Migration and language

Kerswill, Paul (2006). Migration and language. In Klaus Mattheier, Ulrich Ammon & Peter Trudgill
(eds.) Sociolinguistics/Soziolinguistik. An international handbook of the science of language and
society, 2nd edn., Vol 3. Berlin: De Gruyter.

Migration and language
Paul Kerswill
1. Isolating the linguistic and sociolinguistic consequences of
migration

Language change may be thought of as having internal (intra-systemic), external
(contact-based) and extra-linguistic (socio-political and economic) motivations
(Farrar/Jones 2002, 1). It is reasonable to suppose that the migration of people is a
leading cause of contact-induced change; in other words, migration is a key extra-
linguistic factor leading to externally-motivated change. In every case of migration,
except where a homogeneous group of people moves to an isolated location, language
or dialect contact ensues (Thomason/Kaufman 1988; Trudgill 1986).
Migration also has far-reaching consequences for the social fabric of the three
communities affected: the society of origin, the society of destination, and the
migrants themselves (Lewis 1982, 25, summarising Mangalam (1968)). It follows that
migration has profound sociolinguistic consequences, as the demographic balance of
the sending and receiving populations is altered (migrants are typically young and
economically active), and as the migrants are uprooted from familiar social and
sociolinguistic set-ups, perhaps forming an ethnolinguistic minority which has to
relate sociolinguistically to a new, ‘host’ speech community – which in its turn
becomes transformed by their arrival (Kerswill 1994).
Until recently, the study of the (socio)linguistic effects of migration has been
conceptually separated from that of the spread of linguistic features by means of
geographical diffusion, the simplest manifestation of which is the wave-like spread of
a linguistic innovation from its point of origin. This is an example of the geographical
phenomenon known as expansion diffusion (Britain 2003), and by definition does not
implicate population movements. The study of this type of diffusion was central to the
concerns of 19th and 20th century dialect geographers, later refined by the application
of models from human geography, particularly hierarchical diffusion (Trudgill 1983;

- 1 -

---

Britain 2002, 622-627; 2003). The geographical models applied were asocial in
character (Britain 2003), while the linguistic focus was on individual features viewed
within a mechanistic theory of change (the regularity hypothesis (Hock 1991, 35)). A
parallel research tradition, focusing on pidgins and creoles (Schuchardt 1980) but also
on language islands (Bohnenberger 1913; Schirmunski 1928), investigated the
linguistic consequences of relocation diffusion (Britain 2003), by which cultural
elements (including language) are transmitted to non-contiguous locations by human
migration. From the outset, this tradition has taken social conditions into account,
particularly in pidgin/creole linguistics: Holm (2000, 30) cites Schuchardt’s 1882
study of São Tomé Creole Portuguese, which opens with a discussion of the social
history of the language. It was only recently that the relationship between the two
types of diffusion was acknowledged: both involve the varied psycholinguistic and
sociolinguistic processes resulting from language and dialect contact (Trudgill 1994).
At linguistic borders and within “linguistic areas” (Sprachbünde; Chambers/Trudgill
1998, 168-170), expansion diffusion results in language contact, as do almost all cases
of relocation diffusion. In these circumstances, we may find pidginisation,
creolisation, second-language acquisition, multilingualism, borrowing and language
shift. Dialect contact (Trudgill 1986) is characteristic of expansion diffusion within a
dialect area (L. Milroy 2002). It is also characterises relocation diffusion when the
migrants move to a place where the majority language varieties are mutually
intelligible with their own, or when speakers of different, but related varieties
converge on linguistically ‘virgin’ territory, as in a new town or in many colonial
settlements. Here, the processes involved are second-dialect acquisition (Chambers
1992; Britain 2003; Kerswill 1996), accommodation (Trudgill 1986, 1-38), mixing,
simplification, levelling, hyperdialectalisms and reallocation (Trudgill 1986;
Britain/Trudgill 1999). The extent and manifestation of each of these processes
depends on the nature of the contact and the types of communities involved (Trudgill
2002); however, their manifestations will be more extreme in cases of relocation
diffusion than expansion diffusion (Britain 2003).

In this chapter, I shall treat ‘relocation’ as equivalent to ‘migration’. The focus
will be on the parameters of migration as identified by human geographers. The
linguistic and sociolinguistic consequences of each will be illustrated with examples.


- 2 -

---

2. Migration: parameters and consequences

The importance of migration in human affairs is suggested by the observation that:
“… variation in migration levels between places [tends] to be much greater than
differences in births and deaths” (Lewis 1982, 9). Within a single place, it is possible
that the “net migration change [is] generally of much greater importance than natural
change in its contribution to population change” (White 1980, cited in King 1993, 29).
However, migration is not a single process, and there are “definitional problems once
a precise description is attempted” (Boyle/Halfacree/Robinson 1998, 34), though
there is usually agreement on the parameters that must be examined in describing and
categorising cases of migration. These include:

• space
• time
• motivation
• socio-cultural factors
(Lewis 1982, 9-19; Boyle et al. 1998, 34-38)

We will present each of these separately, along with discussions of their
sociolinguistic repercussions.

2.1 Space
Boundaries

The concept of space in migration studies relates, primarily, to whether administrative
boundaries are crossed. Migration is defined as “movement across the boundary of an
areal unit” (Boyle et al. 1998, 34), whereas a move within an areal unit is, simply, a
“local move” (Lewis 1982, 10). Obviously, the larger the areal unit considered, the
fewer moves will be classified as “migrations”, though this may not reflect the impact
of the moves on the communities concerned. A move across a boundary within a
country is termed internal migration, the people involved being in-migrants to the

- 3 -

---

areal unit, those moving out of it (to whatever destination) being out-migrants (Boyle
et al. 1998, 34-5).
Sociolinguistically, the distinction between moves within and across
administrative boundaries within a state is of little consequence except insofar as the
boundaries reflect, or in some cases shape, differing allegiances. For example, Llamas
(2000) reports that younger people in Middlesbrough in northeast England have
ceased to identify themselves as being “from Yorkshire”, the county to which it had
belonged until boundary changes in 1968. This is reflected in the fact that the local
accent has taken on features from the city of Newcastle to the north, even though self-
expressed identities are aligned not with Newcastle but with the town itself. Whether
these changes were caused by the boundary changes or by, say, economic changes is a
moot point. Where the boundary separates states, significant differences of culture,
economic conditions, education and language may be involved, and the impact of the
migration will be greater. Omoniyi (1999) notes that in Idiroko and Igolo, villages on
either side of the border between Nigeria and Benin, language attitudes differ within
the same ethnolinguistic group, the Yoruba. On the Benin side, the population is more
positively disposed towards Yoruba than are the Nigerians, while often sending their
children across the border to be educated through the medium of English instead of
French. At the same time, the twin villages have attracted numerous incomers, such as
traders, smugglers and money-launderers, in addition to the presence of border
officials.

Distance

Space is also reflected in distance. Short-distance migration differs from long-distance
migration in the degree to which individuals can maintain links with the point of
origin, as well as in the amount of personal commitment (resources, motivation)
needed to move and maintain links. As with expansion diffusion (Hägerstrand 1952;
Trudgill 1983; Britain 2003), gravity models have been applied to migration
(Hägerstrand 1957, cited in Lewis 1982, 51-2). The model is the same, and predicts
that migration flows will be a function of the size of populations at the points of origin
and destination, and the distance between them (Lewis 1982, 53). However,
geographers point out that perceived distance is not the same as Euclidean distance,

- 4 -

---

with a logarithmic transformation to some extent matching economic and
psychological distance. This is in line with the claim that nearby places are seen as
strongly differentiated and those further away as more uniform (Lewis 1982, 50).
While short-distance moves enable existing social ties to be maintained, intermediate-
distance moves are often to a socially similar area and allow new ties to be
established. On the other hand, long-distance moves may involve a very different
environment (Lewis 1982, 51), and establishing new ties will prove problematic
(though cultural differences are small in the case of Europe and distant former
European settler colonies such as those of Australasia, Canada, the USA and parts of
Latin America).
The factor of distance is relevant sociolinguistically, but is not an explanatory
variable because there are a number of intervening variables. Primarily, distance
relates to the extent to which social ties can be maintained, as already noted. Weekly
face-to-face contacts will serve to maintain dialect and language better than annual
“home” visits. However, beyond a certain distance (Lewis rather arbitrarily mentions
1,000 miles (1,600 kilometres)) and a certain level of difficulty of travel to and from
the home region, absolute distance is less relevant: direct contact will be relatively
rare, and the cultural difference between the migrant and “host” groups will be
relatively great. There are, however, a number of intervening variables, including
wealth (reflecting ability to pay for travel). Two factors, in particular, are likely to
have powerful sociolinguistic consequences, overriding distance per se: motivation
and socio-cultural factors (2.3.). Distance seen in isolation from other variables is
unlikely to show anything other than a weak association with language behaviour.
Long-distance, long-term labour migration from less to more developed
countries has been characteristic of the period since c. 1950, and is discussed under
“long-term migration”, below (2.2.). Sociolinguistic issues are taken up there as well.

Direction

Finally, space also involves direction. Historically, mass migration in Europe and
North America has been from rural to urban areas, starting in Britain in the late 18th
century with the Industrial Revolution (Boyle et al. 1998, 5-9). The process was
complex, with a good deal of short-stay migration (circulation; see below), as well as

- 5 -

---

return migration over a lifetime (Boyle et al. 1998, 9; see below). This was followed
in the early 20th century by suburbanisation, with commuting made possible by
improved transport links and the motor-car. From the 1960s, we find counter-
urbanisation, with quality-of-life decisions and industrial relocation playing their part.
However, recent years have seen an “urban revival”, with strong population growth in
US metropolitan areas (Boyle et al. 1998, 14).
Sociolinguistically, the critical directional parameter is that of in- vs. out-
migration, since these alter the demographic balance of the location under scrutiny in
terms of age, socio-economic class, ethnicity, other socio-cultural factors and
language. At the same time, social network densities will change, both for the
migrants and the destination societies, with the result that language change and
language shift may be accelerated.
For Europe and North America, the historical picture is one of considerable,
mainly citywards movements, followed by more geographical mobility through
internal migration and circulation in the shape of commuting. Within present-day
western cities, differences have been noted in the pattern of migration among inner-
city residents (mainly local moves, with no predominant direction) and suburban
residents (moves over a greater distance, outwards from the centre within their own
geographical “sector”) (Balderson 1981, quoted in Lewis 1982, 52). In inner cities,
this pattern allows for the maintenance of close-knit networks as well as non-standard,
localised language varieties (L. Milroy 1980), while more mobile outer-city speakers
are more levelled in the sense of using fewer strongly local features (J. Milroy 1982;
1992, 100-109; see Kerswill & Williams 2000a on dialect levelling among mobile
populations). In Great Britain, the establishment of new towns from the 1950s
onwards led to the possibility of koineised (mixed, levelled and simplified – Trudgill
1986, 127) new dialects (Kerswill/Williams 2000b).
In Europe, initial urbanisation, the loosening of individuals’ network ties
following greater geographical mobility and the formation of new towns are thought
to have resulted in regional dialect levelling or dialect supralocalisation, which can
be understood as the rise of distinctiveness at the wider, regional level at the expense
of local distinctiveness, as well as the emergence of regional versions of the standard
(cf. chapters in Foulkes/Docherty, eds., 1999; Milroy/Milroy/Hartley 1994; L. Milroy
2002; Trudgill 1999; Sobrero 1996; Hinskens 1996; Kerswill 2001; 2002;
Andersson/Thelander 1994; Thelander 1982; Auer/Hinskens 1996, 4).

- 6 -

---

In the developing world, rural–urban mass migration is a phenomenon of the
latter part of the 20th century, with Sub-Saharan Africa the latest region to be affected
(Boyle et al. 1998, 20-23). In West Africa, the dominant sociolinguistic effect appears
to be an increase in individual multilingualism and the spread of lingua francas.
Accra, the capital of Ghana, has seen massive in-migration. This has led to the
indigenous ethnolinguistic group, the Ga, becoming a minority in the city (300,000
out of a population of 2 million (Grimes 2000)), with Akan/Twi now the main lingua
franca with considerable numbers of L2 users. However, among northern migrants in
Accra, Hausa is increasing its use as a lingua franca, reflecting its existing lingua
franca status in the North (Kropp Dakubu 2000). Kropp Dakubu (2001) argues that
the influx to Accra of people from the North has led to a conflict of sociolinguistic
practices: the Ga share with other southerners (including the Akan) the custom by
which visitors and hosts exchange news using spokesmen. This is not practised by
northerners, who thereby remain outsiders. In Maiduguri, in northeast Nigeria, mass
in-migration has led to Hausa being used as a lingua franca, particularly by L2 users,
replacing the indigenous Kanuri. A new form of Hausa, separate from L1 varieties
spoken elsewhere, is emerging (Broß forthcoming).
Extreme political circumstances lead to directional mass migration. The
Spanish occupation of Antwerp in 1585 led to the flight of over half the city’s
population to the western provinces of today’s Netherlands, and has had a lasting
effect on the dialects there (Auer/Hinskens 1996, 18). The resettlement in Germany of
German speakers from the eastern provinces of the former Reich after World War II
led to loss of dialect (Auer/Hinskens 1996, 20), while the post-war migration of
people from eastern to western Poland led to dialect levelling (Mazur 1996).
The relatively short-distance in-migration of rural people to local towns/cities
has been the subject of sociolinguistic research. Bortoni-Riccardo (1985) considers
the qualitatively different networks of Caipira (rural) speakers in Brasilia, Brazil;
Kerswill (1994) considers dialect contact, long-term accommodation, network and
integration among rural migrants in Bergen, Norway; Omdal (1994) examines
attitudes and long-term accommodation among rural migrants in Kristiansand,
Norway. A variation on this sociolinguistic approach is van Langenvelde’s (1993)
quantitative migration-based study of the province of Friesland in the Netherlands.
Here, there is in-migration of Frisian speakers to the towns, leading to a temporary
increase in the number of Frisian speakers there. As these people and their

- 7 -

---

descendants are urbanised, many will switch to Dutch. At the same time, there is
counter-urbanisation led by Dutch-speaking town dwellers; this has the effect of
decreasing the proportion of Frisian speakers in the countryside. The result is the
potential for language shift to Dutch both in the country and in the towns.

2.2 Time

“Migration” implies a degree of permanence in the move; migrant groups tend to be
“committed to the project of living in other people’s countries”, despite in many cases
retaining “diasporic yearnings” for a return to the homeland (Rex 1997, 17). An
absolute definition of migration in terms of temporal patterns is, however, not
possible. Four temporal categories have been recognised: daily, periodic, seasonal and
long term (Gould/Prothero 1975, cited in Lewis 1982, 17-18). Daily movements
include commuting, while the latter three categories involve overnight stays.

Circulation

A further category is usually made cutting across these three: this is the concept of
circulation, which includes “a great variety of movements usually short-term
repetitive or cyclical in character, but all having in common the lack of any declared
intention of permanent or long-standing change of residence” (Zelinsky 1971, cited in
Lewis 1982, 18). Examples of circulation include African nomads and western
business people who spend regular periods every year working abroad (Boyle et al.
1998, 35). To these may be added the European Roma (see Réger 1995 and Matras
2000 for approaches integrating circulation and language contact). Students returning
to their home towns during university vacations may find themselves with dual
allegiances resulting in new dialect-mixing patterns that are not characteristic of the
stay-at-homes (Blom/Gumperz 1972). As both insiders and outsiders, such individuals
form a potential bridgehead for the introduction of innovations or for dialect levelling
(“language missionaries” (Trudgill 1986)).


- 8 -

---

Periodic and return migration

The distinction between circulation and migration “proper” is, perhaps, arbitrary; the
United Nations suggests a residence of at least one year in the host community as
defining migration (Lewis 1982, 18). Turkish “guest workers” working in Germany in
the winter but returning to Turkey in the summer are best regarded as temporary
(periodic) migrants or seasonal workers (Boyle et al. 1998, 35). Guest workers and
their families formed sufficiently large and permanent groups for code-switching
norms to emerge (di Luzio 1984).
Guest workers took employment in Germany expecting to leave permanently
at the end of a work contract; for those who did indeed leave, the “myth of return”
(the “failure to complete the intended migration route” (Boyle et al. 1998, 35)) was
transformed into an actual return. Many migrants who have moved from a poorer to a
richer country return to their place of birth at a later life-stage, either having
accumulated enough money, or on reaching retirement; these are known as return
migrants (Boyle et al. 1998, 35; Lewis 1982, 18). Others return, having failed to find
work or an improved lifestyle, as was the case for many after the US stock market
crash of 1929. The scale of return migration is shown by the fact that one quarter of
those who migrated from Norway to the USA after 1880 eventually returned home
(Engesæter 2002).

Sociolinguistically, periodic and return migrations are significant for the
country of origin and for the migrants themselves. Some 20% of the population of
Puerto Rico are returnees from mainland USA, and 10% of children are English-
dominant. This has led to a conflict between the attitudes of Puerto Rican educators
and commentators, who deplore the mixing of Spanish with English, and the return
migrants’ offspring, who believe that it is possible to combine a Puerto Rico identity
with English dominance (Zentella 1990). A rather different example is the Norwegian
Arctic territory of Svalbard (Spitsbergen), where by law residents must remain
registered as domiciled on the mainland and where no one is permitted to remain after
retirement. The average duration of stay is 10 years, with the result that stable
linguistic norms have not emerged. Families normally spend long summer holidays on
the mainland. Children who grow up there speak using often idiosyncratic dialect
mixtures, and express dual allegiance to Svalbard and to their official domicile in
Norway (Mæhlum 1992).

- 9 -

---

Long-term migration

The distinction between return and quasi-permanent migration rests on whether the
individual actually enacts the myth of return. However, there may be no intention to
return. This will be true for religious minorities, such as the German Mennonite
communities in North America (MacMaster 1985; Kraybill 1989) and Russia, which
have practised extensive cultural separation over two centuries or more and have
formed “language islands” (Sprachinseln). But by far the largest category is that of
the migrant with miscellaneous, though mainly economic motivations. The mass
migration from Europe to the USA in the 19th and early 20th centuries is the clearest
example of the intention to establish a “new life”, with no return envisaged.
We can see the circulatory or seasonal migration patterns noted above
merging, over time, with long-term migration. Starting in about 1950, western Europe
saw large-scale long-distance unskilled labour migration from its former colonies
(North Africa, Sub-Saharan Africa, India, Pakistan, the West Indies), from eastern and
southern Europe (Italy, Yugoslavia, Greece) and from Turkey (White 1993). These
migrants were mainly men who came without families, and many, particularly in
Germany, Austria and Switzerland, were on fixed contracts. As Giddens (2001, 260)
points out, this reflected not only macro-level economic circumstances, but also
micro-level decisions taken by individuals, who used information gained from family
and friends and the promise of a support network to inform their decision to migrate
to particular countries and towns. Thus, Turks tended to migrate to Germany, in a
series of chain migrations (Boyle et al. 1998, 36) by which individual “pioneers”
were followed by others who knew them. A well documented case of chain migration
is that of the Sylheti-speaking Bangladeshi community in the London borough of
Tower Hamlets, where their 17,000 children form over 50% of the school population
(Gregory/Williams 2000, 38-9, 154).
From about 1970, restrictions began to be placed on immigration, and a
second stage of migration followed, that of family reunification (White 1993, 49-50).
Initially, agencies such as employers and governments determined the source of the
migrant workers and made provision for their housing. The subsequent reduced need
for unskilled labour coincided with the obligation to support arriving families. With

- 10 -

---