Language Shift in Banglatown?
Evaluating Ethnolinguistic Vitality in East London
Sebastian M. Rasinger*
In this paper, I provide a first evaluation of the Ethnolinguistic Vitality of the Bangladeshi Community in East
London. Despite a considerable number of social, economical and educational research on the Bangladeshi
community, an in-depth study of the community’s vitality has not been carried out yet. This paper summarizes
the findings of two preliminary studies in the borough, and provides first results as well as proposals for future
In particular, I focus on the use of both the first language (L1) Sylheti and the second language (L2)
English in two contexts of interaction: firstly, the home domain, that is, language used in the home/family
environment; secondly, the outside home domain which comprises patterns of daily interaction outside the home
The results indicate a clear distinction of language use between certain domains, with English primarily
used in formal domains, such as administration, and Sylheti used in most other domains of daily interaction.
Nevertheless, data suggests that language use gradually shifts towards the L2, particularly in the home domain. I
argue that this shift is likely to be introduced by children.
The framework of Ethnolinguistic Vitality (henceforth EV) and its application to sociolinguistic
research is by no means new. Since one of its first applications by Giles et al (1977) and Fishman
(1977), EV has been in the focus of attention for linguists and social psychologists. Recently, the
Multilingual Cities Project (MCP) has evaluated in some detail the situation of immigrant
communities in six European cities (Extra and Yagmur, 2004); however, London, with its considerable
immigrant population, has not been included in the MCP.
This paper aims at providing a first evaluation of the ethnolinguistic vitality of the
Bangladeshi community in the East London borough of Tower Hamlets. In particular, based on
qualitative and quantitative data obtained during two studies between 2002 and 2004, I will suggest
that while respondents’ first language Sylheti is the language predominantly used in daily life and
social networks are primarily Bangladeshi-centric, language use gradually shifts towards the use of the
second language English, and interaction becomes increasingly interethnic over time. However, these
developments are clearly restricted by external variables.
* University of Sussex, Department of Linguistics and English Language, Falmer, Brighton BN1 9QN, UK.
2. Setting the scene: Framework and sociocultural context
The concept of Ethnolinguistic Vitality allows to analyse language use and change on a macrosocial
level (Landry and Allard, 1994: 15). It comprises sociostructural factors to explain language
maintenance and shift within a community. According to Giles (1977: 308), “[t]he vitality of an
ethnolinguistic group is that which makes a group likely to behave as a distinctive and active
collective entity in intergroup relations”. That is, the higher the vitality of an ethnolinguistic group,
the higher its chances of survival; the lower its vitality, the higher the risk of its ceasing to exist (1977:
308). In its original version, Giles et al. use three main categories of variables which influence the
vitality of an ethnolinguistic group: firstly, the group’s status contributes to its vitality; the higher its
economic, social and sociohistorical status is, the higher the group’s status as a collective entity (Giles
et al., 1977: 309-311). Furthermore, the status of the group’s language may prove advantageous or
disadvantageous: minority groups whose language is of internationally high status, for example, the
French speaking minority in Quebec, have a clear advantage over minority groups whose language is
internationally hardly recognized (Giles et al., 1977: 311).
Secondly, demographic factors play an important role in generating high group vitality. Large
absolute numbers as well as a high proportion of the overall population support a group’s vitality
(Giles et al., 1977: 309, 312-15). Also, the proportion of group size and space occupied by the group
may influence its survival:
“Minority group speakers who are concentrated in the same
geographical area may stand a better chance of surviving as a
dynamic linguistic community by virtue of the fact that they are in
frequent verbal interaction and can maintain feelings of solidarity.”
(Giles et al., 1977: 313)
Thirdly, institutional support can strengthen Ethnolinguistic Vitality. Formal support can be received
from official institutions, in some cases the respective government directly, while informal support
refers to community-internal institutions, such as pressure groups (Giles et al., 1977: 315-316).
The framework was subject to several amendments and modifications. In particular, it has
been criticized that although the framework allows to objectively measure vitality, that is, from outside
the group, it does not necessarily allow conclusion about the group’s perception of its own vitality.
That is, although the objective vitality of a group might be low, its own subjective perception of its
vitality might be much higher – an important factor in intergroup relationships (Kraemer et al., 1994:
79; Ytsma et al., 1994: 63). Ros et al. (1994) in their study on ingroup bias amongst groups of
different languages in Spain (Castilian, Catalan, Basque, Galician) found that high ingroup vitality
leads to more persistent ingroup bias – while low vitality causes the opposite. Kraemer et al.’s study
on vitality perception and attitudes within the Arab minority in Israel also provided evidence for a
strong relationship between the two variables. As such, the concept of a Subjective Ethnolinguistic
Vitality is closely related to that of a social identity (Ros et al., 1994: 145; Ytsma et al., 1994: 63).
Previous research suggests that an evaluation of the vitality of the Bangladeshi community in East
London is difficult. With about 65,000 people, the Bangladeshi community is relatively large, and
makes up for around 33 percent of the borough’s population (National Statistics Office, 2001); Within
London, 43 percent of Bangladeshis live in the borough. Eade et al (1996: 158) point out that “nearly a
quarter […] of the British total of Bangladeshi residents live in a single London borough”. In addition,
the community consists of close-knit intraethnic social networks (Husain, 1991: 83). Hence, one may
argue that the demographic factors contribute positively to the vitality of the community. In addition,
the community receives considerable support, ranging from mother tongue projects and Bengali
mother tongue classroom assistants, to specifically designed EFL courses and Bengali/Englishi
translation services with focus on particular domain such as healthcare or administration. Tower
Hamlets College offers Bengali GCSE and A-level courses (Tower Hamlets College). Local
authorities provide various services in Sylheti.
However, the group’s status, both socially and, more importantly, economically, provides a
different picture. Research indicates that, from the very beginning of the settlement in London,
Bangladeshis face great social and economic disadvantage, leading to a stark cleavage between
Bangladeshi and white residents (Eade, 1989: 29). In particular the disappearance of the shipping and
docking industry in the East End had a dramatic effect on the Bangladeshi community in terms of high
unemployment and resulting socioeconomic difficulties (Carey and Shukur, 1986: 409); White (2002:
9) reports that around 20 percent of Bangladeshi men who are economically active are unemployed;
unemployment is a particular problem amongst young Bangladeshi men: over 40 percent of male
Bangladeshis under the age of 25 are without employment.
Economic disadvantage is mainly based on low levels of formal qualification. White
concludes that of all ethnic minorities in Britain, Bangladeshis were most likely to be unqualified, with
48 percent of women and 40 percent of men being without formal educational qualification (2002: 12).
This preliminary analysis provides an ambiguous picture of the vitality of the Bangladeshi
community: While demographic data and information on community support indicate that the vitality
is relatively high, socioeconomic variables paint a less positive picture. The main part of this study
will hence focus on the internal organisation of the community, with particular focus on whether and
to what extent the Bangladeshis behave as an “active collective entity” (Giles et al., 1977: 308).
Respondents were asked to fill in a questionnaire, comprising 26 questions with 59 variables
overall. The majority of questions were taken from the ethnolinguistic studies mentioned above, and
were adapted to the needs of this study. The sample comprised 16 respondents. The questionnaire was
divided into six thematic blocks, covering biographical and economic data, data on language use in
both the home and outside home-domain, as well as a question evaluation Bangladeshi self-perception.
Five-point Likert scales were used for quantitative measurement.
Table 1: 5-point Likert scale with allocated adjectival description
A previous study (N=20) aimed primarily at obtaining qualitative data from Bangladeshi respondents,
gained through semi-structured ethnographic interviews; in addition, this first phase of the project was
designed as a pilot study for obtaining quantitative data about community structures.
3.1. Family structures and language use in the home domain
One of the foremost questions of this study was whether the patterns of family structures and language
use within families reported in literature could also be found in this sample. Research in Bangladeshi
communities in London and throughout the UK has shown that family structures bear a striking
resemblance to those in rural Bangladesh: Usually, Bangladeshi households in the UK comprise
several families, spanning several generations; the average size of a Bangladeshi household is
considerably above the British average (5.3 compared to 2.5 people)(Eade et al., 1996: 153). Roles in
families are clearly distributed, with men working in paid employment and women being responsible
for domestic tasks and the upbringing of the children. In addition, similar to Bangladesh, Britain’s
Bangladeshi communities are characterised by close-knit social networks (Husain, 1991: 81).
Marriages take place mainly intra-ethnically, and the vast majority of Bangladeshi women living in
Britain are married to a husband of Bangladeshi origin (Phillipson et al., 2003).
Fourteen of the 16 respondents are married, 13 live in intra-ethnic marriages (92.9 percent), 1
respondent (7.1 percent) is married to an ethnically British partner. A similar picture emerged during
the pilot phase of the study: again, the majority of married informants were married to a Bangladeshi
partner. This supports previous results according to which marriages are predominantly intraethnic. In
addition, both studies indicated that most respondents live with an extended rather than a nuclear
family, a result that was confirmed during the ethnographic interviews: while families do not
necessarily share the same premises, like the homesteads in Sylhet described by Gardner (1995), they
often live in close proximity, such as a ward or estate.
The analysis of language use at home provided similar results in both samples; I will,
however, focus on the main sample here. With a mean of x=4.75 out of a 5-point Likert scale, Bengali,
or rather Sylheti, is the preferred language used at home, compared to a mean usage of English of
x=3.19 points. The respective medians are 5 (Sylheti at home) and 3 (English at home). Also, as
before, the distribution of values clearly shows a preference for Sylheti: the minimum value for Sylheti
is 4 (“much”), while the minimum value for English is as low as 2 (“rarely”). Similarly, no respondent
indicated using only English at home; the maximum value for English is 4 (“much”). However, all
respondents reported using English at home to some extent, with no respondent indicating a Likert
score 1 (“never”) for English.
A Wilcoxon Signed Rank test was conducted to explore whether the difference in language
use is statistically significant. Thirteen respondents use more Sylheti at home than English, while three
respondents use more English. With =0.001, the difference between use of English and use of Sylheti
is highly significant, that is, we can indeed claim that Sylheti is the preferred language used in the
home/family domain. Again, this result was supported by the results of the pilot study.
Communication within the ethnically almost exclusively Bangladeshi home/family domain is mainly
conducted in the L1. This is not surprising: it seems natural that an ethnically homogenous group uses
its L1 rather than L2 in the home/family domain. Nevertheless, a considerable proportion of
respondents indicated that they use English at least sometimes at home: 11 respondents (68.8 percent)
indicated they use English “sometimes”, that is a Likert score (LS) of 3 out of 5, at home, an
additional 4 (25 percent) even reported using English “often” (LS 4). In sum, 98.8 percent use English
“sometimes” (LS 3) or “often” (LS 4), only 6.3 percent use English “rarely” (LS 2) at home.
The analysis also provides interesting insights into which languages are used when addressing
children within the home/family domain, and also which languages children use to address their
parents or each other. Twelve respondents (75 percent) have children; eleven of them have children
who are either currently attending a school, or who were in full-time education in the UK. That is, we
can assume that children have significant English language input from both school and peers, and we
may imply that children have a relatively high English language proficiency.
As would be expected, Sylheti is the preferred language to address children, with 41.7 percent of the
respondents indicating they use Sylheti “always” (LS 5) and a further 33.3 percent of parents using the
Sylheti “often” (LS 4); in sum 75 percent of respondents use predominantly their L1 to address their
children. However, a remarkable 91.7 percent of respondents reported using English “sometimes”
(66.7 percent) and even “often” (25 percent). Only 8.3 percent indicated that they use English rarely.
Moreover, one quarter of the informants indicated they use Sylheti only “sometimes” when addressing
their children. Given that generally no other languages are used in spoken family discourse, it seems to
be the logical consequence that English is being used instead. Unequal group sizes did not allow for a
direct comparison of the two groups (with/without children).
We can thus conclude that if English is used at home, it is predominantly used to address
children, while Sylheti is the main language of communication within the home/family domain in
general. The question remains, is there any evidence about what causes the increased use of English in
communication with children? Two directions of causality are possible: first, parents use English in
order to actively encourage their children to speak English and hence to facilitate their children’s
English language proficiency, which is considered important (cf. Kershen 2000). Secondly, children
could increasingly use English at home, possibly up to a complete refusal of communicating in
Sylheti, hence forcing their parents to use English. This equivocality provides room for further
investigation in a separate study.
3.2. Interactional Patterns and Language use outside home
Having analysed language use within Bangladeshi families to some extent, the remainder of this paper
focuses on interactional pattern outside the family domain. Language use outside home was divided
into three sub-domains: first, language used in informal situations, such as shops, markets etc, where
language is presumably relatively informal. It was assumed beforehand that encounters in these
situations are likely to be with both native speakers of English, Bangladeshi speakers of English, and
speakers of other languages.
Second, language use in formal situations, such as in banks or at hospitals, was explored.
Unlike the first context, it was assumed that the proportion of Bangladeshis (or native speakers of
Sylheti) is considerably lower. Encounters may involve a considerable proportion of native speakers of
English, hence forcing Bangladeshis to use the L2 – simply because Sylheti is not spoken during these
The third question focuses on the language used in all other situations outside the home/family
domain, and hence allows a rather broad spectrum of answers. However, given the dense Bangladeshi
network in the area, the majority of encounters might be intraethnic, that is, between Bangladeshis.
Language use at home was measured by means of one 5-point scale (that is, potential maximum score
is 5), language use outside home was measured by means of 3 questions with a 5-point Likert scale
each, allowing for a maximum of 15 points (that is, score 5 in all three sub-questions).
It is noticeable that unlike in the home/family domain, English is the language used more often
outside home: while 50 percent score 14 and more out of 15 points for using English outside home, the
median for Sylheti is 8 (out of 15) points. As before, a Wilcoxon Signed Ranks test was conducted.
The results indicate that English is used significantly more often outside the home/family domain than
Sylheti ( =0.012); 11 out of 16 respondents use more English than Sylheti outside home, four
respondents use the L1 more frequently.
The above provides a first insight into language use outside the home domain. It was of further
interest whether differences can be identified between the different sub-domains, namely informal
(shopping etc), formal (banks, hospital etc) and any other context outside home/family. It is notable,
though not surprising, that the largest difference between the use of English and Sylheti occurs in
formal situations: 50 percent of the informants indicated they use only English in formal situations
(score 5); for Sylheti, the median is considerably lower, with 50 percent reporting to use Sylheti rarely
(score 2) or not at all (score 1) in formal situations. According to a Wilcoxon test, the difference is
highly significant with ρ=0.007.
In informal situations, English is also used to a significantly higher extent than Sylheti. This is
not surprising: although Bangladeshis are the second largest ethnic group in the borough, much of the
commercial infrastructure is dominated by other ethnic groups, such as Pakistanis (see Kershen, 1997).
Hence, communication between groups with different mother tongues will inevitably take place in the
most common vernacular, English.
The picture is remarkably different for the less defined “other situations outside home”-
domain: the medians for both English and Sylheti are almost identical with 4 and 3.5 respectively.
This difference is statistically not significant. The result is open to interpretation, but the most likely
and reasonable explanation is that English is predominantly used in situations where encounters with
other ethnic groups are likely to happen (shops, market, bank, hospital), that is, domains which are less
Bangladeshi-centric. Based on socioeconomic research in the area, I assume at this point that the
“formal“ domain is dominated by ethnically British people, or at least people who are native speakers
of English or people with a very high command of English. The language of communication in these
contexts would then inevitably be English. The “shopping” domain, although dominated by non-
Sylheti speakers, still requires a high proportion of English usage; however, the proportion of
encounters between Bangladeshis is likely to be higher than in the “formal”-domain.
Lastly, I suggest that the “anywhere else outside home”-domain mainly consists of intra-ethnic
encounters between Bangladeshis, that is, refer to encounters between people who belong to the core
of the social network, such as friends or acquaintances. As a result, the L1 is used as the predominant
language of communication. Other languages are hardly ever used in discourse outside home, and
have not been included in the analysis.
In summary, we have seen that both contact with ethnically British speakers and the use of
English seem to be restricted to a limited number of domains, namely formal-administrative contexts,
which are likely to be dominated by ethnically British people and/or native speakers of English, and
the more informal “shopping”-domain. In other words, these are domains where the interlocutor
cannot be freely chosen, but is determined by the context (for example, bank clerk, shop assistant,
hospital nurse). If respondents have a choice of interaction partners, they seem to tend towards
interlocutors of their own ethnic background.
3.3. Deliberate dissociation or lack of opportunity?
This hypothesis of deliberate isolation, that is, the argument that Bangladeshis deliberately dissociate
themselves from other groups may not be tenable. In fact, many informants reported in the course of
this project that they were actively seeking contact with British people in order to integrate into the
British society and larger community. The following passages are excerpts of the ethnographic
interviews conducted in the course of this project.
In particular, the lack of sufficient competence in communicating in English with either native
or fluent speakers of English seems to constitute a major problem for speakers of lower proficiency;
learners are willing to use English, but are hindered by insufficient L2 competence:
I like it. I’m interested in English language. Speaking, learn speaking. I like
the English language but not speaking very good.
My mother language is Bengali. And (all the) time Bengali speaking. Now,
one and half years (try to) English speaking…
Listening comprehension, and in particular the comprehension of local accents and dialects seem a
major difficulty for speakers of lower proficiency:
Is that a problem you have? That people talk too fast?
Yeah, we have problem. We can't catch they are/ what are they saying.
MMA: Also, we feel problem. When/ Because we can their accent exactly. But- but
we can understand teachers said to us anything (… …) But we can't
understand local languages the difference.
MMA: But- think so- if I try to understand- (… …) Listening is the great problem for
(me). I can speak English something. I'm not good at all English. But I can
speak English. But I'm not good listener. This is a problem for us. (…)
English is hardly used in daily discourse and its usage is often restricted to the EFL classroom and
Do you have any problems in your daily life? Where you have to speak English? Do
you get along or is it very difficult?
This class (for two hours) And it not enough. I want to more- more time.
Do you use English only in the class or do you use it…?
Only- no sometimes (reading for book) (television watching television)
I use in this class and my work and shop and bars (… …)
In particular speakers with lower L2 proficiency almost always mentioned lack of English language
practice together with lack of contact with native speakers of English, although better contact and
practice is highly desired, and informants seem keen on communicating cross-ethnically (or cross-
linguistically) in order to both improve their L2 and facilitate their social networks beyond their own
How many hours a day so you usually spend with English people?
(…) (I) Can’t find English people. (… …) I can’t talk with English people.
So you don’t speak much English…
Sometime is speaking for my child. (.. ..) I like talk to English.
You don’t have English friends?
I have no friend- English friend.
MA: I like English people, I like Bengali, both of them
Do you have much contact with people who speak English? At work, or- Do you have
erm- much- or- in your everyday life do you use the English language quite often or-
MMA: No, no, because- because just when I stay my customer- customers, than I speak
English. And when I'm my living room- my living room (… …) then there I speak
Bengali- Sylheti. They don’t know any others languages. They only Bengali language.
This a great problem for us.
What is it that it's- is it difficult to make contact to English people?
MMA: Sometimes, sometimes.
Yes, we don’t understand what they saying. And we know a little bit. So- when I go to
bank or any office or anywhere erm- I don’t speak much because in- there were some
Bengali people they can underst/ they maybe, you know- they maybe undermined [sic]
me because I can't speak very well.
MMA: Very problem. And- sometimes I feel problem about customers. When they
speaks their local languages-
Interestingly, informants with higher L2 performanceii seem to perceive the situation differently. AZ
reports that she has regular contact with people of other ethnic backgrounds, who she mainly meets at
the community centre she visits regularly. Also, L2 comprehension is less of a problem, and
interaction between native and non-native speakers takes place more effectively:
Do you have– may I ask you– do you have many English friends, or…? Do
you have– your friends, are they more other people from Bangladesh, or are
they also British people…or…?
No. There is more closely working with British people, Mexico (…) British,
Bangladeshi, Pakistani, Indian—mix, you know?
Are people- English people are usually helpful when they realize that you
Yeah. They're really helpful. They can understand you a little bit, then they
cover rest of them. [LAUGHS]
A particularly remarkable statement was made by UIR and other women who were interviewed as a
group at Tower Hamlets College in the course of the first part of this project: although all respondents
reported to be content with their situation in Tower Hamlets and their integration into an almost
exclusively Bangladeshi network, they also expressed a certain discontent about the strong
Are most of your friends and acquaintances Bengali, or do you have much contact
with English people?
This area is too much Bengali. And people just friendly. Bengali. No English
In fact, the group repeatedly stated that there were “too many Bangladeshis in the area” which led to
“not enough contact with English people”. Although this had not been mentioned so explicitly by
other informants, we have seen that the lack of interethnic interaction has been criticized before:
(…) (I) Can’t find English people. (… …) I can’t talk with English people.
It is obvious that there is indeed the desire to meet and interact with non-Bangladeshi people.
However, interethnic contact takes place only infrequently. Lack of English language proficiency is
one possible explanation. We may imagine situations during which contact between native speakers
and non-native speakers of English is initiated, but subsequently fails due to problems in
comprehension. This, in turn, may lead to discouragement of the non-native speakers, as implied by
Whenever you talking, you know, by the time, you talk more. But otherwise
you feel ohh… I’m scared, what (am I) gonna say, what (am I) gonna say.
Moreover, the dense social network within the Bangladeshi community makes interethnic contact not
imminently necessary. Family and social life is mainly centred around people of the same ethnic and
linguistic descent, providing a safe and stable social environment. Although contact with other ethnic
groups, in particular with the dominant British, might be desirable, it is not immediately vital in order
to ensure a stable social network.
3.4. Identifying underlying factors
So far, I have shown that Sylheti is the preferred language spoken in the home domain; English is used
only occasionally. At the same time, it has been illustrated that English is increasingly used in families
with children. In addition, this paper has so far evaluated intra- and interethnic contact, and I have
attempted to sketch a model of a social network for the Bangladeshi community.
One aspect is yet to be addressed, and that is the change in language use and contact over time.
The following section seeks to address the question whether the use of English in both domains (home
and outside home) changes over time. In addition, the extra-linguistic factors gender and children are
considered in the analysis.
In a first step, a Spearman rank correlation was carried out to evaluate whether Length of
Residence (henceforth LoR) correlates with language use at home and outside home, and with contact
with British or Bangladeshi people. In order to increase the chance of detecting a correlation, the 5-
point-scale questions “How much time a day do you interact with British/Bengali people?” and “How
many of your friends are of British/Bengali origin?” were merged into two new variables: Sum of
contact British, and sum of contact Bengali. Each variable has a 10 point scale.
While the general analysis provided no significant results, two constellations of variables
approached significance: contact with ethnically British people increases over time with r=0.416
( =0.061), while simultaneously the use of English outside home increases slightly (r=0.82, =0.381).