- 1 -
In J. K. Chambers, P. Trudgill &N. Schilling-Estes (eds.)
The handbook of language variation and change
.
Oxford: Blackwell. pp. 669-702.
Koineization and accommodation
Paul Kerswill 11/9/00
Koineization as language change
Several chapters in this volume [those by Gordon, Bailey, Klemola and Britain] deal with
language change, from a number of perspectives. In this chapter, we discuss
koineization
, a
contact-induced process that leads to quite rapid, and occasionally dramatic, change. Through
koineization, new varieties of a language are brought about as a result of contact between
speakers of mutually intelligible varieties of that language. Koineization is a particular case
of what Trudgill, in his 1986 work, calls âdialect contactâ; typically, it occurs in new
settlements to which people, for whatever reason, have migrated from different parts of a
single language area. Examples of
koines
(the outcomes of koineization) include the
Hindi/Bhojpuri varieties spoken in Fiji and South Africa, and the speech of ânew townsâ such
as Høyanger in Norway and Milton Keynes in England. Dialect contact, and with it
koineization, is one of the main external causes of language change â âexternalâ here referring
to social factors, in this case migration, which can reasonably be expected to promote change.
Contrasted with this are âinternalâ factors, which have to do with aspects of the structure of a
particular language (its phonology and its grammar) which, perhaps because of structural
imbalances, are predisposed to change.
Because koineization can take place relatively swiftly (though it is probably more
gradual than pidginization â see Siegel, forthcoming), a central theme of this chapter will be
the immediate mechanisms of change rather than the description of longer-term trends that
take place over a century or more (like the English Great Vowel Shift or the rise of the
- 2 -
auxiliary âdoâ). We will be asking questions like, âAre permanent language changes
prefigured in the utterances of the people whose speech communities are undergoing
change?â; âIs it children, adolescents, or adults who are the main agents of change?â; âDo the
social network characteristics of the migrants have an effect?â; âDoes it matter whether the
contributing dialects are very different or very similar?â; âHow long does it take for a koine to
emerge?â; âAre there circumstances in which dialect contact doesnât lead to the formation of a
koine?â; On the more âlinguisticâ side, we shall be asking: âWhich features found in the
melting pot of the early stages of koineization survive in the koine, and which are lost?â; âAre
there particular characteristics of these features that leads to the one outcome or the other?â.
Koineization, as we shall see, typically takes two or three generations to complete,
though it is achievable within one. It is in principle possible for us to observe specific cases,
though this has (to my knowledge) not been achieved for the complete process. Thus, the
literature contains detailed descriptions of koines (the final outcome of koineization) from a
number of parts of the world, together with conjectural reconstructions of the social and
linguistic history of the speakers who contributed to the koine. In the literature, we can also
find a very small number of descriptions of the inception of koines, with direct observations
of the first generation of speakers in new locations. For one established koine, the English of
New Zealand, we even have recordings of the offspring of the original English-speaking
immigrants (albeit as elderly people) to compare with the modern form of the language.
Before we go on, I will outline the general model of language change within which I
shall be operating. Labov (1972) shows that language variation is systematic, in that it can be
related to social divisions within a community, such as class and gender. Change can be
shown to originate with particular social groups based on these divisions. However, a number
- 3 -
of linguists have recently argued that language change lies with the individual (J. Milroy,
1992, Croft, 2000). Thus, the only circumstance under which language change may result is
when the collective use of a new linguistic feature by individual speakers is sufficiently
frequent to be taken up as a new norm. This position need not conflict with that of Labov,
since these individual-speaker behaviors take place against the backdrop of larger social
structures. As we shall see, the individual-as-agent-of-change approach is particularly
relevant in the case of koineization, because this is a process which starts with the first
generation of incomers adapting their speech to the other speakers they encounter. This
adaptation is an example of
speech accommodation
, a research area to which we will return.
- 4 -
Mixing, leveling, simplification and reallocation in established koines
The term âkoineâ (whose Greek meaning is âcommonâ) was first used to refer to the form of
Greek used as a lingua franca during the Hellenistic and Roman periods (Siegel, 1985, p.358;
Bubenik, 1993). It arose as a mixed vernacular among ordinary people in the Peiraieus, the
seaport of Athens, which was inhabited by Greeks from different parts of the Mediterranean
(Thomson, 1960, p.34, quoted in Siegel, 1985, p.358). This kind of âkoineâ is, of course,
rather different from the examples given in the previous section, in that it is not a new variety
used as a vernacular, but rather a compromise dialect used for communication between
speakers of other Greek varieties. This Koine later became the language of the Macedonian
empire, and was widely used as a second language, though it did acquire some native
speakers (Thomson, 1960). According to Siegel (1985, p.358), the Koine was characterized
by
reduction
and
simplification
. âReductionâ refers to âthose processes that lead to a
decrease in the referential or non-referential potential of a languageâ (MĂźhlhäusler, 1980,
p.21), involving, for example, a reduced vocabulary or fewer stylistic devices. To judge from
the recent literature, reduction is not pervasive in koines, though, as we shall see, it may be
present. However, it is a defining feature of pidgins, whose genesis is very different from that
of most koines. (Similarities and differences between these two kinds of contact varieties will
be explored in the final section.) âSimplificationâ, which is a notion we will return to
repeatedly, refers to âeither an increase in regularity or a decrease in markednessâ (Siegel,
1985, p.358, quoting Mßhlhäusler). In practice, this means a decrease in irregularity in
morphology and an increase in invariable word forms (Mßhlhäusler, 1974, cited in Trudgill,
1986, p.103), to which can be added the loss of categories such as gender, the loss of
morphologically marked cases, simplified morphophonemics, and a decrease in the number
of phonemes. Siegelâs recent definition is a useful, very general, reference point:
- 5 -
A koine is a stabilized contact variety which results from the mixing and subsequent
levelling of features of varieties which are similar enough to be mutually intelligible,
such as regional or social dialects. This occurs in the context of increased interaction
or integration among speakers of these varieties. (Siegel, forthcoming)
As Siegel (1985) points out, the term âkoineâ has been variously used to refer to
different aspects of mixed, compromise languages â their form, their function, and their
origin â and there has been disagreement as to what should or should not be included in the
definition. Two categories stand out, already alluded to above:
regional koine
and
immigrant koine
. The original Koine was at first a regional koine, which did not replace the
contributing dialects; on the other hand, a new dialect in a new settlement is an immigrant
koine, which, once established, becomes the vernacular of the new community, replacing the
regional dialects of the original migrants â though not, of course, having any effect on the
dialects in their place of origin.
Between these two categories we find
regional dialect leveling
, which, as we shall
see, shares certain important properties with koineization. âRegional dialect levelingâ refers to
the decrease in the number of variants of a particular phonological, morphological or lexical
unit in a given dialect area, and should be distinguished from
diffusion
, which is the spread
of linguistic features across a dialect area. Leveling leads to a reduction in differences
between dialects and hence a gradual homogenization of the vernacular speech of a region.
For example, in many parts of Italy new regional varieties have emerged, usually centered on
a city. Linguistically, they are a compromise between a number of local dialects and the
standard language. Some scholars, such as Sobrero (1996, p.106), actually refer to these as
- 6 -
âkoinesâ, a use of the term which has some justification since there is evidence that Italian
âkoinesâ do not necessarily supplant the local dialects, with speakers regularly switching
between dialect and koine (Trumper & Maddalon, 1988). By far the more usual case is
dialect leveling entailing the loss, or at least attrition, of dialects. This is widespread in
modern Europe (Auer & Hinskens, 1996; Hinskens, 1996, 1998; Sandøy, 1998; Thelander,
1980, 1982; Williams & Kerswill, 1999) as well as elsewhere (see Inoue, 1986, on recent
changes in Tokyo). Regional dialect leveling may lead to varieties that resemble any koines
that may be spoken in the same region, particularly with respect to simplification â a point we
shall return to.
We will be concerned mainly with immigrant koines, or, to use Trudgillâs term,
new
dialects
(Trudgill, 1986, p.83). In this section, I shall outline some of the key features of
established koines, before, in the remaining sections, tracing the stages through which a
potential koine must pass if it is to reach stability. According to Trudgill (1986, p.127),
koineization is composed of three processes:
mixing
,
leveling
, and
simplification
.
(Elsewhere in his book, he refers just to leveling and simplification â a fact that is
unproblematic since leveling can only take place if, in the new speech community, there has
been prior dialect mixing leading to the presence of more than one form for a particular
linguistic category, such as a vowel, a pronoun, or a suffix.) In koines, we also find what
Trudgill has called
reallocation
, which is defined thus: âReallocation occurs where two or
more variants in the dialect mix survive the levelling process but are refunctionalised,
evolving new social or linguistic functions in the new dialectâ (Britain & Trudgill, 1999,
p.245; cf. Trudgill, 1986, p.110). We turn now to the first of our examples.
- 7 -
One of the major population movements of the late nineteenth and early twentieth
centuries was the shipment of people from the Indian subcontinent to work as indentured
laborers in the European colonies (Mesthrie, 1993). This resulted in new varieties of Indian
languages, particularly Bhojpuri (a Hindi variety of north-east India), being established
across a wide region ranging from the West Indies and the Caribbean to South Africa
(Mesthrie, 1992) and Fiji (Siegel, 1987; Moag, 1977). Table 1 illustrates the mixed nature of
the koine known as Fiji Hindi in one area of its grammar.
- 8 -
Table 1: Indian Hindi dialects and Fiji Hindi definite future suffixes (from Siegel, 1997,
p.115)
Bhojpuri Avadhi
Braj
Fiji
Hindi
1sg
bo
!
,
ab
bu
!
"
,
ab
ihau
!
, u
!
"
gau
ega
!
1pl
ab,
b
ÄŤ
,
iha
ab
iha
ÄŤ
, a
ÄŤ
gai ega
!
2sg (masc.)
be
!
,
ba
be
!
,
ihai
(a)ihai,
(a)igau
ega
!
(fem.)
b
ÄŤ
, bis
2pl
(masc.)
bâ(h)
bo
!
,
bau
(a)ihau,
augau
ega
!
(fem.)
bu
!
3sg
ÄŤ
ÄŤ
, ihai, e
!
(a)ihau,
agau
ÄŤ
3pl
ih,
e
!
, ihen
iha
ÄŤ
, a
ÄŤ
(a)iha
ÄŤ
, a
ÄŤ
gai
ÄŤ
The form
ega
!
clearly comes from Braj; in fact, it appears to be a compromise between the
various forms available in Braj â an example of what Trudgill (1986, p.62) calls an
interdialect
form. The form
ÄŤ
presumably comes
from Bhojpuri or Avadhi. The manner in
which variants have been selected from the range of possibilities provided by the input
dialects is an example of leveling. At the same time, the table shows extensive simplification,
involving the loss of distinct suffixes for the first and second persons singular and plural, the
third person singular and plural, and, predictably perhaps, a failure to adopt the gender
distinction in the second person found in one of the contributing dialects (Bhojpuri). A
gender distinction in verb morphology is functionally redundant, and it is not surprising that
it is lost from overseas Hindi/Bhojpuri varieties generally, including South African Bhojpuri
(Mesthrie, 1993, p.40), Fiji Hindi (Siegel, 1997, p.113), and in Mauritian Bhojpuri except in
the past-tense second-person singular (Domingue, 1980, 1981, cited in Trudgill, 1986, p.109).
- 9 -
While this simplification can be related to the special conditions of language
acquisition in a mixed, or âunfocusedâ speech community (Le Page, 1980; a topic to be
explored in a later section), there is one example of simplification (or, arguably, reduction)
that seems to stem directly from the threatening situation the indentured laborers found
themselves in. Mesthrie explains:
The same [i.e. reduction â PK] is true of the feature ârespect,â which is manifested
systematically in Indic languages in verbal and pronominal paradigms. It seems this
feature did not survive the koineization process in Natal, for there is no systematic
morphological way of signaling respect in SB [South African Bhojpuri]. Power
relations between interlocutors once indexed by pronoun usage must have given way
to the expression of solidarity on the plantations. (Mesthrie, 1993, p.40)
This is a very clear indication that, for a koine to form, the speakers must waive their
previous allegiances and social divisions to show mutual solidarity. Where they do not,
koineization is slowed, or may not result at all, and we will return to this point in the next-to-
final section. The absence of solidarity is also a factor in pidginization, where social divisions
and restricted communication directly contribute to the reduced nature of pidgins. However,
when dialects (and not languages) are in contact as in koineization, speakers can continue to
use their own vernaculars for all informal interaction within a newly-formed community
(Siegel, forthcoming). When this is coupled with solidarity, mutual accommodation on the
part of the speakers results. (See Trudgill, 1994, for a discussion of the different outcomes of
language contact and dialect contact.) (For a further discussion of overseas Bhojpuri/Hindi,
see Trudgill, 1986, pp. 99-102, 108-110.)
- 10 -
Our second example is the development of not one, but two separate koines in Odda
and Tyssedal, small towns just five kilometers apart in south-western Norway. Both grew up
at the beginning of the twentieth century around smelting works located at the head of the
Sørfjord in Hardanger to exploit the plentiful supply of hydroelectric power. People moved to
these new towns from other parts of the country, with the result that each now has a dialect
distinct from surrounding rural varieties. Interestingly, the dialects are radically different, in a
way that reflects the regional origin of the majority of the in-migrants. At the same time, they
share features which do not have their origins either in the contributing dialects or in the
existing speech of the area before industrialization. Sandve (1976) describes the differences
between the two new dialects mainly in terms of morpho-lexical variables (the variant forms
taken by morphological categories, such as the Norwegian suffixed definite article, and
closed-class words, such as pronouns). He finds that the distribution to a considerable extent
reflects the dialects spoken by the original migrants. Table 2 shows the origins of the workers
at the two factories, while Table 3 illustrates some of the morpho-lexical and phonological
features.
- 11 -
Table 2 Origins of factory workers in Odda and Tyssedal shortly after establishment
Origin of people working at Odda Smelteverk in 1916 (from Sandve, 1976, p.19)
Western Norway
Eastern Norway
Norway (other)
Other countries
81%
5%
7%
7%
Origin of people working at Tyssedal Smelteverk in 1916-18 (from Sandve, 1976, p.23)
Western Norway
Eastern Norway
Norway (other)
Other countries
36%
35%
16%
12%
Table 3 Morpho-lexical features in Odda and Tyssedal and in majority West and East
Norwegian dialects (information on Odda and Tyssedal derived from Sandve, 1976)
Odda
Tyssedal
W Norwegian
E Norwegian
i) Odda has West Norwegian, Tyssedal East Norwegian variant:
k
#
st
#
k
#
st
$
k
#
st
#
k
#
st
$
âthrowâ
(infinitive)
j
%
nt
#
j
%
nt
$
j
%
nt
#
j
%
nt
$
âgirlâ
j
%
ntu j
%
nt
#
j
%
nt
#
j
%
nt
#
âthe girlâ
e
&
g j
%
i e
&
g j
%
i
âIâ (pronoun)
kvi
&
t vi
&
t kvi
&
t vi
&
t âwhiteâ
h
%
im
#
j
%
m
$
h
%
im
#
j
%
m
$
âat
homeâ
ii) Both Odda and Tyssedal have leveled towards the East Norwegian variant:
vi
&
vi
&
me
&
vi
&
âweâ
#
l
$
#
l
$
#
dl
$
#
l
$
âallâ
çøt,
g
'
t çøt,
g
'
t çø
&
t, g
'&
t çøt,
g
'
t âmeatâ,
âboyâ
iii) Simplified and/or interdialect forms:
t
#&(
, t
#&($
t
#&(
, t
#&($
t
#&(
, t
#&)$
t
#&(
, t
#&($
âroofâ, âthe roofâ
k
*
m
$
k
*
m
$+
çe
&
m
$
k
*
m
$
r
âcomeâ (present tense)
s
*&
v
$,
s
*&
v
$+,
sø
&
v
$,,
s
*&
v
$-,
âsleepâ (present tense)
v
%
g
#+
v
%
g
$+
v
%.
j
$
r v
%
g
$
r
âwallsâ (masc. noun)
%
lv
$+
/
%
lv
#+
%
lv
$+
%
lv
#
r
%
lv
$
r
âriversâ (fem. noun)
- 12 -
It is clear from Table 3 that the Odda koine closely resembles the majority, mainly rural
dialects of western (strictly speaking, south-western) Norway, from where the vast majority
of migrants arrived. The infinitive suffix is /
#
/, and the indefinite and definite suffixes of
âweakâ feminine nouns are /
#
/ and /u/, respectively, as exemplified by /j
%
nt
#
/ and /j
%
ntu/. The
pronoun âIâ is /e
&
g/, and words such as kvit âwhiteâ and kval âwhaleâ have /kv/. Nonetheless,
this koine contains forms such as /vi
&
/ for western /me
&
/ âweâ, as well as the loss of the south-
western cluster /dl/ in favor of /l/ in words such as alle âallâ. At first sight, this could be
interpreted as straightforward mixing; however, another factor clearly plays a part. One of the
characteristics of leveling is the removal of
marked
forms (Trudgill, 1986, p.98), where
âmarkedâ describes features that are in a minority in the mix, in terms of the number of
speakers who use them, or have a restricted regional currency. The latter is clearly the case
here: /me
&
/ is restricted to the south-west, while /vi
&
/ is found in the rest of Norway, including
the regional center, Bergen, as well as in most forms of written Norwegian. The cluster /dl/
has practically the same geographical distribution as /me
&
/, and is therefore used by a small
minority of Norwegian speakers. It is in any case gradually being lost in the rural dialects. By
contrast, the maintenance of the pronoun form /e
&
g/ and the /kv/ cluster is no doubt supported
by the fact that both are widespread in western and northern Norway, and are also found in
the working-class urban vernaculars in the west, including Bergen.
Markedness (in Trudgill's sense) cannot, however, be a factor in the widespread
substitution of a short vowel in items such as /çø
&
t/ and /g
'&
t/, since the long vowel is found in
almost all western and many southern dialects. However, the short vowel is found in Bergen
- 13 -
and in the east, and (like the loss of /dl/) is beginning to spread throughout the west as part of
regional dialect leveling.
Tyssedal has a mainly eastern dialect, the morpho-lexis being eastern in form. This is
surprising, given that the proportion of east Norwegians among the incomers was only 35%.
Part of the explanation may lie in the fact that other parts of Norway were well represented,
as well the presence of a substantial foreign, mainly Swedish workforce, whose speech would
have been partly mutually intelligible with that of the Norwegians. The mixing situation in
Tyssedal was clearly more complex than in Odda, with greater linguistic differences involved
and no one group predominating. Tyssedal must have been a linguistically highly
diffuse
community, and this may go some way to explaining the eastern character of its koine.
âDiffusionâ is the opposite of
focusing
in Le Pageâs 1980 terminology: It refers to great
linguistic heterogeneity among the population, with variability both between and within
individuals. There is also likely to be an absence of stable norms of any kind, and hence a
lack of adult norms for children to converge on â again, a point we will return to. Unmarked
forms are more likely to survive here than in koineizing communities which have a dominant
group. In this context, it should be noted that, nationally, speakers of various east Norwegian
varieties form by far the largest group. Moroeover, many of the eastern forms coincide with
the majority BokmĂĽl standard. Thus, the eastern and/or standard forms had a better chance of
surviving in Tyssedal than they did in Odda. Standard forms may also have been adopted as a
âstrategy of neutralityâ in a highly diffuse situation (MĂŚhlum, 1992).
Yet in one respect the two koines show continuity with the region in which they were
established: both have the uvular [
+
] for /r/, a pronunciation that has diffusing out from the
towns throughout the west and south of Norway for the past 100 years, replacing an alveolar
- 14 -
articulation (see Chambers & Trudgill, 1998). The pre-new town Odda/Tyssedal area already
used the uvular [
+
] (GjĂrv, 1986, p.28). What is surprising is its adoption in Tyssedal, whose
dialect in almost all other respects has a strongly eastern character. A possible explanation for
this is that it is an early example of the leveling between the two towns which, according to
Sandve (1976), mainly involves the adoption of Odda features by younger people in
Tyssedal.
Finally in our discussion of these Norwegian koines, we look for cases of
simplification and interdialect forms. First, we note the absence in both dialects of the velar-
palatal alternation in nouns whose stems end in /k/, /g/ or /
/
/. In western and central dialects,
the definite form substitutes a palatal for the velar, giving /t
#&)$
/ for âthe roofâ; cf. the
indefinite /t
#&
k/. Both koines have the form /t
#&
k
$
/. However, this apparent simplification
may be the selection of a âsimpleâ feature from among the possibilities offered by the input
dialects. Second, we observe that, in Odda, the forms /k
*
m
$
/ and /s
*&
v
$0
are used for the
present tense of âcomeâ and âsleepâ. These are simple in that they do not show the present-
tense stem change found in some âstrongâ verbs in parts of the south-west (as in /çe
&
m
$
/ and
/sø
&
v
$0
, cf. infinitives /k
*
m
#
/ and /s
*&
v
#0
). It is likely that these are genuine interdialect forms
â it is unlikely that they existed in any of the input dialects, since they combine the
simplified, eastern stem with the western strong-verb suffix /
$
/. Interestingly, similar forms
are increasingly found more generally in western dialects through leveling (Sandøy, 1987,
p.234), an indication that, in all dialect contact situations, the same processes are found. The
third feature does seem to have arisen in the koine itself, since there is only recent evidence
of it in the rural dialects (Helge Sandøy, pc): this is the Odda noun plural system represented
- 15 -
by the forms /v
%
g
#+
/ and /
%
lv
$+
/, which (as Table 3 shows) differ from the majority western
variants. These show an increase in morphological regularity, the reasoning being as follows.
In west Norwegian dialects, masculine and feminine nouns fall into two classes, depending
on whether the plural ending is /
#
r/ or /
$
r/. Most, but not all, masculine nouns, like hest
âhorseâ, take /
#
r/, while feminine nouns, like seng âbedâ, tend to take /
$
r/. What has happened
in Odda is that this pattern has been generalized to all masculine and feminine nouns, leading
to the new, interdialect forms /v
%
g
#+
/ and /
%
lv
$+
/.
All the features mentioned in the paragraph above are identical to developments in
another western Norwegian koine: that of Høyanger, a new town which grew up under very
similar conditions to Odda and Tyssedal (Omdal, 1977; Trudgill, 1986, pp.95-106). The
Høyanger dialect is strikingly similar to that of Odda, a fact which reflects the mainly western
origin of the incomers. However, it contains features characteristic of its somewhat more
northern location in Sogn (from where many of the migrants came), especially an alveolar /r/
and the infinitive ending /
$01
The results from the three towns taken together demonstrate that
the features that survive the leveling prior to koine formation reflect not only the role of
simplification but also the importance of the geographical origins of the original migrants.
The latter has been explored by Trudgill and his co-researchers in an investigation of the
origins of New Zealand English, to be discussed later (Trudgill, 1998; Trudgill, Gordon,
Lewis, & Maclagan, 2000).
Our final example of a koine is, in fact, often not regarded as one at all: the variety of
spoken Hebrew that has emerged in Palestine/Israel since about 1900. The crucial difference
is that the first modern speakers of Hebrew spoke it as a second language. This meant that, in
- 16 -
the Hebrew input to the modern spoken variety, there were substrates reflecting a number of
different languages. What leads Siegel (1997, pp.129-30) to accept it as a koine is that all the
features of koineization can be found. Modern Hebrew is a revived classical language which
now performs all the functions of a community vernacular. In its pre-modern form, it ceased
to be a vernacular around 200 AD, but it continued to be used both as a liturgical language
and as a written and spoken lingua franca among Jews in Europe. The decisive phase in its
modern revival as a spoken language came with the establishment by Eliezer Ben-Yehuda of
the Hebrew Language Council in 1890. Throughout the first half of the 20th century, Hebrew
was promoted by the occupying forces, including the British Mandate of 1922-48. By 1948,
there were Hebrew-using institutions, including a Hebrew radio station. (See Blanc, 1968,
and Ravid, 1995, for further details.)
Today, practically all the Israeli-born population are speakers of Modern Hebrew,
which preserves much of the lexis, morphology, and syntax of the old language. Ravid points
to the rather slow stabilization of the contemporary language (we return to this issue later in
this chapter), and states that, even today, it has âa number of parallel constructions, none of
which has been rendered obsolete by the othersâ (Ravid, 1995, p.5). She gives the following
examples, which all translate as âthe kingâs clothesâ:
1. bigdey ha-mĂŠlex
clothes-of
the-king
2. ha-bgadim
2
el ha-mĂŠlex
the-clothes of the-king
3. bgadav
2
el ha-mĂŠlex
clothes-his of the-king
- 17 -
Ravid states that âthey complement each other and are used in distinct semantic, syntactic and
pragmatic contexts, constituting part of the linguistic competence of the Modern Hebrew
speakerâ (1995, p.5). We can say that the variability in the input to modern Hebrew has not
been leveled in this case, but has been reallocated to new functions.
As already pointed out, a defining feature of Israeli Hebrew is the fact that the input
was a series of second-language varieties with European and other substrates, particularly
Yiddish and Arabic. We must also assume that the first speakersâ proficiency in Hebrew
varied a great deal, showing varying degrees of interlanguage (Selinker, 1992). All of this
took place in a situation where there was no established native spoken norm. A look at the
consonant system shows the effect of these substrate languages. Glinert (1989, p.10) points
out that there has been considerable reduction in the phonological inventory when compared
to the liturgical language. Like many other Semitic languages, Biblical Hebrew had a
distinction between the pharyngeal consonants /
3
/ and /
4
/ and the velar /x/. Neither /
3
/ nor /
4
/
was acquired by most of the (adult) Ashkenazi immigrants, whose first languages were
European. Instead, they merged /
3
/ with /x/ â a phone widely found in European languages â
and deleted /
4
/ altogether. The Sephardic Jews, who had an Arabic substrate, used the
pharyngeals in their Hebrew vernacular. In the majority, high-status Ashkenazi-based
vernacular, the pharyngeals have been leveled out (or never acquired) despite being widely
considered to be correct. This leads to an unusual sociolinguistic situation. According to
Blanc (1968, p.245), when âGeneralâ (or majority Ashkenazi) speakers want to âimproveâ
their speech for whatever reason, they use /
4
/ when the orthography demands it, without
changing other aspects of their speech. On the other hand, /
3
/ is not adopted by these
- 18 -
speakers, because it is apparently too closely associated with Oriental speech (in Trudgillâs
1986 terms, it has âextra-strong salienceâ â a notion to be discussed below). This treatment of
the two pharyngeals â their social evaluation and their sociolinguistic patterning â is a
complex case of reallocation.
Just as with regular sound change, the leveling and simplification of the Hebrew
consonant system has led to complications in the morphology. (See Aitchison, 1991, on the
effect of phonological change on morphology.) Unlike the koines discussed so far, Modern
Hebrew is morphologically more opaque (irregular) than its antecedent. Ravid (1995, p.133)
argues that this is because of the âphonological erosionâ due to its being ârevived as a spoken
medium using a new phonological system only loosely related to that of Classical Hebrew,
with entire phonological classes being obliterated.â This is, of course, the point at which
Modern Hebrew is radically different from other koines, whose speakers are first-language
users of the input varieties. Not surprisingly, she finds among child learners the development
of non-standard reanalyses of morphological classes promoted by the principles of
âTransparency, Simplicity and Consistency.â The ability of these reanalyses to persist into
adult usage and then to become mainstream is, however, constrained by literacy and the
âliterate propensity towards marked structuresâ (1995, p.162). Israeli Hebrew was relatively
slow to stabilize; we will return to reasons for this in a later section.
- 19 -
The pre-koine: Linguistic accommodation by the first migrants in a new settlement
We now deal with koineization itself, viewed as a process with distinct, but overlapping
stages and a variable, but finite time span. If and when it reaches completion, a koine, or a
ânew dialectâ, results. Trudgill identifies the following three stages of new-dialect formation,
in his opinion roughly corresponding to the first three generations of speakers (Trudgill,
1998, Trudgill et al., 2000):
Stage
speakers
involved
linguistic
characteristics
I adult
migrants
rudimentary
leveling
II
first native-born speakers
extreme variability and further leveling
III
subsequent generations
focusing, leveling and reallocation
As we shall see, there is a great deal of variability in the time-depth of koineization, with
focusing being possible already by Stage II, and the absence of focusing sometimes persisting
over several generations of Stage III. In this section, we deal with Stage I, what Siegel calls
the âpre-koineâ. Siegel states:
This is the unstabilized stage at the beginning of koineization. A continuum exists in
which various forms of the varieties in contact are used concurrently and
inconsistently. Levelling and some mixing has begun to occur, and there may be
various degrees of reduction, but few forms have emerged as the accepted
compromise. (Siegel, 1985, p.373)
- 20 -
The question we address in this section is the following: How does this rudimentary mixing
and leveling eventually find its way into the everyday speech of the first generation of koine
speakers? Given that definable changes have occurred as a result of koineization, it follows
that these changes must be foreshadowed in some way in the speech of the pre-koine
generation. Trudgill proposes an extension of
speech accommodation theory
to account for
this process (Giles & Powesland, 1997/1975; Giles & Smith, 1979; Giles & Coupland, 1991;
Giles, Coupland & Coupland, 1991; Trudgill, 1986, pp.1-4). Simply put, accommodation
theory assumes that interlocutors converge linguistically (and on other behavioral
dimensions) when they want to gain each otherâs approval, show solidarity, etc., and that they
diverge when they do not. Accommodation can be mutual, or one-sided. It can be
âdownwardâ (as when a higher-status person uses lower-status forms, or what he or she
believes to be lower-status forms), or it can be âupwardâ (the inverse pattern).
Accommodation is therefore a response to a conversational context (though it can also be
used to define the context). When people speak different varieties, as in a new settlement, the
dialect differences are likely to be exploited â consciously or passively â as part of
accommodation. This can explain the mechanism behind the survival of majority forms in a
koine: There will be more âacts of accommodationâ involving the adoption of majority rather
than minority variants simply because there are more conversational contexts in which this
can take place.
The link between these individual acts and the new dialect, according to Trudgill, is
long-term accommodation
(Trudgill, 1986, pp.11-38), which can be defined as semi-
permanent changes in a personâs habitual speech after a period of contact with speakers using
different varieties. Long-term accommodation results from the cumulative effect of countless
acts of
short-term accommodation
in particular conversational interactions. These changes
- 21 -
are then picked up by the next generation, who will begin the process of focusing. Trudgill
discusses a number of cases of long-term accommodation in the context of the linguistic and
social constraints that promote or inhibit the acquisition of particular features. Before we
consider these below, we must examine two questions: (1) What is the evidence that the
features found in dialect leveling and koines really are foreshadowed in short-term
accommodation? (2) What is the evidence that these features are found in the long-term
accommodation of the original adult migrants?
(1) Evidence that short-term accommodation foreshadows leveling and koine formation. The
link from individual behavior in specific contexts (short-term accommodation) to a future
koine could involve the following mechanisms. First, the features of the future varieties are
adopted by adult and child migrants in individual acts of accommodation to other speakers
who happen to use them already. This has been called a âbehavioral-frequency modelâ (Auer,
1998; Hinskens & Auer, forthcoming). Second, accommodation may not be in response to a
particular interlocutor, but to images, or stereotypes, of the group the interlocutor belongs to,
or of a socially attractive group not actually represented in the immediate context (cf. Bell,
1997, on the role of different kinds of audience). This can be labeled an âidentity projection
modelâ (Auer, 1998). Third, in the case of interdialect forms, we cannot be dealing with
accommodation at all, since such forms are not in the dialect mix; therefore, they must be
created by these speakers. There is clear evidence of the âbehavioral-frequency modelâ in
Trudgillâs account of his own accommodation to his Norwich interviewees (Trudgill, 1986,
pp.7-10), though the changes only took place in the case of markers (see below, on
âsalienceâ). Couplandâs study of speech accommodation by a travel agent to her customers is
another case in point (1984). However, as Auer points out, even though there is demonstrable
accommodation, Coupland interprets this more in line with the âidentity projection modelâ:
- 22 -
Sue [the travel agent] is not attempting to reproduce the actual levels of standardness
for particular variables that she detects in the speech of her interlocutors; rather, she is
attempting to convey via her pronunciation and presumably other behaviors, verbal
and non-verbal, a persona which is similar to that conveyed by her interlocutors.
(Coupland, 1984, p. 65)
Taking this non-interactional approach to accommodation may also help us understand the
spread of dialect features by geographical diffusion where face-to-face contact with users of
the diffusing features is rare, if it is present at all. An example is the rapid and recent spread
of the merger of /f/ and /
5
/ in British English (Williams & Kerswill, 1999; Trudgill, 1986,
pp.53-7).
However, studies which concentrate on dialect leveling and koineization provide only
marginal evidence of the use of the new, leveled features in short-term accommodation
between speakers of different dialects. In an investigation of dialect leveling in the Limburg
region of The Netherlands, Hinskens (1996, pp.447-452) finds that short-term
accommodation on the part of speakers interacting with speakers of other varieties does not
follow the predictions of accommodation theory. The theory predicts that speakers should
reduce dialect features that are not shared with the interlocutor, and preserve features that are.
This turns out not to be the case, accommodation being much less differentiated and more
âacross the boardâ, regardless of the interlocutorâs variety. More striking still are the results of
a study of ongoing dialect leveling in LĂŤtzebuergesch, the Germanic language of
Luxembourg, by Gilles (1996; 1997). He discusses Luxembourg speakersâ belief that they
accommodate each otherâs speech in inter-dialect communication, but finds that this is not the
- 23 -
case (see discussion in Auer, 1998, and Hinskens & Auer, forthcoming). An extremely telling
micro-example of non-accommodation is the following excerpt from a conversation between
two strangers, FA from the north of the Grand Duchy and AN from the south (Gilles, 1996,
p.7):
FA
Bas du och nach am zweete Jor?
âAre you also in the second year?â
AN
Nä am ischte [i
2
te] Jor.
âNo, in the first year.â
FA (echoing) Am ĂŠischten [e
62
t
$
n].
âIn the first (year)â
Both speakers use their own variant, a pattern which is general in Gillesâs data. Despite this,
leveling is present in the language.
In summary, it appears that short-term accommodation as a precursor to dialect
leveling follows the same patterns as speech accommodation generally: a speaker may well
converge with anotherâs dialect, but we should not expect this to be the only pattern.
Accommodation theory and conversation analysis show that interactions are highly complex,
with a number of agendas on the part of the speakers â and the creation of a new dialect is not
one of them. The need to distinguish between actual speech accommodation and other factors
was already recognized by Thakerar, Giles & Cheshire (1982) when they posited a
âpsychological dimensionâ to accommodation. This refers to âindividualsâ beliefs that they
are integrating with and differentiating from others respectively âŚâ [emphasis in original]
(Thakerar et al., 1982, p.222). We need to accept this complexity if we are to understand the
nature of the input to koineization.
- 24 -
(2) Evidence that the features found in dialect leveling and in koines are foreshadowed in the
long-term accommodation of the original adult migrants. While the information on short-term
accommodation is complex and rather unclear, the same cannot be said for the relationship
between long-term accommodation and the outcomes of dialect leveling and koineization.
Our example is the speech of adult rural migrants in the city of Bergen in western Norway
(Kerswill, 1994a). Because they are a linguistic minority in a city, they are not potential
âkoineizersâ. However, as we shall see, their accommodation to Bergen urban vernacular
involves the adoption of some of the features found in Odda and Høyanger as well as in the
increasingly leveled dialects of rural south-west Norway. We start with a transcript of
portions of a recorded conversation between Mr. BS, a 41 year old industrial worker who
moved to Bergen at the age of 24, and another rural migrant.
- 25 -
Table 6 Extracts from conversational data for Mr BS, a rural migrant in Bergen (Kerswill,
1994a, p.148)
Key: Bergen features: single underlining
Rural dialect features: double underlining
Unmarked forms (i.e. both rural and Bergen): normal type
Except for /r/, the features marked are morpho-lexical.
Example Comment
i [
6
f
7*
8
mo
&
lø
6
n
%
]
i frü Müløya
âfrom MĂĽløyaâ
Rural dialect
ii [f
*
ÂĄ
$'
8
s
9&
v
#
]
fĂĽr ikkje du sova?
âCanât you get to sleep?â
Rural dialect
iii [f
97
8
fl
#
sk
$
kl
676"
j
%
8
d
%
n j
6
k
87%
t
86
n son
:
v
#
8
he
%
lt ...]
for flaskeklirringa den gjekk rett inn sĂĽ han var heilt ...
âbecause the rattling of the bottles, it went straight in so it was
completely ...â
Mixture of rural
and Bergen dialect
iv [bj
;<
g
%
n]
Bjørg-the
âBjørgâ
(female personal name + Bergen idiomatic addition of article in
Bergen common gender form)
Mixture of rural
and Bergen dialect
v [o
j
#
8
h
=
n j
#
]
og ja ho ja
âoh yes, herâ
Bergen dialect
vi [
%6
n
g
*/
6
8
f
7%
mti
&%
]
ein gang i framtida
âsome time in the futureâ
Rural and Bergen
dialect, with
within-word
mixing
vii [
8
s
#&
k
%
n '
=
pt
#&
d
%
g]
saka opptar deg
âthe matter at hand keeps you busyâ
Bergen dialect
- 26 -
The most striking facet of this extract is its extreme variability. Not only are features from
the two dialects mixed within an utterance, but they also appear within a single word, as in
[
8
f
7%
mti
&%
], which is [
8
f
7#
mti
&%
] in the rural dialect, and [
8
f
+%
mti
&
dn
:
] in Bergen dialect. The
mixing among this group of speakers is relatively unrestricted, with great variation both
between and within individuals. As we shall see in the next section, this is also characteristic
of the first native-born generations in many koines (Trudgill, 1998; Trudgill et al., 2000;
Omdal, 1977).
While this variability is, of course, not characteristic of established koines, some of
the features involved include those which appear in the west Norwegian koines and in west
Norwegian dialect leveling. Table 7 lists simplificatory features noted in the speech of the
rural migrants recorded for the study (Kerswill, 1994a), along with information as to whether
they occur in koines and dialect leveling in the region.
- 27 -
Table 7 Simplificatory processes found in south-western Norwegian varieties
Process
Presence in the long-
term accommodation
of rural migrants in
Bergen?
Presence in
West
Norwegian
dialect leveling?
Presence in West
Norwegian
koines?
i Simplification
of /dn/, /dl/, and /bm/ to
/n/, /l/, and /m/ in e.g.
/fidn
#
/, /
#
dl
$
/, /kobm
#
/
yes yes
Odda,
Tyssedal
(Høyanger did not
have clusters)
ii Reduction
in
vowel inventory
yes (specifically,
avoidance of /
9
/ and
/
9&
/: /g
9
lv/ -> /g
*
lv/
âfloorâ, /v
9&
r
$
/ ->
/v
*&
r
$
/ âbeenâ (part.)
yes (general
reduction;
SandÂŻy, 1987,
pp.238-9)
yes (general
reduction)
iii
Loss of the
morphophonemic velar-
palatal alternation in
e.g. /t
#&
k/ â /t
#&)$
/
yes yes
(SandÂŻy,
1987, p.234)
yes
iv
Loss of vowel
change in present tense
of some strong verbs,
e.g. /t
%&
k/ > /t
#&7
/, 'take'
(present)
yes yes
(SandÂŻy,
1987, p.236)
yes
- 28 -
It is in fact likely that some of these ostensible simplifications represent the straightforward
borrowing of a Bergen item, since in most cases the Bergen form is identical to the simpler
variant. However, for process (ii) â the loss of /
9
/ and /
9&
/ â this cannot be the case. In
almost all the relevant words, these vowels are replaced by /
*
/ or /
*&
/ â a simple, predictable
substitution. However, this does not always represent a convergence with Bergen dialect,
which has a range of vowels, and often differing lexical forms, for these words. For example,
rural /h
9&
v
$
/ âheadâ is replaced by /h
*&
v
$
/, not by Bergen /h
*
v
$
/ or either of the standard
forms hode or hovud. Similarly, /sk
9&
t
$
/ âshotâ (past participle) appears as /sk
*&
t
$
/, rarely as
Bergen /sk
'
t/ (Kerswill, 1994a, pp. 157, 159-161). Of course, we cannot claim that any of
these simplifications have been arrived at individually by the particular speakers who were
recorded, because there remains the possibility that these simplifications form part of a
ânormâ of migrant speech (Kerswill, 1994a, pp.145-7) â and are therefore spread to new
individuals by borrowing. Yet, the processes are natural, and it is likely that some of the
speakers do respond to the contact situation with simplification. Moreover, the rural migrants
do not acquire complex features of the Bergen dialect (Kerswill, 1994, p.161-2).
We return now to the factors that inhibit or promote the adoption of linguistic features
in long-term accommodation. Trudgill discusses a number of cases, including British people
living in the USA, Americans in Britain, and Swedes in Norway. Dealing mainly with
phonological changes, he finds that accommodation follows similar patterns, the basic order
being:
1. ânaturalâ and phonologically predictable phonetic changes. An example is the early
adoption of the tap [
7
] for intervocalic /t/ by British people in the USA, as in letter â in this
- 29 -
case a feature already present to some extent in many British peopleâs speech (Trudgill, 1986,
p.19).
2. substitutions of phonemes in clearly defined lexical sets. An example is the substitution of
/ĂŚ/ for /
#&
/ in items like dance, last, half, by the same group (Trudgill, 1986, p.18).
3. âcomplexâ changes, some of which may never be acquired. These include:
(i) the reversal of a merger, as when older Canadian children living in Britain
generally fail to separate the sets of cot and caught, which are merged in Canadian English
(Chambers, 1992, pp.687-8)
(ii) the use of phonemes in what are phonotactically impermissible positions in the
speakerâs own dialect, exemplified by the failure of English migrants in the USA to realize /r/
non-prevocalically, e.g., in cart (Trudgill, 1986, pp.15-6)
(iii) the acquisition of lexically unpredictable phonological processes. An example is
the Philadelphia âshort aâ pattern. This refers to the tensing and raising of the vowel in words
like man and bad, which is both phonologically and lexically determined â that is to say,
tensing/raising only takes place before certain consonants, and there are lexical exceptions;
this feature is rarely learned even by young child incomers to Philadelphia (Payne, 1980;
Roberts & Labov, 1995; Trudgill, 1986, 36-7; Kerswill, 1996, pp.186-7).
This ordering is in effect a difficulty hierarchy, with features higher up being
psycholinguistically âeasierâ and, other things being equal, more likely to be involved in
accommodation. (See Kerswill, 1996, p.200, for a more elaborated hierarchy which adds
lexical, grammatical, and prosodic features.) The order predicted by this hierarchy interacts,
however, with the factor of
salience
, invoked by Trudgill to explain why some features are
- 30 -
adopted earlier, or later, than others. Trudgill states that the following factors lead to greater
awareness of a linguistic feature, so that it becomes a
marker
, in Labovâs sense, and
therefore has the potential to become salient:
1. The variable has at least one variant which is overtly stigmatized
2. The variable has a high-status variant reflected in the orthography
3. The variable is undergoing linguistic change
4. Variants are phonetically radically different
5. Variants are involved in the maintenance of phonological contrasts in the accommodating
speakerâs variety
(adapted from Trudgill, 1986, p.11)
Trudgill states (1986, p.37): âDuring accommodation, it is indeed salient features of the target
variety that are adjusted to, except that, in the case of adults at least, a number of factors
combine to delay this modification to different extents.â These factors include those that
come under (ii) and (iii) in the difficulty hierarchy. To complete the model, Trudgill adds a
further inhibitory factor: that of
extra-strong salience
. One of the conditions leading to
salience is the involvement of a phonological contrast. However, in some cases this can lead
to a heightened awareness on the speakersâ part, so that the feature becomes a stereotype and
therefore something to be avoided. This, in Trudgillâs view, explains why Northern English
speakers tend not to acquire the southern vowel /
#&
/ in dance for their own /ĂŚ/, while they
may acquire southern /
>
/ in butter. In the dance case, they are aware that southern speakers
use a different phoneme from themselves, while with butter they are less aware of it, because
/
>
/ is not a phoneme for them (Trudgill, 1986, p.155).
- 31 -
Trudgillâs account is intended as a comprehensive model of long-term
accommodation. However, there are certain problems with it. Most particularly, these
concern the role of extra-strong salience. It seems that the same criterion, that of the presence
of a contrast in the speakerâs dialect, can lead either to a featureâs adoption or to its rejection
(Hinskens, 1996, pp.10-13). One solution to the problem is to look at a feature within the
linguistic system of both dialects, as well as viewing it from a dialect geography and social
perspective. The dance case involves a number of common lexical items, including grass,
bath, chance, last and past. Because of this, and because it involves a phonemic contrast, it is
easily stereotyped and easily labeled: Southerners talk of the northern âflatâ a, while
northerners hear the southern variant as âposhâ, doubtless because it also occurs in Received
Pronunciation. A second example comes from the dialects in the rural hinterland of Bergen in
south-west Norway. As we saw, these dialects have two vowels, /
9
/ and /
9&
/, which are not
found in the city. They are widely considered âuglyâ; speakers, both rural and urban,
comment on them spontaneously. It therefore comes as no surprise that they are being leveled
out, as well as being removed in the long-term accommodation of rural people moving to the
city. Reasons for their extra-strong salience probably lie in the fact that they are both
regionally restricted and phonetically distant from other vowels, typically /
*
/ and /
*&
/, which
are used in the same words in other dialects (Kerswill, 1994a, p.157).
Even with explanations such as these, we come up against difficulties. In a study of
long-term accommodation among eastern German (former GDR) migrants to western
Germany, Auer, Barden, and Grosskopf (1998) apply criteria for salience that are broadly
similar to Trudgillâs, with some additions; in particular, they draw a more explicit distinction
than he does between âobjectiveâ (linguistic) and âsubjectiveâ (social and social
psychological) parameters. The principal additional linguistic criterion is whether or not a
- 32 -
feature is lexicalized â that is, it is not possible to predict on phonological grounds which
lexical items are involved. We have already seen an example of lexicalization in the
Philadelphia âshort aâ pattern. Of the twelve phonological variables they investigated in the
Upper Saxon Vernacular (USV) of the migrants, three will serve to illustrate their point. They
are:
(A:) USV velarized (rounded, back) low vowel: standard [
#&
], USV [
*?&
], as in wahr âtrueâ.
(AI) USV monophthong for the standard diphthong: standard [a
$
], USV [
%?&
], as in kein ânoâ
(determiner). It is lexicalized in that it is restricted to those standard German words
which contain /ai/ derived from MHG (Middle High German) /ei/, not MHG /i
&
/. USV
[
%?&
] merges with /
%&
/, which occurs in another lexical set, e.g., in [l
%??&
bm
:
] leben âto liveâ;
hence, in accommodation to standard German, a merger must be undone.
(P,T) USV syllable-initial voiceless lenis stops instead of fortis stops: standard [p], [t], Saxon
[b
@
], [d
@
], as in paar âsomeâ, Tante âauntâ. The USV feature involves a merger between
standard German /p, t/ and /b, d/. Accommodation entails the undoing of a merger.
Table 4 shows that, by their criteria, (A:) is non-salient, while (P,T) and (AI) are rather more
salient. In the table, a âyesâ entry is evidence of salience. Auer et al. then compare this
classification with the percentage loss of the USV features between the first interview and the
last, two years later. These percentage changes (a positive score represents a loss), shown in
Table 5, do not match expectations at all.
- 33 -
Table 4 Salience of three Upper Saxon variables in contact with standard German â âstrongâ
vernacular realizations (adapted from Auer et al., 1998, p.177, Table 2a)
Variable
Criterion
(A:)
(P,T)
(AI)
Merger to be split?
no
yes
yes
Discrete
variable? no no yes
Lexicalization? no no yes
Style
differences?
yes yes **
Represented in writing?
no
yes
yes
Stereotyping?
no
yes
yes
** data not available
Table 5 Percentage loss of âstrongâ Upper Saxon realizations of three variables over a two-
year period (from Auer et al., 1998, p.180)
(A:)
(P,T)
(AI)
Percentage
loss
65% 72% -3%
The authorsâ predictions that (A:) would shift the least, and (AI) the most, were patently not
borne out. They argue (Auer et al., 1998, p.182) that the fact that (AI), along with a similar
variable, (AU), is lexicalized shelters it from loss; however, they admit that this is not an
explanation for the ârelatively positive prestige of the vernacular realizationsâ of (AI) and
(AU), and refer to the fact that the Berlin vernacular has similar monophthongs. In the case of
- 34 -
(P,T), expectations are met: a salient, stigmatized pronunciation is rapidly dropped. However,
(A:) is subject to massive attrition as well, even though this feature only fulfills one of the
criteria for salience (style shifting, which as an explanation of salience is circular). While (A:)
does not involve phoneme replacement, it can be argued that USV [
*?&
] could be taken by
some listeners for standard German /o
&
/ as in Kor âchoirâ â a possibility not mentioned by the
authors â in which case its behavior is less surprising. The authors conclude that, for this
variable at least, the subjective factors (represented by style shifting) outweigh the objective
ones.
Auer and his colleagues find that there is little match between the objective and
subjective criteria. However, for the lexicalized variables âobjectiveâ criteria do play a part;
for the remainder, different âsubjectiveâ ones seem to take precedence. Moreover, salience
âdoes not indicate the attitudinal polarity (positive or negative) of this [social and
interactional] significance, let alone its precise âideological valueââ (Auer, et al., 1998,
p.184).
Yet, salience of either kind is clearly an extremely significant factor in dialect
accommodation, as a related case study of an individual migrant showed (Auer et al. 1997).
In this study, a man who had accommodated during the first year restored most of the USV
features by the end of the second, as a result of a drastic change in his social network,
attitudes, and degree of integration following an industrial accident. The authors show that
the features which changed, first to standard, then back to USV, were mostly the ones the
main study had already identified as âsalientâ by the six criteria given in Table 4. The non-
salient variables mainly had a âflatâ graph rather than the âzig-zagâ pattern of the salient ones.
- 35 -
The problem with salience is that it may consist of a far more disparate range of
effects than research has hitherto been able to uncover, both linguistic and non-linguistic
(Kerswill & Williams, forthcoming). The fairly wide range of factors discussed in this section
goes some way to address this point.
- 36 -
Focusing: The language of the koineizing generation(s)
Trudgillâs second stage of new-dialect formation involves the first generation of children
born in the new community. As we saw above, he states that this stage is characterized by
âextreme variabilityâ and âfurther leveling.â We will examine four cases to see the extent to
which this characterization is true: New Zealand English, Høyanger, the speech of children in
the English new town, Milton Keynes, Modern Hebrew, and, finally, childrenâs speech in the
Norwegian Arctic territory of Spitsbergen. To anticipate: we find broad similarities among
speakers of this generation, and conclude that focusing usually belongs to the following
generation (the migrantsâ grandchildren). Particular conditions may mean that focusing takes
place earlier, later, or not at all. Moreover, variations we observe are ascribable to a small set
of social and linguistic factors.
The data for Trudgillâs New Zealand study come from recordings made by the
National Broadcasting Corporation of New Zealand in 1946-8. As Trudgill (1998) explains,
â[t]he recordings were oral history pioneer reminiscences, mostly from people who were the
children of the first European settlers in New Zealand ⌠About 325 speakers born between
1850 and 1900 were recorded.â This generation of people represents the first native-born
speakers of English in New Zealand â though of course they were elderly by the time they
were recorded. The most striking fact about this data archive is its tremendous variability,
both between and within individuals. Trudgill (1998) argues that, in situations where there is
no single, stable adult model, children are able to choose from a wider variety of adult models
than otherwise. Also, in the absence of a stable peer-group variety, adults, especially parents
and other caregivers, will have a greater than usual influence on childrenâs speech (Trudgill
et al., 2000). In such a situation, one can expect individuals to make novel selections of
- 37 -
features from the available choice. This turns out to be the case. Thus, Mr. Malcolm Ritchie
has the following features:
1. /
5
/ and /
A
/ are realized as dental stops, [t
B
] and [d
B
], as in Irish English
2. Syllable-final /l/ may be clear (i.e. non-velarized), as in Irish English
3. He has h-dropping in words like home, an English feature absent in Ireland
4. He has a distinction between /
C
/ and /w/, thus distinguishing which and witch. This feature
is never combined with h-dropping in the British Isles.
Not surprisingly, there is great inter-individual variation even between people with near-
identical backgrounds. For example, Mr. Ritchieâs sister-in-law, Mrs. H. Ritchie, attended the
same school at the same time as he did, yet has some quite different features in her speech.
Unlike Mr. Ritchie, she has close realizations of /ĂŚ/ as [
%
] and /e/ as [e], while he typically
has more open variants.
Despite this variability, there is evidence of leveling in this group of speakers. For
example, there is an almost complete absence of the use of the vowel /
=
/, as in
FOOT
, in the
STRUT
set â a feature of the northern half of England. In terms of the demography of the
settlers, northern speakers were certainly in a minority, and it had clearly been leveled out
already in the first generation of native-born speakers.
Trudgill does not provide any information about the transition from Stage II to the
fully-fledged, focused Stage III of present-day New Zealand English. However, he comments
- 38 -
on the relationship between the apparently random speech of the earlier generations and
present day speech, as follows:
The âoriginalâ [i.e. highly individual] mixtures demonstrated by individual
informants such as Mr. Riddle are the result of random selection. But the proportions
of variants present in the accents of groups of second-stage speakers in a particular
location, taken as a whole, derive in a probabilistic manner from, and will therefore
reflect at least approximately, the proportions of the same variants present in the
different varieties spoken by their parentsâ generation taken as a whole.
He examines some of the features of Stage III in the light of the proportions of those features
found in his Stage II corpus. Thus, 75% of the speakers (and, to judge from available
statistics, a majority of the earliest immigrants from the British Isles) did not use h-dropping
in words like house, despite the fact that this is the norm in much of England today; h-
dropping has almost completely disappeared from modern New Zealand English. A similar
explanation can be put forward for the maintenance of the distinction between /
C
/ as in which
and /w/ as in witch, despite its being rapidly lost in England.
Trudgillâs findings on the early speech of New Zealand match Omdalâs comments on
the Norwegian town of Høyanger very closely indeed. Høyanger was founded in 1916 and
received in-migrants from various parts of Norway. Omdal writes:
As it turned out, the first generation to be born and raised in Høyanger, i.e., people
who today [=1977] are in their fifties, do not have a uniform dialect, but have a
spoken language that to a great extent bears the imprint of their parentsâ dialect.
- 39 -
There is a good deal of variation between individuals. (Omdal, 1977, p.7; my
translation)
We can presume that the reasons for the lack of early focusing in Høyanger are
similar to those adduced for New Zealand: But are there any specific, local circumstances
that gave this result? It turns out that, in the early years of Høyangerâs existence, there was
considerable social segregation between the families of managers and professionals and those
of the workers, with housing in different parts of the town. Crucially, while the workers
mainly came from the same county as Høyanger, the managers and professionals came from
the east of the country. This meant that linguistic convergence between the two groups could
only take place later, as social and geographical allegiances became more oriented toward the
new community. A second factor is the relatively large linguistic differences between dialects
in Norway, particularly at that time; the factor of dialect differences would have played a
similar part in New Zealand. (See Kerswill & Williams, 2000, pp.73-4, for a more detailed
discussion of Høyanger.)
Koineization did, however, ensue in the next generation:
To find a uniform spoken variety, we must move a generation on, to people who are
in their 20s or younger. The speech of these people gives the impression that it is just
as âfirmâ as in other similar places with a more stable population growth. (Omdal,
1977, p.7; my translation)
We now have a clearer picture of the relationship between Trudgillâs three stages. The
observations from New Zealand are entirely consistent with what we have seen of the
- 40 -
relationship between the speech of the Bergen rural migrants (who can be taken to represent
Stage I of a west Norwegian leveling/koine formation process), the speech of the first native-
born generation in Høyanger (Stage II), and the features found in the koines which
subsequently developed in those towns (Stage III).
Our next example is the south-east English new town of Milton Keynes, designated in
1967 in a location roughly 80 kilometers from London, Oxford, and Cambridge. From that
date to 1991, the population of the area rose from 44,000 to 176,000. Recordings were made
of children and adults in 1991-2, some 24 years, or one generation, after its foundation.
Further recordings of a different sample were made in 1996 (Kerswill & Williams, 2000;
Kerswill, 1994b; Cheshire, Gillett, Williams, & Kerswill, 1999). Thus, almost all the child
speakers in the samples were the offspring of adult migrants to the town. We consider first
the degree to which this first native generation has focused its speech, in comparison with
that of the caregivers. The variable (ou) refers to the realization of the offset of the vowel /
$=
/
as in
GOAT
, which is currently being fronted in south-east England. The parents of the
children originate from various parts of Great Britain, and would therefore be expected to
show a range of pronunciations for this vowel, from both the south-east and elsewhere. In
order to see whether any focusing among the children has occurred, we can compare the
fronting scores for the parents (only the mothers were recorded in the study) with those of
their children. The variable has the following values:
(ou) - 0:
[o
&
], [o
=
]
score: 0
(Northern and Scottish realization)
(ou) - 1: [
$=
], [
$=D
]
score: 1
(older Buckinghamshire and London)
(ou) - 2: [
$E
]
score: 2
(fronting)
- 41 -
(ou) - 3: [
$6
]
score: 3
(fronting and unrounding)
Figure 1 shows the association of the childrenâs scores (ranked from highest to lowest) with
those of their caregivers. Two points should be noted. First, with two notable exceptions (at
bottom right on the graph), the overall range of the children is much smaller than that of their
caregivers, suggesting a high degree of focusing. The caregiversâ scores reflect their regional
origins, with the six very low scorers coming from outside the south-east. Thus, the
caregiversâ vowel realizations are not reflected at all in their childrenâs scores. On the
evidence of this and other variables (Kerswill & Williams, 2000), Milton Keynes children
seem not to be much influenced by their parentsâ speech â in distinct contrast to the first
generation native speakers in New Zealand and Høyanger. The fact that the two exceptions
just mentioned turn out to be four year olds suggests that it is the older, not the younger
children who are engaged in the focusing â a point to which we will return. Moreover, the
fact that the childrenâs scores are significantly higher also suggests that they are orienting
their focusing towards the new, fronted norm for this vowel (Kerswill & Williams, 2000,
p.101).
- 42 -
Figure 1 Association of Milton Keynes children's (ou) scores with
those of their caregivers (from Kerswill & Williams, 2000, p.102)
0
0.5
1
1.5
2
2.5
3
Subjects
(ou) index
caregivers
children
(Most
fronted)
(Least
fronted)
The Milton Keynes study also allows us to examine which age group is most
involved in the formation of the koine. We saw in the previous section an example of adult
migrantsâ speech, characterized by a high degree of mixing and instability, yet also
anticipating the forms that will appear in a later koine. In Milton Keynes, most of the adults
speak varieties of English that are far more similar to each other than their Norwegian
counterparts. This, we must assume, is because of the extensively leveled nature of British
English, especially in southern England; in any case, it means that investigating their long-
term accommodation is difficult.
However, the Milton Keynes child data allows us further insights into the early stages
of koine formation. It has recently been argued that language change is unlikely to be mainly
due to misanalyses of adult grammars on the part of young children during their acquisition
phase, for two main reasons. First, developmental forms that appear in child language are
rarely the same as those which appear in change (Croft, 2000, p.47). Second, young children,
- 43 -
for sociolinguistic reasons, are not able to be part of the diffusion of changes (Aitchison,
1981, p.180). Instead, it seems likely that older children and adolescents are the main âagents
of changeâ, because of their willingness to innovate and their orientation towards their peer
groups and older adolescents (Eckert, 2000; Kerswill, 1996). Careful examination of the
Milton Keynes childrenâs and adolescentsâ data allows us to draw conclusions about their
contribution to any new dialect that may there. Figure 2 recodes the data from Figure 1 into
two categories: mid/back offset and front offset, and adds information from the adolescents
recorded in 1996.
0
20
40
60
80
100
percent
Caregivers
(1991)
Age 4
(1991)
Age 8
(1991)
Age 12
(1991)
Age 14
(1996)
Figure 2 Percent front/non-front offset of (ou) (GOAT), Milton
Keynes women and girls (data from Cheshire et al., 1999, Kerswill &
Williams, 2000)
mid/back offset
front offset
- 44 -
0
20
40
60
80
100
percent
Caregivers
(1991)
Age 4
(1991)
Age 8
(1991)
Age 12
(1991)
Age 14
(1996)
Figure 3 Percent fronted/non-fronted variants of (u:) (GOOSE), Milton
Keynes Working Class women and girls (data from Cheshire et al.,
1999, Kerswill & Williams, 2000)
mid/mid-back
front/mid-front
As can be seen, the amount of fronting increases with the age of the child subjects from 1991
(the 4, 8, and 12 year olds), while the adults have the lowest score. Interestingly, the 14 year
olds recorded in 1996 show a further small increase. Bearing in mind that the 14 year olds
would have been 9 in 1991, these results strongly suggest that the children themselves
actually increase their fronting as they reach adolescence. Figure 3 shows a rather similar
result for another vowel that is currently being fronted, /u
&
/ as in
GOOSE
, with the process
remaining vigorous into the teens.
On the face of it, focusing has been fully achieved in the speech of Milton Keynes
children. However, there are characteristics of this new speech community which are not
typical of long-established settlements. We return to this point in the final section.
The koines we have looked at in this chapter have generally become focused by the
time of the third generation (the grandchildren of the migrants) â though for Milton Keynes it
is too early to say. As we saw earlier, Modern Hebrew arose under somewhat different
- 45 -
conditions, with second-language speakers forming the input. Despite this, it seems that a
measure of stability was reached by that generation, too, despite the continued massive
immigration to Israel and the fact that most, even Israeli-born, people continued not to be
native speakers. Blanc (1968) sets out the stages of stabilization in terms of âtypicalâ 45-year-
old speakers and their communities at different points in time. They are as follows, with
Trudgillâs stages given in parentheses:
1. 1900: Eliezer Ben-Yehudaâs contemporary. East European, a Yiddish native speaker. He
refers to written sources to guide him in his speech. (Stage I)
2. 1930: Still likely to be of East European birth or background, and not a native speaker.
(Stage II) The children of this group start to diverge, and level their speech, especially in
their informal style (incipient Stage III).
3. 1960: A fifty-fifty chance of being a native speaker. By now, there is considerable leveling
of âcommunal differentiationâ (that is, it is no longer possible to tell peopleâs language
background from their speech). His informal speech is imitated as a matter of course by
many new speakers. (Stage III)
(adapted from Blanc, 1968, pp. 239-40)
It is clear from this that new, leveled norms began to be established 30-40 years after the first
migrations (that is, Blancâs stage 3 speakers when they were children), in other words, in the
speech of the second native-born generation.
Finally, we briefly look at a new community which, despite having existed for over 90
years, has never developed a koine. This is the Norwegian Arctic territory of Spitsbergen
(Svalbard) (MĂŚhlum, 1992), where, because families stay on average only for ten years, there
- 46 -
is no possibility of a stable adult norm. Children there have an âunclear dialect identityâ
(MĂŚhlum, 1992, p.123), expressing identification both with the âhomeâ town or village on
the mainland and with Spitsbergen. Very much in line with findings from New Zealand and
Høyanger, these children apparently retain more influence from their parentsâ speech than
children do in established communities. Consequently, they are much more heterogeneous, as
well as internally inconsistent. MĂŚhlum argues that they use code-switching, dialect mixing,
and a version of standard East (Oslo) Norwegian as âstrategies of neutrality.â
Three main points emerge. First, the kind and level of social integration of the new
community affects the speed of koineization. Thus, a socially homogeneous community is
likely to koineize faster than one with considerable social divisions. Perhaps surprisingly,
continued massive immigration seems to have only a minor inhibitory effect on koineization
â as long as, crucially, there is a stable âcoreâ of speakers who remain after the initial
settlement who can act as a focus for new incomers (cf. Mufweneâs âfounder principleâ,
1996); this factor differentiates Israel from Spitsbergen. Second, childrenâs access to peer
groups is crucial. Child speakers must be able to interact freely with other, perhaps older
children for them to be able to establish norms in the absence of a stable adult model The
development of adolescent norms is likely to be accelerated by compulsory schooling â a
point made by Britain (1997a, p.165), in the context of slow dialect leveling in the Fens of
eastern England following seventeenth-century migrations (on this, see Britain, 1997b).
Schooling in early New Zealand was sporadic and not centralized, because many of the
settlements were remote and communications were poor; this is obviously not true of Milton
Keynes. (See Eckert, 2000, on the role of the adolescent years in socialization and language
change, and the importance of the school.) Third, the degree of difference between the input
varieties will affect the amount of accommodation that individuals have to engage in. In
- 47 -
Milton Keynes (unlike all the other cases considered in this chapter), the dialect differences
are for the most part subtle, being restricted to minor subphonemic variations. As a result,
most of the usual heterogeneity found among first-generation children is simply bypassed,
given sufficient opportunities for contact among children and adolescents, and focusing
toward a new variety is accelerated.
- 48 -
Koineization and continuity
In this final section, we compare koineization with other forms of contact-induced language
change. The reason for doing so is to answer the question, âAre there any characteristics
which distinguish a speech community with a koine as its everyday vernacular from one
which uses a language variety which is the result of ânormalâ transmission across the
generations?â It seems clear that, although koine formation shares some features with pidgin
and creole development, especially in the crucial role of face-to-fact contacts between
speakers of different language varieties, it is very distinct from these. Siegel (forthcoming,
pp.6-7) sets out criteria differentiating koine formation from pidgin and creole genesis, as
follows:
1. Koine formation involves continuity, in that speakers do not need to abandon their own
linguistic varieties. This is not so for pidgin and creole development.
2. In koine formation, there is no âtarget varietyâ. In pidgin and creole development, there is a
target variety.
3. Koine formation requires intimate and prolonged social interaction between speakers. We
must assume that this is not so in pidgin and creole development, where contact is restricted.
4. Koine formation can be a long process; pidgins and creoles are thought to develop rapidly
from an immediate need for communication.
- 49 -
We can take issue with the first of these points. Continuity is not clear-cut, in that a
community using a koine is likely not to have the ânormalâ contact with earlier generationsâ
speech. This places a koine between a pidgin or creole, where transmission is interrupted, and
a dialect that is the result of ânormalâ transmission. Normal transmission is defined by
Thomason & Kaufman (1988, pp.9-10) as taking place when âa language is passed on from
parent generation to child generation and/or via peer group from immediately older to
immediately younger.â Our final example will illustrate this intermediate status of a koine.
As we have already noted, Milton Keynes children use features that are characteristic
of general dialect leveling the south-eastern area, including the fronting of /
$=
/ as in
GOAT
and /u
&
/ as in
GOOSE
. How can we be sure that the developments are the result of koineization,
and not regional dialect leveling by geographical diffusion? We now look at evidence
showing that, even if the outcome of the two processes is similar, the mechanism is different,
because of the discontinuity that exists across the generations in Milton Keynes. We examine
the vowel
0F=0
as in
MOUTH
, which appears to be converging on an Received Pronunciation-
like
GF=H
, moving away from local pronunciations such as
G%6H
and
G%=DH
. Table 6 shows how
this change appears in apparent time in Reading, a town roughly the same size, and distance
from London, as Milton Keynes, but with a long-established local population (Cheshire et al.,
1999). There has clearly been a substantial shift away from the older forms to a leveled
GF=H
.
- 50 -
Table
6 Percentage use of variants of /
F=0
(
MOUTH
), Reading Working Class, interview style
,
G%=DH, G%6H,
G%&H, GF&
$
H, GI=H, GF=H,
Survey of English Dialects (SED)
informants, 1950-60s (Orton et al.,
1968)
â
Elderly (2f, 2m)
53.5
38.1
3.3
0
4.1
0.7
Girls age 14 (n=8)
0
2.3
0
8.0
0
90.4
Boys age 14 (n=8)
3.8
3.2
0
5.7
0
87.1
Table
7 Percentage use of variants of /
F=
/ (
MOUTH
), Milton Keynes Working Class, interview
style
,
G%=DH, G%6H,
G%&H, GF&
$
H, GI=H, GF=H,
Survey of English Dialects (SED)
informants, 1950-60s (Orton et al.,
1968)
â
Elderly (2f, 2m)
63.2
25.6
9.8
0
1.2
0
Women age 25-40 (1991 data; n=48)
0
0
11.7
17.2
38.6
31.5
Girls age 14 (n=8)
0
0
0
5.9
4.7
88.8
Boys age 14 (n=8)
0
0
0
12.3
3.8
83.1
Table 7 shows the corresponding data from Milton Keynes. With the inclusion of data from
the âyoung adultâ generation, this gives an apparent-time snapshot of four generations of the
- 51 -
area, with the new town being established between the âelderlyâ and âwomenâsâ generations.
Despite the similarities, there are differences between the two towns. In Milton Keynes, there
appear to be three stages in the development of this vowel: first, a period of stability in which
G
!
=DH
and
G%6H,
predominated, followed at the height of the Milton Keynes settlement in the
1970s by a period of greater heterogeneity in which
G
"
=H
, the form favored by the majority of
the in-migrants (represented here by the women aged 25-40), was dominant. A âre-focusingâ
finally began with the second-generation migrants (todayâs children), who are settling on
GF=H
. Starting with the âelderlyâ, there is a marked discontinuity in the scores between each
succeeding generation, shown particularly by the total absence of the older forms in the
speech of the women and children. This reflects the lack of social continuity in this town,
where most children have parents as well as grandparents originating elsewhere. In Reading,
young WC speakers are similarly rejecting the regionally marked forms in favor of
GF=H
.
However, it is significant that some young speakers retain the old forms of their grandparents
in a way that is indicative of the strong social continuity in this working-class part of
Reading. It is this distinction between the absence and presence of continuity that marks a
koine from a ânormalâ regional variety: The outcomes may, in the end, be the same, but the
mechanism is quite different.
PAUL KERSWILL
School of Linguistics and Applied Language Studies
The University of Reading
- 52 -
Acknowledgements
I would like to thank Peter Auer, Arne Kjell Foldvik, Frans Hinskens, and Helge Sandøy for
so willingly providing me with last-minute information during the writing of this article, and
Peter Trudgill for answering my queries so promptly.
- 53 -
References
Aitchison, J. (1981). Language change: progress or decay? London: Fontana.
Aitchison, J. (1991). Language change: progress or decay? (2
nd
edn.). Cambridge: Cambridge
University Press.
Auer, P. (1998). Structural convergence and interpersonal accommodation in a theory of
language change. Plenary paper given at the final open conference of the European
Science Foundation Network on âThe convergence and divergence of dialects in a
changing Europeâ, University of Reading, September.
Auer, P., Barden, B., & Grosskopf, B. (1997). Long-term accommodation: A single case
analysis of inner-German West-East migration and its wider implications. Paper given at
the workshop, âMigration as a factor in connection with convergence and divergence
processes of dialects in Europe,â Heidelberg, October. [European Science Foundation
Network on âThe convergence and divergence of dialects in a changing Europeâ]
Auer, P., Barden, B., & Grosskopf, B. (1998). Subjective and objective parameters
determining âsalienceâ in long-term dialect accommodation. Journal of Sociolinguistics,
2, 163-187.
Auer, P. & Hinskens, F. (1996). The convergence and divergence of dialects in Europe. New
and not so new developments in an old area. Sociolinguistica, 10, 1-30.
Bell, A. (1997). Language style as audience design. In N. Coupland & A. Jaworski (Eds).
Sociolinguistics: A reader (pp. 240-50). Basingstoke: Macmillan.
Blanc, H. (1968). The Israeli koine as an emergent national standard. In Joshua Fishman, C.
Ferguson & J. Das Gupta (Eds). Language problems of developing nations (pp. 237-
251). New York: John Wiley.
- 54 -
Britain, D. (1997a). Dialect contact, focusing and phonological rule complexity: the
koineisation of Fenland English. In C. Boberg, M. Meyerhoff and S. Strassel (Eds). A
Selection of Papers from NWAVE 25. University of Pennsylvania Working Papers in
Linguistics, 4, 141-170.
Britain, D. (1997b). Dialect contact and phonological reallocation: âCanadian Raisingâ in the
English Fens. Language in Society, 26, 15-46.
Britain, D. & Trudgill, P. J. (1999). Migration, new-dialect formation and sociolinguistic
refunctionalisation: reallocation as an outcome of dialect contact. Transactions of the
Philological Society, 97, 245-256.
Bubenik, V. (1993). Dialect contact and koineization: the case of Hellenistic Greek.
International Journal of the Sociology of Language, 99, 9-23.
Chambers, J. K. (1992). Dialect acquisition. Language, 68, 673-705.
Chambers, J. K. & Trudgill, P. J. (1998). Dialectology (2
nd
edn.). Cambridge: Cambridge
University Press.
Cheshire, J., Gillett, A., Kerswill, P. E., & Williams, A. (1999). The role of adolescents in
dialect levelling. Final report submitted to the Economic and Social Research Council of
Great Britain, ref. R000236180.
Coupland, N. (1984). Accommodation at work. International Journal of the Sociology of
Language, 46, 49-70.
Croft, W. (2000). Explaining language change: An evolutionary approach. Harlow:
Longman.
Domingue, N. (1980). Syntactic innovations in Mauritian Bhojpuri. Unpublished manuscript.
Domingue, N. (1981). Internal change in a transplanted language. Studies in the Linguistic
Sciences, 4, 151-9.
Eckert, P. (2000). Linguistic variation as social practice. Oxford: Blackwell.
- 55 -
Giles, H. & Coupland, N. (1991). Language: contexts and consequences. Milton Keynes:
Open University Press.
Giles, H., Coupland, N., & Coupland, J. (1991). Accommodation theory: communication,
context and consequences. In H. Giles, N. Coupland, & J. Coupland, (Eds). Contexts of
accommodation: developments in applied linguistics (pp. 1-68). Cambridge: Cambridge
University Press.
Giles, H. & Powesland, P. (1997/1975). Accommodation theory. In N. Coupland & A.
Jaworski (Eds). Sociolinguistics: A reader (pp. 232-239). Basingstoke: Macmillan.
(Reprinted from H. Giles & P. Powesland, P. (1975). Speech Style and Social Evaluation
(pp. 154-70). London: Academic Press)
Giles, H. & Smith, P. M. (1979). Accommodation theory: optimal levels of convergence. In
H. Giles & R. St Clair (Eds). Language and social psychology (pp. 45-65). Oxford:
Blackwell.
Gilles, P. (1996). Virtual convergence and dialect levelling in Luxemburgish. Paper given at
the workshop, âThe role of standard varieties in the convergence and divergence of
dialectsâ, University of Nijmegen, July. [European Science Foundation Network on âThe
convergence and divergence of dialects in a changing Europeâ]
Gilles, P. (1997). [Heidelberg PhD on Luxemburgish]
Gjørv, S. (1986). Mül og samfunn i utvikling. Unpublished Cand. philol. thesis, Department
of Nordic Languages and Literature, University of Bergen.
Glinert, L. (1989). The grammar of Modern Hebrew. Cambridge: Cambridge University
Press.
Hinskens, F. (1996). Dialect levelling in Limburg: Structural and sociolinguistic aspects.
TĂźbingen: Niemeyer.
- 56 -
Hinskens, F. (1998). Dialect levelling: a two-dimensional process. Folia Linguistica, 32, 35-
51.
Hinskens, F. & Auer, P. (forthcoming). Inter-personal convergence and divergence:
implications for language change. In P. Auer, F. Hinskens, & P. E. Kerswill (Eds).
Dialect change. The convergence and divergence of dialects in contemporary societies.
Cambridge: Cambridge University Press.
Inoue, F. (1986). Sociolinguistic aspects of new dialect forms: language change in progress in
Tokyo. International Journal of the Sociology of Language, 58, 73-89.
Kerswill, P. E. (1994a). Dialects converging: rural speech in urban Norway. Oxford:
Clarendon Press.
Kerswill, P. E. (1994b). Babel in Buckinghamshire? Pre-school children acquiring accent
features in the New Town of Milton Keynes. In G. Melchers & N.-L. Johannessen
(Eds). Nonstandard varieties of language: Papers from the Stockholm symposium (pp.
64-84). Stockholm: Almqvist & Wiksell.
Kerswill, P. E. (1996). Children, adolescents, and language change. Language Variation and
Change, 8, 177-202.
Kerswill, P. E. & Williams, A. (2000). Creating a new town koine: children and language
change in Milton Keynes. Language in Society, 29, 65-115.
Kerswill, P. E. & Williams, A. (forthcoming). âSalienceâ as a factor in language change:
evidence from dialect levelling in urban England. In M. C. Jones, E. Esch & A. Lam
(Eds). Contact-induced language change. An examination of internal, external and non-
linguistic factors. Berlin: Mouton de Gruyter.
Labov, W. (1966). The social stratification of English in New York City. Washington, D.C.:
Center for Applied Linguistics.
Labov, W. (1972). Sociolinguistic patterns. Philadelphia: Pennsylvania University Press.
- 57 -
Le Page, R. B. (1980). Projection, focusing, diffusion, or, steps towards a sociolinguistic
theory of language, illustrated from the Sociolinguistic Survey of Multilingual
Communities. York Papers in Linguistics 9, 9-32. Department of Language and
Linguistic Science, University of York.
MĂŚhlum, B. (1992) Dialect socialization in Longyearbyen, Svalbard (Spitsbergen): A fruitful
chaos. In E. H. Jahr (Ed.). Language contact and language change (pp. 117-130). Berlin:
Mouton de Gruyter.
Mesthrie, R. (1992). Language in indenture: a sociolinguistic history of Bhojpuri-Hindi in
South Africa. London: Routledge.
Mesthrie, R. (1993). Koineization in the BhojpuriâHindi diaspora â with special reference to
South Africa. International Journal of the Sociology of Language, 99, 25-44.
Milroy, J. (1992). Linguistic variation and change. Oxford: Blackwell.
Moag, R. (1977). Fiji Hindi. Canberra: Australian National University Press.
Mufwene, S. (1996). The founder principle in creole genesis. Diachronica, 13, 83-134.
Mßhlhäusler, P. (1974). Pidginization and simplification of language. Pacific Linguistics B-
26. Canberra: Australian National University.
Mßhlhäusler, P. (1980). Structural expansion and the process of creolization. In A. Valdman
& A. Highfield (Eds). Theoretical orientations in creole studies (pp. 19-55). New York:
Academic Press.
Omdal, H. (1977). HøyangermĂĽlet â en ny dialekt. SprĂĽklig Samling, 18, 7-9.
Orton, H., Dieth, E., & Wakelyn, M. (1968).The Survey of English Dialects: vol. 4 (the
Southern Counties). Leeds: E.J. Arnold.
Payne, A. (1980). Factors controlling the acquisition of the Philadelphia dialect by out-of-
state children. In W. Labov (Ed.) Locating language in time and space (pp. 143-178).
New York: Academic Press.
- 58 -
Ravid, D. D. (1995). Language change in child and adult Hebrew. A psycholinguistic
perspective. New York: Oxford University Press.
Roberts, J. & Labov, W. (1995). Learning to talk Philadelphian; Acquisition of short a by
preschool children. Language Variation and Change, 7, 101-112
.
Sandøy, H. (1987). Norsk dialektkunnskap. Oslo: Novus.
Sandøy, H. (1998). The diffusion of a new morphology in Norwegian dialects. Folia
Linguistica,
32, 83-100.
Sandve, B. H. (1976). Om talemĂĽlet i industristadene Odda og Tyssedal. Generasjonsskilnad
og tilnÌrming mellom dei to mülføra [On the spoken language in the industrial towns
Odda and Tyssedal. Generational differences and convergence between the two dialects].
Unpublished Cand.philol. thesis, Department of Nordic Languages and Literature,
University of Bergen.
Selinker, L. (1992). Rediscovering interlanguage. London: Longman.
Siegel, J. (1985). Koines and koineization. Language in Society, 14, 357-378.
Siegel, J. (1987). Language contact in a plantation environment. Cambridge: Cambridge
University Press.
Siegel, J. (1993). Introduction: controversies in the study of koines and koineization.
International Journal of the Sociology of Language, 99, 5-8.
Siegel, J. (1997). Mixing, leveling, and pidgin/creole development. In A. K. Spears & D.
Winford (Eds). The structure and status of pidgins and creoles (pp. 111-149).
Amsterdam/Philadelphia: John Benjamins.
Siegel, J. (forthcoming). Koine formation and creole genesis. In N. Smith & T. Veenstra
(Eds). Language contact and creolization. Amsterdam and Philadelphia: John Benjamins.
Sobrero, A. A. (1996). Italianization and variations in the repertoire: the KoinaĂ.
Sociolinguistica, 10, 105-111.
- 59 -
Thakerar, J. N., Giles, H., & Cheshire, J. (1982). Psychological and linguistic parameters of
speech accommodation theory. In C. Fraser & K. R. Scherer (Eds). Advances in the
social psychology of language, (pp. 205-55). Cambridge: Cambridge University Press.
Thelander, M. (1980) De-dialectalisation in Sweden. FUMS Rapport No. 86. Uppsala:
Uppsala University.
Thelander, M. (1982). A qualitative approach to the quantitative data of speech variation. In
S. Romaine (Ed.) Sociolinguistic Variation in Speech Communities (pp. 65-83). London:
Edward Arnold.
Thomason, S. G. & Kaufman, T. (1988). Language contact, creolization and genetic
linguistics. Berkeley: University of California Press.
Thomson, G. (1960). The Greek language. Cambridge: W. Heffer & Sons.
Trudgill, P. J. (1986). Dialects in contact. Oxford: Blackwell.
Trudgill, P. J. (1994). Language contact and dialect contact in linguistic change. In U.-B.
Kotsinas & J. Helgander (Eds). Dialektkontakt, sprükkontakt och sprükfÜrändring i
Norden. FĂśredrag frĂĽn ett forskarsymposium (pp.13-22). Series: MINS, No. 40.
Stockholm: Institutionen fĂśr nordiska sprĂĽk, Stockholms universitet.
Trudgill, P. J. (1998). The chaos before the order: New Zealand English and the second stage
of new-dialect formation. In E. H. Jahr (ed). Advances in historical sociolinguistics (pp.
1-11). Berlin: Mouton de Gruyter.
Trudgill, P. J. Gordon, E., Lewis, G., & Maclagan, M. (2000). Determinism in new-dialect
formation and the genesis of New Zealand English. Journal of Linguistics 36, 299-318.
Trumper, J. & Maddalon, M. (1988). Converging divergence and diverging convergence: the
dialect-language conflict and contrasting evolutionary trends in modern Italy. In P. Auer
& A. di Luzio (Eds). Variation and convergence (pp. 216-258). Berlin: de Gruyter.
- 60 -
Williams, A. & Kerswill, P. E. (1999). Dialect levelling: change and continuity in Milton
Keynes, Reading and Hull. In P. Foulkes & G. Docherty (Eds). Urban voices (pp. 141-
162). London: Arnold.