Joana Marina
-
Upload
ramiromarques -
Category
Documents
-
view
215 -
download
0
Transcript of Joana Marina
-
7/29/2019 Joana Marina
1/15
Code # 18-64-814
Capturing language diversity in quantitative studies: new methodological approaches
AbstractThe present paper intends to provide an overview of commonly used concepts of data collection in
empirical educational research focusing on achievement differences between students with and
without a migration background. Several operationalizations of language use and language
competence, as well as of the concept migration background itself will be outlined and discussed
with respect to their challenges and limitations. Furthermore, an empirical study on the variation in
results using different operationalizations will be provided.
Keywords: migration background, language use, quantitative research, multivariate analysis1. IntroductionA considerable number of school performance studies has repeatedly found evidence for the
significantly poorer results of students with a migration background in most European countries in
several educational areas (OECD 2010; Bos et al. 2009). Specifically, these disparities exist with
respect to both language-related tests (such as reading-comprehension in the PISA-Study, see OECD
2010) and natural sciences or mathematics (for example in the TIMSS Study, see Bos et al. 2009).
Furthermore, disadvantages seem to grow cumulatively throughout the school career.An aspect that is increasingly referred to in the quantitative operationalization of the concept of
migration background is related to the language use within the families. The new migrati on
phenomena since the end of the Cold War, brought about by increasing globalization movements
and characterized by an intensification of migration typologies (in terms of countries of origin,
language use, ethnicity and religion, as well as of motives, patterns and itineraries of migration,
processes of integration into host communities, etc.), has exerted a strong effect on the complexity
of language practices among the migrant population. This diversification of diversity (Martiniello
2004) has been defined under the umbrella-concept of super-diversity (Vertovec 2007). Whereas
sociolinguistic studies call for a more complex and encompassing description and understanding of
what has been called super-diverse language repertoires of migrants (Blommaert & Rampton2011), most large-scale studies use few simplified variables to gather data about language use within
migrant families. Against this backdrop, this paper tackles the issue of capturing complex forms of
linguistic super-diversity and aims at giving an overview of language-related explanations for the
educational gap between students with and without a migration backgorund (section 1), focusing on
the consequences of migration-induced linguistic super-diversity for data collection in quantitative
empirical studies (section 2). Based on an empirical study of 273 9th
and 10th
graders in Hamburg,
Germany, we address the question of whether and to what extent different operationalizations of
family language use and the choice of reference groups in multivariate analyses may matter in terms
-
7/29/2019 Joana Marina
2/15
of conclusions that are drawn with respect to the impact of family language practices on students
majority language competence.
2. Explaining the educational gap: consequences for researchGenerally speaking, it is reasonable to expect an educational system to play an important improving
role with respect to new migration-induced forms of super-diversity. Yet, achievement studies of
students of linguistic minority backgrounds do not make reassuring reading. Contrarily, indicators
generally held to be appropriate benchmarks for educational systems suggest that European
educational systems are falling short of reasonable targets with respect to their linguistic minority
populations. Even when socioeconomic status and parental educational background are controlled
for, a disproportionately high amount of students born outside the country of residence or whose
parents were born abroad fail to reach levels of reading, mathematical or science literacy that are
comparable to those achieved by their native peers. Thus, addressing aspects leading to the
achievement gap entails at the same time addressing issues of social justice and equity in educationand society.Several attempts have been made to grasp the gap by isolating and analyzing specific aspects leading
to educational disparities. As a consequence, a considerable amount of studies has focused on
characteristics of the migrant groups themselves, such as their social, religious or cultural background
which has consistently been shown to differ from that of natives. However, the disadvantages of
students from migrant families cannot be satisfactorily explained by the assumption that their
cultural or religious predispositions do not match the expectations of the schools or by the
comparatively poor socio-economic situation of their families in empirical studies (Diefenbach 2010).Another set of studies has attempted to describe legal and political measures leading to better or
worse school results (see Gorard & Smith 2004 for a comparison of equity issues). For example,
Canada is a multilingual nation by definition and has immigration laws and acculturation strategies
that differ from those of officially monolingual countries, so that empirical results are necessarily
different from those in other contexts. However, although many policies and measures have been
adopted from successful PISA countries, for instance, most legal and political features of educational
systems cannot easily be transferred to another context and if so, do not necessarily guarantee
higher educational outcomes.In addition, many researchers have conducted sociological studies based on Bourdieus theory of
class distinction (Bourdieu 1991). Bourdieu theorizes that class fractions are determined by acombination of varying degrees of social, economic, and cultural capital across the population. He
emphasizes the dominance of cultural capital early on by stating that differences in cultural capital
mark the differences between classes. This perspective would imply that societies reproduce
inequalities across time, and that those occupying lower positions within a society pass on their
position to their children. However, when controlling for differences in capital endowments,
significant differences are consistently found to remain with respect to the educational performance
of students with and without a migration background (Diefenbach 2010).The fourth set of factors identified as a possible cause for educational inequalities is related to
several systemic aspects of teaching and learning. For example, the fact that an educational systemhas a monolingual self-understanding although its population is largely multilingual is one of the
subtle mechanisms leading to educational failure (see the concept of the monolingual Habitus in
-
7/29/2019 Joana Marina
3/15
Gogolin 1994). Migrant languages are thus often seen as an obstacle to the learning of the language
of the host society rather than as a valuable instrument in the acquisition of other languages. It has
also been shown that institutional discrimination mechanisms within educational systems, such as
the early tracking of students, affect migrants more often than their monolingual peers (Gomolla &
Radtke 2002). Furthermore, research has shown that the linguistic register used at school is often
inaccessible for second language learners and is not explicitly taught (Gogolin & Lange 2010).Another major factor of educational differentiation is immigrant generational status. In almost all
contexts, the second generation tends to outperform the first (OECD 2006). Some second-generation
groups have been found to even outperform native-born students (Chiswick & DebBurman 2004).
Contrarily, other groups show restricted progress across generations: a well-known example in the
literature is the experience of Mexican descendants in the United States (Telles & Ortiz 2008).Most empirical studies agree with respect to the evident influence of family characteristics in
accounting for extensive parts of educational disparities, though there are significant variations
across countries (Heath, Rothon, and Kilpi 2008). Research on equity in educational opportunitiesusually operationalizes family background in the form of variables such as parental occupational and
educational level. However, socioeconomic background is equally associated with an ample range of
other family features with (partially reciprocal) effects that are not easily distinguishable from one
another. For example, low socioeconomic status has been linked to weak family structures
(McLanahan & Percheski 2008) as well as to reduced cognitively stimulating resources in the home
environment (Lahaie 2008).While measures of parental education tackle some of the cultural factors relevant for childrens
educational success in general, other aspects seem to play a more crucial role within the scope of
families with a migration background. Theoretically, language competencies denote one of the most
important types of human capital characteristics of immigrant families and have been considered a
central predictor for educational attainment (Esser 2006). However, in practical terms, and especially
in educational contexts, it is often the case that the languages of migrants are perceived as a threat
to the monolingual self-understanding of most European nation-states (Gogolin 2002).Boudons (1974) distinction of primary and secondary effects which originally aimed to explain
differences in educational outcomes across social strata has gained increasing attention in attempts
to explain educational disparities between natives and students with a migration background. While
the former refer to conditions which affect students probability of educational success, such as
capital endowments, the latter refer to systematically different educational decisions which result
from the familys social position in society. This concept has been extended to explain differences ineducational outcomes between natives and immigrant-background students and is discussed in
terms of primary and secondary effects of ethnic origin (e.g. Heath & Brinbaum 2007; Kristen &Dollmann 2010), referring to factors that influence students probability of success and their
educational decisions which are specific to the migration situation. In this context, both majority
language competences and the factors that are thought to influence these competences, such as
family language use, have been identified as central predictors of students probability of success in
terms of primary effects of ethnic origin (e.g. OECD 2010).In general, a need for multi-level approaches (i.e. involving the educational system, the school, the
community, the classroom and the individual level) in tackling the achievement gap has beenrepeatedly expressed (OECD 2004). Furthermore, not many studies have the capacity to focus both
on interaction effects and on reciprocal or cumulative effects over time of several aspects leading to
-
7/29/2019 Joana Marina
4/15
disparities in school outcomes. However, before starting to design complex empirical studies, it is
necessary to reflect on the operationalization of the above-mentioned crucial aspects which may
trigger performance differences between students with and without migration background: the
familys socioeconomic background, migration background and language use. Specifically, we argue
that results with respect to the impact of language use on language competences considerably vary
depending on the concrete operationalization of these variables, in particular in large-scale studies
with a limited amount of items to define very complex concepts and practices. Forms of linguistic
super-diversity thus seem to make existing instruments for data collection rapidly obsolete, and
currently used operationalizations for family language use should be carefully used and results
should be critically interpreted.
3. Linguistic super-diversity: consequences for researchThe past two decades, more concretely the period since the early 1990s, have been marked by a rise
both in the amount and complexity of migration flows worldwide. It is estimated that there areapproximately 214 million migrants worldwide at present (Vertovec 2009, UN-DESA 2008). When
compared to the identifiable migrants of the 1950s to the 1970s, current migrant groups are smaller
in numbers, more mobile, socially more stratified and their legal status is more differentiated. The
term super-diversity has been used to designate these global changes in migration flows and forms
which have occurred in the past twenty years (Vertovec 2006). An ethnicity-based approach (for
example the Turks or the Somalis) to understanding minority groups, as applied in many models,
thus seems insufficient and inappropriate. Yet, a methodological obstacle for planning research
within super-diverse contexts arises from these complex constellations. On the one side, ethnic-
based homogenization of groups which are per se diverse has to be avoided; on the other, by adding
manifold variables such as legal status, milieu and language-related aspects, research designs
become tremendously complex, specially if these variables simply constitute control variables
(Larsen-Freeman & Cameron 2008).The investigation of speakers, languages and communicative practices in super-diverse settings thus
causes various methodological challenges. To focus on the dynamic interplay of the relevant
variables, as well as on their development across time, longitudinal designs of both quantitative and
qualitative nature are called for. Regarding quantitative research, Lynn (2009) describes distinct
features of longitudinal surveys: the focus on individual-level change; the employment of measures
of stability or instability; the inclusion of time-related characteristics of events and circumstances;
the enabling of analysis of expectations and outcomes that would not be possible with any otherdata source. Such designs are complex and require long-time planning and financing.In a qualitative paradigm, designed to capture the complex nature of communicative practices,
ethnographic methods are usually applied (Blommaert & Rampton 2011; see also Creese &
Blackledge 2010). The complexity of language(s), their speakers and interactions may require
researchers to use a combination of methods to study one and the same phenomenon (triangulation
or mixed-method designs) (Flick, 2011, Leech and Onwuegbuzie, 2009).The OECD-PISA-Studies offer an example for the complexity of capturing language use in migrant
families (OECD 2010). Specifically, it includes the most frequently spoken language(s) at home by 15-
year-olds as a proxy for language use within the family. The DESI (Deutsch Englisch Schlerleistungen
InternationalInternational comparison of German and English student performance) study (Klieme
et al. 2006) has been particularly intensive in the endeavor to describe inner-familiar language-
-
7/29/2019 Joana Marina
5/15
related learning opportunities by means of a detailed questionnaire in which parents were asked
about language-related knowledge, attitudes and experiences. A major topic were language practices
in the family, such as dealing with the German and English languages (for example with respect to
German: "In our family there are conversations related to language issues, for example, can I say or
write this?; related to the use of English: We talk about English texts). DESI also included several
variables to capture the linguistic home environment in relation to learning, such as language-related
skills in the family, the importance of language in the occupation of the parents, language-related
forms of support, dealing with German or English in the family, the value of language in the family
and parental interest in German or English language teaching. In the classification of language use
DESI takes on a developmental perspective classifies students according to the first language learned
(in a tripartite model: 1) German, 2) multilingual, i.e. German and another language were
simultaneously acquired, and 3) other language, i.e. another family language was learnt first). Results
show that the categories of first language learnt and migration background are closely related. The
group of students with German as a first language is largely identical to the group without a
migration background as traditionally defined with respect to the students and parents country ofbirth. Students with a first language other than German are mostly found in the group of students
with foreign-born parents. Only the group of multilingual students is distributed relatively evenly
with regard to the origin of the family members.The German Socio-Economic Panel (SOEP), which constitutes one of the most important sources for
individual-level investigations of the migrant population, investigates family language usage in the
mother-child questionnaire only and specifically asks whether household members speak German
only, German and other languages, or other languages only with the child (Socio-Economic Panel
2013). The National Educational Panel (NEPS), which allows for analyses based on a large-scale data
set which also comprises productive language data for certain migrant groups, makes an effort to
capture family language use in a more detailed manner. Students are not only asked which
language(s) they have learned in the family, but also what languages are used in different
constellations of communication. Specifically, students are asked what languages they use when
talking to their mother, father, brothers and sisters, and what languages are used by their parents
when talking with each other, where students can differentiate between only German, mostly
German, mostly the other language, and only the other language. Further, the NEPS assesses which
languages are used by the students when carrying out different activities, such as reading books or
writing text messages (National Educational Panel Study 2013).Although when speaking about students with a migration background it seems to be clear what
exactly is meant, similar to the issue of language use the operationalization of this concept differs
across the different studies. Kemper (2010), for instance, illustrated this inconsistency using data
from the German school statistics. He demonstrated which problems stem from the heterogeneity in
the definition of migration background, particularly with respect to the limited comparability of
results from different studies. Furthermore, many studies focus only on particular groups of
migrants, for example on former labor migrants and their families (Venema & Grimm 2002; Babka
von Gostomski 2008; Weidacher 2000) or on ethnic German repatriates (Haug & Sauer 2007), while
special research projects focus on female migrants (Boos-Nnning & Karakasoglu 2006), the "second
generation" (Haug & Diehl 2005) or Muslims (Brettfeld & Wetzels 2007).The question of the measurement of language competence is also dealt with in various ways, makingresults difficult to compare. While the PISA-Study tests reading comprehension, DESI, for example,
also includes text production, reading comprehension (only for the English language), language
-
7/29/2019 Joana Marina
6/15
awareness, vocabulary, writing and orthography. Thus, different results may emerge in relation tothe tested areas and also forms (receptive vs. productive; free production vs. multiple choice, etc.).Summing up, it can be stated that empirical research dealing with the achievement of immigrant-
background students commonly uses various key-concepts, of which the operationalization is 1)
highly heterogeneous both in terms of the amount of items used to grasp certain phenomena as wellas with regard to the complexity and range of the areas included, and thus 2) do not allow for general
comparisons between studies.
4. Research questions and hypothesesThe present study aims to empirically address two issues with respect to attempts to estimate the
effect of language use in migrant families on students majority language skills. Firstly, the question is
addressed whether and to what extent estimates differ when applying different operationalizations
to capture language use in the family. Secondly, we assess how far it matters what reference group is
chosen in interpreting results. Specifically, while studies like PISA typically compare test scores of
natives and immigrant-background students from different language environments, it may be more
reasonable to compare the achievements among immigrant-background students that are
characterized by different language use strategies. In sum, we hypothesize that the impact of
students language use in the family will considerably differ when applying different
operationalizations of this concept, suggesting that commonly used dummy variables that simply
reflect whether German is used, or used most often, in the family or not may not adequately capture
students home language environment. In addition, we hypothesize that results can be strongly
manipulated by altering the reference group in regression analyses.
5. MethodologyTo empirically address our research questions, we collected data from 370 9
thand 10
thgraders from
three general schools in Hamburg in the end of the school year 2010/11, when students were at the
point of transition into further general schooling to obtain higher certificates or into vocational
education and training. Students were asked to fill out a detailed questionnaire in which one major
focus was set on language use in the family. On the one hand, information was collected on the
language use of different family members and in different communicative constellations. On the
other hand, we collected data on the languages and frequency with which different topics are
discussed in the family, and what languages students mostly use when carrying out different
activities. Further, students were asked to participate in a text production test in German to assess
their (academic) language skills. Additionally, we collected information on various background
variables and students educational aspirations to be able to later relate language issues to students
educational outcomes not only in terms of primary, but also in terms of secondary effects of ethnic
origin.Of primary relevant with respect to the present research question, however, is the matter of
language use of different family members in different communicative constellations as well as
students language test scores. In a first step, we will assess whether and to what extent students
with a migration background perform differently than natives in the language test. In the analysesbelow, students with a migration background are referred to as those students with at least one
parent born abroad. In a second step, we estimate regressions to explain the observed variation in
-
7/29/2019 Joana Marina
7/15
the language test score not by further differentiating the migrant population based on characteristics
of the family language use. Specifically, besides the question of whether only German or other
languages as well, or exclusively, are used in the family, we will apply five alternative
operationalizations of language use in the family to assess in how far results are sensitive to such a
change in perspectives. Finally, we will present the results of the same regressions using a different
reference group in order to investigate whether a change in the point of reference will allow the
manipulation of our results. Specifically, we will first use native students as the reference point and
then change it to immigrant-background students that grow up using only German most often in the
family.
Language use in the familyAs far as students language use is concerned, we apply the following operationalizations to
empirically address the research questions outlined above, which refer to the language(s) used in the family
the student acquired first in the family
used most often among the parents when communicating with each other
used most often by the student when communicating with his or her mother
used most often by the mother when communicating with the student
used by both parents when communicating with the students, i.e. whether German only is
used most often when communicating with the student by both parents, by one parent or
not at all. As concerns the precise operationalizations used, the first two aspects relate to the question whether
German only, German and other languages, or languages other than German only are used in the
family on the one hand, and what language(s) were first acquired by the student, on the other. As
regards the latter aspects, questions refer to whether the student indicated that only German,
German and other languages, or languages other than German only are used most often in
communications among different family members.
Language competenceLanguage competence was tested by applying the Fast Catch Bumerang language test for the
German language1 (Reich, Roth & Dll 2009). This test intends to give a complex insight into severalareas of linguistic proficiency as it is based on the method of profile analysis (Clahsen 1986). The test
is a picture-based elicitation of productive textual data, based on eight pictures depicting the various
steps of building a boomerang. The analysis of language samples includes four main aspects: textual
pragmatics (textual structuring, task accomplishment, etc.), vocabulary (technical and general nouns,
verbs and adjectives), academic language (nominalizations, use of passives, compound-words, etc.)
and syntax (text cohesion in terms of sentence connectors). From these indicators, an overall score
was built to reflect students language skills.
1There are also versions available in Turkish and Russian.
-
7/29/2019 Joana Marina
8/15
6. ResultsIn a first step, we present regression analyses to explain the variation in the achieved language test
scores by whether students have a migration background or not while controlling for sex and the
familys socioeconomic background (HISEI) as a baseline model (table 1). The lowest test score that
was achieved was as low as 9.99 points, the highest as much as 55.82 points. Only 273 out of the 370students were included in the analyses below as only those students provided information on all
relevant questions for our analyses, participated in the language test, and lived in one household
with their mother (and/or father) at the point of data collection.Firstly, we can observe that boys perform significantly worse than girls and score, on average, about
six points lower in the test than girls. While students socio-economic status is not significant, we
observe significant differences between students with and without a migration background: the
former group scores about four points lower in the language test (model 1a). The second model
includes a variable that further differentiates the group of migrants by the number of parents born
abroad and suggests that the gap observed between natives and migrants can be primarily attributed
to differences between natives and students from families where both parents are born abroad
(model 2a)2. While this group scores about five points lower than natives, no significant differences
can be observed in the language competences of natives and students with one parent born abroad.
This finding may, at first sight, suggest that differences in the language use in the family may indeed
be a major underlying mechanism which causing lower levels of majority language skills in immigrant-
background students. Models 3a 8a were estimated using different operationalizations of students
language use in the family to address this question empirically. The reference group in these models
is students without a migration background.Model 3a includes a variable that reflects what language(s) is/are used in the family and suggests
that immigrant-background students from families where German is used only do not perform
significantly worse than natives, while the other migrant groups do. Specifically, students from
families where German and other languages are used score about four points lower, and students
from families where languages other than German are used only score more than 7 points lower than
natives. As regards the language acquired first by the student, model 4a suggests a similar result.
While immigrant-background students who acquired German as their first language do not score
significantly lower than natives, students who acquired both German and other languages as their
first language or languages other than German only are characterized by significantly lower levels of
majority language skills. Specifically, the former group scores about 4 points lower and the latter
more than 6 points lower than natives.
As concerns the language use among parents (model 5a), we obtain different results. According to
this operationalization, neither the group of students whose parents use German most often when
communicating with each other, nor the group whose parents use both German and other languages
most often score significantly lower than natives. The group of students whose parents use
languages other than German only most often, however, is characterized by significantly lower levels
of performance and scores about 6 points lower than natives.As concerns the language(s) used most often by the student when communicating with his or her
parents, we find that all three groups of immigrant-background students score significantly lower
than natives when differentiating by the language usage of the student when communicating with2
Model 2a includes 271 observations only as in two cases students provided information on the country of birth of oneparent only.
-
7/29/2019 Joana Marina
9/15
the mother (model 6a). While students that use German most often score about three points lower
than natives, the difference is about 5 points in the case of students that use German and other
languages most often, and almost as high as seven points in the case of students that use languages
other than German only most often when communicating with their mothers. With respect to the
language use of the mother when communicating to the child (model 7a), we make a different
observation. While students whose mother uses German only most often when communicating with
them do not perform significantly worse than natives in the language test, students whose mother
does not only use German do. Specifically, students whose mother uses German and other languages
most often score about four points lower, and students whose mother uses languages other than
German most often score more than six points lower than natives.Finally, model 8a assesses the language use of both parents in communicating with the student and
suggests that neither students from families where both parents use German most often, nor from
families where one parent uses German most often only, score significantly lower than natives. The
group of migrants from families where neither parent uses German most often in communicating
with the student, however, score as much as five points lower in the language test than natives.In sum, we make several observations with respect to our first research question regarding the
matter of different operationalizations to capture students language use in the family. On the one
hand, we seem to indeed find a tendency that students who are exposed to German exclusively or
mostly in the family show levels of performance which are very similar to those of natives. Students
who are only or mostly exposed to languages other than German, on the other hand, show
significantly lower levels of language competence in all models, the difference to natives varying
between 5 and 8 points depending on the respective operationalization. As concerns the group of
students who are exposed to both German and other languages most often, the evidence is
ambiguous. Specifically, when using certain opeartionalizations this group performs significantlyworse than the group of natives, whereas no significant differences can be observed when using
other operationalizations. This finding suggests that the way in which family language use is captured
may indeed matter with respect to the impact ascribed to language use in explaining the
performance gap between natives and students with a migration background. Also, we find that all
models explain the variation observed in the language competence test to a very similar extent
(between 17 and 18 percent).In a second step, we address the question whether the choice of the reference group in our
multivariate analyses matters with respect to the impact of language use strategies on majority
language competencies (table 2). Specifically, while the reference group above consisted of nativestudents, it is now the group of immigrant-background students that is exposed to German
exclusively or most often only. Nothing changes with respect to the comparison of students without
a migration background and the reference group. As above, we observe significant differences only in
model 6b, reflecting the language(s) used most often by the student when communicating with his or
her mother. As concerns the other two groups, however, we observe several interesting changes in
coefficients and with respect to their significance levels. While the group of students who are mostly
exposed to language(s) other than German only in the family were shown to score significantly lower
than natives no matter which operationalization was used to capture family language use above, this
pattern turns out to be less clear now. In fact, this group scores significantly lower than immigrant-
background students mostly exposed to German only in the case of two out of six models, specificallyin the case of students that acquired language(s) other than German as a first language only (model
4a) on the one hand, and students from families where the mother uses languages other than
-
7/29/2019 Joana Marina
10/15
German most often when communicating with the child on the other hand. In all other models,
immigrant-background students that are exposed to German only most often and students that are
primarily exposed to other language(s) are characterized by similar levels of majority language
competence. As concerns the group of students that are exposed to both German and other
languages most often, we find no significant differences compared to the competence levels shown
by immigrant-background students that are exposed to German most often only in all models.Table 1: Different operationalizations of family language use, reference group natives. Language competence (German) Model 1a Model 2a Model 3a Model 4a Model 5a Model 6a Model 7a Model 8asex -6.05 *** -6.06 *** -6.21*** -6.09*** -6.01 *** -6.07*** -6.34*** -6.23 ***(reference category: female) 1.006 1.005 1.008 1.016 1.007 1.007 1.003 1.007
0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000HISIEI 0.03 0.02 0.03 0.02 0.02 0.02 0.02 0.02
0.027 0.027 0.027 0.027 0.027 0.027 0.027 0.0270.3050 0.3750 0.3050 0.4410 0.4310 0.5400 0.5400 0.5080
Migration background (one or both parents born abroad) -4.10 ***(reference category: no migration background) 1.025
0.0000One parent born aborad -1.81(reference category: no migration background) 1.496
0.2270Both parents born abroad -4.94 ***(reference category: no migration background) 1.145
0.0000Langauge at home: German only -0.92(reference category: no migration background) 2.4110.7020
Langauge at home: German and other language(s) -4.30***(reference category: no migration background) 1.063
0.0000Language at home: language(s) other than German only -7.34*(reference category: no migration background) 3.346
0.0290Language first acquired: German only -1.11(reference category: no migration background) 1.724
0.5220Language first acquired: German and other language (s) -3.87**(reference category: no migration background) 1.278
0.0030Language first acquired: language(s) other than German only -6.58***(reference category: no migration background) 1.467
0.0000Parents language usage: German most often -2.48(reference category: no migration background) 1.666
0.1370Parents language usage: German and other language(s) most often -2.45(reference category: no migration background) 1.636
0.1360Parents language usage: language(s) other than German most often -5.65 ***(reference category: no migration background) 1.24
0.0000Langauge usage student to mother: German most often -2.94*(reference category: no migration background) 1.289
0.0230Langauge usage student to mother: German and other language(s) most often -4.72***(reference category: no migration background) 1.339
0.0010Langauge usage student to mother: language(s) other than German most often -6.64**(reference category: no migration background) 2.1
0.0020Langauge usage mother to student: German most often -1.54(reference category: no migration background) 1.508
0.3080Langauge usage mother to student: German and other language(s) most often -4.21**(reference category: no migration background) 1.377
0.0020Langauge usage mother to student: language(s) other than German most often -6.49***(reference category: no migration background) 1.457
0.0000Both parents to student: German most often -3.32(reference category: no migration background) 2.113
0.1180One parent to student: German most often -2.16(reference category: no migration background) 1.524
0.1570No parent to student: German most often -5.40 ***(reference category: no migration background) 1.222
0.0000Constant 35.38 *** 35.58*** 35.45*** 35.77*** 35.71 *** 35.97*** 36.11*** 35.97 ***1.599 1.601 1.599 1.59 1.599 1.633 1.609 1.624
0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000R 0.16 0.17 0.17 0.18 0.18 0.17 0.18 0.17Adj. R 0.15 0.15 0.15 0.17 11.34 0.15 0.17 0.16F-Statistic 17.11 13.249 10.90 11.99 0.16 10.92 11.91 11.11Prob > F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Number of observations 273 271 273 273 273 273 273 273
-
7/29/2019 Joana Marina
11/15
Table 2: Different operationalizations of family language use, reference group migrants that are exposed to
German exclusively or mostly.Language competence (German) Model 3b Model 4b Model 5b Model 6b Model 7b Model 8bsex -6.21 *** -6.09*** -6.01 *** -6.07*** -6.34*** -6.23 ***(reference category: female) 1.008 1.016 1.007 1.007 1.003 1.007
0.0000 0.0000 0.0000 0.0000 0.0000 0.0000HISIEI 0.03 0.02 0.02 0.02 0.02 0.02
0.027 0.027 0.027 0.027 0.027 0.0270.3050 0.4410 0.4310 0.5400 0.5400 0.5080Students without a mig ration background 0.92
(reference category: migrants that use German only i n the family) 2.4110.7020
Langauge at home: German and other language(s) -3.38(reference category: migrants that use German only i n the family) 2.429
0.1650Language at home: language(s) other than German only -6.41(reference category: migrants that use German only i n the family) 3.981
0.1090Students without a mig ration background 1.11(reference category: migrants who acquired German only as a first language) 1.724
0.5220Language first acquired: German and other language (s) -2.76(reference category: migrants who acquired German only as a first language) 1.924
0.1520Language first acquired: language(s) other than German only -5.47**(reference category: migrants who acquired German only as a first language) 2.015
0.0070Students without a mig ration background 2.48(reference category: migrants whose parents use German most of ten) 1.666
0.1370Parents language usage: German and other language(s) most often 0.04(reference category: migrants whose parents use German most of ten) 2.1270.9860Parents language usage: language(s) other than German most often -3.17(reference category: migrants whose parents use German most often) 1.822
0.0830Students without a mig ration background 2.94*(reference category: migrants who use German most often to their mother) 1.289
0.0230Langauge usage student to mother: German and other language(s) most often -1.78(reference category: migrants who use German most often to their mother) 1.576
0.2600Langauge usage student to mother: language(s) other than German most often -3.70(reference category: migrants who use German most often to their mother) 2.23
0.0980Students without a mig ration background 1.54(reference category: migrants whose mother uses German most oft en) 1.508
0.3080Langauge usage mother to student: German and other language(s) most often -2.68(reference category: migrants whose mother uses German most oft en) 1.806
0.1400Langauge usage mother to student: language(s) other than German most often -4.95**(reference category: migrants whose mother uses German most oft en) 1.843
0.0080Students without a mig ration background 3.32(reference category: migrants whose parents use both German most often) 2.113
0.1180One parent to student: German most often 1.16(reference category: migrants whose parents use both German most often) 2.427
0.6340No parent to student: German most often -2.08(reference category: migrants whose parents use both German most often) 2.198
0.3450Constant 34.52*** 34.67*** 33.23*** 33.03*** 34.57*** 32.66 ***
2.552 1.992 1.973 1.71 1.871 2.2910.0000 0.0000 0.0000 0.0000 0.0000 0.0000
R 0.17 0.18 0.18 0.17 0.18 0.17Adj. R 0.15 0.17 0.16 0.15 0.17 0.16F-Statistic 10.90 11.99 11.34 10.92 11.91 11.11Prob > F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Number of observations 273 273 273 273 273 273
7. Summary and discussionIn sum, our data clearly shows that both the operationalization used to capture family language use
and the reference group lead to significantly different results with respect to the negative influence
ascribed to the use of languages other than German in migrant families. As concerns the
operationalization of family language use, our data suggests that it may strongly vary among
different family members and in different constellations of communication, and differently affect
students majority language skills. Consequently, the question of which language is used most often
in the family, as used in large-scale studies such as PISA and DESI, will most likely reflect the studentssubjective judgment of which of these constellations or persons the question may be most relevant
in this context. Further, we find that the choice of the reference group in multivariate analyses
-
7/29/2019 Joana Marina
12/15
indeed matters in drawing conclusions with respect to the effect of family language use on majority
language competences. While language use strategies in the family seem to play a major role when
using natives as a reference group, we hardly find any significant differences when using immigrant-
background students that are mostly or exclusively exposed to German in the family context only as a
point of reference.As the present study is of cross-sectional nature, however, no causal conclusions can be drawn in
terms of the actual influence of language usage on majority language competence. For example, it is
possible that reverse causalities may play a role in the estimations above. Specifically, it may not only
be the case that family language use influences students language competences in German, but the
opposite relation may also occur, i.e. the family language use may as well be influenced by students
language skills.Another limitation is the fact that relevant variables, such as the generation or age of migration, are
not included in our analyses. Similarly, it is possible that the operationalizations suggested to capture
family language use may not be of high explanatory value themselves but rather proxy relatedphenomena. For example, it must not necessarily be the case that the language used among parents
largely explains the variation in the students test scores, but that this pheno menon is rather related
to the languages used in communications between parents and the student him- or herself, which
may be much more important. A simultaneous consideration of these different aspects and
constellations, however, requires larger data sets which allow the introduction of several variables
into one model along with language competence measures other than self-assessed data. In sum, our
findings strongly suggest that results of and conclusions drawn from studies that consider family
language use to explain differences in language competence, or academic performance in general,
may be easily manipulated by choosing different reference groups and different operationalizations
to capture family language use.Our study has attempted to show the centrality of a critical reflection of the key-concepts and
reference groups when collecting, analyzing and interpreting data in super-diverse constellations.
According to Vertovec (2009), the complexity and amount of migration forms will continue to
increase, thus posing further challenges to empirical educational research. Our results show a
considerable amount of discrepancies deriving from the different forms of operationalization and the
choice of reference groups. As a consequence, we suggest that empirical research on the
achievement of immigrant-background students should be more explicit in explaining the challenges
and limitations of the chosen form of operationalization and should address differences in results
regarding different reference groups.
-
7/29/2019 Joana Marina
13/15
8. ReferencesBabka von Gostomski, C. (2008). Trkische, griechische, italienische und polnische Personen sowie
Personen aus den Nachfolgestaaten des ehemaligen Jugoslawien in Deutschland. Nrnberg: BAMF,
Working Paper No. 11.Blommaert, J. & Rampton, B. (2011). Language and superdiversity: A position paper.WorkingPapersin Urban Language & Literacies. Universities of Ghent, Albany, Tilburg and King's College.Boudon, R. (1974). Education, Opportunity, and Social Inequality: Changing Prospects inWesternSociety. New York: Wiley & Sons.Bos, W., Bonsen, M., Kummer, N., Lintorf, K. & Frey, K. (eds.) (2009). TIMSS 2007. Dokumentation der
Erhebungsinstrumente zur Trends in International Mathematics and Science Study. Waxmann:
Mnster.Boos-Nnning, U. & Karakasoglu, Y. (2006). Viele Welten leben. Mnster: Waxmann.Bourdieu, P. (1991). Language and Symbolic Power. Cambridge: Harvard University Press.Brettfeld, K. & Wetzels, P. (2007). Muslime in Deutschland. Berlin: Bundesministerium des Inneren.Chiswick, B.R. & DebBurman, N. (2004). Educational attainment: Analysis by immigrant generation.Economics of Education Review23 (4), 361-379.Clahsen, H. (1986). Die Profilanalyse. Ein linguistisches Verfahren fr die SprachdiagnoseimVorschulalter. Berlin: MarholdCreese, A. & Blackledge, A. J. (2010). Translanguaging in the Bilingual Classroom: A Pedagogy forLearning and Teaching? The Modern Language Journal94 (1), 103-115.Diefenbach, H. (2010). Kinder und Jugendliche aus Migrantenfamilien im deutschen Bildungssystem:Erklrungen und empirische Befunde. Wiesbaden: VS Verlag fr Sozialwissenschaften.Esser, H. (2006). Sprache und Integration. Die sozialen Bedingungen und Folgen desSpracherwerbsvon Migranten. Frankfurt am Main/ New York: Campus Verlag.Flick, U. (2011). Triangulation in der Bildungsforschung: Zum Stand der Diskussion - Aktualitt,
Anstze und Umsetzungen der Triangulation. In Triangulation in der Bildungsforschung, J. Ecarius &
I. Miethe (eds.). Opladen: Leske & Budrich. 19-40.Gogolin, I. (1994). Der monolinguale Habitus der multilingualen Schule. Mnster: Waxmann.Gogolin, I. (2002). Linguistic and Cultural Diversity in Europe: a challenge for educational researchand practice. European Educational Research Journal1(1), 123-138.Gogolin, I. & Lange, I. (2010). Bildungssprache und Durchgngige Sprachbildung. In: Frstenau, S. &
Gomolla, M. (eds.). Migration und schulischer Wandel: Mehrsprachigkeit. Wiesbaden: VS-Verlag.
107-127.Gomolla, M. & Radtke, F.-O. (2002). Institutionelle Diskriminierung. Die HerstellungethnischerDifferenz in der Schule. Opladen: Leske & Budrich.Haug, S. & Diehl, C. (2005).Aspekte der Integration. Wiesbaden: Verlag fr Sozialwissenschaften.Haug, S. & Sauer, L. (2007).Zuwanderung und Integration von (Spt-)Aussiedlern. Nrnberg:BAMF, Forschungsbericht Nr. 3.
-
7/29/2019 Joana Marina
14/15
Heath, A. & Brinbaum, Y. (2007). Explaining Ethnic Inequalities in Educational Attainment.Ethnicities 7, 291-305.Heath, A., Rothon, C. & Kilpi, E. (2008). The second generation in Western Europe: Education,
unemployment and occupational attainment.Annual Review of Sociology34 (1), 211-235.Kemper, T. (2010). Migrationshintergrund - eine Frage der Definition! Die deutsche Schule102(4), 315-326.Klieme, E. et al. (2006). Unterricht und Kompetenzerwerb in Deutsch und Englisch. Zentrale
Befunde der Studie Deutsch-Englisch-Schlerleistungen-International (DESI). Frankfurt am Main:
DeutschesInstitut fr Internationale Pdagogische Forschung.Kristen, C. & Dollmann, J. (2010). Sekundre Effekte der ethnischen Herkunft: Kinder aus trkischenFamilien am ersten Bildungsbergang. In B. Becker & D. Reimer (eds.). Vom Kindergarten bis zurHochschule: Die Generierung von ethnischen und sozialen Disparitten in der Bildungsbiographie.Wiesbaden: VS Verlag fr Sozialwissenschaften. 117-144.Lahaie, C. (2008). School readiness of children of immigrants: Does parental involvement play a role?Social Science Quarterly89 (3), 684-705.Larsen-Freeman, D. & Cameron, L. (2008). Complex Systems and Applied Linguistics. Oxford: OxfordUniversity Press.Leech, N. & Onwuegbuzie, A. (2009). A Typology of Mixed Methods Research Designs. QualityandQuantity43(2), 265-275.Lynn, P. (2009). Methodological Research for Longitudinal Surveys. Chichester: John Wiley & Sons.McLanahan, S. & Percheski, C. (2008). Family structure and the reproduction of inequalities.Annual
Review of Sociology34, 257276.Martiniello, M. (2004). The many dimensions of Belgian diversity. CanadianDiversity/DiversitCanadienne 3(2), 43-46.National Educational Panel Study (2013). https://www.neps-data.de/de. Accessed February 14,2013.OECD (2010). PISA 2009 Results: Learning Trends. Changes in Student Performance since 2000. Paris:Organisation for Economic Co-operation and Development.OECD (2004). Education at a glanceOECD Indicators. Paris: Organisation for Economic Co-operation
and Development.Reich, H., Roth, H.-J. & Dll, M. (2009). Fast Catch Bumerang. Deutsche Sprachversion.
Auswertungsbogen und Auswertungshinweise. In: D. Lengyel, H. Reich, H.-J. Roth & M. Dll (eds.). Von
der Sprachdiagnose zur Sprachfrderung. FRMIGEdition Band 5. Mnster: Waxmann. 209-241.Socio-Economic Panel (2013)., http://panel.gsoep.de/soepinfo2011/. Accessed February 14, 2013.Stanat, P. & Christensen, G. (2006). Where Immigrant Students Succeed: A Comparative Review ofPerformance and Engagement in PISA 2003. Paris: OECD.Telles, E. & Ortiz, V. (2008). Generations of exclusion: Racial assimilation and Mexican Americans.New York: Russell Sage Foundation.
http://link.springer.com/book/10.1007/978-3-531-92105-1http://link.springer.com/book/10.1007/978-3-531-92105-1http://link.springer.com/book/10.1007/978-3-531-92105-1http://link.springer.com/book/10.1007/978-3-531-92105-1https://www.neps-data.de/de.%20%20Accessed%20February%2014https://www.neps-data.de/de.%20%20Accessed%20February%2014https://www.neps-data.de/de.%20%20Accessed%20February%2014http://panel.gsoep.de/soepinfo2011/http://panel.gsoep.de/soepinfo2011/http://panel.gsoep.de/soepinfo2011/http://panel.gsoep.de/soepinfo2011/https://www.neps-data.de/de.%20%20Accessed%20February%2014http://link.springer.com/book/10.1007/978-3-531-92105-1http://link.springer.com/book/10.1007/978-3-531-92105-1http://link.springer.com/book/10.1007/978-3-531-92105-1 -
7/29/2019 Joana Marina
15/15
UN-DESA (2008). Trends in international migrant stock: The 2008 revision. United Nations
Department of Economic and Social Affairs.
http://esa.un.org/migration/index.asp?panel=1. Accessed February 13, 2013.Venema, M. & Grimm, C. (2002). Situation der auslndischen Arbeitnehmer und ihrer
Familienangehrigen in der BRD. Reprsentativuntersuchung 2001. Mnchen/ Offenbach:Bundesministerium fr Arbeit und Sozialordnung.Vertovec, S. (2006). The Emergence of Super-Diversity in Britain. Centre on Migration, Policyand Society Working Papers, 25. London: Oxford University.Vertovec, S. (2007). Super-diversity and its implications. Ethnic and Racial Studies 30 (6): 1024-1054.Vertovec, S. (2009) Conceiving and researching diversity. MMG Working paper 09-01:
www.mmg.mpg.de/fileadmin/user_upload/documents/wp/WP_09-01_Vertovec_Diversity.pdf.
Accessed February 12, 2013.Weidacher, A. (2000). In Deutschland zu Hause. Opladen: Leske & Budrich.
http://esa.un.org/migration/index.asp?panel=1http://esa.un.org/migration/index.asp?panel=1http://esa.un.org/migration/index.asp?panel=1http://www.mmg.mpg.de/fileadmin/user_upload/documents/wp/WP_09-01_Vertovec_Diversity.pdfhttp://www.mmg.mpg.de/fileadmin/user_upload/documents/wp/WP_09-01_Vertovec_Diversity.pdfhttp://www.mmg.mpg.de/fileadmin/user_upload/documents/wp/WP_09-01_Vertovec_Diversity.pdfhttp://www.mmg.mpg.de/fileadmin/user_upload/documents/wp/WP_09-01_Vertovec_Diversity.pdfhttp://esa.un.org/migration/index.asp?panel=1http://esa.un.org/migration/index.asp?panel=1