Constructing a clientele in need: The field of German as a Second Language

Chapter 6 Constructing a clientele in need: The field of German as a Second Language Jana Gampera, Christoph Schroederb, Julia Schlaucha& Dorotheé Steinbockb aUniversity of Giessen bUniversity of Potsdam When studying multilingual children and adolescents, it is crucial to ask about the age of onset and/or length of exposure to the languages in question before classifying them as L2 learners or not. In our semi-systematic review of 138 papers from the field of research on German as a Second Language (GSL), we ask to what extent this distinction is drawn in the field. GSL is concerned with at least two fundamentally different groups. One group is multilingual children and adolescents who acquire German as one of their first languages or as their early L2. The other group is the first generation of newly immigrated children, who clearly acquire German as their second language. We reviewed a data sample of n=138 papers that were coded according to explicit and implicit features related to GSL speakers. With the help of relative frequencies and association analyses we show that GSL speakers are mainly conceptualized as “bior multilingual” and speakers with “languages other than German” in their repertoire with “insufficient German competence” experiencing “educational disadvantages” in schools. Common psycholinguistic L2-criteria such as Age of Onset (AoO) occur surprisingly rarely in our data. These findings suggest a deficit-oriented conceptualization of GSL speakers and reinforce tendencies of Othering. In light of these findings, we argue for a narrowing of the concept of GSL and a better disclosure of L2-learner-specific metadata. 1 Introduction We ask here to what extent “Othering” applies to studies in the field of German as a Second Language (Deutsch als Zweitsprache, henceforth GSL), with a focus Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock. 2025. Constructing a clientele in need: The field of German as a Second Language. In Artemis Alexiadou, Claudio Scarvaglieri, Christoph Schroeder & Heike Wiese (eds.), The construction of multilinguals as Others: Do we practice what we preach?, 103–133. Berlin: Language Science Press. DOI: 10.5281/ zenodo.17132449 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock on studies related to child and adolescent learners. We combine this with a second question, asking, how academic studies in the academic field of GSL conceptualize their “clientele” of learners. We conduct our research in the form of a systematic literature review. Our research questions originate in the observation that GSL is a rather vague concept, tending to take a deficient perspective on its clientele. This is related, in our view, to an unclear differentiation between the psycholinguistic notion of GSL in the frame of second language (henceforth L2) acquisition on the one hand, which has or should have (in our view) certain specific and empirically based implications. On the other hand, the academic field of GSL has a particularly applied tradition in Germany that differs from (established) second language acquisition (henceforth SLA) research in particular ways. Our chapter is structured as follows: In section 2, we substantiate our research question. In section 3, we develop and conduct our review. Section 4 discusses our findings. 2 Theoretical considerations To substantiate our research question, we first briefly introduce our understanding of multilingualism and second language acquisition (2.1), before turning to the academic field of GSL (2.2) and some initial qualitative observations (2.3), which lead us to point out certain dangers of “Othering”, which may originate in the mismatch between the scope of the academic field and the psycholinguistic notion of GSL. 2.1 Multilingualism, second language acquisition and age of onset Acquiring an L2 is one way an individual can become multilingual, and in the individual-oriented use, the term “multilingualism” is an umbrella term for various forms of language acquisition in the course of an individual’s life, as well as for linguistic practices and regulations in everyday life, in working life and institutions (cf. Grosjean 2010). Individuals become multilingual by acquiring more than one language. There is no doubt that the age of onset (AoO) and the length of exposure (LoE) to the language(s) are decisive for the course of the acquisition process, but also for its speed and to some extent for its success (cf. Gagarina et al. 2021, Grimm & Cristante 2022, Meisel 2009). Differences in AoO, then, are important indications of the (expected) qualitative course of acquisition. The acquisition of more than 104 6 Constructing a clientele in need: The field of German as a Second Language one language simultaneously as first languages is commonly called “bilingual first language acquisition”. The acquisition of a second language after the age of 3 is called “successive second language acquisition”. Speaking of German as an L2, this is where “German as a Second Language” begins.1Successive second language acquisition up to the age of 7 is summarized as “early childhood second language acquisition”. Finally, a distinction must be made between whether the language(s) are acquired as a child or as an adult. Adult acquisition of another language is considered to start with the onset of puberty (cf. Hyltenstam & Abrahamsson 2003). All in all, language development in German as L2 can be summarised well by saying that the earlier the acquisition of German begins, the more similar its course and dynamics are to monolingual first language acquisition. Work by Thoma & Tracy (2007) and Tracy (2008) makes it quite clear that children who begin acquiring German before the age of 3-4 go through the same stages of acquisition, at least in the area of syntactic development, as children who acquire German as their first language. A first qualitative break becomes visible at the age of onset into second language acquisition around age 4. In particular, the work of Haberzettl (2014) and Gagarina et al. (2021) show that children who begin second language acquisition at this age acquire the language differently than children for whom the process begins earlier (see also Grimm & Cristante 2022). These age-related differentiation of learner groups are no more than approximate guidelines for the connection of onset and course of L2 acquisition, but AoO is a crucial defining factor when it comes to L2 learner identification. However, there are no reliable figures for Germany on how many of the multilingual individuals can be classified as L2 learners. For example, there are no figuresavailable on the proportion of families who speak only German as the family language (=monolingual German language acquisition), or speak German and another language as family languages (=simultaneous bilingual acquisition). Childcare rates in kindergartens could provide a clue, although in official state reports only data on “migration background” is available,2but not linguistic data. Still, if we take 1This distinction is by no means uncontroversial. There are certainly approaches that place the beginning of successive second language acquisition (in contrast to bilingual first language acquisition) later, namely at the age of four or five (Meisel 2009, Schulz & Grimm 2019). Note that the fact that the exact formulation of the criteria might be controversial does not diminish the importance of clear-cut criteria. 2“Migration background” (Germ. Migrationshintergrund) is a sociodemographic characteristic used in demographic statistics in Germany in order to describe the migrant community in the country. According to the Federal Statistical Office in Germany, individuals have a migration background if they themselves or at least one of their parents were born with a non-German citizenship. See for an overview and critical evaluation Will (2020). 105 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock the start of kindergarten attendance as the start of the acquisition of German for children with a “migration background”, then, the (cautious) conclusions about the age at the start of acquisition are as follows: For the vast majority of children born in Germany with a migration background, we can assume bilingual first language acquisition or early second language acquisition.3 What does this say with regard to expected competencies in German? On the one hand, the idea that second language acquisition is completed when the speaker has mastered the second language at the same level as a monolingual speaker is not tenable: Bilingualism or multilingualism does not mean multiplied monolingualism, as the famous citation from Grosjean (1989) goes. At the same time, the variation in language use and competence among monolingual speakers is so great that a normative assumption of monolingual language use and competence is not possible (Shadrova et al. 2021). However, there is a growing body of research results which compare the academic language competencies of children and adolescents born in Germany with a migration background and/or with a family language other than or besides German with those of children and adolescents without a migration background or with a monolingual acquisition biography. These allow the following conclusion: The vast majority of children and adolescents born and raised in Germany with a family language other than or alongside German do not differ in their linguistic competencies from monolingual children and adolescents when it comes to academic language performance (see among others Haberzettl 2016, Ricart Brede 2020, Petersen 2014). If there was successive second language acquisition at all (and not bilingual first language acquisition), then this process reaches completion in primary school (Gagarina et al. 2021). At the same time, this does not necessarily have to mean that the linguistic competencies and practices of multilinguals are indistinguishable from the linguistic competencies or practices of monolinguals. However, potential support needs are no different from those faced by monolinguals and correlate primarily with social factors that limit access to the academic register of school (Reiss et al. 2019, Rauch 2019). If we summarise the discussion above for our research questions, two points become evident: First, in the study of multilingual children and adolescents, it 3According to the figures of the Federal Statistical Office, just over 20% of 0-3 year old children with a migration background attend kindergarten and the rate rises to over 80% for 3-6 year olds (with certain differences between the federal states), see Statistisches Bundesamt, figures for March 2020, at https://www.destatis.de/DE/Themen/GesellschaftUmwelt/Soziales/Kindertagesbetreuung/Tabellen/betreuungsquote-migration-unter6jahrennach-laendern.html;jsessionid=8F81E7EAE4F5065DB0917E1313242370.live742 [accessed 2024/04/09]. 106 6 Constructing a clientele in need: The field of German as a Second Language is essential to ask for biographical details concerning the exposure to the newly acquired language, before classifying them as L2 learners. On a minimal level, this is best done through inquiring about the age of onset of acquisition (AoO), and/or the length of exposure (LoE), possibly also the length of schooling (LoS) in the language. And second, for those we classify as “early second language learners”, we can assume that they stop being “second language learners” at the latest at primary school age. 2.2 The academic field of German as a Second Language The academic field of Deutsch als Zweitsprache (DaZ / GSL) is relatively new. It emerged from the period of the first “guest worker” immigration to Germany in the 1970s. Germany experienced an economic boom in the late 1960s, and the demand for labor exceeded domestic resources. As a result, workers from Southern Europe and North Africa were recruited in large numbers. Initially, these so-called “guest workers” were to stay only for a few years, but it soon became clear that the need for labor continued, and companies did not want to let their trained workers leave again. Consequently, the workers were invited to settle for good, together with their families. Research projects focussing on the linguistic aspects of this development (initially carried out under the name of Deutsch als Fremdsprache (DaF), that is, “German as a foreign language”, and only from the 1980s on “DaZ”, cf. Baur 2001) concentrated, on the one hand, on untutored second language acquisition of (mostly young) adults (Ahrenholz & Rost-Roth 2021). On the other hand, a strong applied orientation developed, which turned to the socially important question of how the school system dealt or should deal with students whose knowledge of German was initially too low to be able to follow the content of instruction (Menk 2000). In this line of research, a focal topic of GSL research became the particular school register of academic German (Bildungssprache, cf. Lange 2020) that is presumed to open doors for social participation. Almost half a century has passed since the emergence of GSL as an academic field. On the one hand, the original field of activity of GSL has changed considerably. The study of adult second language acquisition has diversified considerably because adult immigration has changed greatly in form and scope. After all, the new German Immigration Act of 2005 has introduced institutionalized language teaching in the form of integration courses (cf. Gamper et al. 2021). At the same time, the former “children of foreigners” (Germ. Ausländerkinder) are now adults “with a migration background”; they are parents or grandparents; their children and children’s children were born in Germany, go through German educational 107 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock institutions and, as long as at least one parent was born abroad with a foreign citizenship, are again children or young people “with a migration background”. Let us underline the fact that a history of family migration does not tell us anything about whether German is an L1 or L2, especially for descendants of the first generation of immigrants (see section 2.1). On the other hand, immigration of children and adolescents continues. Particularly with the large numbers of refugee migration in recent years, first in 2015 following the Syrian civil war, and then after the Russian attack on Ukraine in February 2022, questions about forms of schooling for newcomers with little or no knowledge of German, studies on the second language acquisition of this group at school, and on the conditions and possibilities for support have once again come to the fore (Gamper et al. 2021). Thus, the academic field of GSL in Germany deals with at least two fundamentally different groups: On the one hand, some multilingual children and adolescents are born in Germany and acquire German as one of their first languages or as their early L2, and, on the other hand, the first generation of immigrants, who acquire German as their L2. It is crucially important to distinguish between these two groups. The second group needs support in learning German, but the first group does not, at least not in the same way. Support here means that newly immigrated children and adolescents need help to acquire basic knowledge of German to gain access to the school-specific registers. Secondly, lumping the two groups together is meaningless if one wants to generalize the dynamics of second language acquisition. And thirdly, not distinguishing between the groups may have the consequence that the first group slips into a “problem group” categorization. This is not only wrong but it also has discriminating tendencies, in other words, tendencies of “Othering”.4 So far, we have seen that the umbrella term GSL is ambiguous. On the one hand, it covers learners who are new immigrants and meet the criterion of having started L2 acquisition after the age of 3. However, due to the historical development of the field of GSL, it also covers learners for whom it is unclear whether German is a (second) L1 or an L2. This conceptual vagueness5has consequences for how L2 learners are conceptualized and what attributions are made to this (unclearly defined) group. 4We speak of “Othering” as the differentiation of a group or a person from another group by describing the former group or person as different and alien. Language is a possible tool of differentiation here (cf. Szymczyk et al. 2022). 5See for a similar discussion in Great Britain Szymczyk et al. (2022). 108 6 Constructing a clientele in need: The field of German as a Second Language 2.3 Observations Given the previous observations, we have the impression that the academic field of GSL does indeed have a tendency to develop a certain blind spot and treat bilingual young people who were born and raised in Germany as GSL learners and in need of special language support. Three examples should suffice for a start, and we give them without a reference, because we do not want to put individual authors on a spot: • A more didactically oriented introduction to teaching German in multilingual classrooms from 2017 equates being multilingual with acquiring German as the second language. • One of the most successful introductions to German as a second language (4th edition, 2020) uses “migration background” as the central defining criterion for L2 learners of German. Later, “insufficient knowledge of German as a second language” is identified as a decisive factor for the educational failure of this group, in addition to “lower class membership”. • A study on the reading behavior of adolescents, published 2014 by a recognized publisher in the field, defines L2 learners of German on the basis that they have “non-German mother tongue”. At no point does AoO or LoE play a role, neither are the linguistic competencies of the adolescents topic at any point. Our review is intended to answer the question, whether the above examples represent a general trend in the sense that features like migration background, multilingualism and L2 learner are mixed and additionally attributed with insufficient academic performance, or whether they are exceptions. Let us make quite clear at this point that our study is not intended as just another act of Othering. The authors of this study by no means exclude themselves from this critical (self-)reflection but have in the past engaged themselves in publications which defined German as a L2 through implicit and vague categories. So when asking here, what kind of discursive construction of a clientele emerges in scholarly publications on GSL, we also ask, what kind of discursive constructions we ourselves have engaged in. 3 Review The focus of the empirical work that follows is on the question of whether our selective observations stand up to empirical verification. These observations in109 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock clude on the one hand the hypothesis that different target groups, namely L2 learners in the narrow sense as well as multilingual speakers in general, are mixed under the umbrella term GSL. On the other hand, the construct GSL seems to be associated with insufficient knowledge of German and the need for support. To test these assumptions, we conducted a semi-systematic review. Our research question was: How do academic papers on GSL published in Germany, and written in German for a German academic audience, conceptualize child and adolescent GSL learners? 3.1 Methods 3.1.1 Sample The sample of our review includes 138 papers, of which 92 are empirical studies and 46 are non-empirical (i.e. conceptual). To compile the sample, we conducted an automatic literature search in the database “FIS Bildung”,6which includes monographs, collective work contributions and journal articles from different fields of acquisitionand learning-related research. The database includes more than 1 million subject-specific titles. Within this database, we searched for papers that carried the term “German as a Second Language” (in German) in the title in different variants (i.e., German as a Second Language, Second Language German, L2 German, German as a Foreignand Second Language) and were published between 2011 and 2021. The first sample compiled in this way included 113 titles, which were reduced in a second step excluding works focusing on out-of-school learning, teacher education, development of teaching material, as well as on adult learners. The final sample extracted from the FIS database comprised 79 titles. In a second step, we included a book series in the sample that has established itself in the German-speaking community as relevant to the topics of GSL, multilingualism, and L2 acquisition. The series is the result of the workshop Kinder mit Migrationshintergrund (Children with a Migration Background), which has existed since 2009 and was renamed Workshop Deutsch als Zweitsprache, Migration und Mehrsprachigkeit (German as a Second Language, Migration and Multilingualism Workshop) around 2015. The workshop is held annually and invites researchers conducting GSL research. The series was included in the review because it reflects current research projects and discourses and can thus be classified as discourse-shaping. The review includes 14 workshop volumes (volume 1 to double volume 14/15) from the period 2009 to 2021. The volumes comprise a 6https://www.fachportal-paedagogik.de/literatur/erweiterte_suche.html?checkFormParams= 1&herkunft[]=fis [accessed 2024/04/09]. 110 6 Constructing a clientele in need: The field of German as a Second Language total of 189 contributions, of which we only included the empirical papers in the review (n = 89). Of the 89 empirical workshop papers, 30 were excluded because they did not meet the above mentioned inclusion criteria, leaving 59 titles in the workshop series subsample. The reason for including only empirical papers from the book series is that we assume that mainly empirical studies can shed light on how GSL learners are operationalized. For the overall sample, however, this means that there is a surplus of empirical studies. 3.1.2 Data preparation The sample was coded using a double-coding approach (cf. O’Connor & Joffe 2020). The codes were first developed deductively from the psycholinguistic literature and then extended on the basis of a smaller data sample. In this way we were able to identify a total of 19 features, of which we considered three to be explicit and 16 to be implicit or secondary. We recorded both sets of characteristics (i.e., explicit vs. implicit) once with regard to the theoretical description of GSL in the entire sample and once exclusively with regard to the operationalization of subjects in empirical studies. Table 1 gives an overview of the features we identified within the sample. Under “explicit”, we have grouped features which unambiguously clarify that German is an L2 in the psycholinguistic sense. The AoO, LoE and LoS features can be used to measure the onset and age of acquisition.7Under “implicit” we have grouped characteristrics that may be a consequence of (e.g., bilingualism or multilingualism, other or additional language spoken at home)8or a reason for L2 acquisition (e.g., migration, refuge status), but these are not sufficient criteria to determine whether German is indeed an L2 for the learners described in the papers. Among the implicit features, moreover, there are also numerous ones that have no relation to GSL in the narrow sense, most notably the features “insufficient competence” and “educational disadvantage”. Such characteristics may apply to all learners in the school, regardless of their individual language biographies. The list also includes features that suggest an Othering of GSL speakers (e.g., specific ethnic background or the feature “foreign”). The term “unspecific” was used when GSL learners were mentioned but not described in detail. “Other” includes characteristics that are too rare to record separately. 7We included Length of Schooling (LoS) under explicit features since we assume that reporting the length of schooling allows us to infer the length of exposure and also the age of onset. 8“Other language” refers to the assumption that the speakers speak a different language and therefore do not speak German at home or as L1. “Additional language”, on the other hand, means that German is spoken in addition to another family language. 111 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock paratively low overall occurrence, we find studies that remain unspecific when it comes to operationalizing of GSL. Overall, the data review so far suggests a strong tendency to frame GSL in terms of implicit rather than explicit features. Within these implicit features, there is a tendency toward very general characteristics of speakers and their biographies (e.g., bilingual, multilingual, immigration) that allow little inference about whether German is an L2. Likewise, characteristics such as “insufficient competence” and “educationally disadvantaged” are often found, which in themselves do not necessarily have any relation to GSL and which in turn are not or hardly ever the subject of empirical studies. What has been shown based on the general feature frequencies can be consolidated with the help of Figure 2 and Figure 3 as follows: Implicit features not only dominate the construction of GSL, they are also found more often for general description of speakers and less often for operationalization of subjects in the context of empirical studies. This discrepancy is particularly pronounced for those features that hardly allow any conclusions to be drawn about whether German is the L2 (i.e., for the features “insufficient competence”, “immigration”, and “educationally disadvantaged”). So far, we have used relative frequencies to get an exploratory overview of the data. What the frequency calculations do not tell us, however, is how the individual features relate to each other and whether, for example, certain features occur particularly frequently with others. To identify such forms of feature cooccurrence, we performed an association analysis. 3.2.2 Feature associations An association analysis (Agrawal et al. 1993) is a statistical method of exploratory data analysis (comparable to data mining techniques) that makes it possible to identify patterns and relationships in large data sets. It is particularly useful for finding associations between different characteristics that are not directly related. The goal of an association analysis is to uncover relationships between different variables and find possible rules or patterns. In the previous section, we have seen that the majority of papers in our data sample use implicit features to describe GSL speakers and operationalize participants. As stated several times, these implicit features are not suitable for making statements about whether German is an L2 for the speakers in question. We have also seen that the general description of potential GSL speakers are often based on features that can be classified as deficient and have no explicit reference to the psycholinguistic notion of L2 acquisition (i.e. features such as “insufficient competence” or “educationally disadvantaged”). For the empirical studies, we also observed a discrepancy 118 6 Constructing a clientele in need: The field of German as a Second Language between general properties assigned to speakers and the operationalization of participants. We argue that the construction of a speaker group via implicit and also partly deficient features contributes to the construction of Othering. In the third and final step of our analysis, we are therefore interested in whether papers that rely on explicit and psycholinguistically motivated features (i.e., AoO, LoE, LoS) are more likely to refrain from implicit features and deficit attributions than those that do not mention these explicit features at all. To make such potential differences more visible, we divided our data sample into two new subsets: For the first subset (subset 1, n=29), we have selected only those studies that use explicit features to operationalize participants. This subset therefore contains only empirical papers. The second subset (subset 2, n=87), on the other hand, includes all papers that do not use explicit features at all – neither in the general description of GSL speakers nor for the operationalization of participants. In this subset, we included both empirical and non-empirical papers. Here we would like to identify whether the implicit features used to describe and operationalize speakers or participants are linked to each other in a specific way. We use an association analysis to determine whether implicit (and deficient) attributions are particularly prevalent where explicit defining features are not used.9 We performed an association analysis using the Apriori algorithm from the arules package in R (Hahsler et al. 2005) . Using the algorithm, we identified tendencies that describe which features tend to occur together across studies. Such tendencies of co-occurrences form rules (comparable to different weights) within the data set. Two characteristic values are usually calculated as part of the association analysis: Support and Confidence. “Support” is a measure of how often a specific combination of features occurs in the data. If support is high, it means that combination occurs frequently. For instance, if papers within our data sample contain a feature A and also B, then the support for the combination of A and B is high. “Confidence” is a measure of how likely it is for B to occur when A occurs. Together, these values help us identify patterns in the data and understand which features often occur together and how strong their relationship is. In addition to Support and Confidence, other parameters can be considered when evaluating association rules (e.g., Coverage and Lift). In our concrete example, we restricted ourselves to Support and Confidence, because these parameters have the highest informative potential. 9A more common data mining technique is cluster analysis. Cluster analyses, however, are based on identifying similarities between observations. In our case, a cluster analysis would have identified papers that share specific features. An association analysis, however, enables us to find co-occurrences across single data points (i.e., papers in our case). Also, cluster analyses usually require a predefinition of clusters to be identified in a data set. Association analysis is not restricted in such a way. 119 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock Association analysis requires the definition of a confidence threshold. We decided on a comparatively high threshold of 0.8 which means that we get an output of those rules whose probability of occurrence is at least 80%. By setting this, we have ensured that the rules we observe have a high probability of features co-occurring. In addition, an association analysis requires the setting of a support threshold which is a value that is set to determine which combinations of elements in a data set should be considered relevant. It defines the minimum frequency with which a combination must occur for it to be considered. In our case, we initially set a support threshold of 0.2 meaning that we are only interested in combinations of features that occur at least 20% of the time in our data. In other words, we focus on patterns that occur fairly frequently. For subset 2, however, we found that there is only one rule that meets the Confidence and Support thresholds. To make more rules visible, we therefore lowered the Confidence threshold for subset 2 to 0.6 and Support to 0.1. The two results of subsets 1 and 2 thus differed in the frequency of the rules we observed. In consequence, the rules observed for subset 2 are weaker than those for subset 1. Below we first see the results of the identified rules in a table format, each for subsets 1 and 2. The “lhs” (= left-hand side) column in the Table 2 and Table 3 lists the features that were identified as conditionor if-elements of a rule. If more than one feature is listed in this column, it means that these features occur together as a condition for the rule. The order of the variables in the column is not meaningful and has no significance for the interpretation of the rule. The column “rhs” (= right-hand side) refers to the variables that are included in the rule as inferenceor then-elements. The tables thus show the frequency (= Support) and the likeliness (= Confidence) of if-then-elements (that together form rules) within the data set. The column “count” states the number of papers in which the observed rule occurs. As we can see for both Table 2 and Table 3, there is a clear contrast between the Support and Confidence values. We find comparatively high Confidence values, but rather low Support values. This means that the if-then-relations we found in the data are fairly strong in the sense that the likeliness of co-occurrence is very high. At the same time, the co-occurrences do not occur very frequently. This is especially true for subset 2: since we had to lower the Confidence and Support thresholds, the if-then-relations are generally lower. This, in turn, may imply, that there is a certain arbitrariness as to which features are associated with GSL and that there is semantic vagueness and ambiguity in the construct. To better understand the numeric results from Table 2 and Table 3, the results of the association analyses are presented in Figure 4 and Figure 5. Figure 4 represents the results based on Table 2 and Figure 5 those based on Table 3. Both 120 6 Constructing a clientele in need: The field of German as a Second Language Table 2: Results association analysis subset 1, n=29 papers which mention explicit features condition/if-elements inference/then-elements support confidence count {op_LoS} →{op_LoE} 0.21 1.00 6 {th_disadvanataged} →{th_biling} 0.21 1.00 6 {th_multiling} →{op_otherL} 0.25 1.00 7 {th_multiling op_LoE} →{op_otherL} 0.21 1.00 6 {th_biling op_otherL} →{op_LoE} 0.25 1.00 7 {th_immigration} →{op_LoE} 0.25 0.87 7 {th_otherL, op_otherL} →{op_LoE} 0.25 0.87 7 op_otherL op_LoE} →{op_otherL} 0.25 0.87 7 {th_biling op_LoE} →{op_otherL} 0.25 0.87 7 {th_multiling} →{op_LoE} 0.21 0.86 6 {th_insufficientcomp} →{op_LoE} 0.21 0.86 6 {th_multiling op_otherL}→{op_LoE} 0.21 0.86 6 Table 3: Results association analysis subset 2, n=87 papers which do not mention explicit features condition elements (= lhs) inference element (= rhs) support confidence count {th_disadvantaged} →{th_biling} 0.14 0.81 13 {op_multiling} →{th_multiling} 0.13 0.75 12 {op_multiling} →{op_otherL} 0.13 0.75 12 {th_multiling, op_multiling}→{op_otherL} 0.10 0.75 9 {op_otherL, op_multiling} →{op_multiling} 0.10 0.75 9 {th_multiling, op_otherL} →{op_multiling} 0.10 0.75 9 {th_biling, op_otherL} →{th_otherL} 0.12 0.68 11 {th_multiling} →{th_biling} 0.20 0.64 18 {th_disadvantaged} →{th_otherL} 0.11 0.62 10 {th_immigration} →{th_otherL} 0.13 0.60 12 121 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock figures were drawn manually, but are based on a visualization of results using the packages arulesViz and gggraph in R (cf. for application of packages for association analyses Hahsler & Karpienko 2017). The if-conditions are colored in white, and the then-conditions in grey circles. Arrows show the if-then relations between the features. The width of the arrows represents the support values, which are also indicated in the figure legend. As can already be seen from Table 2, we see in Figure 4 two main nodes in those empirical papers that use explicit features to operationalize participants: “other language” and “length of exposure”. It is striking that the two nodes go hand in hand with each other: “LoE” is an if-condition for “other language” and vice versa. The rather strong connection between these two features is largely plausible: If LoE is recorded, it also seems logical to include other language(s) of participants for operationalization. Also, the “th”-features used to describe GSL speakers within the theoretical section of the papers follow a largely consistent pattern: we predominantly find “th”-features such as “multilingual” and “bilingual” for the general description of GSL speakers. One exception is the “th”- feature “insufficient competence”. In addition, the isolated pair of nodes at the bottom right of the figure is also striking: completely detached from the operationalization features LoE and “other language”, we find a connection between the “th”-features “bilingual” and “disadvantaged”. Together with the “insufficient competence” feature, this is an indication that even studies that make use of explicit and psycholinguistically motivated features for the operationalization of participants use implicit (and deficient-oriented) features to describe GSL speakers. However, such attributions seem to be rather rare in these papers. Where speakers are identified as GSL learners based on the feature LoE, features that are a logical consequence of L2 acquisition (i.e., bi-/multilingualism, presence of other languages in addition to L2 German) predominate. This is different in papers that do not mention the explicit features AoO and LoE at all (i.e. neither for the general description of GSL speakers nor for operationalization of participants). In such papers, we see four main nodes in Figure 5: “other language” (as an “op”-feature), “multilingual”, “bilingual” and “other language” (all three as “th”- features). With respect to the operationalization of participants with the help of the feature “other language”, we find co-occurrences with “th”-features like ‘multilingual’ and “bilingual”. Although these co-occurences are in line with the papers that refer to more explicit features (mainly LoE, see Figure 4), it remains unclear whether an operationalization of GSL speakers through the feature “other language” is justified. Furthermore, we see in Table 3 and Figure 5 that the “th”- feature “educationally disadvantaged” triggers the features “bilingual” as well as “multilingual”. As already pointed out in section 3.2.1, the feature “educationally 122 6 Constructing a clientele in need: The field of German as a Second Language Figure 4: Associations of features in subset 1 123 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock Figure 5: Associations of features in subset 2 124 6 Constructing a clientele in need: The field of German as a Second Language disadvantaged” is never used for operationalization. Whether the relations of multilingualism, GSL and educational disadvantage suggested by such regularities are empirically verified and tenable remains open. Overall, the association analysis shows no particularly large differences between papers that use explicit features to operationalize subjects and those that do not mention the features at all. In both data subsets, we see that GSL speakers are predominantly described as bior multilingual and as speakers of an “other language”. There is also a tendency in both subsets to use implicit deficit characteristics such as “insufficient competence” and “educationally disadvantaged” as part of general descriptions of speakers. Finally, the main difference between the two subsets is that those who rely on explicit features to operationalize participants can state with a high degree of certainty that their participants are real L2 learners. At the same time, this does not justify the deficient attributes that we nevertheless find in the data. For those papers that do not contain any explicit features, we only know that the participants or speakers are bior multilingual and speak a language other than German. Whether they are L2 learners in the narrower sense remains completely open. 4 Discussion Our review started with the question of how German as a Second Language is conceptualized in scholarly publications on the subject, which concentrate on child and adolescent learners. The research question is based on the observation that GSL seems to be a rather vague concept, implying a deficient perspective on its clientele. To test our research question, we reviewed a data sample of n=138 papers that were coded according to explicit and implicit features related to GSL. In the first exploratory analysis, we determined which features are generally frequent and whether there are differences between empirical and non-empirical studies. We also assessed the relationship between general descriptions of GSL speakers and the operationalization of participants in the studies. In the second step, we used association analyses to determine whether there are regular relations between features in studies that use explicit features for operationalization (= subset 1) and those that do not mention explicit features at all (= subset 2). Our results can be summarized as follows: 1. Implicit features occur noticeably more often than explicit ones, regardless of the type of paper (i.e., empirical vs. non-empirical). 125 Jana Gamper, Christoph Schroeder, Julia Schlauch & Dorotheé Steinbock 2. The most frequent features are linguistic features, but mainly those that do not allow to conclude whether German is an L2 (i.e., “bilingual”, “multilingual” as well as “other language”). These features shape the general description of GSL speakers to a large extent and are also frequently used for the operationalization of participant groups in empirical studies. 3. The features “insufficient competence” (here concerning German) and “educationally disadvantaged” are also frequently used. These non-linguistic parameters are mainly found in the field of theoretical descriptions of GSL but are hardly ever used for operationalization in empirical studies. 4. Both papers that use explicit features for operationalization and those that do not mention them at all base their general description of GSL speakers and the operationalization of participants on the features from point 2. Regardless of whether the papers use explicit features to operationalize participants, we find deficient attributions using features from point 3. Based on these results, the initial observation that GSL seems to be a rather vague and unspecific construct can be confirmed, at least for our data sample. It is uncertain whether German can be considered an L2 for the speakers described in the non-empirical papers and the participants in empirical studies, in most cases. This is because only a small number of papers are based on comparatively strict acquisition-theoretical criteria (i.e., AoO, LoE and LoS). In the majority of the papers, GSL speakers are conceptualized as a group of speakers who are biand/or multilingual, speak a language other than German at home, are educationally disadvantaged, and/or have insufficient German language skills. The fact that explicit criteria for the definition of GSL speakers are rather rare is problematic for two reasons: First, the group’s second language proficiency in German is uncertain due to the lack of explicit criteria. Second, the lack of explicit criteria allows for the creation of a construct of GSL speakers that may not exist in reality or may be highly biased. Our data analysis suggests an answer to our second research question: The GSL construct, as pursued in the academic field of GSL in Germany indeed has tendencies toward a deficit view of GSL speakers in particular and multilingualism in general. The fact that biand multilingualism, inadequate German language skills, and educational disadvantage appear to be entangled in our data in such a dominant way suggests that a discourse is being reiterated here that we originally found in non-expert educational policy debates. Large-scale studies such as PISA or educational trends for German, among others, have suggested, at 126 6 Constructing a clientele in need: The field of German as a Second Language least in their early results, that migration-related multilingualism is problematic when it comes to achieving adequate German competencies and high educational attainment (cf. Artel et al. 2003: 51ff.). This initial interpretation has since been significantly weakened (cf. Reiss et al. 2019: 158f.). Instead of migration-related multilingualism, social factors (independent of students’ linguistic background) seem to influence the attainment of certain language competencies and degrees. GSL research should correct this form of bias with the help of accurate methodological designs, instead of using labels such as GSL or multilingualism to support proxy debates (precisely because important causal factors such as social inequality seem to be more important reasons for different school performance). There is reason to be concerned that such proxy debates stigmatize GSL learners as well as (certain) multilingual speakers. Such an interpretation of our data does not mean that there are no studies that do not address this issue - there certainly are. However, in our data set, we find a clear tendency toward methodological as well as conceptual imprecision coupled with the aforementioned correlation with more deficit education-related features. Another tendency that is evident in our data is the conceptualisation of GSL as linguistic Otherness. The dominance of the characteristic ”other language” (than German) in our data shows that the mere existence of a language other than German in the family context leads to a classification in which the multilinguals are not competent speakers of German - and thus ’other’ from the monolinguals (cf. also Szymczyk et al. 2022). Although we did not test whether the characteristic correlates with concepts of Othering, it would be worthwhile investigating such correlations in future studies. Our findings highlight problematic tendencies in the GSL research community of participating in parts of a deficit discourse and share responsibility for discursively treating multilingual speakers in general and GSL learners in particular as Other. This is a harsh accusation. Let us stress again that the authors of this study by no means exclude themselves from this problem. Our studies, too, often define GSL through implicit and vague categories and engage in a form of Othering.10 Therefore, our review in no way intends to defame individuals or research approaches. Rather, it aims to make certain, in our view problematic, approaches 10We are aware of the fact that it is state of the art to include a monolingual control group in empirical studies. However, there are numerous problems to this, one of them being that monolinguals hardly ever constitute a homogeneous group of speakers but tend to show high degrees of heterogeneity (see also Alexiadou 2025 [this volume]). It is therefore necessary to carefully evaluate whether a comparison between multilinguals and monolinguals is theoretically grounded and purposeful. In any case, a control group should be well justified and should not be an automatism, as this can contribute to conceptualizing GSL and multilingual speakers as others. 127