Professional Documents
Culture Documents
(Hamburg Studies On Linguistic Diversity) Joana Duarte, Ingrid Gogolin-Linguistic Superdiversity in Urban Areas - Research Approaches-John Benjamins Publishing Company (2013) PDF
(Hamburg Studies On Linguistic Diversity) Joana Duarte, Ingrid Gogolin-Linguistic Superdiversity in Urban Areas - Research Approaches-John Benjamins Publishing Company (2013) PDF
Editors
Peter Siemund Ingrid Gogolin
Linguistic Diversity Management in Linguistic Diversity Management in Urban Areas
Urban Areas (LiMA)
(LiMA) Comparative and Intercultural Education Research
English and General Linguistics University of Hamburg
University of Hamburg
Volume 2
Linguistic Superdiversity in Urban Areas. Research approaches
Edited by Joana Duarte and Ingrid Gogolin
Linguistic Superdiversity
in Urban Areas
Research approaches
Edited by
Joana Duarte
Ingrid Gogolin
University of Hamburg
List of contributors ix
Capturing superdiversity
1. Preamble
The new migration phenomena since the end of the Cold War, brought about by
increased globalisation and characterised by the intensification of migration ty-
pologies (in terms of countries of origin, language, ethnicity and religion; motives,
patterns and itineraries of migration; processes of integration into host communi-
ties, etc.) provoked a new discourse on the appropriateness of the multiculturalism
paradigm (Vertovec 2010). Critics of this paradigm that focu s on so-called “ethnic
minorities” (Blommaert & Rampton 2011) perceive it to be incapable of capturing
the actual diversity of current migration and as being responsible for social break-
down, class-based inequality and the increase in ethnic tensions (Vertovec 2010).
In this post-multicultural globalised era, the term superdiversity is about to re-
place the multiculturalism paradigm in its attempt to describe and apprehend
migration-induced phenomena of the last two decades (Vertovec 2006)1. Accord-
ing to Vertovec, “the time has come to re-evaluate – in social scientific study as
well as policy – the nature of contemporary diversity” (Vertovec 2009: 86)
The superdiversity framework relates to the image of “the world in one city”
which has, on occasion, been used to describe London’s growing diverse composi-
tion (Vertovec 2007, Vertovec & Baumann 2011). Observing the complex
phenomena of superdiversity was mostly carried out from sociological and an-
thropological perspectives. More recent research, however, has been conducted on
linguistic practices amongst multilingual speakers (Blommaert & Rampton 2011,
Creese & Blackledge 2010). Within this framework, little attention is given to the
repercussions of superdiversity (in particular concerning language use) on educa-
tional settings and systems; in many respects, we can argue that nowadays the
world exists within one school.
Our contribution aims at providing an overview of the consequences of lin-
guistic superdiversity for educational institutions. We will start by explaining the
concept of superdiversity, as presented by Vertovec. We then give an overview of
what has been described up to now as “linguistic superdiversity”. This perspective
collides, to a certain extent, with the majority of approaches to teaching and learn-
ing in schools with a “multilingual” population. In general, such approaches refer
to a binary concept of dealing with “bilingual” students, on one hand, and
“monolingual” students, on the other. We will illustrate the shortcomings of this
In the period since the early 1990s, there has been a rise in migration flows
(including refugees) worldwide. It is estimated that there are approximately 214
million migrants worldwide at present (Vertovec 2009). The number of places of
origin, the forms and aims of migration have become increasingly diversified,
giving rise to a so-called ‘diversification of diversity’ (Martiniello 2004). When
compared with the large immigrant groups that were identifiable in migration
movements of the 1950s to the 1970s, current migrant groups are smaller
in numbers, more mobile, socially more stratified and legally more differentiat-
ed. The term superdiversity has been used to designate these global changes in
migration flows and forms which have occurred in the past twenty years
(Vertovec 2006).
Traditional views on diversity assumed that the heterogeneity of migrants
could be adequately captured by separating them according to their country of
origin (Hopf 1987) or their ethnicity. In light of superdiversity, such methodologi-
cal categorisations make little sense. Vertovec (2006: 17) uses Somalis in the UK to
exemplify the inadequacy of such approaches: “[W]e will find British citizens,
refugees, asylum-seekers, persons granted exceptional leave to remain, undocu-
mented migrants, and people granted refugee status in another European country
but who subsequently moved to Britain. A simple ethnicity-focused approach to
understanding and engaging minority groups in Britain, as taken in many models
and policies within conventional multiculturalism, is inadequate and often inap-
propriate”. However, a methodological hurdle arises for research within super-
diverse contexts. On the one hand, the homogenisation of groups, which per se are
diverse, has to be avoided. On the other hand, by adding manifold variables such
as legal status, milieu, language, etc., the research designs become very complex –
possibly too complex for most forms of empirical research (Larsen-Freeman &
Cameron 2008).
From a sociological perspective, attempts have been made to capture the rising
differentiation of migration forms and practices through the concept of
“transmigration”. This can roughly be defined from the standpoint of migrants as
on-going migration processes (Pries 2004, Gogolin & Pries 2004). Gogolin and
Pries presented the variables influencing migration outcomes – such as the
Joana Duarte and Ingrid Gogolin
relationship to the countries of origin and destination, the reasons for migrating
and the envisaged length of stay in the new country – and characterised them vis-
à-vis the types of migrants, thus designing a typology of categories of current mi-
gration (see Table 1).
The concept of transmigration draws attention to the dynamism of migration
processes, resulting in an understanding of such movements as fluid and mallea-
ble. The rapid development of technical possibilities for instantaneous communi-
cation, as well as the growth in more cost-effective travel options, serve to reinforce
such assumptions. This understanding crosses the basic patterns ruling migration
regimes up to now, which focus on migration ending in an act of complete
“integration” whereby a former migrant transfers to a member of the group of
non-migrants2. The differentiation of migration is related to the variation in ways
of living, identification forms and social positioning processes in which different
variations of diversity partially overlap. Steven Vertovec has attempted to grasp
these phenomena and their consequences for the formation of current societies
with the notion of superdiversity.
Table 1. Four ideal types of international migrants (Gogolin & Pries 2004: 9)
2. Esser (2006) for example speaks of the integration of migrants in the sense of systemic
(general social system such as a society) and social integration (at the level of the individual
actors). In his concept, the idea of multiple integration of a migrant in both host community
and ethnic group is a rare phenomenon, which cannot happen in most migration situations.
Migrants can thus only assimilate (at least functionally) or live separately from the “host
community”.
Introduction
3. The author refers to Great Britain, but his conclusion can be applied to other European and
probably even more contexts.
4. For further explanation see also the research cluster “Globaldivercities”, directed by
Vertovec. Its core research question is: “In public spaces compared across cities, what accounts
for similarities and differences in social and spatial patterns that arise under conditions of diver-
sification, when new diversity-meets-old diversity?” http://www.mmg.mpg.de/research/all-
projects/globaldivercities/(16 February 2012).
Joana Duarte and Ingrid Gogolin
The concept of language within the superdiversity framework has entered the dis-
course on the linguistic texture of societies. Here, Blommaert and Rampton (2011)
identify three main areas where a revision or paradigm shift has taken place over
the last decades. These concern essential ideas about (a) languages themselves, (b)
language groups and speakers and (c) actual communication. The conceptualisa-
tion of research in these three areas has, according to the authors, been subject to
a fundamental change: “Rather than working with homogeneity, stability and
boundedness as the starting assumptions, mobility, mixing, political dynamics
and historical embedding are now central concerns in the study of languages, lan-
guage groups and communication” (2011: 4). This research rises in opposition to
what the authors define as notions of linguistic diversity still transmitted as a
“hegemonic force in public discourse, in bureaucratic and educational policy and
practice, and in everyday common sense, as well as in some other areas of lan-
guage study” (2011: 5), but considered non-operational in light of new superdi-
verse phenomena. An example of such an “archaic notion” still prevailing in many
of the aforementioned institutions is that of a “language” as a homogeneous and
clearly defined or definable object, which can be linked to a likewise identifiable
“people”. While this notion may no longer be reflected in state of the art research
on language phenomena, it prevails in the political sphere, educational practice
and as a common belief in European societies.
Concerning (a) languages: most research conducted under the “multicultural-
ism paradigm” (Vertovec 2010) has been based on traditional categories, whereby
a language is often associated with a given (usually monolingual) nation-state.
This association thus has consequences for the hierarchisation of languages in so-
ciety – accepting monolingualism as the rule implies that multilingual forms of
practice, particularly those that are migration-induced, acquire the status of devi-
ant or “illegitimate” practices (Gogolin 2007). In the superdiversity framework,
the conceptualisation of languages as “bounded systems linked to bounded com-
munities” (Blommaert & Rampton 2011: 5) is seen as an ideological artefact.
As regards (b) language groups and speakers, the superdiversity framework
calls for a shift towards research on “communities of practice” (Blommaert &
Rampton 2011) where linguistic repertoires of a fluid and fragmentary nature may
Introduction
features are relevant, such as the degree of privacy in the communication process
or the status of the speakers.
Blommaert and Backus (2011) argue that previous definitions of repertoires,
based on the triad of language resources, knowledge of language (“competence”)
and community, must equally be revised in light of the complexity of strategies and
channels through which (not only) migrants acquire language. According to the
authors, “(...) different learning modes lead to very different degrees of knowledge
of language, from very elaborate structural and pragmatic knowledge to elementary
“recognizing” languages, whereby all of these resources in a repertoire are function-
ally distributed in a patchwork of competences and skills” (2011: 2). In their typol-
ogy of language learning trajectories, a variety of factors is taken into account:
length of socialization, contexts of acquisition, purposes, types of languages learned,
and levels of competence. They thus speak of “superdiverse repertoires” (2011: 22).
Communicative practices have increasingly become the focus of research,
resulting in the coinage of new termini for characterising the dynamic use of lan-
guages by multilingual speakers (for a summary see Blommaert & Rampton 2011).
Examples of such denominations are “crossing” (Rampton 1995), “polylingualism”
(Jørgensen 2008) and “translanguaging” (Creese & Blackledge 2010, and in this
volume). While multilingual speakers actually engage in dynamic and fluid com-
municative practices, the discourses on such practices make use of “obsolete and
conclusively discredited models of language knowledge” (Blommaert & Backus
2011: 4).
An impressive compilation of research on communicative practices in super
diverse settings is to be found in the volume The Languages of Global Hip Hop
(Terkourafi 2010), in which communication in the hip hop genre across the world
is examined. Another example is the investigation of the role of migrant languages
in German hip hop lyrics by Androutsopoulos (Androutsopoulos 2010). He
reaches the conclusion that the use of “a mixture” of languages – German, English
and minority languages – as well as the purpose of language choice and mixing are
conducted according to “a complex mobilization of linguistic resources that goes
beyond code-switching in a narrow sense” (2010: 37). In his analysis and albeit the
many differences found in the lyrics of the corpus, he concludes that the distribu-
tion of migrant languages is of a symbolic nature as they work as “ethnic identity
symbols” (2010: 39) rather than as contributing to content. Androutsopoulos sug-
gests that this variant of language switching and mixing cannot be interpreted as
spontaneous and a fluid private phenomenon, but must be taken as signs of public
discourse.5
5. For one of the first attempts to uncover the systematic status and significance of such com-
munication see Hewitt (1986).
Introduction
6. In some European states, such as Belgium or Switzerland, we find the concept of monolin-
gualism transferred to the regions (see Mettewie & Janssens 2007: 117–143 for a discussion on
the situation in Belgium).
Introduction
focused on “general” linguistic abilities that are suited to everyday colloquial com-
munication. The German model programme, FörMig (Förderung von Kindern
und Jugendlichen mit Migrationshintergrund), advocates a change of focus to-
wards the academic register, i.e. “Bildungssprache” (Gogolin et al. 2011). Here
another feature of linguistic diversity is addressed: the difference of registers with-
in a language to which learners have to adjust themselves. Research has confirmed
that a general linguistic proficiency is less responsible for better school outcomes
than the mastery of a particular register referred to as “academic language”
(Leseman et al. 2007; Leseman et al. 2009; Thomas & Collier 1997)7.
Jim Cummins (Cummins 2000; Cummins 2002) first described the differ-
ences of both registers and their functions for school careers. He stated that ev-
eryday language proficiency – basic interpersonal communication skills (BICS)
– is usually acquired by a second language learner who enters a new linguistic
Lebenswelt relatively easy and fast. This type of discourse is typical of familial
and informal interactions; it is highly contextualised and can be managed with
little cognitive effort. The acquisition of CALP (cognitive academic language
proficiency) skills on the other hand, takes a longer period of time. These skills
comprise explicit discourse forms, largely decontextualised language and demand
a higher cognitive involvement of the speaker. Studies which have examined the
acquisition of academic language skills show that this process can take five to
eight years in the case of second language learners (Thomas & Collier 1997;
Hakuta et al. 2000). In comparison, the acquisition of colloquial competence by
second language learners only takes between six months and two years (MacSwan
& Pray 2005). Cummins originally focused on this aspect of linguistic diversity
from the viewpoint of a learner who enters a new linguistic environment and its
school. “Bildungssprache” goes further by addressing the fact that the specific
registers of school and learning may also be a hurdle for learners who, in a con-
ventional sense, are considered monolingual – given the fact that registers used in
private and those used in educational discourse differ greatly. Here, the aspect of
cultural capital of the child’s environment takes effect: the closer the language use
of a family is to the standard, and the more it involves literacy practices, the small-
er the gap that a child has to bridge in order to acquire the register of a “successful
student”.
7. For the German discourse, the term “Bildungssprache” was introduced by the FörMig-
team as the more appropriate expression than “academic language”. This is due to the fact that
“academic”/“akademisch” in German is closely related to education in the tertiary sector,
namely university.
Introduction
One of the approaches that was established in different education systems with the
aim of coping with linguistic diversity is the creation of so-called bilingual educa-
tion models (for a general discussion see Gogolin 2011). Such models are found in
a variety of schemes in many countries of the world, and their effectiveness is
highly contested (see Gogolin & Neumann 2009). They represent an attempt to
react to non-monolingual students’ linguistic “needs” by trying to cope with lan-
guage diversity within a conceptual framework of monolingualism.
Here we provide an example of bilingual school models that were established
in the German Federal States of Hamburg and Saxony. Our research group carried
out empirical observations of the models, and the following illustration is based
on this research (Gogolin et al. 2004; Roth et al. 2007; Duarte 2011).
Bilingual classes in Hamburg were established in the year 2000, in a small
number of schools and in one or two of the parallel classes in each participating
school – the other parallel classes follow the “regular” monolingual curricula. In
the “bilingual” classes, students from different linguistic backgrounds get the op-
portunity to become bilingual and biliterate, and to additionally receive English as
their first “foreign language” as it is regularly taught in Hamburg from grade one
(Roth et al. 2007). The language pairs of the schools are Italian-German, Spanish-
German, Portuguese-German and Turkish-German. This model came into exis-
tence in 1999, based on a joint initiative of the Ministry of Education of the State
of Hamburg8 and the Consulates of Italy, Spain, Portugal and Turkey. The arrange-
ment indicates that this bilingual model is based on the concept of a quasi-natural
relation between a state and “its” language.
Consequently, classes were designed for students of migrant background, de-
riving from the respective country of origin that co-finances the model, as well as
for German monolingual children, aiming at a 50:50 composition. In reality, this
was never met, largely due to the composition of the population in Hamburg. The
classes were composed of German-only students, a number of bilingual students
with more or less advanced knowledge of the so-called partner language and a
number of bi- or multilingual students with other heritage languages. For some of
these students, the schools can be classified as dual language models or two-way
bilingual programs, where two groups from a given society learn each other’s
8. Hamburg is one of the 16 German Federal States which are independently responsible for
the education systems; see for an overview of the German education system: http://www.kmk.
org/information-in-english/the-education-system-in-the-federal-republic-of-germany.html.
Joana Duarte and Ingrid Gogolin
heterogeneous population. These were similar to the ones found for some of the
Hamburg schools and will be summarised here.
Several of the Hamburg schools developed didactic arrangements that are
speaker-centred and reflect pupils’ multilingual realities, thus intending to explore
their heteroglossic repertoires and not reduce them to a bilingual dichotomy. In
some of the schools, this was successfully achieved by scaffolding pupils from every
day colloquial registers to academic language, by using varieties or dialects as use-
ful bridges, or in allowing and encouraging several forms of multilingualism in the
classroom (through materials such as dictionaries, but also in terms of language
comparison and transfer; see Duarte 2011). As a result, pupils with other languag-
es not included in the project could also actively use them in class. In other cases,
however, a negative attitude towards dialectal varieties and the use of other lan-
guages was registered (Dirim et al. 2009). In these schools, language choice was
kept to a minimum and there were strong regulations on which language to use
where, when and with whom.
Cross-linguistic communicative practices of all sorts, as productive resources
rather than as a sign of linguistic deficits, have also been found to be a key feature
in adjusting to growing diversity in dual language programmes (Purkarthofer &
Mossakowski 2011). Instances of code-switching in content matter, as well as lan-
guage classes (foreign or not), were accepted in some schools as useful resources
for learning and as a means of ensuring active participation in classroom dis-
course. Although such practices tended to rapidly disappear in general classroom
discourse, they remained common practice in group or pair work. Similar to the
Carinthian schools, language use in some of the Hamburg schools was regulated
by respecting a balance between pupils’ language choice and a proactive support
for the use of the other languages. This sort of approach which allows language
learning across language boundaries (García et al. 2006), aims at fostering meta-
linguistic knowledge through repeated language comparison.
In addition, and also similar to the successful Carinthian schools, some of
the bilingual schools in Hamburg documented language use and proficiency of
their speakers over time by applying a variety of different methods. Some made
use of selected materials available in the European Language Portfolio, while
others introduced language diaries, for instance. This allowed for progress to be
visible for all and for a common method of establishing new aims for language
development.
In the most successful bilingual schools, a wider choice of foreign languages
was also registered. In addition to English and French, some schools added to their
curriculum the learning of typical immigrant languages, particularly throughout
secondary education. The Portuguese-German school, for example, offered
Turkish and Spanish classes. Some courses in the afternoon were open for teachers
Joana Duarte and Ingrid Gogolin
and parents. This measure opens the school profile to a sort of multilingualism
closer to that of the surrounding society, on the one hand, and helps in attracting
a more heterogeneous school population on the other, as school profiles with a
larger amount of foreign languages may be attractive for parents from a range of
social classes.
This aspect is linked to another important issue of dual language models in
their adaptation to linguistic superdiversity; the attempt to construct teachers as
role models and competent speakers of many languages, thus avoiding a concep-
tion of teachers as monolingual speakers (Balboni 1997). As Purkarthofer and
Mossakowski (2011) put it, their role is to reinforce the students’ self-construction
as proficient multilingual speakers and not to further sustain the myth of the
monolingual habitus. In some of the Hamburg schools, teachers were bi- or mul-
tilingual themselves and were encouraged to make use of their linguistic reper-
toires for instruction. In the Portuguese-German case, for example, the Portuguese
teacher was also responsible for the English classes (and later also for German),
turning these language classes into multilingual spaces of language comparison
and co-construction. Often, and depending on the programme, she would not
separate classes according to the language, but use time slots to teach one aspect
through two or three languages.
This approach to language teaching was also reflected in the attitudes towards
language separation. Again, similar to the findings of the Carinthian model,
Hamburg schools opted for a time- or situation-based language use policy and not
for a person-oriented language use. This, of course, only worked in cases where
teachers were proficient in two or more languages. In some schools where the
same teacher used different languages, depending on time or situation, language
changed every week; in others it depended on the room where the class took place.
According to Purkarthofer and Mossakowski (2011) this system offers more rou-
tine and malleability in learning through different languages of instruction.
The evaluation of the bilingual schools in Hamburg (Duarte 2011; Roth et al.
2007; Gogolin et al. 2004) showed that there were no achievement gaps for bilin-
gual pupils with language combinations not matching those of the project. In some
cases, particularly in schools where measures for dealing with linguistic diversity
were practiced, these pupils performed exceptionally well. However, due to rela-
tively small sample size, no conclusive assertions on school performance can be
made. This would definitely constitute an aspect for further research.
To summarise, traditional bilingual models, although mostly based on mono-
lingual norms and occasionally reflecting traditional views of bilingual speakers as
“two monolinguals”, can in fact adapt to growing linguistic superdiversity. How-
ever, this means overcoming the usual double monolingual habits found in these
models and moving towards methods already known from successful schools in
Introduction
dealing with diversity (Gogolin et al. 2011). In some of the Hamburg schools, these
approaches have been found, while in other cases the traditional model persisted.
In general, the schools that adopted the measures described in this article were
more successful in compensating for language problems and in improving school
outcomes of all pupils (for an overview of the Portuguese-German case, see
Duarte 2011).
contribution along three main trends for the study of dialectal variation and pro-
vides empirical results for each, as well as the most marked methodological
features. Several aspects that are relevant to the broader study of languages in su-
perdiverse constellations are discussed with respect to dialects in urban areas, con-
sidering issues of code-switching, borrowing, variability and dialect loss, as well as
dialect acquisition, or the stability of dialect boundaries. The author shows that
ethnolect variation is heterogeneous and not only dependent on substrate features,
and he calls for the inclusion of contextual features in studies of variation: per-
sonal aspects such as language attitudes as well pragmatic (e.g. appropriateness of
use) and socio-cultural aspects (e.g. education, language policy). This is compatible
with the superdiversity framework, which refers to the complex interplay of vari-
ables that constitute diversity.
The contribution of Suzanne Aalberse and Pieter Muysken addresses the con-
troversial issue of language proficiency in relation to the age of acquisition from
the perspective of heritage language speakers. Similarly to Kapia’s contribution,
the issue of norm when evaluating bilinguals is also discussed and reflected against
a background of other sociological and educational aspects that influence profi-
ciency or loss of heritage languages. The contribution tackles the important issue
of vulnerability and attrition of heritage languages, as well as the consequences
this brings for individuals, educational institutions and societies.
Against a background of the sociolinguistics of globalization, Sjaak Kroon, Jan
Blommaert and Dong Jie, on the last contribution of the volume, focus on language
use in China, taking order notices in Beijing as an example. In a combination of
two perspectives – language for mobility/mobile people and languages – the con-
tribution presents several examples of the presence of English in the Beijing public
space. New hybrid forms of language use are discussed in light of two theoretical
concepts within the sociolinguistics of globalization: the need to move from
language(s) to varieties or repertoires and the need to reflect on language
hierarchies within mobile societies, as well as on the ways these (re)shape social
positioning.
All contributions to the volume reinforce the need for, as well as profit from,
international comparative research, and all show the advantage of interdisciplin-
ary approaches in research on language in superdiverse settings. Furthermore, the
relevance to include a historical perspective in the studied phenomena is
illustrated, not least with respect to the necessity to reflect upon the research para-
digms which explicitly or implicitly frame an approach. The dynamics of superdi-
verse constellations calls for the constant control of research paradigms and con-
cepts which may owe their evidence to a past historical constellation.
Introduction
5.1 Acknowledgements
Twelve papers cannot be brought into a coherent shape without the assiduous as-
sistance of collaborators. We would like to extend our gratitude to Juliane Lewin
and João Ferreira Duarte for struggling with and finally overcoming a plethora of
inconsistencies and idiosyncrancies. Moreover, we wish to thank all referees who
participated in the reviewing process. Their comments helped enormously to
increase the quality of the volume. Last but not least, we would also like to ac-
knowledge the financial aid made available by the Behörde für Wissenschaft und
Forschung and the University of Hamburg.
References
Friedrich-Ebert-Stiftung, Abt. Wirtschafts- und Sozialpolitik Friedrich (ed.) 2007. Schule in der
Einwanderungsgesellschaft. Gesprächskreis Migration und Integration. Dokumentation im
Auftrag der Friedrich-Ebert-Stiftung. Bonn.
Androutsopoulos, J. 2010. Multilingualism, ethnicity and genre in Germany’s migrant Hip Hop.
In The Languages of Global Hip Hop, M. Terkourafi (ed.), 19–43. London: Continuum.
Balboni, P.E. 1997. Ladino-Italian bilingual programs in Italy. In Bilingual Education
[Encyclopaedia of Language and Education 5], J. Cummins & D. Corson (eds), 143–50.
Dordrecht: Kluwer.
Baumert, J., Stanat, P. & Watermann, R. 2006. Herkunftsbedingte Disparitäten im Bildungswesen:
Vertiefende Analysen im Rahmen von PISA. Wiesbaden: VS Verlag.
Blommaert, J. & Backus, A. 2011. Repertoires revisited: “Knowing language”. In Superdivertsity.
Working Papers in Urban Language & Literacies 67. London: King’s College. <www.kcl.
ac.uk/innovation/groups/ldc/publications/workingpapers/67.pdf>
Blommaert, J. & Rampton, B. 2011. Language and superdiversity: A position paper [Working
Papers in Urban Language & Literacies 70]. London: King’s College. <www.kcl.ac.uk/in-
novation/groups/ldc/publications/workingpapers/70.pdf>
Bos, W., Hornberg, S., Arnold, K.-H., Faust, G., Fried, L., Lankes, E.-M., Schwippert, K. & Valtin,
Renate (eds) 2007. IGLU 2006. Lesekompetenzen von Grundschulkindern im internationalen
Vergleich. Münster: Waxmann.
Bos, W., Bonsen, M., Baumert, J. Prenzel, M., Selter, C. & Walther, G. 2008. TIMSS 2007.
Mathematische und naturwissenschaftliche Kompetenzen von Grundschulkindern in
Deutschland im internationalen Vergleich. Münster: Waxmann.
Bybee, J. 2006. From usage to grammar: The mind’s response to repetition. Language 82(4):
711–733.
Cenoz, J. & Gorter, D. 2006. Linguistic landscape and minority languages. International Journal
of Multilingualism 3(1): 67–80.
Council of the European Union 2009. Council conclusions of 12 May 2009 on a strategic frame-
work for European cooperation in education and training (ET 2020). Official Journal C
119: EURLex, 0002 – 0010.
Creese, A. & Blackledge, A. 2010. Translanguaging in the bilingual classroom: A pedagogy for
learning and teaching? The Modern Language Journal 94(1): 103–115.
Joana Duarte and Ingrid Gogolin
Cummins, J. 2000. Language, Power and Pedagogy. Bilingual Children in the Crossfire. Clevedon:
Multilingual Matters.
Cummins, J. 2002. BICS and CALP. In Encyclopedia of Language and Teaching, M. Byram (ed.),
76–79. London, Routledge.
Dehn, M. 2011. Elementare Schriftkultur und Bildungssprache. In Migration und schulischer
Wandel: Mehrsprachigkeit, S. Fürstenau & M. Gomolla (eds), 129–151. Wiesbaden: VS
Verlag.
Deutsches PISA-Konsortium 2001. Basiskompetenzen von Schülerinnen und Schülern im inter-
nationalen Vergleich. Opladen: Leske + Budrich.
Diefenbach, H. 2010. Kinder und Jugendliche aus Migrantenfamilien im deutschen Bildungssys-
tem. Erklärungen und empirische Befunde. Wiesbaden: VS Verlag.
Dirim, I., Döll, M., Neumann, U. & Roth, H.-J. 2009. Abschlussbericht über die Türkisch-
Deutschen Modellklassen. Hamburg: Hamburg University. <http://www2.erzwiss.uni-
hamburg.de/personal/neumann/Bericht-2009-bilinguale%20Grundschulklassen.pdf>
Duarte, J. 2011. Bilingual Language Proficiency. A Comparative Study. Münster: Waxmann.
Esser, H. 2006. Sprache und Integration. Die sozialen Bedingungen und Folgen des Spracherwerbs
von Migranten. Frankfurt: Campus.
Extra, G. 2008. Immigrant languages in multicultural Europe: Comparative perspectives. In Wieser
Encyclopaedia, U. Ammon & E. Haarmann (eds), 489–519. Klagenfurt: Wieser Verlag.
Flick, U. 2011. Triangulation in der Bildungsforschung. Zum Stand der Diskussion – Aktualität,
Ansätze und Umsetzungen der Triangulation. In Triangulation in der Bildungsforschung,
J. Ecarius & I. Miethe (eds), 19–40. Opladen: Budrich.
García, O., Skutnabb-Kangas, T. & Torres-Guzman, M. (eds) 2006. Imagining Multilingual
Schools. Languages in Education and Glocalization. Clevedon: Multilingual Matters.
Göbel, K., Vieluf, S. Hesse, H.-G. 2010. Die Sprachentransferunterstützung im Deutsch- und
Englischunterricht bei Schülerinnen und Schülern unterschiedlicher Sprachlernerfahrung.
Zeitschrift für Pädagogik Beih., 55: 101–122.
Gogolin, I. 1987. “Muttersprache”: Zweisprachigkeit. Sprachliche Bildungsvoraussetzungen der
Kinder aus ethnischen Minderheiten. Pädagogische Beiträge 12: 26–30.
Gogolin, I. 1994. Der monolinguale Habitus der multilingualen Schule. Münster: Waxmann.
Gogolin, I. 2007. Sprachlich-kulturelle Differenz und Chancengleichheit. In Bildung – Lernen.
Humanistische Ideale, gesellschaftliche Notwendigkeiten, wissenschaftliche Erkenntnisse,
D. Lemmermöhle & M. Hasselhorn (eds), 167–182. Göttingen: Wallstein.
Gogolin, I. 2011. Bilingual education. In The Routledge Handbook of Applied Linguistics,
J. Simpson (ed.), 229–242. London: Routledge.
Gogolin, I., Neumann, U. & Roth, H.-J. 2004. Bericht 2004: Stand der Untersuchungen,
Schulorganisatorische und didaktische Perspektiven, Leistungsbeurteilung, Schwerpunkt:
Biliteralität – Schreiben und Lesen in zwei Sprachen. Hamburg: Hamburg University.
<http://www.epb.uni-hamburg.de/files/Bericht2004.pdf>
Gogolin, I. & Pries, L. 2004. Stichwort: Transmigration und Bildung. Zeitschrift für Erziehun-
gswissenschaft 7(1): 5–19.
Gogolin, I. & Neumann U. 2009. Streitfall Zweisprachigkeit. The Bilingualism Controversy.
Wiesbaden: VS Verlag.
Gogolin, I., Dirim, I., Klinger, T., Lange, I., Lengyel, D., Michel, U., Neumann, U., Reich, H. H.,
Roth, H.-J. & Schwippert, K. 2011. Förderung von Kindern und Jugendlichen mit Migration-
shintergrund FÖRMIG. Bilanz und Perspektiven eines Modellprogramms [FÖRMIG-Edition
vol. 7]. Münster: Waxmann.
Introduction
Hakuta, K., Butler, Y. & Witt, D. 2000. How Long Does it Take English Learners to Attain Proficiency?
Santa Barbara CA: University of California Linguistic Minority Research Institute.
Hewitt, R. 1986. WhiteTalk Black Talk: Inte-racial Friendship and Communication amongst
Adolescents. Cambridge: CUP.
Hobsbawm, E.J. 1990. Nations and Nationalism since 1780: Programme, Myth, Reality. Cambridge:
CUP.
Hopf, D. 1987. Herkunft und Schulbesuch ausländischer Kinder. Berlin: Max-Planck-Institut für
Bildungsforschung.
Hult, F.M. 2010. Swedish Television as a mechanism for language planning and policy. Language
Problems and Language Planning 34(2): 158–181.
Jørgensen, N. 2008. Poly-lingual languaging. Evidence from Turkish-speaking youth. In:
Multilingualism and Identities across Contexts. Cross-Disciplinary Perspectives on Turkish-
Speaking Youth in Europe, V. Lytra & N. Jørgensen (eds), 129–150. Copenhagen: University
of Copenhagen.
Klieme, E. et al. (ed.) 2010. PISA 2009: Bilanz nach einem Jahrzehnt. Münster: Waxmann.
Larsen-Freeman, D. & Cameron, L. 2008. Complex Systems and Applied Linguistics. Oxford:
OUP.
Leech, N. & Onwuegbuzie, A. 2009. A typology of mixed methods research designs. Quality and
Quantity 43(2): 265–275.
Leseman, P.P.M., Scheele, A., Mayo, A. & Messe, M. 2007. Home literacy as special language
environment to prepare children for school. Zeitschrift für Erziehungswissenschaft 10(3):
334–355.
Leseman, P.P.M., Scheele, A., Mayo, A. & Messe, M. 2009. Bilingual development in early
childhood and the languages used at home: competition for scarce resources? In Streitfall
Zweisprachigkeit – The Bilingualism Controversy, I. Gogolin & U. Neumann, 289–316.
Wiesbaden: VS Verlag.
Levitt, P. & Glick Schiller, N. 2004. Conceptualizing simultaneity: a transnational social field
perspective on society. International Migration Review 38(3): 1002–1039.
Lindholm-Leary, K.J. 2000. Dual language Education. Clevedon: Multilingual Matters.
Lynn, P. 2009. Methodological Research for Longitudinal Surveys. Southern Gate: John Wiley &
Sons.
MaCall, L. 2005. The complexity of intersectionality. Signs: Journal of Women in Culture and
Society 30(3): 1771–1800.
MacSwan, J. & Pray, L. 2005. Learning English bilingually: Age of onset of exposure and rate of
acquisition of English among children in a bilingual education program. Bilingual Research
Journal 28(3): 687–712.
Martiniello, M. 2004. The many dimensions of Belgian diversity. Canadian Diversity/Diversité
Canadienne 3(2): 43–46.
Mettewie, L. & Janssens, R. 2007. Language use and language attitudes in Brussels. In Multilin-
gualism in European Bilingual Contexts: Language Use and Attitudes, D. Lasagabaster &
A. Huguet (eds), 117–143. Clevedon: Multilingual Matters.
Mühlhäusler, P. 2003. Language of Environment – Environment of Language. A Course in
Ecolinguistics. London: Battlebridge.
Müller, A. G. 2007. Aspekte schulbezogener Sprache als Barriere für schulischen Erfolg von
Schülerinnen und Schülern deutscher und nichtdeutscher Herkunftssprache. PhD
dissertation, Freie Universität Berlin.
Joana Duarte and Ingrid Gogolin
Müller, N., Kupisch, T., Schmitz, K. & Cantone, K. 2006. Einführung in die Mehrsprachigkeits-
forschung. Deutsch – Französisch – Italienisch. Tübingen: Narr.
OECD 2010. OECD Reviews of Migrant Education – Closing the Gap for Immigrant Students:
Policies, Practice and Performance. Paris: OECD.
Pries, L. 2004. Transnationalism and migration: New challenges for the social sciences and
education. In Migration, Education and Change, S. Luchtenberg (ed.), 15–39. London:
Routledge.
Purkarthofter, J. & Mossakowski, J. 2011. Bilingual teaching for multilingual students? Innova-
tive dual-medium models in Slovene-German schools in Austria. International Review of
Education 57: 551–565.
Rampton, B. 1995. Language crossing and the problematization of ethnicity and socialization.
Pragmatics 5(4): 483–513.
Roth, H.-J., Neumann, U. & Gogolin, I. 2007. Bericht 2007: Abschlussbericht über die italie-
nisch-deutschen, portugiesisch-deutschen und spanisch-deutschen Modellklassen.
Hamburg: Hamburg University. <http://www.epb.uni-hamburg.de/files/Bericht2007.pdf>
Schnepf, S. 2007. Immigrants’ educational disadvantage: an examination across ten countries
and three surveys. Journal of Population Economics 20(3): 527–545.
Schrauf, R.W. 2009. Longitudinal designs in studies of multilingualism. In Language Develop-
ment across the Lifespan, K. de Bot & R.W. Schrauf (eds), 245–70. New York NY: Routledge.
Shohamy, E., Ben-Rafael, E., & Barni, M. (eds.) 2010. Linguistic Landscape in the City. Bristol:
Multilingual Matters.
Stanat, P. 2003. Schulleistungen von Jugendlichen mit Migrationshintergrund: Differenzierung
deskriptiver Befunde aus PISA und PISA-E. In PISA-2000 – Ein differenzierter Blick auf
die Länder der Bundesrepublik Deutschland, J. Baumert et al. (eds), 243–260. Opladen:
Leske + Budrich.
Stanat, P., Rauch, D. & Segeritz, M. 2010. Schülerinnen und Schüler mit Migrationshintergrund. In
PISA 2009: Bilanz nach einem Jahrzent, E. Klieme et al. (eds), 200–30: Münster, Waxmann.
Terkourafi, M., (ed.) 2010. The Languages of Global Hip Hop. London: Continuum.
Thomas, W.P. & V. Collier 1997. School Effectiveness for Language Minority Students. Washington
DC.: National Clearing House for Bilingual Education. 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>
Vertovec, S. 2006. The Emergence of Super-Diversity in Britain. Oxford: Centre of Migration,
Policy and Society.
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>
Vertovec, S. 2010. Towards post-multiculturalism? Changing communities, conditions and
contexts of diversity. International Social Science Journal 199: 83–95.
Vertovec, S. & Baumann, G. 2011. Multiculturalism in the Public Sphere: City and School, Mar-
kets and Media. London: Routledge.
Capturing superdiversity
Using correspondence analysis to model
immigrant multilingualism over time
Robert W. Schrauf
Pennsylvania State University
Modeling multilingualism
Basic concepts
A field of interactive elements lends itself to representation and analysis via the
cross-tabulation of data in contingency tables. In correspondence analysis, the cell
values in the rows of a contingency table are computed as proportions of their
rows totals (and similarly for columns). By dividing each cell value in a row by the
row total, a vector of numbers called the row profile is generated. A row mass is
calculated by dividing the total frequencies in each row by the total frequencies of
the whole table. This latter is an indication of the proportional importance of each
row relative to the whole table, and of course their sum is equal to 1.00. The aver-
age row profile, called the centroid, is computed by dividing the total of cell values
in each column by the total for the table as a whole, and this vector represents the
‘homogeneity assumption’ or the ‘no association assumption.’ Given a basic con-
tingency table, the chi-square can be calculated for the table as a whole. If the ob-
served frequencies differ markedly from expected frequencies (given the degrees
Robert W. Schrauf
of freedom), then we may conclude that the pattern of observed values is signifi-
cantly different from the ‘homogeneity’ or ‘no difference’ assumption.
To generate the graphic representation of row and column profiles in
n-dimensional coordinate space, the chi-square distance is used. This is a weighted
Euclidean distance which standardizes the variances, so that in the mapping of
points large differences between larger proportions do not swamp the effect of
small differences between smaller proportions. The variance of row profiles around
the centroid is expressed as inertia. In simplest terms, inertia is the chi-square sta-
tistic divided by the total frequencies in the table. The larger the inertia (variance) the
more the row profiles differ from the centroid or average row profile.
Given chi-square differences between each pair of points, it then becomes possible
to map the various row and column profiles in n-coordinate space. However, the
advantage of correspondence analysis lies in markedly reducing the high dimen-
sionality necessary for plotting numerous row profiles to a more visually intuitive,
low dimensional space – usually in two or three dimensions. By analogy to regres-
sion, in which the least squares criterion is used to fit a trend line to a series of
points, correspondence analysis uses a similar metric to find a plane or cube that
cuts n-dimensional space in such a way that the profile points may be projected
onto it with as little distortion as possible. Each of the two or three resulting
dimensions ‘accounts for’ some proportion or percentage of the total inertia. The
proportion of total inertia accounted for by any one dimension may be further
broken down by the contributions made by each point to that particular dimension,
and of course points may make different contributions to different dimensions.
The end result is a map that represents geometrically the associations between
row profiles on given dimensions in two- or three-dimensional space. The mean-
ings of the dimensions are not given by the statistical procedure but must be char-
acterized by the researcher. Clues to the meaning of a dimension come from the
relative inertia or contributions made by particular points to that dimension.
Typically, points with highest inertia can be found at opposite ends of the dimen-
sions (right-to-left, or top-to-bottom), and the semantic opposition between these
can be used to interpret the dimension. In sum, the researcher interprets the re-
sults of a correspondence analysis by (1) testing whether the observed pattern dif-
fers significantly from the homogeneity assumption, via the chi-square statistic,
(2) determining the number of dimensions necessary to represent the row profiles
in space, via inspection of the shared variance accounted for by the first few eigen-
values, (3) assessing the relative importance of each dimension, via the percentage
of inertia accounted for by each dimension, (4) interpreting the meaning of each
Using correspondence analysis to model immigrant multilingualism over time
Contingency tables that cross-tabulate more than two variables are analysed via
multiple correspondence analysis (MCA). Statistically, MCA is based on a symmetric
matrix in which all variables are cross-tabulated with all other variables, with sub-
matrices along the diagonal representing the values of each variable cross-tabulated
with itself (called a Burt Matrix). Greenacre (2007: 141–144) notes that percentages
of inertia are artificially low and tend to underestimate the true value of the maps.
MCA facilitates the analysis of the inter-relation of variables, attending to the qual-
ity and strengths of association between them. It is also important to note that the
initial analysis can be supplemented by other variables of interest that, strictly speak-
ing, may not have the same structure (i.e. response categories) or substantive focus
as the initial field. Logically enough, these are called supplementary variables, and
they are fitted to the original solution without affecting the inertia of that original
solution. Thus, for instance, having used MCA to structure the field of proficiency-
and-use, we might fit additional biodata, such as ‘where language learning took
place,’ or social data such as ‘neighbourhood linguistic diversity,’ to the total map.
In the adaptation of multiple correspondence analysis called geometric data
analysis (Le Roux & Rouanet 2004; Rouanet, Ackermann, & Leroux 2000), the
points on the map are referred to as ‘clouds’ and divided into the cloud of modali-
ties and the cloud of individuals. Modalities (traditionally variables) may be active
modalities, encompassing the primary field of interest, and passive modalities,
which are the supplementary points (understood to be ‘explanatory’ in some
sense). The participants whose data is represented by the modalities can also be
placed on the map, and these form the cloud of individuals.
The data used in this chapter come from a larger study on bilingual autobiographi-
cal memory (Schrauf 2009a, Schrauf & Hoffman 2007; Schrauf & Rubin 2001,
2004). Sixty older Spanish-English bilingual Puerto Ricans living in ethnic neigh-
bourhoods in Chicago, Illinois, had a mean age of 69.35 (SD = 6.90), and had emi-
grated from the island of Puerto Rico to the US at a mean age of 22.13 (SD = 7.09).
On average, they had 7.12 (SD = 3.95) years of formal education. They filled out a
language background questionnaire in which they provided sociodemographic and
Robert W. Schrauf
other information. The field of English language-use was queried via self-ratings of
use of English and Spanish on likert scales (1 = none, 5 = all the time) with specific
classes of interlocutors (spouse, children, in-laws, own family, friends, neighbours,
and workmates) and in specific mental activities (writing a note to oneself, thinking
in words, expressing feelings, dreaming, swearing, telling jokes, doing math, and pray-
ing). In the analyses below, the data include only ratings of English (and not
Spanish) in social domains (7 domains) and self-directed speech (9 domains), for a
total of 16 variables. Reported years of education were transformed into a categori-
cal variable (<4 years, 4–8 years, 9–12 years, >12 years). Participants answered one
question on how they acquired English (streets, work, school). They completed the
Bicultural Puerto Rican Scale (Cortes, Rogler & Malgady 1994) as a measure of ac-
culturation, with separate scales for American acculturation (high, medium, low)
and Puerto Rican acculturation (high, medium, low). Participants’ postal codes, in-
dexed to the US Census (2000), were used to determine percentage of Hispanics in
their neighbourhoods of residence (1–25%, 26–50%, 51–75%, and 76–100%).
Participants also completed the Adult Language Assessment Scales (ALAS) (Duncan
& De Avila 1991), which assess English comprehension and production via picture
naming, generation of antonyms, sentences describing line drawings, comprehen-
sion of a recorded conversation, and production of a summary of a recorded
‘newscast.’ The ALAS scores were grouped into beginner, low intermediate, high in-
termediate, and fluent categories (according to test manual recommendations).
The analysis proceeds in four steps. First, all the data are scaled and mapped,
with attention to inertia values and interpretation of the dimensions. Second, the
maps of modalities and supplementary points are generated, again with attention to
the interpretation of dimensions. Finally, the map of individuals is generated. Using
the location of particular individuals relative to the modalities, we can describe in
more detail how particular cases or life-histories are related to the whole field.
A two-dimensional solution, with total inertia of 3.75, explains 52.88% of the in-
ertia (variance). The mapped points are too dense to allow for meaningful inter-
pretation, and so we will systematically decompose the map to interpret it. Suffice
for now to note that the points describe a kind of horseshoe or arch effect, and this
is common in correspondence analysis (Greenacre 2007: 127). In general, the total
solution (Figure 1) shows that higher levels of English-use are concentrated to-
wards the right of the map, with ratings of 4s and 5s in the far upper right. A very
dense cloud of points occupies the centre of the map, with lowest ratings in the
upper left. This right-to-left distribution is easily interpreted as degree of language
Using correspondence analysis to model immigrant multilingualism over time
competence-in-use, and it accounts for the majority of the inertia in the two di-
mensional solution (39.73% out of the total 53.36%). However, it is much less clear
what the second dimension (along the y-axis) might represent. This is not an un-
common situation in correspondence analysis.
The dense cloud of points in Figure 1 makes the graph difficult to interpret,
and therefore Figure 2 shows only the extreme values for each query (e.g. 1s and 5s
for the likert scales, only ‘low intermediate’ and ‘fluent’ scores on the ALAS, only
‘high’ and ‘low’ American acculturation scores, etc) as these are found within the
circles (A and B) on Figure 1. (Note that the x- and y-axes have been shortened on
Figure 2 relative to Figure 1 to make the groupings more salient). Thus, in the area
labelled A, falling from the upper- to the lower-left quadrants, are found the low
values of self-reported language use (1’s) and the lowest ALAS category (‘low in-
termediate’). These are associated with acquisition of English on the ‘streets and at
work,’ ‘low’ American acculturation, fewer years of education (<4), higher age-at-
immigration (26–41), and higher levels of percentage of Hispanics in the neigh-
bourhoods (75–100%). As language use values rise into the 2’s and 3’s (not shown),
the points fall into the lower left quadrant and shift toward the lower right quad-
rant. Then, in the area encircled as ‘B,’ rising from the lower right- into the upper
right-quadrant are found the highest ALAS category (‘fluent’), ratings of 5’s on
self-reported language use in various domains, ‘high’ American acculturation,
greater education (>12 years), low concentrations of other Hispanics in the neigh-
bourhoods (1–25%), and lower age-at-immigration (<16).
4.5
Eng w Spouse-4
3.5 B
Think-4
Pray-5
Eng w Friends-4
2.5 Dream-5
Inertia : 13.62 %
A SelfTalk-5
1.5
Math-4
Feel-4
Eng w Child-5
Swear-5
Eng w Child-1 EngEng
w Neighbor-5 Math-5
0.5 Note-1
EngwatFriends-1
Eng Work-1 at Work-5
Educ->12
Postal-1-25%Hisp Feel-5
ALAS-Lo Inter Jokes-1 AgeImm-16-20 PuertoRican-Low Jokes-5
Eng w Spouse-3
Swear-1
Feel-1
Think-1 American-Low Note-4
Engw
Eng
Math-1
Eng wDream-1
Spouse-1
Eng w Fam-1
Neighbor-1
Pray-1
at Work-2 Learn-School Eng w In-Laws-3
SelfTalk-1 Learn-Street&Work
Eng w Jokes-2
In-Laws-1
Eng w Swear-2.5
AgeImm-26-41
Educ-<4
Postal-75-100%Hisp Educ-9-12
ALAS-Hi Eng
PuertoRican-High
Educ-4-8
Math-2 w Fam-4
Postal-25-50%Hisp
Inter Neighbor-3
American-High
Note-3
EngEng w Child-2
w Friends-2 American-Medium
AgeImm->16
Note-5 Dream-3 Think-3
Postal-50-75%Hisp
PuertoRican-Medium
Eng w Neighbor-2 ALAS-Fluent
Pray-3 Feel-3 Eng w Fam-3
Swear-2 Feel-2
Learn-NightSchool Eng w Child-4
Eng w Friends-3
–0.5 Learn-Army AgeImm-21-25
Eng at Work-3
Think-2 Eng Pray-2
Eng wEng ww Neighbor-4
Spouse-2
EngSwear-3
w Jokes-3
Child-3
SelfTalk-2
Fam-2 Eng w Friends-5 EngDream-4
w In-Laws-5
EngNote-2
Eng at Work-4 Math-3 SelfTalk-3
w Dream-2
In-Laws-2 Eng w Spouse-5
Jokes-4
Swear-4
–1.5
–1.5 –0.5 0.5 1.5 2.5
Inertia : 39.73 %
Pray-5
3
B
Dream-5
2 SelfTalk-5
A
Inertia : 13.62 %
1 Eng w Child-5
Swear-5
Eng at Work-1
Eng w Child-1 Eng w Friends-1 Eng w Neighbor-5 Math-5
Feel-1 American-Low Eng at Work-5
Swear-1Note-1 Jokes-1 Educ->12 Feel-5
Street&Work Jokes-5
ALAS-Lo Inter Think-1 Dream-1 1-25%Hisp
Spouse-1 Eng w Fam-1 Learn-School
0 Math-1 Eng w Neighbor-1Eng w In-Laws-1
75-100%Hisp SelfTalk-1 ALAS-Hi Inter
American-High AgeImm-<16
Pray-1 Note-5
AgeImm-26-41 ALAS-Fluent
Educ-<4 Eng w In-Laws-5
Eng w Friends-5
Eng w Spouse-5
–1
–1.2 –0.2 0.8 1.8
Inertia : 39.73 %
Maps of individuals
The map of individuals is abstracted from the total solution and is shown in
Figure 3. This maps shows the location of the participants in the same space (on
the same axes as Figure 1). In the tradition of geometric data analysis (Rouanet,
Ackerman & Leroux 2000), this is the ‘cloud of individuals.’ One method of inter-
preting this cloud of individuals is to examine specific participants represented by
extreme points in the regions as modal representatives of their clusters. This is
more illustrative if other biographical data is also available for these individuals.
For instance, Observation 53 in the high language competence cluster in the upper
right region is a 61 year old female secretary with over 12 years of education, who
immigrated at age 5. She learned English at school and lives in a neighbourhood
with quite a low percentage of other Spanish speakers (6%). At the other extreme,
Observation 15 in the low language competence quadrant at the upper left is a 63
year old female who is a housekeeper with six years of formal education. She im-
migrated at age 16, learned English on the streets and at work, and she lives in a
neighbourhood where 1 in 4 residents speak Spanish (26%).
Reviewing these several maps, we can see that the values of English profi-
ciency and use cluster consistently together. By selectively showing the low vs.
high ALAS scores and self-reports of language use in various domains (Figure 2),
it becomes clear that they too cluster separately from one another. This suggests
that the overall map (Figure 1) reflects increasing proficiency along the x-axis,
Using correspondence analysis to model immigrant multilingualism over time
4.5
4
3.5
3
2.5
Inertia : 13.62%
1.5 53
1
0.5 1545 19
11 837
3910
526748
59 1640 5112 35 50
0 22
43
47 444
56 30
34 46
33
2112 2723 39
1458
2860 36 54
–0.5 24 26 57 42
20 3249
–1
–1.5
–1.5 –1 –0.5 0 0.5 1 1.5 2 2.5
Inertia : 39.73 %
accounting for about 40% of the variance. Further, the supplementary points –
psychocultural factors such as acculturation, education, age-at-immigration, and
sociocultural factors such as percentage of Hispanics in the neighbourhoods – also
cluster logically with the levels of proficiency and use. If we think of these two
mapped groupings as a cross-sectional ‘take’ on language proficiency and use, there
are at least two putative immigrant ‘pathways’ evident in the data. In one, later age-
at-immigration, language learning on the streets and at work, and residence in
primarily Hispanic neighbourhoods is associated with lower levels of English pro-
ficiency (as measured psychometrically on the ALAS) and language use in social
domains and in practices such as thinking, praying, dreaming, counting, etc. In the
other, earlier age-at-immigration, formal language learning, and residence in
neighbourhoods not predominantly Hispanic, are associated with higher profi-
ciency in English and more extensive use of English across a range of domains and
practices. Importantly, the visualization of factors does not commit us to assuming
that one causes other, but allows us to represent multiple directions of influence.
(1987; 2005) and Greenacre (2003). Given that correspondence analysis results
from a chi-square analysis of a contingency table, the question arises as to how to
code time into the contingency table. I explore two possibilities here: coding times
of data collection into the columns of the table and constructing a transition
matrix.
The first method is to code times of data collection into the columns of the contin-
gency table and queries (responses, behaviours, etc.) into the rows, so that cells
represent the frequencies of responses or behaviours at each successive time point
(see also, Schrauf & Iris 2012). Let us assume that the data are in the form of re-
sponses to likert style questions. Rows then represent the frequency data for each
response option (1, 2, 3, 4, 5), cross-tabulated with columns representing 1st data
collection, 2nd data collection, 3rd data collection, etc.
The data: A constructed longitudinal data set with three time points
Hispanic concentration. The columns were Time 1, Time 2, and Time 3. Cells
contained the frequencies of each variable at its respective time point. In the fol-
lowing, I present the statistical results of the correspondence analysis. Again, a
map representing all of the column and row profile points together shows an un-
interpretable cloud of points. Hence, to better demonstrate change over time, I
show the same map twice, each time with the same column profile points, but se-
lectively representing low (Figure 4) and high (Figure 5) values on the variables.
0.5
Child1 Neighbors1
T1
Inertia : 20.36%
Street&Work T3
Friends1 Work1
0 Spouse1
Family1
InLaws1
T2
–0.5 NightSchool
–1
–1.5 –1 –0.5 0 0.5 1 1.5 2
Inertia : 79.64%
Friends5
0.5 Spouse5
InLaws5
School
T1 Child5
Inertia : 20.36%
Work5 T3
0
T2 1-25%Hisp
–0.5
Family4
Neighbors5
–1
–1.5 –1 –0.5 0 0.5 1 1.5 2
Inertia : 79.64%
For purposes of demonstration Figure 4 shows the clustering of low values on self-
rated language use, informal modes of acquisition, and high neighbourhood con-
centrations of Hispanics around Time 1. Figure 5 shows that high values cluster
around time 3. Again, this is constructed data, but the coordination of rising levels
of use-of-English corresponds to the representation of successive ‘data collection’
points, and careful inspection of values within classes of interlocutors tends to
show the same progression.
A second way of examining change over time is to represent the final state
(last moment of data collection) in the columns and previous moments of data
collection in the rows (see for example (Andres, Adamuti-Trache, Yoon, Pidgeon
& Thomsen 2007; van der Heijden 2005). A transition matrix (van der Heijden
1987, 2005) is formed by placing all of the values from time 1 in the rows and all
of the corresponding values from time 2 in the columns. The cells contain the
cross-tabulations of rows and columns. In the cross-tabulation of a likert scale,
for instance, one can read across the rows to see how often people who selected
the first response option at time 1, selected the second, third, fourth, etc. option
at time 2.
Using correspondence analysis to model immigrant multilingualism over time
The data: A constructed longitudinal data set with two time points
To test this kind of transition matrix, I again constructed a ‘longitudinal’ data set
from the same data, but in order to maintain a workable sample size I did a me-
dian (instead of a tertile) split and matched participants across the two sets to
construct a ‘sample’ of 30 participants. Specifically, I rank ordered all 60 partici-
pants according to their average score on all of their responses to ‘language use in
social domains,’ and then performed a median split. I matched the first participant
in the first ranked set of participants with the first participant in the second ranked
set of participants to construct a new participant with scores from ‘two moments’
of data collection, and so on. This yielded a data set of 30 participants, each
‘measured’ twice on each variable.
0.2
A-Child3
B-Child-4
A-Child1
Inertia: 2.11%
B-Child-2
B-Child-5
A-Child2
B-Child-3
–0.2
–1 –0.6 –0.2 0.2 0.6 1
Inertia: 97.89 %
(for a similar but technical demonstration, see van der Heijden 2005). In this
example the first dimension accounts for nearly all of the variance (97.90%), and
clearly the first dimension shows a progression from little English with one’s
children on the left to a majority of English on the right. Again, given the con-
structed nature of the data, it is not surprising to see that self-ratings of English at
time 1 “not at all” (A-Child1) are associated at time 2 with English “sometimes”
(B-Child2) and “half the time” (B-Child3) in the left hand quadrants. Similarly, at
the right hand of the map at time 1, English “sometimes” (A-Child2) and English
“half the time” (A-Child3) are associated at time 2 with English “often” (B-Child4)
and English “all the time” (B-Child5). In essence, from time 1 to time 2, partici-
pants rate themselves as speaking more English with their children.
The focus of this chapter has been on the methodological advantages of corre-
spondence analysis and multiple correspondence analysis for modelling immi-
grant multilingualism. A cross-sectional demonstration was performed on
empirical data, and two longitudinal analyses were conducted on constructed/
simulated data.
First, the cross-sectional analysis using multiple correspondence analysis
shows how the method can facilitate the visualization and analysis of immigrant
multilingualism as an interactive field including abilities, social contexts, and prac-
tices. This is a holistic view in which language proficiency (as a set of abilities) is
intimately linked to language use (in differing social contexts) and language prac-
tices (a range of expressive domains). The modalities maps (Figures 1–2) show
how this field is structured by increasing English proficiency/use/practices from
left to right. Of course, this scaling is the result of the variable responses of the
participants, and each of these is a particular ‘case’ with a unique response pattern.
The individuals’ map (Figure 3) plots the cases along the same dimensions as the
modalities map. From additional biographical and language history data (perhaps
from qualitative interviews) the researcher can situate particular life-stories
(i.e. trajectories or pathways) in the total proficiency/use/practices field and rela-
tive to other cases/life-histories. In the cross-sectional analysis above, I described
two individuals located at the extremes, in upper left and upper right quadrants.
Correspondence analysis also differs from regression-based models that priv-
ilege linear causality in which many factors predict one outcome. Rather,
correspondence analysis dispenses with the distinction between independent and
dependent variables and provides instead a visualization the whole set of interre-
lated factors in differential association. The visualization allows the researcher to
Using correspondence analysis to model immigrant multilingualism over time
assess which factors seem to be more closely associated than others (via chi-square
distances) without losing the sense of multiple interactive and potentially bi-
directional influences. As is evident in the cross-sectional demonstration, for ex-
ample, greater language abilities and more extensive language-use are associated
with residence in more multicultural neighbourhoods, but the visualization does
not commit us to assuming either that the one causes the other or that other fac-
tors are dropped from sight. These two features – the simultaneous visualization
(scaling) of all relevant factors and the recognition of mutual, though unequal, as-
sociations among them – offers a more ecologically valid representation of immi-
grant multilingualism because it
A cross-sectional analysis however remains weaker than a longitudinal analy-
sis, because it compares different people along the same path, whereas ideally we
want to compare the same person on his or her path at successive time points
(Schrauf 2009). While there is some literature on longitudinal correspondence
analysis (van der Heijden 1987, 2005; Greenacre 2003), further developments
would be welcome. In characterizing paths of immigrant multilingualism, we es-
pecially want to pay attention to collecting and analysing data on both developing
abilities-and-use and shifts in the social/institutional/environmental contexts,
which recursively result from the ongoing development of abilities-and-use. In
short, we require a model that attends to the interplay between changes in the so-
cial environment as well as changes in the individual. Using constructed data, we
looked at two methods for mapping change over time. In the first, successive mo-
ments of data collection were entered as columns in the contingency table and
mapped as anchors on the map itself. Row profiles (and corresponding points on
the map) represented data on several domains of language use with particular
classes of interlocutors. Logically enough, the maps (Figures 4–5) provide a visu-
alization of increasing language use over time, and in particular they allow us to
see differences across domains across time. But this falls short of showing indi-
vidual trajectories.
A second method of longitudinal analysis involved constructing a transition
matrix in which ‘performance’ at time 1 and performance at time 2 are cross-
tabulated. The example above was limited to only one domain (‘English spoken
with one’s child’), and the resulting map (Figure 6) showed associations between
‘no English with one’s children’ at time 1 to ‘some’ English at time 2, but interest-
ingly there seemed to be a more frequent jump from ‘some English with one’s
children’ at time 1 to ‘English half the time,’ ‘English often,’ and ‘only English’ at
time 2. Were this real (and not constructed) data, we might investigate whether
there are two paths here, critically distinguished by the amount of English spoken
with children at time 1. Thus, the transition matrix approach allows us to charac-
terize particular pathways more precisely, but because of the limitation to a single
Robert W. Schrauf
References
Alptekin, C. 2010. Redefining multicompetence for bilingualism and ELF. International Journal
of Applied Linguistics 20(1): 95–110.
Andres, L., Adamuti-Trache, M., Yoon, E.-S., Pidgeon, M. & Thomsen, J.P. 2007. Educational
expectations, parental social class, gender, and postsecondary attainment. Youth & Society
39(2): 135–163.
Clausen, S.-E. 1998. Applied correspondence analysis: An introduction. Thousand Oaks CA: Sage.
Cortes, D.E., Rogler, L.H., & Malgady, R.G. 1994. Biculturality among Puerto Rican adults in the
United States. American Journal of Community Psychology 23: 339–356.
Duncan, S.E. & De Avila, E.A. 1991. Adult Language Assessment Scales. Monterey CA: CTB
Macmillan/McGraw-Hill.
Franceschini, R. 2011. Multilingualism and multicompetence: A conceptual view. The Modern
Language Journal 95: 344–355.
Glick-Schiller, N., Caglar, Y., & Guldbrandsen, T.C. 2006. Beyond the ethnic lens: Locality,
globality, and born-again incorporation. American Ethnologist 33(4): 612–633.
Greenacre, M.J. 1984. Theory and Application of Correspondence Analysis. London: Harcourt
Brace.
Greenacre, M.J. 1992. Correspondence analysis in medical research. Statistical Methods in
Medical Research 1: 97–117.
Greenacre, M.J. 1994. Correspondence analysis and its interpretation. In Correspondence Analysis
in the Social Sciences, M. Greenacre & J. Blasius (eds.), 3–22. New York NY: Academic Press.
Greenacre, M.J. 2003. Singular value decomposition of matched matrices. Journal of Applied
Statistics 30(10): 1101–1113.
Greenacre, M.J. 2007. Correspondence Analysis in Practice. Boca Raton FL: Chapman & Hall/
CRC.
Hall, J.K., Cheng, A. & Carlson, M.T. 2006. Reconceptualizing multicompetence as a theory of
language knowledge. Applied Linguistic 27(2): 220–240.
Using correspondence analysis to model immigrant multilingualism over time
Kecskes, I. 2010. Dual and multilingual systems. International Journal of Multilingualism 7(2):
91–109.
Le Roux, B. & Rouanet, H. 2004. Geometric Data Analysis: From Correspondence Analysis to
Structured Data Analysis: Dordrecht: Kluwer.
Lebaron, F. 2009. How Bourdieu “Quantified” Bourdieu: The geometric modeling of data. In
Quantifying Theory: Pierre Bourdieu, K. Robson & C. Sanders (eds.), 11–29. Berlin: Springer.
Rouanet, H., Ackermann, W. & Leroux, B. 2000. The geometric analysis of questionnaires: The
lesson of Bourdieu’s La Distinction. Bulletin de Methodologique Sociologique 65: 5–18.
Schrauf, R.W. 2009a. English use among older bilingual immigrants in linguistically concen-
trated neighborhoods: Social proficiency and internal speech as intracultural variation.
Journal of Cross-Cultural Gerontology 24: 157–179.
Schrauf, R.W. 2009b. Longitudinal designs in studies of multilingualism. In Language Develop-
ment over the Lifespan, K. de Bot & R.W. Schrauf (eds), 245–270. New York NY: Routledge.
Schrauf, R.W. & Iris, M. 2012. Very long pathways to diagnosis among African Americans and
Hispanics with memory and behavioral problems associated with dementia. Dementia. 11,
no. 6: 726–746.
Schrauf, R.W. & Hoffman, L. 2007. The effects of revisionism on remembered emotion: The
valence of older, voluntary immigrants’ pre-migration autobiographical memories. Applied
Cognitive Psychology 21: 895–913.
Schrauf, R.W. & Rubin, D.C. 2001. Effects of voluntary immigration on the distribution of
autobiographical memories over the lifespan. Applied Cognitive Psychology 15: 75–88.
Schrauf, R.W. & Rubin, D.C. 2004. The ‘language’ and ‘feel’ of bilingual memory: Mnemonic
traces. Estudios de Sociolinguistica 5(1): 21–39.
Van der Heijden, P.G.M. 1987. Correspondence Analysis of Longitudinal Categorical Data.
Leiden: DSWO Press.
Van der Heijden, P.G.M. 2005. Correspondence analysis of longitudinal data. In Encyclopedia of
Biostatistics, P. Armitage & T. Colton (eds), 1230–1234. Chichester: John Wiley.
Vertovec, S. 2007. Super-diversity and its implications. Ethnic and Racial Studies 30(6):
1024–1054.
Capturing diversity
Linguistic land- and soundscaping
1. Introduction
individual’s or a people’s reality, actual social practice has to be studied. So far, such
studies have yet to be performed.
This statement also holds true for the study of linguistic diversity. We know
about diversity with regard to the home languages of pupils in a variety of
European cities (Fürstenau et al. 2003, Extra & Yagmur 2004, Barni & Extra 2008).
Furthermore, studies in the framework of linguistic landscaping have revealed
data concerning the statistical distribution of languages found on public signs in a
number of (mostly) urban areas around the world (s. e.g. Shohamy et al. 2010,
Jaworski & Thurlow 2010). What we do not know a lot about is how people actu-
ally make use of the increasingly diverse linguistic means at their disposal. Only
few studies have looked at the communicative practicing of linguistic diversity,
and even they usually do not discuss their findings with regard to their implica-
tions on the societal level (see, however, Rehbein 2010). There is therefore a gen-
eral need 1) for studies that describe and analyse how linguistic diversity is
‘performed’ on the micro-level of linguistic action1 and 2) the development of
methods of connecting these micro-level analyses to the macro-level of social di-
versity. In the following sections, we will outline ways in which we think these
methodological and theoretical gaps can be bridged. In doing so, we will present a
pilot study which investigates multilingual communication in St. Georg, a central
district in the city of Hamburg.
We begin by offering some general data on linguistic diversity in Hamburg.
We then take a brief look at the district St. Georg, recognized by inhabitants and
visitors as a focal point of urban diversity in Hamburg (Section 2). Section three
sketches the theoretical background that our analyses of St. Georg’s linguistic and
social diversity are based on, namely functional pragmatics. In sections four and
five, we present our methodological approach as well as the data and findings con-
cerning diversity in St. Georg. First, statistical data on the distribution of languag-
es in St. Georg will be given (4.1), before the approach of linguistic landscaping is
addressed. We point out some of the limitations in the way in which linguistic
landscaping has traditionally been conducted and propose means of coping with
these limitations (4.2). Adopting a functional-pragmatic perspective in the analy-
sis of languages on public signs is, in our opinion, a suitable way of compensating
for some of the shortcomings of traditional linguistic landscaping research.
Section five introduces linguistic soundscaping as an innovative multi-method ap-
proach designed to capture oral linguistic diversity. Linguistic soundscaping ex-
pands the scope of research on the audible environment in urban areas as it is
conducted in a variety of disciplines to include linguistic sounds. We will propose
1. Cf. Blommaert & Rampton 2011, who emphatically call for research on communicative
language practice.
Capturing diversity
“Hamburgers”
from around the
world:
Figure 1. Nationalities of inhabitants in Hamburg (Source: Hamburg: Das Magazin aus
der Metropole. Dec. 2009: 4)
The numbers vary slightly with regard to the administrative unit of Hamburg Mitte
(‘Hamburg Centre’), where our pilot study was conducted (Table 1). First, there is
an even larger number of people with personal or familial histories of migration
(44% in Hamburg Mitte compared to 29% in Hamburg overall). A second fact
concerns diversity as a whole: the largest group in Table 1 is represented by people
from “other” countries, with each of the subgroups in this category accounting for
less than 1.9% of the total number of immigrants in Hamburg. This shows just
how diverse and multi-faceted Hamburg Mitte’s composition is. The distribution
of the other groups more or less mirrors their overall statistical representation in
Hamburg. We find a predominance of people from Turkey and from countries
in which Slavic languages are spoken (Poland, Bulgaria, Russia and former
Yugoslavia).
Hamburg on the whole as well as Hamburg Mitte are therefore characterised by
significant ethnic diversity. This phenomenon, however, is even more prominent
for the district of St. Georg, located in the centre of Mitte near Hamburg’s
Capturing diversity
Table 1. Ethnic Diversity in Hamburg Mitte 2010 (Statistisches Amt für Hamburg
und Schleswig-Holstein 2011b)
main station. It is known as a hot-spot for social, ethnic and linguistic diversity
and it is for that reason that we chose this district as the object of our pilot study
on linguistic diversity in practice.
Judging from municipal statistics (Statistisches Amt für Hamburg und Schleswig-
Holstein 2011a) on the districts of Hamburg, there are a number of facts that stand
out about St. Georg: 1) the population density in St. Georg is among the highest of
all of the districts; 2) the inhabitants are very young, with a high percentage (37.7%)
of people with a familial history of migration among the residential population
(just as for Hamburg Mitte, persons from Turkey and from Middle and Eastern
Europe are heavily represented). While great density and a young age of the resi-
dent population are typical for central districts in attractive urban areas, another
fact decreases St. Georg’s attractiveness significantly: 3) its crime rate is more than
ten times higher (1458 crimes compared to 129 crimes per 1000 inhabitants per
year) than that of Hamburg on average. Not surprisingly, 4) the percentage of fam-
ilies living in St. Georg is significantly lower than in the rest of the city. These
numbers prove St. Georg to be a densely populated, dynamic inner-city district
facing a variety of social and communal problems.
These issues become clearer when focussing more closely on the social
composition of St. Georg. The district is in itself divided into two sharply con
trasting areas: The gentrified Lange Reihe-area and the low price, lower-class
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
2. Other important groups of immigrants, e.g. people from Poland or countries of the former
Soviet Union, are not as strongly represented in St. Georg. They tend to cluster in other parts of
Hamburg that were not part of the pilot study (s. Brehmer 2010), but will be included in a re-
search project aiming to establish a discursive topography of multilingualism (Redder 2013).
Capturing diversity
In our pilot study, we used the methods of linguistic landscaping to gain initial
insight into the visual manifestations of multilingualism in the two areas of in-
terest. Following the seminal work by Landry and Bourhis (1997), linguistic
landscaping has become a well-established approach in the sociolinguistic study
of multilingualism. Still, the study of public signage predates the coining of the
term “linguistic landscape”. As early as in 1972, Yasuo Masai assembled a “living
map of Tokyo” focussing on script-types on shop signs in centres of metropoli-
tan Tokyo. In 1978, Stella Tulp published a study on language distribution in
Brussels that, like the work of Rosenbaum et al. (1977) on Jerusalem,4 started the
4. For a more in-depth overview of these and other early studies on the linguistic landscape,
see Backhaus 2007: 12f.
Capturing diversity
tradition of studying public signage in areas where the use of language is so-
cially and politically contested. Other studies have since followed suit, tackling
highly controversial language use on public signs in Quebec (Monnier 1989,
Landry & Bourhis 1997, Dagenais et al. 2009), the Basque region (Cenoz &
Gorter 2006) and Israel (Spolsky & Cooper 1991, Ben-Rafael et al. 2006,
Trumper-Hecht 2010, Guilat 2010, Waksman & Shohamy 2010, Ben-Rafael &
Ben Rafael 2010). While some scholars have also studied traditional monolin-
gual areas (see, for example, Backhaus 2007 on Tokyo and Huebner 2006 on
Bangkok), where language choice in itself is not a contentious issue, especially
early studies on linguistic landscapes were carried out in regions in which lan-
guage groups struggled over the presence of “their” language on official and in-
official public signage. It is precisely in these regions that language use on public
signs is often seen as a political statement. Language choice is then not only de-
termined by the language the author of a sign or his/her supposed addressees
know (cf. Spolsky & Cooper 1991; Spolsky 2009: 33), but can also be a “state-
ment” about the language group the author wishes to be associated with. Scholars
who have studied the linguistic landscape in areas of language conflict have thus
often identified the use of a particular language on a “bottom-up”- sign5 as an
expression of “solidarity” with a language group, whereas the presence of a lan-
guage on “top-down” signs is usually regarded as a display of the “power” of the
corresponding language group. While these categories may be useful for areas of
political struggle over language use, they are inappropriate for other places,
such as Hamburg, where traditionally only one language is used. Thus, the use
of Turkish on a poster with a mug shot (Figure 2), where only the first line
(Wir bitten um Ihre Mitarbeit, ‘We ask for your cooperation’) and the logo
(Polizei Hamburg) are kept in German, certainly does not signify the power of
the Turkish community in Hamburg.
Considering the notion of “solidarity”, we would similarly argue that the pur-
pose of the Arabic script on the top of the bottom-up sign in Figure 3, which says
Halal, is to let readers know that they can order food in this restaurant prepared
5. Auer has criticized the use of the dichotomy between “bottom-up” and “top-down” signs in
research on linguistic landscape because it would insinuate that public spaces are shaped by hi-
erarchical relations of “above” and “below” (Auer 2010: 295). He argues that private actors can
acquire the right to change parts of the public space, whereas municipalities retain these rights
in all other areas. We agree with Auer’s argument, but would also like to point out that even the
distinction between private and public is not selective enough to account for all cases of public
signage (e.g. what about private businesses like snack shops or bakeries that operate within a
state-owned building such as a railway station?). However, since most of the literature on lin-
guistic landscape draws on the dichotomy between “bottom-up” and “top-down”, we adhere to
this terminology in this section for the sake of comparison (cf. Ben-Rafael et al. 2006: 14f).
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
according to the tenets of the Quran. A claim that the use of Arabic can also be
interpreted as a symbol of solidarity with the Arabic-speaking population of Ham-
burg would need to be supported by further evidence (like stickers or oral state-
ments explicitly supporting the Arabic-speaking community).6
6. By using Arabic to advertise for an Indonesian restaurant, the sign also reminds us of a fact
obvious to most scholars of multilingualism, but all too often ignored in public discourse on
social diversity: the common linking of a particular language to a single ethnic group and fur-
ther to a single country of immigration is most often wrong: in this case, for example, Arabic is
used to offer Indonesian food to all kinds of people, irrespective of whether they understand
Arabic themselves or simply recognize the form of the Halal-lettering.
Capturing diversity
7. For a discussion on the unit of analysis in the study of the linguistic landscapes see also
Cenoz and Gorter 2008: 351.
8. Ehlich (2007b/III, K1, K3) defines ‘text’ as a functional ensemble of speech actions which
bridges the systematic gap between the situation of production by a speaker and that of recep-
tion by a hearer, and hence copes with challenges of “linguistic tradition”. Texts are characterized
by the lacking co-presence of speaker and hearer; when speaker and hearer are co-present, the
macro-unit of speech action is called ‘discourse’.
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
of languages on public signage in Hamburg St. Georg are therefore based on the
“sign” as the unit of analysis.
We first offer a general overview of the visual data we gathered and then go
into more detail by analysing specific examples of the linguistic landscape of
Hamburg St. Georg.
the Steindamm-area (61.3%) than in the Lange Reihe-area (32.2%). This finding is
mirrored by the percentage of monolingual German signs (row four). Still, the
linguistic diversity in the Steindamm-area has not lead to the displacement of
German: nearly three-quarters of all signs contain German. Not surprisingly,
the Lange Reihe-area features an even larger proportion of German-language sig-
nage (80.4%).
When it comes to particular languages aside from German, English is the lan-
guage most frequently used on signs in St. Georg. It prevails in both the Stein-
damm-area (30.6% of the signs contain English) and the Lange Reihe-area (17.2%).
These high numbers are probably due to the fact that English has a huge influence
on contemporary German. A lot of English words are used (such as “phone” or
“shop”) and commonly understood, but are not (yet) part of the German language
system, i.e. are still commonly recognized as English. Thus, these words were not
classified as loanwords but rather as instances of the use of English.
In the Lange Reihe-area, English accounts for the bulk of signs with languages
other than German. In addition to English and in line with our expectations, we
found languages with high ‘symbolic value’ such as Italian or French, though in a
relatively small number. These languages are mostly used in restaurants and shops
offering goods from the respective countries.9 In the Steindamm-area, the analysis
reveals a strong presence of immigrant languages: 13.1% of the signs contain
9. We cannot go into detail about the relationship between language and the various eco-
nomic sectors in the linguistic landscape of Hamburg St. Georg, but see Breckner et al. (2013)
for more information on this point.
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
Turkish, 12% contain Farsi and 5.6% Arabic. From these numbers, it is obvious
that immigrant languages play an important role in the general linguistic
landscape of the Steindamm-area. This certainly does not hold for the Lange
Reihe-area: the total proportion of the three immigrant languages displayed in
Table 3 accounts for 30.7% in the Steindamm-area and only 0.8% in the Lange
Reihe-area.10
Summarising our statistical findings on the linguistic landscape, we can con-
clude that St. Georg is a linguistically diverse district. However, the diversity
found differs between the Lange Reihe and Steindamm-area, with the latter
accounting for the majority of the multilingual (more than one language) signs
as well as for most of the signs containing a foreign language. In the Lange
Reihe-area, multilingualism mainly consists of the use of languages of high sym-
bolic value, whereas in the Steindamm-area we find languages used almost exclu-
sively by immigrants.11 Thus, there appears to be a rift between the two areas, not
only in the degree of visible linguistic diversity but also when it comes to the
languages that constitute this diversity. Fitting to its social profile, the Lange
Reihe-landscape is characterized by a ‘gentrified’ multilingualism that depends
on the dominant domestic language as well as languages of high symbolic value.
The linguistic landscape of the Steindamm-area, on the other hand, is shaped by
migration-induced multilingualism that complements the domestic language
with significant contributions by immigrant languages. However, despite all of
the differences between the two areas, there is one feature that both of them
share: German is by far the most commonly used language. To account for this
fact and to gain deeper insight into the complex functional texture created
through the use of multiple languages in St. Georg, we need to take a closer look
at selected instances of public signage and the communicative processes they are
involved in.
10. Other languages are more or less irrelevant in both areas. Accordingly, we found only three
instances of Russian (two in the Lange Reihe, one in the Steindamm) and only one sign using
Polish (in the Lange Reihe) (but cf. Brehmer 2010, who found a cluster of public signs in Slavic
languages near the central bus station not far from the Steindamm-area). The same holds true
for languages from India (one in the Lange Reihe, four in the Steindamm), Africa (only three
instances of Twi in the Steindamm, as an example) and for Urdu (two in the Steindamm). Thus,
the “migration-induced multilingualism” of the Steindamm-landscape seems to consist mainly
of Turkish, Farsi and Arabic.
11. For the sake of simplicity, these languages will be called “immigrant languages” in the rest
of the paper, even though it is clear that at least some of them may be spoken by people with a
monolingual German background.
Capturing diversity
One instance of public signage and its context is depicted in Figure 4. The figure
shows the front of the Turkish grocery Sönmez Markt. The name itself is bilingual,
since it combines the Turkish word Sönmez (‘never dies’) with the German Markt
(‘market’). Below the name there is a slogan saying ‘Our strength comes from
quality’ in Turkish. The next semiotic level is also bilingual, this time Turkish-
English. The sign advertises the brand “Meray Nuts” with the Turkish phrase ‘shell-
breaking tastes good!’ Here, non-domestic languages such as Turkish and English
are used for the speech action of advertising. Below the advertisements there are
boxes with fruits and vegetables marked with yellow signs with the names of the
products, their prices, the respective unit of pricing and other additional informa-
tion. Interestingly, these signs are not bi- or multilingual, but are kept solely in
German instead. The products are named in German (e.g. Limetten), the unit of
pricing is written in German (St as a kind of abbreviation for Stück, ‘per piece’ and
Kg as an abbreviation of Kilogramm, ‘kilogram’). Additional information is also
provided only in German (e.g. Angebot, i.e. ‘bargain’, on top of Limetten).
Figure 5 provides a similar impression of the public signage in the Steindamm-
area. It shows the front of the grocery Anssar (Arabic for ‘supporter’), where the
name is presented in both Roman and Arabic script (at the top right corner it says
“Anssar company” in Arabic). Again, the name of the food item, its country of
origin as well the unit of pricing are given exclusively in German.
Another interesting combination of signs is shown in Figure 6. This photo-
graph was taken in the Bremer Reihe, a crossroad to Steindamm, and shows the
shop window and the main sign of “Maharaja”, a shop for Indian goods. The main
sign at the top is kept mostly in English (“Fancy Suits”, etc.) but also uses some
German, too (Mode Schmuck (‘fashion jewellery’)). Below the main sign there
is an advertisement for another Indian shop, “Saira India Hous”, in English.
Then, on the shop window, we find three different languages expressed by three
scripts. The upper lettering is Hindi (Devanagari script), saying ‘Sahraaz India
House’. The lower lettering is Dari (Arabic script), saying ‘clothes from India for
men, women, kids; jewellery, DVDs’. Hindi, Dari and English are used to convey
the shop’s name, its ethnic affiliation and the goods it has to offer. German does not
feature prominently in this mix of languages, but it is used on the white sign sub-
sequently hung in the shop window with adhesive tape. As in the other cases, the
German sign again realizes a different type of text: it is a prohibition sign. The sign
says in German (containing a ‘mistake’ (cf. the notion of “peripheral normativity”
introduced by Blommaert et al. 2005) in spelling out the compound noun
Rauchverbot): “Absolutely no smoking in this area and the consumption of alcohol
is most strictly forbidden” (Hier gilt absolutes Rauch Verbot und der Verzehr von
Alkohol ist strengstens untersagt).
12. Figures 2–6 also highlight an important component of texts in a linguistic landscape that
we cannot elaborate upon in this context: the material quality of signs (cf. Auer 2010: 282f). For
the communicative process, it obviously makes a difference whether a text is hastily written on
a random piece of paper or whether it is printed on a poster or a billboard. After examining our
data, we believe that the functional division of the district as described above may also be re-
flected in the materiality of the signs: the percentage of pencil-and-paper signs seems to be
much higher in the Steindamm-area than in the Lange Reihe-area.
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
The audible part of one’s everyday environment was first investigated by the musi-
cologist Murray Schafer (1977). He coined the term “soundscape” to describe an
acoustic environment, and thereby tried to draw the attention of social science
scholars towards the audibility of social events. The term “soundscape” was intro-
duced by Schafer in a rather broad sense: “The soundscape is any acoustic field of
study” (Schafer 1977: 7). Since this vague definition makes it hard to delineate
what a soundscape actually is, it has allowed for a very wide interpretation of the
term in a variety of disciplines (see Rösing 2000).
Particularly deficiency-oriented approaches most readily adopted the notion
of soundscapes: The fact that everyday urban life is strongly affected by its sonic
environment – often in a negative way – led scholars to the conclusion that a
soundscape can and must be shaped (Schulte-Fortkamp & Dubois 2006). Further-
more, the idea of improving the sonic environment led to concepts of “acoustic
designing”. While this kind of ‘applied soundscaping’ (cf. Acta Acustica 2006) may
be best known and features most prominently in current public debates on sound,
a variety of other research questions are linked to the study of soundscapes:
– Why do certain sounds appear in particular places?
– How do we perceive and experience sounds?
– How and why do individual sounds or soundscapes as a whole change over
time?
– How is the development of a soundscape interconnected with social
developments?
– How can the arts (music, film, sound-design installations) change our think-
ing and feelings about sounds?
Research on these issues has been conducted in the fields of anthropology, acous-
tics/acoustic ecology, ethnology and sociology (see Rösing 2000 for a summary).
Soundscapes have also been used for artistic purposes, especially in music, film
and sound installations. These projects have taken advantage of the fact that every
place, and especially every city, features its own distinct sound profile. Musicians
have used soundscapes as inspiration by documenting the sounds they encoun-
tered and integrating them into their compositions.13 Furthermore, authentic
13. See e.g. the project about sounds and music in Istanbul, Dubai, Johannesburg and the Pearl
River Delta conducted by the Ensemble Modern from 2008 to 2010, http://www.ensemble-
modern.com.
Capturing diversity
sounds have been used as raw material for music such as Tape music, Musique
concrete, and electronic music (cf. Rösing 2000).
While research on soundscapes usually ignores the human voice and its lin-
guistic productivity, it represents a good starting point for the investigation of oral
linguistic diversity (cf. Samuels et al (2010: 339), who make a similar argument
from an anthropological perspective). As Schafer already pointed out, sound-
scapes offer insights not only into our sonic environments but also into society:
“the general acoustic environment of a society can be read as an indicator of social
conditions which produce it and may tell us much about the trending evolution of
that society” (Schafer 1977: 7). Adopting this idea, linguistic soundscaping should
start off with the hypothesis that the linguistic soundscape offers insights into lan-
guage use trends within a given society. This should especially hold true for a so-
ciety like the German one, which is in large part influenced by migration and a
growing degree of social diversity (cf. Vertovec 2011).
In trying to link linguistic action and the acoustic shape of urban areas on a theo-
retical level, we propose to adopt Eckert’s (2009) interpretation of Bourdieu’s
(1985) differentiation between social and physical space.14 At first glance, all
linguistic actions are performed in concrete, distinct physical spaces. At the same
time, an oral or written verbal action transforms a physical space into social spac-
es in which people interact in certain ways. Institutions (to be understood as pur-
pose-driven structural types of repetitive societal actions (Redder 2008: 145f.;
Ehlich & Rehbein 1994)), for example, heavily regulate the kinds of action that
can be performed. These regulations are the products of linguistic actions
(laws, directions, action knowledge), but they are only peripherally influenced by
the physical shape of the building in which the institution is located. So it is lin-
guistic action that serves as a bridge between a physical space and its soundscape
and the social space in which people live and interact, thereby creating the sounds
of a city.
Consequently, one of the aims of linguistic soundscaping is to document how
social spaces are created and developed over time through linguistic action.15 In
doing so, linguistic soundscaping addresses the following research questions:
14. Cf. Läpple 1991 for an attempt to organize different conceptual layers of space.
15. These types of data can then be used to set up a “discursive topography of multilingualism”
that develops an integrated functional understanding of multilingualism from a synchronic and
diachronic perspective (Redder 2013).
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
public as a whole. Since they are not part of private conversations, there are no
legal restrictions (apart from the obligation to depersonalize the data) on record-
ing these public utterances. Although linguistic actions of this kind may occur
only infrequently and in specific places, they comprise a particularly prominent
feature of the soundscape because of the speech volume and the universal audi-
ence. As Baumann showed, analysing these public utterances reveals important
details about the linguistic structuring of urban areas.
All of the methods mentioned here can be used to identify actors that shape
the soundscape of certain urban districts or other physical spaces. However, they
will be complemented by narrative interviews with these actors addressing their
language choices in order to gain deeper insight into the causes and structure of
the linguistic soundscape.
So far, our knowledge of the linguistic soundscape of St. Georg stems mainly from
the ethnographic observations and gathered quantitative data concerning the dis-
tribution of languages spoken in this district (but see Redder 2013 for an analysis
of transcribed multilingual conversations). We first present data on the frequency
of language occurrences and then discuss some representative instances of lan-
guage use in St. Georg.
16. By contacts we mean instances of language use, irrespective of length or the number of
interlocutors participating in the exchange.
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
the data in the Lange Reihe; Arabic and Slavic Languages make up 2% each, with
no other language or language group accounting for more than 1% of the total of
languages charted. These numbers are roughly in line with those from another
case study on languages spoken in the subway system of Hamburg (Brehmer 2010),
where we also found a predominance of German (83%), with Turkish as the sec-
ond most important language (3%). The relatively high percentage of English in
the Lange Reihe (3% compared to only 1.5% in the subway system) is probably due
to the Lange Reihe’s reputation as a hot-spot for tourists.17 One the one hand, the
numbers for the Lange Reihe show that the area is not characterized by prominent
multilingualism, since more than four-fifths of the encounters were in German.
On the other hand, even in this gentrified, touristic area, we find considerable
proportions of Turkish and Arabic contact as two of the most important migrant
languages in Germany. The relatively high percentage of English indicates the role
of tourism for this area.
17. Obviously English in general is not restricted to the domain of tourism, but can serve a
variety of functions (cf. Philipson 2006: 80, Lüdi 2007: 142f). Our conclusion here is based upon
the number of high-priced restaurants and hotels in the Lange Reihe area as well as on the fact
that tourists, carrying cameras and trolley cases, are a common sight in this area.
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
Summarizing our preliminary findings on an empirical level, we can state that St.
Georg’s functional and social division is generally well-reflected in the data on its
linguistic diversity. The Steindamm-area is characterized by a migration-induced
multilingualism consisting of languages of immigration and German. In the Lange
Reihe-area, the dominance of German is much stronger and is complemented
regulations allow for that, thus adding another piece of evidence in order to get a
comprehensive picture of multilingual language practice in public places.
Acknowledgements
This work has been made possible through financial support to the Landesexzel-
lenzcluster (State of Hamburg Excellence Initiative) Linguistic Diversity Manage-
ment in Urban Areas – LiMA by the Forschungs- und Wissenschaftsstiftung
Hamburg. The responsibility for the contents of this study lies with the authors.
The authors wish to thank Prof Dr. Tatiana Oranskaia, Trang Lam Thuy and Dana
Ofosu for contributions to this study. We thank Dr. Kasper Juffermans and an
anonymous reviewer for their comments that helped improve this paper. Warm
thanks are also due to Audrey McDougall for proofreading an earlier version of
the paper.
References
Acta Acustica united with Acustica 2006. The Journal of the European Acoustics Association
(EAA). 91(6).
Ammon, U. 2001. Verkehrssprachen in Europa – außer Englisch. Tübingen: Niemeyer.
Auer, P. 2010. Sprachliche Landschaften. Die Strukturierung des öffentlichen Raums durch die
geschriebene Sprache. In Sprache intermedial. Stimme und Schrift, Bild und Ton,
A. Deppermann & A. Linke (eds), 271–298. Berlin: De Gruyter.
Backhaus, P. 2007. Linguistic Landscapes. Comparative Study of Urban Multilingualism in Tokyo.
Clevedon: Multilingual Matters.
Barni, M. & Extra, G. (eds) 2008. Mapping Linguistic Diversity in Multicultural Contexts. Berlin:
De Gruyter.
Bauman, R. 2001. The ethnography of genre in a Mexican market: Form, function, variation. In
Style and Sociolinguistic Variation, P. Eckert & J. Rickford (eds), 57–77. Cambridge: CUP.
Ben-Rafael, E., Shohamy, E., Amara, M., & Trumper-Hecht, N. 2006. Linguistic landscape as
symbolic construction of the public space: The case of Israel. In Linguistic Landscape: A
New Approach to Multilingualism, D. Gorter (ed.), 7–30. Clevedon: Multilingual Matters.
Ben-Rafael, E. & Ben-Rafael, M. 2010. Diaspora and returning diaspora: French-Hebrew and
vice-versa. In Linguistic Landscape in the City, E. Ben-Rafael, E. Shohamy, M. Amar &
N. Trumper-Hecht (eds), 328–345. Clevedon: Multilingual Matters.
Blommaert, J., Muyallert, N., Huysmans, M. & Dyers, C. 2005. Peripheral normativity: Literacy,
English, and the persistence of hidden inequality in a South African township. Linguistics
and Education 16(4): 378–403.
Blommaert, J. & Rampton, B. 2011. Language and Superdiversity: A position paper [Working
Papers in Urban Language & Literacies 70]. London: King’s College.
Capturing diversity
Bourdieu, P. 1977. The economics of linguistic exchange. Social Science Information 16(6):
645–668.
Bourdieu, P. 1985. Sozialer Raum und “Klassen”. Leçon sur la leçon [Suhrkamp-Taschenbuch
Wissenschaft 500]. Frankfurt: Suhrkamp.
Breckner, I., Peukert, H. & Pinto, M. 2013. The delicate search for language in spaces:
Multilingualism as a resource in urban development? In Multilingualism and Language
Diversity in Urban Areas [Hamburg Studies on Linguistic Diversity], P. Siemund, I. Gogolin,
J. Davydova & M. Schulz (eds). Amsterdam: John Benjamins.
Brehmer, B. 2010. Acoustic presence and visual absence of minorities in multilingual cities –
The case of Slavic speaking minorities in Hamburg. Paper presented at the conference
“Multilingual individuals and multilingual societies”, 7 October 2010. Hamburg: Hamburg
University.
Bühler, K. 1934. Sprachtheorie: Die Darstellungsfunktion der Sprache. Jena: Fischer.
Bührig, K. 2005. “Speech Action Patterns” and “Discourse Types”. Folia Linguistica 39(1–2):
143–171.
Cenoz, J. & Gorter, D. 2006. Linguistic landscape and minority languages. In Linguistic
Landscape: A New Approach to Multilingualism, D. Gorter (ed.), 67–80. Clevedon: Multilin-
gual Matters.
Cenoz, J. & Gorter, D. 2008. The linguistic landscape as an additional source of input in second
language acquisition. International Review of Applied Linguistics in Language Teaching
46(3): 267–287.
Dagenais, D., Moore, D., Sabatier, C., Lamarre, & Armand, F. 2009. Linguistic landscape and
language awareness. In Linguistic Landscape. Expanding the Scenery, E. Shohamy &
D. Gorter (eds), 253–269. New York NY: Routledge.
Eckert, P. 2010. Who’s there? Language and space in social anthropology and interactional
sociolinguistics. In Language and Space. An international Handbook of Linguistic Variation,
P. Auer & J. Schmidt (eds), 163–178. Berlin: De Gruyter.
Ehlich, K. 2007a. Thrifty monolingualism and luxuriating plurilingualism? In Language Regimes
in Transformation. Future Prospects for German and Japanese in Science, Economy, and
Politics [Contributions to the Sociology of Language 93], F. Coulmas (ed.), 19–32. Berlin:
De Gruyter.
Ehlich, K. 2007b. (ed.) Sprache und sprachliches Handeln, 3 Vols. Berlin: De Gruyter.
Ehlich, K. & Rehbein, J. 1972. Zur Konstitution pragmatischer Einheiten in einer Institution:
Das Speiserestaurant. In Linguistische Pragmatik, D. Wunderlich (ed.), 209–254. Frankfurt:
Athenaion.
Ehlich, K. & Rehbein, J. 1975. On effective reasoning. In Proceedings of the IVth AILA Congress
of Applied Linguistics, C. Candlin (ed.), 313–338. Stuttgart: Hochschulverlag Stuttgart.
Ehlich, K. & Rehbein, J. 1994. Institutionsanalyse: Prolegomena zur Untersuchung von
Kommunikation in Institutionen. In Texte und Diskurse. Methoden und Forschungsergeb-
nisse der funktionalen Pragmatik, G. Brünner & G. Gräfen (eds), 287–327. Opladen:
Westdeutscher Verlag.
Ehlich, K. Rehbein, J., & Thije, J. D. (eds) 1993. Kennis, taal en handelen: Analyses van de
communicatie in de klas. Assen: Van Gorcum.
Extra, G. & Yagmur, K. (eds) 2004. Urban Multilingualism in Europe. Immigrant Minority
Languages at Home and School. Clevedon: Multilingual Matters.
Fürstenau, S., Gogolin, I. & Yagmur, K. 2003. Mehrsprachigkeit in Hamburg. Ergebnisse einer
Sprachenerhebung an den Grundschulen. Münster: Waxmann.
Claudio Scarvaglieri, Angelika Redder, Ruth Pappenhagen and Bernhard Brehmer
Guilat, Y. 2010. The holy ark in the street. Sacred and secular painting of utility boxes in the
public domain in a small Israeli town. In Linguistic Landscape in The City, E. Shohamy,
E. Ben-Rafael & M. Barni (eds), 37–56. Clevedon: Multilingual Matters.
Huebner, T. 2006. Bangkok’s linguistic landscape: Environmental print, code mixing and lan-
guage change. In Linguistic Landscape: A New Approach to Multilingualism, D. Gorter (ed.),
31–51. Clevedon: Multilingual Matters.
Hult, F. 2009. Language ecology and linguistic landscape analysis. In Linguistic Landscape.
Expanding the Scenery, E. Shohamy & D. Gorter (eds), 88–103. New York, NY: Routledge.
HWF Hamburg Business Development Center 2011. Where the world is at home. <http://
www.hamburg-economy.de/why-hamburg/2222584/where-the-world-is-at-home.html>
(28 November 2011).
Jaworski, A. & Thurlow, C. (eds) 2010. Semiotic Landscapes. Language, Image, Space. London:
Continuum.
Kokot, W., Gandelsman-Trier, M., Wildner, K. & Wonneberger, A. (eds.) 2008. Urban Studies.
Port Cities as Areas of Transition: Ethnographic Perspectives. Bielefeld: Transcript.
Landry, R. & Bourhis, R. 1997. Linguistic landscape and ethnolinguistic vitality. An empirical
study. Journal of Language and Social Psychology 16(1): 23–49.
Läpple, D. 1991. Essay über den Raum: Für ein gesellschaftswissenschaftliches Raumkonzept. In
Stadt, Raum und Gesellschaft, Vol. 1: Stadt und Raum. Soziologische Analysen,
H. Häußermann, D. Ipsen & T. Krämer-Badoni (eds), 1–47. Pfaffenweiler: Centaurus.
Lüdi, G. 2007. Basel: einsprachig und heteroglossisch. Zeitschrift für Literaturwissenschaft und
Linguistik 148: 132–157.
Masai, Y. 1972. Tokyo no seikatsu chizu [Living map of Tokyo]. Tokyo: Jiji Tsûshinsha.
Monnier, D. 1989. La langue d’accueil et la langue de service dans les commerces à Montréal. Con-
seil de la langue française [Notes et documents 70]. Québec.
Phillipson, R. 2006. Figuring out the englishisation of Europe. In Reconfiguring Europe. The
Contribution of Applied Linguistics, C. Leung & J. Jenkins (eds), 65–86. London: Equinox.
Redder, A. 2008. Functional pragmatics. In Handbook of Interpersonal Communication, G. Antos,
E. Ventola & T. Weber (eds), 133–178. Berlin: De Gruyter.
Redder, A. 2013. Multilingual communication in Hamburg metropolitan area – A pragmatic
approach. In Multilingualism and Language Diversity in Urban Areas [Hamburg Studies on
Linguistic Diversity 1], P. Siemund, I. Gogolin, J. Davydova & M. Schulz (eds). Amsterdam:
John Benjamins.
Rehbein, J. 1977. Komplexes Handeln: Elemente zur Handlungstheorie der Sprache. Stuttgart:
Metzler.
Rehbein, J. 1994. Theorien, sprachwissenschaftlich betrachtet. In Texte und Diskurse. Methoden
und Forschungsergebnisse der funktionalen Pragmatik, G. Brünner & G. Gräfen (eds), 25–67.
Opladen: Westdeutscher Verlag.
Rehbein, J. 2010. Sprachen, Immigration, Urbanisierung – Elemente zu einer Linguistik
städtischer Orte der Mehrsprachigkeit. In Recerca i gestió del multilingüisme. Algunes
propostes des d’Europa – Mehrsprachigkeitsforschung und Mehrsprachigkeitsmanagement,
P. Comellas & C. Lleó (eds), 81–116. Münster: Waxmann.
Rehbein, J. & Kameyama, S. 2004. Pragmatik. In Sociolinguistics/Soziolinguistik. An Internation-
al Handbook of the Science of Language and Society, U. Ammon, N. Dittmar, K. Mattheier & P.
Trudgill (eds), 556–588. Berlin: De Gruyter.
Capturing diversity
Hagen Peukert
University of Hamburg*
Introduction
* This research was carried out within the research group on multlingual spaces in urban
areas at the LiMA excellence cluster initiative.
Hagen Peukert
reason, and McIntosh (1967) takes the square root from the Simpson Index to
receive a better representation of points in a coordinate system. Mathematically,
all these indices are based on Simpson’s idea and share an equal core, that is,
they add the squares of all proportions of the members of each of the species.
Such a relation is easy to understand for its simplicity and less error-prone
interpretation. However, it only captures distributional information of one par-
ticular kind.
The Shannon Weaver Entropy (1949) is different in that it weighs each propor-
tion by its own logarithm. This is a major contribution because the logarithm
comprises some unique properties that cannot be found in any other function, as
shown by Shanon and Weaver (1949). In equal distributions, the sum of the loga-
rithms becomes highest, indicating that the entropy − the state of disorder − is at
its maximum. It is this quantity that makes the index suitable to measuring diver-
sity. The state of highest disorder is equal to the highest diversity, i.e. all members
of a set are equal in number and so cannot be predicted on the basis of differing
probabilities. As for Simpson’s Index, Shannon’s idea is the point of origin for many
other indices (Ricotta & Avena 2003: 362), but the logarithmic relation also bears
a disadvantage. A negative consequence of the logarithmic properties is some
skewing between the edges. Attention therefore has to be paid when interpreting
the data. The Shannon Weaver Entropy gives reasonable results for the two ex-
tremes − very equally distributed and very unequally distributed − but it is skewed
if values fall between these extremes (Pielou 1966). To illustrate a similar effect, we
could mention the bias occurring when reading the arithmetic mean without
looking at the spread of the population. In theory, we can control such effects by
testing the data for deviations and interpreting the results in light of additional
indicators analogous to considering the median or standard deviation when inter-
preting the arithmetic mean.
In this paper, we attempt to lay down how language diversity can be reliably
captured based on the ideas of Simpson (1949) and Shannon and Weaver (1949)
by trading off their disadvantages. We adjust the indices to language and see how
the diversity of languages can be easily compared, especially when distributed
across the urban space.
In this section, the Simpson Index and the Shannon Weaver Entropy are dis-
cussed. It should become clear how and what the indices measure and why they
should be considered in diversity measurements. In addition, we propose de-
scendants of these two indices as established measures in other scientific fields,
Measuring language diversity in urban ecosystems
such as ecology. These derivations are especially useful for a more thorough
evaluation of data deviations in time and space. We argue here that all numbers
together provide a better picture of both biodiversity as well as language
diversity, whereas neither all peculiarities of language nor of biology can be
accounted for.
It is a coincidence that the frameworks of both indices were published in the
same year and that both discoveries are driven by probability distributions.
Whereas Simpson became known for finding a statistical paradox rather than
his diversity index and was known to a limited number of scientists only,
Shannon’s information entropy is well-known and is applied to many disciplines,
even though his work was never meant to be used as an index as implied here.
In what follows, we look at the Simpson Index first and elaborate somewhat
more on the Shannon Entropy. While the indexical relation is straightforward
and intuitive in the former, the latter needs more explanation in terms of its
evolution.
Simpson’s idea is based on “the probability that two individuals chosen at ran-
dom and independently from the population will belong to the same group”
(Simpson 1949: 688). He shows that the exact probability λ1 can be approximated
to ι under the assumption that G marks the size of groups, N the number of indi-
viduals, whereas n = N underlying an adjusted Poisson distribution:
(2) i1
ni ni 1
N N 1
Provided that samples of size 0 (infinite diversity) or 1 (no diversity at all) are
excluded,2 the approximation, λ ≈ ι, is also true for varying sample sizes. Indeed, it
is legitimate to simplify the Simpson Index further: n − 1 and N − 1 are set to n or
N respectively when species from two different areas are compared. In the field,
however, this version of the index is quite established for measurements without
spatial distinctions. Consequently, data for small sample sizes are skewed some-
what because the oversimplification denies that every specie is always in its own
group and should not be calculated as an additional event of its own probability.
1.
1
z
k 1 d ...d k 1
2
i
(1) ...
e
0 0 k 1
1... z
2 1 z
Zk 1
i
2. Therefore the λ of the Poisson distribution as shown in footnote 1 has to be adjusted for
leaving out the initial term.
Hagen Peukert
As the case study of St. Georg in the next section will reveal, the Simpson Index is
less reliable for small samples and is therefore suited to studying data of larger
quantities. The bias is not present in large samples. Substituting pi for ni/N as the
squared proportion of abundance, the formula now reads:
G
(3) S p 2
i
i1
The index is defined between zero and one, which marks the two extreme scenar-
ios: infinite diversity and no diversity at all. All values in between gradually mea-
sure the degree of diversity, that is, the higher the index, the less diverse the entity
under investigation. The reciprocal behaviour of the index is intricate and so the
inverse (1/S) is often used for ease of interpretation, i.e. the index rises when the
diversity rises. It is also good practice to subtract the index value from one. The “1”
represents the complete homogeneity of an environment. The remainder gives the
distance either to complete diversity where the inverse is used, or to full homoge-
neity where the simple index value is used.
The framework of the Simpson Index can be a starting point for many applica-
tions, such as distances and distance relations to densities of qualitative and quan-
titative data (e.g. McIntosh 1967). We restrict ourselves to the general framework
here since it best serves to understand ideas for measuring language diversity. In
this sense, we now turn to a second measure. The Shannon Weaver Index deserves
more elaboration on where it comes from in order to understand its appropriate-
ness for measuring language diversity.
Since Shannon thought in terms of binaries as the basis of electronic circuitry
used to transfer information through a channel of limited capacity, his original
derivation is somewhat intuitive to start with. Emanating from the idea3 that each
digit only possesses two states – zero and one – and that each information can be
transferred in terms of n digits, the maximum size of an information channel can
only be 2n by definition. The information represented by 2n seems redundant be-
cause it represents every possible sequence of n digits. To achieve the original goal,
i.e. coping with scarce resources of information transmission, one can minimize
the channel capacity by reducing the information content to the least possible
representation. The channel capacity for only one sequence (i.e. its probability of
occurrence) is much less than for all possible sequences, namely its inverse 2-n or
3. Shannon himself refers to Nyquist (2002) for deriving the logarithmic relation between the
analogical signal and the discrete information content: max. bit rate = 2W log2L, where W is the
maximal frequency and L is the analogical signal.
Measuring language diversity in urban ecosystems
1/2n. Taking the negative binary logarithm of the sequence probability, would
again give us the number of digits n.
1
(4) n log 2
2n
(5) I c log 2 p
This equation is the first ingredient in the entropy measure. It is important to note
that syntactic information − the specific order of the digits − and its probability
become “isomorphic” (Lyre 2002: 18) through the logarithmic connector.
The information content, Ic, as of yet could represent any information load in
binary digits. It would make sense to measure the most elementary piece of infor-
mation as one binary digit (bit). From the logarithmic relation described above,
follows that one unit of Ic with its two possible states having the same probability
(0.5) equals log22.4 log22 carries the same information content as tossing a coin.
Winning the German lottery needs about 23.74 bits and rolling a double costs 5.16
bits of information. These examples intuit that information complexes of any one
system can be added up. Shannon’s contribution here is not only to see that infor-
mation is additive and is composed of information of smaller contents/sizes, but
also that Pascal’s expectation applies.5 The expectation of a variable is the sum of
its values multiplied by their probabilities:
EV X xi pi x2 p2 ... xn p2 or rewritten as
(7) n
EV X p x i i
i1
The information entropy is really the expected value of the information content.
So putting equations 7 and 5 together gives us the information entropy. That means
we substitute Ic for x in equation 7 and, following convention, label the informa-
tion content Ic with the Greek letter eta.
n
(8) H p log
i 2 pi
i1
1
0,8
0,6
H(p)
0,4
0,2
0
0 0,2 0,4 0,6 0,8 1
P
Figure 1.
is changed to the Euler number e since binary calculations are no longer necessary
(see e.g. McIntosh 1967: 400) and the ease of calculating derivatives from the natu-
ral logarithm is substantial. Derivatives are needed for finding extrema and carry-
ing out proofs.
Shannon’s information entropy as it is now falls short of being comparable in
different temporal and spatial settings. Hence, normalizing the entropy value
would enable the researcher to make meaningful comparisons across various enti-
ties, be they biotopes, ecotopes, or urban areas. In ecological studies, the observed
occurrences of animal or plant individuals of one group describe their probability
and so they are equated with those observations. Applied to languages in a city
district, all observed occurrences of one language within one spatial entity are
added and divided by the total of all languages in that district. The number repre-
sents the proportion of that language in the district. At the same time, it gives the
probability of occurrence.
To achieve a normalized measure that enables us to compare different spaces,
but also to see where one index skews the data, we consider the Rényi Entropy
(Rényi 1961) in equation 9, which describes entropies more generally. Henceforth,
the Shannon Entropy is treated as a special case of the Rényi Entropy. For limα→1
the maximum entropy value (Hmax) of the Shannon Entropy is reached.
N
1
(9) H
1
log p
i
i1
0 ; 1
The advantage of the generalised form of the Shannon Entropy, the Rényi Entropy,
for the present case is its universality. It gives us the means to depict the data in a
well-arranged and compact way. Hill (1973) proposes to take the antilog of the
Rényi Entropy by exponentiating to the basis of the log, here e. Such a conversion
is also useful for measuring language diversity because it compresses all indices
needed here in one formula. Moreover, relations of various α can be used as nor-
malizations. For reasons of clarity, Hα is set to Nα = exp(Hα).
1
N 1
(10) N
pi
i1
The equation now fulfils important properties for measuring linguistic diversity in
a compact form. The following specifications are of interest:
Hagen Peukert
N
(11) E ,
N
Object of investigation
Major European cities are typical destinations for immigrants. Migration impacts
the cityscape in various ways. Urban spaces dominated by high shares of different
Measuring language diversity in urban ecosystems
We use an adjusted version of linguistic landscaping as our main method, that is,
systematically taking pictures in a predetermined manner according to well-
defined criteria. In short, 173 houses in 15 streets over six blocks in two adminis-
trative units were investigated. 17 different languages from a total of 303 language
cases were identified. The indices proposed above were calculated at street level
and at district level. Calculations on the house level as the smallest unit seemed
unreasonable for the large variance in the data that was expected (most houses
reveal only German or English, very few have two or three different languages). In
Hagen Peukert
what follows, the exact procedure is described in terms of the established stages of
collecting, preparing, and analysing data.
First, data collection comprises taking a preferably representative amount of
written material from the areas of interest. For the city district, we follow the spa-
tial categorizations of houses, streets, blocks, and administrative units. The small-
est spatial unit is the house, which is clearly determined by its number. For each
house, pictures are taken showing the entire facade and, if present, details from all
written signs visible from the outside.
Second, all photo material is processed. Languages are identified and counted
as types and tokens. Ambiguous cases, e.g. German transcriptions in Arabic or
vice versa, are treated as exceptions and allotted to respective categories. Illegible
instances like menu displays or handwritten tags in showcases are excluded. No
distinctions are made between bottom-up and top-down signs or handwritten and
printed signs.
Third, the data is arranged in a spreadsheet. The matrix specifies three spatial
units (house, street, administrative unit) and the indices (see Table 1). The indices are
defined as functions of all languages found, and so they can be automatically calcu-
lated for each spatial unit (examples are given in Table 1 for administrative units and
in Table 2 for the street level). The distance of occurrence between the languages
determines the set of dominant languages required for the Berger Parker Index.
Reaching this decision can be tedious and sometimes involves a biased justification.
In the present case, however, the data is unambiguous (see Figure 2) and therefore
unproblematic (English and German). Lastly, we compare and interpret the data.
Results
5007 5004
German German
English
English
Farsi
Italian
Turkish
Arabic French
Chinese Portuguese
Italian Russian
Urdu Hindi
Afghan Arabic
Japanese
Chinese
Spanish
Thai
French
Hindi Turkish
Figure 2.
The closer N1 and N2 to either extreme, the more the index tends to reflect domi-
nant or non-dominant languages respectively. This is important to keep in mind
when interpreting the indices that follow. These distances are supposed to uncover
the skewed data of the Shannon Entropy.
The next group of indices, marked by capital E, are normalized by either the
number of languages encountered in the specified area (total: E1,0 or log: ln(E1,0)
and Eln(1,0) or by the Simpson Index itself E2,1). Finally, the last four lines show the
traditional measures as components of the complex index forms.
Table 1. Dimensions of linguistic diversity in two administrative units in St. Georg
Table 1 makes it easy to compare and contrast the differences, strengths and weak-
nesses of each index. It becomes apparent that one index is insufficient for a com-
plete picture of linguistic diversity. To illustrate, scrutinizing the N-dices in relation
to each other reveals how minority languages are assimilated in the index. So look-
ing at unit 5007 tells us that minority languages have a greater effect there than in
5004. This is confirmed by the E-dices in the second group where we find that the
most dominant languages impact less (Theil Index: 1.18 in 5004 versus 1.07 in
5007). All other E-dices do not contradict each other, which cannot be derived
from the simple forms alone (the last group in Table 1).
Yet Table 2 contains several examples where the indices hint at peculiarities in
the distribution of multilingual signs, thus highlighting the shortcomings when
only one index is applied. To be precise, the indices in the Steindamm prompt us
to look at the data more carefully. The J-value (Pielou’s evenness) is larger than the
more conservative E2,1-value that equals out much of the distributional outliers.
This is the only street where this is the case. Compare for example the respective
values for the Danziger Straße. The distance between J and E2,1 here is almost dou-
ble as much, although the dominance index is equal (ln(E1,0)). We can conclude
that not a single outlier would raise the Pielou evenness so drastically above
the “entropy-simpson-relation” in the Steindamm. Checking the data, we do
indeed find a subcluster of languages occurring with almost equal frequency.
Table 2. Indices for various streets displaying linguistic dimensions of multilingualism
German 35 German 40
English 20 English 15
Arabic 5 Italian 5
Turkish 6
Farsi 5
Spanish 1 Thai 1
Urdu 2 Arabic 1
Afghan 2 Chinese 1
Chinese 3 Turkish 1
Russian 2
French 2
Hindi 2
Portuguese 3
In comparison to Lange Reihe, one outlier raises J, but does not have the same
strong effect to outperform E2,1 (see Table 3). According to the indices, the outlier
of five occurrences of Italian is not sufficient to reflect a new distributional pattern
in the Lange Reihe. On the other hand, for the Steindamm, the distance of the
J-value and E2,1 indicates that structure that cannot be smoothed out is existent. As
an extreme case of monolingualism, Kirchenweg is added to Table 2. In this in-
stance, only German occurred on three signs.
Discussion
Linguistic landscaping is not originally designed to give reliable data on the actual
distribution of multilingual speakers. The method is supposed to deliver a
description of the visually discernible symbolic environment. Arguments have
been raised on retrieving representative samples of the linguistic capabilities of
individuals, however we do not discuss these attempts here. In how far the written
language reflects the density of multilingual speakers is part of a different debate
(see Breckner et al. forthcoming, 2013).
In the present study, we merely map the existing linguistically diverse land-
scapes to indices and enquire whether the indices support our initial intuition that
diversity has many different flavours. We have seen that this is superficially the
case. But at the same time, some surprises arise in the following analysis.
Hagen Peukert
11,00
Lange Reihe
Index value
Steindamm
6,00 Kirchenweg
Complete diversity
1,00
N0 N1 N2 N∞
Figure 3.
The question to be unravelled here aims at whether the comparisons between the
chosen spatial units make sense. To demonstrate this idea we refer to the specific
example of St. Georg in Hamburg. The first two columns in Table 2 exhibit the data
for Steindamm and Lange Reihe in this district.
The data can be construed in two different ways as we find two different kinds
of distribution. At first glance, the simple indices seem to outline a clear picture.
For Steindamm the Simpson Index (S) is smallest or its inverse N2 is highest, which
would classify the street as more diverse than Lange Reihe. The entropy (H) con-
firms this measure in both variants (N1 and H). The skewing of the indices is neg-
ligible, but still larger for Lange Reihe. If we depict it graphically (Figure 3), the
N-dices draw quite a straight line in the middle of the two extremes represented by
the monolingual Kirchenweg and the upper straight line as an ideal of complete
diversity for Lange Reihe.
Turning then to the complex E-dices, usually we would leave Hill’s evenness
aside because it has different population sizes. These values would have to be nor-
malized by street length, some other spatial measure, or e.g. the number of houses
in that street. Otherwise, data can be easily misinterpreted, that is, for both ex-
tremes Hill’s evenness would be seen as equal, namely 1. Since the two streets that
we look at now are about equal in size in terms of space and population, we can use
this number and get an initial indication that diversity in Steindamm is not preva-
lent in every respect.
The dominance (the second E-number) also reveals a more differentiated pic-
ture. Although in absolute numbers the dominant languages are exactly the same,
they have a much stronger effect in the Steindamm, where we have, of course,
fewer language types. German dominates significantly more in Lange Reihe,
whereas there are fewer differences between German and English in Steindamm.
Essential to grasping the diversity met in both streets are the last two E-dices.
Again, due to their normalization, it is essential to note that the smaller the values
Measuring language diversity in urban ecosystems
become, the larger the diversity will be. Moreover, a small distance between the
two suggests that the majority of the languages occur with similar frequencies. For
Lange Reihe the numbers are quite close to each other and Pielou’s evenness (J) is
smaller. This indicates that languages other than the dominant ones (i.e. German,
English) are rather equally distributed since J is more sensitive to outliers.
In comparison, Steindamm has fewer language types, however, occurring at
higher token frequency. J outperforms the conservative E2,1 measure and thus
guides us to a different kind of diversity. In this case, the languages cluster to more
and less frequent ones. The former are more robust, the latter show a similar pat-
tern as in Lange Reihe (see Table 3).
The central question is which area is more diverse? Should it be the robust-
ness, the symmetries in the distribution, or the pure number of languages encoun-
tered? All of these dimensions of multilingualism are reasonable criteria. We
would argue for Lange Reihe although it loses ground in one of the dimensions and
is slightly more skewed to unilingualism in figure 3. That means we trade off the
dimensions as equally weighted qualities.
Lange Reihe is more diverse because all E-dices point in the same direction,
i.e. more languages are distributed more equally along the street. In comparison,
Steindamm contains more instances of fewer language types. Of course, these fre-
quencies weigh the simple indices much more. So they display a different dimen-
sion of multilingualism. Since Lange Reihe has the advantage of outperforming
Steindamm not only in the number of different languages but also in the distribu-
tion criteria, it is fair to attribute a higher diversity there.
It is also possible that a street dominates all dimensions identified here: more
types and tokens are equally distributed. To this end, the dimensions would not
have to be balanced against each other. As an example, the reader may contrast any
other street in Table 2 to either Lange Reihe or Steindamm.
General remarks
We have seen that the measurement of diversity in language and ecological do-
mains bears some resemblances, which lead us to assimilate and adjust research
carried out in the natural sciences. Indices development for describing the diver-
sity of plants and animals are also suitable for understanding linguistic diversity
under the assumption that biological species and languages share common
properties. To a certain extent this overgeneralization is acceptable. However, we
cannot deny that there are differences between both domains. It would be illumi-
nating to see if these differences could be reflected in index values and, where
possible, to assimilate those features in a new index. The indices proposed here do
Hagen Peukert
not account for all characteristics of language. Biological and linguistic systems
bear differences that remain unconsidered at the present stage of research.
First and foremost, no index pays tribute to the various forms of plurilingual-
ism of individuals. They can be manifold. For example, individuals may speak
more than one language, whereas species as physical entities have no such quality
of representing more than themselves. Not all of these languages present in the
speaker’s mind are represented to the outside world, be it in oral or written form.
They may, however, take effect subconsciously.
Second, languages of minority groups may become dominant and disappear
in short time intervals as well. Also, language mixing and attrition are fundamen-
tally different processes from the procreation of biological species that are in gen-
eral predictable on the basis of environmental factors. The dynamics of language
change, in contrast, depend on both intrinsic and extrinsic influences.
Third, several cases of language death and language revival are reported in the
literature. As an illustration, for biological species, there is usually no resurrection
from death to life. Languages may disappear and revive without disclosing notice-
able attrition. In fact, for reasons of social or political expediency all speakers of a
language might stop speaking it for years and begin using the standard language
exclusively. Once the depression ceases, the language may be revived. During the
time of the depression, the language still exists in the heads of the speakers, al-
though the researcher cannot find any traces of it. This property of language, as of
any other cognitive ability, is a general problem for proving the existence of the
phenomenon.
Lastly, for reasons of social prestige and economics, displayed languages may
not always represent the ethnicities present in an area. For example, Romanians
use Italian in their restaurant because it sells better. To the best of the author’s
knowledge this occurs in an unparalleled way in urban ecosystems.
Summary
The present study proposes a row of indices each of which is parameter instantia-
tions of one general equation. These are normalized versions of the Simpson Index
and the Shannon Weaver Entropy being crucial in pointing out the differences in
linguistic diversity. However, they have to be checked for consistency using a range
of further indicators. To understand how linguistically diverse a specified urban
area is, all of the presented indices should be considered. Research carried out in
the natural sciences prompted the present study in linguistics.
Coming back to the initial argument mentioned in the introduction, we
have to ask how knowledge of linguistic diversity facilitates decision makers. In
Measuring language diversity in urban ecosystems
response, we raise three arguments as the main results of the present study. If large
scale data are available, a reliable average number for linguistic diversity can be
calculated. The variances thereof can be cross-correlated with data on e.g. socio-
economic background such as income, profession, or educational background.
Positive or negative effects of linguistic diversity can then be grounded in empiri-
cal facts.
The numbers imply a much more differentiated picture of linguistic diversity.
Areas that appear to be less diverse might turn out to have a much higher degree
of linguistic diversity than previously assumed. This knowledge forces decision-
makers to view multilingualism from different angles, be it in specific situations
(e.g. street level) or wider areas (e.g. district). Both spaces can now be compared.
Finally, we refer to the nature of quantitative analyses in general. Individual
cases do not play a major role and weaken the bias towards intuition, stereotypes,
and prejudices. Individual cases and outliers are not taken as standards, but
decrease in significance with a large enough sample. In conclusion, there should
be little doubt that quantitative studies permit us to arrive at a more balanced un-
derstanding of linguistic diversity.
References
Fisher, A.R., Corbet, A.S., & Williams, C.B. 1943. The relation between the number of species
and the number of individuals in a random sample of an animal population. Journal of
Animal Ecology 12: 42–58.
Gibbs, J.P. & Martin, W.T. 1962. Urbanization, technology, and the division of labor –
International patterns. American Sociological Review 27(5): 667–677.
Gorter, D. 2006. Linguistic Landscape: New Approach to Multilingualism. Clevedon: Multilingual
Matters.
Hill, M.O. 1973. Diversity and evenness: A unifying notation and its consequences. Ecology
54(2): 427–432.
Hunter, P.R. & Gaston, M.A. 1988. Numerical index of the discriminatory ability of typing sys-
tems: An application of Simpson’s index of diversity. Journal of Clinical Microbiology 26(11):
2465–2466.
Johansson, V. 2008. Lexical diversity and lexical density in speech and writing: a developmental
perspective. In Working Papers, D. o. L. a. P. Lund University (ed.), 61–79. Lund: Lund
University.
Koerner, E.F.K. & Asher, R.E. (eds) 1995. Concise History of the Language Sciences: From the
Sumerians to the Cognitivists. Oxford: Pergamon.
Krebs, C.J. 1989. Ecological Methodology. New York NY: Harper & Row.
Lyre, H. 2002. Informationstheorie: Eine philosophisch-naturwissenschaftliche Einführung.
München: Wilhelm Fink Verlag.
Magnussen, S. & Boyle, T.J.B. 1995. Estimating sample size for inference about the Shannon-
Weaver and the Simpson indices of species diversity. Forest Ecology and Management 78:
71–84.
Malvern, D. 2008. Lexical Diversity and Language Development: Quantification and Assessment.
Basingstoke: Palgrave Macmillan.
Malvern, D.D. & Richards, B.J. 1997. A new research tool: mathematical modelling in the mea-
surement of vocabulary diversity. Child Language Bulletin 18: 2–2.
Malvern, D.D. & Richards, B. 2000. A new method of measuring lexical diversity in texts and
conversations. TEANGA 19: 1–12.
McCowan, B., Hanser, S.F., & Doyle, L.R. 1999. Quantitative tools for comparing animal com-
munication systems: information theory applied to bottlenose dolphin whistle repertoires.
Animal Behaviour 57: 409–419.
McIntosh, R.P. 1967. An index of diversity and the relation of certain concepts to diversity.
Ecology 48(3): 393–404.
Mlodinow, L. 2009. Wenn Gott würfelt. Hamburg: Rowohlt.
Nyquist, H. 2002 [1928]. Certain topics in telegraph transmission theory. Proceedings of the
IEEE 90(2): 280–305.
Pielou, E.C. 1966. Shannon’s formula as a measure of specific diversity: Its use and misuse. The
American Naturalist 100(914): 463–465.
Pielou, E.C. 1969. An Introduction to Mathematical Ecology. New York NY: Wiley.
Pielou, E.C. 1975. Ecological Diversity. New York NY: Wiley.
Pinker, S. 1994. The language Instinct. New York NY: Morrow.
Rényi, A. 1961. On measures of entropy and information. In Proceedings of the 4th Berkeley
Symposium on Mathematics, Statistics and Probability, J. Neyman (ed.), 547–561. Berkeley
CA: University of California Press.
Ricotta, C. & Avena, G. 2003. On the relationship between Pielou’s evenness and landscape dom-
inance within the contexts of Hill’ss Diversity profiles. Ecological Indicators 2: 361–365.
Measuring language diversity in urban ecosystems
L3 acquisition has been a topic of increasing interest throughout the past decade.
Nevertheless, linguistically-oriented work has so far paid little attention to the
acquisition of foreign languages by second generation minority (or heritage)
speakers, who have acquired two languages in early childhood. Our paper
investigates article acquisition by German-Turkish speakers, who have grown
up in Germany. Specifically, we address the question whether these speakers’ L3
English shows cross-linguistic influence (CLI) from Turkish (their chronological
L1) or German (their chronological L2). Results indicate the absence of negative
CLI from Turkish and positive CLI from German, their L2. The study can be
seen as supporting the view that typological proximity is crucial for CLI, while
also allowing for the conclusion that language dominance plays a role.
1. Introduction
Throughout the past decade there has been an increasing number of studies inves-
tigating L3-acquisition (see Falk & Bardel 2010, for an overview).1 While many of
the early studies have focused on the acquisition of the lexicon (e.g., Cenoz 2001;
De Angelis 2005; Tremblay 2006; Jessner 2003, 2006), recent years have seen
1. The term L3 is used here in terms of a language that has chronologically been acquired after
the second language. It can be the fourth, fifth, etc., language, although in our case it happens to
coincide with the third language acquired.
Tanja Kupisch, Neal Snape and Ilse Stangen
growing interest in the area of syntax and morphology (e.g., Bardel & Falk 2007;
Flynn, Vinitskaya, & Foley 2004, Leung 2005, 2006). Our paper will be concerned
with article use in obligatory contexts – a primarily syntactic phenomenon which
is dependent on semantic, pragmatic and (to a lesser extent) lexical factors.
It is more or less uncontroversially accepted that cross-linguistic influence
(CLI) occurs both at the onset and the final state of L3 acquisition. One of the most
prominent issues in generative L3 studies concerns the interplay between previ-
ously acquired linguistic systems with the target L3. In other words, which will be
the source of CLI into the L3? Is the L1 the only possible source of CLI (Leung
2006), does CLI occur from the chronologically last system acquired (the L2)
(Bohnacker 2006; Bardel & Falk 2007), or can CLI come from either L1 or L2
(Flynn, Vinitskaya, & Foley 2004), perhaps depending on typological distance
(De Angelis 2005, Rothman & Cabrelli Amaro 2010; Rothman 2011) or perceived
(psycho-typological) distance (Kellermann 1979)?
Several factors that could potentially determine the source of CLI have been
discussed. These include L2 status (a tendency to transfer from the most recently
acquired language), recency and frequency (how recently was/how frequently is
the language used?), age of acquisition and proficiency. There is evidence from
individual studies, in part contradictory, that each of these is relevant.
To our knowledge, linguistically-oriented research has so far not explicitly
dealt with the issue of CLI into L3 in cases of child L2 acquisition, i.e., cases where
the L2 has been acquired in early childhood and where the L1 constitutes a minor-
ity language in the larger national community. By contrast, there has been some
research in the domain of education showing that multilingualism is beneficial to
the acquisition of foreign languages at school. For example, Hesse & Göbel (2009)
have shown that teenagers with a multilingual background perform better in
L3-English than monolingual German learners of English, although lagging
behind their monolingual peers in their abilities in German. Other studies
(e.g., Kessler & Paulick 2010) confirm that children growing up bilingually per-
form similarly to monolingual children in L3-English, if one of their languages is
German, while children who came from multilingual non-German speaking fam-
ilies have disadvantages. Although these studies are impressionistic from a linguis-
tic viewpoint (by not targeting specific phenomena)2, they raise the interesting
question whether an L1 can be beneficial for the acquisition of an L3 despite hav-
ing linguistic properties that deviate from both L2 and L3.
This is a study into the acquisition of articles in L3-English by 15 adult speak-
ers who have acquired German and Turkish during their childhood in Germany.
2. Another problem which makes direct comparisons difficult is that definitions of multilin-
gual and bilingual diverge and are sometimes even left to the participant to decide.
Foreign language acquisition in heritage speakers
For practical purposes, we will follow De Houwer (1995), using the term bilingual-
ism for cases in which the child has been exposed to the second language no later
than one month after birth with regular (daily) exposure to both languages, while
counting cases with exposure after the 1-month age limit as instances of early L2
acquisition (De Houwer 1995: 223). When adopting this definition we do not mean
to imply that there are any cut-off points during this early period that determine
ultimate acquisition outcomes. The participants of our study all grew up with two
Turkish-speaking parents. Since they were exposed to Turkish from birth but to
German between the ages of 1 and 7 years, they can be considered early L2 learners
of German. As adults, most are active users of both languages, using German more
often than Turkish in their daily lives and considering themselves to be more pro-
ficient in German – the dominant language of their environment.
Our study will show that there is no negative CLI from L1-Turkish to L3-English
in article production. The absence of such influence is unexpected, given that pre-
vious research on articles in child learners of German (L1-Turkish) has shown
problems with article production (e.g., Pfaff 1992, Schönenberger 2010), possibly
due to CLI from Turkish. Provided that L1-Turkish influences L2-German nega-
tively, negative CLI on L3-English may be expected by implication.3
The paper is structured as follows. Section 2 will provide an overview of the
nominal domains of Turkish, German and English (in relation to our participants’
L1, L2 and L3). Section 3 summarizes previous research on the acquisition of
English by 3 groups of learners, including L1 speakers of articleless languages
(Turkish and Russian) as well as L1 speakers of German. Section 4 provides our
predictions for CLI. Section 5 presents our study. Section 6 discusses the results.
We draw conclusions from our findings in Section 7.
3. This paper includes no analysis of CLI from L1-Turkish to L2-German, although our study
allows for some conclusions on this matter.
Tanja Kupisch, Neal Snape and Ilse Stangen
Since Abney (1987), it is assumed that articles are functional elements which
occupy the head position (D) of Determiner Phrases (DPs) and take nouns (NPs)
as their complements. In the following, when referring to nouns without articles
or other determiners (e.g., car) we will use the term ‘bare NP’ and when referring
to nouns preceded by articles (e.g., a car, the car) or other determiners (e.g., de-
monstratives) we use the abbreviation DP.
In German and English, articles can appear in different contexts which can be
specific or non-specific: the [–definite, +specific] context is when the speaker uses
the indefinite article to signify a specific referent in the discourse which has not
been introduced previously (e.g., I bought a new car); the use of the indefinite
article in the [–definite, -specific] context means there is no specific referent
(e.g., I want to buy a new car). The speaker uses the definite article in a [+definite,
+specific] context which indicates that the speaker and hearer share knowledge of
the referent in the discourse situation (e.g., the car that I told you about). In
Turkish, these distinctions are not expressed by means of articles.
2.1 Turkish
Öztürk (2005) argues that Turkish is a language which lacks the syntactic category
DP, unlike German and English which have fully specified DPs. In Turkish there is
no definite or indefinite article that would correspond directly to the articles found
in German and English. Öztürk (2005) proposes that Case in Turkish is the
counterpart of determiners in other languages. Definiteness is marked only with
accusative DPs, which take an accusative case suffix when they are definite, as in
example (1).
(1) Ali kitab-ı okudu
Ali book-Acc read
‘Ali read the book.’
In subject position in example (2), bare nouns can be either definite or indefinite
(with arbitrary reference), although bare subject NPs are usually definite, because
the sentence initial position is generally occupied by topics.
(2) Çocuk güldü.
child laugh
‘The child laughed.’ (taken from Öztürk 2005)
Though there is no direct equivalent of an indefinite article there is (optional) in-
definite bir which attributes an indefinite reading to the noun when it is unstressed
(see 3a).
Foreign language acquisition in heritage speakers
(3) a. kitáp
‘(the/a) book’
b. bir kitáp
‘a book’ (taken from Öztürk 2005)
2.2 German
German has a fully specified DP and the articles (der/die/das, etc. and ein/eine,
etc.) are functional elements which head the category D. German articles (and
other determiners) are marked for gender, number and Case, but this is not rele-
vant to our study. Singular count nouns (CNs) in subject and object positions must
be syntactically licensed by an article, as in (4).
(4) a. Ein/der Junge spielte mit einem/dem Freund von nebenan.[±definite]
[singular CN]
a/the boy played with a/the friend of next door
If no definite or indefinite article precedes the singular CN Junge or Freund the
sentence is ungrammatical, as indicated with the use of *, in example (5).
(5) *Junge spielte mit Freund von nebenan. [singular CN]
boy played with friend of next door
With plural CNs and singular mass nouns (MNs), bare nouns are allowed if the
reading is nonspecific, as in (6).
(6) a. Im Garten spielen Jungen mit Autos. [–definite] [plural CN]
in-the garden play boys with cars
b. Wir brauchen Brot und Wein. [–definite][singular MN]
we need bread and wine
Plural count and mass nouns are also bare if the reading is generic, as shown in
(7).
(7) a. Katzen sind Säugetiere.
cats are mammals
b. Wein ist gesund.
wine is healthy
Although bare NPs are generally restricted to plural CNs and singular MNs, there
are some exceptions in German where singular CNs can appear as bare NPs. Typ-
ically, they are found after prepositions (8a–c), with incorporated nouns (8d), or in
Tanja Kupisch, Neal Snape and Ilse Stangen
predicative position (8e). As the translations show, these often require an article in
English.
(8) a. Der Wecker funktioniert ohne Batterie.
the alarm works without battery
‘The alarm works without a battery.’
b. Sie sah einen Mann mit Hut.
she saw a man with hat
‘She saw a man with a hat.’
c. Ich schicke den Brief per Post.
I send the letter by mail
‘I’ll send the letter by mail.’
d. Wir werden Auto/Boot/Bus/Fahrrad/Kanu fahren.
we will car/ boat/bus/bike /canoe go
‘We will go by car/boat/bus/bike/canoe.’
e. Er möchte Lehrer werden.
he wants teacher become
‘He wants to become a teacher.’
2.3 English
Similar to German, English has a fully specified DP and also encodes definiteness
through the use of the definite article (the) and the indefinite article (a). Singular
CNs in subject and object positions must be syntactically licensed by an article, as
in (9).
(9) a/the boy played with a/the friend from next door
[±definite] [singular CN]
If no definite or indefinite article precedes the singular CN boy or friend the sen-
tence is ungrammatical, as in example (10).
(10) *boy played with *friend from next door [singular CN]
As in German, with plural CNs and singular MNs, bare NPs are allowed if the
reading is nonspecific, as in (11), or generic, as in (12).
(11) a. Boys are playing with cars in the garden.
b. We have to buy wine and bread.
(12) a. Cats are mammals.
b. Wine is healthy.
Foreign language acquisition in heritage speakers
Again, similar to German, there are some exceptions in English where singular
CNs can appear bare in subject (13a and 13b) and object positions (13c and 13d),
as in the following examples from Stvan (2007). However, the contexts in which
English unusually allows bare singular CNs differ from those where German al-
lows them. Stvan (2007) does not consider these contexts to be idiomatic because
they are quite productive (Stvan 2007: 174). Nevertheless, we assume that such
cases are not subject to general rules of article use because generally bare NPs must
be preceded by determiners and are not permitted with any kind of noun,
(cf. (14e)).
(13) a. School never closed.
b. Church has become very important in my life.
c. Need help setting up camp?
d. Which days will you be on campus?
e. *Need help setting up tent?
In fact, all literal translations of (13) into German are ungrammatical. This means
that contexts such as (13) may lead learners to produce errors of article overuse in
L3-English if there is transfer from German.
3.1 L1-Turkish
Studies by White (2003) and Goad and White (2004) have provided a detailed
account of an L1-Turkish L2-English speaker named SD by focusing on article
production. SD is a very advanced learner of English but despite her advanced
proficiency she continues to experience problems in article use in production. In a
series of interviews conducted by White (2003) SD spoke about a range of topics
and upon transcribing the conversations with SD the interviewer found a number
of article omissions and articles which received non-target like stress (see Snape &
Kupisch 2010 for details). White (2003) argues for a missing surface inflection ac-
count as the problem SD faces is the mapping of inflection features to surface
phonology which results in omission.
An alternative account for SD’s article omissions is offered by Goad and White
(2004) who argue that prosodic transfer from L1-Turkish is the likely cause. A
prosodic account goes further than a missing surface inflection account as Goad
and White (2004) show that higher article omission is predicted when the article
precedes an adjective, e.g., the young man, as the Turkish prosodic structure and
the English prosodic structure differ in this type of structure. There is no equivalent
Tanja Kupisch, Neal Snape and Ilse Stangen
3.2 L1-German
4. Note, however, that there are different views on this issue (see White et al. 2011).
5. An anonymous reviewer commented that one finds considerable amount of words of
French origin in Turkish. Given the impact of French on English, one could speculate that
Turkish could indeed help when learning English. However, this would not be the case in the
acquisition of articles.
6. Evidence to support the claim that German shows properties of Turkish in some domains
of grammar would be consistent with studies showing disadvantages of bi- or multilingual
children as compared to monolinguals in their German proficiency (e.g., Hesse & Goebel
2009).
Foreign language acquisition in heritage speakers
favour such a view. Pfaff (1992) compares the development of two Turkish-
dominant children with a German-dominant (“essentially monolingual”) child
between the ages of 1 and 5 years, growing up in Berlin. She reports (ibid., 282)
that “the proportion of zero article usage for Turkish children is much higher than
for the German monolingual child and persists much longer”. Schönenberger
(2010) has studied production data from four successively bilingual Turkish-
German children, showing that after 30 months of exposure to German the
children still omitted articles between 20% and 30% and the omission rate stag-
nated. She speculated that article use may be subject to incomplete acquisition in
German. Unfortunately, the data collection stops at that point and it remains a
matter of speculation whether these children eventually converged to the target
system or whether they failed to fully acquire the phenomenon within the relevant
critical phase.7 Moreover, if the two studies are on the right track, problems with
article use in L3-English are expected: CLI from the speakers’ L1 could overrule L2
influence, because properties of the L1 have become part of the L2 system. An
undesired implication of this finding would be that bilingualism has also negative
effects on the acquisition of subsequent languages in life.
Our study sets out to explore whether the acquisition of L3-English shows
CLI from L1-Turkish or L2-German. In the remainder of this section we formu-
late the predictions for CLI from a quantitative and qualitative perspective. CLI
should lead to different acquisitional results, depending on the source of
influence.
Quantity of influence
i. CLI from German to L3-English should be positive and result in a relatively
low rate of errors; CLI from Turkish should be negative and result in a higher
rate of errors.
ii. Since monolingual L1-Turkish speakers have problems with article use in spo-
ken production and L1-German speakers do not, participants who speak
Turkish frequently should be more likely to transfer from Turkish than par-
ticipants who speak Turkish less frequently.
iii. Relatively high proficiency in Turkish (and relatively low proficiency in
German, possibly due to a late AoO) might go along with negative CLI into
L3-English.
7. To our knowledge, nobody has even defined a critical age for article acquisition. It is rea-
sonable to assume that different article functions are acquired at different ages, which would be
consistent with a view of multiple sensitive periods. However, since articles cease to be omitted
at a very early age, this sensitive period, if it exists, should be associated with a fairly early age,
perhaps age 3–4 (see also Section 5.2 for discussion).
Tanja Kupisch, Neal Snape and Ilse Stangen
4. Our study
The data was collected as part of a project on L3 acquisition carried out within the
LiMA cluster (Linguistic Diversity Management in Urban Areas) at the University
of Hamburg.9 The following study is based on naturalistic data that was gathered
in pre-structured interviews of approximately 20 minutes in duration. All but 2
interviews were carried out by native speakers of English. The interviews were re-
corded with a Microtrack II digital recording device and subsequently transcribed
with FOLKER (Schmidt & Schütte 2010). The recordings contained a total of 7778
utterances. 5562 (71.5%) of these utterances were produced by the participants
and 1528 (19.6%) of them contained one or several NPs and/or DPs.
4.2 Participants
8. One could further speculate that problems with indefinite articles are restricted to particular
contexts. According to classic work in the grammaticalization of articles (e.g., Givón 1981), in-
definite articles grammaticalize from numerals. The contexts in which they become obligatory
first are contexts of specific reference, as in I saw a (specific) book [–definite, +specific]; only at
more advanced stages of grammaticalization, they become obligatory in [–definite, -specific], as
in I would like to buy a book (but I don’t know which one). Since bir has not yet become obligatory
it is still in an incipient stage of grammaticalization. As a consequence, CLI from Turkish to
English should lead to higher omission rates in [–definite, -specific] contexts. As we will see,
participants have very few problems with indefinite articles, so the discussion becomes obsolete.
9. We wish to thank Katrina Walsh and Marina Zielke for their participation in the data col-
lection and transcription. Special thanks to Deniz Akpinar, who started the project with us, and
Cigdem Güney, who helped continue it!
Foreign language acquisition in heritage speakers
families in the city. None of the participants grew up in a bi-national family. Par-
ticipants were between 20 and 34 years old (mean = 24 years). There were differ-
ences with respect to the age at which participants had their first intensive contact
with German, ranging between 1 year (nursery school) and 7 years (primary
school), and in the extent to which they continued using Turkish during adoles-
cence and adulthood.
We originally intended to assess relative language proficiency (i.e., the weaker
and stronger language) through a timed cloze test with 30 blanks which required
filling in free morphemes and content words. According to the cloze test 4 of the
participants scored better in German, 6 scored better in Turkish, while 4
participants performed at ceiling (90% or more) in both Turkish and German.10
We suspect, however, that the cloze test was not a good indicator of proficiency, as
it did not match the participants’ self-assessed proficiency.11 On a 5-point scale
(1 = medium proficiency, 2 = good, 3 = very good, 4 = excellent, 5 = -native-like)
participants rated themselves 4.7 on average in German. 11 out of 15 participants
considered themselves native-like in reading, writing, speaking, comprehension
and overall abilities, and these included speakers with an AoO before and after age
3. By contrast, self-estimated proficiency in Turkish was only 3.1 on average and
the few speakers who considered themselves to have native-like abilities in
Turkish did not coincide with the latest learners of German.12 In terms of relative
use of the two languages, most participants (n = 10) reported to use German more
often than Turkish daily, and most also had more German- than Turkish-speaking
contacts.
All participants started to learn English at school at the ages between 6 and 12
(mean = 10 years). Self-assessed proficiency was either intermediate (n = 8)
or advanced (n = 7). Accuracy in the cloze test ranged between 20 to 85%
(mean = 69%). Most participants (n = 9) have been to an English-speaking coun-
try for a maximum of two weeks, one for 6 weeks, one for a year, and four have
never been to any English-speaking country.
4.3 Methods
All utterances were coded for the presence and correct use of articles, bare NPs,
proper names, and determiners other than articles (e.g., demonstratives, posses-
sive pronouns, numerals, etc.). In the following, we will focus on definite and in-
definite article use as well as bare nouns, excluding determiners other than articles
as well as proper names.
Some utterances were discarded from the analysis. We excluded isolated
(non-argumental) noun phrases from the counts because most of the time these
cannot be clearly coded in terms of whether an article is required or not. We ex-
cluded nouns that were preceded by hesitation fillers (e.g., uhm), as they are not
clearly distinguishable from articles. We discarded instances where the word
preceding the article has been repaired or interrupted (e.g., I think more about
d- German culture). Finally, we also excluded utterances containing mixed noun
phrases (e.g., it’s a agglutinierende Sprache).
5. Results
Our first analysis (Table 1) shows the distribution of appropriate definite and in-
definite marked DPs and bare NPs, as well as the overall number of inappropriate
definite and indefinite DPs and bare NPs. The analysis is based on 1528 instances
of NPs and DPs. Overall, only 5.5% of these were inappropriate uses. As Table 1
shows, definite articles are used incorrectly more often than indefinite articles and
the most common mistake was to use a bare NP where it is not required. A com-
parison between definites and indefinites shows a significant difference (χ2 = 11.29,
p<0.001) in incorrect uses. The Fisher’s exact probability test (two-tailed) reveals
that there is no significant difference between incorrect definites and bare NPs
(χ2 = 1.93, p= 0.164), while there is a significant difference between incorrect uses
of indefinites and bare NPs (χ2 = 18.04, p<0.0001).
Table 1. Appropriate and inappropriate article use and bare NPs
The contexts in which articles were omitted would have required definite articles
18 times, indefinite articles 17 times, and other determiners 5 times.
Examples of inappropriate article use are provided in (14) through (16) for
incorrect use of indefinite articles, definite articles, and bare NPs, respectively. For
many of the relevant nominals, the translation equivalent in German would be
grammatical (e.g., 14c, 15a, 15b, 15c, 15d, 15e, 16a, 16d, 16e). Moreover, inappro-
priately used articles also involve instances of overuse (e.g., 14a, 14b, 14c, 15a, 15b,
15c, 15e).
(14) a. ... I always try to &er to to &hm to make a progress in my English
language. (D05)
b. &oh yeah they made a study &uhm or a field work. (D03)
c. and now when I’m just like finished to study like a social science.
(D19)
d. yeah I’m looking in a news in in &umm internet. (D16)
(15) a. i began at th- &umm class three at the primary school. (D02)
b. the American English I don’t like much but +... .] (D02)
c. that’s all what I’m doing all the day. (D01)
d. ... this semester I’m going to &er make the master in linguistic. (D07)
e. and decide whether you go to the islam or y- you go to the christianism.
(D07)
(16) a. Long time my father worked at the harbour. (D04)
b. Also now here in Latin at university. (D07).
c. Because I always go to Turkey in summer. (D09)
d. My mother is just housewife. (D10)
e. I’m &er studying &er French and Maths in order to become teacher.
(D04)
As mentioned previously, it may be argued that appropriate article use and omis-
sion reflect different types of knowledge (specifically lexical vs. syntactic and se-
mantic), depending on whether articles are part of fixed expressions or not. We
therefore recalculated the data, excluding all expressions that may have been learnt
as chunks (e.g., at the end, at the moment, in the middle of, a little, a lot of, a few,
once a month, at home, at school, on holiday, by plane, by car, etc.) This recalcula-
tion decreased the rate of errors to 5.2% overall, which means that some errors
were lexically driven.
In summary, the analysis has shown that the overall rate of inappropriate ar-
ticle use and omission is very low in the dataset (~5%). Since Turkish does not
have a fully grammaticalized article system, one would have expected higher rates
of article omission if there had been negative CLI from Turkish, as reported for
Tanja Kupisch, Neal Snape and Ilse Stangen
5.2 Inappropriate article use vis-à-vis frequency and recency of Turkish use
Our first analysis explores whether individual performance in English can be re-
lated to frequency and recency in the use of Turkish. Specifically, if there were CLI
from Turkish, one would expect it to be strongest for those speakers who are fre-
quent and recent users of Turkish. There were two ways in which we assessed
frequency of use in Turkish. First, we asked each participant to indicate the num-
ber of Turkish-speaking contacts giving them a choice between the categories 0,
1–5, 6–10, 11–20, and more than 20. Second, we asked them to indicate the fre-
quency of their daily use of Turkish relative to their daily use of German, giving
them a choice between 100% Turkish, 75% Turkish (25% German), 50% both lan-
guages, 75% German (25% Turkish), and 100% German.14 Tables 2 and 3 show the
number of speakers in each category as well as the corresponding mean rates of
inappropriate NP/DP-use. Note that a high number of Turkish-speaking contacts
did not always coincide with a higher frequency in the daily use of Turkish.
13. Another question to be addressed in the future is whether or not prosodic transfer from
L2-German has anything to do with article use in spoken production in L3-English, given that
our participants rarely omitted articles. Perhaps being early learners of German, with English
being their third language, these speakers have prosodic structures available to them unlike the
L1-Turkish L2-English speaker in Goad and White’s (2004) study.
14. Since there was only one speaker who claimed to have no Turkish-speaking contacts, we
collapsed the categories 0 contacts and between 1–5 contacts.
Foreign language acquisition in heritage speakers
No. of participants 5 3 4 3 15
inappropriate NP/DP use (%) 30/534 24/472 17/254 13/268 84/1528
(5.6%) (5.1%) (6.7%) (4.9%) (5.5%)
no. of participants 0 5 8 2 15
inappropriate NP/ – 26/602 53/726 5/200 84/1528
DPs use (%) (4.3%) (7.3%) (2.5%) (5.5%)
Finally, one could also be tempted to assume that a late AoO in German correlates
with a higher amount of inappropriate NP/DP-use in English. One line of reason-
ing could go as follows: a speaker’s Turkish will be stronger if the AoO in German
is late. This is because in the typical context of heritage language development, the
heritage language receives less support (i.e., it is heard and used less frequently)
with entry in kindergarten or school, where typically only German is spoken. In
other words, the later a bilingual child comes into contact with German, the more
time there is for Turkish to develop. It is reasonable to assume that a highly profi-
cient speaker of Turkish is more likely to transfer from Turkish into L3-English
than a less proficient speaker of Turkish, because we know that L1 speakers of
Tanja Kupisch, Neal Snape and Ilse Stangen
Turkish have problems with articles in L2-English (see above). If this line of rea-
soning is along the right track, late onset of German should coincide with lower
accuracy in English article use.
Yet, a problem with the rationale just outlined is that general language pro-
ficiency and AoO in German appears to be unrelated in our participants’ two
languages. Recall that participants rated themselves to be more proficient in
German (4.7) than in Turkish (3.1), and the speakers who considered them-
selves to be native-like in Turkish were the ones with the latest AoO in
German.
Of course, one could assume that proficiency in article use is related to AoO in
German, without being related to overall proficiency. In other words, speakers
may have excellent language skills overall, but not in the domain of articles. In fact,
articles are among the earliest function words that children acquire and omission
generally ceases before age 3 in German (e.g., Kupisch 2008). Given the assump-
tion of multiple sensitive periods in language acquisition, one could assume that
the sensitive phase for articles is around the ages between 3 and 4, as has been
proposed for some aspects of morpho-syntax in Meisel (2011). Provided there is
indeed such a sensitive period between age 3 and 4, some of the participants in our
study may have missed it.
As Figure 1 shows, however, this assumption is not entirely borne out. Al-
though the 5 participants with a relatively late AoO in German (after age 3) per-
form less target like (mean 8.8%) than the 10 participants with an earlier AoO
(ages 1 to 3) (mean 5.8%), there is no linear increase in the rate of errors with in-
creasing AoO in German.
On the other hand, proficiency in L3-English article use seems to be related to
general proficiency in English, as measured by the cloze test (Figure 2).
25
Target deviant article
20
use/omission (%)
15
10
0
0 2 4 6 8
AGE of onset in German
Figure 1. Target deviant article use and omission with respect to AoO in German
Foreign language acquisition in heritage speakers
18
use/omission (%)
12
9
6
3
0
0 5 10 15 20 25 30
Cloze test results
Figure 2. Target deviant article use and omission with respect to AoO in German
and general proficiency in L3-English
The aim of this study was to find out whether German-Turkish bilinguals who
acquired German as an early L2 show CLI from their chronological L1 (Turkish)
or from their chronological L2 (German) when acquiring English as a third lan-
guage. Our analysis was based on the nominal domain where Turkish is different
from English and German by not having a fully grammaticalized article system.
As for the quantity of influence, we formulated the following predictions:
(i) CLI from L2-German to L3-English should have resulted in a low rate of inap-
propriate NP/DP use, while CLI from L1-Turkish should have yielded a relatively
high rate of inappropriate uses. (ii) Speakers who use Turkish frequently should be
more prone to making errors (due to negative CLI from Turkish) than speakers
who use Turkish less frequently. (iii) High proficiency in German should go along
with positive influence into L3-English, while high proficiency in L1-Turkish
(possibly do due an early AoO in German) should go along with negative CLI
from L1-Turkish to L3-English and the absence of positive CLI from L2-German.
Moreover, it was speculated whether late AoO in German would coincide with a
higher rate of inappropriate article use or omission.
As for (i), the rate of inappropriate NP/DP-uses was relatively low (~5%),
contradicting the assumption of negative CLI from Turkish. As for (ii), frequen-
cy of Turkish use appeared to be unrelated to the rate of inappropriate article use.
As for (iii), there did not seem to be a direct relationship neither between AoO
in German and overall language proficiency nor between AoO in German and
(in)appropriate article use or omission. Therefore, regardless of whether or not
there is a critical period for article use around the age of 4 years (which is a mat-
ter of debate), learners seem to benefit from early exposure to German when
Tanja Kupisch, Neal Snape and Ilse Stangen
learning English, no matter whether they start to acquire German before or after
age 4.
With regard to the quality of cross-linguistic influence, we expected the fol-
lowing: (iv) If there had been CLI from Turkish, article omissions should have
been more frequent in contexts where definite (and not indefinite) articles are re-
quired, and (v) if there had been CLI from German, problems should have been
most severe in contexts of lexically determined article use since these show differ-
ences between German and English. As for (iv), inappropriate omissions were as
frequent in contexts where definites were required (n = 18) as in contexts where
indefinites were required (n = 17), again, contradicting influence from Turkish.
Moreover, inappropriate realizations of articles were often not substitution errors
but errors of overuse, again, contradicting Turkish influence. Further, it was in-
deed the case that participants were less successful in lexically determined con-
texts, even though the difference was minimal (5.5% vs. 5.2%). This result shows
further that our participants have acquired the semantic and the syntactic princi-
ples of article use (or successfully transferred them from German). Moreover,
some of the (few) instances of incorrect article use looked like direct translations
from German. On the other hand, the fact that inappropriate article use was sig-
nificantly less frequent with indefinite than with definite DPs may be taken to
suggest that the learners are also influenced positively by their Turkish grammar,
which has indefinite bir, but no definite article.
Taken together, the results of the individual analyses suggest there is little to
no negative CLI from L1-Turkish to L3-English, no matter how frequent and how
proficient our participants spoke their L1.
We have not tested participants’ use of German articles, but by implication it
is likely that they have also successfully acquired the German article system, at
least as far as syntactically and semantically appropriate article use and omission is
concerned. If there had been incomplete acquisition of German due to negative
influence of Turkish, participants should have failed to successfully acquire Eng-
lish articles, contrary to what we have shown. However, it should be noted that
previous studies have shown that multilingual children, despite showing advan-
tages over monolinguals in learning English as a foreign language, nevertheless lag
behind their monolingual peers in German (cf. Hesse & Göbel 2009). The latter
raises the question about our participants’ skills in German, which is something to
be explored in a future study.
Overall, our study implies that successively bilingual acquisition has no nega-
tive consequences for foreign language acquisition, regardless of whether the L1 is
typologically distinct or not. It remains open whether CLI into L3 is determined by
typological proximity or language dominance, because our participants seem to be
more proficient in German than in Turkish (even though our picture, of which
Foreign language acquisition in heritage speakers
language is stronger, based on a cloze test and self-assessment, is not perfectly clear).
In other words, if there is CLI from German (and not Turkish), as our results sug-
gest, the question remains whether this happens because German is typologically
closer to English (than Turkish), or spoken with higher proficiency than Turkish.
Admittedly, our results are based on a relatively small number of participants
and it needs to be confirmed on the basis of a larger sample. Nevertheless, if these
results can be generalized, they have implications for teaching English in class-
rooms (as an L2 or an L3), suggesting that articles are not a vulnerable domain for
German-Turkish speakers (at least not if they are highly proficient speakers of
German), even though Turkish does not have a fully grammaticalized article
system. Moreover, if there can be positive CLI from German to English, as our
results indeed suggest, instruction on this grammatical domain could be alike for
German-Turkish and monolingual German children, focussing on those aspects
of article use where German differs from English. However, as just pointed out,
our participants had acquired German and Turkish successively. Even those who
acquired German relatively late had had four years of contact with German and
were proficient speakers of the language before starting to acquire English. The
situation may be different in cases where AoO in German and English coincide, as
is the case in schools where English is taught from the first grade on.
References
Abney, S.P. 1987. The English Noun Phrase in its Sentential Aspect. PhD dissertation, MIT.
Bardel, C. & Falk, Y. 2007. The role of the second language in third language acquisition: The
case of Germanic syntax. Second Language Research 23: 459–484.
Bohnacker, U. 2006. When Swedes begin to learn German: from V2 to V2. Second Language
Research 22: 443–486.
Cenoz, J. 2001. The effect of linguistic distance, L2 status and age on cross-linguistic influence in
third language acquisition, In Crosslinguistic Influence in Third Language Acquisition:
Psycholinguistic Perspectives, J. Cenoz, B. Hufeisen, & U. Jessner (eds), 8–20. Clevedon:
Multilingual Matters.
De Angelis, G. 2005. Interlanguage transfer of function words. Language Learning 55(3):
379–414.
De Houwer, A. 1995. Bilingual language acquisition. In The Handbook of Child Language,
P. Fletcher & B. McWhinney (eds), 219–250. Oxford: Blackwell.
Falk, Y. & Bardel, C. 2010. The study of the role of the background languages in third language
acquisition: The state of the art. International Review of Applied Linguistics in Language
Teaching (IRAL) 48: 185–219.
Flynn, S., Vinnitskaya, I., & Foley, C. 2004. The Cumulative Enhancement Model for language ac-
quisition: Comparing adults and children’s patterns of development in first, second and third
language acquisition of relative clauses. International Journal of Multilingualism 1: 3–16.
Tanja Kupisch, Neal Snape and Ilse Stangen
Givón, T. 1981. On the development of one as an indefinite marker. Folia Linguistica Historica 2:
35–53.
Goad, H. & White, L. 2004. Ultimate attainment of L2 inflections: Effects of L1 prosodic struc
ture. In EuroSla Yearbook, S. Foster-Cohen, M. Ota, M. Sharwood Smith, & A. Sorace (eds),
119–45. Amsterdam: John Benjamins.
Goad, H. & White, L. 2009. Prosodic transfer and the representation of determiners in Turkish-
English interlanguage. In Representational deficits in SLA: Studies in Honor of Roger
Hawkins [Language Acquisition and Language Disorders 47], N. Snape, Y-k. I. Leung & M.
Sharwood Smith (eds), 1–26. Amsterdam: John Benjamins.
Hesse, H.-G. & Göbel, K. 2009. Mehrsprachigkeit als Kapital: Ergebnisse der DESI Studie. In
Streitfall Zweisprachigkeit? The bilingualism controversy, I. Gogolin & U. Neumann (eds),
281–287. Wiesbaden: VS Verlag für Sozialwissenschaften.
Jaensch, C. 2009. Article choice and article omission in the L3 German of native speakers of
Japanese with L2 English. In Second Language Acquisition of Articles: Empirical Findings
and Theoretical Implications [Language Acquisition and Language Disorders 49], M. del
Pilar García-Mayo & R. Hawkins (eds), 233–263. Amsterdam: John Benjamins.
Jessner, U. 2003. On the nature of crosslinguistic interaction in multilinguals, In The Multilin-
gual Lexicon, B. Cenoz, B. Hufeisen, & U. Jessner (eds), 45–55. Dordrecht: Kluwer.
Jessner, U. 2006. Linguistic Awareness in Multilinguals: English as a Third Language. Edinburgh:
EUP.
Kellermann, E. 1979. Transfer and non-transfer: Where are we now. Studies in Second Language
Acquisition 2: 37–57.
Kessler, J.-U. & Paulick, C. 2010. Mehrsprachigkeit und schulisches Fremdsprachenlernen:
Englischunterricht bei Lernern mit Migrationshintergrund. In Fachunterricht und Deutsch
als Zweitsprache, B. Ahrenholz (ed.), 257–278. Tübingen: Narr Francke.
Kupisch, T. 2008. Determinative, Individual- und Massennomen im Spracherwerb des Deutschen:
Diskussion des Nominal Mapping Parameters. Linguistische Berichte 214: 129–160.
Leung, Y-k. I. 2005. L2 vs. L3 initial state: A comparative study of the acquisition of French DPs
by Vietnamese monolinguals and Cantonese-English bilinguals. Bilingualism: Language
and Cognition 8(1): 39–61.
Leung, Y-k. I. 2006. Full transfer vs. partial transfer in L2 and L3 acquisition. In Inquiries in
Linguistic Development. In Honor of Lydia White, R. Slabakova, S. Montrul, & P. Prévost
(eds), 157–88. Amsterdam: John Benjamins.
Meisel, J.M. 2011. First and Second Language Acquisition. Parallels and Differences. Cambridge:
CUP.
Öztürk, B. 2005. Case, Referentiality and Phrase Structure. Amsterdam: John Benjamins.
Pfaff, C.W. 1992. The issue of grammaticalization in young bilinguals’ second language
development. Studies in Second Language Acquisition 14: 273–296.
Rothman, J. & Cabrelli Amaro, J. 2010. What variables condition syntactic transfer?: A look at
the L3 initial state. Second Language Research 26(2): 189–218.
Rothman, J. 2011. L3 syntactic transfer selectivity and typological determinacy: The typological
primacy model. Second Language Research 27(1): 107–127.
Schmidt, T. & Schütte, W. 2010. FOLKER: An annotation tool for efficient transcription of
natural, multi-party interaction. In Proceedings of the Seventh Conference on International
Language Resources and Evaluation [LREC’1], N. Calzolari & K. Choukri (eds), Valletta,
Malta: European Language Resources Association (ELRA).
Foreign language acquisition in heritage speakers
Introduction
Wortham (2001) notes that speaking with a certain voice means using words that
presuppose some social position because these words are characteristically used
by members of a certain group. In this chapter we will look at the voices of multi-
lingual people studying and teaching in complementary schools and describe how
key linguistic practices and words presuppose particular identity positions and
allegiances. However, we will also show how these are negotiated and refined by
participants in these voluntary school settings. We will draw on Bakhtin’s (1994)
concept of heteroglossia because is serves the purpose of highlighting the utter-
ance as socially, politically and historically ‘entangled’ with the voices of others.
Adrian Blackledge and Angela Creese
participants in complementary schools are imbued with the voices of others which
have gone before, but which are negotiated anew in the here-and-now.
Blommaert 1999; Blommaert & Verschueren 1998; Gal 1998; Gal & Woolard 1995;
Kroskrity 1998; Woolard 1998).
In the research reported here the relations between ‘language’ and ‘ideology’
are far from straightforward. Complementary schools exist in relation to, in re-
sponse to, and perhaps even in spite of, a strongly-felt public discourse of mono-
lingualism and homogeneity in the multilingual, heterogeneous state. The impetus
towards the erasure of minority immigrant languages is resisted where comple-
mentary schools have been set up by communities which have gathered whatever
resources are at their disposal to teach and maintain the heritage/community lan-
guage. At the same time, complementary schools often appear to argue for a static,
reified version of ‘culture’ and ‘heritage’, which may be remote from their students’
experience (Blackledge & Creese 2008). In this complex ideological context com-
plementary schools become sites where subtle, nuanced negotiations of identities
frequently occur. We will see that negotiations are performed using a range of
linguistic resources which do not fit a neat categorization into one language or
another.
Translanguaging
Recently a number of terms have emerged, as scholars have sought to describe and
analyse linguistic practices in superdiversity, in which meaning is made using
signs flexibly. These include, but are not limited to: Flexible bilingualism
(Creese & Blackledge 2011); Codemeshing (Canagarajah 2011); Polylingual lan-
guaging (Jørgensen 2010; Madsen 2011); Contemporary urban vernaculars
(Rampton 2011); Metrolingualism (Otsuji & Pennycook 2011); and Translanguag-
ing (García 2009; Creese & Blackledge 2010). The shared perspective represented
in the use of these terms considers that meaning-making is not confined to the use
of ‘languages’ as discrete, enumerable, bounded sets of linguistic resources. Rather,
signs are available for meaning-making in communicative repertoires (Rymes
2010) which extend across ‘languages’ and varieties which have hitherto been as-
sociated with particular national, territorial, and social groups.
García uses the term translanguaging to describe the usual and normal prac-
tice of “bilingualism without diglossic functional separation” in New York class-
rooms (García: 2007: xiii). García (2009: 45) proposes that “when describing the
language practices from the perspective of the users themselves, and not simply
describing bilingual language use of bilingual contact from the perspective of the
language itself, the language practices of bilinguals are examples of
translanguag-
ing”. For García “translanguagings are multiple discursive practices in which bilin-
guals engage in order to make sense of their bilingual worlds” (2009: 45 – emphasis
Adrian Blackledge and Angela Creese
Heteroglossia
Blommaert (2012) argues that the contemporary semiotics of culture and identity
need to be captured in terms of complexity rather than in terms of multiplicity or
plurality. Indeed he argues that “a vocabulary including ‘multi-lingual’, ‘multi-
cultural’, or ‘pluri-’, ‘inter-’, ‘cross-’, and ‘trans-’ notions all suggest an a priori
existence of separable units (language, culture, identity), and they suggest that the
encounter of such separable units produces peculiar new units: ‘multilingual’ reper-
toires, ‘mixed’ or ‘hybrid’ identities and so forth.” Blommaert argues that a perspec-
tive which focuses on ‘code-switching’ is emblematic of this view. Bailey (2012: 501)
engages with the limitations of an approach to linguistic analysis which empha-
sizes code-switching, arguing that a focus on “constellations of linguistic features
that are officially authorized codes or languages, e.g. ‘English’ or ‘Spanish’ can con-
tribute to neglect of the diversity of socially indexical resources within languages.”
Bailey points out that if the starting-point is social meanings, rather than the code
or language in use, it is not central whether a speaker is switching languages, alter-
nating between a dialect and a national standard, register shifting, or speaking
monolingually in a variety that highlights language contact. Language, whether
monolingual or multilingual, carries social meanings through phonological, lexi-
cal, grammatical, and discourse level forms: “these forms index various aspects of
individuals’ and communities’ social histories, circumstances, and identities”
(Bailey 2012: 506). The stem ‘hetero-’ may, of course, be viewed as contiguous with
‘multi-’, ‘pluri-’, and so on. However, whereas previously sociolinguistic research
Heteroglossia in English complementary schools
Bakhtin views the word not as a material thing but rather “the eternally mo-
bile, eternally fickle medium of dialogic interaction” (1984: 197). For Bakhtin the
life of the word is contained in its transfer “from one mouth to another, from one
context to another context, from one social collective to another, from one genera-
tion to another generation” (1984: 202). Bailey (2012: 504) summarizes ‘hetero-
glossia’ as “the simultaneous use of different kinds of forms or signs, and the
tensions and conflicts among those signs, on the sociohistorical associations they
carry with them”. What we propose, then, is an analytic gaze that explicitly joins
the linguistic utterance in the present and the sociohistorical relationships that
give meanings to those utterances; a gaze that takes as its focus speakers as social
actors using heteroglossic linguistic resources to negotiate the social world. Such
an analytic gaze “encourages us to interpret the meanings of talk in terms of the
social worlds, past and present, of which words are part-and-parcel, rather than in
terms of formal systems, such as ‘languages’, that can veil actual speakers, uses, and
contexts” (Bailey 2012: 502).
Research context
The data presented here come from a research project which investigated the mul-
tilingual practices and identity performances of young people and their teachers
in complementary schools in four of Britain’s linguistic minority communities.
The research design is of four interlocking case studies focusing on Bengali schools
in Birmingham, Chinese (Cantonese and Mandarin) schools in Manchester,
Gujarati schools in Leicester and Turkish schools in London (Blackledge & Creese
2010). Data were simultaneously collected and shared across the full team through
the writing and exchange of fieldnotes throughout the data collection period. Each
case study identified two complementary schools in which to observe, record, and
interview participants. We also collected key documentary evidence, and took
photographs. After four weeks two key participant children were identified in each
school. These children were audio-recorded during the classes observed, and
where possible also for 30 minutes before and after each class over a six week pe-
riod. We interviewed key stakeholders in the schools, including teachers and ad-
ministrators, and the key participant children and their parents.
For the Chinese case study, we selected two schools in Manchester2 city centre:
one taught Cantonese and the other Mandarin Chinese (Creese et al. 2007). We
observed and collected linguistic data in a range of settings, including inside class-
rooms, at break times and in other formal school contexts such as prize-giving
ceremonies. Two ‘key participant’ students were selected from each class, in nego-
tiation with the teachers. We audio-recorded two key participant children in the
Cantonese school over four weeks, and four key participant children in the
Mandarin school over seven weeks. The teachers allowed us to record classes with
an audio-recorder placed on the teacher’s desk. We also video-recorded teachers
working with the key participant children, and interviewed the young people,
their parents, the teachers and members of the schools’ management committees.
In addition, we collected documents relating to each school’s policy, planning and
curriculum, and took photographs in a range of settings.
The long-established Cantonese school, founded in the 1970s, operated every
Sunday of the school year, from1.30 p.m. until 3.30 p.m. There were around 350
students on roll, and student recruitment was mainly through word of mouth.
Some of the pupils lived many miles away from the school. Around 30% of the
pupils’ families came from mainland China (the Cantonese-speaking region) and
70% of parents originated from Hong Kong. There were forty teachers and teach-
ing assistants. The school hired the facilities of the local further education college.
In contrast the Mandarin Chinese school in Manchester was newer, having been
established in 1996. The school used the facilities of Manchester University rent
free. It had around 180 students at the time of the study. Student recruitment was
mainly by word of mouth. There were 18 teachers altogether. All the language
teaching staff were female. Most of the teachers were mature students either doing
Masters or doctoral research degrees at universities in Manchester, or were settled
professionals. None of the teachers worked in the mainstream sector. The school
used textbooks published by People’s Education Press of China, purchased through
a London-based Chinese company.
At one point during a 10-week data collection period we recorded the following
interaction in the Cantonese school. The teacher is discussing identity in relation
to the Chinese national anthem.
(1) [students have been writing the words of the Chinese national anthem]
S1: [finishes writing and puts down his pen] I don’t know any of that.
S2: my parents forced me to sing the song
T: you are Chinese
S3: even if you live in England.
T: you are Chinese
S2: technically I am not Chinese
(classroom audio-recording, Cantonese school)
Adrian Blackledge and Angela Creese
Here the teacher makes available some options and limits others. She legitimizes
and values identities linked to Chinese nationality. However, it is clear even in such
a short example that the identity options available to individuals at a given mo-
ment in history are subject to negotiation and change. Indeed, the identity position
imposed by the teacher was immediately contested by her students, who provided
her with an alternative reading. In the process of negotiating identities, language
users may seek new social and linguistic resources which allow them to resist iden-
tities that position them in undesirable ways, produce new identities, and assign
alternative meanings to the links between identities and linguistic varieties. The
example here captures one view of identity which we heard in the complementary
schools in our study – one in which identity is inseparably associated with a terri-
tory (Freeland & Patrick 2004). However, ethnographic study of the school re-
vealed alternative discourses and voices. Despite the rather stark and seemingly
entrenched position taken around identity by the teacher here, an alternative dis-
course was also available in these schools, in which identity was more flexible, and
open to negotiation. Identity performance was revealed to be about speaker voice
rather than linguistic codes and boundaries. Different constructions of bilingual-
ism, ethnicity and language came into play as young people and their parents and
teachers interacted in and around these Chinese complementary schools.
Among the voices which argued against the unitary and territorially bounded
identity position of Example One were those which valued heteroglossic practices
such as translanguaging. Here the additional value and resource that bilingualism
brings to identity performance was highlighted. The next three examples are ex-
tracts from interviews with a parent, a teacher and a child, all of whom spoke of
the importance of drawing across languages to make meaning. The first example
comes from a parent (P) who is reminding the researcher (W) that code-switching
has a long and historical influence on those with diasporic links to Hong Kong.
(2) W: 您自己对这样子情形的看法怎么样?您喜欢人家这样用吗?
<What do you think of that? Do you like people doing that?>
P: 什么?< What?>
W: 就是这样汉语、英语、汉语、英语这样用?
<Just like this, using Chinese, English, Chinese, English?>
P: 我说是不对的,所以我不想要孩子这样做的。但是我自己看
到,在香港很多人都是这样的。你说是吗?因为,这样很奇
怪,这样都混在一起,大家讲不通啊,是吗?。。。给我一个
很怪的感觉,就是。。什么事都是乱糟糟的。应该要分开嘛,
中文就是中文,英语就是英语。不应该混在一起。但是我想在
香港长大,香港工作很多时候都是这样。我很少看到只光说广
东话的。。里面都是有很多的英语。我想就是一个普遍的问
题。
Heteroglossia in English complementary schools
I’d say that’s not right, so I don’t want kids to do that. But I can see that myself,
many people do that in Hong Kong. Don’t you say so? Because this is very strange,
everything’s mixed together, you cannot really communicate, can we? Give me a
very strange feeling, that is ... Everything’s in a confused manner. Should be sepa-
rated. Chinese is Chinese, English is English. They shouldn’t be mixed together.
But I think a lot of the time people do that growing up in Hong Kong, working in
Hong Kong. I rarely saw anyone just speak Cantonese only, there was always a lot
of English. I think it is a common problem.
The interview extract captures the simultaneous but conflicting constructions
of bilingualism referred to earlier. The speaker is troubled by the mixing of
Cantonese and English. She articulates a powerful ideology which insists on keep-
ing languages pure and separate, otherwise confusion and disorder will ensue.
However, she is also aware that both in Manchester and Hong Kong heteroglossic
practices persist, which do not insist on clear boundaries around languages. In-
deed heteroglossic practice in Hong Kong may also be evident in the practices of
multilingual young people in a Manchester complementary school classroom.
The next example articulates the importance of voice and repertoire, in an
interview with the head teacher (D) of the Cantonese school. Here the head teach-
er is talking to the researcher (W) in English about using his bilingualism in what-
ever ways work best to communicate with his daughters.
(3) D: In detail, my kids found it difficult to understand me fully in
Cantonese so in between I’d bring in some English.
W: What do you mean they don’t understand you fully? Some terms they
don’t understand? Or
D: I mean something maybe like obviously they do understand me in
simple conversation. But the detail – for instance, when my daughter’s
having problem with boys, asking “Daddy, do you like that boy?”
(inaudible) in more detail. Or maybe school which will have a long
long term effect on her. If it’s something important, I prefer the lan-
guage they prefer.
(Interview with head teacher, Cantonese complementary school)
In the head teacher’s account the enforcement of boundaries around languages
has no salience as he converses with his daughters about sensitive issues. Rather
the emotion, commitment and support the father wishes to show as he talks to
his daughters requires a heteroglossic practice which moves across linguistic
boundaries. Empathy and understanding are his priority. In the face of main-
taining good family relationships, in this instance at least, the importance of
language boundaries dissolves. In the next example one of the young people
makes a similar point.
Adrian Blackledge and Angela Creese
(4) KP: Yah. Or sometimes my mum couldn’t exactly understand English. So,
sometimes I just say .. uh. . a bit in English and then she asked me and
then I say in Mandarin.
R: Do they say anything about you doing that? You know, mixing the
two languages?
KP: Well. My dad and mum want to me to learn English and Mandarin.
R: So, they’re happy for you to use them mixed?
KP: Yes. Yes.
(Interview with student, Mandarin Chinese complementary school)
In this example the heteroglossic practices of drawing on different signs, typically
categorized as associated with particular languages, is presented as common and
usual practice in this bilingual home. Indeed the mixed vernacular itself appears
to be the home language rather than ‘English’ and ‘Mandarin’.
These interviews all provide accounts of linguistically diverse heteroglossic
practices in which a range of semiotic signs are used for meaning making. Like
family settings, complementary schools provide a multilingual implementational
space (Hornberger 2005) for the endorsement of this kind of heteroglossic and
flexible bilingualism. Although we have written elsewhere (Blackledge & Creese
2010) about the varied and sometimes oppositional constructions of bilingualism
present in complementary schools, bilingual practices which ignored language
boundaries in rapid translanguaging practices were common-place. In the next
example students are about to pack up at the end of the school day. On this par-
ticular Sunday a money box went missing at the school. This was announced by
teachers to their classes, and children were asked whether they knew anything
about it. Teachers were shocked and concerned that this should happen in their
school, and attempts to locate the missing money box occupied much of the after-
noon. At least one teacher searched the students’ bags. In the example, the chil-
dren are chatting to one another and playing with the digital voice recorder at the
end of the school session.
(5) (S: non-specific student, G: non-specific girl, B: non-specific boy,
T: teacher.)
G1: Is this recording?
G2: I think it is.
T: 糖紙掉入垃圾筒,不要留在班裏。<Put the sweet wrappers in the
bin, don’t leave them in the class.>
G1: do get off. ... just do.
S: why?
G1: it’s not yours.
S: so what?
Heteroglossia in English complementary schools
In an interview with the head teacher of the Chinese Mandarin school, we heard
of the importance of culture in the curriculum:
Adrian Blackledge and Angela Creese
(6) I think the Chinese school can provide children a chance to experience
the Chinese culture.... For example, we did this before, during Mid-
Autumn festival, we’d get some moon cake. And during Chinese New Year,
we’d talk about this and that.
(Interview with head teacher, Mandarin Chinese School).
In our interviews with parents of students in the school we frequently heard that
the Chinese school was both a place to learn Chinese language and culture, and
also a place for pupils to learn from a taught curriculum which was not available
in a home environment:
(7) I think on one hand, parents hope they won’t lose the language and culture
of their ancestors. And Chinese school being here, on one hand, people
have somewhere to go. If it’s not here, we’d probably have to hire a tutor or
something (Interview with parent, Mandarin Chinese School)
The cultural curriculum is a core undertaking of complementary schooling
(Blackledge & Creese 2010). The teaching about festivals was a particularly com-
mon activity and an aspect of this was the teaching of folk stories associated with
the festival (Creese et al. 2009). Example Eight provides a typical example of a
teacher using festivals to teach language and culture.
(8) T: 好,这是什么节日? <OK, what is this festival?> 我们上次讲了
<We talked about this last time>,中国的四大节日<The four major
Chinese festivals>。 四大节日里边有哪四大 <Four major festivals
include which four?> 我们再回忆一下<Let’s recall them again>。我
们有<We have> Spring Festival 是什么节日 <What festival is it>
S: 春节 <Spring Festival>。
T: 好<OK>,春节<Spring Festival>。然后下一个呢<What is the next
one> The next one?
Ss: 元宵节 <Lantern festival>
T: 元宵节<Lantern festival>,下一个<next one>。
Ss: 端午节<Dragon boat festival>。
T: 最后一个<Last one>...
Ss: 中秋节<Mid-Autumn festival>。
T: 中秋节<Mid-Autumn festival>,Mid-Autumn festival. 中秋节大家
吃[什么<What do people eat at Mid-Autumn festival>?
S: 月饼 <Moon cake>
T: 还做什么呢<What else do we do?>,一边吃月饼<Eating moon
cake>,一边
<While we?>
Heteroglossia in English complementary schools
very different example of ‘longing for’ and in doing so sets himself apart from
the social position adopted by his teacher, and the voices she represents. The
extract below is a continuation of Example Nine above. We see the teacher at-
tempt to bring together the different worlds of her students in another slightly
exaggerated example, but one which is less political. We also see the students
continue to make hay with what is a routine language teaching exercise as the
teacher moves on to teach another vocabulary item, ‘big and round’, related to
describing the moon.
比如说<for example>,我们都盼望家人团聚<we all long for/
(10) T:
look forward to family reunion>。For example, if you are here in
Manchester, your parents are back in China, and you have been
separated for years, you are looking forward to the reunion of the
family.
S: 盼望天气<long for/look forward to the weather >...
T: 好<OK>,下一个词 <next term>,又大又圆 <big and round>。
我们什么东西又大又圆 <what can we say is big and round>
S1: 月饼 <moon cake>。
S2: 月亮 <the moon>。
T: 月饼是不是又大又圆 <are moon cakes big and round>? [jokingly]
we got a special moon cake 又大又圆<big and round>。好<OK>
,还有什么东西又大又圆
<what else is big and round>?
S1: 月亮 <the moon>。
T: [除了月亮<apart from the moon>
Liu: 太阳<the sun>。
T: 太阳也是又大又圆<the sun is also big and round>。
Din: my head
T: 你的头<your head>。丁锐说他的头又大又圆 <D said that his
head is big and round>。It’s true. 对了<that’s right>。[Students
laugh] 好<OK>
(Classroom audio-recording, Mandarin Chinese school)
Here the students introduce their own examples of ‘long for’ and ‘big and round’.
There are different degrees of challenge and acquiescence to teacher authority in
this interaction, but the teacher refuses to acknowledge the challenges as serious
or oppositional. The teacher persists with the official curriculum of vocabulary
teaching while Liu introduces an element of parody, using the vocabulary item
itself to appropriate the activity and entertain his classmates. The students perform
a serious and non-serious stance toward the cultural teaching about the festival,
and are able to ‘have it both ways’ (Rampton 2006: 367). Bakhtin (1986: 88) noted
Heteroglossia in English complementary schools
that words enter our speech from the utterances of others, “thereby retaining to a
greater or lesser degree the tones and echoes of individual utterances”. For Bakhtin
the transmission and assessment of the discourse of others was one of the most
widespread and fundamental topics of speech: ”every conversation is full of trans-
missions and interpretations of other people’s words” (Bakhtin 1981: 337). In the
examples above we see multiple voices brought into the classroom. We see how
these voices circulate and how in the interactional moment utterances are shot
through with ideological underpinnings.
Conclusion
References
Kroskrity, P. 1998. Arizona Tewa Kiva speech as a manifestation of a dominant language ideol-
ogy. In Language Ideologies: Practice and Theory, B. Schieffelin, K. Woolard, & P. Kroskrity
(eds), 103–122. New York NY: OUP.
Li W. & Wu, C.-J. 2009. Polite Chinese children revisited: Creativity and the use of codeswitch-
ing in the Chinese complementary school classroom. International Journal of Bilingual
Education and Bilingualism 12(2): 193–212.
Madsen, L.M. 2011. Social status relations and enregisterment in contemporary Copenhagen.
Working Papers in Urban Language and Literacies 72. London: King’s College Publications.
Martin, P., Bhatt, A., Bhojani, N., & Creese, A. 2006. Managing bilingual interaction in a Gujarati
complementary school in Leicester. Language and Education 20(1): 5–22.
May, S. 2005. Language rights: Moving the debate forward. Journal of Sociolinguistics 9(3):
319–347.
Maybin, J. 2006. Children’s Voices. Talk, Knowledge and Identity. Basingstoke: Palgrave.
Otsuji, E. & Pennycook, A.D. 2011. Social inclusion and metrolingual practices. International
Journal of Bilingual Education and Bilingualism 14(4): 413–426.
Rampton, B. 2006. Language in Late Modernity. Cambridge: CUP.
Rampton, B. 2011. From “Multi-ethnic adolescent heteroglossia” to “Contemporary urban
vernaculars”. Language & Communication 31: 276–294.
Rymes, B.R. 2010. Classroom discourse analysis: A focus on communicative repertoires. In,
N. Hornberger & S. McKay (eds), Sociolinguistics and Language Education, 528–546. Bristol:
Multilingual Matters.
Vertovec, S. 2007a. Super-diversity and its implications. Ethnic and Racial Studies 30(6):
1024–1054.
Vertovec, S. 2007b. Introduction: New directions in the anthropology of migration and
multiculturalism. Ethnic and Racial Studies 30(6): 961–978.
Wang, S.C. 2008. The ecology of Chinese language in the United States. In Encyclopedia of
Language and Education, 2nd edn, Vol. 9, A. Creese, P. Martin, & N.H. Hornberger (eds),
169–182. New York NY: Springer Science & Business Media LLC.
Woolard, K. 1998. Introduction: Language ideology as a field of inquiry. In Language Ideologies:
Practice and Theory, B. Schieffelin, K. Woolard, & P. Kroskrity (eds), 3–47. New York NY:
OUP.
Wortham, S. 2001. Ventriloquating Shakespeare: Ethical positioning in classroom literature
discussions. Working Papers in Educational Linguistics 17(1–2): 47–64.
Wortham, S. & Locher, M. 1996. Voicing on the news: An analytic technique for studying media
bias. Text 16: 557–585.
Wu, Chao-Jung 2001. Learning Cultures: The Example of Learning Chinese as Community
Language in Chinese Schools in the UK. PhD dissertation, University of Leicester.
Wu, Chao-Jung 2006. Look who’s talking: Language choices and culture of learning in UK
Chinese classrooms. Language and Education 20(1): 62–75.
Enough is enough
The heuristics of authenticity in superdiversity
Introduction
1. This paper emerged in the context of the HERA project “Investigating discourses of inheri-
tance and identity in four multilingual European settings”, and was first discussed during a
meeting in Birmingham, May 2011. We are grateful to the participants of that meeting, and in
particular to Adrian Blackledge, Angela Creese, Jens-Normann Jörgensen, Marilyn Martin-
Jones, for stimulating reflections on that occasion. A preliminary version of this paper was cir-
culated as Working Papers in Urban Language and Literacies, paper 76.
Jan Blommaert and Piia Varis
intends to offer a realistic, yet generalizable, approach to inquiries into the com-
plexities of contemporary identity practices. Such practices now evolve in online
as well as in ‘offline’ contexts, and connections between both social universes are
of major importance for our understanding of what superdiverse society is about.
In this paper, we present a first and tentative set of generalizations based on
several recent pieces of ethnographic research into identity processes in a variety
of online and offline superdiverse contexts (e.g. Blommaert 2005, Chapter 8; Dong
2009; Wang 2010; Varis & Wang 2011; Vigouroux 2011; Li & Juffermans 2011; Li
et al. 2012; Van der Aa 2012; Williams 2012). In such work, several common fea-
tures could be distinguished. We saw (a) that contemporary identity work devel-
ops along ‘miniature’ aspects of life, leading to (b) complex repertoires of separate
orientations towards norms for being X or Y; (c) often formulated as ‘authentici-
ties’ rather than ‘identities’. People search for ideal articulations of small sociocul-
tural roles and positions, not for totalizing categorical identities. (d) And such
practices occur in online as well as offline environments; while the Internet, and
especially social media, offer rich environments for the production and
reproductions of such authenticities, the processes we observe are not confined to
online contexts. The scope of such processes cannot be determined at present;
more extensive and comparative work is needed for that, and we hope in this pa-
per to provide suggestions for such work.
This paper, thus, does not itself offer ethnographic narratives, but sketches
tentative and hopefully productive lines for future ethnographic work and pro-
poses an equally tentative vocabulary for it. The opening sections of the paper will
be general in tone and approach, and will contain the theoretical proposals we
intend to make, while the closing ones will offer illustrative vignettes supporting
these proposals.
Micro-hegemonies
The complexities of contemporary identity work are baffling, yet perhaps not en-
tirely new; what is new is the awareness of such complexities among academic and
lay observers. Late Modernity – the stage of Modernity in which the emergence of
superdiversity is to be situated – has been described as an era of hybridized,
fragmented and polymorph identities (e.g. Deleuze & Guattari 2001; Žižek 1994),
often also subject to conscious practices of ‘styling’ (Rampton 1995; Bucholtz &
Trechter 2001; Coupland 2007). Prima facie evidence appears to confirm this:
people do orient towards entirely different logics in different segments of life –
one’s political views may not entirely correspond to stances taken in domains such
as consumption, education or property. So here is a first point to be made about
Enough is enough
The range of features that can be employed in identity work in order to produce
authenticity can be wide and include a number of different, and sometimes very
elaborate semiotic means. However, in actual practice the features that produce
recognizable identities can be reduced to a very limited set, and here we see the
salience of what we called ‘dosing’. That is, mobilising an authentic identity dis-
course about oneself can be a matter of attending to the most infinitesimally small
details – sometimes even only observable to those ‘in the know’ – and a very small
number of recognizable items, such as a piece of clothing. 2
In enregistering such features, certain rules need to be observed for the process
to be successful – to be recognized by others as what was intended. These are the
rules that ‘newcomers’, ‘beginners’ and ‘wannabes’ need to observe and mobilize in
their own identity work in order to ‘pass’ as authentic someone (cf. e.g. Kennedy’s
2001 account on racial ‘passing’). This is where the Internet, for all the freedom
and opportunity it is seen as offering for creative identity-play, appears not only as
a useful instructional, normative source for the ‘wannabe’ but also as a space rife
with regulatory discourses on ‘how to’ be or become someone. YouTube, for in-
stance, features plenty of ‘how to’ videos – videos providing viewers with instruc-
tions on the minute details of how to be an ‘authentic’ gangsta or emo – that is, the
features that should be employed for an authentic identity as a gangsta or an emo
to be produced. The Internet now offers an infinite range of identity assembly kits
and complements them with volumes of users’ guides. Such identities are not nec-
essarily offered to replace others; they are offered as additional niches, and one can
walk in and out of them ad lib. The users’ guides, therefore, are the micro-
hegemonies we mentioned above.
The chav culture – a form of working-class British youth culture – is one ex-
ample of a subculture very visible on the Internet. A search online for anything
‘chav’ provides plenty of material for someone wishing to ‘chavify’ oneself
(although it should be perhaps pointed out that this is not amongst the most de-
sired and aspired to identities to be displayed by young Brits) – from YouTube
videos to images that put forward a ‘chav-semiotics’ where certain features are
iconic of a chav identity.
2. We can see ‘dosing’ also in the many studies of ‘styling’ now available in the literature since
Rampton’s (1995) pioneering work. Homeopathic doses of features – one sound sometimes, or
one word – can be enough to redefine the speakers in an interaction as well as the whole situa-
tion itself.
Enough is enough
3. All the images and web information appearing in this paper were retrieved on June 28, 2011.
Jan Blommaert and Piia Varis
As we can see from Figure 2, in the case of producing ‘chav’ this ‘active sub-
stance’ is the fashion feature we mentioned above: the British luxury brand
Burberry, with its signature tartan pattern. Burberry manufactures a wide range of
products, such as clothing, shoes and accessories, and as a brand has become em-
blematic of the working-class chav culture. The fact that this often takes place in
the form of counterfeit Burberry products is of no major importance as such: it may
in fact be essential for the products themselves to be ‘fake’ in the production of an
authentic ‘chav’. Whether ‘real’ Burberry or not, the brand itself is indeed emblem-
atic of ‘chav’ to the extent that the Burberry check is also enough in itself to turn
other cultural products into ‘chav’ – as in the case of Chav Guevara (Figure 3.).
Turning Che Guevara into Chav Guevara by presenting him in the Burberry
check pattern points to a significant, more general aspect in identity work. Admin-
istering the right amount of specific semiotic features is at the core of authenticity:
being an authentic someone requires orientations towards certain resources that
emblematically index a particular desired identity, and, as with chav identity, the
dose of resources can be minimal, almost homeopathic. The dose can be small, but
the only thing that is required is that it is enough – enough to produce a recogniz-
able identity as an authentic someone. And as the illustrations here make clear:
this single emblematic feature can be applied in an almost infinite range of cases,
redefining every object into a ‘chav’ object. On the Internet, we find underwear,
cars and houses coated in the Burberry tartan, along with almost every imaginable
cartoon figure and superhero. The ‘active substance’ of chavness can be blended
with almost any other substance to produce the same ‘real’, ‘true’, ‘authentic’ and,
above all, instantly recognizable image: the ‘chav’.
This vignette, thus, demonstrates the basic mechanism of identity construc-
tion: one isolates a set of emblematic features that are seen and understood as
bestowing an identity of X or Y, and while ‘full’ identities would require the enact-
ment of all the features, the display of one feature can trigger the full emblematic-
ity of certain identities. Here, the Burberry tartan becomes the mobile and passe
partout emblem of Chavism.
The second vignette illustrating the processes described in this paper engages with
a globalized social and cultural icon, to be found at present in almost any large and
middle-size city of the Western world and many parts of the non-western world as
well. Wherever it occurs, the Irish pub is instantly recognizable. And as we have
seen in the previous example, this recognizability is triggered by the use of a small
set of ‘active substances’ that, when present in the right dose, lend a pub its instant
identification as ‘Irish’. The active substances are objects and artefacts people ori-
ent towards in an attempt to construct authenticity.
The Irish pub is undoubtedly an instance of the ‘invention of tradition’
(Hobsbawm & Ranger 1983), something which has been developed quite recently
as a particular iconic place breathing a kind of fundamental ‘Irishness’ inscribed in
its layout, spatial organization, furniture and products on offer. As for the latter,
there is little doubt that the Guinness beer brand has been instrumental in devel-
oping and promoting this worldwide ‘standard’, so to speak. The Irish pub is an
artefact of globalized commodification.
As a globalized commodity, it has become extraordinarily successful. In
Belgium alone, 86 ‘Irish pubs’ are listed on www.cafe.be, a directory of cafés in
Belgium. Most, if not all of them are of course run by Belgian publicans; customers
would be served in the language of the place and some of the staff working in such
pubs have never visited Ireland. Such Irish pubs do in fact present a blend of local
and global features; the presence of the globalized features turns them into in-
stantly recognizable Irish pubs; the local features ensure that the overwhelmingly
local customers do not feel out of place in such pubs.
Let us now turn to the globalized features, the ‘active substances’ as we called them.
Running through about one hundred Irish pub websites (and having visited a good
number of such pubs ourselves), we see that a small handful of emblematic features
appear in almost every case; we can list them. But before we do that, let us have a
look at one illustration, in which we see several of the emblematic items. In
Figure 4, we see a coaster from an Irish pub in the small Belgian town of Zottegem:
1. Pubs have a recognizable Irish name. This name can be a family name. From the
list of Belgian Irish pubs, we note: Blarney, Conway, Fabian O’Farrell, Finnegan,
Kate Whelan, Kelly, Kitty O’Shea, Mac Sweeney, Mac Murphy, Maguire,
McCormack, Molly Malone, Murphy, O’Fianch, O’Malley, O’Reilly, O’Dwyer,
Paddy, Patrick Foley, Scruffy O’Neill, Sean O’Casey. Apart from a name, an
identifiably Irish word can be used, such as ‘An Sibhin’ or ‘The Ceilidh’. Alterna-
tively, the pub’s name refers to Irish symbols such as the Shamrock, or to well-
known characters from Irish literature such as Molly Bloom. And words such
as ‘Irish’ and ‘Celtic’ can also be used to flag the Irishness of the pub.
2. There would be a preference for a particular kind of Celtic lettering in shop
signs and advertisement boards; this can be done ‘completely’ or by approxi-
mation. In Figure 4 we see a rather amateurish attempt in ‘Zottegem’, where
especially the ‘e’, the ‘g’ and the ‘m’ have a Celtic twist.
3. Some stock symbols of Ireland would be present. Among others, the website
http://www.of-ireland.info/symbol.html lists the following canon of five
‘symbols of Ireland’: the flag, the shamrock, the harp, the Celtic cross and the
ring of Claddagh. The three-leaf clover, shamrock, would be present in almost
every case – see Figure 4. The Irish harp would also be quite frequently used,
certainly when Guinness beer is advertised; less used are the Irish flag and the
Celtic cross. We have not found instances of the use of the ring of Claddagh.
Also quite widespread as a symbol of elementary Irishness is the color green –
see again Figure 4 above, where green dominates the coaster as well as the
clothes of the figures depicted in it.
4. Irish pubs would almost always advertise music as part of their character and
attractiveness. Evening shows with live bands, often performing folk music, are
quite a widespread feature of Irish pubs, and one Belgian Irish pub is called af-
ter the legendary folk band ‘The Dubliners’. Other Irish stars such as U2 would
be mentioned, and theme nights would be organized around their music.
5. Finally, some products are omnipresent. Guinness beer is undoubtedly the in-
dispensible commodity on offer in any Irish pub. Jameson whiskey is another
very frequent item on offer, and both would often be visibly advertised from
the outside of the pub. Other ‘typical’ products would be Kilkenny’s beer and
Irish cider; when food is offered, Irish lamb stew and Irish steak would very
often be found on the menu.
Enough is enough
These five elements dominate the Irish pubs in Belgium; no doubt, they will be
found elsewhere around the world as well.4 They combine in a rustic, dark wood
and brass interior to form a kind of cosiness welcomed by customers. The Zottegem
Paddy’s Pub summarizes its character as follows:
The Paddy’s pub is an Irish pub where everyone feels at home and makes oneselfs
comfortable (sic). It’s got everything you can find in an Irish pub, nice music, Irish
whiskeys, Guiness, Kilkenny, Adam’s Apple, Irish food combined with European
dishes, in a word, a part of Ireland in Zottegem Belgium.
Note the Belgian accent in Paddy’s English, and observe the statement “Irish
food combined with European dishes” – which summarizes what we intended to
demonstrate. Irish pubs blend a small dose of emblematic globalized Irishness
with a whole lot of local and other features. Guinness and cider are flanked by
solidly Belgian beers such as Jupiler and Leffe on tap. So too with food: apart
from the Irish lamb stew and the Irish Angus beef, Irish pubs in Belgium offer
the same snacks and meals as those offered by non-Irish pubs in many places
around the world. O’Reilly’s in Brussels, for instance, offers some iconic Irish
food along with buffalo wings, beef and veggie nachos, hamburgers (with Irish
beef), as well as the very English fish and chips and Sunday roast (http://oreillys.
nl/brussels/menu/7-food-menu.html). And in many pubs, a choice of Irish whis-
kies would be complemented by a rich variety of original Scotch malts. Irishness
can be extended, as we can see, into a broader realm of Anglo-Saxon-ness.
Needless to say, nevertheless, that almost every Irish pub advertises itself as au-
thentically Irish.
Is there a critical limit to the amount of emblematicity that a place ought to
display in order to be a recognizably ‘Irish’ pub in Belgium? When is a pub ‘Irish
enough’ to pass credibly as such? From our observations, we see that at least some
of the features listed above are mandatory. Contrary to what we saw in the case of
the Chav, one feature is not enough: a pub called ‘Sean O’Reilly’s’ but not serving
Guinness or other Irish products would not easily be recognized as ‘Irish’ (“What!
You don’t have Guinness?!”); in the opposite case, it is not enough to serve
Guinness to qualify as an Irish pub. Irish pubs need to look and feel Irish, and they
achieve that by means of a bundle of emblematic features: a name, a choice of
products, displays of the shamrock or the harp, the color green, and so forth. The
bundle, however, should not be too big. A pub which is so Irish that customers are
required to speak English – or Irish – in order to get their orders passed, for in-
stance, would not be too long in business in a town such as Zottegem. The same
4. A possible sixth feature could be this. Almost all pubs would organize a St Patrick’s Day
event. Customers are requested to dress in green colors that day, and substantial drinking at
discount prices would be the backbone of the event; live bands would perform as well.
Jan Blommaert and Piia Varis
would apply to pubs that would only welcome Irish customers.5 Irish pubs are
globalized in a familiar way: a small but highly relevant bundle of globalized em-
blematic features is blended with a high dose of firmly local features. Customers
can feel at home in Zottegem while they are, simultaneously, savoring an ‘authenti-
cally Irish’ pub ambiance. By entering an Irish pub, the local customers do not
become Irish; nor would they have to, and that is the whole point: one merely en-
ters a niche of Irishness.
This vignette, thus, illustrates another defining feature of identity work: the
fact that it revolves around the production of blends, in which emblematic features
can be mixed with non-emblematic ones. Provided there is ‘enough’ emblematic-
ity, the mix will work as an instance of the desired identity. The mix – in itself evi-
dently ‘inauthentic’ – can even emanate authenticity, an exalted purity in identity
and an enacted extension of an identity tradition. The pubs we looked at are all
fully ‘authentic’ as Irish pubs in their own eyes (and those of many of their
customers, we can submit), in spite of being, at the same time, authentically
Belgian pubs as well. It is this blending that, perhaps, explains the remarkable dy-
namics of authenticity in superdiverse (and thus essentially polycentric) contexts:
being authentic does not revolve around being just one thing; it consists of the
capacity to create rapid realignments with a variety of different identity positions
and resources.
Enough is enough
5. A very small number of exceptions exist, mostly in larger cities such as Antwerp and
Brussels, targeting English-speaking expatriates.
Enough is enough
cultural lives, we arrange features in such a way that they enable others to identify
us as ‘authentic’, ‘real’ members of social groups, even if this authenticity comes
with a lower rank as ‘apprentice’ within a particular field. We enter and leave these
niches often in rapid sequence, changing footing and style each time and deploy-
ing the resources we have collected for performing each of these identities – our
identity repertoire is the key to what we can be – or can perform – in social life. In
so doing, we produce complex blends and hybridities; but the ‘impurity’ presup-
posed in such forms of mixing does not appear to preclude judgments of full
authenticity.
Enoughness judgments determine the ways in which one can rise from the
apprentice rank to higher, more authoritative ranks – apprentices orient towards
the ‘full’ authenticity while they start building their own restricted versions of it.
Fans of Irish pubs, for instance, would begin to exhibit and develop their fanship
by collecting ‘Irish’ objects: green top hats, shamrock coasters, Guinness beer
glasses, Irish national team soccer or rugby jerseys and so on. They gather objects
that culturally bespeak authentic ‘Irishness’ – such Irishness that can align them
with the propotypical templates they orient towards, the Irish pub and beyond it,
an imaginary essential Irishness.
Throughout all of this we see that ‘culture’ appears as that which provides
(enough) meaning, i.e. makes practices and statements sufficiently recognizable for
others as productions of identities. And throughout all of this, we see such cultures
as things that are perpetually subject to learning practices. One is never a ‘full’
member of any cultural system, because the configurations of features are perpetu-
ally changing, and one’s fluency of yesterday need not guarantee fluency tomor-
row. In the same move, we of course see how such processes involve a core of
perpetually shifting normativities (the things that enable recognizability and, thus,
meaning), and because of that, power – power operating at a variety of scale levels
in a polycentric sociocultural environment in which all of us, all the time, are re-
quired to satisfy the rules of recognizability.
All of this can be empirically investigated; it enables us to use an anti-essen-
tialist framework that, however, does not lapse into a rhetoric of fragmentation
and contradiction, but attempts to provide a realistic account of identity practic-
es. Such practices, one will observe in a variety of domains, revolve around a
complex and unpredictable notion of authenticity, which in turn rests on
judgments of enoughness. The concise framework sketched here can serve as a
heuristic for engaging with this enormous and rapidly changing domain of
authenticity.
Jan Blommaert and Piia Varis
References
Vigouroux, C. 2011. Magic marketing: Performing grassroots literacy. Diversities 13(2): 54–69.
Wang, X. 2010. “I’m not a qualified dialect rapper”: Genre innovation as authenticity [Working
Papers in Urban Language and Literacies 64]. <http://www.kcl.ac.uk/innovation/groups/
ldc/publications/workingpapers/download.aspx>
Williams, Q. 2012. The enregisterment of English in rap “braggadocio”: A study from English-
Afrikaans bilingualism in Cape Town. Tilburg Papers in Culture Studies 20. <http://www.
tilburguniversity.edu/research/institutes-and-research-groups/babylon/tpcs/>
Žižek, S. 1994. Mapping Ideology. London: Verso.
The primary classroom as a superdiverse
hetero-normative space
Massimiliano Spotti
Tilburg University
1. Introduction
It is against this background that this paper deals with language repertoires
and with the micro-dynamics of identity ascription and subscription taking
place in a regular multicultural primary school classroom in the Netherlands.
More specifically, the first set of data to be discussed features the evaluative dis-
course of the teacher and it focuses on the ways in which this classroom teacher
indexes pupils’ identities on a continuum ranging from foreigner to near-native
speaker of Dutch. The second set of data to be discussed, instead, features the
evaluative discourse of Moroccan girls of both Berber and Arabic origin where
their identities are strongly anchored on a continuum of purity versus impurity,
established on the basis of their language skills in Berber. The third set of data
features an act of sabotage engaged in by two pupils with a Turkish background,
who cross into ‘fake’ Arabic to escape the pressure exerted on them by the class-
room researcher.
The paper concludes by proposing an understanding of multilingualism that
can do justice to the complexity of the pupils’ own sociolinguistic repertoires
and identity performances that are not necessarily bound to ethnic group
membership.
One’s identity, broadly speaking, is who one is to oneself and to others. In other
words, or rather, in the words of John Gumperz (1968: 383), identities are a “field
of semiotic action where the distribution of linguistic variants is the reflection of
social facts”. Thus, at a symposium we may be asked our name and surname, pos-
sibly followed by further enquiries about our place of origin. We are subsequently
expected to give our institutional affiliation(s) and naming our field of expertise.
The latter may vary according to the audience in front of us. In fact, the author of
this article might be ascribed the identity of a ‘sociolinguist’ by an audience con-
sisting primarily of anthropologists, while he might be considered more of an
anthropologist’ by an audience of sociolinguists. Attributions like the ones de-
scribed above contribute to identity construction, both of the ‘inhabited’ (or as
others would say, ‘subscribed’) kind as well as ascribed by others. Further, these
attributions show that identity is the product of a process of meaning making in
a socio-cultural space. In that space, in fact, we at once provide and give away –
through the deployment of semiotic resources, including our language use – in-
formation about ourselves, information which in turn is accepted, refused, nego-
tiated and/or re-defined in interaction with others. Our identity is thus the
product of a semiotic performance, and it is through the evaluation of this perfor-
mance by others – tied in turn to the norms grounded in a certain socio-cultural
The primary classroom as a superdiverse hetero-normative space
space – that determines whether or not we have succeeded in acting out an ap-
propriate identity and have positioned ourselves adequately, within the socio-
cultural space at hand. Language elements such as language varieties, accents,
styles, registers are among the semiotic resources at our disposal that allow us to
act out identities, and all of these resources are deeply entrenched and hierarchi-
cally organised in a value system. When played out in the right space and at the
right time, language performances enable their users to present the ‘right’
(or desired) identity, which subsequently grants them the possibility of being
heard and understood in a space as someone holding a certain place (see also
Spotti 2007).
It is now time to move on to ethnolinguistic identity, which is central to the
argument that runs through this paper. If we take Silverstein’s referential theory of
language (1996) and apply it to ethnicity, we see that both language and ethnicity
– constituent elements of ethnolinguistic identity – are often presented in the me-
ta-pragmatic discourse of lay people as a large set of transparent, structured, non-
contextual, denotational entities that bind people together in an ethnic group
where language is often addressed as the main factor of ethnic belonging. A case
in point is presented in the utterance ‘I am Italian and thus I am a native speaker
of Italian’. The association of ethnic group membership with language and native
speakerhood is quite straightforward. But there is more to it than that. In response
to an ill-articulated utterance, we might hear a remark like ‘Speak English, boy’ or
‘You cannot say that in English, it is grammatically incorrect’, where both respon-
sive utterances, although varying in their degree of courtesy, indicate that a
language called ‘English’ is depicted as being composed of a finite set of (gram-
matical) structures that one can strive to master and apply. Failing to do so may
result in the speaker being ascribed the identity of a ‘sloppy’ language user, ‘gram-
matically ignorant’, ‘poorly educated’ or ‘non-native’. In other words, no matter
how basically neutral the elements of a language (essentially) are, the use of these
elements never is. Rather, language use is indexically loaded being the link be-
tween the linguistic sign and the social meaning that the sign holds according to
the institution in which a certain sign is used (Silverstein 1998: 404).
From the above, we can gather an understanding of language, ethnicity and
ethnolinguistic identity that is socio-culturally, historically and politically in-
vested, emphasizing that the very existence of a language, of an ethnic affiliation
and of an ethnolinguistic identity is the result of the work of indexicalities with-
in a certain space that either grant or refuse someone’s identity a certain place.
The situations in which these categories are forged, the meanings that they take
on board and the implications of their indexical loads for individuals may
vary quite dramatically ending up producing inequality of access and voice
(see Hymes 1996).
Massimiliano Spotti
On February 15th, 2005, Group 8a, one of the two eighth-grades’ classrooms at
St. Joseph’s Catholic Primary comprised eighteen pupils in total. These were eight
boys and ten girls. The age of the pupils ranged from eleven to thirteen due to the
fact that some pupils were repeating the year. None of the pupils had been added to
the group in the course of the current school year; thirteen of them had been at
St. Joseph’s Catholic Primary since their first grade. According to the school register,
all pupils in Group 8a had been assigned an educational weight of 1.90 – i.e., almost
twice the weight of ‘regular’ 1.0 pupil, which tend to be indigenous Dutch pupils –
the only exception being Walid, who was assigned an educational weight of 1.0,
probably on account of his parents, both of them born in Morocco, being highly
educated. In Group 8a’s home language survey, we found that all pupils came from
immigrant minority groups and that at home they all spoke a language other than
or alongside Dutch. As regards their country of birth, thirteen of the eighteen pupils
were born in the Netherlands. Of the remaining five, three were born in the Dutch
Antilles, one in Bosnia-Herzegovina and one in Morocco. As regards their parents’
country of birth, Affifa was the only pupil born to a second-generation Moroccan
father and a first-generation Moroccan mother; all other pupils had parents who
were both born outside the Netherlands. Table 1 shows the home languages, gender
and names of the pupils as gathered from the home language survey.
In what follows, I present and analyse the meta-pragmatic discourse of Miss
Sanne, the regular classroom teacher of Group 8a, when dealing with the sociolin-
guistic profile as well as the ascribed sociolinguistic behaviour of three of her pu-
pils: Mohammed, Walid and Micheline.
Table 1. Gender, names and home languages of the pupils according to Form 8a home
language survey
Max: (hmm)
Sanne: from the Somali language.
vanuit de Somalische taal. (S02: 323–329)
Mohammed not only uses ‘strange sentences’ in Dutch because his speech is based
on the structure of Somali – a language that has SOV- order in its main clause
while Dutch has SVO (cf. Saeed 1999). Also, as introduced by the causative con-
junction ‘so’, both Mohammed’s parents are responsible for the syntactic interfer-
ence from Somali to Dutch. The father, in fact, spoke no Dutch and the mother
also ‘suffered’ from Somali sentence structure in her use of Dutch, both language
situations that feed the discourse of immigrant minority pupils’ language disad-
vantage because of their parents’ lack of Dutch proficiency.
Walid, who had already been singled out by Miss Sanne as the only pupil assigned
an educational weight of 1.0, is also worth focusing on. In the second long inter-
view with Miss Sanne, she expands on Walid’s case:
Sanne: Right, so there is one pupil with a one point zero weight. But
when they [his parents MS] came to an advisory meeting in prep-
aration to secondary education that mother asked like how is that
possible. I say well that is only possible if you at the enrolled
Walid: ‘I want to register my child as a one point zero’. But it is (...)
Walid is also simply the child of two Moroccan parents.
[Nou, er is dus eentje met een punt nul alleen. Toen ze op advies-
gesprek kwamen voor het voortgezet onderwijs vroeg die moeder
dus van hoe kan dat nou. Ik zeg, nou, dat kan alleen maar als u bij
het inschrijven van Walid aangeeft van ik wil mijn kind als een
punt nul aangeven. Maar het is (...) Walid is ook gewoon een kind
van twee Marokkanse ouders.]
Max: (hmm)
Sanne: and also he himself is therefore completely just one hundred per-
cent Moroccan so in principle he is also simply one point nine.
[en hij is zelf ook gewoon dus helemaal honderd procent Marok-
kaan dus in principe is hij ook gewoon een punt negen.]
(S02: 287–289)
Miss Sanne constructs Walid’s identity as an exception to the mechanistic educa-
tional weight system and singles him out – ‘so there is one pupil with one point
zero’ (S02: 287). In the adversative clause that follows – ‘but it is (...) Walid is also
The primary classroom as a superdiverse hetero-normative space
simply the child of two Moroccan parents’ (S02: 288) –, Walid is ascribed, on the
basis of ius sanguinis, to the Moroccan immigrant minority group. This ascription
is confirmed in the co-ordinate clause ‘and as for himself, he’s simply one hundred
per cent Moroccan’ (S02: 289), which (as the causative connective ‘so’ makes clear)
automatically qualifies him as a 1.90 pupil. Further on, through the use of an
adversative clause, Miss Sanne states:
Sanne: But his parents are indeed very highly educated so (...) I do not
know whether it has been a little mistake or that at the time they
have asked for a one point zero but she was rather puzzled in my
view so I think that it is a little mistake because all the children
here in the class are offspring of foreign parents.
[Maar zijn ouders zijn wel heel erg hoog opgeleid dus (...)Ik weet
niet of het een foutje is geweest of dat ze destijds hebben gevraagd
van een punt nul maar ze was vrij verbaasd in mijn ogen dus ik
denk dat het een foutje is want alle kinderen hier in de klas zijn
afkomstig van buitenlandse ouders.]
Max: (hmm) okay so it could very much be that he is a one comma
nine then.
[(hmm) oké dus het zou best wel kunnen zijn dat dan hij een een
komma negen is.]
Sanne: Yes, yes sure, technically seen he is a one comma, one point nine.
[Ja, jawel, technisch gezien is hij een een komma, een punt negen.]
Max: What do you mean? Tech (...) technically seen (/)
[Hoe bedoel je? Tech (...) technisch gezien (/)]
Sanne: Yes he (...) his parents are simply both foreigners.
[Ja hij (...) zijn ouders zijn gewoon allebei buitenlands.]
Max: Yes okay but prac (...) in practice?
[Ja oké maar prakt (...) praktisch gezien?]
Sanne: (uh) he is a really clever boy and his parents are also both (uh)
highly educated and you really experience that.
[(uh) het is een hele slimme jongen en zijn ouders zijn ook allebei
(uh) goed opgeleid en dat merk je toch wel]
Max: (hmm)
Sanne: they speak also correct Dutch at home so he also gets it taught
well so yes.
[die spreken thuis ook correct Nederlands dus hij krijgt het ook goed
aangeleerd dus ja.]
Max: (hmm) and what do you mean by correct Dutch?
[(hmm) en wat is volgens jou correct Nederlands?]
Massimiliano Spotti
Miss Sanne also addresses the case of Micheline, an eleven-year-old Antillean girl
born on the island of Curaçao to Antillean parents. Micheline is the only pupil in
Miss Sanne’s class who did slightly better on the national test than the score the school
had predicted for her. She is also the only grade 8a pupil who was recommended to
go on to HAVO-level secondary education (Senior General Secondary Education)
(all the others being recommended to go on to lower vocational training). In dealing
with Micheline’s case, Miss Sanne tries to unravel the reasons why she does not suffer
from the language problems other grade 8a pupils are faced with:
Sanne: She [Micheline MS] is she is also Antillean so she also has, right,
on Curaçao they also speak Dutch.
[Ze is ze is ook Antilliaans dus ze heeft ook, hè, in Curaçao spreken
ze ook Nederlands]
Max: (hmm)
Sanne: so her parents can also both speak Dutch.
[dus haar ouders kunnen ook allebei Nederlands.]
Max: (hmm) (S02: 425–428)
The primary classroom as a superdiverse hetero-normative space
At first, Miss Sanne groups Micheline together with Joshua and Rhonda on the
basis of her origin as “she is also Antillean (...)” (S02: 425). She subsequently con-
nects her being Antillean with the fact that “on Curaçao they also speak Dutch”
(S02: 425), which means that “so her parents can also both speak Dutch” (S02:
427). If we were to follow what Miss Sanne has so far proposed, i.e., that having
parents who speak Dutch implies the Dutch (also) being spoken at home, and
Dutch at home in turn would explain Micheline’s good Dutch language skills. In
the adversative clause that follows, Miss Sanne expands on Antillean children and
their Dutch language skills:
Sanne: But I also notice that simply very often (...) with Antillean children
that that language [Dutch MS] they simply posses it a bit better al-
ready because that, because Dutch is not really a second language
[Maar dat merk ik gewoon ook heel veel (...) bij Antilliaanse
kinderen dat die taal er ook al gewoon wat beter in zit omdat dat
omdat Nederlands niet echt een tweede taal is]
Max: (hmm)
Sanne: Dutch is also their mother tongue with Papiamentu.
[Nederlands is ook hun moedertaal me[:]t het Papiaments.]
(S02: 431–433)
By stating that with Antillean children Dutch is “a bit better already [...]” (S02: 431),
Miss Sanne is implying a comparison. If the comparison concerns Mohammed, the
pupil previously used as an example of the current limitations of Group 8a pupils in
Dutch, then it would follow that Antillean pupils are better at Dutch than other
pupils with an immigrant minority background. This is so because, as Miss Sanne
states, for Antilleans “[...] Dutch is not really a second language” (S02: 431), mean-
ing that Dutch for them is closer to a first language since “Dutch is also their moth-
er tongue together with Papiamentu” (S02: 433). The sociolinguistic position of
Dutch on Curaçao sketched by Miss Sanne is not that of a foreign language, as it is
for most of the other immigrant minority pupils who have parents who do not
speak Dutch at home. Rather, on Curaçao, education has to cater for learning Dutch
and learning in Dutch (cf. Kook & Vedder 1989). In addition, Dutch is used as a
vreemde voertaal in society, i.e., a foreign language that is used in education and in
other official situations but that in practice has little sociolinguistic relevance in
people’s everyday lives (cf. Narain 1998: 7). The generalisation drawn by Miss Sanne
has two consequences. First, given that it is common to all Antillean pupils to be ‘a
bit better at Dutch’, it confirms that Micheline is good at Dutch because Dutch is
almost a first language for her. Second, it confirms the insight that sees parental
Dutch language skills as supportive of pupils’ good Dutch. Micheline, however, is
not the only Antillean pupil in the class. It is at this point that erosion comes into
Massimiliano Spotti
play. Miss Sanne’s meta-pragmatic judgements about the Dutch of Antillean pupils
and Micheline’s good results were based on Micheline’s parents’ use of Dutch. The
other two Antillean pupils in grade 8a, Joshua and Rhonda, were also born in the
Netherlands and, like Micheline, they both speak Dutch at home with their parents.
However, they both attend Leerweg Ondersteunend Onderwijs (Education for chil-
dren in need of Learning Support) and are both ascribed by Miss Sanne as “weak”
pupils (S03: 127). The mismatch among the profiles of the Antillean students in her
class, prompts her to add a second explanation for Micheline’s good results: “And
yes where else does it come from, I think that it is also simply part of her nature”
(S02: 429). This second explanation interprets the difference as not being related to
Micheline’s ethnolinguistic background and parental language skills. Rather, it boils
down to Micheline’s own intrinsic nature.
The following discussion with Lejla, Samira, Lemnja, Hajar and Siham, stems from
these pupils’ initial responses in the essays about themselves and their language
repertoires. It starts with Lejla explaining her affiliation to Bosnia and continues as
follows:
Lejla: Them live in the Netherlands also and are a hundred per cent
Moroccan too.
Hun wonen toch ook in Nederland en zijn ook honderd procent
Marokkaans.
Max: Is that so girls?
Is dat zo meisjes?
[The other girls agree loudly]
Max: Yeah right, behave properly.
Ja hallo, doe normaal.
Lemnja: What do you mean a hundred, what do you mean a hundred, a
thousand right?
Wat honderd wat honderd, duizend of niet?
Max: A thousand percent Moroccan, what do you mean?
Duizend procent Marokkaans, hoe bedoel je?
Hajar: Simply.
Gewoon.
Lemnja: A million.
Miljoen.
Hajar: My whole life.
M’n heel leven.
The primary classroom as a superdiverse hetero-normative space
Max: So he is an Arab?
Hij is een Arabier dus?
Lemnja: Yes and the rest are all Berber.
Ja de rest allemaal Berber.
Hajar: Yeah and I’m not. [With an upset tone]
Ja en ikke nie.
Max: Yes, yes. [Pointing at Siham]
Ja, ja. (GD01: 1066–1074)
In the above, we see macro language politics at play within an immigrant minority
community group becoming sedimented in micro-processes of misrecognition,
sanctioning and contestation. Samira, who was ascribed by Siham as ‘half Arabic’,
objects to being constructed at the periphery of the Berber group and reiterates
her ‘Berber Arabic’ affiliation, where the purpose of using Berber first and Arabic
second may be to underline the contrast, to set it over against Siham’s initial as-
cription as ‘half Arabic’. Hajar’s remark, on the other hand, stresses her being upset
at being ascribed as non- Berber.
6. Beyond big and small languages: ‘A real Turk’ says ‘shouf, shouf habibi’
Along the same lines, Osman’s self-reported use of Turkish is motivated by the
pragmatic goal of communicating ‘a secret’ and using it at playtime for telling
‘something important’. However, Turkish assumes different connotations among
pupils of Turkish background as is shown by the following transcript.
Max: And do you do you do that often, using Turkish, at school?
En doe je doe je dat vaak, Turks, op school?
Osman: Yes
Ja.
Max: Oh.
Osman: In the breaks.
In de pauze.
Max: You do?
Ja?
[Turning to Cemal]
Max: So do you also speak Turkish with everyone because it makes
you look really cool?
Dus spreek je ook met iedereen het Turks want het maakt je heel
stoer?
Massimiliano Spotti
Through the interviews with Miss Sanne and the group discussions held with her
pupils. Both of which were supported by prolonged periods of ethnographic
observation, I believe to have given the reader some insight into the hetero-
normativity of this primary school classroom, and into the workings of these pu-
pils’ sociolinguistic repertoires and of their consequent identities reifications.
Someone’s sociolinguistic repertoire comes from a space. This means that his/her
repertoire – whether in terms of language varieties or in terms of the use of those
fine-grained linguistic elements that make up for a genre and a style – are all prod-
ucts of someone’s biographical project (cf. Johnstone et al. 2006 but also Lippi-
Green 1997 and more recently Blommaert & Backus 2011). Though, we should not
overlook the fact that as well as coming from a socio-cultural space, a sociolinguis-
tic repertoire is always acted out within a socio-cultural space. It is within this
space, e.g. a classroom like Group 8a, that someone’s repertoire is confronted with
hetero-normative language ideologies and hierarchies of language use. It is there-
fore within this space that the language someone uses and the identity someone
acts out may qualify either as ‘big’/valuable/central or as ‘small’/less valuable/mar-
ginal to the space at hand (see Blommaert et al. 2006 on being declared illiterate
and on the (mis)recognition of pupils repertoires as a ‘non-languages’).
As it emerges from the classroom teacher’s discourse when reflecting upon the
language repertoires of her pupils, we realise that these pupils are to a greater or
lesser extent competent users of the Dutch language, but if we want to fully under-
stand classroom multilingualism. These pupils’ use (or non-use) of community lan-
guages serves us here as a case in point. First, a community language like
Berber becomes a ‘big’ language in that it is an index of authenticity in terms of a
pupil ‘pure’ membership (or lack thereof) to an ethnic group. This authenticity, as-
sociated with Berber in this case, it causes the authentic members of this commu-
nity to address those who either use Arabic or are not fully proficient in Berber, as
only ‘half ’ pure. Second, a community language like Turkish, becomes in these pu-
pils’ reported judgements a ‘small’ language. This is so, in that Turkish does not hold
Massimiliano Spotti
the same position that international languages like English or French do. This leads
to Turkish being perceived as not worth the effort because ‘it is not cool’. Further,
though, the Turkish language is also presented as an index of authentic (male) mem-
bership of the Turkish community. This symbolic belonging to a language, which
constructs an ‘authentic male Turk’ on the basis of Turkish language use, is chal-
lenged by Cemal. By using an Arabic sentence popular in the Netherlands and there-
by claiming some knowledge of Arabic – a stratagem that result in repeating tokens
of the same highly popular Arabic sentence – Cemal sabotages the interactional or-
der with the authority who wants to know about their language use. This ‘sabotage’
(Goffman 1974: 426), though, is not a matter of outspoken contrast but a language
device that, once employed, does not have a long lasting effect. It serves the purpose
of eluding a situation in which authority is seeking to elicit a specific routine in lan-
guage behaviour. These pupils therefore are not just pupils caught between ‘big’ and
‘small’ languages. Rather, they are multilingual experts whose expertise is to be found
righteously so in the sharp awareness that they have of the linguistic market that
unfolds in front of them within the school classroom. It is through this understand-
ing of the linguistic market that they manage and negotiate their daily identity prac-
tices and it is through this management that they either re-enforce or cross over
ethnic community boundaries of belonging.
References
Bezemer, J. 2003. Dealing with Multilingualism in Education – A Case Study of a Dutch Primary
School Classroom. Amsterdam: Aksant.
Blommaert, J. & Backus, A. 2011. Repertoires revisited: “Knowing language” in superdiversity
[Working Papers in Urban Language & Literacies 67]. London: Kings College. <http://
www.kcl.ac.uk/innovation/groups/ldc/publications/workingpapers/67.pdf>
Blommaert, J., Creve, L., & Willaert, E. 2006. On being declared illitterate: Language-ideological
disqualification in Dutch classes for immigrants in Belgium. Language & Communication
26(1): 34–54.
Goffman, E. 1974. Frame Analysis. An Essay on the Organization of Experience. New York NY:
Harper & Row.
Gogolin, I. & Kroon, S. (eds). 2000. “Mann schreibt, wie mann spricht”. Ergbenisse einer interna-
tional-vergleichenden Fallstudie über Unterricht in vielsrachingen Klassen. Münster:
Waxmann.
Gumperz, J.J. 1968. The speech community. In International Encyclopedia of the Social Sciences,
D. Sillis (ed.), 381–386. New York NY: Macmillan.
Gumperz, J.J. 1982. Discourse strategies. Cambridge: CUP.
Hymes, D. 1996. Ethnography, Linguistics, Narrative Inequality: Towards an Understanding of
Voice. London: Taylor & Francis.
Jaspers, J. 2005. Tegenwerken, belachelijk doen. Talige sabotage van Marokkaanse jongens op een
Antwerpse middelbare school. Brussels: VUB Press.
The primary classroom as a superdiverse hetero-normative space
Johnstone, B., Andrus, J. & Danielson, A.E. 2006. Mobility, indexicality and the enregisterment
of “Pittsburghese”. Journal of English Linguistics 34(2): 77–104.
Kook, H. & Vedder, P. 1989. Antiano i Arubano den skol. De onderwijssituatie van Antilliaanse en
Arubaanse kinderen en hun klasgenoten. Utrecht: POA.
Lippi-Green, R. 1997. English with an Accent: Language, Ideology and Discrimination in the
United States. London: Routledge.
Narain, G. 1998. Moedertaal, tweede taal en vreemde taal. Het leren van talen in het onderwijs
in de Nederlandse Antillen. In Proceedings of the Colloquium “Nederlands in de wereld” –
“Nederlands in het Caribisch gebied”, G. Narain (ed.), 138–142. Curaçao: NTU.
Pujolar, J. 2001. Gender, Heteroglossia and Power. A Sociolinguistic Study of Youth Culture. Berlin:
Mouton de Gruyter.
Saeed, J.I. 1999. Somali Reference Grammar. Wheaton MD: Donwoody Press.
Silverstein, M. 1996. Monoglot “standard” in America: Standardization and metaphors of
linguistic hegemony. In The Matrix of Language, D. Brenneis & R. Macaulay (eds), 284–306.
Boulder CO: Westview Press.
Silverstein, M. 1998. Contemporary transformations of local linguistic communities. Annual
Review of Anthropology 27: 401–426.
Spotti, M. 2007. “What lies beneath?” Immigrant minority pupils’ construction in a multicul-
tural Flemish primary classroom. Journal of Language, Identity, and Education 6(1): 31–51.
Sturm, J. 2000. “Swimming in the garden”. Key incident analysis as an ethnographic tool to con-
struct knowledge on (multilingual) education. In“Mann schreibt, wie mann spricht”.
Ergbenisse einer international-vergleichenden Fallstudie über Unterricht in vielsrachingen
Klassen, I. Gogolin & S. Kroon (eds), 139–171. Münster: Waxmann.
Van der Craats, I. 2000. Conservation in the Acquisition of Possessive Constructions: A Study of
Second Language Acquisition by Turkish and Moroccan Learners of Dutch. PhD
dissertation, Tilburg University.
Assessing narrative development
in bilingual first language acquisition
What can we learn from monolingual norms?
Enkeleida Kapia
University of Hamburg
1. The study with the Albanian-speaking monolingual children was conducted with Anila
Kananaj, researcher at the Center for Albanian Studies in Tirana, whereas the study with
Albanian-Greek bilingual children was conducted in collaboration with Prof. dr. IanthiTsimpli
and one of her doctoral students, Maria Papakonstantinou at Aristotle University. Special thanks
go to both teams in Tirana and in Thessaloniki. The Greek data is not reported in this study since
the study focuses on the developmental trajectory of Albanian as L1 in monolingual children
and Albanian as L1 in bilingual children.
Enkeleida Kapia
Introduction
In light of the ever increasingly superdiverse world that we live in (as per the notion
of superdiversity as defined in Vertovec 2006, 2007 which will be summarised be-
low), assessing language development and identifying language delay in children
with more than one language has become an important task for researchers, educa-
tors, and clinicians around the globe. The task is particularly difficult because there
are almost no tests available that may be used with multilingual children (with the
exception of a few such as the Bilingual Syntax Measure and the Taal Toets Alle
Kinderen which were developed, respectively, for the Spanish-English kindergar-
ten children and non-Dutch school aged children attending the Dutch school sys-
tem). Yet, it is quite important to know and understand whether these children are
on track in terms of language development in order to better assist their education
and their development. Importantly, identifying early language delay helps im-
prove later language development and academic achievement (Marchmann &
Fernald 2008; Bornstein, Putnick & De Houwer 2006; Bornstein, Tamis-LeMonda
& Haynes 1999; Thal, Bates, Goodman & Jahl-Samilo 1997; Hart & Risley 1995).
Superdiversity has been defined as the social condition that presupposes an
interplay of complex and ever changing variables that have to do, among others,
with the speakers’ countries of origin, migration channels and legal status
(Vertovec, 2006). Within the framework of superdiversity, understanding language
development is an issue that requires not just the study of language per se, but also
demands the study of the effects of the interconnectedness of all the complex vari-
ables at play. Many have argued that using knowledge we have gained from mono-
lingual contexts as norms for superdiverse speakers is theoretically inappropriate
due to this multifaceted interaction, characteristic of superdiverse environments.
Despite the strength of these arguments and before casting aside useful knowledge
that we may have already gained, I believe that it is still important to investigate
whether language development of speakers in superdiverse environment shares
any of the milestones of the language development of monolingual speakers. In
other words, it is important to understand whether there are specific areas in which
these two look-alike developmental paths are similar or different and whether
these similarities or differences can be useful in educational and clinical contexts.
In light of this need, this paper offers a comparison of the study of language
development in monolingual children with that of bilingual children in superdi-
verse environments. While similar type of work has focused on three important
language domains: lexical development, morphosyntactic development, and nar-
rative development with a larger focus being on lexical development, this study
will focus on narrative development. The study of narrative development allows
for tapping into both micro and macro structure language features. The study’s
Assessing narrative development in bilingual first language acquisition
main aim will be to determine how the profile drawn from narratives of monolin-
gual children can be compared to the profile built from bilingual children by using
the same methodology across both populations. Ultimately, the comparison aims
to shed light on whether specific aspects of monolingual norms should be used to
effectively determine language development in children with more than one lan-
guage in superdiverse environments. The next two sections will focus on reviews
of cross language acquisition of narratives for both the monolingual and the
bilingual child.
Monolingual children
Bilingual children
Early work on narrative development in bilingual children has showed that bilin-
gual children used both language-specific forms and story-telling styles in each
language (Dart 1992). These results were echoed in later work, which maintained
Enkeleida Kapia
As clearly seen from the literature reviewed above, the field has learned a great deal
about the developmental path of narratives in the monolingual and bilingual con-
text. Yet, in light of the fact that the field lacks testing tools to determine language
development in bilingual children, the question of whether norms established for
monolingual narrative development can be used for bilingual narrative develop-
ment appears quite useful for at least two important reasons.
First, studies comparing monolingual norms with bilingual norms have sys-
tematically looked at vocabulary or lexical development as a means of assessment.
For instance, one of the more important earlier studies in the field concluded that
when looking at total vocabulary and/or conceptual vocabulary, there is no basis
for concluding that bilingual vocabulary development is slower than monolingual
vocabulary development (Pearson et al. 1993). However, while lexical develop-
ment is an important domain, other domains such as morphosyntax and the
narrative domain are equally as important in aiding us to understand the devel-
opmental trajectory of early language development in bilinguals. Specifically, nar-
ratives are important because they allow for the study of both micro and macro
structure elements of language in ways that the other two domains do not.
Second, studies looking at the comparison of norms from both monolinguals
and bilinguals have produced differing results. For instance, one quite recent study
suggested that monolingual norms can be effectively used to identify language
delay in bilingual children (De Houwer 2010). Specifically, this study recommend-
ed that single language measures should be sufficient in order to determine
whether a bilingual child has a developmental delay in terms of his or her lexical
development. Other studies, however, have shown that monolingual norms differ
significantly from bilingual norms and that double language measures are neces-
sary to avoid an inappropriate language evaluation (e.g., Pearson et al. 1993)
Assessing narrative development in bilingual first language acquisition
Participants
This study relies on narrative data collected from two different groups: “the mono-
lingual group” comprising of 6 Albanian-speaking monolingual children aged 6–7
years old residing in Tirana, Albania and “the bilingual group” comprising of
6 Albanian-Greek speaking bilingual children aged 6–7 years old residing in
Thessaloniki, Greece. The study itself was part of a large Cost Action project2, which
followed children and their language development in different countries with the
specific aim of determining patterns of SLI across bilingual children. All of the
monolingual children in this study were typically developing children and were not
exposed to any other language on a regular basis. In addition, all of the bilingual
children in this study were also typically developing and had been exposed to both
Greek and Albanian from birth. A priori variables of similar language background,
early onset of bilingualism, age, and similar levels of parental education provided
2. This study was conducted in the context of a larger study undertaken as part of COST
Action ISO804, “Language Impairment in a Multilingual Society, Linguistic Patterns and the
Road to Assessment” and in cooperation with Linguistic Diversity Management in Urban Areas
at University of Hamburg.
Enkeleida Kapia
at least somewhat comparable groups. The sample was purposeful due to the
required characteristics for each subject, such as age (6–7 years), onset of bilingual-
ism (from birth), level of parental education (at least high school diploma). Sub-
jects were recruited through local schools in the city of Tirana and Thessaloniki.
Materials
Children were presented with two picture sets comprising of 6 sequential pictures
each. Each set formed a story on its own with a beginning, a development, and an
end. Both picture sets were counterbalanced across all the participants. All the
pictures in the two sets were designed to be as systematically similar to one an-
other as possible. They all depicted three goals, three attempts to achieve these
goals, and three outcomes. They were also conceived as needing about the same
number of mental state terms to be narrated accordingly. Both picture sets were
piloted with 5 monolingual children (not participants in the experiment) before
they were used in the experiment.
Procedure
Participants were asked to view all the pictures of one set once and then tell the story
in their own words by looking at the pictures two by two. Children’s narrative pro-
ductions were recorded, and then later on transcribed and coded. The task included
an introductory session followed by a test session. Bilingual children that were un-
able to participate in the introductory session were excluded from the experiment.
Participants’ answers were coded for the following micro structure elements: num-
ber of words (presented as “w” in the tables below), number of clauses (presented
as “c” in the tables below), number of utterances (presented as “u” in the tables
below), number of content words (presented as “cw” in the tables below), number
of function words (presented as “fw” in the tables below), number of coordinating
conjunctions (presented as “cc” in the tables below), number of subordinating
conjunctions (presented as “sc” in the tables below), number of passives (presented
as “p” in the tables below), and number of relatives clauses (presented as “rc” in the
tables below). In addition, participants’ answers were also coded for the following
macro structure elements: goal, attempt, outcome, and mental state terms. Partici-
pants were given 1 point for each goal, attempt, outcome and mental state term
that they produced.
Assessing narrative development in bilingual first language acquisition
In the next sections several different analyses are presented. First, I will pres-
ent a global overview of the performance with regard to the micro structure ele-
ments in monolingual children. Then, the raw scores for the same category will be
presented for the Albanian of the bilingual children. This will be followed by an
Independent Samples T Test analysis of the difference between the two aforemen-
tioned performances. The final session will present the raw numbers of the macro
structure elements for both monolinguals and bilinguals followed by an analysis
that determines whether these raw numbers differ statistically from one another
(also an Independent Samples T Test analysis).
This section reviews findings on the different analysis performed for the micro and
macro elements of the narratives of both monolinguals and bilinguals. Table 1
indicates the raw results of the micro structure elements across Albanian-speaking
monolingual children.
Table 2 shows the raw results of the micro structure elements across bilingual
children (for their Albanian only).
Participant w C u cw fw cc sc p rc
1 81 18 17 33 48 2 4 2 0
2 107 22 23 54 53 9 5 1 1
3 131 28 28 63 68 8 9 1 0
4 141 29 31 66 75 5 12 1 3
5 92 21 21 52 40 4 5 2 0
6 76 16 17 33 43 7 2 4 0
Participant w C u cw fw cc sc p rc
1 40 7 4 18 22 2 1 0 0
2 22 6 5 9 13 1 1 0 1
3 52 13 9 23 29 9 0 0 0
4 77 15 13 46 33 3 4 0 4
5 50 10 9 27 23 14 0 0 0
6 72 19 12 33 39 15 0 0 0
Enkeleida Kapia
w c u cw fw cc sc p rc
Table 3 indicates the means for micro structure elements for both monolinguals
and bilinguals according to all the micro structure elements.
An Independent Samples T test indicates that there is no significant difference
between monolinguals and bilinguals with respect to the categories of “coordinat-
ing conjunctions” and “relative clauses”. In other words, children seem to perform
similarly for both of these categories. On the other hand, the Independent Samples
T Test shows that there is a significant difference between monolinguals and bilin-
guals with regard to all of the following categories: words, clauses, utterances,
content words, function words, subordinating conjunctions, and passives. Indeed,
as seen from the raw numbers in Tables 1 and 2, monolinguals produce a higher
number of these than bilinguals do.
This section reviews findings on the different analysis performed for the macro
elements of the narratives of both monolinguals and bilinguals. Table 4 provides
the raw numbers for the macro structure of the narratives produced by the mono-
lingual children. Table 5, on the other hand, presents the raw numbers for the
same macro structure elements in bilingual children’s narratives.
Table 6 indicates the means for macro structure elements for both monolin-
guals and bilinguals according to all the macro structure elements.
An Independent Samples T test indicates that there is no significant difference
between monolinguals and bilinguals with respect to any of the macro structure
elements.
1 1 3 3 6
2 3 1 1 8
3 3 3 3 6
4 3 2 2 11
5 2 2 2 7
6 1 2 2 2
Assessing narrative development in bilingual first language acquisition
1 1 0 1 2
2 1 0 1 6
3 0 1 1 6
4 2 2 2 6
5 1 1 0 3
6 2 3 0 4
to the micro structure elements of their narratives. This result agrees with some
previous findings, but does not with other findings reported in the field. Differ-
ently, these previous studies have demonstrated either no difference found be-
tween the two groups (Umbel et al. 1992), or a higher number of micro structure
elements, such vocabulary words, for bilinguals than monolinguals (Rosenblum &
Pinker 1983). However, this result does resonate with results found by other work,
which seems to suggest that bilinguals fall within a lower norm than monolinguals
when it comes to lexical development (Fernandez et al. 1992; Allman 2005).
Even though the reasons why bilinguals show a lower performance on micro
structure elements when compared to monolinguals are not entirely clear and
straightforward at all times, there are at least a couple of attempts in the literature
that target the explanation of these observed differences. One of them maintains
that bilinguals’ developing cognitive capacities impose limitations on the amount
of information that can be stored in the accessible memory (Bialystok 2001).
Another account suggests that lower performance is related to lower frequencies
of micro structure elements in either language (Ben Zeev 1977). These two factors
alongside the factors that affect the linguistic environment of bilinguals may fully
explain the lower performance observed here.
What is striking, however, and unexpected in a way, is the fact that bilinguals
and monolinguals here are seen to perform at about the same levels with respect
to the macro structure elements of a narrative. In other words, monolinguals and
bilinguals aged 6–7 years old show the same developmental patterns for all the
macro structure elements investigated here. While such similar results have not
yet been reported in the literature (with the exception of results on lexical develop-
ment, for this see de Houwer 2010), these results are, at the very least, proof that
being bilingual does not harm conceptual development. The bilingual children in
this study show the same narrative developmental milestones as observed for
monolingual children even when it comes to macro structure elements. The devel-
opmental path of their macro structure elements of their language(s) does not
suffer at all as a result of their being bilingual, i.e. having to deal with the process-
ing and computational load of two languages at the same time.
These results have important implications towards the development of better
language assessment tools for bilinguals. As seen here, bilingual children do not
produce errors per se. What we see as differences in performances here can be es-
sentially attributed to distributed knowledge across the two languages. As such, it
becomes important to understand that these differences do not mark delay or an
atypical developmental path, but rather a normal way of development for such
children. To this end, knowing and understanding whether aspects of develop-
mental paths of monolingual children and children in superdiverse worlds differ
becomes a question of interest not just from the point of view of research, but also
Assessing narrative development in bilingual first language acquisition
from the point of view of educational and clinical matters. As a result, the develop-
ment of appropriate assessment tools and practices for determining language de-
velopment or language delay should use comparisons to appropriate normative
groups and appropriate language targets. Although it has been a general tendency
to use assessment tools that target lexical or vocabulary development in bilinguals
in order to determine delay, such measures may not be the only effective ones, and
as this study showed, such measures may not even be the most effective ones at all.
In particular, the results from this study seem to suggest that monolingual norms
should not be used to assess bilinguals’ micro structure language development.
However, monolingual norms are quite effective in assessing bilinguals only if they
investigate macro structure elements. In practical terms, a bilingual child who per-
forms well according to monolingual macro structure norms does not need any
help by a speech language pathologist.
In conclusion, although it is important to understand why norms for mono-
lingual children cannot be taken as norms that represent the development of chil-
dren growing up in superdiverse environments, it is also useful to understand that
these developmental paths overlap in both of these populations. Determining
which areas overlap can serve as an educational and clinical tool in order to ac-
curately determine language delay and enable early intervention.
References
Allman, B. 2005. Vocabulary size and accuracy of monolingual and bilingual preschool chil-
dren. In ISB4: Proceedings of the 4th International Symposium on Bilingualism. Somerville
MA: Cascadilla Press.
Bedore, L. & Peňa, E. 2008. Assessment of bilingual children for identification of language im-
pairment: Current findings and implications for practice. International Journal of Bilingual
Education and Bilingualism 11(1): 1–30.
Ben Zeev, S. 1977. The influence of bilingualism on cognitive strategy and cognitive develop-
ment. Child Development 48(100): 9–18.
Berman, R.A. & Slobin, D.I. 1994. Relating Events in Narrative: A Crosslinguistic Developmental
Study. Hillsdale NJ: Lawrence Erlbaum Associates.
Bialystok, E. 2001.Bilingualism in Development: Language, Literacy, & Cognition. Cambridge:
CUP.
Bornstein, M., Putnick, D. & De Houwer, A. 2006. Child vocabulary across the second year:
stability and continuity for reporter comparisons and a cumulative score. First Language 26:
299–316.
Bornstein, M., Tamis-LeMonda, C. & Haynes, M. 1999. First words in the second year:
Continuity, stability, and models of concurrent and predictive correspondence in vocabu-
lary and verbal responsiveness across age and context. Infant Behavior and Development
22(1): 65–85.
Enkeleida Kapia
Dart, S.N. 1992. Narative style in the two languages of a bilingual child. Journal of Child Lan-
guage 19: 367–387.
De Houwer, A. 2010. Assessing lexical development in bilingual first language acquisition: What
can we learn from monolingual norms? In Multlingual Norms, M. Cruz-Ferreira (ed.),
279–322. Frankfurt: Peter Lang.
Emslie, H. & Stevenson, R. 1981. Pre-school children’s use of the articles in definite and indefi-
nite referring expressions. Journal of Child Language 8: 813–828.
Fernandez, M.C., Pearson, B.Z., Umbel, V.M., Oller, D.K. & Molinet Molina, M. 1992. Bilingual
receptive vocabulary in Hispanic preschool children. Hispanic Journal of the Behavioral
Sciences 14: 268–276.
Fiesta, C. & Peňa, E. 2004. Narrative discourse in bilingual children: Language and task effects.
Language, Speech and Hearing Services in Schools 35(2): 155–168.
Hart, B. & Risley, T. 1995.Meaningful Differences in the Everyday Experiences of Young American
Children. Baltimore MD: Brookes.
Marchmann, V. & Fernald, A. 2008. Speed of word recognition and vocabulary knowledge in
infancy predict cognitive and language outcomes in later childhood. Developmental Science
11(3): 9–16.
Papakonstantinou, M., Kapia, E. & Tsimpli, I. 2011.Narrative structures in Albanian-Greek
bilinguals. Malta 2nd Conference of Cost Action ISO804. Assessing Language Impairment
in a Multilingual Society: Linguistic Patterns and the Road to Assessment, University of
Hamburg.
Pearson, B.Z., Fernandez, S. & Oller, D.K. 1993. Lexical development in bilingual infants and
toddlers: Comparison to monolingual norms. Language and Learning 43(1): 93–120.
Rosenblum, T. & Pinker, S.A. 1983. Word magic revisited: Monolingual and bilingual children’s
understanding of the word-object relationship. Child Development 54: 773–380.
Silliman, E.R., Huntley Bahr, R., Brea, M.R., Hanth-Chisolm, T. & Mahecha, N.R. 2001. Spanish
and English proficiency in the linguistic encoding of mental states in narrative retellings.
Linguistics and Education 12(4): 1–37.
Thal, D., Bates, E., Goodman, J. & Jahl-Samilo, J. 1997. Continuity of language abilities: An explor-
atory study of late and early talking toddlers. Developmenal Neuropsychology 13: 239–273.
Umbel, V., Pearson, B.Z., Fernandez, M.C., & Oller, D.K. 1992. Measuring bilingual children’s
receptive vocabularies. Child Development 63: 1012–1020.
Vertovec, S. 2006. The Emergence of Super-Diversity in Britain. Oxford: Centre of Migration,
Policy and Society.
Vertovec, S. (2007). Super-diversity and its implications. Ethnic and Racial Studies 30(6):
1024–1054.
Examples of language contact and change
Detecting historical continuity
in a linguistically diverse urban area
The present perfect in modern Singapore English
Julia Davydova
University of Hamburg
1. Introduction
The goal of the present study is to explore the problems related to detecting histori-
cal continuity and effects of language contact in a linguistically diverse urban area
by focusing on the historical variety of Irish English (henceforth IrEng) and pres-
ent-day Singapore English (henceforth SingEng). More specifically, we seek to dis-
cover which type of evidence or which indicators may provide basis for diagnosing
the processes of diffusion associated with historical continuity in New Englishes.
In so doing, we take recourse to the methodological inventory adopted in the
quantitative paradigm of sociolinguistics (e.g. binominal step-up/step-down anal-
ysis) and apply it to the morphosyntactic domain of the present perfect, while
Julia Davydova
1.1 Terminology
The variants alternating with the have perfect include the preterite (such as
asked, went, came, e.g. I watched it on TV once it’s very good, ICE-Singapore: S1A–
012), be perfect (such as is gone, is failed, is come, e.g. He is gone off himself this
morning, HCIE: McKayJ01), the present (such as is studying, are buying, does, goes,
e.g. the poverty and distress I am going through here this six months, HCIE:
BurkeW01), lone past participle (such as gone, done, been, e.g. I shall not enter into
a history of many things she done and said, HCIE: SprouA01), lone present parti-
ciple (such as studying, going, e.g. I am from Banglore, teaching in college for past
one year, ICE-India: S1a–021), medial-object perfects (such as have a letter writ-
ten, e.g. We have not the Buildings finished, HCIE: BrownW07), and after perfects
(e.g. We are just after receiving a letter from Kate of Farside, HCIE: Mahony02).
IrEng and SingEng have both emerged as a result of extensive bilingualism, and
can by and large be described as products of second-language acquisition (see also
Hickey 2004: 95 for IrEng and Schneider 2007: 153 for SingEng). In addition to L1
influence and universal learning strategies, other factors, which arguably underlie
formation of New Englishes including SingEng, are input varieties, otherwise
known as superstrate(s) in the relevant literature (see, for instance, Mesthrie &
Bhatt 2008: 188ff.). These are dialects spoken by different native speaker groups
(e.g. missionary English, sailor English, English of American Methodists, etc.) in a
language contact situation.
Studies on second-language acquisition, however, show that the domain under
investigation, i.e. the present perfect, is to a large extent determined by L1 influ-
ence or substrate influence. Slobin (1996) makes a strong case for a contact-induced
scenario arguing that tense and aspect categories in speakers’ L1 shape and con-
strain the subsequent acquisition of the L2 tense and aspect system. In a similar
vein, Bao (1995, 2005) offers a substratist explanation for some tense and aspect
features attested in SingEng. The study sets out to explore inter alia the possibility
of a substrate-induced scenario in the domain of the present perfect marking.
need to take a closer look at the main stages of its historical development. To start
with, SingEng is a textbook definition of a contact variety. After becoming a trad-
ing post and a free port of the British East India Company, Singapore attracted
Chinese, Malay and Arab traders (Schneider 2007). Various dialects of Chinese
such as Hokkien, Teochow, and Cantonese (cf. Gramley & Pätzold 2004), Malay
pidgins, Tamil, and English have been in constant contact ever since (Gupta 1998;
Ansaldo 2004, 2009; Lim 2004; Deterding 2007).
The population grew steadily throughout the 19th century, new arrivals com-
prising predominantly traders, colonial agents, and contract labourers of Chinese
and Indian origin. The spread of English was continuous and “[n]othing hindered
a fundamentally exonormative orientation. [...] Bilingualism was spreading but
had an elitist touch” (Schneider 2007: 154).
The 20th century saw an increase in English-medium schools and concomi-
tant spread of bilingualism. The input varieties included Scottish English, English
spoken by American Methodists and, finally, Irish English (cf. Schneider 2007:
154; Gupta 1998: 107). In a similar vein, Platt & Weber (1980: 5) mention upper
class English, Scottish, Welsh and Irish regional varieties in their list of super-
strates relevant to the formation of SingEng.
World War II interrupted Singapore’s prosperous history and when the British
authority was restored, the situation had changed considerably in terms of language
attitudes and identity constructions. After gaining its independence in 1965,
Singapore experienced a stunning economic development resulting in a highly mod-
ernised nation. English has come to play a unique role in the context of multicultur-
alism. Firstly, not being associated with a particular ethnic group, English has be-
come a common linguistic asset linking various population groups (cf. Lick &
Alsagoff 1998: 127ff.). Secondly, a policy of additive bilingualism pursued by the
government made English-medium education obligatory. To put it slightly differ-
ently, children are to be educated in their respective ethnic language and in English.
As a variety, SingEng has undergone the process of restructuring or nativisa-
tion, which is particularly visible in informal registers or vernaculars. Despite the
existence of these distinct local forms of SingEng, “exonormative standards con-
tinue to define the study of English in the classroom” (Ooi 2001: x).
While setting out on the search for effects associated with historical continuity in
New Englishes, it is important to be clear about the details of the sociohistorical
context accompanying their emergence. The importance of a careful examination
of sociohistorical details in tracing the roots of a variety has been emphasized in,
for instance, Montgomery (1989: 235–237).
Detecting historical continuity in a linguistically diverse urban area
were sent to Asia in 1968 (Hally 1971: 340). The Catholic missions still run pri-
mary and secondary schools in Singapore, which are highly regarded.
With the increase in popularity of English-medium schools, non-native English
teachers from Sri Lanka and India also found employment in Singapore. The im-
portance of Indian languages on SingEng has been documented in, for instance,
Ramish (1969, cited in Ho & Platt 1993). Since this study assesses the relative con-
tribution of the Irish English input to the development of SingEng, an excursion
into the history of the Irish community in Singapore is due at this point.
Although the history of the Irish community in Singapore is filled with blank
spots, we know that the Methodists and the Catholics were among the first groups
to have arrived in Singapore after it became a crown colony and soon turned out
to be an important source in English-medium education. Whereas Irish Catholics
are linked to the Singaporean system of education, Irish Protestants are to this day
associated with an important theological centre in Singapore (Hally 1971: 346).
A somewhat informal but perhaps the most detailed account of the Irish mi-
gration to Singapore can be found in Lim (2008), who reports, for example, that
out of 54 national monuments in Singapore five were created by the Irish archi-
tects. She also mentions that six Irishmen held prominent posts in Singapore at
various times. In her description the Irish community emerges as vibrant and
actively shaping Singaporean scenery. Traditions brought from Ireland to
Singapore include Gaelic football and the St. Patrick’s Day Parade celebrated year-
ly in Singapore. The St. Patrick’s Society provides social connections for the Irish
community in Singapore. Last but not least, according to Lim (2008) Guinness
beer has been in Singapore for around 140 years.
Given that the emergence of SingEng as a distinct variety of English was accompa-
nied by extremely heterogeneous, that is to say linguistically diverse, inputs not
only from the respective mother tongues of its speakers but also from different
superstrate varieties dating back to the 19th century, two questions need to be ad-
dressed: (i) which types of linguistic influence is it necessary to consider in the
account of SingEng and (ii) how can we disentangle the effects of various types of
influence?
As far as the first question is concerned, SingEng has at least three major com-
ponents that need to be considered in order to determine which influences may
have come from contact with Irish English:
Detecting historical continuity in a linguistically diverse urban area
2. Note that this definition of diffusion is different from that found in Labov (2007). Labovian
definition of diffusion is related to the wave model, which views diffusion as gradual dispersion
of linguistic features from one community to the next. See Labov (2007) for discussion and
Tagliamonte (2012) for a detailed overview.
Julia Davydova
Whereas the outcomes of the selection are relatively easy to point to because
of the form-function saliency of some features (take, for instance, the after perfect
in IrEng), detecting results of diffusion requires sophisticated techniques that
would enable the analyst to trace the subtle effects of contact-induced change.
Thus, that the have perfect is a feature stemming from English is beyond any doubt.
However, to be able to ascertain whether the patterns of its use are conditioned by
an intrusion from indigenous language systems, whether they are constrained by
universal processes or whether they are inherited from an older superstratal sys-
tem through diffusion, we need to carry out multivariate analyses of data. It is
precisely at this point where the variationist approach (as elaborated in what fol-
lows) comes to play a prominent role.
The variationist sociolinguistic method is a suitable tool for inferring
(historical) relatedness or, conversely, a lack of (historical) relatedness of language
varieties (cf. Poplack & Tagliamonte 1989; Meyerhoff 2009: 301). This method is
also very useful in language contact studies because the most basic components of
a variationist analysis – identification of statistically significant factor groups,
establishing a hierarchy of factor effects for all factor groups, and identification of
factor weights within each factor group – allow for detailed and therefore meticu-
lous comparisons of data sets under study. Furthermore, constraint rankings and
significant factor weights are probably too subtle to be picked up by speakers de-
liberately and therefore may provide the right type of evidence for historical con-
tinuity in New Englishes.
While carrying out comparisons between 19th-century IrEng and modern
SingEng, the study draws on the variationist model for a contact-induced sce-
nario provided in Meyerhoff (2009: 303). The model is concerned with cases in
which inherent variability is a feature of what is called a model language and a
replica language in studies on language contact (Heine & Kuteva 2005). The for-
mer is a variety in which a pattern originates historically; the latter represents a
variety that adopts (and adapts) the pattern as a result of contact with the model
(Meyerhoff 2009: 298). When applied to the language-contact situation of the type
found in Singapore, the dichotomy between a ‘model language’ and a ‘replica
language’ can be understood as a dichotomy between an ‘input variety’ and a
‘New English’.
In order to establish which patterns of variation can be attributed to contact-
induced transfer and to what effect, Meyerhoff (2009) proposes three principles
for incorporating variation into the study of language contact. These principles are
as follows (Meyerhoff 2009: 303):
Detecting historical continuity in a linguistically diverse urban area
(2) i. Where the same factor groups are significant constraints on a variable
in the model and in the replica varieties, let us call this weak transfer
or replication;
ii. Where the same factor groups are significant in both model and rep-
lica, and the ordering of these factors groups is the same in both mod-
el and replica, let us call this (strong) transfer;
iii. Where the same factor groups are significant in both model and rep-
lica, and the ordering of these factor groups is the same in both model
and replica, and the factors within groups have the same ranking in
model and replica, let us call this calquing.
Meyerhoff (2009) designed her model to establish the degree of relatedness be-
tween an indigenous language and a new variety emerged as a result of a language
contact situation (see Meyerhoff 2009 for further details). Taking this idea one step
further, this paper argues that this model can effectively be used to detect effects of
different kinds of input, not just those related to the mother tongue influence. In
other words, we can use this model to place a New English vis-à-vis other relevant
varieties of English, including historical Englishes, in order to see whether they
contributed to its formation, to which degree and with which effects.
Meyerhoff ’s work also suggests that “unless we seriously evaluate variation
[...], we may erroneously attribute inheritance, or mistakenly infer historically
shared development, to varieties that are not in fact “sisters under the skin””
(Meyerhoff 2009: 301). This article argues in much the same vein that a data-driven
approach to studying variation is crucial for understanding contact-induced effects
in varieties emerged in linguistically complex situations.
Data
The major source for the treatment of IrEng is the Hamburg Corpus of Irish English
(HCIE), “a collection of dialectal, sub-literary English texts of Irish provenance”
(Pietsch 2007: 9). It consists of letters written in the context of emigration to the
United States and Australia and of the social and political upheaval (represented
by prisoners’ letters, anonymous threatening letters, petitions, etc.) This collection
is complemented with additional sources, such as Fitzpatrick (1994) and Miller et
al. (2003). The data used in this study amount to some 261,062 words of text. It
covers the 18th and 19th centuries, although the latter is better represented due to
a much better availability of sources (cf. Pietsch 2007: 10).
Julia Davydova
The corpus has some limitations of its own. Probably the biggest problem is
the heterogeneity of the sample at hand. In other words, for some authors English
was already a native idiom by the time these texts were written, whereas others
used it as a second language they acquired as adults. Indeed little is known about
the exact amount of contact that those writers had with the medium of English.
Although the letters present their own difficulties as most historical texts do
(cf. Hickey 2010), they can nevertheless be said to “reflect the authentic linguistic
intentions of their writers” (Pietsch 2007: 10) and can thus be argued to mirror – at
least, to some extent – the vernacular style, a necessary prerequisite for the study
of variation and change (cf. Labov 1972, Schneider 2002: 69). More importantly,
the texts under analysis comply with the majority of criteria for texts suitable for a
variationist analysis listed in Schneider (2002: 71):
i the texts display a variability of phenomena under investigation, i.e. they
contain functionally equivalent variants of a linguistic variable;
ii the texts reflect the diversity of extra-linguistic parameters: they stem from
different authors of both sexes and, as mentioned above, represent varying
stylistic levels;
iii the texts provide large token frequencies of individual variables and are suit-
able for quantitative analyses of different linguistic phenomena;
iv in terms of regional distribution, the corpus covers major dialectal areas of
Ireland: out of the total of 384 texts, 214 originate from Ulster, 66 from
Munster, 56 from Leister and 18 from Connaught; in 30 cases the author’s
provenance remains unknown.
Finally, the corpus provides information on the religious background of its writ-
ers, as demonstrated in Table 1.
Although religious background remains unknown in many cases, a significant
proportion of writers in the sample were Catholics – 48 writers, which accounts
for 50% of all known backgrounds. This is quite remarkable given the records of
Anglicans 16 writers
Catholics 48 writers
Presbyterians 16 writers
Protestants 9 writers
Quakers 6 writers
Methodists 2 writers
Religion unknown 42 writers
Total 139 writers
Detecting historical continuity in a linguistically diverse urban area
Catholics missions from Ireland in Singapore and their fairly conspicuous pres-
ence in the Singaporean system of education. Taking all the pieces of evidence and
lines of argumentation into consideration, we argue that the corpus can be used as
a suitable baseline allowing the analyst to draw comparisons between present-day
SingEng and its 19th-century input from Ireland.
The SingEng data was obtained from the Singaporean component of the Inter-
national Corpus of English (ICE). The corpus comprises 100 texts amounting to
some 200,000 words. The focus is on informal vernacular as found in direct con-
versations (S1A–001 to S1A–090) and telephone calls (S1A–091 to S1A–100).
In order to enlarge the scope of our comparison, we adduce the Indian and the
East African component of the ICE family to the analysis. Both data sets comprise
the same genre, i.e. direct conversations (S1A–001 to S1A–090) and telephone
calls (S1A–091 to S1A–100) for ICE-India and direct conversations/classroom dis-
cussions (S1A001K-S1A030K and S1A018T-S1A020T) for ICE-East Africa. Simi-
lar to SingEng, Indian English (IndEng) and East-African English (EAfEng) are
products of second-language acquisition and emerged largely through the educa-
tion system in multilingual settings. These varieties provide proper baselines for
drawing comparisons with SingEng and will help us find out whether patterns of
use attested in the data are due to universal processes underlying second-language
acquisition or whether they are indeed to be ascribed to an historical input. This
comparison in turn will improve our understanding of the relationship between
19th century IrEng and modern SingEng.
The London-Lund Corpus (LLC) of spoken English represents another yard-
stick against which the non-native and historical data is measured and compared.
The corpus represents a mainstream vernacular spoken by native educated speak-
ers in England (henceforth, StEngEng). The LLC is a collection of spontaneous
surreptitiously recorded conversations and non-surreptitious private and public
conversations and is therefore largely compatible with the data from the ICE cor-
pora. Similar to the ICE data, the LLC data does not include analyses of texts con-
taining monitored speech (e.g. radio commentaries, parliamentary debates, and
political speeches), its major focus being on the vernacular data.
The addition of the LLC data considerably improves the methodological de-
sign of the study. Like many other New Englishes, SingEng owes a great deal to
classroom acquisition, which typically has an exonormative orientation towards
the native standard. It is thus an interesting task for the analyst to find out which
patterns of use of the have perfect are likely to have emerged as a result of exposure
to Standard English, thereby excluding this type of influence from other super-
strate effects.
Julia Davydova
Dependent variable
The study focuses on the investigation of the variants surfacing in resultative, ex-
tended-now, experiential contexts as well as contexts of recent past (see Davydova
2011 for a detailed explanation of the circumscription of the variable context).
These contexts are illustrated in (3) through (6)
(3) resultative
I have eaten a cupcake but I am still hungry.
(4) extended-now
It has been raining incessantly all day long.
(5) experiential
Have you ever seen such a miracle?
(6) recent past
The Prime Minister has resigned recently.
The dependent linguistic variants comprise the have perfect, the preterite, the
present tense, be perfects, lone present participles, lone past participles, the medi-
al-object perfect, and the after perfect. Whereas some of these forms are circum-
scribed to specific contexts or varieties (e.g. after perfect in context of recent past
in IrEng only), there are two forms that alternate with each other, albeit to differ-
ent degrees, in all contexts and in all varieties studied here. These forms are the
preterite and the have perfect. They arguably represent a set of linguistic items that
share the sameness of cognitive meaning in that they both relate an event to the
past (one variant, the have perfect, links a past situation to speech time, whereas
the other, the preterite, does not) and can thus be conceived of as variants of the
linguistic variable called the present perfect here.
These variants formed a basis for the multiple regression models (binominal
step-up/step-down), which were constructed with the help of Goldvarb 2001.
Because the multivariate methodology requires that variants occur often enough
in all contexts under study, all infrequent items and items restricted in their distri-
bution, i.e. be perfects, bare verb stems, lone present participles, etc., were exclud-
ed from the multivariate analyses (cf. Winford 1993; Tagliamonte 2006). They are
presented and discussed, however, in the distributional analyses of data.
In the next step, we explain why the present perfect was chosen for the explo-
ration of the issues addressed in this study. First and foremost, the category of the
present perfect represents a domain with a high degree of language-internal varia-
tion and is therefore well suited for studying contact-induced phenomena and
their patterns of use. Secondly, exhibiting high levels of complexity in the sense of
L2 difficulty (cf. Trudgill 2001; Kusters 2003, 2008; Szmrecsanyi & Kortmann
2009), the present perfect needs special attention in terms of instruction and is
Detecting historical continuity in a linguistically diverse urban area
being addressed again and again by English teachers in the classroom. Given the
importance of the classroom acquisition in the emergence of New Englishes and
the popularity of Christian English-medium schools in Singapore, it seems rea-
sonable to focus on the investigation of those language categories that are associ-
ated with that type of environment.
Independent variables
Table 2. Linguistic factors claimed to affect the use of the have perfect and related forms.3
3. Note that some time adverbials such as yet, now, up until now, etc. exhibit a close associa-
tion with the have perfect. Yet these distinctions are not accounted for in this study.
Detecting historical continuity in a linguistically diverse urban area
6. Results
As a first step, we assess the most general way in which the linguistic variable un-
der study operates in the relevant data sets. Table 3 demonstrates the overall distri-
butions of variants in present perfect contexts in SingEng (ICE), IrEng (HCIE) and
StEngEng (LLC).
As we can see, the have perfect is extremely robust in the data set representing
the mainstream native-speaker variety (StEngEng). It accounts for 90% of all
relevant cases, leaving only a modest 9% for its major contender: the preterite.
Julia Davydova
The have perfect is the most frequent variant in both SingEng and IrEng. The other
robust variant is the preterite; it remains stable in both varieties. In addition, the
IrEng corpus contains a fair amount of be perfects and present tense tokens, none
of which occupy a prominent position in the SingEng data. In contrast, SingEng
manifests a fairly conspicuous number of bare verb stems, which represent a clear-
cut case of substrate influence from Chinese substratum. Since these cases have
been meticulously studied elsewhere (cf. Kwan-Terry 1989; Bao 1995; Deterding
2007; Mesthrie & Bhatt 2008; Davydova 2011), they will not be dealt with here.
In order to reveal which factor groups condition the occurrence of the have
perfect in modern Sing Eng, 19th-century IrEng and StEngEng, we perform mul-
tivariate analyses of these data sets. In doing so, we seek to discover which patterns
of variability attested in the data can point to similarities obtaining between which
varieties and to what extent.4
4. Because the factor groups ‘Aktionsart’ and ‘semantic context’ showed significant interac-
tions in the data on IrEng, I chose to run two separate analyses of this data set (cf. Tagliamonte
2006: 229ff.). In the first run, I excluded ‘semantic context’ from the analysis. In the second run,
I excluded ‘Aktionsart’. I then used the log likelihood to compare both analyses, thus revealing
which analysis was better (cf. Tagliamonte 2006). The analysis from which ‘semantic context’
was absent yielded a log likelihood of –830,86. By contrast, the analysis excluding ‘Aktionsart’
exhibited a log likelihood of –756,00. The latter analysis is therefore a better fit for the model
since its log likelihood is closer to zero. I furthermore redefined time adverbials ever, never and
always as a special case of an experiential context, including it into the factor group ‘semantic
Detecting historical continuity in a linguistically diverse urban area
The first observation is that two factor groups were selected as statistically
significant in three data sets: semantic context and time adverbial specification.
The ordering of these factor groups is the same in all varieties: semantic context
exerts a strong effect, whereas time adverbial specification has the least strong ef-
fect. Moreover, the constraint ranking for each factor group remains stable across
the board. Whereas resultative and extended-now contexts favour the have perfect
in SingEng, IrEng and StEngEng, experiential contexts and contexts of recent past
have a much lesser impact on the occurrence of the present perfect. These results
are consistent with the evidence obtained from some non-standard L1 vernaculars
(e.g. Scottish English as reported in Miller 2004) and are in line with the hypoth-
eses outlined in Table 2.
Furthermore, temporally unspecified contexts unequivocally display a signifi-
cant association with the have perfect, as predicted by the hypothesis. We can thus
see that a resultative context with no time adverbial specification is the strongest
predictor of the occurrence of the have perfect in different – albeit historically re-
lated – varieties of English. In a way, it is this specific pattern of use that renders
these language systems unmistakably English. Moreover, this finding empirically
corroborates the descriptivist claim that a resultative perfect is a central manifesta-
tion of the English present perfect (Comrie 1976; Siemund 2004).
On closer inspection we notice, however, that vivid similarities exist between
SingEng and IrEng, which forms a sharp contrast to the reference variety. To start
with, the factor groups transitivity and negation are statistically significant con-
straints in both SingEng and its historical input variety: IrEng. Note that none of
these factor groups was chosen as a significant variable for the reference variety of
StEngEng.
In both varieties, transitive and intransitive predicates are associated with the
have perfect much more than with mutative verbs, that is verbs of motion and
change-of-state verbs, e.g. change, leave, come, vanish, die, etc. In other words,
there is a divide between transitive and intransitive predicates on the one hand and
mutative predicates on the other in the two data sets. However, note also that the
copula BE exerts the strongest impact on the occurrence of the have-perfect in
SingEng, whereas in IrEng this role is assigned to intransitive verbs. In all, the
hypotheses spelled out for this factor group in Table 2 are at least partially borne
out in both data sets.
context’. Here is a linguistic explanation for this slight rearrangement of factors: the semantics of
adverbs ever, never and always are compatible with the semantics of an experiential context.
Moreover, these temporal specifiers are limited to experiential environments. We can thus view
experientials with ever, never and always as a subtype of an experiential context (e.g. I have
never been to America), as opposed to an experiential context not containing any time adverbial
specification (e.g. I have been to the dentist, but not very often).
Julia Davydova
Table 4. Contrastive multivariate analyses of the contribution of factors selected as significant to the probability of the have perfect in Sing
SingEng (ICE), IrEng (HCiE) and StEngEng (LLC)
have perfect
SingEng (ICE) IrEng (HCIE) StEngEng (LLC)
Input value FW % N FW % N FW % N
have perfect
SingEng (ICE) IrEng (HCIE) StEngEng (LLC)
Input value FW % N FW % N FW % N
Further evidence can be adduced in favour of the contention that there are,
indeed, close parallels between modern SingEng and 19th-century IrEng: it is de-
rived from the factor group negation (or polarity). Within this factor group a dis-
tinction is drawn between affirmative clauses and negative clauses, which are then
subdivided into two further categories: clauses consisting of the auxiliary (HAVE
or DID) followed by the negative particle not and (the past participle of) the main
verb and clauses featuring negative words such as never, no, none, nothing. The
former type of negation is referred to as not-negation in this study; the latter type
has been termed as no-negation (cf. also Cheshire 1999: 34; Mazzon 2004: 22).
Table 4 makes clear that the not-negation (e.g. He hasn’t come home yet) is a
strong favouring context for the have perfect in both SingEng and IrEng; the effect
of this factor group is not significant in StEngEng. This finding corroborates the
previous research on the English present perfect in L1 vernaculars and earlier forms
of English, which suggests that “the long association of negation and the English PP
is reinforced by the dual role of the auxiliary HAVE attracting the negator (I have
not seen it) while also serving as the perfect auxiliary” (Van Herk 2008: 63).
Table 4 also highlights the fact that whereas not-negation is highly preferred
with the have perfect, the contexts featuring no-negation favour the preterite. The
pattern, which is consistent in both data sets, is illustrated in (7) through (8). Note
also that in StEngEng both not-negation and no-negation are highly frequent with
the have perfect (see Table 4), a finding that is consistent with the previous research
on the present perfect in native-speaker English (see, for instance, Elsness 1997).
(7) negation in IrEng (HCIE: CarroJ03; Doorle01)
a. I got no answer.
b. He has not lived with her this three years.
(8) negation in SingEng (ICE: S1A–004; S1A–031)
a. Well at least you never had a problem of being drunk.
b. Oh so they have not passed you the grades lah.
In sum, factor groups of transitivity and negation are statistically significant vari-
ables underlying the occurrence of the have perfect in both SingEng and 19th-
century IrEng. Constraint hierarchies for these factor groups remain constant in
both data sets with the exception of the copula BE constraint, which has a highly
significant impact on the occurrence of the have perfect in SingEng but not in
IrEng. Moreover, these similarities in the patterns of use of the have perfect stand
in stark contrast to those attested for the reference variety of StEngEng.
However, to be able to ascertain whether these similarities are indeed due to
the influence from the historical IrEng input or whether they are just a result of a
mere cross-varietal coincidence, we need to consider the data presented here in
Detecting historical continuity in a linguistically diverse urban area
the light of the three major components that have been constraining the genesis of
SingEng. Given the peculiarities of the sociohistorical context accompanying the
emergence of this variety, these components are (i) substratal, or L1-related, influ-
ence, (ii) L1 independent, universal processes and (iii) (historical) inputs from the
superstrates. We now examine major possible explanations for the patterns of use
of the have perfect in SingEng in terms of these components.
7. Analysis
5. Note that an ideal-case scenario would be to carry out a corpus-based quantitative study
disentangling effects related to transitivity and negation. This undertaking would, however, jus-
tify another research project (see, for instance, Odlin (2009) for the discussion of the method-
ological problems and challenges implicated in discerning substrate influence). I therefore restrict
myself here to the description of the indigenous tense and aspect systems found in the literature.
Julia Davydova
Whereas compound tenses featuring iru-kka are not related to any specific ef-
fects associated with transitivity, being common with both transitive and intransi-
tive predicates (see examples in Steever 2005: 176–177), a connection has been
reported between compound tenses featuring iru-kka and negation. Steever (2005)
shows that the forms of iru-kka occur without any restrictions in the affirmative
contexts and are usually not allowed in the past and present so that the Tamil sim-
ple present tense, the simple past tense and the compound tenses corresponding to
the English present perfect and past perfect are collapsed into one form, as in
(11) from Steever (2005: 173)
nī pirakka.υ illai
you-S-nom be.born-inf ind-neg
‘You weren’t/hadn’t/aren’t/haven’t been born’
The pattern correlating verbal forms to the feature of negation attested in Tamil is
distinct from that attested in the data on SingEng: It is not the case that the Tamil
perfect compound tense is associated with a specific negation marker and disas-
sociated with another. We therefore conclude that the patterns of use of the have
perfect in our data cannot be explained in terms of influence from Tamil.
In Malay, verbs are not inflected for gender, number or tense. Tense distinc-
tions are rendered lexically with the help of time adverbials, e.g. sudah ‘already’ or
belum ‘yet’, semalam ‘yesterday’, sekarang ‘now’, etc. (cf. also Omar & Subbiah 1968:
4). The Malay grammar does not seem to contain linguistic structures whose
morphosyntax or semantics could foster cross-linguistic identification with the
English present perfect.
To sum up, although the indigenous languages do not exhibit morphosyntactic
structures similar to the English structure consisting of the auxiliary HAVE and a
past participle, some of them, i.e. Mandarin Chinese and Tamil, possess language
items whose functions overlap with those of the English have perfect. A careful
examination of the indigenous grammars, however, indicates that linguistic be-
haviour of these language forms does not match the patterns of use attested for the
SingEng have perfect. It is therefore unlikely that the patterns of use of the have
perfect in SingEng derive from the indigenous substratum.
We have mentioned that both SingEng and IrEng are varieties of English that have
emerged as a result of prolonged language contact. What this observation implies
is that the common patterns of use related to transitivity and negation may have
arisen in both varieties independently of one another, simply by virtue of the fact
that both are contact varieties involving significant amounts of bilingualism and
second-language acquisition.
Julia Davydova
Table 5.╇ Contrastive multivariate analyses of the contribution of factors selected as significant to the probability of the have perfect in
SingEng (ICE), IndEng (ICE) and EAfEng (ICE)
have perfect
SingEng (ICE) IndEng (ICE) EAfEng (ICE)
Input value FW % N FW % N FW % N
have perfect
SingEng (ICE) IndEng (ICE) EAfEng (ICE)
Input value FW % N FW % N FW % N
3.╇ Negation
Not-negation .80 88% 113/128 [.56] 65% 91/139 [.48] 59% 31/52
affirmative .45 57% 411/715 [.49] 59% 624/1047 [.50] 61% 216/354
No-negation .20 20% 8/39 [.56] 65% 91/139 [.48] 59% 31/52
range 60
4.╇ Time adverbial
spec.
unspecified .53 65% 367/564 .51 62% 574/918 [.49] 60% 191/317
time adverbial .43 51% 165/318 .39 49% 42/85 [.50] 62% 56/89
range 10 12 15
5.╇ Aktionsart
stative [.53] 62% 147/234 .57 59% 158/266 [.51] 65% 60/91
dynamic [.54] 61% 197/319 .53 60% 232/384 [.44] 55% 79/143
accomplishment [.46] 56% 37/65 .44 60% 325/536 [.54] 62% 108/172
achievement [.42] 57% 151/264
range 13
Julia Davydova
In order to check whether or not these patterns of use reflect some universal pro-
cesses at work in language contact situations, we extend the scope of our analysis
to comprise those forms of English that evolved in similar sociolinguistic settings.6
Indian English (IndEng) and East African English (EAfEng) are varieties that have
emerged in post-colonial settings as a result of intensive language contact and are
characterised by high levels of adult second-language acquisition. In what follows,
we present and analyse patterns of use of the have perfect in IndEng and EAfEng
as represented by the spoken components of the ICE-corpora. We also contrast
them with the patterns attested in SingEng.
Table 5 points out that whereas semantic context universally underlies the oc-
currence of the have perfect in all varieties studied here, neither transitivity nor
negation was selected as statistically significant factors in either IndEng or EAfEng.
What this finding implies is that neither transitivity nor negation can claim a sta-
tus of universal factors underlying the occurrence of the have perfect in L2
Englishes. We can thus safely exclude the universalist hypothesis as a possible ex-
planation of these particular patterns of use of the have perfect.
In the light of the foregoing discussion, the patterns of the have perfect use in
SingEng and IrEng are indeed striking, thereby suggesting an historical-input ex-
planation as not an unlikely scenario.
But what exactly can be said about the degree of this historical-input influ-
ence? In other words, just how pervasive was that influence? Moreover, what can
be said about the position of the latter vis-à-vis mainstream English, a model in
the classroom setting for Singapore? We turn to a discussion of the results of our
study in the light of the model for contact-induced language change proposed in
Meyerhoff (2009) and reproduced for convenience in (2) in Section 4 here.
8. Discussion
SingEng: 3–2–1–4.
IrEng: 1–3–2–4.
StEngEng: 1–4.
Figure 1. Contraint hierachies of linguistic factor groups
Julia Davydova
7. While acquiring or replicating the patterns of use of the have perfect, SingEng speakers also
adjust them. Thus, copula BE has a very strong association with the have perfect in SingEng but
not in IrEng. This finding is in accordance with the principle of “transformation under transfer”
postulated by Meyerhoff (2003) whereby speakers “cannot replicate [...] the linguistic details of
the model in the replica language” and thus transform the existing/old model in the new variety.
Detecting historical continuity in a linguistically diverse urban area
8.2 The have perfect and other verb forms in present perfect contexts
9. Conclusions
The present study has been concerned with the problems involved in detecting
historical continuity in a linguistically diverse area by investigating the relation-
ship between modern Sing-Eng and one of its input varieties – IrEng – in the do-
main of the present perfect marking. It has been shown that in contrast to the
processes of selection, which are relatively easy to point to, the processes of
(historical) diffusion in newly emerged forms of English are best detected through
the subtle techniques offered by the quantitative analysis of variationist sociolin-
guistics: identification of statistically significant factor groups, establishing a
Julia Davydova
hierarchy of factor effects for all factor groups, and identification of factor impacts
relative to each other within a factor group.
The article has also attempted to demonstrate the importance of a careful,
qualitative examination of the details of the sociohistorical context as they often-
times help us understand the role played by the given superstrate in the formation
of a new variety.
Quantitative and qualitative approaches advocated in the study fit well with
the research on super-diversity encouraging new quantitative techniques coupled
with qualitative (ethnographic) undertakings (Vertovec 2007). When combined,
both approaches foster a better understanding of the dynamics and processes
underlying the evolution of language systems in linguistically heterogeneous
urban areas.
The effects related to transitivity and negation that are implicated in the occur-
rence of the SingEng have perfect can be tentatively linked to the superstratal input
as manifested by the variety of 19th century IrEng studied here. While drawing
this conclusion, we rely on the multivariate evidence supported by the document-
ed evidence of contact between the two varieties as a basis for detecting historical
continuity in SingEng. Further quantitative investigations into other historical su-
perstrates relevant to the formation of SingEng will enhance our understanding of
the relationship obtaining between the two varieties, shedding light on whether
the commonalities in the patterns of use of the present perfect in relation to tran-
sitivity and negation are IrEng specific and are to be traced back to this superstrate
or whether they are more general trends observable in other regional (non-
standard) superstrates of British English.
Finally, the study has proposed a way of uniting historical research in the study
of New Englishes with the variationist tradition specializing in contact-induced
language change. The model proposed in Meyerhoff (2009) is a suitable tool for
inferring (historical) relatedness or lack of (historical) relatedness of language va-
rieties (cf. Poplack & Tagliamonte 1989; Meyerhoff 2009: 304). Analysis of data in
terms of this model has shown that effects of language contact may manifest them-
selves in an adjustment of patterns since an exact replication of the incoming
language patterns seems to be too demanding a task for the speaker (see also dis-
cussion in Meyerhoff 2009: 313). It has been suggested, however, that the transfer
of patterns from one variety into the other may have been facilitated by the mode
of acquisition characterising the emergence of SingEng.
Following the idea explicit in Meyerhoff (2009), we conclude by saying that a
detailed examination of patterns of variation is a means of, and a key to, under-
standing the degree of contact amongst related forms of language.
Detecting historical continuity in a linguistically diverse urban area
References
Ansaldo, U. 2004. The evolution of Singapore English: Finding the matrix. In Lim (ed.), 127–150.
Ansaldo, U. 2009. The Asian typology of English. Theoretical and methodological consider-
ations. English World-Wide 30: 133–148.
Arden, A.H. 1962. A Progressive Grammar of Common Tamil. Madras: The Christian Literary
Society.
Bao, Z. 1995. Already in Singapore English”. World Englishes 14(2): 181–188.
Bao, Z. 2005. The aspectual system of Singapore English and the systemic substrate explanation.
Journal of Linguistics 41: 237–267.
Bybee, J., Perkins, R., & Pagliuca, W. 1994. The Evolution of Grammar: Tense, Aspect and Modality
in the Languages of the World. Chicago IL: University of Chicago Press.
Cheshire, J. 1999. English negation from an interactional perspective. In Negation in the History
of English, I.T.-B. van Ostade, G. Tottie & W. van der Wurff (eds), 29–49. Berlin: Mouton de
Gruyter.
Comrie, B. 1976. Aspect. Cambridge: CUP.
Davydova, J. 2008. Preterite and present perfect in Irish English: Determinants of variation
[Working Papers in Multilingualism 84. Series B]. Hamburg: University of Hamburg.
Davydova, J. 2011. The Present Perfect in Non-Native Englishes [Topics in English Linguistics].
Berlin: Mouton de Gruyter.
Deterding, D. 2007. Singapore English. Edinburgh: EUP.
Elsness, J. 1997. The Perfect and Preterite in Contemporary and Earlier English. Berlin: Mouton
de Gruyter.
Fenn, P. 1987. A Semantic and Pragmatic Examination of the English Perfect. Tübingen: Günter
Narr.
Fitzpatrick, D. 1994. Oceans of Consolation: Personal Accounts of Irish Migration to Australia.
Cork: Cork University Press.
Gramley, S. & Pätzold, K.-M. 2004. A Survey of Modern English. London, New York NY:
Routledge.
Gupta, A.F. 1998. The situation of English in Singapore. In English in New Cultural Contexts.
Reflections from Singapore, J.A. Foley, T. Kandiah, Z. Bao, A.F. Gupta, L. Alsagoff, H.C. Lick,
L. Wee, I. Taib & W. Bokhorst-Heng (eds), 106–126. Singapore: OUP.
Hally, C. 1971. A hundred years of Irish missionary effort. The Furrow 22(6): 335–349.
Heine, B. & Kuteva, T. 2005. Language Contact and Grammatical Change. Cambridge: CUP.
Hickey, R. 2004. Development and diffusion of Irish English. In The Legacies of Colonial English.
Studies in Transported Dialects, R. Hickey (ed.), 82–121. Cambridge: CUP.
Hickey, R. 2010. Linguistic evaluation of earlier texts. In Varieties of English in Writing. The
Written Word as Linguistic Evidence [Varieties of English Around the World G 41], R.
Hickey, (ed.), 1–14. Amsterdam: John Benjamins.
Ho, M.-L. & Platt, J.T. 1993. Dynamics of a Contact Continuum. Oxford: Clarendon.
Housen, A. 2002. The development of Tense-Aspect in English as second language and the
variable influence of inherent aspect. In The L2 Acquisition of Tense-Aspect Morphology
[Language Acquisition and Language Disorders 27], R. Salaberry & Y. Shirai (eds), 155–197.
Amsterdam: John Benjamins.
Julia Davydova
Kwan-Terry, A. 1989. The specification of stage by a child learning English and Cantonese
simultaneously: A study of acquisitional processes. In Interlingual Processes, H. W. Dechert
& M. Raupach (eds), 33–48. Tübingen: Günter Narr.
Kusters, C.W. 2003. Linguistic Complexity: The Influence of Social Change on Verbal Inflection.
Utrecht: LOT.
Kusters, C.W. 2008. Complexity in linguistic theory, language learning and language change. In
Language Complexity. Typology, Contact, Change [Studies in Language Companion 94],
M. Miestamo, K. Sinnemäki, & F. Karlsson (eds), 3–21. Amsterdam: John Benjamins.
Labov, W. 1972. Sociolinguistic Patterns. Philadelphia PA: University of Pennsylvania Press.
Labov, W. 2007. Transmission and diffusion. Language 83: 344–387.
Li, C.N. &. Thompson, S.A. 1981. Mandarin Chinese. A Functional Reference Grammar. Berkley
CA: University of California Press.
Lick, H.C. & Alsagoff. L. 1998. English as the common language in the multicultural Singapore.
In English in New Cultural Contexts. Reflections from Singapore, J.A. Foley, T. Kandiah,
Z. Bao, A.F. Gupta, L. Alsagoff, H.C. Lick, L. Wee, I. Talib & W. Bokhorst-Heng (eds),
201–218. Singapore: OUP.
Lim, L. 2004. Singapore English: A Grammatical Description [Varieties of English around the
World G33]. Amsterdam: John Benjamins.
Lim, R. 2008. An Irish Tour of Singapore. Singapore: Two Trees.
Mazzon, G. 2004. A History of English Negation. Pearson: Longman.
Mesthrie, R. &. Bhatt, R.M. 2008. World Englishes. The Study of New Linguistic Varieties [Key
Topics in Sociolinguistics]. Cambridge: CUP.
Meyerhoff, M. 2003. Formal and cultural constraints on optional objects in Bislama. Language
Variation and Change14(3): 323–346.
Meyerhoff, M. 2009. Replication, transfer, and calquing: Using variation as a tool in the study of
language contact. Language Variation and Change 21: 297–317.
Miller, J. 2004. Problems for typology: Perfects and resultatives in spoken and non-standard
English and Russian. In Dialectology Meets Typology: Dialect Grammar from a Cross-
linguistic Perspective, B. Kortmann (ed.), 305–334. Berlin: Mouton de Gruyter.
Miller, K., Schrier, A., Boling, B., & Doyle, D. 2003. Irish Immigrants in the Land of Canaan:
Letters and Memoirs from Colonial and Revolutionary America, 1675–1815. Oxford: OUP.
Montgomery, M. 1989. Exploring the roots of Appalachian English. English World-Wide 10(2):
227–278.
Norman, J. 1988. Chinese [Cambridge Language Surveys]. Cambridge: CUP.
Odlin, T. 2009. Methods and inferences in the study of substrate influence. In Vernacular
Universals and Language Contacts: Evidence from Varieties of English and Beyond,
M. Filppula, J. Klemola & H. Paulasto (eds), 265–279. London: Routledge.
Odlin, T. & Alonso-Vázquez, C. 2006. Meanings in search of the perfect form: A look at
interlanguage verb phrases. Rivista di Psicolinguistica Applicata 6: 53–63.
Omar, A.H. & Subbiah, R. 1968. An Introduction to Malay Grammar. Kementerian: Dewan
Bahasa Dan Pustaka.
Ooi, V.B.Y. (ed.). 2001. Evolving Identities. The English Language in Singapore and Malaysia.
Singapore: Times Academic Press.
Pakir, A. 1991. The range and depth of English-knowing bilinguals in Singapore. World Englishes
10(2): 167–179.
Pietsch, L. 2007. The Irish English “After Perfect” in context: Borrowing and syntactic productiv-
ity [Working Papers in Multilingualism 82. Series B]. Hamburg: University of Hamburg.
Detecting historical continuity in a linguistically diverse urban area
Platt, J. & Weber, H. 1980. English in Singapore and Malaysia. Status. Features. Functions. Oxford:
OUP.
Platt, J., Weber, H. & Ho, M.L. 1983. Singapore and Malaysia [Varieties of English around the
World G4]. Amsterdam: John Benjamins.
Platt, J., Weber, H., & Ho, M.L. 1984. The New Englishes. London: Routledge and Kegan Paul.
Poplack, S. & Tagliamonte, S.A.. 1989. There’s no tense like the present: Verbal -s inflection in
early Black English. Language Variation and Change 1: 47–84.
Siemund, P. 2004. Substrate, superstrate and universals: Perfect constructions in Irish English.
In Dialectology Meets Typology: Dialect Grammar from a Cross-linguistic Perspective, B.
Kortmann (ed.), 401–434. Berlin: Mouton de Gruyter.
Schneider, E.W. 2000. Feature diffusion vs. contact effects in the evolution of New Englishes:
A typological case study of negation patterns. English World-Wide 21(2): 201–230.
Schneider, E.W. 2002. Investigating variation and change in written documents. In The Hand-
book of Language Variation and Change, J.K. Chambers, P. Trudgill, & N. Schilling-Estes
(eds), 67–96. Malden MA.: Blackwell.
Schneider, E.W. 2007. Postcolonial English. Varieties around the World. Cambridge: CUP.
Shirai, Y. &. Andersen, R.W. 1995. The acquisition of tense-aspect morphology: a prototype
account. Language 71(4): 743–762.
Slobin, D. 1996. From “though and language” to “thinking for speaking”. In Rethinking Linguistic
Relativity, J.J. Gumperz & S.C. Levinson (eds), 70–96. Cambridge: CUP.
Steever, S.B. 2005. The Tamil Auxiliary Verb System. London: Routledge.
Szmrecsanyi, B. & Kortmann, B. 2009. Between simplification and complexification: Non-
standard varieties of English around the world. In Language Complexity as a Variable
Concept, G. Sampson, D. Gil, & P. Trudgill (eds), 64–80. Oxford: OUP.
Sun, C. 2006. Chinese. A Linguistic Introduction. Cambridge: CUP.
Tagliamonte, S. 1996. Has it ever been “Perfect”? Uncovering the grammar of early Black English.
York Papers in Linguistics 17: 351–396.
Tagliamonte, S. 2000. The grammaticalization of the present perfect in English. In Pathways of
Change. Grammaticalization in English [Studies in Language Companion Series 53], O.
Fischer, A. Rosenbach & D. Stein (eds), 329–354. Amsterdam: John Benjamins.
Tagliamonte, S. 2006. Analysing Sociolinguistic Variation. Cambridge: CUP.
Tagliamonte, S. 2012. Variationist Sociolinguistics. Change, Observation, Interpretation. Malden
MA: Wiley-Blackwell.
Trudgill, P. 2001. Contact and simplification: Historical baggage and directionality in linguistic
change. Linguistic Typology 5: 371–373.
Van Herk, G. 2008. Letter perfect: The present perfect in early African American correspon-
dence. English World-Wide 29(1): 45–69.
Vertovec, S. 2007. Super-diversity and its implications. Ethnic and Racial Studies 30(6):
1024–1054.
Winford, D. 1993. Variability in the use of perfect have in Trinidadian English: A problem of
categorical and semantic mismatch. Language Variation and Change 5: 141–188.
Winford, D. 2009. The interplay of “Universals” and contact-induced change in the emergence
of New Englishes. In Vernacular Universals and Language Contacts: Evidence from Varieties
of English and Beyond, M. Filppula, J. Klemola, & H. Paulasto (eds), 204–227. London:
Routledge.
Four decades of study of synchronic
variation in varieties of Dutch. A sketch*
Frans Hinskens
Meertens Instituut (KNAW) and VU University, Amsterdam
1. Introduction
* This contribution evolved out of an invited talk on (what the organizers had circumscribed
as) ‘Language variation: synchronic perspectives’. The author would like to thank the organisers,
the audience as well as the editors for their valuable input. A special Thank You to reviewer
Jürgen Erfurt for his highly useful remarks and suggestions as well as to Roeland van Hout and
Johan Taeldeman for a considerable amount of significant input to the content of this chapter.
All remaining flaws are the author’s responsiblity.
Frans Hinskens
For reasons which can hardly be considered specific to the relevant develop-
ments in the Low Countries, the starting point for this sketch of the innovation of
the linguistic study of language variation is the appearance in 1966 of Labov’s The
Social Stratification of English in New York City. Even more important was
Weinreich, Labov & Herzog’s 1968 Empirical foundations for a theory of language
change. These studies triggered scholarly attention for areas of language variation
which had previously been largely neglected, as well as for new views on and ap-
proaches to older areas. As the perfect intermediary, Jo Daan, the then head of the
department of dialectology at the Meertens Instituut, introduced Labov’s work to
the Dutch community of linguists (for the first time in 1967 in a paper in a booklet
entitled Taalsociologie, edited by Daan and Weijnen).
Roughly until the second half of the 20th century, traditional dialectology
largely (though by no means exclusively) had a cultural historical orientation and
looked upon dialect variation from a diachronic perspective. Yet, some of the more
important older studies also had a sociological aspect in that next to the geograph-
ical dimensions language variation was also explicitly related with specific social
and/or cultural groups, notably in Van Ginneken’s two-volume Handboek der
Nederlandsche taal, ‘Handbook of the Dutch language’, from 1913 and 1914 and
Kloeke’s De Hollandsche expansie in de zestiende en zeventiende eeuw en haar weer-
spiegeling in de hedendaagsche Nederlandsche dialecten, ‘the Hollandic expansion
during the 16th and 17th century and its reflection in the present-day Dutch dia-
lects’ from 1927, to mention but two titles.
This paper addresses three of the most influential general tendencies in the
recent history of the linguistic study of language variation in the Dutch language
area: the social turn, the re-orientation on theoretical debates in linguistics and –
thirdly – the improvement and miniaturisation of recording equipment and the
rapid rise of digital research techniques. To each of these tendencies and their main
manifestations in the Low Countries a section will be dedicated. An additional
section will be devoted to the incipient widening of the research object to include
externally motivated variation. Unlike the preceding sections, which have the form
of succinct overviews, this last one goes into some depth in that a recent research
project is presented, along with a brief discussion of some preliminary findings.
Of course, the spectrum of sociolinguistic research is considerably wider than
the aspects that will be discussed below. No attention will here be paid to such areas
as language policy, language conflict, the study of creole languages, including lexi-
cally Dutch based creoles, register-related variation in connection with the distinc-
tion between oral – written language use, interaction and discourse, etc. All these
areas are well represented in recent linguistic research in the Low Countries. Yet,
they will not be discussed in this contribution and for this there are three weighty
reasons: first, a practical one, namely space limitations, second, a principled one:
not all research of synchronic variation is primarily sociolinguistic in orientation,
Four decades of study of synchronic variation in varieties of Dutch. A sketch
and, third, another principled one: in most investigations in the areas which are not
discussed here, (at least in the Dutch language area) variation is not the object per
se, if at all. Linguistic heterogeneity is what binds most studies in these areas to-
gether, but in most cases that perspective hardly overlaps with the generally ac-
cepted conception of language variation as an inherently quantitative phenomenon
and the theories and methods that have been developed in that domain.
Most of the studies resulting from the younger approaches have a relatively strong
synchronic orientation, zooming in on the modern state of the dialects (which are
no longer implicitly assumed to be homogeneous) and ongoing developments.
The social turn was largely brought about by the very outspoken growth in the
scholarly interest in urban dialects, which were considered off-limits by many rep-
resentatives of classical dialect geography (Schuchardt Brevier – Spitzer 1922) and
which tend to show social rather than geographical differentiation. The study of
urban dialects required methods from the social sciences (cf. Section 2.5 below)
and this is one of the key characteristics of the rise of sociolinguistics in the late
1960’s, along with basic assumptions such as:
1. language variation is structured heterogeneity;
2. intra-systemic variation is an instantaneous exposure of ongoing language
change. The main research questions concern the actuation, the embedding
(both in linguistic structure and in the speech community) and the evaluation
of innovations;
3. the principles and mechanisms that can be shown to guide language variation
and change today have always applied (Labov’s 1972 ‘uniformitarian princi-
ple’). They can lead to a deeper insight in processes of language change and
are, hence, crucial to historical linguistics.
The main exponent of assumption (1) is the so-called variable rule (Cedergren &
Sankoff 1974), which was originally conceived as a refinement the optional trans-
formational rule, cf. the classical format of the phonological rule
(1) A → B/C __ D
i.e. in the linguistic environment between C and D a phonological feature A is
changed into B. E.g., in the rule describing the process of syllable-final devoicing,
A is the feature specification [–son] which targets obstruents, B is [–vce], C is not
applicable, while D is ]σ, indicating a right syllable edge. In the case of optional
and variable phenomena, the actual conditioning (factors C_ and _D) are typi-
cally not categorical (making the rule apply either always or never) but rather
Frans Hinskens
probabilistic (making the chances for the rule to apply either smaller or bigger).
Moreover, the conditioning factors C_ and _D do not only concern the linguistic
context, but also the extra-linguistic context, such as the age or the social-econom-
ic background of the speaker and features of the interactional situation, such as the
relative degree of formality, and the like.
Evidently, this type of research is usually quantitative in method, reflecting the
fact that most processes of linguistic innovation are gradient.
that urban dialects and vernaculars can be important sources of change in re-
gional varieties is corroborated in Rogier’s (1994) and Tops’ (2009) studies of the
presently ongoing diffusion of the uvular realisation of /r/ in Flanders.
theoretical linguistic discussion on the basis of the findings is either ‘thin’ or alto-
gether lacking.
The independent variables typically include dimensions such as age, sex/gen-
der and socio-economic background factors such as educational level, income
level and the dialect background of the speakers. Sometimes the type of linguistic
output has also been systematically varied; in those cases the design usually
provides both (experimentally or otherwise) elicited and conversational speech
material.
Linguistic independent variables are not always part of the design. If they are,
then the choice is usually determined by the nature of the phenomenon/-a at issue;
in the case of phonetic or phonological variables, attention is often paid to the
segmental left-hand and right-hand environment. Prosodic or metrical factors are
not commonly included.
In the research sketched so far, the language varieties and their dynamics were
central. However, as has been the case elsewhere, in the Dutch language area espe-
cially from the sociolinguistic perspective attention has also been paid to other
aspects of language variation. An important one was and is social psychology; in
this context studies have zoomed in on the evaluation of language varieties and
linguistic innovations. Language attitudes were a weighty object of study in the
1980’s (Deprez & De Schutter 1980, Knops 1982, Van Bezooijen 1988, Van Hout &
Vallen 1984; Van Hout & Knops 1988) and they made a comeback in the first de-
cade of the 21st century; cf. Grondelaers (forthcoming).
A related area is often referred to as perceptual dialectology. Many of the ques-
tions and methods which feature in this area have been inspired by studies by
Dennis Preston (1989) and his students who, in turn, often build on Weijnen’s
(1946) pijltjesmethode, lit. ‘little arrow method’; speakers of a local dialect are given
a map of the region and asked to draw arrows to surrounding villages and towns
where dialects are spoken which they feel are similar to theirs. The areas on the
map which are not covered with arrows are dialect boundaries in the speakers’
awareness. On the basis of data from the first systematic investigation of the per-
ception of dialect distance by non-linguists in the Netherlands (by Willems around
1886)2, Goeman (1989) constructed a historical arrow map.
These and similar endeavours focus on the relative degree of alleged similarity
of the surrounding dialects with one’s own dialect and –in some cases- the degree
of intelligibility of the surrounding dialects; cf. Gooskens et al. (forthcoming).
In both lines of research, the object is formed by ‘folk linguistic’ concepts and
notions, as well as the potential impact they may have on language variation and
change.
From the 1980’s onwards many reports appeared on smaller or larger surveys of
reported behaviour, more in particular code choice, in domains of language use à
la Fishman (1965). An important publication is Gorter et al. (1984) on the position
of Frisian in the Dutch province of Friesland (Fryslân). Some studies include
surveys of the aptitude and (reported) use of dialect versus standard
language (e.g. Münstermann & Van Hout 1986; Cucchiarini & Hinskens 1989).
Jongenburger & Goeman (2009) discuss the findings from a large, nationwide
survey on dialect use.
Most of these studies also concern opinions regarding the appropriateness of
dialect use in specific interactional situations; conceptually, these studies are thus
closely related to social psychological issues. M. Jansen (2010) is one of the studies
which very clearly show the relevance of this type of data to the study of func-
tional dialect loss (or dialect shift).
Remarkably, social networks have hardly ever been central in the study lan-
guage variation in the Dutch language area.
One of the oldest areas of sociolinguistic concern regarding the societal relevance
of language variation is education. In the late 1960s and early 1970s, the sociolin-
guistic debates centred on the opposition between the ‘deficit view’ and ‘difference
view’ on non-standard varieties. In the Dutch society the ‘deficit view’ (which is
associated with the work of the British sociologist Bernstein – 1971), specifically
his distinction between ‘elaborated’ and ‘restricted’ codes) resulted in the develop-
ment and implementation of language compensation programs, such as the
Grandia-project in Rotterdam (advocated by the local councillors of PvdA, the
social-democratic party), and the Innovatie-project in Amsterdam (advocated by
the local councillors of CPN, the communist party). These projects were support-
ed by psychologists such as Kohnstamm and Gestel. Variationist research into the
Bernsteinian code concept was done by J. Van den Broeck (1977) in the town of
Maaseik in the Belgian province of Limburg.
Four decades of study of synchronic variation in varieties of Dutch. A sketch
RBrul) are not popular in the Netherlands; most researchers use more general
software such as SPSS and – to a lesser degree – R. The SPSS or R equivalent of
variable rule analysis in the original, strict sense (multivariate analyses with nom-
inal variables) is logistic regression.
The synchronic emulation of diachrony or ‘real time’ is the ‘apparent time’
method, in which language use of representatives of different age groups is system-
atically compared (a technique which is comparable to cross-sectional methods in
studies of language acquisition). Thus language change is studied through age-
related patternings in language variation.
Only little real time research has been done in the Dutch language area – some
of the exceptions being a range of studies presented in Gerritsen (1979),
Taeldeman (1980) – all on the basis of the comparison of older and younger ques-
tionnaire data. For the Dutch language area, Van de Velde (1996) is the first study
that was based on recent and historical recordings (i.e. of radio broadcasts), albeit
standard language production. No replication studies (à la LANCHART in Copen-
hagen – Gregersen 2009) have been conducted in the Dutch language area yet.
In social-psychological studies of language variation, techniques such as
matched guise experiments and Likert scale items have become part and parcel of
the methodology. In perceptual dialectology, variants of the ‘pijltjesmethode’
(cf. Section 2.2 above) are in use and – recently – web experiments (Heeringa &
Hinskens 2011).
In the first decades of the 20th century, some of the more important scholars of
dialectology were oriented towards – then young – structuralism in linguistics. In
the hey-days of structuralism, Dutch dialectology was engaged in fruitful
interactions with theoretical work on language structure. The best-known repre-
sentative of this group is probably the Nijmegen scholar Jacques van Ginneken
(1877–1946).
A period of relative theoretical abstinence followed, but from the late 1970’s
onwards the contact with theoretical linguistics was intensified. Two approaches
can be discerned:
a. dialect data are confronted with generative theories
b. findings concerning one or more Dutch dialects are checked with comparable
descriptions of a range of language systems from all over the world, in other
words: with the findings from linguistic typology.
Four decades of study of synchronic variation in varieties of Dutch. A sketch
The following subsections sketch some of the main work for the sound compo-
nents and grammar sensu stricto, indicating how theoretical orientations vary both
in direction and in importance. The fact that the discussion in the remainder of
this section will be limited to these two broad areas also reflects the fact that so far
in the Dutch language area the theoretically inspired study of linguistic variation
has almost entirely been confined to phonology and (morpho-)syntax.
The third main general trend in the study of synchronic variation concerns the
hardware which is used for fieldwork or other types of data collection as well as for
various steps in the analysis of the material thus gathered.
In the past two decades audio (as well as video) recording equipment has be-
come much smaller in size (and thus less ‘intrusive’), while the technical quality of
Frans Hinskens
the recordings has considerably improved and the prices have come down. This
makes collecting digital material in fieldwork not only more affordable, but also
more feasible in practical respects, while chances have grown that the recorded
material is relatively natural.
It is becoming good practice to store recordings in sound archives, which are
sometimes part of large (sometimes distributed, mutually connected) databases.
For the labelling of metadata, standards (such as IMDI – cf. http://www.mpi.nl/
IMDI/) are being developed.
User-friendly tools (such as ELAN – freeware; cf. http://www.lat-mpi.eu/tools/
elan/) for the transcription, annotation and coding of recorded data and software
(such as PRAAT – equally freeware) for phonetic analyses are being applied more
and more.
Geo-browsers and applications such as Google Earth and Google Maps bring
automatic cartography within the reach of many.
These and similar developments facilitate technologically advanced research
on a scale that was hitherto almost inconceivable. The Royal Netherlands Academy
of Arts and Sciences (KNAW) has made available special budgets (called ‘Compu-
tational humanities’ – cf. e-humanities, digital humanities and related labels) to
stimulate the exploitation of these new technologies in research in the humanities.
With regard to the linguistic effects of language contact, a general distinction can
be drawn between interference (and/or transfer) versus borrowing, corresponding
to individual and incidental versus speech community-wide, pervasive and struc-
tural, respectively. In general and a fortiori for the Dutch language area, it can be
established that hardly any research has been carried out into the structural effects
of language contact, specifically borrowing, as a source of synchronic variation.
With few exceptions, such as C. Smits’ (1996) study of dialects of Dutch spo-
ken in Iowa, hardly any work has been done on varieties of Dutch in Sprachinseln
(enclaves of ‘roofless’ dialects) elsewhere in the world, potential koineisation or the
structural changes they undergo as a result of contact with varieties of the domi-
nant language.
Relatively much research has zoomed in on the acquisition of (mainly
Netherlandic) Dutch as a second language; some of this research is of a pronounced
applied linguistic nature. So far, one of the most important themes was the relative
success in the acquisition of Dutch (the standard variety, which is not necessarily
the variety spoken in the immediate environment) as a second language by migrant
Four decades of study of synchronic variation in varieties of Dutch. A sketch
workers and their families, refugees and similar groups. In this connection atten-
tion was paid to the effects of linguistic properties of the first language (sometimes
on the basis of contrastive analysis) and speaker-specific parameters such as
– the age of onset;
– language choice in specific domains;
– the type of motivation;
– the degree of exposure to/integration into the surrounding Dutch society, and
– educational background.
Another important theme was language loss or language attrition in the individu-
al (e.g. Jaspaert & Kroon 1987, 1988; El Aissati 1996). Typically, the research popu-
lation was identical to that of the work on Dutch as a second language and the
attrited language was usually the original language of the migrant group at issue.
A third area of research is code switching and mixing and examples of relevant
studies are Nortier (1989) on Arabic – Dutch code switching among Moroccans in
Utrecht and Treffers’ (1991) study of the mixing of the local dialect varieties of
Dutch and French in Brussels. Both studies draw up the findings for a range
of linguistic and extra-linguistic factors in quantitative terms.
In his dissertation, Backus (1996) paid attention to properties of bilingual
speech of Turkish immigrants in the Netherlands, focussing on language contact
phenomena, rather than, say, on the degree of success in the acquisition of Dutch.
Like the work carried out in the three areas sketched so far in this section, this is
not research of synchronic variation either. However, Backus (2001) was the first
Dutch linguist who pointed at the potential importance of ethnicity as a sociolin-
guistic factor in language variation.
The first research project which pays systematic attention to language contact
(substrate effects) and universal traits of second language (L2) acquisition as
sources of synchronic variation in indigenous varieties of Dutch is ‘The roots of
ethnolects. An experimental comparative study’.3 This project zooms in on
synchronic variation, not only in the speech of bilinguals, not only in the speech of
members of the ethnic minority groups involved, but also in monolingual speak-
ers of Dutch. Ultimately, the data are analysed quantitatively.
3. Conceived and supervised by Pieter Muysken (Nijmegen) and the present author. The oth-
er researchers involved in this project, which was financed 2005–2010 by the Netherlands
Organisation for Scientific Research (NWO), the Meertens Instituut, and Radboud Universiteit
Nijmegen, were/are Esther van Krieken, Wouter Kusters, Hanke van Buren, Arien van
Wijngaarden, Linda van Meel and Roeland van Hout.
Frans Hinskens
Table 1. Three relevant demographic facts about the Dutch cities of Amsterdam
and Nijmegen
Table 2. The speakers design for the Roots of ethnolects project
12 years old 18 to 20
Amsterdam Moroccan 6 6
Turkish 6 6
‘white’ Dutch, strong inter-ethnic ties 6 6
‘white’ Dutch, weak inter-ethnic ties 6 6
Nijmegen Moroccan 6 6
Turkish 6 6
‘white’ Dutch, strong inter-ethnic ties 6 6
‘white’ Dutch, weak inter-ethnic ties 6 6
Four decades of study of synchronic variation in varieties of Dutch. A sketch
Except for the ‘white’ Dutch boys who have but few friends from other ethnic
groups (‘weak inter-ethnic ties’), three recordings are being made of every single
speaker; the recordings concern conversations, one with a speaker whose main
background is Moroccan, one with a speaker with a Turkish background and one
with a ‘white’ Dutch boy with friends from other ethnic groups.
Exploration of the conversational speech has brought to light a range of fea-
tures of Moroccan and Turkish Dutch. They concern (morpho-)syntax (e.g. the
marking of grammatical gender in various environments; pronominal adverbs
containing the ‘floating’ particle er or its unreduced equivalents hier and daar) as
well phonology/phonetics (examples of which will be discussed below).
Some of the features are rooted in the surrounding local/regional Dutch dia-
lect (such as the use of 3rd plural dative hun, ‘them’, as a subject pronoun or the
dialect-specific realisation of the voiced velar fricative /γ/) while some are exotic
– in the sense that they do not occur in any indigenous variety of Dutch. Examples
of the latter category include SVO order in embedded clauses.
The Moroccan and Turkish varieties of Dutch investigated are not only marked
by more linguistic variables, also the number of different variants of prior-existing
linguistic variables is sometimes higher than in indigenous varieties. An example
concerns the segment /γ/ which has two main variants, one of which is velar
(typical of the north-western varieties of Dutch) and the other one is palato-velar
in articulation (typical of the southern and south-eastern varieties of Dutch).
Speakers with a Turkish and – especially – Moroccan background add a uvular or
guttural variant to the pool of variants.
The remainder of this section will zoom in on some main findings concerning
variation in the realisation of /z/. Speakers with a Turkish and Moroccan background
have been found to use the standard variant [z] along with the indigenous non-
standard voiceless variant [s]. In the speech of these groups several exotic variants of
/z/ occur as well, one of the most salient ones being overlong [zù] (needless to add
that there is no phonemic length contrast in Dutch consonants). Another exotic vari-
ant of /z/ is (what has been impressionistically labelled) ‘sharp’ (cf. below) and a third
remarkable fact about the realisation of /z/ following an obstruent (as in ik zie, ‘I see’)
by speakers with a Moroccan or Turkish background is the regressive voice assimila-
tion (or ‘sandhi voicing’: i[gz]ie), which in indigenous standard and non-standard
Dutch never occurs in obstruent clusters in which the second consonant is a fricative.
This abundance of variants constitutes ‘super-diversity’ in micro-cosmos.
In the case of the ‘sharp’ realisation of /z/, especially of the voiced realisation
[z], the corona, roughly the tip and the foremost part of the tongue blade, is high-
er and almost against the upper front teeth, such that the space for resonance is
smaller. The resulting dental realisation of /z/ is more constricted. It is a ‘hisser’
with relatively much frication.
Frans Hinskens
Amsterdam
0,40
0,35
Mean_sharpness of voiced z
0,30
0,25 D_interlocutor
0,20 M_interlocutor
0,15 T_interlocutor
0,10
0,05
0,00
D_speaker M_speaker T_speaker
Nijmegen
0,40
0,35
Mean_sharpness of voiced z
0,30
0,25 D_interlocutor
0,20 M_interlocutor
T_interlocutor
0,15
0,10
0,05
0,00
D_speaker M_speaker T_speaker
Figures 1 and 2. The proportion of ‘sharp’ (dental) realisations of [z] in 18 to 20 year old
Moroccan, Turkish and white Dutch speakers of Dutch in Amsterdam and Nijmegen
Four decades of study of synchronic variation in varieties of Dutch. A sketch
The ‘sharp’ [z] does not seem to have spread to the ‘white’ Dutch speakers (as yet
– ?). There is no statistically significant difference between the two cities. In the
data from the Amsterdam speakers there is no statistically significant effect of the
language background of the speakers, nor for that of the interlocutors; the interac-
tion between the variables is not significant either. In the data from the Nijmegen
speakers, however, the language background of the interlocutors does have a sig-
nificant effect (F2,12 = 4,233, p = .041), while that of the speakers does not. The
interaction effect between both variables is significant (F4,12 = 5,958, p = .007),
with a high peek in the curve for Moroccans talking to Moroccans – so the ‘sharp’
(dental) realisation of Dutch /z/ may be a Moroccan in-group phenomenon.
It thus appears that for Moroccan speakers of Dutch in Nijmegen, /z/ is subject
to stylistic manipulation, i.e. the sharpness of [z] varies according to the situation,
more in particular the language background of the interlocutor (cf. Bell’s 1984
audience design). The fact that these speakers systematically manipulate the de-
gree of [z] sharpness implies that they are aware of the Dutch norms and capable
to behave accordingly, so at least in this respect Nijmegen Moroccan Dutch cannot
be considered as a learners’ variety.
5.2 Sizing up
References
Backus, A. 1996. Two in One. Bilingual Speech of Turkish Immigrants in the Netherlands. PhD
dissertation, Tilburg University.
Backus, A. 2001. Etniciteit als sociolinguïstische factor. Levende Talen Tijdschrift 3(1): 3–10.
Bakkes, P. 1996. Variatie en verandering in het Montforts: Taalstructurele en sociolinguïstische
aspecten van een veranderend dorpsdialect. PhD dissertation, Nijmegen University.
Barbiers, S., Bennis, H.J., de Vogelaar, G., Devos, M. & van der Ham, M.H. (eds). 2005[2008].
SAND: Syntactische Atlas van de Nederlandse Dialecten/Syntactic Atlas of the Dutch
Dialects, Vol. 1 & 2. Amsterdam: Amsterdam University Press.
Belemans, R. 2009. Taal of tongval? De gespleten Limburgse kus, oraal erfgoed en taalpolitiek.
PhD dissertation, Leuven University.
Bell, A. 1984. Language style as audience design, Language in Society 13(2): 145–204.
Bloemhoff, Henk. 1991. Fonologie en morfologie van het Stellingwerfs. Een toetsing van de
natuurlijke generatieve fonologie. PhD dissertation, Groningen University.
Bernstein, B. 1971. Class, Codes and Control: Theoretical Studies Towards a Sociology of
Language. London: Routledge & Kegan Paul.
Boves, T. 1992. Speech Accomodation in Co-operative and Competitive Conversations. PhD
dissertation, Nijmegen University.
Brouwer, D. 1989. Gender Variation in Dutch; A Sociolinguistic Study of Amsterdam Speech.
PhD dissertation, Nijmegen University.
Bybee, J. 2001. Phonology and Language Use. Cambridge: CUP.
Cedergren, H. & Sankoff, D. 1974. Variable rules: Performance as a statistical reflection of
competence. Language 50: 333–355.
Cornips, L. 1996. Syntactische variatie in het Algemeen Nederlands van Heerlen. PhD
dissertation, University of Amsterdam. Amsterdam: IFOTT.
Cucchiarini, C. & Hinskens F. 1989. De plaats van het dialect in het Land van Axel, Mededelin-
gen van de Nijmeegse Centrale voor Dialect- en Naamkunde 20: 1–31.
Croft, W. 2000. Explaining Langage Change. An Evolutionary Approach. Harlow: Longman.
Daan, J. 1967. Communicatie-taalkunde. In Taalsociologie, J. Daan & A. Weijnen (eds),
Amsterdam: Bijdragen en Mededelingen der Dialectencommissie van de Koninklijke
Nederlandse Akademie van Wetenschappen te Amsterdam.
Daan, J. 1971. Verschuiven van isoglossen. Taal en Tongval 23: 77–79.
Daan, J. 1987. Ik was te bissie... Nederlanders en hun taal in de Verenigde Staten. Zutphen:
De Walburg Pers.
Daan, J. & Weijnen, A. 1967. Taalsociologie. Amsterdam: Noord-Hollandsche Uitgevers-
maatschappij.
Den Besten, H. & Hinskens, F. 2005. Diversificatie van het Nederlands door taalcontact.
Nederlandse Taalkunde 10: 283–309.
De Rooij, J. 1965. Als-Of-Dat. Enkele conjuncties in ABN, dialect en Fries PhD dissertation,
Nijmegen University.
De Rooij, J. 1967. Van hebben naar zijn. Het gebruik van hebben en zijn in de voltooide tijden
(actief) van zijn, gaan, vergeten en verliezen in stadaardtaal, ouder Nederlands en dialect.
Amsterdam: Publikaties van het P.J. Meertens Instituut.
De Rooij, J. 1988. Lummel dat je bent, Lummel dat je bent. Nederlandse Taal- en Letterkunde
83: 108–121.
Four decades of study of synchronic variation in varieties of Dutch. A sketch
De Tier, V., van Hout, R., & Dehaes, G. (eds). 2010. Taalvariatiebeleid! Groesbeek: Stichting
Nederlandse Dialecten.
De Vink, L. 2004. Dialect en dialectverandering in Katwijk aan Zee. PhD dissertation, Leiden
University.
De Vogelaer, G. 2008. De Nederlandse en Friese subjectmarkeerders: geografie, typologie en
diachronie. Gent: Koninklijke Academie voor Nederlandse Taal- en Letterkunde.
Deprez, K. & De Schutter, G. 1980. Honderd Antwerpenaren en honderd Rotterdammers over
dertien Nederlandse taalvariëteiten. Een attitude-onderzoek. Leuvense Bijdragen 69:
167–256.
Docherty, G. & Foulkes, P. 2000. Speaker, speech, and knowledge of sounds. In Phonological
Knowledge: Conceptual and Empirical Issues, N. Burton-Roberts, P. Carr & G. Docherty
(eds), 105–129. Oxford: OUP.
Driessen, G. 2005. In Dutch? Usage of Dutch regional languages and dialects. Language, Culture
and Curriculum 18(3): 271–285.
Elias, M. 1980. Enige aspekten van het Haagse stadsdialekt. In Sociolinguïstische studies, 1:
Bijdragen uit het Nederlandse taalgebied, G. Geerts & A. Hagen (eds), 80–96. Groningen:
Wolters-Noordhoff.
El Aissati, A. 1996. Language Loss Among Native Speakers of Moroccan Arabic in the
Netherlands. PhD dissertation, Nijmegen University.
Fishman, J. 1965. Who speaks what language to whom and when? La Linguistique 2: 67–88.
Fournier, R. 2008. Perception of the Tone Contrast in East Limburgian Dialects. PhD dissertation,
Nijmegen University.
Gerritsen, M. (ed.). 1979. Taalverandering in Nederlandse dialekten: honderd jaar
dialektvragenlijsten: 1879–1979. Muiderberg: Coutinho.
Gerritsen, M. 1980. Een kwantitatief onderzoek naar sekseverschillen in het Amsterdams. In
Sociolinguïstische studies, 1: bijdragen uit het Nederlandse taalgebied, G. Geerts A. & Hagen
(eds), 154–179. Groningen: Wolters-Noordhoff.
Giesbers, C. 2008. Dialecten op de grens van twee talen: Een dialectologisch en sociolinguïstisch
onderzoek in het Kleverlands dialectgebied. Phd dissertation, Nijmegen University.
Giesbers, H. 1989. Code-switching tussen dialect en standaardtaal. PhD dissertation, Nijmegen
University.
Goeman, A. 1989. Dialectes et jugements subjectifs des locuteurs. Quelques remarques de
méthode à propos d’une controverse. In Espaces Romans. Études de dialectologie et de
géolinguistique offertes à Gaston Tuaillon, Vol. II, 532–544. Grenoble: Ellug.
Goeman, A. 1999. T-deletie in Nederlandse dialecten. Kwantitatieve analyse van structurele,
ruimtelijke en temporele variatie. Amsterdam : LOT.
Gooskens, C., van Bezooijen, R., & Nerbonne, J. Forthcoming. Perception of geographically
conditioned linguistic variation. In Language and Space: Dutch, F. Hinskens & J. Taeldeman
(eds). Berlin: Mouton de Gruyter.
Goossens, J. 1992. Dialecten in het centrale zuidnederlandse stedennetwerk. In Stadsdialecten,
A. Hagen, J. Goossens, R. & van Hout (eds). Special issue of Taal en Tongval 44: 29–47.
Gorter, D., Jelsma, G. & van der Plank, P. 1984. Taal yn Fryslan (Undersyk nei taalgedrach en
taalhalding yn Fryslan). Leeuwarden: Fryske Akademy.
Gregersen, F. (ed.). 2009. Language Change in Real Time: Evidence from the Danish Laboratory.
Acta Linguistica Hafniensia 41.
Grondelaers, S. forthcoming. Attitude measurements in the Low Countries. In Language and
Space: Dutch. F. Hinskens & J. Taeldeman (eds). Berlin: Mouton de Gruyter.
Frans Hinskens
Gussenhoven, C. Forthcoming. Complex intonation near the tonal isogloss in the Netherlands.
In Prosodic Typology, II: The Phonology of Intonation and Phrasing. Sun-Ah Jun (eds).
Oxford: OUP. (Sound files).
Haegeman, L. 1992. Theory and Description in Generative Syntax: A Case Study in West Flemish.
Cambridge: CUP.
Haeseryn, W. 1990. Syntactische normen in het Nederlands. Een empirisch onderzoek naar
volgordevariatie in de werkwoordelijke eindgroep. PhD dissertation, Nijmegen University.
Haslinger, I. 2007. The Syntactic Location of Events: Aspects of Verbal Complementation in
Dutch. PhD dissertation, Tilburg University.
Haugen, E. 1950. The analysis of linguistic borrowing. Language 26(2): 210–231.
Heeringa, W. 2004. Measuring Dialect Pronunciation Differences Using Levenshtein Distance.
PhD dissertation, Groningen University.
Heeringa, W. & Hinskens, F. 2011. The measurement of Dutch dialect change: Lexicon versus
morphology versus sound components. The Formation of Regiolects in the Low Countries.
Special issue of Taal en Tongval 63(1): 79–98.
Heeringa, W. & Nerbonne, J. forthcoming. Dialectometry. In Language and Space: Dutch.
F. Hinskens & J. Taeldeman (eds). Berlin: Mouton de Gruyter.
Hermans, B. 1994. The Composite Nature of Accent. With Case Studies of the Limburgian and
Serbo-Croatian Pitch Accent. PhD dissertation, VU University Amsterdam.
Hinskens, F. 1986. The selection of linguistic variables in empirical research on variation and
change in dialects. In Language Attrition in Progress, B. Weltens, K. de Bot & T. van Els
(eds), 53–74. Dordrecht: Foris.
Hinskens, F. 1992. Dialect Levelling in Limburg. Structural and Sociolinguistic Aspects. PhD
dissertation, Nijmegen University.
Hinskens, F. 2004. Nieuwe regenboogkleuren. Jonge typen niet-standaardtaal en hun taalkundig
belang. Amsterdam: Serie Oraties gehouden aan de VU Amsterdam. <http://www.meertens.
nl/medewerkers/frans.hinskens/rede.pdf>
Hinskens, F. 2011a. Lexicon, phonology and phonetics. Or: Rule-based and usage-based
approaches to phonological variation. In Linguistic Universals and Language Variation,
P. Siemund. (ed.), 416–456. Berlin: Mouton de Gruyter.
Hinskens, F. 2011b. Koineization in the present-day Dutch dialect landscape: Postvocalic /r/ and
more. In The Formation of Regiolects in the Low Countries, W. Heeringa & G. de Vogelaer
(eds). Special issue of Taal en Tongval 63(1): 99–126.
Hinskens, F. & Muysken, P. 2007. The talk of the town: Languages in Amsterdam 1507–2007. In
Im Dickicht der Städte, I: Sprache und Semiotik, R. Franceschini (ed.). Special issue of
Zeitschrift für Literaturwissenschaft und Linguistik (LiLi) 148: 7–23.
Hinskens, F. & van Oostendorp, M. 2009. Sources of phonological variation in a large database
for Dutch dialects. In Language Variation – European Perspectives II. Selected papers from
the Fourth International Conference on Language Variation in Europe (ICLaVE 4) [Studies
in Language Variation 5], S. Tsiplakou, M. Karyolemou, & P. Pavlou (eds), 103–118.
Amsterdam: John Benjamins.
Hoppenbrouwers, C. 1982. Language Change: A Study of Phonemic and Analogical Change
with Particular Reference to S.E. Dutch Dialects. PhD dissertation, Groningen University.
Hoppenbrouwers, C. 1983. Het genus in een Brabants regiolect. Tabu 13(2): 1–25.
Hoppenbrouwers, C. 1990. Het regiolect: Van dialect tot Algemeen Nederlands. Muiderberg:
Coutinho.
Four decades of study of synchronic variation in varieties of Dutch. A sketch
Howell, T. 2006. Immigration and koineization: The formation of Early Modern Dutch urban
vernaculars. Transactions of the Philological Society 104(2): 207–227.
Jansen, F. 1981. Syntaktische konstrukties in gesproken taal. Amsterdam: Huis aan de Drie
Grachten.
Jansen, M. 2010. Language Change on the Dutch Frisian Island of Ameland. Linguistic and
Sociolinguistic Findings. PhD dissertation, VU University, Amsterdam.
Jaspaert, K. & Kroon, S. 1987. The relationship between global proficiency tests and language
loss. In Linguistics in the Netherlands 1987, F. Beukema & P. Coopmans (eds), 91–100.
Assen: Van Gorcum.
Jaspaert, K. & Kroon, S. 1988. The relationship between language attitudes and language choice.
In Language Attitudes in the Dutch Language Area, R. van Hout & U. Knops (eds), 157–171.
Dordrecht: Foris.
Johnson, K. 1997. Speech perception without speaker normalization: An exemplar model. In
Talker Variability in Speech Processing, K. Johnson & J. Mullennix (eds), 145–165. San Diego
CA: Academic Press.
Jongenburger, W. & Goeman, A. 2009. Dimensions and determinants of dialects in the
Netherlands at the individual and regional levels at the end of the twentieth century. Inter-
national Journal of the Sociology of Language 196/197: 31–72.
Kloeke, G. 1927. De Hollandsche expansie in de zestiende en zeventiende eeuw en haar
weerspiegeling in de hedendaagsche Nederlandsche dialecten. The Hague: Martinus Nijhoff.
Knops, U. 1982. Attitudes van Vlamingen tegenover de Nederlandse standaardtaal. PhD
dissertation, Nijmegen University.
Kraaykamp, G. 2005. Dialect en sociale ongelijkheid. Een empirische studie naar de sociaal-
economische gevolgen van het spreken van dialect in de jeugd. Pedagogische Studiën
82: 390–403.
Kremer, L. 1979. Grenzmundarten und Mundartgrenzen. Untersuchungen zur wortgeographis-
chen funktion der Staatsgrenze im ostniederländisch-westfälischen Grenzgebiet. Köln: Böhlau
Verlag.
Kremer, L. Forthcoming. Varieties of Dutch/Dutch as a minority language in Germany. In
Language and Space: Dutch, F. Hinskens & J. Taeldeman (eds). Berlin: Mouton de Gruyter.
Kruijsen, J. 1995. Geografische patronen in taalcontact. Romaans leengoed in de Limburgse
dialecten van Haspengouw. PhD dissertation, Nijmegen University.
Labov, W. 1966. The Social Stratification of English in New York City. Washington DC: Center for
Applied Linguistics.
Labov, W. 1972. Sociolinguistic Patterns. Philadelphia PA: University of Pennsylvania Press.
Labov, W. 1977. The Unity of Sociolinguistics [University of Trier: Linguistic Agency, Papers, Ser.
B. No. 22].
Labov, W. 2001. Principles of Linguistic Change, Vol. 2: Social Factors. Oxford: Blackwell.
Münstermann, H. 1989. Dialect loss in Maastricht: Attitudes, functions and structures. In
Language and Intergroup Relations in Flanders and in The Netherlands, K. Deprez (ed.),
99–128. Dordrecht: Foris.
Münstermann, H. & van Hout, R. 1986. Taalattitudes contra geschiktheid en gebruik. In Werk-
in-uitvoering; Momentopname van de sociolinguïstiek in Nederland en Vlaanderen, J. Creten,
G. Geerts & K. Jaspaert (eds), 235–249. Leuven: Acco.
Neuckermans, A. 2008. Negatie in de Vlaamse dialecten volgens de gegevens van de Syntac-
tische Atlas van de Nederlandse Dialecten (SAND). PhD dissertation, Ghent University.
Frans Hinskens
Nijen Twilhaar, J. 1990. Generatieve fonologie en de studie van Oostnederlandse dialecten. PhD
dissertation, Utrecht University.
Nortier, J. 1989. Dutch and Moroccan Arabic among Moroccans in the Netherlands. PhD
dissertation, University of Amsterdam.
Nortier, J. 2001. Murks en straattaal. Vriendschap en taalgebruik onder jongeren. Amsterdam:
Prometheus.
Pietersen, L. 1969. De Friezen en hun taal. Een onderzoek naar de lees- een spreekgewoonten in
Friesland en naar de houding ten aanzien van het Fries. Drachten: Laverman.
Preston, D. 1989. Perceptual Dialectology: Nonlinguists’ Views of Areal Linguistics. Dordrecht:
Foris.
Rindler Schjerve, R. 1987. Sprachkontakt auf Sardinien. Soziolinguistische Untersuchungen des
Sprachwechsels im ländlichen Bereich. Tübingen: Narr.
Rogier, D. 1994. De verspreiding van een social hooggewaardeerd taalkenmerk: De huig-R rond
Gent. In R – zes visies op een kameleon, G. de Schutter et al. (eds). Special issue of Taal en
Tongval 46: 43–53.
Ryckeboer, H. 2000. The role of political borders in the millenial retreat of Dutch (Flemish) in
the north of France. In Dialect Convergence and Divergence across European Borders, J.L.
Kallen, F. Hinskens & J. Taeldeman (eds). International Journal of the Sociology of Language
145: 79–108.
Ryckeboer, H. Forthcoming. A West Flemish dialect as a minority language in the north of
France. In Language and Space: Dutch, F. Hinskens & J. Taeldeman (eds). Berlin: Mouton de
Gruyter.
Rys, K. 2007. Dialect as a Second Language: Linguistic and Non-linguistic Factors in Secondary
Dialect Acquisition by Children and Adolescents. PhD dissertation, Ghent University.
Schatz, H. 1986. Plat Amsterdams in its Social Context: A Sociolinguistic Study of the Dialect of
Amsterdam. Amsterdam: Publicaties van het P.J. Meertens-Instituut.
Schouten, M., Crielaard, T. & van Dijk, S. 1999. De open klinkers in het Utrechts en het Amster-
dams. Taal en Tongval 50(2): 101–115.
Smits, C. 1996. Disintegration of Inflection. The Case of Iowa Dutch. PhD dissertation, VU
University Amsterdam.
Smits,T. 2011. Strukturwandel in Grenzdialekten: Die Konsolidierung der niederländisch-
deutschen Staatsgrenze als Dialektgrenze. Wiesbaden: Steiner.
Spitzer, L. (ed.). 1922. H. Schuchardt-Brevie: Ein Vademekum der allgemeinen Sprachwissen-
schaft. Halle: Max Niemeyer.
Taeldeman, J. 1985. De klankstruktuur van het Gentse dialekt: Een synchrone beschrijving en een
historische en geografische situering. Gent: Rijksuniversiteit te Gent.
Taeldeman, J. 1980. Pluralisation in the Flemish (and the Brabantish) dialects. A diatopical
survey and a sociolinguistic close-up of morphological change. In Linguistics in the
Netherlands 1980, S. Daalder & M. Gerritsen (eds), 29–47. Amsterdam: North Holland.
Taeldeman, J. 2005. The influence of urban centres oj the spatial diffusion of dialect phenomena.
In Dialect change. Convergence and Divergence in European Languages, P. Auer, F. Hinskens
& P. Kerswill (eds), 263–283. Cambridge: CUP.
Tops, E. 2009. Variatie en verandering van de /r/ in Vlaanderen. Brussels: VUB Press.
Treffers, J. 1991. French-Dutch Language Mixture In Brussels. PhD dissertation, University of
Amsterdam.
Four decades of study of synchronic variation in varieties of Dutch. A sketch
Van Alphen, I. 1999. Variatie inverbale interactie (v/m). Een sociolinguïstisch onderzoek naar
de vorm en de functie van het taalgebruik van adolescentemeisjes en jongens. PhD
dissertation, University of Amsterdam.
Van Bezooijen, R. 1988. The relative importance of pronunciation, prosody, and voice quality for
the attribution of social status and personality characteristics. In Language Attitudes in the
Dutch Language Area, R. van Hout & U. Knops (eds), 85–103. Berlin: De Gruyter Mouton.
Van Bree, C. 1981. Hebben-constructies en datiefconstructies binnen het Nederlands taalge-
bied. PhD dissertation, Leiden University.
Van Bree, C. 1985. Structuurverlies en structuurbehoud in het dialect van Haaksbergen en
Enschede: een onderzoek naar verschillen in resistentie. Leuvense Bijdragen 74(1): 1–35.
Van Craenenbroeck, J. 2004. Ellipsis in Dutch Dialects. PhD dissertation, Leiden University.
Van de Velde, H. 1996, Variatie en verandering in het gesproken Standaard-Nederlands
(1935–1993). PhD dissertation, Nijmegen University.
Van den Broeck, J. 1977. Class differences in syntactic complexity in the Flemish town of
Maaseik. Language in Society 6: 149–181.
Van der Plank, P. 1985. Taalsociologie: Een inleiding tot de rol van taal in het maatschappelijk
verkeer. Muiderberg: Coutinho.
Van Ginneken, J. 1913–1914. Handboek der Nederlandsche taal, Deel I: De sociologische structuur
der Nederlandsche taal. Deel II. De sociologische structuur onzer taal. Nijmegen: Malmberg.
Van Hout, R. 1995. Kwantitatieve taalkaarten en variabele regels. In Lingua Theodisca. Beiträge
zur Sprach- und Literaturwissenchaft. Jan Goossens zum 65. Geburtstag, L. Kremer &
H. Niebaum (eds), 341–348. Münster: Zentrum für Niederlande-Studien.
Van Hout, R. 1989. De structuur van taalvariatie: een sociolinguistisch onderzoek naar het
stadsdialect van Nijmegen. PhD dissertation, Nijmegen University.
Van Hout, R. & Knops, U. (eds), 1988. Language Attitudes in the Dutch Language Area. Foris:
Dordrecht.
Van Hout, R. & Vallen, T. 1984. Sociolinguïstiek: Januskop of kop van Jut? Tijdschrift voor
Taal- en Tekstwetenschap 4: 279–286.
Van Koppen, M. 2005. One Probe – Two Goals: Aspects of Agreement in Dutch Dialects. PhD
dissertation, Leiden University.
Van Meel, L. 2011. Variation in the realization of /z/ within and across Dutch adolescents of
Moroccan, Turkish and Dutch descent. Paper delivered at the yearly TIN-dag [‘Linguistics
in the Netherlands day’]. Utrecht, February 5.
Van Oostendorp, M. 1995. Vowel Quality and Phonological Projection. PhD dissertation,
Tilburg University.
Vandekerckhove, R. 2000. Structurele en sociale aspecten van dialectverandering. De dynamiek
van het Deerlijkse dialect. Gent: Koninklijke Academie voor Nederlandse Taal- en
Letterkunde.
Vandekerckhove, R., de Houwer & A. Remael. 2009. Between language policy and linguistic
reality: Intralingual subtitling on Flemish television. Pragmatics 19(4): 609–628.
Voortman, B. 1994. Regionale variatie in het taalgebruik van notabelen. Een sociolinguïstisch
onderzoek in Middelburg, Roermond en Zutphen. PhD dissertation, University of
Amsterdam.
Vousten, R. 1995. Dialect als tweede taal: linguistische en extra-linguistische aspecten van de
verwerving van een Noordlimburgs dialect door standaardtalige jongeren. PhD dissertation,
Nijmegen University.
Frans Hinskens
1. Introduction
* This study was conducted with support from ERC Project #230310 ‚Traces of Contact‘. We
would like to thank two anonymous reviewers and the editors for their helpful suggestions.
Suzanne Aalberse and Pieter Muysken
Subsection 1.4 relates studies to HLs to language change, while 1.5 links HLs to the
notion of superdiversity.
In Section 2 various earlier research projects on immigrant languages in the
Netherlands are briefly presented from various perspectives: second language ac-
quisition (2.1), code-mixing and code-switching (2.3), loss and attrition (2.3), and
superdiversity (2.4). Section 3 presents the initial results of a case study on the
Chinese languages as heritage languages in the Netherlands. In 3.1 the complexi-
ties of the Chinese languages are hinted at, in 3.2 the migration of the Chinese to
the Netherlands is sketched, in 3.3 the overlapping identities in Chinese organisa-
tions, and in 3.4 language use and change. Finally, Section 4 briefly presents some
conclusions.
match native speakers with heritage speakers in their dialect background and to
include lower educated monolinguals in a sample.
Apart from the quantity of the input, the quality of the input is often consid-
ered as the most important factor in explaining differences between heritage
speakers and native speakers. O’Grady et al. (2011) refer to the study of HL acqui-
sition as a classic experiment in nature: a naturally occurring event that sheds light
on the effect of factors that for ethical or practical reasons could not be controlled
for in a laboratory setting. The study of HL learners can help to resolve the debate
on the role of age and input in language acquisition. Many heritage learners are
like monolingual child learners in the sense that they acquire the language in an
implicit way in a family setting at a young age, but they are different from the
monolingual children in terms of the amount of language use. Moreover, HL
speakers can show transfer from the dominant language.
A central claim in the work of Montrul (2008) is that there is not only a critical
period for language learning, but also for language loss. She describes a positive
correlation between age and proficiency in language loss. Foreign adoptees who
cease to use their mother tongue at a very young age are likely to completely forget
the language. Simultaneous bilinguals who speak two languages from an early age
(before three), have less exposure to the HL, and are second in line in the hierarchy
of language forgetters. Sequential bilinguals acquire much more of the language,
but still their knowledge is incomplete when compared to monolingual speakers.
Last in the hierarchy of forgetters are the first generation immigrant speakers who
arrived in their new country after puberty. Montrul (2008) claims that when a
language is used until the age of nine it will not be forgotten anymore. Speakers
might have problems with lexical retrieval and there may be superficial differenc-
es, but according to her the language system stays intact. The observation that
disuse of a language can lead to language loss especially if it begins at an early age
is an important finding from HL studies.
Another important issue in the study of the knowledge of heritage speakers is
whether these speakers manifest incomplete acquisition or attrition of grammar.
In other words, do heritage speakers fail to learn certain structures or do these
structures get acquired and then undergo subsequent degradation due to lack of
use or transfer from the dominant language? This question plays a central role in
the work of Polinksy (2011). She argues, for example, that in the domain of relative
clauses, adult heritage speakers undergo attrition. While heritage children per-
form like monolinguals, only heritage adults do not perform native like. The as-
sumption is that these heritage adults were once like the heritage children and
since the children perform native like in this respect, the non-native behaviour of
the adults must be due to attrition.
Suzanne Aalberse and Pieter Muysken
In the last section we saw that heritage speakers do not always perform native like.
In order to ameliorate language skills in these cases HL speakers sometimes enroll
Language contact in heritage languages in the Netherlands
in the likelihood of profiting from a special heritage class. The attitude of the par-
ents towards the HL also affects the likelihood of success.
Carreira (2004) describes speakers that would not qualify for a HL course
based on the assessment of their linguistic abilities, but who have overheard the
language and who can speak it a bit. From a linguistic standpoint these learners are
like L2 learners, but they have affective and intellectual needs that are generally not
addressed in a course for second language speakers. To make these students feel
more comfortable in class, special attention can be paid to the culture of the group,
to talk about bilingualism, to discuss prejudice on dialectal variation and in gen-
eral to pay attention to the topics that interest these students. Carreira (2004) also
notes that it is important to treat these students as resources. They know things
other second language learners do not know, and students are more likely to be
successful if they contribute knowledge from their own experience in class.
A last question in the education domain is why special heritage classes should
be funded. In this respect the title of the first conference on HLs in the United
States in 1999 is telling: Heritage languages in America: Preserving a National
Resource. The idea is that with proper instruction, HL learners are much more
likely than any second language learner to achieve near-native linguistic and so-
ciocultural fluency. And therefore the heritage learner’s capacity should be used as
much as possible, educating those speakers that will most likely reach success in
the language, thereby opening opportunities for trade and international coopera-
tion. Kagan and Dillon cite the agenda of UCLA steering Committee Heritage
Language Research Priorities Conference Report (2001) stating that ‘a multidisci-
plinary approach is needed including economist, scientists and social scientist
would..... have important roles in measuring the effects of heritage language learn-
ing on the individual, the family, the community, and the nation.’
Another factor that influences shift is language prestige. Clyne & Kipp (1997)
argue that Mandarin speakers retain their language more often than Cantonese
speakers. Mandarin is the Chinese language with the highest market value.
Language prestige is related to other factors that stimulate language maintenance
such as the availability of schooling in the language and the fact that the language
has a long writing tradition. The Chinese written system is based on Mandarin
grammar. All Chinese schooling in Australia is in Mandarin and the same goes for
the education of Chinese speakers in the Dutch school system (cf. Extra et al. 2002:
206, 214).
A third factor in language maintenance is numerical strength in a specific re-
gion. If there is a high concentration of a language group in a certain area the
likelihood of language maintenance is highest. Clyne and Kipp (1997) show that it
is relative concentration rather than absolute numbers that affects language main-
tenance. This result is replicated in Extra et al. (2002: 144), who show that for most
immigrant languages they study language vitality is highest in the cities of The
Hague or Utrecht. The relevance of a network of fellow speakers can also explain
why there are differences in early migration waves and later migration waves in
Australia. In the first migration waves the migrants were often low in number,
isolated from other speakers of the language and mostly married outside their
group. They lost their language quickly, whereas in later migration waves when
immigrants had a larger network of fellow speakers the language was retained.
In most cases language shift is stronger in the second generation than in the
first generation. Children from endogamous languages (where both parents speak
the same language) shift less than children from exogamous marriages (where the
two parents speak different languages). The second generation is most likely to
show shift in the age 5–14 when they go to school and in the cases of some Asian
groups in Australia the age group 15–23 also shows high instances of language
shift as children leave their home surroundings. Language shift is halted when
there are strong ties with the home country such as in doing business with the
home-country that involves travelling back and forth.
A last important factor in language maintenance is the cultural value a commu-
nity attaches to the language. Smolicz et al. (2001) show that in some cultures lan-
guage is a core value of the culture. If this is the case language maintenance is likely.
In the Dutch culture language is assumed to be less important and this can in part
explain the high incidence of language loss in the Dutch community in Australia.
To what extent is language loss like language change occurring in more stable
monolingual communities? Before trying to answer this question, we need to
Suzanne Aalberse and Pieter Muysken
– The recency effect has to do with how long ago a particular item or structure
was used by a speaker; the more recently used, the better retained. It is clearly
something most directly relevant to individual language use.
– Frequency of reinforcement has to with the way frequent use reinforces a
particular item or structure in the linguistic competence of individuals.
– Elements with a high pragmatic load will be retained longer, since they are
important to patterns of use.
A second set of factors concerns properties of the language structures
themselves:
– The complexity of structures likewise will play a role: highly complex struc-
tures will be subject to more rapid attrition.
– The degree to which elements or structures contrast also influences the rate of
attrition, as does their linguistic distinctiveness. Distinctive and contrasting
elements will be retained more often.
– The irregularity of a pattern will certainly contribute, everything else being
equal, to its more rapid loss.
– Another factor that may play a role is the centrality of a certain structure to
the core of the linguistic message; more central elements will be retained
longer.
– Related to this, functional load concerns the idea that certain contrasts or
positions in language play a more central role in distinguishing words and
structures, and hence in conveying meaning, than others.
– The animacy hierarchy is based on the idea that certain categories are more
animate than others (the pronoun ‘I’ > nouns; nouns for humans > nouns for
objects; etc.). Presumably elements high on the animacy hierarchy would be
retained longer.
Obviously, all these factors, and the interaction between them, require much more
elaboration and further study.
The social anthropologist Steven Vertovec has introduced the notion of ‘superdi-
versity’ (2007, 2010) to indicate the form of highly complex and dynamic diversity,
the ‘diversification of diversity’, that emerges in contemporary societies. This su-
perdiversity is argued to be the result of two independent but coinciding social
processes: (a) the shift in migration patterns from a once and for all unilateral
movement from region or country A to region or country B, as in traditional
migration, to a constant going back and forth between A and B, and to social
Suzanne Aalberse and Pieter Muysken
networks of individuals spanning A and B over extended periods; (b) the emer-
gence of internet and mobile phone communication, which has produced much
more trans-local networks, not bound to a specific place.
This superdiversity has many manifestations, and proponents of superdiver-
sity argue that traditional narratives of migrants as going through a generational
cycle:
– first generation newcomer, trying to find its way
– second generation struggling to become part of mainstream society
– third generation already settled and assimilated
– that such narratives are less and less applicable. Of course, it needs to be stud-
ied whether indeed patterns of migration have changed significantly for the
majority of migrants. It is clear that some migrant groups indeed do not follow
the traditional pattern, and for all groups communication patterns have shift-
ed dramatically.
Blommaert and Rampton have explored the consequences of superdiversity for
language studies, and conclude (2011: 4):
‘Rather than working with homogeneity, stability and boundedness as the start-
ing assumptions, mobility, mixing, political dynamics and historical embed-
ding are now central concerns in the study of languages, language groups and
communication.’
Building on this, Blommaert and Backus (2011) have reviewed the consequences
for the central notion of ‘linguistic repertoire’, arguing for a flexible and dynamic
perspective.
Although HLs are intensively being studied in the United States and Canada, they
have received relatively little attention in the Netherlands earlier, with the excep-
tion of the exploratory studies from Marian Schonemakers-Klein Gunnewick
(1989) on Portuguese migrants in the Netherlands, and those of Jaspaert & Kroon
(1991) on Italian.
An exception to this observation is the work by Backus’ research group in
Tilburg, who have been studying Turkish spoken in the Netherlands (e.g. Backus
et al. 2011). This section discusses the focus of some crucial earlier approaches to
languages in the immigrant communities in the Netherlands.
Of course it is not possible to survey the vast amount of research that has been
carried out on the sociolinguistic patterns of the Dutch immigrant communities.
A number of centres and groups have been involved in this, starting around 1976.
Language contact in heritage languages in the Netherlands
While researchers from most Dutch universities need mentioning, it is fair to say
that sociolinguistic research in this domain has been concentrated particularly in
the universities of Amsterdam, Nijmegen, and Tilburg.
Another feature of research in this domain in the Netherlands has been the
fairly even-handed concentration on two large HL communities: immigrants from
Turkey and Morocco. There have been countless projects of the type: ‘X in the
Turkish and Moroccan communities’, where X is L1 or L2 development, code-
mixing, attrition and loss, etc. This is justifiable numerically, since these two com-
munities are among the largest, as can be seen in Table 1. Taking the country
of birth of either mother or father (but not necessarily nationality) as measure,
Turkey and Morocco rank quite high, but so do the former colonies Suriname and
Indonesia.
One thing that should be taken into account is that migration from the former
colonies was a much more complex process, sociolinguistically as well. We return
to this issue below, in the context of the discussion about superdiversity. To be fair,
however, the fully multilingual character of the Moroccan and Turkish
communities was not taken into account until much later. Berber and Kurdish
were not much studied; thus the Berber specialist and enthusiast El Aissati wrote
a doctoral thesis about Moroccan Arabic rather than Berber.
Table 1. Population of the Netherlands on 01-01-2007 on the basis of the combined
criterion of country of birth (mother/father) and the nationality criterion (CBS 2008).
People from the (Netherlands) Antilles have Dutch nationality. Adapted from Extra (2011)
It is fair to say that there has been very extensive research on the acquisition of
Dutch as a second language, in the area of grammar, vocabulary, and perhaps
slightly less on pronunciation. Both children and adults were and continue to be
studied. Dutch was a member of the ESF project on untutored adult second lan-
guage acquisition directed by Wolfgang Klein and Clive Purdue. An overview of
research is given in Bossers et al. (2010).
There has been much less work on ethnolects of Dutch (a recent survey is
provided in Muysken 2013; see also Muysken 2010 and Muysken & Rott 2013). A
few isolated studies were done on Indonesian Dutch and Suriname Dutch, but
particularly the latter merits a great deal more investigation, both in the Nether-
lands and in Suriname (see also de Kleine 2013). Yiddish Dutch in Amsterdam has
been studied in a historical perspective. Nortier (2001) and Nortier and Dorleijn
(2005), among others, have studied street language, particularly among young
people of Moroccan and Turkish descent. Currently under way is a broad survey
project supervised by Hinskens and Muysken on 12-year olds and 20-year olds of
Moroccan and Turkish descent in Nijmegen and Amsterdam. Both phonetic and
morpho-syntactic variables are studied (cf. e.g. Muysken 2013).
Although there has not been systematic work in all communities, there has been
in depth research on a number of code switching and mixing settings, particularly
Moroccan Arabic-Dutch (Nortier 1990; Boumans 1998) and Turkish (Backus
1992, 1996). Nortier found very dense code-mixing patterns particularly among
highly competent bilinguals, while Boumans found evidence from incipient con-
ventionalization. Backus has traced code-mixing and -switching patterns across
several generations of speakers, ultimately linking code-mixing to language
change, as in his subsequent work.
Some studies concern e.g. Sranan-Dutch (Bolle 1994), but this area needs
much more exploration.
Language loss and attrition research in the Netherlands has taken two tracks,
which only incidentally come together. There have been a number of studies with
a primarily psycholinguistic focus, often involving De Bot as the supervising re-
searcher (Weltens et al. 1986; de Bot & Weltens 1991), including the recent work
by Schmid (2011). A variety of languages was involved.
Language contact in heritage languages in the Netherlands
A second group of studies has focused on children and on the interaction be-
tween HL and acquisition, with a particular focus on Turkish: Boeschoten (1990)
and Schaufeli (1991).
The third tradition has been more sociological in nature, focusing on lan-
guage loss. Typical studies include Jaspaert and Kroon (1991) on Italian, El
Aissati (1997) on Moroccan Arabic, and Van der Avoird (2001). The latter intro-
duced a comparative dimension with Hindustani speakers in the Netherlands and
the UK.
Carmen Fought (2006), to illustrate the relation between language and ethnicity,
has sketched the complexity of speech styles and registers in the continuum from
American English to South American Spanish. This continuum includes Chicano
English, local Chicano Spanish varieties, various forms of second language
English, varieties involving heavy mixing, etc. The continuum is also in the field
of gravity of varieties like African American Vernacular English, various tradi-
tional forms of Spanish spoken in the US, the Spanish of several Latino immi-
grant groups, second language Spanish of non-Latino’s, Mexican Spanish as
transmitted through the media, etc. Developing identities and new allegiances,
regional and ethnic, as well as new migration waves, trigger a continuous dy-
namic in this continuum. Individuals may move through it or remain in a certain
ethnolinguistic niche.
This situation illustrates the effect that superdiversity has on language. It is
tempting to put all varieties spoken in a given broad language community on a
two-dimensional map, but the situation is almost too complex for that. One of the
cases we are studying also involves Spanish. Pablo Irizarri van Suchtelen has been
recording first and second generation immigrants (many of them refugees) from
Chile, and the dynamics of the development of Spanish within their community
not only reflects influence from Dutch but also the echoes of the larger Latino
community that has came into existence in the urban western rim (de Randstad)
of the Netherlands. The situation of Chinese varieties discussed in the next section
is a prime example of this superdiversity.
Clyne and Kipp (1997) refer to people from different migration waves as vintages;
Kloss (1966) uses the term stocks. The first vintage of Chinese immigrants arrived
in the Netherlands before World War II (Geense & Tsui 2001, Li 1999). In 1898, 11
Chinese seamen registered as foreign employees in Amsterdam. There were 196 sea-
men registered in 1910 and this number jumped to 765 in 1911 rising to 2165 in
1915 through 1930 the numbers stayed between 2000–3000 to decrease again to
about 400 in the second world war (Li 1999). The sudden increase of Chinese in
1911 can be related to a strike in the harbour of Rotterdam. Apart from the sailors
a group of Chinese immigrants arrived via Marseille. They were from Wenzhou
and Qingtian districts and they were pedlars selling cheap ties, small toys from
door to door. When the sailors could not directly find employment on a ship they
stayed in lodgings close to the harbours of Rotterdam and Amsterdam. Chinese
hairdressers and shops arose around these lodging places creating small China
towns in Rotterdam in the area of Katendrecht, and in Amsterdam in Nieuwmarkt.
These first migrants were mainly men. Many men returned to China, but some
migrants like hair dressers or lodging owners settled more permanently and mar-
ried Dutch women. If the mixed couples had children, the children spoke Dutch.
During the crisis in the 1930s life was very difficult. As an alternative to being
sailors many of the Chinese started to sell sweetened peanuts on the streets. When
discussing Chinese with older Dutch people, their first association with the
Chinese is usually the stereotypical phrase pinda pinda lekka lekka ‘peanut peanut
Language contact in heritage languages in the Netherlands
tasty tasty’ a slogan said to be used by the peanut vendors. Other immigrants
started restaurants.
The Chinese restaurants became popular during the 1950s. People became
richer and wanted to try new food and the Chinese restaurant holders started to
cook Indonesian food for the people who had come from the former Dutch colony
Indonesia. The restaurants became so popular that new workers were attracted
from China starting the second immigration wave (1950–1975) (Li 1999). After
1955 immigration from the PRC was forbidden and therefore all immigrants came
from Hong Kong. The Hong Kong Chinese had English passports, allowing them
to immigrate. Cantonese became the dominant Chinese language of the Dutch
Chinese community. Geense and Tsui (2001) states that Cantonese came to be as-
sociated with Dutch Chinese identity. To become part of the Chinese community
Cantonese was requisite. Li (1999) writes:
In fact some Zhejiang interviewees commented that, because the only workers
they could recruit were Hong Kong people, the Zhejiang restaurateurs had to learn
the Hong Kong dialect (i.e. Cantonese) to communicate with their employees.
Consequently the Hong Kong culture has since then dominated the Chinese
community.
Apart from the Hong Kong immigrants, there also was a group of Indonesian
Chinese, referred to as Peranakan Chinese. These Chinese spoke Bahasa Indonesia
and Dutch and they integrated into the Dutch culture very quickly. Some of the
ancestors of these immigrants had already studied at Dutch universities in the
period between 1911 and 1940 (Li 1999).
In the mid-1970s new groups of Chinese immigrants arrived. In 1975 after the
independence of Suriname, many Surinamese, including Surinamese Chinese
came to the Netherlands. The Surinamese-Chinese spoke Hakka mainly. In 1976
Chinese Vietnamese refugees came, and in 1977 the PRC started allowing emigra-
tion. Note that especially the Surinamese Chinese have a complex migration his-
tory. They were originally from Northern China, moved to Hong Kong, then from
Hong Kong to Suriname (some via Java), and via Suriname they came to the
Netherlands.
The number of immigrants from the PRC has increased enormously
(Gijsberts et al 2011: 29, 190). This new immigration wave consisted mainly of
highly educated immigrants. A very different group from the restaurant Chinese,
who frequently had no education after elementary school (Gijberts et al 2011: 48).
Since 2000 many Chinese students studied at Dutch universities, creating yet an-
other sub-community, usually Mandarin speaking. In number immigrants from
the PRC are now the dominant group of Chinese in the Netherlands. Gijberts et al
(2011: 29) report that on January 1st 2010 there were 71500 Chinese migrants and
their offspring in the Netherlands of whom 53300 came from the PRC.
Suzanne Aalberse and Pieter Muysken
In the project Traces of Contact a start is made with the elicitation of data from
Chinese heritage speakers. Shi (2011) compared seven heritage speakers of
Mandarin to six native speakers of Mandarin and Chau (2011) compared two
speakers of Cantonese in Venlo with two speakers of Cantonese in Amsterdam.
This section presents the first results of this investigation which is still underway.
Both Shi (2011) and Chau (2011) used a video elicitation kit geared at eliciting dif-
ferent argument structures and aspectual situations (www.ru.nl/linc). Shi focused
on Mandarin speakers in the Netherlands. Four of her participants began to speak
Dutch before the age of four. Two participants started to speak Dutch at a later age,
6 and 8 respectively. She studied the use of four aspect markers in these speakers,
namely le (perfective), guo (experiential), zhe (durative) and zai (progressive). She
compared the speech of the young bilinguals with the speech of Chinese students
studying in Nijmegen. The children who began to speak Dutch at the age of six or
later spoke like the Chinese students. The young bilinguals performed differently.
They used significantly less aspect markers. The low use of aspect markers is also
observed in English second language learners of Chinese. Monolingual child learn-
ers tend to overuse aspect markers (Jin & Hendriks 2005).
Language contact in heritage languages in the Netherlands
The only aspect particle the young bilinguals overused was zai, an infrequent
particle in monolingual input. Zai marks the progressive. It is associated with the
features [+durative] and [+dynamic]. So it is fine to use zai with activities like in
(1) and with accomplishments (2), but it cannot be combined with achievements
because of the feature [+durative], and it cannot be combined with states because
of the feature [+dynamic].
(1) Wo zai youyong. (Activity) Jin & Hendriks (2005)
I zai swim
‘I am swimming’.
(2) Wo zai kan yi ben shu. (Accomplishment) Jin & Hendriks (2005)
I zai read one cl book
‘I am reading a book’.
English second language learners overuse zai to achievements if they overuse the
form. Monolingual children overgeneralize the form to states. Again the heritage
speakers perform like the second language speakers, they overgeneralize zai to
achievements. In example (3) one of the early bilinguals in Shi (2011) overuses zai:
shuibian (waterside) implies an endpoint which goes against [+durative] and
therefore zai is usually out.
(3)* You yi-ge xiao chuan zai you-dao shuibian.
exist one-cl small boat zai swim-dir.p waterside
‘A small boat is sailing to the waterside.’
Although the number of participants is still low at this moment, an overall chi-
square test shows that there was a significant association between the type of
aspect marking and the group a speaker belonged to (e.g. native speakers, the
HL’s who became bilingual before the age of two and the HL’s who became bilin-
gual in school (χ2 (10) = 144.470, p < 0.00001). The results by Shi (2011) show
that in the case of aspect markers, young bilinguals are a special group. Unlike
the Chinese children who came to speak Dutch at in school, they use less aspect
markers, and if they overgeneralize zai, they overgeneralize in the direction of
achievements. Both the frequency of use and the direction of the overgeneraliza-
tion are like second language learners and unlike monolingual first language
learners.
Chau’s (2011) study on speakers of Cantonese and Venlo is very modest given
that only four participants were involved. More data are needed but the basic as-
sumption of her study is worthwhile, Chau’s hypothesis based on her own experi-
ence as a heritage speaker of Cantonese was that there would be more language
change in case of a smaller network of fellow speakers and less change in larger
Suzanne Aalberse and Pieter Muysken
cities where people had access to more Cantonese oriented parties and other ac-
tivities. She compared speakers from Amsterdam with a high density of Cantonese
Chinese speakers, with two speakers from Venlo (a provincial town in the south of
the Netherlands), where the number of Chinese is much lower. The speakers from
Amsterdam frequently watched Cantonese soaps and they attended Asian parties.
One of the participants from Amsterdam was active in dragon boat rowing. Both
participants are able to use Cantonese in many different situations. The two par-
ticipants from Venlo had watched Cantonese TV in the past and they had listened
to Cantopop, but they stopped, and the number of situations where they used
Cantonese was limited to visits to their family. During the elicitation the partici-
pants were asked to speak as much Cantonese as possible. This felt unnatural.
Usually Cantonese and Dutch are mixed in their speech.
The speakers in Venlo show innovative patterns in the domain of measure
words, and overgeneralized the most common one. Moreover, they used adverbs
preverbally rather than postverbally and one participant even used Dutch inflec-
tion in her Chinese. In (6) an example is given of the use of the third person
singular inflection in Dutch that has become a sort of particle in Cantonese:
(4) Keoi5 go3 snaar laan6 zo2-t
佢 個 snaar 爛 咗-t
3s cl string tear asp-t
‘His (guitar) string is broken.’
It is possible that this participant was trying to suppress Dutch, because we asked
for Cantonese only and that she was unable to suppress inflection. Although the
number of participants makes it impossible to draw conclusions about the system-
aticity of this type of interference and of its spread through the community, the
fact that the use of Dutch inflection was used throughout the elicitation, suggest a
closeness between Dutch and Chinese in this speaker, perhaps facilitated due to
intensive code switching.
4. Conclusions
References
van der Avoird, T. 2001. Determining Language Vitality. The Language Use of Hindu
Communities in the Netherlands and the United Kingdom. PhD dissertation, Katholieke
University Brabant, Tilburg
Backus, Ad. 1992. Patterns of Language Mixing. A Study in Turkish-Dutch Bilingualism.
Wiesbaden: Otto Harassowitz.
Backus, Ad. 1996. Two in One. Bilingual Speech of Turkish Immigrants in the Netherlands. PhD
dissertation, Katholieke Universiteit Brabant, Tilburg.
Backus, A., Dogruöz, A.S. & Heine, R.J. 2011. Salient stages in contact-induced grammatical
change: Evidence from synchronic vs. diachronic contact situations. Language Sciences 33:
738–752.
Benmamoun, E., Montrul, S. & Polinsky M. 2010. White Paper: Prolegemena to Heritage
Linguistics. Cambridge MA: Harvard University.
Bley-Vroman, R. 1988. The fundamental character of foreign language learning. In Grammar
and Second Language Teaching: A Book of Readings, W. Rutherford & M. Sharwood Smith
(eds), 19–30. Rowley MA: Newbury House.
Blommaert J. & Rampton B. 2011. Language and superdiversity: A position paper. Working
Papers in Urban Language & Literacies 67.
Blommaert J. & Backus, A. 2011. Repertoires revisited: ‘Knowing language’ in superdiversity.
Working Papers in Urban Language & Literacies 70.
Boeschoten, H. 1990. Acquisition of Turkish by Immigrant Children. A Multiple Case Study of
Turkish Children in the Netherlands aged 4 to 6. Wiesbaden: Harassowitz
Bolle, J. 1994. Sranan Tongo – Nederlands. Code-wisseling en ontlening. MA thesis, Universiteit
van Amsterdam.
Bossers, B., Kuiken, F. &Vermeer A. 2010. Handboek Nederlands als Tweede Taal in het
volwassenenonderwijs. Bussum: Coutinho.
de Bot, K. & B. Weltens 1991. Recapitulation, regression, and language loss. In First Language
Attrition, H.W. Seliger & R.M. Vago (eds.), 31–52. Cambridge: CUP.
de Bot, K. & Clyne, M. 1994. A 16-year longitudinal study of language attrition in Dutch im-
migrants in Australia. Journal of Multilingual and Multicultural Development 15(1): 17–28.
Boumans, L. 1998. The Syntax of Code switching: Analyzing Moroccan Arabic/Dutch
Conversation. PhD dissertation, Nijmegen University.
Carreira, M. 2004. Seeking explanatory adequacy: A dual approach to understanding the term
“heritage language learner”. Heritage Language Journal 2. <http://www.heritagelanguages.
org/> (25 July 2011).
CBS 2008. Allochtonen in Nederland 2007. Den Haag: Centraal Bureau voor Statistiek.
Chau, W. 2011 The Influence of Dutch on Cantonese in the Netherlands. Rural Versus Urban
Areas. MA thesis, Leiden University.
Clyne, M. & Kipp, S. 1997. Trends and changes in home language use and shift in Australia
1986–1996. Journal of Multilingual and Multicultural Development 18(6): 451–473.
El Aissati, A. 1997. Language Loss among Native Speakers of Moroccan-Arabic in the Netherlands
[Studies in Multilingualism 6]. Tilburg: Tilburg University Press.
Extra, G., Aarts, R., van der Avoird, T., Broeder, P. & Yagmur, K. 2002. De andere talen van
Nederland. Thuis en op school. Bussum: Coutinho.
Suzanne Aalberse and Pieter Muysken
1. Introduction
environment in which new discursive genres and patterns can be developed along
with new practices (such as online gaming and blogging; Leppänen et al. 2009;
Androutsopoulos 2006) and identities (as in life writing on Facebook and other
virtual social networks); (2) the new force with which different (and new) variet-
ies of English get spread and adopted by groups all over the world; (3) the way in
which a lot of this language spread and change is connected to the worldwide
surge of popular culture formats such as hip hop music, computer games and so
forth (Pennycook 2007); and (4) the way in which languages as a consequence of
their globalization-related presence on the internet but also as a consequence of
their use in linguistically super-diverse contexts, undergo changes in their ap-
pearance and structure, e.g. by using Latinized instead of the languages’ original
scripts on the internet or by using features of different languages in one commu-
nicative exchange.
Second, there is the effect of globalization on migration patterns and émigré
communities. On the one hand, there are new forms of migration. Refugees and
short-term or itinerant labour migrants have changed the face of large urban cen-
tres all over the world, leading to what Vertovec (2006) called ‘super-diversity’: a
mixture of resident and non-resident migrants in urban centres, having different
migration motives and purposes and displaying different patterns of organization
of their migrations, including the rise of ‘networks’, ‘global communities of prac-
tice’ or ‘super-groups’ rather than ‘language communities’ as the dominant format
of social organization (Castells 1996). On the other hand, and connected to this,
there is the effect of globalization on the sociolinguistics of migration. There is
more language diversity now in urban migration centres than ever before, and
technologies such as Skype and mobile phones enable migrants now to remain in
close contact with communities elsewhere in the world, usually in their home lan-
guages. Patterns of diasporic multilingualism have been changed and the discus-
sion on language maintenance and loss is entering a new level in which language
loss is becoming an obsolete concept.
Third, there is the effect of globalization, notably of the spread of English and
other global languages within and between nation states, on language hierarchies,
in particular on the position and predicament of linguistic minorities. Issues of
language endangerment, minority language maintenance and loss have acquired a
new dynamic in the context of the developments sketched above. Lingua franca
use of global languages has acquired a new dimension, leading to new patterns of
multilingualism among speakers of minority and majority languages and creating
new opportunities for language survival (Mufwene 2005, 2008).
These three primary domains of the sociolinguistics of globalization are
accompanied by a fourth domain, i.e. the domain of communities that, be it on
the basis of experiences or expectations, have serious doubts regarding their
Chinese and globalization
Now that some of our theoretical bearings have been set, we can move to consider
possible topics in a sociolinguistics of globalization in China. The range of actual
topics is, of course, infinitely vast, but we will try to delineate two large complexes
of topics. Both complexes have to do with the theme with which we concluded the
previous section: mobility. As we said, a sociolinguistics of globalization will have
to be a sociolinguistics of mobility, and this means that we must look at language
Chinese and globalization
phenomena in which mobility issues are a key feature. The blocks of topics,
consequently, could be defined as language for mobility and mobile people and
language.
To start with the first topic, the leading question here is: what sort of mobility
potential do particular languages, varieties and features offer to people? And it is
clear right from the start that such mobility potential will be significantly different
for different languages, varieties and features. And it is also clear that the general-
ized spread of English in China has clearly reshuffled the cards in this respect as
well. While Mandarin used to be, in Chinese society, the language of widest mo-
bility – given that the nation-state was the range of mobility for the overwhelming
majority of the people – we now see that the growing social diversification of the
population, with the rise of new globalized elites, creates a more chequered pat-
tern. For some groups in the population, English has now become available as the
language that offers global mobility. It does so by enabling physical relocation to
most other places in the world, as well as virtual mobility through communica-
tion technologies such as mobile phones or the internet. Proficiency in English
allows you to conduct your business in Mumbai and Rome, New York and Buenos
Aires, and to contact people all over the world. The mobility potential of English
is, seen from that perspective, virtually unlimited, be it that, of course, in most
places outside the native English environments, English would be a language in
which only a minority would be fully proficient. The fact is that English offers
global mobility among English-proficient elites all over the world, and it occurs in
more niches worldwide than perhaps any other language. (At the same time of
course, English creates a new social division between ‘haves’ and ‘have-nots’,
between those who engage in the global world through English and those who are
excluded and cut off.
This means that Mandarin has now been firmly relegated as a language of
mobility to China and the Chinese diaspora in the world. It has become, or has
remained, a national language (be it that this language is spoken by a quarter of the
world’s population). Less people in the world will learn Mandarin than people
who learn English. The fact is, however, that Chinese is gaining currency as a mod-
est globalized language. The export of Chinese goods all over the world makes, for
instance, Chinese writing something which has now become part of almost any-
one’s household in the West – in the form of printed product labels, user manuals
of appliances and so forth. But while the language has, thus, been spread over most
of the globe, it has been spread without triggering communication. It has been
spread, in short, as an empty sign system that for non-Chinese speakers does not
communicate meanings (other than “this is from China”).
The national sociolinguistic order in China, however, has effectively been
reorganized due to the rising importance of English. While Mandarin remains
Sjaak Kroon, Jan Blommaert and Dong Jie
The characters in Figures 1a and 1b are printed in a slightly different style. To the
eyes of a non-native speaker of Chinese, Figure 1b seems to represent a more re-
cent production than Figure 1a and accordingly also the characters in Figure 1b
look more modern, i.e. stylised. Or at least the box in 1b has a newer appearance
and is less worn out than that of 1a, suggesting that it probably is a more recent
one. Still another fire extinguisher box in the Forbidden City seemed to escape
the difficult task of splitting up the words ‘fire extinguisher box’ into units that in
one way or another match the Chinese characters. The result is a message that
does not contain a single typographic error but represents the English without
any spatial distinction: FIREEXTINGUISHERBOX (picture missing; informa-
tion from our field notes). In producing these signs the ‘writer/printer/painter’
not only made small spelling mistakes (UISHR for UISHER and FIRF for FIRE)
but also showed limited knowledge or awareness of the meaning and form of the
English words: English words are split up in a way that does not follow the
conventions of English, but that does match the number of Chinese characters in
the sign, and they are put under the Chinese characters without matching their
meaning.
Two other examples of a fire extinguisher box were found on both sides of a
door at the Institute of Anthropology and Ethnology of the Chinese Academy of
Social Sciences in Beijing (pictures taken August 2011). The characters in
Figures 2a and 2b are both more stylized than the ones in Figure 1 but at the
same time they differ: Figure 2a appears to be more stylized in the sense that it
paints in a more hand-written way which suggests a higher level of sophistica-
tion. Figure 2a shows yet another solution of the mismatch between four
Chinese characters and three English words that was solved until now by
introducing non normative spatial writing of English (‘fire exting uishr’
and ‘fireextinguisher’) by simply adding ‘of ’ leading to BOX OF FIRE
EXTINGUISHER. Figure 2b on the other hand represents conventional English
writing. Apart from the unnecessary hyphen in all figures, Figure 2a contains a
typographic error: FIRF ALARM.
A final observation here relates to the multimodal character of the boxes. We
already referred to the colour red that is a universal reference to a semantic com-
plex of danger, alarm and fire (fighting). Two of the boxes in addition contain an
image of a (meanwhile) almost archetypical phone, off the hook in Figure 1b and
on the hook in Figure 2b. We already mentioned the differences in the way the
Chinese characters were printed. Also the English text shows different fonts: clas-
sical letters with serifs in Figures 1a, 1b and 2b, and more modern sanserif letters
in 2a (which also contains the most stylized characters).
Chinese and globalization
The fire extinguisher and fire hydrant boxes analysed thus far can be considered
semiotic artefacts in which specific resources are being blended in an attempt to
make sense to mobile people, i.e. foreign tourists to whom ‘English’ appears more
accessible than Chinese. As such they point backwards to their producers and
their conditions of production, i.e. a public authority (government) that wants to
convey meaning in English without having full access to the language. At the same
time they also point forward, towards their intended audiences, i.e. foreign non-
Chinese speaking tourists, and their intended consequences.
In addition to the above bilingual signs, we also found a somewhat different
sign on a fire extinguisher box (see Figure 4a). The Chinese characters are the
same as in the above examples. They are however not accompanied by English but
by the Pinyin1 equivalent of the message in Chinese characters: MIE HUO QI
XIANG (literally: kill-fire-device-box). In case of the Chinese-English signs it is
clear that Chinese and foreign, English proficient, visitors of the Forbidden City
are the intended audience. The combination of Chinese characters and Pinyin,
however, seems to suggest that it is necessary for a certain audience to explain the
way in which the characters have to be pronounced. For Chinese speakers how-
ever this information seems to be rather superfluous since they know how to
pronounce the characters whereas for non-Chinese speakers the ‘translation’ in
Pinyin is not necessarily very helpful since it only helps them to be able to read,
i.e. vocalize, the Chinese sign but they still would not know what it means. They
only have the colour red and an emergency phone number as clues to create mean-
ing and act accordingly. It is highly possible that the ‘translation’ in Pinyin is only
added because it introduces Latin alphabet to the sign and as such conveys an in-
ternational, western image of the sign to its Chinese readers. It might be no Eng-
lish, but it is close enough. It might also be a result of the national movement of
promoting the use of Pinyin across the country since the 1950s. A similar sign
(Figure 4b) was found at Beijing Foreign Studies University. The sign reads XIAO
HUO SHUAN XIANG (literally: kill-fire-material-box) without any multimodal
indication of what is hidden behind the white door.
We observe this specific use of English in the public space of Beijing system-
atically, which shows that the introduction of English as a language of and for
mobility in China is not totally unproblematic. In most cases, it can be considered
English with a Beijing accent. Mobility of signs and sign users involves complex
processes of decoding and interpretation. When signs travel, their shape moves in
a rather unproblematic way, whereas other features – meaning, indexicals, social
values etc. – do not travel too well. We distinguished three different types of
Chinese accents in English signs (Kroon, Dong, Van Bochove, Blommaert 2011).
1. Pinyin is the Roman alphabetic representation of standard Mandarin Chinese, which was
put in place in the 1950s.
Sjaak Kroon, Jan Blommaert and Dong Jie
Figures 4a and 4b. ‘Mie huo qi xiang’ and ‘Xia huo shuan xiang’
Chinese and globalization
First, there is English with a Chinese accent that is related to existing local re-
sources. These include self-evident orthographic and linguistic features but also
cultural modes of speech. Examples of the latter occur when existing Chinese
modes of speech are used as a blueprint for English expressions that make good
sense in Chinese but sound strange in English. Secondly there is also English with
a Chinese accent that originates from (total) absence of access to resources,
i.e. standard English being beyond reach of most or many people. This leads to
attempts towards English, resulting in unfamiliar expressions such as misnomers,
cluttered orthography and syntax and English translations that result in a ‘soup of
words’, each closely or remotely equivalent to the Chinese text but hardly making
sense when put together in what is at first sight conventional English orthography
and syntax.
The fire extinguisher box can be considered a semiotic artefact in which
specific resources are being blended in an attempt to make sense to mobile people,
i.e. foreign tourists to whom ‘English’ appears more accessible than Chinese. It
represents, we would argue, problems at the level of English orthography, i.e. the
rules for ‘writing English correctly’. As we have seen, these rules are violated in two
ways. First, the English writing contained spelling errors (the missing ‘e’ from
‘uishr’, the ‘f ’ instead of ‘e’ in ‘firf ’ and the unwarranted hyphen in ‘fire-alarm’).
Second, the English words were also graphically ordered in a way that violated
their conventional morphosyntactic boundaries (’exting-uishr’ and ‘hyd-rant’).
The addition of ‘of ’ in Figure 2a is interesting here because although it does not
violate the rules of English its equivalent cannot be found in the Chinese charac-
ters. It goes beyond the scope of this paper to fully analyse these examples in detail
here. We for example do not go any further into the possible underlying meaning
of the different versions of the simplified Chinese characters that are printed on
the boxes.
We argue that what we see here is the emergence of a distinct Beijing or even
Chinese dialect of the super-vernacular English, a dialect that is part of the Beijing
sociolinguistic profile. This dialect emerges out of the local sociolinguistic envi-
ronment and can only be understood within this environment. Treating it as just
English does not account for its local features nor for its local function. The emer-
gence of a Chinese English accent reflects on a micro scale the patterns of spread
of English as a super-vernacular that we see on a global scale as well. It also reflects
worldwide patterns of inequality in a local context. English with a Chinese accent
in the Beijing linguistic landscape might be perceived as valuable in its local
context and its use can be analysed as indexical for Beijing’s increasing participa-
tion in the global world, in the United Kingdom or the United States it may be
perceived as indexing low levels of education and migrant identity. This is what a
sociolinguistics of globalization should and can account for.
Sjaak Kroon, Jan Blommaert and Dong Jie
5. Conclusion
deserve attention. Often they enter the country with relatively low educational or
other qualifications, and they find themselves in the lower ranks of society. Pro-
cesses of sociolinguistic adaptation among such groups, too, may inform us in
very significant ways about the changing nature of the Chinese sociolinguistic
landscape. Studies along these research lines are far from being exhausted. On the
contrary, it is an emerging school of scholarship and the study of Chinese and
globalization will shed new light on China’s transforming society as well as the
globalizing process of Chinese in motion.
Before leaving this paper to the judgment of the reader, we should underscore
one major point. A sociolinguistics of globalization, ideally, looks at the total pic-
ture, not just at aspects of it. As mentioned earlier, it would be wrong to just focus
on the new global elites, on the rise of the internet and of popular culture, and on
English as a world language. We need to keep an eye on all the different objects and
actors in the field, because globalization changes the whole of society, not just
some visible parts of it. It changes, in effect, the whole of the world, even if in most
places in the world one would not see any direct evidence of globalization pro-
cesses. China is not immune to this process of change, is in fact a very central
agent in the worldwide process.
This reminder of the holistic nature of a sociolinguistics of globalization, of
course, means that sociolinguists must work in teams, preferably in large and in-
ternational teams, operating in different parts of the world and willing and able to
share and to engage with each other’s work. The work is important scientifically,
because we will be compelled to innovate our theoretical and methodological
frameworks. It is, however, also practically important because globalization is, at
heart, a very unfair process that creates (apart from a small category of winners)
many losers. A child who does not learn English now is a child that risks being
barred in the future from important roads towards upward social mobility. We
have a responsibility towards such potential and effective losers as well. If in taking
that responsibility we also refine our theoretical apparatus and become better sci-
entists, so much the better.
References
Androutsopoulos, J. 2006. Multilingualism, diaspora, and the Internet: Codes and identities on
German-based diaspora websites. Journal of Sociolinguistics 10(4): 520–547.
Blommaert, J. 1999. State Ideology and Language in Tanzania. Cologne: Rüdiger Köppe.
Blommaert, J. 2007. Sociolinguistics and discourse analysis: Orders of indexicality and
polycentricity. Journal of Multicultural Discourses 2(2): 115–130.
Blommaert, J. 2010. The Sociolinguistics of Globalization. Cambridge: CUP.
Sjaak Kroon, Jan Blommaert and Dong Jie
Dong, J. 20, 144, 279, 282, 283, Goeman, A. 230, 233, 234, 237 Hunter, P. 77
289 Goffman, E. 176 Hymes, D. 51, 163, 279
Dorleijn, M. 264 Gogolin, I. 1, 3, 4, 6, 7, 10, 12–14,
Driessen, G. 235 16, 17, 126, 161 I
Duarte, J. 13–17, 21 Goodman, J. 180 Iris, M. 36
Dubois, D. 62 Goossens, J. 230
Duncan, S. 32 Gorter, D. 9, 53, 55, 85, 234, 235 J
Gramley, S. 196 Jaensch, C. 107
E Greenacre, M. 27–29, 31, 32, 36, Jahl-Samilo, J. 180
Eckert, P. 50, 63 37, 41 Jansen, F. 239
Ehlich, K. 51, 55, 63, 68 Gregersen, F. 236 Jansen, M. 231, 234
El Aissati, A. 241, 263, 265 Grondelaers, S. 233 Janssens, R. 10
Elias, M. 230 Guattari, F. 144 Jaspaert, K. 241, 262, 265
Elsness, J. 205, 212 Guilat, Y. 53 Jaspers, J. 175
Erbrecht, L. 77 Guldbrandsen, T. 29 Jaworski, A. 46
Esser, H. 4 Gumperz, J. 51, 162, 174 Jessner, U. 99
Extra, G. 10, 46, 259, 263 Gupta, A. 196, 197 Jin, L. 268, 269
Gussenhoven, C. 239 Johansson, V. 77
F Johnson, J. 254
Fairclough, N. 51, 277 H Johnson, K. 238
Falk, Y. 99, 100 Haegeman, L. 238 Johnstone, B. 175
Fenn, P. 220 Haeseryn, W. 239 Jongenburger, W. 234
Fernald, A. 180 Hagen, A. 235 Jørgensen, N. J. 8, 127, 279
Fernandez, M. 188 Hakuta, K. 12 Juffermans, K. 70, 144, 149, 283
Fiesta, C. 182 Hall, J. 28
Fisher, A. 77 Hally, C. 197, 198 K
Fishman, J. 51, 234 Hamann, C. 106 Kagan, O. 257, 258
Fitzpatrick, D. 201 Hart, B. 180 Kameyama, S. 51
Flick, U. 9 Haslinger, I. 238, 239 Kapia, E. 19, 183
Flynn, S. 100 Haynes, M. 180 Kecskes, I. 28
Foley, C. 100 He, A. 125 Kellermann, E. 100
Fought, C. 265 Heeringa, W. 232, 236 Kennedy, R. 150
Foulkes, P. 238 Heine, B. 200 Kerswill, P. 245
Fournier, R. 239 Heller, M. 126, 129 Kessler, J.-U. 100
Franceschini, R. 28 Hendriks, H. 268, 269 Kipp, S. 258, 259, 266
Francis, B. 125 Hermans, B. 237 Klieme, E. 7, 10
Freeland, J. 132 Herzog, M. 228 Kloeke, G. 228
Fürstenau, S. 46 Hesse, H.-G. 18, 100, 108, 118 Kloss, H. 266
Hewitt, R. 8 Knops, U. 233
G Hickey, R. 195, 202 Koerner, E. 76
Gal, S. 127 Hill, M. 83, 84, 87, 90 Kokot, W. 47
Gao, Y. 278 Hinskens, F. 19, 231, 232, 234, Kook, H. 169
García, O. 15, 127–129 236–238, 245, 264 Köpke, B. 260
Gaston, M. 77 Ho, M.-L. 197, 198 Kortmann, B. 204
Geense, P. 266, 267 Hobsbawm, E. 10, 153, 277, 278 Kraaykamp, G. 235
Gerritsen, M. 235, 236 Hoffman, L. 31 Krebs, C. 77
Gibbs, J. 77 Hopf, D. 3 Kremer, L. 232
Giesbers, C. 232 Hoppenbrouwers, C. 231, 237 Kroon, S. 20, 161, 241, 262, 265,
Giesbers, H. 231 Hornberger, N. 134 283, 289
Gijsberts, M. 267 Housen, A. 205 Kroskrity, P. 127, 280
Givón, T. 110 Howell, T. 230 Kruijsen, J. 232
Glick-Schiller, N. 7, 29 Huang, A. 283 Kupisch, T. 18, 105, 116
Goad, H. 105–107, 114 Huebner, T. 53 Kusters, C. 204, 241
Göbel, K. 18, 100, 118 Hult, F. 10, 55 Kuteva, T. 200
Author index
Schleicher, A. 76 Tagliamonte, S. 194, 199, 200, Van Meel, L. 241, 244
Schmid, M. 260, 264 204, 205, 208, 222 Van Oostendorp, M. 237
Schmidt, T. 110 Tan, P. 278 Vandekerckhove, R. 231, 235
Schneider, E. 195, 196, 199, 202, Terkourafi, M. 8 Varis, P. 19, 144, 145, 149, 278
221 Thal, D. 180 Vedder, P. 169
Schnepf, S. 11 Theil, H. 84, 87 Verschueren, J. 127
Schoenmakers-Klein Gunnew- Thomas, J. 82 Vertovec, S. 1–6, 29, 45, 63, 75,
iek, M. 256 Thomas, M. 108 125, 126, 180, 194, 199, 222,
Schönenberger, M. 101, 106, Thomas, W. 12 261, 276
107, 109 Thompson, S. 213, 214 Vigouroux, C. 144
Schouten, M. 230 Thomsen, J. 38 Voeykova, M. 106
Schrauf, R. 9, 17, 27, 28, 31, 36, 41 Thurlow, C. 46 Von Schlegel, F. 76
Schulte-Fortkamp, B. 62 Tops, E. 231 Voortman, B. 230
Schütte, W. 110 Trechter, S. 144 Vousten, R. 232
Scollon, R. 55 Treffers, J. 241
Scollon, S. 55 Treichler, M. 106, 107 W
Semidor, C. 64 Tremblay, M.-C. 99 Waksman, S. 53
Shannon, C. 78, 79, 80–84, Trudgill, P. 204 Walters, P. 278
87, 92 Trumper-Hecht, N. 53 Wang, J. 278
Shi, M. 268, 269 Tsimpli, I. 183 Wang, S. 125
Shirai, Y. 205 Tsou, B. 260 Wang, X. 144, 278
Shohamy, E. 9, 46, 53, 85 Tsui, T. 266, 267 Watkins, R. 77
Siemund, P. 209 Tulp, S. 52 Weaver, W. 78
Silliman, E. 182 Weber, H. 196, 197
Silverman, S. 77 U Weijnen, A. 228, 233
Silverstein, M. 149, 163, 279 Umbel, V. 188 Weinreich, U. 228
Simpson, E. 77–80 Weltens, B. 264
Slobin, D. 181, 195 V White, L. 105–108, 114
Smits, C. 240 Valdés, G 257 Wierenga, A. 231
Smits, T. 232 Vallen, T. 233, 235 Wiley, T. 257
Smolicz, J. 259 Van Alphen, I. 235 Willemyns, R. 235
Snape, N. 18, 105 Van Bezooijen, R. 233 Williams, C. 77
Spitzer, L. 229 Van Bochove, L. 283, 289 Williams, Q. 144
Spolsky, B. 53 Van Bree, C. 231, 238 Winford, D. 204, 205, 218, 220
Spotti, M. 19, 163 Van de Velde, H. 236, 237 Woolard, K. 127
Stanat, P. 11 Van den Broeck, J. 234 Wortham, S. 123, 137
Steever, S. 214, 215 Van der Aa, J. 144 Wu, C.-J. 125
Stevenson, R. 181 Van der Avoird, T. 265
Stroud, C. 67 Van der Craats, I. 165 Y
Sturm, J. 161 Van der Heijden, P. 29, 35, 38, Yagmur, K. 46
Stvan, L. 105 40, 41 Yoon, E.-S. 38
Subbiah, H. 215 Van der Plank, P. 235 Yule, G. 77
Sun, C. 213, 214 Van Ginneken, J. 228, 236
Szmrecsanyi, B. 204 Van Herk, G. 205, 212 Z
Van Hout, R. 227, 230, 233, Žižek, S. 144
T 234, 241
Taeldeman, J. 227, 230, 236, 237 Van Koppen, M. 238, 239
Subject index
Dutch 19, 149, 162, 164–171, 180, French 57, 87, 89, 108, 111, 113, inappropriate article use 110,
228, 231–246, 256, 258–260, 174, 176, 232, 235, 241 112–115, 117, 118
262–270 frequency 80, 88, 91, 100, 108, indefinite article(s) see article(s)
Antilles 164 114, 115, 256, 261 independent variable(s) 28,
based creoles 228 Friesland 234 29, 205, 233, 239 see also
Frisian 231, 234, 235 dependent variable(s)
E functional pragmatics 46, 50, 51 index, indices 77–80, 82–92
educational indigenous 161, 164, 199–201,
settings 2, 187 G 215, 241–243
success 11, 235 German 8, 11–18, 50, 54, 56–61, interaction 205
weight 164, 166, 168 66–69, 86, 87, 89–91, 100–119 system 52, 61
education systems 10, 13 Germany 11, 14, 47, 106, 110 interference 165, 166, 240
emblematic Ghent 237 Internet 144, 150, 152, 275, 276,
features 146–148, 152, 154–156 globalization see sociolinguistics 283
resources 147 of globalization Irish 153–157, 195–198, 201, 221
English 18, 32, 35–37, 40, 41, 57, Italian 50, 57, 69, 87, 89, 92,
67, 86, 87, 100–102, 105–111, H 262, 265
115–119, 133, 134, 196–199, Hamburg Corpus of Irish
212–215, 218–221, 278–282, English 201 K
284–292 Hamburg 13, 15–18, 46–49, 53, Katwijk aan Zee 231
East African 203, 218, 220 59, 66, 67, 75, 85, 90, 110 koineisation 231, 240
Indian 197, 203, 218 Heerlen 239 Kurdish 47, 50, 67, 263
Irish, IrEng 195, 196, 198, 200, heterogeneity 129, 193, 199,
201, 203, 207–212, 218–222 202, 229 L
Scottish 196, 209 heteroglossia, heteroglossic 18, language
Singapore, SingEng 193–200, 19, 123, 124, 128–130, 132–135, acquisition see acquisition
203, 207–210, 212–216, 139 attrition see attrition
218–222 Hindi 60, 89 choice 8, 15, 18, 52, 53, 65, 69,
South African 197 Hispanics 32, 33, 35, 36, 38 128, 241
Standard 199, 203, 221, 291, historical continuity 193, 196, contact 19, 65, 115, 128, 194,
292 198, 200, 221, 222 195, 199–201, 215, 218, 222,
New Englishes 194–197, 199, hybridity, hybrid 20, 128, 157, 230, 240, 241, 245
200, 203, 205, 222 144, 292 development 9, 15, 115, 165,
enoughness 146–150, 153, 157 180–183, 187
enregistering 146, 148, 150 I diversity 13, 75, 76, 78–80,
entropy 78, 79, 81–84, 88, 90 identity 124, 127, 128, 130–132, 83–85, 276
ethnicity 3, 5, 76, 125, 126, 163 139, 143–152, 156, 157, 162, group 53, 67, 68, 258, 259
ethnic 163, 279 hierarchies 161, 175, 275, 276,
diversity see diversity construction 125, 162, 196, 282, 292
group 69, 126, 162, 163, 175, 279 proficiency 7, 14, 28, 31, 35–37,
196, 241–245 practices 144, 146, 147, 157 40, 111, 116, 117, 280
ethnolect 20, 241, 242, 245, 264 processes 144, 146, 277 repertoires 162, 170, 175, 280,
ethnolinguistic 163, 170, 171, 265 see also linguistic identities 292 see also repertoire
Europe 1, 10, 47, 126, 258, 268 ideology, ideologies 124–127, speakers 7, 8, 254–260, 269
European 1, 3, 5, 6, 10, 11, 15, 46, 129, 137, 139, 280 variation 76, 227–230, 233–
75, 84, 143, 155, 235, 268 immigrant(s) 11, 15, 28, 29, 47, 236, 241 see also variation
European Charter for Regional 48, 50, 58, 125, 253, 262, 267, community language 123,
or Minority Language 235 279 see also migrant(s) 127, 175
evenness 77, 84, 87, 88, 90, 91 immigrant dual 13–16
languages 15, 57, 58, 61, 66, foreign 13–18, 56–58, 169
F 69, 127, 259 heritage 7, 10, 13, 14, 115, 125,
Farsi 57, 58, 66, 89 multilingualism 27, 29, 40, 41, 253–258
Flemish 231, 232, 235, 238, 239 50 see also multilingualism model 200
Subject index
replica 200, 220 see also monolingual present perfect 193–195, 204,
replication children 13, 100, 109, 179–181, 205, 207–209, 212, 215, 221, 222
vehicular 68, 69 183–188, 255, 269 procedure(s) 52
legal status 3, 5, 180 habitus 1, 10, 16 Psychology 233, 235, 239
Limburg 230–232, 234, 237, 239 narrative acquisition 181 see Puerto Rico 31
Linguistics 17, 51, 75–77, also acquisition
92, 227–229, 236 see also norms 16, 179, 181–183 R
Sociolinguistics speakers 7, 16, 180, 241, 255 recency 100, 108, 114, 261
linguistic Morocco 164, 171, 263 register 12, 128, 148, 163
diversity 10, 13, 18, 45–47, 52, Moroccan 162, 166–168, 171, regression 28, 30, 40, 204, 236
56–58, 61–63, 76, 86–88, 241–245, 263–265 repeated measures 36, 37, 42,
92, 93, 126, 195 see also morphology 76, 100, 205, 238, 239
diversity 257 repertoire(s) 6, 8, 126–128,
identity 123, 132, 139, 163 see morphosyntax 182, 215, 257 133, 144, 145, 162, 170,
also identity multicompetence 28 262, 279, 280, 292 see also
landscape, landscaping 9, 52, multilingualism 27, 28, 50, sociolinguistic repertoires
53, 56–59, 61, 64, 67, 85, 89 58, 63, 67–69, 88, 100, 124, replication 201, 218–220
soundscape, 126, 276, 278, 282 see also Rimburg 231, 237
soundscaping 45–47, immigrant multilingualism Rotterdam 234, 237, 266
62–65, 69 multilingual Russian 49, 58, 87, 89, 101, 106,
London 2, 124, 130, 131, 206, signs 56, 57, 88 107
283 speakers 2, 7, 8, 89, 128, 219
London-Lund Corpus 203 multiple correspondence S
longitudinal analysis 9, 27, 29, analysis 27, 31, 32, 40 see also selection 199, 200, 221
35–41 correspondence analysis Shannon Weaver Entropy 78, 92
Low Countries 227, 228 multivariate Simpson Index 78–80, 84, 87,
analyses 200, 204, 208, 210, 90
M 216, 236 Slavic 47, 48, 58, 66, 67
Maasbracht 237 methodology 204, 205 Sociolinguistics 193, 199, 227,
Maaseik 234 229, 230, 275–280, 282, 283,
Maastricht 230, 231 N 291–293 see also Linguistics
macro structure 179, 180, narrative 180–183, 185–188, 262 of globalization 277–280,
182–188 development 179–182 282, 283, 292, 293
Malay 196, 199, 215, 253, 270 see also acquisition of sociolinguistic repertoires 162,
Maldegem 238 narrative structures 175, 278, 283
Mandarin 213–215, 220, 259, native speakers 7, 163, 168, 254, Sociology 62, 77, 235
266–268, 281–283, 289 255, 268, 269 Somali 164–166
map 28, 30–32, 34, 40, 41 Natural Sciences 11, 91, 92 Spanish 13, 15, 31, 32, 34, 87, 89,
media 149, 283 negation 205, 211–215, 217–220, 164, 181, 265
mental state terms 183, 184, 222 speech action 50–52, 55, 61
186, 187 Netherlands (The) 240–242, St. Georg (Hamburg) 46–50,
micro-hegemonies 143–146, 262–264, 266, 268, 270, 282 56–59, 61, 65–69, 75, 80, 84–87,
150, 156 Nijmegen 242, 244, 245, 264, 90
micro structure 179, 183–188 268 substrate 195, 199, 208, 213,
migration 2–5, 259, 261, 265 276 norms 16, 144, 162, 180–183 see 214, 241
channels 5, 45, 180 also monolingual norms superdiversity 2–5, 9, 125, 126,
migrant(s) 3, 4, 11, 126, 161, 259, 144, 180, 261–263, 265
262, 282 see also immigrant(s) P superstrate(s) 195, 198, 199, 213,
minority Papiamentu 164, 169 219, 222
languages 88, 127, 253, 292 performance gap 11 syntax 76, 100, 165, 238, 256, 291
group(s) 3, 85, 92, 164, 167, phonology 76, 105, 233, 237, 243
173, 241 Polish 58 T
mobility 280–283, 292 polylingualism 8 Tamil 196, 199, 214, 215
monolingualism 6, 10, 13, 89 Portuguese 13–17, 50, 57, 87, 89 Thai 89
Linguistic Superdiversity in Urban Areas Research approaches
Tilburg 237, 262, 263, 283 United States 29, 201, 258, 262, variant(s) 90, 93, 162, 195, 204,
transfer 106, 107, 114, 181, 182, 291 208, 221, 243
201, 218, 219 urban variation 200–202, 219–222,
transition matrix 36, 38, 39, 41 areas 9, 10, 14, 49, 63–65, 76, 227–230, 234–237, 239–241,
transitivity 207, 209, 210, 83, 194, 222 245
212–216, 218–220, 222 ecosystems 75, 77, 92 varieties 194–196, 198–201,
translanguaging 123, 127, 128, dialects 229–231 209, 215, 218–222, 230–235,
132, 134 Urdu 58, 89 237, 239–243, 265, 266, 276,
transmigration 3, 4 usage-based models 28, 227, 238 279–282, 292
Turkish 13, 15, 47, 50, 53, 57–59, Utrecht 241, 259 vernacular 202, 203, 221, 291
66, 67, 100–102, 105–111, vocabulary 128, 137, 138, 168, 182
113–119, 164, 173–176, 241–244, V voice(s) 123–125, 132–135,
262–265 variables 5, 31, 32, 36, 81, 125, 137–139, 163, 243
194, 212, 229, 232, 235, 236,
U 243, 245 see also dependent W
United Kingdom 124, 125, 265, variable(s), independent Wambeek 238
291 variable(s) Welsh 196