“Boundary-Maintenance” or “Boundary-Crossing”? Name-Giving Practices among Immigrants in Germany

This study investigates why some immigrants choose names for their children that are common in their home country whereas others opt for names used by natives in the host country. Drawing on the sociological literature on symbolic boundaries, the first strategy can be described as boundarymaintenance whereas the second can be classified as boundary-crossing. Using data from the German Socio-Economic Panel Study and applying bivariate and multivariate methods, two broader explanations for name-giving practices are tested: (1) cultural proximity and the permeability of the symbolic boundary between home and host country; and (2) immigrants’ levels of linguistic, structural, social, and emotional integration in the host country. Overall, the theoretical model explains the differences very satisfactorily. Whilst both sets of factors proved relevant to immigrants’ name-giving practices, the immigrants’ level of integration in the host country was less important than the cultural proximity between the origin group and host country.


Introduction
First names often act as markers of social identity. They typically reveal information about the person's gender and sometimes even about their generation, social class, or ethnic origin (Lieberson and Mikelson 1995;Lieberson, Dumais, and Baumann 2000;Sue and Telles 2007). In Germany, a person called Eylül or Pavlov might be inferred to come from an immigrant family based on the name. Hence, parents who choose a name that is common in their home country may mark their child's ethnic origin, while those who choose a name that is common in the host country may disguise their child's ethnic origin and indicate assimilation into the majority society. Drawing on the sociological literature on symbolic boundaries, the first strategy can be described as "boundary-maintenance" and the second as "boundary-crossing" (Horowitz 1975;Zolberg and Woon 1999;Lamont and Bail 2008;Wimmer 2008a).
This paper analyzes the strategies of ethnic boundary-making in the name-giving of immigrants coming from 41 different countries and living in Germany. 1 It seeks to explain why some immigrants adopt boundary-crossing strategies while others choose boundary-maintenance strategies. It is proposed that immigrants are more likely to opt for boundary-crossing when (1) the symbolic boundary between the home and host countries is "blurry"; and (2) they are better integrated into the host society in linguistic, structural, social, and emotional terms. As will be shown in the bivariate and multivariate analyses presented, variation in these factors helps to explain differences in immigrants' name-giving practices very well.

Theoretical Framework: Explaining Immigrants' Naming Practices
Cultural Proximity: "Blurred" or "Bright" Boundaries The article draws on the sociological literature on symbolic boundaries that commonly use the terms "bright" and "blurred" in order to describe the characteristic of boundaries (e.g., Alba 2005;Wimmer 2008aWimmer , 2008bLamont and Bail 2008). 2 With reference to naming, the brightness of the boundary between the home and host country can be determined by establishing the extent of the overlap between the two countries' name pools. The cultural distance is manifested in the share of names that is common in both countries. The more such names exist, the more blurred the boundary between the countries is, and the easier it is to find a first name that is not exclusive to either the home or host country. For instance, a family from Italy can choose from a wide range of first names that are common in both Italy and Germany, such as Emilia or Marco. In contrast, if Turkish parents wish to select a name from the German name pool, the boundary they have to overcome is a bright one, since there is little overlap between the Turkish and German name pools. If the parents cross the boundary to the host country, the first name might demarcate the child from their own origin group. In addition, the phonetic and structural characteristics of their child's first name might even be unfamiliar to them. We assume that two aspects of cultural proximity are particularly relevant with regard to the degree of the overlap of first names: language and religion (Gerhards and Hans 2009).

Language
The immigrants in our study come from 41 different countries in which different languages are spoken. Some of these languages are similar to German (such as Dutch), while others show a large linguistic difference, such as Hungarian or Turkish. As shown in other studies, linguistic proximity has an impact on commonly used first names (Gerhards and Hans 2009). The name pool overlap is therefore positively associated with the proximity in languages. The more similar two languages are, the greater the overlap between the name pools is. We therefore started from the hypothesis that immigrants who came from a country with a language that is similar to German are more likely to choose a name for their children that is common in Germany than immigrants who came from a country with a language that is different from German. Higher rates of boundary-crossing in name-giving were therefore expected for those immigrants who came from a country with a language that is similar to German (hypothesis 1a).

Religion
A second aspect of cultural proximity determining the name-pool overlap is religion. If immigrants' religious affiliation is congruent with the dominant religion in the host country, they will have a shared pool of first names. The Christian names common in Germany often come from the Old and New Testaments as well as from the pool of Christian saints. In contrast, Koranic names are largely uncommon in Germany (Gerhards 2005). Thus, immigrants living in Germany and coming from a historically Christian country have a larger pool of potential first names that are common in both the home and host countries; in such cases, the symbolic boundary between the two countries is blurred.
Even if it is assumed that religion influences the first-name pool for all members of a given cultural group, an individual's affiliation with the respective religious community increases their likelihood of selecting a religious name. It is hypothesized that, at the individual level, belonging to a Christian religious denomination will increase the likelihood of engaging in boundary-crossing, while belonging to a Muslim religious denomination will increase the likelihood of boundary-maintenance (hypothesis 1b).

Integration into the Host Society
Based on classical theories of immigration and integration and in line with previous studies (Lieberson 2000;Gerhards and Hans 2009), this study assumes that immigrants who are better integrated into the host society are more likely to choose names from that host country. Consistent with the relevant literature (Gordon 1964;Alba and Nee 1997;Esser 2001;Levitt and Jaworsky 2007;Hans 2010), four dimensions of integration are identified: linguistic, structural, social, and emotional integration into the host country. We assume that the better immigrants are integrated in linguistic, structural, social, and emotional terms, the more likely they are to adopt a boundary-crossing strategy (hypotheses 2a-d). The different dimensions of integration are not independent of each other. Hence, while the analyses here do not primarily concern these interdependencies, they are simultaneously included in the multivariate model in order to determine the independent effect of each individual dimension of integration.

Linguistic Integration
Migrating to another country often requires the acquisition of a new language. It can be assumed that parents with a good knowledge of the German language would choose first names that are common in Germany more often than parents with few or no German language skills (hypothesis 2a). Immigrants who have acquired the language of the host country have crossed an important boundary to the majority society; their linguistic integration then may be followed by a symbolic boundary-crossing regarding the name choice. Host language acquisition may not merely influence the associated integration process but may also familiarize immigrants with the phonetic characteristics of first names commonly used in the host country.

Structural Integration
The term "structural integration" refers to the acquisition of (1) relevant educational credentials; and (2) host-country citizenship and thus of legal equality with the citizens of the majority society (Esser 2001;Hans 2010). Structural integration should promote boundary-crossing in naming (hypothesis 2b) for two reasons. First, individuals' educational level is associated with labor market integration and income (Esser 2001;Hans 2010). To the extent that immigrants' socio-structural position correlates with their satisfaction with the host society, education can be expected to have a direct positive effect on their adoption of the host-society culture, for instance, in the form of host-society first names (Kalter 2005). Second, structural integration can be expected to have an impact on parents' motivation for advancement of their children. Crossing the boundaries into the majority society by giving a name that is common in the host society might be advantageous for the child later on, because it could implicitly indicate a willingness to adapt to the majority society's habits and styles.

Social Integration
The term "social integration" refers to migrants' embeddedness in majority society networks in the host country. These include friendships, marriages, or partnerships (Hans 2010). Here, it is assumed that immigrants with regular contact with members of the majority society will be more familiar with hostcountry symbols, including the associated first names, their meanings, and their phonetic characteristics (hypothesis 2c). This association should be stronger for partnerships and marriages than for friendships, as spouses (who are typically also the child's parents) will likely jointly decide on the first name.

Emotional Identification
One can assume that immigrants who feel that they belong to the host country are more likely to adopt its symbols and customs, including first names (hypothesis 2d). By choosing a name that is common in Germany, immigrants can express their emotional identification with the host society.
In recent years, however, the hypothesis that better integrated immigrants are more likely to adopt a boundary-crossing strategy has been challenged by studies on the "integration paradox." The integration paradox describes the (seemingly counter-intuitive) empirical finding that immigrants identify less with the host country as they become increasingly integrated into it (Verkuyten 2016;Steinmann 2019). Two explanations for this are discussed in the literature. First, with increasing levels of integration, immigrants' sensitivity to discrimination changes. As the level of integration increases, immigrants' expectations of being recognized as full members of the majority society and their awareness of discrimination grow (sensitivity). Second, integration in the host society changes the opportunity structures for discrimination. Better integrated immigrants are more exposed to members of the majority population and are therefore more likely to have encounters that involve discrimination (exposure). If perceived discrimination by the majority population leads immigrants to emphasize their ethnic origins more strongly, one could expect the likelihood for boundary-crossing to decrease with increasing integration. Thus, the integration paradox might make highly integrated immigrants more likely to opt for a boundary-working strategy that Andreas Wimmer refers to as "transvaluation" (Wimmer 2008a) or "inversion" (Wimmer 2008b). As better integrated immigrants are more likely to perceive rejection by the majority society, they develop a strategy that enhances their origin group. Disappointed by the perceived lack of acceptance by German citizens, they increasingly orient themselves towards the culture of their country of origin. This might make them more likely to choose first names from their origin country. Hence, two different and contradicting hypotheses can be derived from the literature: immigrants who are better integrated into the host society are either more or less likely to choose names from the host country and adopt a boundarycrossing strategy.

Dataset and Sample
The data used is from the German Socio-Economic Panel Study (GSOEP), a nationally representative survey of private households in Germany that is repeated on an annual basis and was established in 1984 (Goebel et al. 2019). Version 32 of GSOEP, which contains data collected between 1984 and 2015, is analyzed. The sample is restricted to first-generation immigrants who had at least one child in Germany (N=1,367). 3 The level of analysis is the child whose first name is the dependent variable in the models, although most of the independent variables relate to the parents. The respondents in our sample come from 41 countries.

Onomastic Coding of First Names
To make it possible to link together the data collected for each person on an ongoing basis, the GSOEP collects the first names of all individuals in a household.
For data security reasons, the first names are not part of the scientific use file. For this project, an anonymous list with all first names ever collected in the GSOEP was provided. The assignment of a first name to a specific country is mainly based on the empirical distribution of first names in different countries. The question of whether a first name is perceived as a "common German name" or as a name that indicates that the name bearer has an immigrant background is a categorization, and therefore, a social construction, which is exhibited by the members of a society. Even if names such as Mehmet and Mustafa have become widespread in Germany, they are still widely considered to be typical immigrant names.
To begin with, all first names were assigned to one or several countries or regions based on their prevalence in different countries and their historical origin by a professional onomastic company (Humpert and Schneiderheinze 2000). This coding provided information on whether a first name is traditionally used in Germany, Turkey, Russia, etc. First names that could not be assigned or that were obviously not given names (such as nicknames) were excluded from the analyses (2.96%). The aim was to identify all first names in GSOEP that belonged to the pool of names that are commonly used by the German majority population. However, the onomastic coding had some weaknesses. Some names that did not originate in Germany but are now common in Germany were not assigned to the country code for Germany, despite being likely to be considered "typically German" by the majority population. An example is the name Kevin, which is coded as a foreign name in the onomastic analysis conducted by the company, but which has been part of the German first-name pool since the 1980s.
In order to address this problem, two members of the project checked all first names that were marked as "not from Germany" with regard to the question of whether a name-regardless of its originsis a common name in Germany or not. Two criteria were applied here using two classification questions: Would members of the German majority population commonly give this name? Likewise, would a member of the German majority population recognize the name's bearer as having an immigrant background based on his or her first name? This recoding procedure goes beyond the original coding, as it classifies names that originate from outside of Germany and were not historically "common in Germany" as German provided that the names have now become part of the name pool used by the German majority population. In order to check the reliability of the recoding procedure, five German native speakers were asked to recode a randomly selected sample of 100 first names. The intercoder reliability (agreement rate between different coders) was 8.34 percent. Of course, this two-fold coding procedure still left a few ambiguous cases, but most first names could be assigned to the boundary-crossing or -maintenance categories. The proportion of cases that had to be excluded due to the fact that they either could not be assigned or were obviously not given names was 1.43 percent. These country codes were then merged with the personal IDs in the GSOEP, while the actual first names were deleted in order to guarantee respondents' anonymity. The result was a dataset that included information on the first names of all individuals in the GSOEP, stating whether or not their first name is common in Germany. 4

Dependent, Independent, and Control Variables Dependent Variable
The first names that GSOEP respondents gave their children born in the host country were differentiated according to whether those names were common (boundary-crossing) or uncommon (boundarymaintenance) in Germany.

Independent Variables
The first independent variable, cultural proximity, was operationalized by two indicators: linguistic proximity between German and the language spoken in the immigrants' country of origin, and respondents' religious affiliation. In order to determine the linguistic distance between German and other languages, we made use of the calculations from elinguistics.net. This website presents a computerized model for comparative linguistics. Based on a comparison between the basic vocabularies, the linguistic distance between two languages was calculated. The values for linguistic distance could vary between zero and 100. The lower the value the more similar the languages. 5 Scores were z-transformed and express standard deviations from the original scores. Unfortunately, it was not possible to measure linguistic proximity on the individual level because the respondents in our survey were not asked about their mother tongue. However, we do know which country the respondents came from. Correspondingly, we assigned the language that is the dominant language in the country of origin as mother tongue to each respondent. However, this approach can lead to errors, as some immigrants belong to a minority group in their country of origin and do not speak the dominant language of their country as their mother tongue. The second indicator for cultural proximity was the respondents' individual religious affiliations. The analysis differentiated between (1) Roman Catholic, (2) Protestant, (3) Christian Orthodox, (4) Muslim, (5) other, and (6) no religious affiliation.
The second independent variable, integration, was subdivided into four dimensions of integration: linguistic, structural, and social integration; as well as emotional identification with the host country. Where possible, the most recent information at the time of the child's birth was used. If information was not available in the last wave, data from the years preceding the child's birth was used, up to a maximum of seven years.
To measure linguistic integration, information about the respondents' subjective evaluation of their language proficiency in German in the domain of speaking was used. Respondents' subjective assessment of language proficiency was assessed on a five-point scale ranging from "not at all" to "very good." As the potential signaling function of language proficiency with regard to immigrant background is of concern here, the variable was coded into a binary measure with a value of 1 for good or very good knowledge of German and 0 for fair, poor, or no knowledge of German.
Two items were used to assess the level of structural integration, namely German citizenship and educational level. German citizenship was ascribed if at least one of the parents had German citizenship.
Educational level was measured by the ISCED 2011 (International Standard Classification of Education) score of both parents. The following categories were used: (1) low level of education (ISCED 1-2); (2) medium level of education (ISCED 3-5); and (3) high level of education (ISCED 6-8). If the spouses had different scores, the higher one was used in the analyses.
The level of social integration was measured using two indicators, namely intermarriage and German friends and acquaintances. Intermarriage was ascribed if one parent of the child was born in Germany. For the German friends and acquaintances variable, respondents were asked about the three most important persons from their circle of friends and acquaintances. Unfortunately, no distinction was made between friends and acquaintances. 6 In a follow-up question, respondents were asked to provide information on the country of origin of the three most important persons mentioned before. Based on this information, a dummy variable was coded as 1 if at least one friend or acquaintance was born in Germany and 0 if none of the three closest friends or acquaintances were born in Germany. Of course, measuring friends' and acquaintances' origins based on their birth country may be problematic as it may result in second-and third-generation ethnic minority friends being coded as "German friends and acquaintances." Therefore, the variable has to be interpreted with caution.
Emotional identification was assessed with a question regarding the extent to which respondents "felt German" on a five-point scale ranging from "not at all" to "completely." The average score of both parents on this quasi-metric measure was used. If information for one parent was missing, valid information for the other parent was used.

Control Variables
The child's sex was used as a control variable. Several studies have identified gender differences in naming (Lieberson 2000;Sue and Telles 2007;Becker 2009). Parents engage in boundary-crossing more often when choosing girls' names than boys' names. Additionally, controls were applied for the specific subsample of the GSOEP that each household belonged to.

Analytic Strategy
First, the associations between the independent variables and boundary-crossing in name-giving were subjected to bivariate analyses. At this stage, boundary-crossing proportions were analyzed by respondents' religious affiliation, language skills, citizenship, educational level, social integration, and child's sex in percentage terms. To obtain a quasi-metric measure of linguistic distance between German and the language of the immigrants' country of origin on the one hand and respondent's identification with the host country on the other, the strength of the associations with boundary-crossing were measured using a point-biserial correlation coefficient. While these bivariate analyses offered an indication of the associations between the independent variables and boundary-crossing in naming, it was necessary to conduct multivariate analyses because some associations might have diminished following the introduction of other independent variables. For instance, the association between cultural proximity and boundary-crossing might not have remained following the addition of the respondents' individual level of integration to the model. Therefore, stepwise logistic regression analyses were used in the second step to estimate the effects of the independent variables in multivariate terms. Logistic regression analysis allowed the estimation of the effect of a certain independent variable, such as linguistic distance, on the likelihood of boundary-crossing, while holding other dependent variables constant. The first set of models focused on the effects of cultural proximity. Admittedly, the two indicators of cultural proximity are rather diffuse and cannot directly quantify the degree of overlap in first names between the country of origin and the country of destination. In addition, linguistic proximity and religious affiliation can correlate strongly with one another. For this reason, the two indicators of cultural proximity were introduced in stepwise fashion. In model 1a, the effect of linguistic proximity was estimated while controlling for the child's sex and the subsample of the GSOEP. In Model 1b, the respondents' religious affiliation was added. In the second model, the level of integration in the host society was included (Model 2).
A word about the interpretation of statistical significance is necessary at this point. More recently, there have been controversies regarding the use of p-values and statistical significance. Some statisticians have argued in favor of stopping the conventional use of p-values to decide whether a result supports or rejects a hypothesis (Wasserstein and Lazar 2016). Instead, they advocate focusing on whether the sign of an association is positive or negative and the strength the relationship between two factors. We followed this suggestion in the interpretation of our data.

Bivariate Analysis Cultural Proximity: Blurred and Bright Boundaries
As Table 1 shows linguistic proximity impacts parents' strategy of boundary-making. Immigrants who came from a country whose language is similar to German are more likely to choose a name for their children that is common in Germany than immigrants who came from a country with a language that is different from German. The negative sign of the correlation coefficient indicates that hypothesis 1a is confirmed. The analyses also revealed substantial differences in boundary-crossing levels with regard to the second indicator of cultural proximity, religious affiliation. As expected, the proportion of children with common names in Germany was highest among Roman Catholic or Protestant parents, followed by parents from "other" Christian denominations. In non-religious families, about one fifth of the children received a name that was common in Germany. The lowest boundary-crossing rates were found in families belonging to an unspecified "other" religion and in Muslim families. In the latter group, only 1.19 percent of the children were given a name that is common in Germany. The bivariate pattern between religious affiliation and boundary-crossing in name-giving supports the expectations set forth in hypothesis 1b.

Integration
Overall, hypotheses 2a to 2d are confirmed by the bivariate analyses. The better the immigrants were integrated, the more likely they were to engage in boundary-crossing. This holds true for all five dimensions of integration. Hence, the results do not support the hypotheses that can be derived from studies on the "integration paradox," as highly integrated immigrants were more likely to opt for and not against a boundary-crossing strategy.

Multivariate Analyses
Do the patterns observed in the bivariate analyses hold in the multivariate models? In the multivariate analysis (see Table 2), average marginal effects (AMEs) were computed. These were positive when the variable in question increased the probability of boundary-crossing and negative if it decreased this probability. The AMEs indicated the differences in the likelihood of boundary-crossing in percentage points between the respective category of a variable and its reference category for categorical variables, and for an increase by one unit in case of continuous variables, respectively. Model 1a estimated the effects of linguistic proximity on name-giving practices, while controlling for the child's sex. The result confirms the finding from the bivariate analysis. The higher the linguistic distance, the lower the probability of boundary-crossing was in terms of choosing a name that is common in Germany. With every increase by a standard deviation in the linguistic distance, the likelihood of boundary-crossing decreased by 17 percentage points.
In Model 1b, religious affiliation was added. Compared to non-religious immigrants, Roman Catholic or Protestant immigrants were significantly more likely to choose a name that is common in Germany. This finding also applies to immigrants who were affiliated with other Christian religions, even if the effect was weaker than for Catholics and Protestants. This result is in line with our hypothesis. An additional finding was that, when religious affiliation was included in the model, the AMEs for linguistic proximity disappeared. We theorize that this result could be traced back to the fact that religion and linguistic similarity are highly correlated. For example, Christian immigrants very often come from countries in which a language is spoken that is more similar to German than, for example, Turkish, Arabic, or Chinese. Furthermore, religious affiliation might be a more accurate measure, as this information was collected at the individual level of the respondents, whereas the score of linguistic proximity was assigned based on the immigrants' origin country and the dominant language spoken in this country. This might have led to inaccurate assignments for immigrants whose mother tongue was not identical to the dominant language in their country of origin. Therefore, it is not possible to evaluate whether religion is more important than language proximity, or whether this is due to the different modes of measurement. A comparison of Models 1a and 1b also showed a substantial rise in the pseudo R 2 , indicating the importance of religious affiliation for naming practices. Data source: GSOEP v32; N= 1,367; all models include control for subsample; ***p < 0.01; *p < 0.1.
In Model 2, the second set of explanatory variables-pertaining to integration into the host societywas added. The analysis showed that linguistic integration in terms of German language skills was no longer a decisive factor when choosing a child's first name after controlling for cultural proximity and other dimensions of integration. Intermarriage also had no effect in the multivariate model. In contrast, the association between German citizenship and emotional identification on the one hand and boundary-crossing on the other remained statistically significant, net of cultural proximity and other dimensions of integration. With regard to education, highly educated immigrants were still more likely to opt for boundary-crossing than immigrants with a low educational level; this difference remained marginally significant at the 10 percent level. The difference between medium and low education pointed in the expected direction but was not statistically significant anymore. The number of German friends increased the probability of choosing a German name for a child. However, the effects were also no longer significant.
The extent of the increase in the pseudo R 2 from Models 1b to 2 indicated that religious affiliation had a stronger impact on naming strategies than the degree of integration in the host society. Hence, cultural proximity between the home and host countries (net of integration) seemed to be more influential than the degree of integration in the host country (net of cultural proximity).
It is also noteworthy that the effect of the child's sex was stable throughout the models. As expected, boys were significantly less likely to receive first names that are common in Germany than girls. This gender gap may reflect gender roles insofar as male descendants are predominantly associated with the intergenerational transmission of family traditions and thus receive names from the country of origin more often than girls. However, gender differences in the assignment of names can be observed not only among immigrants, but also among native citizens, as previous studies have shown (Rossi 1965;Lieberson and Bell 1992).

Discussion
There are various studies that have tried to explain why some immigrants decide on boundary-crossing and give their children names that are common in the host country while others choose boundarymaintenance strategies and opt for names that are common in their country of origin. The results of these studies indicate that where the likelihood of boundary-crossing is larger, the better immigrants are integrated into the host society socially in terms of education, occupation, citizenship, friendships, or intermarriage (Watkins and London 1994;Lieberson 2000;Gerhards and Hans 2009). In this article, we used these studies as a starting point but expanded the existing literature. We differentiated between explanations related to immigrants' context of origin and the cultural distance between their country of origin and the country of destination, on the one hand; and to their level of integration in the host country, on the other hand. Previous studies have investigated only one immigrant group; hence, they were not able to analyze the impact of the context of origin in naming systematically. Our sample consisted of immigrants from 41 countries. This respondent diversity provided us with sufficient feature variation in the context of origin to enable us to analyze the impact on the immigrants' naming practices. A number of conclusions could be drawn on the basis of the research findings presented here.
First, the overall theoretical model explains the differences in naming very well. In the stepwise logistic regression models, the pseudo R 2 increased to over 35.45 percent. Although this quantity cannot be interpreted as the straightforward percentage of explained variance (as would be the case had a linear regression analysis been performed) the pseudo R 2 did indicate that the model has good explanatory power.
Second, many of the hypotheses are supported by the results. The greater the cultural proximity between the home and host country and the better integrated immigrants in this study were in the host country (citizenship, emotional identification, education, German friends and acquaintances), the more likely they were to choose boundary-crossing strategies. Linguistic integration and intermarriage with a person from Germany, however, had no independent effects on name-giving after controlling for the other explanatory variables. In addition, the analyses found that the immigrants investigated were more likely to cross the symbolic boundary in naming when choosing a name for a girl.
Third, a comparison of the impact of the various explanatory factors on naming reveals that immigrants' religious background especially determined the strategies used when naming their children. We argued that religion and to a lesser extent language define the pool of names that are available for parents. Hence, immigrants coming from a historically Christian country were found to have a larger pool of potential first names that are common in the country of origin and at the same time in Germany. In such cases, the symbolic boundary between the two countries was blurred, which increased the likelihood of their opting for a boundary-crossing strategy. Cultural proximity in the sense of religious similarity had a stronger impact on naming strategies than the degree of integration in the host society.
There is probably a further mechanism that could shed light on why religion was found to be so important and why Muslim immigrant respondents, who mostly came to Germany from Turkey, were less likely than all the other groups investigated here to choose boundary-crossing strategies. A previous qualitative study conducted by the present authors examined 55 immigrants from different countries, reconstructed the immigrants' experiences with their own names in Germany, and explored their strategies for handling the symbolic boundary with the majority society (Gerhards and Buchmayr 2018). The study found that Muslim immigrants from Turkey and the Arab world were more likely to report discrimination than any other immigrant group, although many of them were structurally well-integrated into German society. Their feeling of being rejected by the majority society stood in stark contrast to their own perception of being well-integrated. Consequently, they developed a special strategy of boundary work. Frustrated by the lack of acceptance by German citizens, they increasingly oriented themselves towards their parents' culture of origin and chose their children's names from their parents' country (also see Skrobanek 2009). If the results of that study are combined with those of the present article, one can envisage the following mutually reinforcing process as a likely mechanism. Muslim immigrants from Turkey and Arabic countries, unable to find children's first names that are common in both their home and host counties due to the cultural distance, consequently, follow a strategy of boundary-maintenance in name-giving. The majority society interprets this behavior as a lack of willingness to adapt to the host society and reacts in a discriminatory manner. This, in turn, leads the immigrants to turn even further away from the first names that are common in the host country.
To conclude, we would like to point out some limitations of our analyses. A first limitation is related to the coding of our dependent variable. Unfortunately, we were not able to differentiate first names that are common in both the country of origin and destination. We were unable to determine whether parents opted for such a "hybrid" first name; or whether they chose a name that is indeed only common in the country of destination, but not in their home country. Strictly speaking, only the latter strategy can be considered boundary-crossing, while choosing a hybrid name is better understood as a strategy of boundary-blurring. Due to the data restrictions, we had to subsume both strategies under the concept of boundary-crossing. The lack of differentiation between boundary-crossing and boundary-blurring is a drawback of our analysis, as for some immigrants, more such "hybrid" names are available than for others due to differences in the overlap of the first-name pools between home and destination country. In addition, we assumed that the degree of cultural distance between the country of origin and the destination country affected the overlap in the pool of names available in both countries.
We measured cultural proximity by using two indicators: linguistic distance and religious affiliation. Unfortunately, direct measurement of the degree of overlap between the two name pools was not possible. Finally, one has to keep in mind that our quantitative study calculated correlations between different variables. We inferred from the statistical associations to the name-giving motives of immigrants without having directly analyzed the motives themselves. It is suggested that future research complement such quantitative analyses with qualitative investigations to shed more light on the motives that drive immigrants to choose particular names.

Notes
1 In 2018, the share of the population with an immigrant background in Germany amounted to 25.5 percent. A person is considered having a migration background when he or she or at least one parent does not have German citizenship. Most of the people with an immigrant background in Germany come from Turkey (14.4 percent), Poland (10.9 percent), Russia (7.2 percent), Italy, and Romania (4.5 percent each) (Bundeszentrale für politische Bildung 2018). Laws and regulations about first names in Germany are more restricted than in the US. Certain names are not permitted in Germany, such as 1) names that are the same as siblings' names; 2) offensive, ludicrous, or otherwise burdensome names as well as names taken from consumer products; and 3) names that are not gender-specific. Although since 2008 it has been possible to choose a gender-neutral first name, it must be accompanied by a further name which clearly identifies the sex of the child. First names may be rejected by the registry office if they do not conform to the regulations.
2 Although we acknowledge that these terms do not constitute linguistic antonyms, we sought to maintain the link to the scholarship in this field and therefore have kept the original terms.
3 Interested readers will be provided with an anonymized version of the data set for replication on request.
4 As supplementary coding could only be conducted with regard to the question of whether or not a name is common in Germany, it was unfortunately not possible to generate a third category of hybrid names that are common in both the home and host country. For this purpose, it would have been necessary to do the ex-post coding for all languages that are represented in the sample, which was beyond the scope of this project. 5 See http://www.elinguistics.net. For instance, the distance between Swedish and German is 25.0, between Polish and German 68.1, and between Turkish and German 95.5. 6 Unfortunately, the wording of the question is vague in another aspect. In the explanation provided for the question, it was specified that the three persons a respondent could mention could be non-relatives as well as relatives who live outside the respondent's household.