• CODE SWITCHING REFERENCES
• Missouri State University - BearWorks - Code-Switching in Spoken Indian English:
A Case Study of Sociopolitical TalkSociopolitical Talk
• National Center for Biotechnology Information - Pubmed Central - Acculturation
and attitudes toward code-switching: A bidimensional framework
• Frontiers - Code-Switching Strategies: Prosody and Syntax
• CORE - The Role of Code-Switching as a Communicative Tool in an ESL Teacher
Education Classroom
• Verywell Mind - Code Switching: How and Why it Happens
• Social Sciences LibreTexts - Code Switching
• ScholarWorks at UMass Amherst - What Code-Switching Strategies are EOective
in Dialogue Systems?
• NPR - How Code-Switching Explains The World
Chapter 4. Code switching in its social contexts. (WINFORD, Donald. An
Introduction to Contact Linguistics. Blackwell Publishing, 2003, pp. 110-137)
1. Introduction.
We saw in Chapter 3 that bilinguals play an active role in the kinds of structural
diOusion that lead to convergence of linguistic systems. In this chapter, we examine the
actual performance of bilinguals who exploit the resources of the languages they
command in various ways, for social and stylistic purposes. Bilinguals achieve this by
alternating between their two languages, or by mixing them in diOerent ways. These kinds
of behavior are referred to as code switching.
The phenomenon known as code switching involves several types of bilingual
language mixture, including the alternating use of relatively complete utterances from two
diOerent languages, alternation between sentential and/or clausal structures from the two
languages, and the insertion of (usually lexical) elements from one language into the other.
These kinds of language mixture have long been the norm in many communities, and have
become increasingly common as a result of various socio-historical forces that have led
to increasing contact among diOerent language groups within the same national and local
communities. The sociolinguistic situations that are characterized by code switching are
quite varied. First, there are stable long-term situations such as those in Switzerland,
Belgium, etc., where bilingualism is the norm across wide sections of the community.
Second, there are situations, particularly in Africa, South East Asia, the Caribbean and
South America, where colonization introduced European languages to serve as oOicial
vehicles of administration, education and other public activities, alongside the preexisting
indigenous languages. Third, the increasing flow of immigrants into more
industrialized nations in Europe, North America, and elsewhere has led to the
establishment of linguistic minority groups who must become bilingual in the host
community's language, and in some cases shift entirely to the latter in the course of a few
generations. Finally, there are situations in which speakers of non-standard dialects are
required to learn the standard variety of their language for purposes of educational and
social advancement. The result is increasing bi-dialectalism, accompanied by code
switching between the varieties.
Code switching is therefore a cover-term for quite varied types of bilingual and
bidialectal language mixture, resulting from quite diOerent social circumstances and
motivations. Ludi (1987) has proposed that the relevant sociolinguistic situations can be
categorized in terms of two inter-dependent distinctions - exolingual vs endolingual
interaction, and unilingual vs bilingual interaction. Exolingual interaction involves
speakers of diOerent languages, while exolingual interaction involves speakers with the
same language background. Either type may involve bilingualism or unilingualism. The
combination of these distinctions yields the following typology of situations:
Bilingual Unilingual
Exolingual Interaction among speakers Interaction between native and
with diOerent languages. non-native speakers of the same language.
Endolingual Interaction among bilinguals. Interaction among monolinguals.
All of these situations involve code switching of one form or another. Yet the tendency
in the literature has been to focus mostly on the types of language mixture found in
"endolingual bilingual" (i.e., stable bilingual) situations, and the types of code switching
behavior that is typical of highly proficient bilinguals in such communities.
2. Defining code switching.
Code switching has been defined as "the alternate use of two or more languages in the
same utterance or conversation" (Grosjean 1982:145). This definition is broad enough to
encompass just about any kind of language alternation or mixture. Researchers, however,
don't always agree on precisely what kinds of alternation should be included under the
designation "code-switching". Most scholars would exclude the use of diOerent languages
in diOerent situations or activity types, of the sort found in so-called diglossic speech
communities. It's questionable, however, whether such situations do in fact involve a
strict complementarity of functions between the codes involved, with no language
alternation within the same situation or event. Some researchers also exclude cases
involving "non-contiguous stretches of talk, for example, one occurring at the beginning,
the other at the end of the conversation" (Auer 1995:116). In general, the criterion of
juxtaposition of elements from the two codes is a pre-requisite for code switching. For
Auer, this also means that style shifting involving gradual transitions from dialect into
standard cannot be included, since it works very diOerently from bilingual code switching
(ibid.). Most researchers also exclude those types of mixture, often referred to as
"interference phenomena", which occur in the speech behavior of persons acquiring a
second language (i.e. in "exolingual bilingual" situations). The relationship between code
switching and these interlanguage phenomena will be discussed further in section 2.2,
below.
In general, then, code switching is taken as referring only to those cases where
bilingual speakers alternate between codes within the same speech event, switch codes
within a single turn, or mix elements from two codes within the same utterance. Auer
(1995:124) identifies four patterns of code switching. Pattern I involves switches from
code A to code B, as in (1), or switches within a single speaker's turn, as in (2).
In (1), from Myers-Scotton 1993a:134), a bus conductor switches from Swahili to
English (in italics) in an exchange with a passenger.
(1) Conductor: Umelipa nauli ya basi?
"Have you paid the bus fare?"
Passenger: (No response)
Conductor: Unaenda wapi?
"Where are you going?"
Passenger: Nafika Jerusalem.
"I'm going to Jerusalem [housing estate]."
Conductor: You must always say clearly and loudly where you are going to alight,
OK?
In (2), a market-vendor and a customer alternate between Swahili and English (in
italics) as they negotiate prices (Myers-Scotton 1993a:40-41).
(2) Vendor: Habari, mheshimwa. Have some vegetables.
"Hello, respected sir"...
Customer: Mboga gani? Nipe kabeji hizi. How much is that?
"Which vegetables?" Give me these cabbages"...
Vendor: Five shillings only.
Customer: That's too much. Sina pesa.
"I don't have [much] money."
These types of switching often mark a shift in topic, role-relationship or activity type.
Auer refers to this pattern as "discourse-related" code switching or "conversational" code-
switching.
Pattern II involves a negotiation of a language of interaction, with participants
switching from one code to another until consensus is reached on the medium of
exchange. Milroy & Li Wei (1995:149) oOer the following example of this pattern, which
Auer (1995:125) refers to as "preference-related switching". Here a mother switches
between English and Cantonese in addressing her son, who finally responds in English.
(3) Mother: Finished homework?
Son: No response (2.0 sec.)
Mother: Steven, yiu mo wan sue?
"Steven, do you want to review your lessons?"
Son: (1.5 sec) I've finished.
Further examples of language negotiation from Heller (1985) will be discussed later.
Pattern III involves switching between languages in a turn such that no single language
can be identified as the base language. This pattern is typical of code switching as an
"unmarked choice" (see below), used either to express "strategic ambiguity" (Heller 1988),
or a strategy of neutrality (Myers-Scotton 1993a:147). In these cases, a speaker may
switch from sentence to sentence or from clause to clause within the same sentence,
thus leaving language choice open. Sentence (4) is an example of inter-sentential
switching from Myers-Scotton (1993a:123). A Luyia man is interviewing a Luyia woman
who works as a nurse in Nairobi. She switches from English to Swahili, to their shared
ethnic language, Lwidakho. English elements are in italics, while those from Lwidakho are
underlined.
Interviewer: Unapenda kufanya kazi yako lini? Mchana au usiku?
"When do you like to do your work? Days or nights?"
Nurse: As I told you, I like my job. Sina ubaguzi wo wote kuhusu wakati
ninapofanya kazi. I enjoy working either during the day au usiku yote ni
sawa kwangu. Hata family members wangu wamezoea mtindo huu. There
is no quarrel at all. Obubi bubulaho.
"As I told you, I like my job. I have no difficulty at all regarding when I do
work. I enjoy working either during the day or at night; all is okay as far as I'm
concerned. Even my family members have got used to this plan. There is no
quarrel at all. There is no badness.”
Example (4), from SankoO & Poplack (1981:11), illustrates inter-clause switching in the
performance of a Spanish/English bilingual in New York city. Spanish is in italics.
(4) There was a guy, you know, que [that] he se montó [got up]. He started playing with
congas, you know, and se montó y empezó a brincar [got up and started to jump]
and all that shit.
Finally, Type IV alternation refers to momentary switches which do not really change
the language of the interaction. This kind of alternation is referred to as intra-sentential, or
more accurately, intra-clause code switching. It produces utterances which have most of
their lexicon as well as morpho-syntactic apparatus from one language, referred to as the
matrix language (ML), with insertion of single words or phrases from the other language,
referred to as the embedded language (EL). This type of mixture produces three kinds of
constituents: mixed constituents made up of materials from both languages, EL 'islands'
or phrases incorporated from the EL, and ML 'islands', that is, phrases entirely in the ML.
Example (5) illustrates a mixed constituent consisting of an English stem (decide) with
Swahili aOixes in an otherwise Swahili utterance (Myers-Scotton 1993b:4).
(5) Hata siku hizi ni-me-decide kwanza kutumia sabuni ya miti.
even days these 1s-PERF-decide first to use soap of stick.
"[But] even these days I've decided first to use bar soap."
Example (6) illustrates a mixed English/Swahili constituent (ni-ta-try) English EL
islands (throughout the day) as well as Swahili ML islands (Myers-Scotton 1993b:146).
(6) Mimi mi-ta-try kuwa nyumbani throughout the day.
EMPH 1s-FUT-try to be home
"As for me, I try to be at home throughout the day."
There is disagreement among researchers as to whether all types of intra-sentential
alternation should be included within code switching proper. Auer, for instance, refers to
this type of mixture as "transfer" or "insertion" and distinguishes it from code switching.
Other researchers, such as Kachru (1978) and Singh (1985) refer to such mixture as "codemixing",
reserving the term "code-switching" for types I , II and III above, in which diOerent
codes are used on diOerent occasions, or correspond to diOerent stages in an interaction.
But it is often diOicult to draw clear boundaries between the two. Poplack (1990), while
accepting the distinction between inter-sentential and intra-sentential code switching,
reserves the latter only for cases where entire constituents from the two languages
alternate in the same utterance (i.e., cases involving EL islands). Hence Poplack &
Meechan (1995:200) define code switching as "the juxtaposition of sentences or sentence
fragments, each of which is internally consistent with the morphological and syntactic
(and optionally, phonological) rules of its lexifier language." This definition explicitly
excludes single morpheme switches, which are treated as "nonce borrowings". This will
be discussed further in section 3 below. Poplack also distinguishes a third type of
alternation which she calls "tag-switching", that is, the insertion of a tag in language B into
an utterance which is otherwise in language A, as in the following examples from
Tagalog/English (Bautista 1980:247) and English/Spanish (Poplack 1980:596).
(7) The proceedings went smoothly, ba [Tagalog tag]?
"The proceedings went smoothly, didn't they?"
(8) I could understand que [that] you don't know how to speak Spanish, verdad? [right?].
Finally, Myers-Scotton (1993a, b) also distinguishes between inter-sentential and
intra-sentential code switching; however, contra Poplack, she includes single-morpheme
switches within the latter category. Also, in her linguistic analysis of code switching
(1993b), she focuses mainly on intra-sentential alternations, and her definition of code
switching seems to present this type as prototypical.
"Code switching ... is the selection by bilinguals or multilinguals of forms from
an embedded variety (or varieties) in utterances of a matrix variety during the
same conversation" (1993a:3).
These linguistically based definitions of code switching seem to imply a certain level
of skilled behavior that only practiced bilinguals can manage. As we shall see, the
linguistic frameworks developed for the analysis of code switching generally focus on this
kind of skilled performance. The tendency to reify code switching as a unitary and clearly
identifiable phenomenon has been questioned by Gardner-Chloros (1995:70), who
prefers to view code switching as a "fuzzy-edged concept." For her, the conventional view
of code switching implies that speakers make binary choices, operating in one code or the
other at any given time, when in fact code switching overlaps with other kinds of bilingual
mixture, and the boundaries between them are diOicult to establish. Moreover, it is often
impossible to categorize the two codes involved in code switching as discrete and
isolatable. This indeterminacy is reflected in the diOiculty of distinguishing code switching
from "borrowing" on the one hand, and "mixing/interference" on the other.
Exercise: Examine the code switching behavior of bilinguals you know, and try to
determine which of the four patterns described above they tend to use. How easy is it to
distinguish among the patterns?
2.1. Code switching versus borrowing.
The single-morpheme switching which is typical of intra-clause code switching is very
common across bilingual communities. Researchers have attempted to distinguish this
from borrowing, but there is no consensus on the boundary between the two. The chief
criteria that have been used to distinguish them include:
(a) Degree of use by monolingual speakers; and
(b) Degree of morphophonemic integration.
According to the first criterion, established loans are commonly used by monolingual
speakers, whereas code-switches tend to be transitory phenomena. Some researchers
argue, however, that frequency counts are inconclusive, and that the distinction between
a switch and a borrowing is not transparent to bilinguals. The criterion of
morphophonemic integration is also problematic, since both borrowings and wordswitches
may or may not be morphologically and phonologically adapted to the ML or
recipient language (Myers-Scotton 1993b: 177-91). Poplack and her associates, however,
claim that borrowing involves diOerent mechanisms than code switching, and that the
latter involves more than single-morpheme insertions. SankoO et al (1986) argue that if an
utterance has the syntax and morphology of one language, then any lexical item not native
to that language must be a borrowing. This lead them to treat all single word switches as
"nonce borrowings". Poplack & Meechan (1995) support this view with evidence from
noun-modification patterns on single French switches in both French/Wolof and
French/Fongbe bilingual discourse. The single nouns are fully integrated into the
morphosyntax of the recipient languages, and are indistinguishable from other wellestablished
borrowings. They are quite diOerent from multi-word French switches, which
show the internal structure of French NPs, and represent true instances of code switching.
Other researchers (Myers-Scotton 1993b, TreOers-Daller 1991, Gardner-Chloros 1995)
view single-word switching and borrowing as essentially similar processes which fall
along a continuum, based on degree of integration or assimilation. For Myers-Scotton,
code switching is in fact a major conduit for borrowing, since single switches can become
borrowed forms through increased frequency of use and adoption by monolingual
speakers of the ML (1993b:182). The only clear basis for distinguishing them is frequency
of occurrence - an admittedly arbitrary criterion, but one that seems to have some
empirical support (ibid.). Other criteria that fail to distinguish the two phenomena include
(a) the extent to which native synonyms are displaced by either; and (b) the types of
grammatical category that tend to be borrowed as opposed to being switched. As
Gardner-Chloros (1995:73-74) notes, both single switches and borrowings may fill lexical
gaps in the recipient language and also oOer further options to native equivalents.
Moreover, there is no diOerence in the types of lexical categories that can be switched or
borrowed. Both processes display a similar hierarchy of incorporation of items, with nouns
most likely to be incorporated, followed by adjectives, then verbs, prepositions, and so on,
as discussed in Chapter 2.
Exercise: Collect data on “single word switches” from various studies and compare
them with cases of lexical borrowing. What similarities and diOerences can you observe
in the processes of integration that apply in each case? [Myers-Scotton (1993b:177-91)
and Poplack & Meechan (1995) discuss the general relationship between borrowing and
code-switching. TreOers-Daller (1994) and Gardner-Chloros (1991) also compare the two
in the same community.]
2.2. Code switching vs "interference"
Most researchers also draw a distinction between code switching as practised by
competent bilinguals and the kinds of mixture found in the "interlanguage" of persons
acquiring a second language (See Ludi's "exolingual bilingual" category above.) Hamers &
Blanc (1989:149) refer to the former as "bilingual code-switching" and the latter as
"incompetence code-switching". They note that the latter type is typical of certain
immigrant populations who have acquired a limited functional competence in L2 but have
to resort to their L1 to compensate for their lack of knowledge of L2" (ibid.). There are also
cases where immigrants who have lost some of their competence in L1 resort to the L2 to
fill the gaps. Poplack (1987:72) also distinguishes the kinds of mixture associated with
fluent bilinguals from "speech errors which involve elements of both languages, and which
may be properly considered 'interference'". It may seem somewhat arbitrary to reserve the
term "code-switching" only for skilled bilingual behavior, and exclude phenomena
associated with processes of shift and second language acquisition. However, there are clear
differences, both linguistic and sociolinguistic, between the two kinds of behavior. The
linguistic differences recall those between changes due to borrowing and changes due to
L1 influence on L2 acquisition. Sociolinguistically too, the language mixture found in
interlanguage does not follow the rules associated with bilingual code switching (Gumperz
1982).
This may explain why it is often diOicult to describe interlanguage phenomena in the
frameworks employed for bilingual code switching. Myers-Scotton skirts the issue, noting
that "I do not pretend to detail language shift facilitated by code-switching; my only
purpose here is to suggest that the mechanism exists" (1993b:223). Attempting to treat
cases of language shift (SLA) in terms of gradual replacement of ML morpheme order and
system morphemes, etc., by their EL counterparts (ibid.) seems to oversimplify the nature
of the learning process, especially in the earlier stages of acquisition. Further discussion
of language mixture in situations of SLA and language shift will be reserved for chapter 7.
Advanced learners, however, do produce code switching phenomena similar to those
of (other) competent bilinguals. In some cases, such learners may employ either the L1 or
the L2 as the matrix language. For example, Nishimura (1986) describes how fluent
Japanese/English bilinguals in Toronto and San Francisco produce code-switched
utterances some of which have the syntactic structure of English, as in (9), while others
have the structure of Japanese, as in (10). Japanese items are in italics.
(9) The ones we've seen are bimboo na kodomo.
poor children
(10) Kaeri ni wa border de we got stopped, eh?
return on Topic on
"On the way home, we got stopped at the border."
In some cases, Nishimura shows, it is not possible to assign code-switched utterances
unambiguously to one language or the other. This will be discussed further below.
The rest of this discussion will focus on code switching as practised by fluent
bilinguals. Both inter- and intra-sentential alternation, including single morpheme
switches, are included here under the umbrella of code switching. The type of
performance displays properties similr to those of monolingual discourse. As Romaine
(1989:111) notes:
In code-switched discourse, the items in question form part of the same speech act. They
are tied together prosodically as well as by semantic and syntactic relations equivalent to
those that join passages in a single speech act."
This suggests that the code switching performance of bilinguals is associated with an
underlying competence that can be described by a system of rules and constraints
analogous to those that regulate monolingual performance. The linguistic aspects of code
switching will be discussed more fully in the following chapter. The rest of this chapter
explores the social functions and meanings of this type of language mixture. The
discussion here is also relevant to the various kinds of language behavior that are
discussed throughout this book, all of which are subject to similar social forces and
motivations.
3. Social motivations for code-switching.
The socio-cultural factors that influence code switching have been investigated
primarily within three major frameworks: the Sociology of Language, Linguistic
Anthropology, and the Social Psychology of language choice. Each of them has
contributed to our understanding of the social motivations for code switching, and their
insights complement one another in providing a comprehensive picture of how macrolevel
societal norms interact with micro-level factors to constrain the linguistic behavior
of bilinguals. The Sociology of Language provides insight on how macro-level social
institutions and group relationships influence patterns of code switching. Anthropology
investigates how micro-level interpersonal relationships, participant goals and types of
interaction aOect cs behavior in specific encounters. Within Social Psychology,
Communication Accommodation theory sheds light on how attitudes and group
consciousness influence individuals to accommodate to one another through cs or other
linguistic compromises, or to diverge and maintain language boundaries.
3.1. Code switching and sociolinguistic domains.
In two ground breaking papers, Fishman (1964, 1965) introduced the concept of
"sociolinguistic domains" to represent the contexts of interaction into which social life is
organized, and which influence the language of interaction. Examples of domains include
"family", "work", "religion", "friendship", "education" and others. Such domains are
associated with specific language varieties which are seen as appropriate for a particular
interaction. Fishman (1972:441) defines domains as "institutional contexts and their
congruent behavioral co-occurrences". Domains are abstract constructs, made up of a
constellation of participants' statuses and role relationships, locales or settings, and
subject matter (topic). As Breitborde (1983:18) notes, "A domain is not the actual
interaction (the setting), but an abstract set of relationships between status, topic and
locale which gives meaning to the events that actually comprise social interaction." The
relationship between "domain" and "situations" is analogous to that between a phoneme
and its allophones. In other words, a domain is "a cluster of interaction situations, grouped
around the same field of experience, and tied together by a shared range of goals and
obligations" (Mioni 1987:170).
Domain analysis has revealed a great deal about the situational factors that influence
language alternation in diOerent settings. While this kind of alternation is diOerent from
code switching proper, it is often the starting point for the latter, particularly since it is
doubtful whether a neat separation of languages is always achieved in all the relevant
domains. An often-cited example comes from studies of a New York City Puerto Rican
community by Greenfield & Fishman (1968) and Fishman & Greenfield (1970). They
identified five domains of language behavior for the community: family, friendship, work,
religion and education. Table 1 illustrates the interrelationships among domains, their
components (status/role relationship, locale and topic), and the corresponding language
choices in the community. Note that the choices indicated here are majority ones; in other
words, not all subjects agreed that Spanish or English was the preferred choice in the
relevant domain.
Table 1: Selected domains of language choice among NYC Puerto Ricans.
(Adapted from Greenfield and Fishman 1968)
Domain Role relationhship Locale Topic Language
Family Parent/child Home Family matters Spanish
Work Employer/employee Workplace Job performance English
EducationTeacher/student School Homework English
As Breitborde (1983:20) notes, the lack of unanimity in responses is significant in its
own right, since it points to diOerences in individual interpretations of what is salient in a
particular domain (e.g., topic vs status), as well as diOerences in the way societal norms
constrain individual choice. In addition, in actual interactions, the components do not
always correspond so neatly to the configurations shown in Table 1. For instance,
discussion of a work-related topic in the home setting may lead to conflicting language
choice, as do other mismatches among the components of a domain. In such cases, the
neat compartmentalization of language choice by domain often breaks down, leading to
code switching proper. An example from Swiss German will be discussed in the following
section.
3.1.1. Domains, diglossia and code switching.
The concept of sociolinguistic domains actually goes back to the work of Charles
Ferguson (1959), who introduced the notion of "diglossia" (adapted from Marçais' (1930)
diglossie) to describe situations where two related language varieties are employed in
complementary distribution across diOerent situations. In diglossic communities, one of
the varieties, designated the H(igh) language, is employed in more oOicial, public domains
such as government, education, literature, etc., while the other, designated the L(ow)
language, is used in more private and informal domains such as the family, friendship,
neighborhood, etc. Ferguson's term was intended to describe a special type of
sociolinguistic situation that diOered from the usual standard-with-dialects scenario on
the one hand, and from strictly bilingual situations on the other. The varieties involved in
diglossia, while related, are still quite divergent in structure and lexicon, and only one of
them, the L variety, is typically acquired as a first language, while the H variety has to be
acquired as a second language, usually at school. A variety of other characteristics of
diglossia is summed up in the following definition from Ferguson (1964:435)
Diglossia is a relatively stable language situation in which, in addition to the
primary dialects of the language (which may include a standard or regional
standards). there is a very divergent, highly codified (often grammatically more
complex) superposed variety, the vehicle of a large and respected body of
written literature, either of an earlier period or in another speech community,
which is learned largely by formal education and is used for most written and
formal spoken purposes but is not used by any sector of the community for
ordinary conversation.
To illustrate his concept, Ferguson cited cases such as the alternation of vernacular
and classical Arabic in Middle-Eastern countries; Schwyzertüütsch and Standard German
in (German-speaking) Switzerland; Dhimotiki and Katharevousa in Greece, and Kreyòl and
Standard French in Haiti. Fishman adopted the concept of diglossia, but extended it to
include bilingual situations characterized by a similar compartmentalization of languages
across public and private domains. As Fasold (1984:53) suggests, it is useful to distinguish
"classic" diglossia in Ferguson's sense, from "broad" diglossia in Fishman's sense.
An excellent example of how classic diglossia works is provided by Keller's (1982)
account of language use in German-speaking Switzerland, where the diglossia is relatively
stable, and everyone is aware of its existence. Keller points out that there are deliberate
attempts "to cultivate diglossia by making the distinction between standard and dialect
as clear as possible" (1982:75). This is the policy of the Bund für Schwyzertüütsch, with its
headquarters in Zürich (ibid.). The following example from Keller (1982:79) illustrates the
structural diOerences between the varieties (Gloss and translation by Charlotte
Schaengold).
(11) H. Sie konnten jenes Häuschen dort drüben nicht kaufen
they can(past) that house (dimin.) over there not buy
Es war ihnen zu teuer.
it be(past) for-them too expensive.
L. Si händ säb Hüüsli deet äne nid chöne chauOe.
they have that house(dimin.) over there not can buy
S isch ene z tüür gsy.
it is for-them too expensive be(pp)
“They couldn't buy that little house over there. It was too expensive for them"
On the whole, the complementary distribution of the functions of H and L is wellmaintained,
with H functioning as the medium of science and technology in lectures, at
conferences, in books and written instructions, as well as in other public domains such
as legislation, administration, education and so on. L, on the other hand, is obligatory for
"ordinary conversation", in personal and family relationships, and so on. What is
particulary interesting, however, is the fact that the pattern of separation of varieties is
neither neat nor consistent in all situations. As Keller (1982:88) notes, when there is a
conflict between function and situation, (for example, if a topic involving science or
technology is discussed at home), "the result tends to be a mishmash of both language
forms, in other words there is a breakdown of diglossia." One manifestation of this is
extensive borrowing of vocabulary from L into H, in cases where "the situation is too
private for 'High", yet 'Low' is unsuited for the function. Where such a conflict occurs, the
whole thought process and the mental syntax tend to be standard, the lexicon will be half
'High' and half 'Low'" (ibid.). These types of alternation have resulted in significant change
in the lexicon of L, with native items continuously being replaced by 'High' items, or being
adapted phonologically under influence from the latter. Moreover, some structural
features such as the Standard German future tense and present participle, are slowly
being introduced into the L variety. Keller does not discuss whether L also influences H,
but notes that "the German standard language has ... a Swiss variant, the Schweizerhoch
Deutsch, which is not just characterized by dialectalisms" (p. 83). This variety diOers from
Standard German "in pronunciation, stress, orthography, lexis (semantic and lexical) and
grammar (morphological and syntactical)" (p. 84). Most diOerences appear to be due to
the separate evolution of the Helvetian standard, but it seems likely that some of its
features are due to "interference" from Schwyzertüütsch in the process of second
language acquisition.
Keller's account makes it clear that the neat compartmentalization of functions
associated with diglossia does not apply even in "classic" cases like this. The Swiss
German situation oOers some insight into how changing constellations of the
components of domains, can lead to varying degrees of code switching, resulting in
contact-induced changes in both varieties.
Exercise: Compare situations of stable diglossia such as that in German Switzerland
with unstable situations like that in Greece, where diglossia has broken down and the L
language has spread into public and “high” domains. What social, political and
ideological factors favor the maintenance or loss of diglossia? [On Switzerland, see Keller
(1982). On Greece, see Frangoudaki (1992) and Kazazis (1993). An extensive bibliography
on diglossia can be found in Alan Hudson (1992)}
3.2. Micro-level analysis: Code switching and conversation.
Domain analysis has provided much insight into the general patterns of language
choice in bilingual communities. The concept provides a link between the macro-level
organization of society, with its "socio-cultural norms and expectations" and the microlevel
organization of language use manifested in "individual behavior at the level of faceto-face
verbal encounters”(Fishman 1972:441). On the one hand, it is clear that the social
structure constrains language choice in certain ways. On the other, individuals can and do
exploit the choices available to them to manipulate situations and redefine the
relationships pertinent to a particular interaction. Research in social and linguistic
anthropology has greatly enhanced our understanding of these individual choices.
Hymes (1962) presented a comprehensive framework for the investigation of language
choice in speech communities. His "ethnography of speaking" distinguishes a range of
components that are relevant to the analysis of the speech economy of a community.
Some of these, such as 'situation' and 'participants' have to do with the social organization
of the community. Others such as 'ends', 'act sequence', 'key', and 'norms of interpretation'
have to do with the transactional nature of interactions, as negotiations aimed at
achieving particular goals. In addition, they portray such interactions as ordered
exchanges which rely for their success on shared understanding of socio-cultural norms
and the meanings assigned to specific language choices.
The major impetus for investigation of code switching as socially meaningful linguistic
behavior came from Gumperz' research on Norwegian communities in the 1960's, and
particularly from the highly influential study by Blom & Gumperz (1972) of the town of
Hemnesberget. They found that residents of the town employed two dialects: Ranamål
(the vernacular), and Bokmål (the standard variety). The scenario they present is clearly
reminiscent of diglossia, with B as High and R as L language. The residents of the
community fall into 3 socio-economic strata. First there are locally born laborers and
skilled workers with strong local network ties, including kinship. Then there is an
intermediate group of wholesale/retail merchants and plant managers who have relatives
within the group of artisans, and who must cultivate ties with customers of various
backgrounds. Finally, there it a diversified group of service personnel, including clerks,
business owners, school teachers, doctors and the like, who are often migrants, or have
stronger ties outside the town. Interestingly, Blom & Gumperz found that language choice
was not directly correlated with socio-economic divisions, but rather depended on
network ties that often cut across those divisions.
Those linked primarily to local networks and values include manual workers and
artisans, as well as lower-range white-collar workers. Merchants and plant managers have
both local and non-local network ties, and hence divided loyalties to local versus external
values. Finally, the retail owners, school teachers, doctors and other professionals
identify more with the non-local middle-class value system of the pan-Norwegian elite. It
is primarily the interactions between the members of the first two networks that involve
code switching. This takes two forms. First, diOerent situations are associated with
diOerent varieties - a pattern referred to as "situational" code switching. For instance,
oOicial business between a local resident and an employee in a government oOice is
conducted in Bokmål, but if the two engage in an informal chat, they switch to Ranamål.
In general, Bokmål is used in church services, school matters, interaction with strangers,
and other formal contexts. Ranamål is used among locals at home, in the workplace, and
in public meeting places. This situational switching involves changes in participants' rights
and obligations, that is, their status and role relationships.
But Blom & Gumperz observed that speakers often switched codes on the same
occasion, particularly when there were changes of topic. For instance, locals would greet
government clerks and discuss family matters in vernacular, but switch to standard for the
business transaction. But even within the latter, they used the vernacular for informal
comments and asides. (Compare the Swiss German situation discussed earlier.) Blom &
Gumperz termed this "metaphorical" switching. The distinction between 'situational' and
'metaphorical' switching triggered a great deal of research on the social motivations for
code switching. The distinction itself has been challenged on the grounds that it isn't
based on clear criteria. As Pride (1979) pointed out, 'metaphorical' switches involve the
same re-definitions of status and role-relationships as 'situational' switches, hence the
two are not distinguishable. But Breitborde (1983:14) argues that there is "both empirical
and intuitive validity in distinguishing the two." He suggests that what is involved is the
diOerence between "two statuses operating simultaneously in one situation, and two
statuses each operating in its own situation." However, Breitborde does not distinguish
status from role-relationship, which would add more weight to his position.
The Blom & Gumperz study was important for several reasons. It demonstrated that
choice of a code is not just a mechanical product of a static domain, but that domains
themselves can change as role-relationships do, and interlocutors assume diOerent
stances toward the subject matter of the transaction. The study was also innovative in its
use of participant-observation techniques and social network analysis to investigate the
code alternation in the community. Blom & Gumperz showed that, in order to understand
the social meaning of code switching, we must ask "who the speakers are, and how the
codes are used. The former requires that we comprehend the social identities and social
relations obtaining among speakers; the latter, that we figure out how these social
identities (which are linked to linguistic varieties) are brought to bear in social interaction"
(Breitborde 1983:7).
3.2.1. Taxonomies of factors aOecting code switching.
Micro-level approaches to code switching have also attempted to construct
taxonomies of the functions of code-switches, to determine the various factors that trigger
such switches within a conversation. Gumperz (1977) proposed three general motives for
code-switching:
• Choice of H to add authoritativeness to an utterance.
• Choice of H to highlight the point of a narrative in L.
• Choice of H to add seriousness to commands directed at a child.
Since then, fuller taxonomies have been developed. Auer (1995:120) provides the
following list of "conversational loci in which switching is particularly frequent".
• Reported speech.
• Change of participant constellation, particularly addressee selection.
• Parentheses or side-comments.
• Reiteration, i.e., quasi-translations into the other language, for purposes of
emphasis, clarification, attracting attention, etc.
• Change of activity type, also called 'mode shift' or 'role shift'.
• Topic shift.
• Puns, language play, shift of 'key'.
• Topicalization, topic/comment structure.
But such taxonomies are of limited use, since, in general, they fail to provide clear
explanations for specific switches within the context of an interaction. In other words, as
Auer explains, such listing does not bring us closer to a theory of code-alternation or tell
us why code-alternation may have a conversational meaning or function. What is at stake
here is a theoretical framework which can explain (and ultimately predict) patterns of code
switching within a conversation.
3.3. Toward a theory of the social meaning of conversational code switching.
There have been two approaches to this problem - one focussing on the sequencing of
utterances and the sequential environment of code-switches (Gumperz 1982; Auer 1995);
the other focussing on code choices as "indexing rights-and obligations sets (RO sets)
between participants in a given interaction type" (Myers-Scotton 1993a:84).
The former approach builds on Gumperz' (1982) idea of code switching as a
"contextualization cue" similar to monolingual choices of prosodic, paralinguistic,
syntactic and lexical features which signal a particular intent on the part of the speaker. In
this approach, the meaning of a code-switch depends both on its sequential position in
the discourse context itself, and on the broader situational and socio-cultural context
which make up the background knowledge of the participants. The latter embraces the
community norms for the type of interaction, and the values assigned to diOerent
language choices. The approach is therefore similar to those types of conversational
analysis which attempt to explain how intended meanings are conveyed through
implicatures signaled by the speaker, and inferences drawn by the listener. Hence this
approach attempts to link the micro-level of conversational interaction to the macro-level
societal setting in which it occurs, and without which it cannot be interpreted.
Myers-Scotton (1993a) takes a somewhat diOerent approach to the social motivations
for code switching. Her view is that speakers do not exercise linguistic choices solely
because of their social identities or in response to situational factors, but rather use such
choices to convey "intentional meaning of a socio-pragmatic nature" (1993a:57). This
approach stresses the role of the speaker as creative actor who uses language choice to
negotiate changes in the nature of the situation and the social relationships among
participants. Myers-Scotton's "markedness" model is thus a model of speakers' sociopsychological
motivations for code switching (1993a:75). It draws on concepts from a
variety of disciplines, including Sociology of Language (domains), Social Anthropology
(transactions, negotiations), Linguistic Anthropology (communicative competence) and
Pragmatics (implicatures and intentional meaning). Basically, the model presents code
choices as "indexing" (pointing to) rights-and-obligations sets (RO sets) between the
participants in a particular interaction. An RO set "is an abstraction which is based on
situational factors, and represents the attitudes and expectations of participants toward
each other" (1993a:85). The model interprets code choices as negotiations of RO sets.
The idea of "markedness" relates to the community norms which apply to interaction
types and which individuals may either obey or violate to achieve conversational goals.
"Unmarked" choices of code are those which conform to community norms and
participants' expectations. Examples would include instances of so-called "situational"
switching where diOerent RO sets are associated with diOerent code choices. There are
also instances of switching within the same conversation which signals simultaneous
adherence to two positively valued social identities (for instance the types of switching
practised among educated bilinguals in African cities like Nairobi and Dakar). The latter
type of code switching is typically a strategy of neutrality that picks the middle ground
between two identities and their related RO sets. "Marked" code switching on the other
hand, represents a departure from the normal, expected choice, and therefore has "shock
value", signaling some ulterior intent on the speaker's part. According to Myers-Scotton,
the general eOect of marked code switching is "to negotiate a change in the expected
social distance holding between participants, either increasing or decreasing it"
(1993a:132). Marked code switches may be used to convey anger or authority, to exclude
outsiders from in-group interaction, to "flag" or emphasize messages via repitition, and so
on. They may also be used to achieve more familiarity and solidarity with an addressee in
a situation which normally calls for a diOerent code choice. Though Myers-Scotton
focusses on individual choices and motivations, her model requires a thorough
understanding of macro-level societal norms and evaluations to interpret the
transactional meaning of code switches. As she notes, such interpretation depends on
"the framework of markedness which is provided by societal norms" (1993a:109). Hence
both macro- and micro-level factors are included in her model.
Most of the empirical frameworks proposed for the study of code switching attempt
this kind of integration. None would claim to have achieved predictive power, though
several inductive generalizations have emerged about why speakers switch on specific
occasions in certain socio-cultural contexts.
3.4. Code switching and Communication Accommodation Theory.
Another important contribution to our understanding of the social meaning of code
switching comes from Communication Accommodation Theory (CAT), developed by
Howard Giles and his associates. A major objective of CAT is to explain the cognitive and
aOective factors that influence individuals to change their speech (and other forms of
communication) in ways that either converge with or diverge from that of their
interlocutors. Convergence (convergent Accommodation) can be described as a strategy
by which interlocutors accommodate to each other's speech in a variety of ways - by
adjusting pronunciation and other linguistic features, or even paralinguistic features such
as speech rate, pauses and length of utterances. Divergence (divergent Accommodation)
refers to the ways in which speakers emphasize linguistic diOerences between
themselves and others. The central idea behind CAT is that speakers may be motivated to
adjust their speaking styles in order to evoke the listener's social approval, to increase
communicative eOiciency, and to maintain a positive social identity.
CAT is based on social psychological processes such as similarity attraction, social
exchange, causal attribution and inter-group distinctiveness. Speech convergence is
explained primarily in terms of the first three processes, while divergence is explained in
relation to inter-group processes. For instance, in convergent Accommodation, a speaker
minimizes the linguistic diOerences between herself and her interlocutor with a view to
increasing social attraction. The notion of social exchange refers to the speaker's
assessment of the costs and rewards of accommodating. The process of causal
attribution has to do with the listener's interpretation of the speaker's intention in adapting
his style. Such accommodation is valued more positively if it is attributed to the speaker's
desire to reduce social distance.
Finally, the notion of inter-group distinctiveness comes into play when a speaker
employs distinctive linguistic markers of his own group to emphasize his own group
membership and dissociate himself from his interlocutor. Divergent accommodation is
therefore a strategy for maintaining social and psychological distance. Analysis of
divergence has benefited from Tajfel & Turner's (1979) social identity of inter-group
relations and social change. According to this theory, the more individuals define
encounters in inter-group terms and wish to preserve a positive in-group identity, the more
likely they are to diverge linguistically. The extent of the divergence depends on factors
such as the individual's perception of the social forces operating in favor of their own
group, and their group's status vis a vis other groups (Giles et al. 1987:29). These factors
are closely linked to the concept of ethnolinguistic vitality - a construct introduced by
Giles, Bourhis & Taylor (1977). The concept refers to the degree of autonomy and
distinctiveness a group enjoys by virtue of factors such as demography (size, distribution,
natural increase, etc.); status (economic and political power, prestige, etc.); and
institutional support (degree of support in societal institutions such as the mass media,
education, etc.). The assumption is that the higher the degree of ethnolinguistic vitality,
the more likely it is that the group will preserve its distinctiveness and its own language.
Individuals' perceptions of the ethnolinguistic vitality of their group influence the way they
behave in inter-ethnic encounters. Hence the concept oOers ways of linking micro-level
individual behavior to the macro-level societal structures which define inter-ethnic group
relations. This and other additions to the theory have helped CAT evolve from a strictly
socio-psychological model of speech accommodation to an interdisciplinary framework
for describing and explaining language choices in inter-ethnic interpersonal interactions.
The basic propositions of CAT were first summarized by Street & Giles (1982:213-14)) as
follows:
(1) Speakers will attempt to converge linguistically towards the speech patterns
believed to be characteristic of their interlocutors when (a) they desire their social
approval and the perceived costs of so acting are lower than the rewards anticipated;
and/or (b) they desire a high level of communicational eOiciency and (c) social norms
and/or linguistic competence are not perceived to dictate alternative speech
strategies.
(2) The degree of linguistic divergence will be a function of (a) the extent of the speakers'
repertoires, and (b) factors (individual diOerences and situation) that may increase
the need for social approval and/or communicational eOiciency.
(3) Speech convergence will be positively evaluated by recipients when the resultant
behavior is (a) perceived as such psychologically, (b) perceived to be at an optimal
sociolinguistic distance from them, and (c) attributed internally with positive intent.
(4) People will attempt to maintain their speech patterns or even diverge linguistically
from those believed characteristic of their recipients when they (a) define the
encounter in inter-group terms and desire a positive ingroup identity, or (b) wish to
dissociate personally from another in an interindividual encounter, or (c) wish to
bring another's speech behaviors to a personally acceptable level.
(5) The magnitude of such divergence will be a function of (a) the extent of speakers'
repertoires, and (b) individual diOerences and contextual factors increasing the
salience of the cognitive or aOective functions in proposition 4.
(6) Speech maintenance and divergence will be negatively evaluated by recipients when
the acts are perceived as psychologically diverging, but favorably reacted to by
observers of the encounter who define the interaction in inter-group terms and who
share a common, positively valued group membership with the speaker.
These propositions have been revised and reformulated in more detail by Giles et al.
(1987), to take account of the findings of recent research.
3.4.1. Code switching as accommodation.
Much of the earlier work in the CAT framework consisted of experimental laboratory
studies, but there has been increasing attention to observation of language behavior in
natural settings (for an overview, see Giles et al. 1991). Some of the latter studies are
highly relevant to understanding the social motivations for code switching. For instance,
Coupland (1984) analyzed how a travel agent converged phonologically to clients of
diOerent educational and socio-economic backgrounds. Berg (1985) studied code
switching in Taiwan by examining over 8,000 interactions in markets, shops, banks and
department stores. In most cases, it was found that interlocutors accommodated
mutually to each other's code choice, which varied according to the setting. For instance,
customers in banks would converge upwardly to the clerks, who converged downwardly
to them. In the marketplace, however, customers converged downwardly to salespersons
who in turn converged upwardly to them. These patterns of convergence seem typical of
interactions like these, where the customer's approval is sought. Other kinds of
interaction are characterized by diOerent patterns of Accommodation.
Studies like these support the proposition that the greater a speaker's need to achieve
social approval, social integration or some instrumental goal, the greater the degree of
convergence will be. Research of this kind can therefore shed light on patterns of code
switching as a form of accommodative behavior in situations of actual or potential
language shift, as well as in situations of stable bilingualism where code switching is the
'unmarked' choice. As Giles et al. (1991:20) point out,
"Much of the literature on long and mid-term language and dialect
acculturation can also be interpreted in convergence terms, whereby
immigrants [and other linguistic minorities - DW] seek the economic
advantage and social rewards (though there are clearly also costs) that
linguistic assimilation sometimes brings."
Language divergence has also been investigated from a CAT perspective in controlled
experiments and in natural settings. An example of the former is Bourhis et al.'s (1979)
study of how diOerent groups of trilingual Flemish students (Flemish-French-English)
reacted when recorded in 'neutral' as distinct from 'ethnically-threatening' encounters
with a French (Walloon) out-group speaker. Neutral encounters between Flemish and
French speakers are often conducted in English. But when the out-group speaker posed
ethnically threatening questions in English to the listeners, they initially responded in
English, but half of them later switched to their in-group language, Flemish. In a follow-up
study, when the Walloon speaker posed his threatening question in French, almost all
listeners switched to Flemish. Studies like this demonstrate how, in such inter-group
encounters, speech maintenance or divergence are (often deliberate) acts of maintaining
group identity (Bourhis 1979).
Speech communities in which diOerent ethnolinguistic groups are in conflict or wish
to preserve their own distinctiveness are likely to be characterized by language
divergence. This typically results in non-reciprocal patterns of code switching which
disfavor the kinds of language mixture that Accommodation produces in other situations.
A well-documented example of this is the situation in Quebec, Canada, as described by
Heller (1985; 1995). Here, inter-ethnic interactions typically involve a process of language
negotiation in which individuals attempt to choose between English and French as the
language of conversation. The patterns of code switching that accompany such
exchanges shed light on "how language use is bound up in the creation, maintenance or
change in relations in power" (Heller 1995:164). To understand this, one must first
comprehend how power relationships between Anglophones and Francophones in
Quebec have changed over the years. Until the 1960's, Anglophones were dominant in the
political and economic spheres of activity, and this dominance was reflected in the
imposition of English as the language of public and inter-ethnic private communication
(op. cit. 167). After World War II, however, Francophones became more and more
upwardly mobile, penetrating the arenas of politics, higher education and business
enterprises. As a result, French became increasingly favored as the medium of interaction
in these settings. A key factor in this development was the passing of Bill 101 by the
government of Quebec in 1977. This bill declared French to be the oOicial language of the
province and required that it be used in various public domains, including government,
education and occupational spheres such as pharmacy, nursing, engineering, etc. This
law in eOect gave French speakers access to all areas of social and economic activity in
the province, and made French the language of the work sphere. As a result, French and
English came to compete directly in various domains, especially in public service
encounters and in the workplace (Heller 1985:79).
This redefinition of the power relationships between the groups and their language led
inevitably to a redefinition of the norms of language use and evaluation that had
characterized the older social order. Under the new dispensation, choice of one or the
other language can signal a variety of meanings and intentions (presentations of self) on
the speaker's part. Exclusive use of French may signal commitment to French nationalist
identity, while exclusive use of English may indicate resistance to that nationalism (ibid.).
Alternatively, such exclusive choices may simply be due to monolingualism. Hence interethnic
interactions can proceed only when interlocutors have determined each other's
ethnic aOiliation, language ability and language preference, and particularly the values
they associate with diOerent language choices.
Speakers react to such choices in diOerent ways. In some cases, they may opt for
deliberate divergence, emphasizing group distinctiveness in the encounter. Heller
(1985:78-79) presents the following example of a conversation at a government bureau
between an Anglophone man who has come to take a French proficiency test, and a
bilingual receptionist:
Man: Could you tell me where the French test is?
Receptionist (in French) Pardon?
Man: Could you tell me where the French test is?
Receptionist: En français? (In French?)
Man: I have the right to be addressed in English by the government of Quebec.
Receptionist (to 3rd person) Qu'est-ce qu'il dit? (What's he saying?)
In other cases, speakers continually switch languages until some consensus on one
or the other is reached, as in the following example from Heller (1985:81) - a telephone
conversation between a patient and a clerk at a hospital.
Clerk: Central Booking, may I help you?
Patient: Oui, allô?
Clerk: Bureau de rendez-vous, est-ce que je peux vous aider?
("Appointments desk? Can I help you?")
Patient: (French) (The patient begins to try to make an appointment)
Clerk: (French)
Patient: (English)
Clerk: (English)
Patient: (French)
Clerk: (French)
Patient: Êtes-vous français ou anglaise? ("Are you French or English?")
Clerk: N'importe j'suis ni l'une ni l'autre.
("It doesn't matter, I'm neither one nor the other")
Patient: Mais... ("But...")
Clerk: Ç ne fait rien. ("It doesn't matter")
Patient: (French)
(The conversation continues in French)
Code switching in such cases becomes a "strategy of neutrality" by which speakers
avoid committing themselves to a distinct ethnic identity, choosing a middle path instead
(Myers-Scotton 1993a:147). The situation in Quebec, then, represents one way in which
code alternation is linked to social processes and interactions between groups, and
becomes a means of expressing relationships of power or solidarity. In other situations,
the forces that promote divergence may explain why certain minority groups resist
acquiring a host language with complete native proficiency (Giles et al. 1991:31), settling
instead for an ethnolect that reflects their separate status. This may also partly explain the
emergence and maintenance of local or "indigenized" varieties of European languages in
ex-colonial settings (see for example the "New Englishes" of Africa, South-east Asia, the
Caribbean etc.) It is clear that CAT oOers a rich theoretical framework within which to
investigate not just code switching, but other types of convergent or divergent linguistic
behaviors. With respect to code switching in particular, it oOers a way of explaining how
and why code selection "can be the vehicle of quite opposite tendencies, from
accommodation to divergence, and from language maintenance to language shift"
(Gardner-Chloros 1995:80).
Summary:
This chapter examined various types of bilingual language mixture to which the label
“code switching” has been applied. First we considered how best to define this term.
Code switching manifests itself in various ways, from alternation between utterances in
diOerent languages to various kinds of intra-sentential and intra-clause mixture.
There is a tendency to restrict the definition of code switching only to those kinds of
language mixture practised by skilled bilinguals. Hence those kinds of mixing that
characterize the interlanguage of learners acquiring a second language tend to be treated
as distinct phenomena. In addition, not all researchers regard the type of intra-sentential
mixture that involves momentary switches of content morphemes as “true” code
switching. Some see it as a kind of lexical borrowing. It is clear that code switching is a
continuum of language behaviors, and there are no sharp boundaries between it and other
kinds of language mixture.
The other central focus of the chapter was on the social meanings and motivations of
code switching. This kind of language performance is a reflection of the way speakers
perceive their social identities and relationships to one another, in the broader context of
their community’s social structure. Choices of code are typically associated with diOerent
situations or sociolinguistic domains. A perhaps extreme example of this are communities
characterized by diglossia, a situation in which diOerent code choices are employed in
complementary distribution in separate domains. Code switching can often be an act of
identity (Le Page & Tabouret-Keller 1985) by which speakers locate themselves in social
space and in relation to their interlocutors. In other cases, it may be a communicative
strategy by which they attempt to achieve transactional goals. In these respects, it is
similar to other kinds of stylistic variation (for instance, in monolingual communities)
which are conditioned by social norms of interaction.
Choices of code are also regulated by factors such as language attitudes that reflect
individuals’ and groups’ perception of one another. Such attitudes can lead them to
accommodate to one another via code switching or other kinds of convergence, or to
diverge by emphasizing linguistic diOerences and avoiding code switching or other
accommodative behavior. Degrees and types of code switching are also constrained by
other social factors, such as the extent of contact between (members of) the groups, and
the degrees of bilingual competence that individuals display.
However, in the final analysis, it is the interplay between social and structural factors
that determines the actual types of code switching practised by individuals and groups.
As Myers-Scotton (1993c:476) points out, structural factors determine permissible forms
of code switching, while social factors regulate the choices among the various types
permitted. The role that structural factors play is the subject of the following chapter.