbid language disorder (CLD) can be expected, for example, in the case of autism spectrum disorder (ASD, ICD-10:F84, [6]). In other cases, the primary disease is not directly associated with an LD, for example in some metabolic diseases. However, children with primary diseases must often manage a challenging daily life, so that the primary disease can be paramount. In case their primary disease is accompanied by a CLD, the children additionally have to compensate for their linguistic deficits in communication.²

An evaluation of a suspected LD in multilingual children is important. A falsely ascribed LD might have far-reaching consequences for the child’s self-esteem and academic performance, while children with an actual LD need speech therapy. Unfortunately, in our clinical experience in Austria, clinical routines and psychological practice often only involve the assessment of the L2 German according to monolingual norms. However, the mere focus on the L2, and, specifically, on monolinguals as reference can lead to an overdiagnosis of LD (see also, e.g., [7]). Therefore, many studies on multilingual children with a focus on differentiating between typical development (TD) and LD suggest the combination of L2 measures (e.g., nonword or sentence repetition tests) with a parent questionnaire about the L1 development [8]. This procedure is readily applied by professionals because they do not need any knowledge of the L1. For children without any observed sensory, motor, or neurological impairment who are able and willing to cooperate in the test procedure, this combination of L2 assessment and parent questionnaires might be sufficient to establish whether their language development is typical or not (ICD-10:F80 [20]). Generally, the validity of parent questionnaires is debatable. Even in cases of L1 development, parent questionnaires often lead to false assessments [9, 10]. In an L2 setting, many families are in a psychosocially burdening situation because of their low social status in the majority society, due to socioeconomic problems or sometimes to traumatic experiences related to their flight. In this situation, parents’ information about their children’s language development might be even less reliable. Many studies show that in order to speak of an LD, the disorder must occur in all languages and in all communication situations [3]. Therefore, testing in all languages must be seen as the gold standard. It allows for the classification of the language development of multilingual children in TD and LD.

In Vienna, multilingual children with a suspected LD can be examined in a specialized counselling hour. During this counselling hour, a linguistic evaluation of the children’s L1 and L2 development is conducted and interpreted based on information about the children’s individual multilingual situation. These results are combined with results of a psychological evaluation and integrated into a pediatric perspective.

The goal of the present study is to present an overview of the children who visited this counselling hour between 2013 and 2020. By looking at their language, medical, sociodemographic, and psychological parameters, we aim to provide a solid basis for diagnosis and thereby to recommend the best possible support.

The aim of the present study was to describe the group of multilingual children who are suspected of having an LD and are therefore sent to a specialized counselling hour. In addition to an overall description of the whole group of multilingual children who were examined, two groups of children were differentiated: one without primary disease (47% of all children) and one with primary disease (53% of all children).

As described above, a primary disease can often be directly or indirectly associated with CLD. In fact, our results show that a suspected LD was confirmed more often in children with primary disease (90% of these children) than in children without primary disease (38% of these children). This reflects the differences of the prevalence of LD described in the literature. For ICD-10:F80, a prevalence between 5–8% is indicated [22], whereas for children with primary disease, a prevalence of CLD of up to 20% is suggested [23]. Even though the suspected LD can be rejected for only 10% of children with primary disease, they have a wide spectrum of diseases, with some of them more strongly associated with LD than others. Since an LD is expected in many young children with ASD [6], the L1 evaluation only provides information about the pattern of the disorder [19]. Many further diseases are characterized by a high incidence of CLD (e.g., epilepsy [24]), so that the probability of confirming a CLD in the children’s L1 is high. In contrast, many somatic diseases like juvenile diabetes are not directly associated with LD. However, a general association of psychological disorders with juvenile diabetes is observed (e.g., depression, anxiety disorders, [25]), which might enhance the probability of CLD. Nevertheless, for 39% of children with somatic diseases in our study, the suspected LD could be rejected, as well as for about 20% of children with neuropediatric diseases, psychiatric diseases (except ASD and cognitive disorders), or hearing impairments. Thus, it is important to critically investigate the suspicion of an LD, despite a generally strong association between primary diseases and LD.

Significant predictors for the linguistic result for children without primary disease were heredity of ICD-10:F80, age at first word production, and age at first examination. Children without heredity have better chances of TD than children with known heredity, which is in line with the literature (e.g., [26]). Children who produce their first words later have a higher risk of having ICD-10:F80. This result was also expected, as typically developed children mostly produce first words around 12 months, whereas children with LD often produce first words later (e.g., [27]), despite large individual variation. Results further showed that the older the children were at their first examination, the more likely they were to receive the result TD. Older children are probably more often examined as a precaution, often ordered by the school. In contrast, if younger children are sent for language assessment, there might be a higher suspicion of an LD, often noticed by the parents themselves.

The overall distribution of linguistic results (TD, ICD-10:F80, CLD) was independent of L1 and gender. This reflects that “language development in children is remarkable for its regularity across individuals and different languages, even when the languages are quite diverse” [28]. Children’s L1s were distinctly heterogeneous. The most frequent L1s were Turkish, Bosnian/Croatian/Serbian, Arabic, Russian, and Romanian, which roughly reflects the overall distribution of non-German L1s in Vienna [1]. Typologically, these languages differ at various linguistic levels, leading to different cross-language influences on L2 German. This variety of L1–L2 patterns is further highlighted by the many L1s that were represented with sometimes as few as one child per language. In addition, 26% of the children examined in this study are regularly exposed to more than two languages. Although this increases the possibilities of cross-language interference (L1–L2–L3) and likely reduces the amount of input in each of the languages, the number of languages spoken did not predict the number of children referred to the counselling hour (nor did it predict the linguistic results, looking at the children without primary disease).

In Austria the majority language has a high value in the educational system. Because L2 German assessments are part of the obligatory school entrance tests [29], a large proportion of the children present at the counselling hour in temporal context with school entry. Our results show that these children are more often unwarrantedly suspected of having LD than younger children. A comprehensive language evaluation including an L1 evaluation might correct this misjudgement without further negative consequences.

To our opinion the common practice to focus on an L2 assessment cannot completely meet the children’s needs. An LD must be verified in each of the children’s languages. Many children in our study received the result TD on the basis of their L1 and their specific multilingual setting. Children with a L1 other than German represent 59% of all children between 3 and 5 years in Viennese day care [1]. Therefore, all languages should be recognized and valued (see also [30]). A mere assessment of the L2 is not sufficient, because many multilingual children are actually behind monolingual children in their L2 German (see, e.g., [22, p. 33]), which has also been confirmed by the evaluations in the counselling hour. However, this does not justify the diagnosis of an LD. Moreover, none of the investigated parameters, that is, neither the respective primary disease nor the various sociodemographic parameters, can explain the outcome of the linguistic result (TD, ICD-10:F80, CLD) by itself. These parameters increase the chances of a given result, but the linguistic evaluation, including an L1 assessment, cannot be replaced by this information. In addition to the classification of the language disorder, the linguistic assessment allows the description of patterns of the disorder and thereby the development of strategies for therapy. Whenever possible, native speakers should be integrated in order to analyze the child’s language together with linguists and the child’s parents. We assume that the inclusion of native speakers would indeed be possible at many locations. A high percentage of (university) students acquired other first languages than German and might therefore contribute to the L1 assessment process, as well as a high percent of day care staff. Moreover, children and their families might benefit from the integration of native speakers as a cultural broker.