‘International Journal of Bilingualism’ • Volume 9 • Number 1Running

• 2005, 85– 102|

Head 85

Child codeswitching
and adult content contrasts
Susan Ervin-Tripp 1 and Iliana Reyes 2
1 University of California, Berkeley
2 University of Arizona

Abstract Key words

This paper discusses links between the conditions of developing child bilin- borrowing
gualism and the adult outcomes in semiotic contrast in elicited speech and
codeswitching.
child
Analysis of interaction of children raised as bilinguals shows that from the bilingualism
beginning they can recognize the appropriate language for addressees. When
the lexical repertoire is inadequate, borrowing occurs, and codeswitching of
longer segments appears before age two. Throughout childhood and adoles-
codeswitching
cence, codeswitching has increasing interactional functions as children’s
pragmatic skills grow.
Some adult codeswitching relies on semiotic differences implied by language. It is likely that
both such codeswitching and the dual selves shown in elicited discourse in different languages are
limited to specific sociolinguistic situations and personal histories. These include education in a
second language, adult immigration, and frequent contact with a monolingual community.

1 Introduction
The differences in content and perspective related to language separation in the bilingual
adult are the other side of language choice during codeswitching. In this paper, we
propose to examine the context and development of codeswitching in children as an
aspect of pragmatic development, and then to link developing codeswitching practices
to the adult division of labor between languages.

2 Child bilingualism
The requirements for the acquisition of any language are: capacity, motivation to
understand, be understood and/or sound like a group member, and access or sufficient
time in situations where a symbolic system and understandable information (meanings)
co-occur systematically and permit inference and mapping.
The factor of access is the most important one at any age to acquire a language.
Motivation has its effects primarily through varying access. In infancy motivation is
not a variable factor since exposure is out of the control of the child, but older learners

Address for correspondence

Susan Ervin-Tripp, Psychology Department, University of California, Berkeley CA 94720 -1650, USA;
e-mail: < ervintrp1@socrates.berkeley.edu > .

‘International Journal of Bilingualism’ is ©Kingston Press Ltd. 2005

 86 S. Ervin-Tripp and I. Reyes

can ignore or avoid access to one of the varieties available to them. By preschool,
children begin to consider it more important to understand some speakers than others
and tend to pay more attention to them and try to spend time with them. They come
to care more about being understood by or sounding like specific others. These are the
factors that account for the fact that children usually sound more like their peers than
like their parents.
Children also often learn the language of their peers in the neighborhood. To go
beyond just understanding a language, one also needs a reason to speak. This might be
sociability, which makes one want to initiate interaction, or a need to ask for necessities.
Beyond this, learners can have a desire to sound like others and to be seen as an insider
who shares the same values. Sociolinguistic studies of dialect (e.g., Berthele, 2000; Eckert,
1989) have shown that in children and teenagers these can be powerful factors in making
children’s speech like that of others located nearby in a peer network.
These factors mean that most infants with enough exposure to more than one
language will become competent, regardless of how the interaction conditions are set
up. In the case of bilinguals, there is not equal access to both languages. Whether a
bilingual child will grow in knowledge of each language, develop complex syntax, a
variety of registers, and a mature vocabulary, depends both on the changing family usage
and on wider social networks as time passes. It is very easy for a childhood language
to be lost without family and community support (e.g., Hakuta, 1994; Wong-Fillmore,
1991). Later use of one of a child’s languages often declines in relatively monolingual
societies like the United States because the child hears the dominant language of the
community more often, wants to sound like a speaker of that variety, and gets no school
support for his or her first languages if they are of low prestige.

3 Contexts of child bilingualism

3.1
Person-prescriptive families
The first studies of child bilingualism were often done by linguists studying their own
families (Leopold, 1939 – 1949; Ronjat, 1913). In these cases, the prescriptive beliefs of
linguists affected their family practices. A prototype of this kind of study is Ronjat’s
family, in which one parent spoke only German, the other spoke only French, and
visitors were admonished to stick to one language only (Grammont’s principle).
In such practices, the output of each parent is prescribed, but both parents must
understand each other. In the strictest case, speech between the parents involves asym-
metry of use, with one speaking language A and the other speaking language B. While
the child must know that the parents can each understand the speech of the other, the
child is admonished to do situational switching; that is, to change language by addressee,
even when the child is with both parents.
In a three-person, isolated family, such prescriptive conditions can be described
as valuing maximum separation of languages and setting up the optimal experimental
conditions for the study of any mutual influences on the linguistic systems of the child
that might occur. However, families are normally not isolated. There are others who
The International Journal of Bilingualism
 Child codeswitching 87

come into the household as playmates, relatives, guests, employees or repairmen. In

many communities it cannot be assumed that these all are bilingual, so what happens
theoretically is that the child must act as interpreter, since only the child is allowed to
speak both languages. Valdés (2003) provides more discussion on the child as interpreter
to outsiders. Language choice between siblings in such a family is not predictable, since
their interaction must sometimes occur outside of parental control.
In the real world, of course, it is extremely difficult to maintain rigid prescriptive
separation, and even adults following a prescriptive system are likely to slip and use
codeswitching or brief insertions to each other or to the child or to find it necessary to
use their passive language to outsiders.

3.2
Asymmetric parent bilingualism
In some families, there is both a bilingual and a monolingual parent or other relative.
The outcome depends very much on whether the monolingual speaks the dominant
societal language or a minority language, and is unwilling or unable to learn the other
language.
In such families, the child hears the parents talk to each other in the shared variety.
Since bilinguals often feel it is rude to speak in a language others do not understand, there
is a problem in conversation between the child and bilingual parent if the monolingual
parent is present. For this reason, the excluded parent sometimes may not actually
want the child to be bilingual, so power dynamics in the family can play a role. If the
monolingual speaks the community’s dominant language, there can be a strong risk of
language loss. This issue of fear of exclusion and of a threat to power is reminiscent of
the problem in workplaces where monolingual owners/supervisors prohibit speech in
the minority language.
On the other hand, the presence in the family of a monolingual speaker of a
minority language that is not supported in the community can be a very strong induce-
ment to bilingualism in children. In an unpublished instructional classroom study on
hundreds of families through a questionnaire project in the community, Ervin-Tripp
and Guo (1992) found that death or departure from the household of a monolingual
first generation immigrant was among the strongest predictors of minority or immigrant
language loss after childhood.

3.3
Bilingual dual use families
In the majority of bilingual families, both languages are known and used by everyone.
In immigrant families competence changes through time, as a result of both schooling
of children and language learning on the job by parents (Hakuta, 1994), changing both
language attitudes and codeswitching practices in the family (Valdés, 1996).
Although sibling talk is an understudied topic, a few researchers have stressed the
importance of siblings’ social role in the development of a family’s bilingual skills (e.g.,
Fantini, 1985; Grosjean, 1982). Most studies show that at a given age it is first-borns that
know the parents’ language best. Yet older children also bring the dominant language
The International Journal of Bilingualism
88 S. Ervin-Tripp and I. Reyes

from the community into the home and switch or translate when siblings or parents
do not understand.

3.4
Bilingualism from the outside
When a family uses a single language variety inside the household, the bilingualism
of a child comes from outside the family. In immigrant communities, many children’s
first exposure to the dominant societal variety is from school and neighborhood peer
interaction. For many, the first contact is at school age. In such cases, there is an inside
language and an outside language, which are likely to have quite different pragmatic
and semantic domains.
Because the pragmatic and social features of language are learned in contexts of
social differentiation, children whose knowledge of one language is limited to hearing
their parents can have embarrassing gaps in their ability to use the language in the
larger community. An example is formal style or social marking. Jun (1992) found
that Koreans brought up in the U.S. did not learn obligatory addressee honorifics
unless they spent many hours in Korean-speaking gatherings of varying age and rank.
This frequency variable, in addition to family attitudes toward correctness in Korean,
predicted pragmatic proficiency. The presence of nearby older relatives might be the
key, since Korean parents must use polite honorific forms to their own parents and
older siblings, and children learn from observation. The nuclear family is not enough
to give full social competence in a language.

3.5
Multilingual communities
When there is multilingualism in the local community, a child will encounter more
than one variety of language in school or neighborhood. Community use of the family
variety can provide the child with a formal oral style, literacy, and access to wider
register variation than the home offers (Lowie, 1945).
In the larger community, the family language may be heard in a local neighborhood,
and may even carry prestige on the playground, yet this community may be relatively
powerless in the larger society. In other conditions, the child’s home variety may be the
language of a powerful minority. An example of the latter would be children of families
of colonial rulers, as in Victorian India or South Africa, or Americans in Puerto Rico.
It remains to be seen in future research at what age children make use of macro-societal
prestige locally in their language use, for example in play.
In the case of communities where different ethnic groups live, children learn
different varieties of a language. For a decade, Zentella (1997) followed a group of
Puerto Rican “Nuyorican” children who were growing up in New York City’s East
Harlem. These “Nuyorican” children grew up learning up to seven varieties of Spanish
and English, such as Popular Puerto Rican Spanish, English-dominant Spanish, Puerto
Rican English, and African American Vernacular English. The children’s repertoire
was directly influenced by the community members and their social networks.
The International Journal of Bilingualism
 Child codeswitching 89

The simple measurement of relative lexical or syntactic proficiency does not begin
to exhaust the range of difference in possible pragmatic, social, and semantic develop-
mental histories that can affect research studies of bilinguals.

3.6
Instructional strategies
Family language values and the family strategies for dealing with mixing and switching
provide a contextual background for all code choice issues. Because of the inevitability
of lexical gaps, nonce borrowing must occur in every child, even in prescriptive families.
Several observers have reminded us that what parents say they do is not always what it
turns out on tape that they actually do. Inaccuracy of informal observation or recall,
especially where there are values about mixing or about the varieties used, is a familiar
caution in sociolinguistic research, (e.g., Gumperz, 1982, p.64). For this reason, we can
only trust recorded observations.
Lanza (1992) has provided us with a convenient list of increasingly permissive
instructional contexts. She suggests a cline in adult responses to child errors in language
choice, going from (a) minimal grasp by the hearer (such as a clarification request), (b)
expressed guess which reveals some comprehension, (c) repetition in the other language
(translation), (d) move-on (ignore), and (e) codeswitching at the instigation of the child.
It is not surprising to learn that most of the time families go about their business and
do not stop to do instruction on language each time a switch or insertion occurs (Juan-
Garau & Perez-Vidal, 2001).
Given the problem of lexical access for young children and the exigencies of daily
life, it is not altogether practical to respond at the instructional level to each instance.
However, we can expect that differences in family discourse strategies at the extremes
of the cline can have consequences for the child’s practices in the family. For example,
Juan-Garau and Perez-Vidal (2001) and Lanza (1992) observed that in asymmetric parent
bilingualism the parent who speaks the minority language is the one who strives the
hardest by using clarification requests because the threat of language loss is greatest. The
parent who speaks the least threatened language is likely even to accept codeswitching
as a communicative strategy.

4 Types and functions of codeswitching

The literature on the functions of various kinds of code alternation directs us to four
slightly different processes. One is discourse-related switching, as described by Auer
(1984), Halmari and Smith (1994), Wei (1998), and others. In the case of discourse-related
switching, it is a change itself that does the marking, just as in monolingual speech
lexical and prosodic discourse markers do their work by signaling change. Marking
of quotations, interjections, reiteration, clarifications, and emphasis on a request, or
marking teasing or punchlines by language switching can be of this kind. Alfonzetti
(1998) explicitly separates switching where direction matters from switching purely for
interactional functions such as self-repair, reformulation, topic change, story-telling,
and pre-closing of conversation.
The International Journal of Bilingualism
 90 S. Ervin-Tripp and I. Reyes

Lexical insertion or borrowing involves triggering of brief codeswitching, which is

often due to a second process that Zentella (1990) calls crutching, producing the most
immediately accessible lexical item in a semantic field. Such insertions are often related
to topic specializations (Lowie, 1945).
In some cases, codeswitching dependent on topic specialization is correlated with
well-defined language choice in the history of the speaker. The use of English textbooks in
many countries for science and professional instruction can lead to special language use
by field. People who do not live in bilingual communities or go to schools or universities
where situational switching occurs can nevertheless have a personal history which leads to
lexical specializations, such as professional or humanities education in a different country.
In his analysis of the socio-pragmatics of switching, Gumperz (1982) identified socially
predictable language choices based on addressee, location, event frame or genre, and
topic. He dubbed these all situational codeswitching. In established societal bilingualism,
situational codeswitching can be so predictable that deviations become marked (Myers-
Scotton, 1993) and available for tactical use (Myers-Scotton & Bolonyai, 2001).
A fourth function of codeswitching is what Gumperz (1982) has called metaphorical
or conversational switching, in which the switch is a contextualization cue. These are
switches that occur without any other cues, as part of the normal flow of conversa-
tion, “to communicate metaphoric information about how they intend their words to
be understood” (p.61). Among his examples is a type he called personalization versus
objectivization, the contrast being drawn between we and they, the family variety and
other varieties, emotional aspects usually being in the family variety. However, just
which semiotic contrasts are indexed by a switch may vary widely. For instance, in
public monologs, comedians play on the audience’s beliefs about the groups alluded to
by linguistic features — ignorance/education, conservatism/radicalism, cold/warm, and
so on, (Ervin-Tripp, 2001; Woolard, 1987). In these switches, direction matters since
the code alludes to or indexes social features, or involves conversational divergence
or convergence. Gumperz regards this switching as an optional semiotic device for
generating implicature.
These processes are not entirely independent. In a conversation between bilinguals,
it is not always clear when a particular switch is just interactional topic change marking, is
triggered by lexical specialization and accessibility, or is a contextualization tactic. Even in
the discourse-related switching described by Wei (1998) the particular language did matter
to the local interactional process, in ways that became clear through the exchanges.
Almost all studies of codeswitching over the years have been done on fluent adult
bilinguals and the claim has been made that codeswitching requires equal competence.
However, we will demonstrate that codeswitching occurs even before second-language
learners have achieved equal proficiency in both languages.

5 Developmental changes in bilingualism

5.1
Infant code choice
Studies of bilingual infants in person-prescriptive families reveal that from the beginning
the children differentiate the languages (Goodz, 1989; Lanvers, 2001; Lanza, 1992), and
The International Journal of Bilingualism
 Child codeswitching 91

if they know the necessary lexical items, make appropriate choices to a significant degree.
Controlled evidence of this skill has been developed by Quay (1995) and by Genesee,
Nicoladis and Paradis (1995) in Canada, using French/English bilinguals around two
years old. A particularly important feature of this body of work is the careful recording
of the lexical repertoire of each child for each language, which makes it possible to
identify when borrowing is based on gaps in lexical repertoire. What is not clear in such
research is how children normally cope with addressing both parents at the same time,
when the family rules oblige them to do situational switching according to addressee.

5.2
Borrowing in early child bilingualism
The insertion of single words or short phrases is very common in early child bilingualism.
The linguists who have looked at cases of infant bilingualism (Jisa, 2000) have noted
that when there is a difference in dominance, the dominant language is no different
from that of child monolinguals, but the weaker language may be deviant from that of
monolinguals of the same age. In addition, there is a language typology factor, in the
earlier emergence of morphosyntax in some languages. Thus a child who is bilingual
may be at a more advanced stage in one language than in the other.
In the case of Siri Lanza (1992; 1997) the child’s stronger language was Norwegian,
the language of the larger community. The language of the mother, and between the
parents, was English. Siri inserted lexemes across languages, but the basic system was
always Norwegian. She could borrow English words into Norwegian sentences, but
she never integrated Norwegian words into English morphology with English function
words. Petersen (1998) reports a Danish-English bilingual dominant in English who did
the same with English as the basic language.
At early stages of language development, insertional processes may be quite different
from those that occur later. A good comparison is provided by Jisa (2000), whose two
French-English bilingual daughters (2:3 and 3:6), after starting out in France with
French dominance and strong passive competence in English, were immersed for two
months in English at their differing stages of language development. As in the Lanza
and Peterson examples, these children did not insert French lexical material into English
bound and free grammatical morphemes, thus showing French dominance. One piece
of evidence that mixed utterances were pushing the limits of the children’s capacities
was that they were longer than the MLU in one-language utterances.
Jisa’s examples of borrowings by young children are conspicuously different from
those we see in older children and adults; for example in the case of pronouns, which
are freely inserted as if they were nouns.

(1) French-English bilingual age 3:8

tu DO WHAT avec le table
‘you do what with the table’
(2) tu SEE
‘you see’
The International Journal of Bilingualism
92 S. Ervin-Tripp and I. Reyes

(3) WHAT THIS tu GOT

‘what this you got’
(4) moi I GET DOWN
‘me, I get down’
(Jisa, 2000 p.1374)
The last example reflects the practice of Francophone children who often start
sentences with moi je as a strong assertion about the self. Paradis, Nicoladis and Genesee
(2000) show similar examples in Montreal, such as HE a pomme, I vas taxi, and ME
cacher loup. These examples show us that young children treat pronouns as content
words to be borrowed.

5.3
Functions of codeswitching
Evidence of situational switching reflects what is salient to young children as they
develop: people, place, activity, and genre. Person switching has proven to be the most
robust pragmatic factor in children’s language choices. Lanza (1997) noted that from
the earliest age, Siri would speak the appropriate language according to addressee, or
to get the hearer’s attention. Lanvers (2001) found self-corrective switching to parents
in a parent-prescriptive family by 1:7 and 1:10. According to experimental evidence of
Genesee, Boivin and Nicoladis (1996), bilingual children in Montreal are capable of
adapting their language choice to match the language used by an unfamiliar interlocutor
by two years of age. Jisa found slips in her two-year-old, who occasionally spoke English
to a French grandmother after her return from America, but none in the three-year-old.
The sources of this skill appear to be the language used by the other and communicative
breakdowns (Comeau & Genesee, 2001). Eventually children generalize to strangers
similar in appearance (Fantini, 1985; McClure, 1981).
A second frequent factor in early situational switching is physical setting. Children
come to recognize that language is different at school, at church, at the doctor’s and
at home, though these changes in setting typically also include personnel changes, so
a careful study would use controlled conditions. Jørgensen (1998) argues that until
eight years of age, person and setting are the main determinants, and that therefore
conversational and discursive switching start late.
Kwan-Terry (1992) identified activity as a third situational selector for her son, who
insisted that narratives from his bilingual mother always occur in Cantonese.
Whether metaphorical switching occurs in children has been under dispute. The
evidence suggests locally constructed cases of contrast. Young children engage in fantasy;
one could see rules about speech variety according to role as an extension of participant
situational switching. For example, Kwan-Terry’s son insisted that play characters and
animals speak and receive English. Language changes can be used to mark the voices in
play, the director’s voice, the role voice, the child’s voice (e.g., Halmari & Smith, 1994),
just as very young monolinguals vary pitch and phonological features of roles.
Discourse-related switching appears quite early. Lanvers (2001) found in early
German-English switching before age two the same functions of crutching, appeal, and
The International Journal of Bilingualism
 Child codeswitching 93

emphasis that Zentella (1990) described in older children. Topical switching appeared
before two, with quotational switching later.

5.4
Pragmatic changes in middle childhood
Around seven or eight, children develop new pragmatic skills such as being able to infer
intentions from hints or from sarcasm and to make requests that take into account
obstacles in another’s situation (Ervin-Tripp, 1982; Shantz, 1983). We expect that at this
age there will be more complex conversational strategies that may take the listener’s
perspective into account.
In a study of middle childhood peer talk, Ervin-Tripp (2000) developed a method
of collecting data which lets children mutually select one or two friends and leaves
them alone while being monitored with radio microphones. The conversational data
gathered using this methodology is rich in content and does not interfere with children’s
normal peer talk.
First, Ervin-Tripp examined what children at these ages are able to do within
monolingual speech or single language turns, by way of marking of conversational
structure. The markers used varied with function. They include prosodic changes,
vocatives, discourse markers like hey, look, but, OK, and repetition. It is known from
studies of early social interaction (Kyratzis & Ervin-Tripp, 1999), and from puppet play
(Andersen, Brizuela, DuPuy, & Gonnerman, 1999) that by age four children can already
use stylistic variation and choice of discourse markers to indicate social information.
The various markers we identified in the peer talk served most often to mark episodes
in an activity such as quoting or role-playing, to shift frames, and less often to attract
attention or grab a turn. Changing frame recalls Goffman’s discussions of footing (1974,
1979). What is shifted in the frame change can be genre, topic, speech act, or perspective.
The identifiers after each example indicate grade level, gender, and group. Asterisks
precede stressed words.
(5) Marking genre/activity change (seven-year-olds)
[boys discuss cleaning up lunch garbage]

a. Alan: [blows into microphone]

b. sh-h-h
Ë c. {[funny voice] hello = is this *on? = }
d. Len: = he-
e. {[funny voice] my name is *Lenny.}
UCDisclab T2M3
Sometimes the child appeared to be shifting perspective in a new idea or in a reply.
An example is marking the frame of a quotation, frequently by speeding up. Quoted or
imagined speech of course is itself represented by changes, usually stylistic, but also
often prosodic.
The International Journal of Bilingualism
94 S. Ervin-Tripp and I. Reyes

After frame change, the next most common function in the markers was a change
in role, shown by a vocal shift of some sort, often prosodic. This might be a shift from
narrator to speaker, or from narrative line to coda.
Reyes has used the same research method to observe bilingual children’s peer
talk (Reyes, 2004). The children in this sample are Spanish speakers from immigrant
families in northern California who were in a bilingual education program. All of the
children had first learned Spanish at home. The children we observed heard English
and Spanish unpredictably in the classroom, and their own language choice was left
uncontrolled in class. Teachers generally supported the use and development of both
Spanish and English at school, but many teachers and schoolmates were monolingual
English speakers.
When the children talked about science materials and bilingual worksheets in
the observational situation, they typically increased their use of Spanish. Clearly, one
cannot say that English is marked for school uses for these children.
Borrowing
These children all were competent in Spanish but differed in their relative skill in
English. The children with the least English skill did the least codeswitching and more
often inserted English lexical items into their Spanish. This borrowing was typically
of items we judged to have greater relative frequency of use in this community in the
source language, or which had unique semantics in that language. In these examples,
the identifiers indicate grade level, gender, and group.
(6) P: y luego el GUINEA PIG que tiene se salió a escondidas
‘and then the guinea pig that she has got out and hid’
UCDisclabW2F3
It is unlikely that a child would have encountered ‘guinea pig’ in California outside of
the school context.
(7) R: veniamos aquí cuando eramos STUDENT COUNCIL
‘we came here when we were [part of the] student council’
UC Disclab W5M3
(8) a. J: aquí esta la bolita
‘here’s the ball.’
Ë b. I: OH CO-O-OL
c. J: = =damelo, damelo
‘give it to me, give it to me’
UCDisclab W2M1
Many slang phrases that children use, such as oh cool have no equivalents and
carry connotations of hipness specifically in American mass culture. Brief phrases
can be acquired quickly as a first entry point into another language, before the child’s
productive skills make it comfortable to switch longer sentences. Many were from music.
The younger children (seven-year-olds) who were still developing their skills in English
were the ones who used nonce borrowing the most.
The International Journal of Bilingualism
 Child codeswitching 95

Codeswitching
Halmari and Smith (1994) and Cromdal and Aronsson (2000), observing bilingual
peer play in middle childhood, found many discursive uses of codeswitching, often at
the service of social dynamics such as teasing or anger, or to mark by-play. In group
interaction, codeswitching can be used to select participants or to display power.
The codeswitching in the Oakland dyads was most frequent by the most competent
and most balanced bilinguals. In the observational situation, there were three contexts
which might have lent themselves to situational switching — sociable talk while waiting
for the science task, school task-oriented talk about the science materials we brought,
including a worksheet in English and Spanish, and the sociable talk that was interspersed
into the task activity.
Given that the two languages occurred in the school both in the classroom and at
play, we did not find evidence of situational switching, for example between the on-task
science talk and sociable exchanges. However, the terminology for magnets, though
the objects were labeled in print in both languages, tended to be in English, so there
was English lexical borrowing during the science task when Spanish was preferred. The
largest amount of codeswitching in both age groups occurred during the sociable talk
that frequently alternated with the science task talk.
In example (9) codeswitching marked a change in footing that occurred during
the alternation of social and science talk, making use of the terminology for magnets
embedded in the children’s figurative speech play. L’s speech accelerates as she talks.
(9) Fifth grade girls during activity with magnets.
a. L: mira {[ac]mira mira mira mira mira}
‘look, look, look, look, look’
b. si lo pongo asi y lo quiero acercar ese se va
‘if I put it like this and I try to put it close to the other one it pulls apart’
c. E: [laugh]
Ë d. L: {[role voice] I DON’T WANT TO BE WITH YOU. I DON’T WANNA BE
e. WITH YOU *NORTH. I WANNA BE WITH *SOUTH MY *LOVE}
[both laugh]
Ë f. hora (‘ahora’) un SOUTH con un SOUTH
‘now a south with a south’
g. {[high, role voice] OH MY GOODNESS I LOVE NORTH.
h. I DON’T WANT YOU SOUTH}
i. {[crying] OH GET AWAY FROM ME}
UCDisclab W5F5
The example above shows how two fifth-grade girls use the concept of the polarity
between positive and negative electric charges of the magnets as they test the attraction
and repulsion of the magnets. They switch between languages as they switch footing
between the science task involving object manipulations and social imaginative play with
the objects as characters. Throughout the transcript, the two girls used both languages
The International Journal of Bilingualism
96 S. Ervin-Tripp and I. Reyes

in a variety of contexts, so it was not predictable which would be chosen to role play
the voice of magnetic poles.
Even in dyads, the children used switching to mark interactional processes, as
Halmari and Smith (1994) and Cromdal and Aronsson (2000) proposed for larger
play groups. The primary function of switching marking the sociable talk during the
science task was to indicate an off-task frame shift. Within sociable talk at both ages,
switching most often marked topic changes. Secondarily, switching marked emphasis for
commands, or clarification. In the science activity there also was some language change
to mark interactional function shifts such as those between assertions and questions,
and in the older children’s sociable talk there also was inter-turn accommodation to
follow up on a partner’s language choice. In (e) we see attention shifting, in (h) shifting
for a clarification, and in (g) the shift of O from the language of his prior turn seems
to be an accommodation to the prior shift of A.
(10) Fifth grade boys’ talk about magnets
a. O: con el *compas asi que da *vueltas la flecha?
‘with the compass this way the arrow turns around’
b. las *demas cosas de *clase?
‘the other things from class?’
c. A: = =*ira (mira).
‘look.’
d. O: se *pegan.
‘they stick together’
Ë e. A: *ira (mira) … LET’S DO *THAT ONE … LET’S SEE.
f. {[low] GOO::D … *POWER UP} … AAH::
Ë g. O: ROUND *THEM UP.
Ë h. son *MAGNETS. tienen MAGNETS *adentro.
‘they’re magnets … they have magnets inside’
i A: YUP.
UCDisclab W5M4
The codeswitching observed in child second language learners is very similar to
codeswitching observed in bilingual adults in terms of interactional functions, and some-
what expands the functions observed in infants. The variety of interactional functions
served by the switches increased with age. As children achieve higher levels of second
language fluency and social adaptability, their pragmatic and conversational repertoire
increases to achieve their more complex communicative goals, like accommodation.
Zentella (1997) has described the prevalence of codeswitching in Puerto-Rican
families in New York. Within families and in schools, values differ with respect to
codeswitching. Peers can have their own norms about switching. A very rich set of group
practices, including gender difference in norms, can be seen in Woolard’s description of
teenagers in Barcelona (Woolard, 1997). Woolard argued from her interview data that
there were different ideas about the social meaning of accommodation and divergence
for the boys and for the girls.
The International Journal of Bilingualism
 Child codeswitching 97

Identity functions in switching

Gumperz pointed out that “in bilingual situations the participants’ awareness of alter-
native communicative conventions becomes a resource, which can be built on to lend
subtlety to what is said” (Gumperz, 1982, p.65). This he referred to as contextualization.
In a family situation Myers-Scotton and Bolonyai (2001) argued for a case of tactical
language choice by a child which was marked because of family preference. However,
in our child peer data, there were as yet no clear semantic differentiation between the
two languages, only interactional functions for shifting, independent of direction as
far as we could tell. Sebba and Wootton argue that, even for adults,
although speakers may systematically differentiate the codes in conversational practice,
it is not possible to make a priori assumptions about which code carries the putative ‘we’
functions and which the putative ‘they’ functions. This can only be decided by looking
at the functions which the codes serve, which may vary from situation to situation and
cannot be treated as given. (Sebba & Wotton, 1988, p.276)

6 Changes in content with changes in language

The semiotic separation of languages implied in certain types of metaphorical code-
switching in adults suggests that languages (and related cultures) may call up different
semantic contexts. In the mid-fifties, the possibility that bilinguals may have different
experiences in their two language contexts that would be reflected in discourse in each
language had not yet been explored. Impelled by the report of bilingual friends that
they felt that they had several personalities according to language, Ervin undertook
two projects using the social psychological tools of the day. The first (Ervin, 1964b)
used the collection of three-minute stories about ambiguous Thematic Apperception
Test pictures. The participants were all French-born bilinguals who had learned the
languages in entirely different cultural contexts. English was a second language for all
of them. Many were married to Americans, and had raised families in the U.S. The
sessions were six weeks apart. Half the bilinguals told French stories first, and half told
English first. They were told that if they remembered the first story, they should think
of a new one. Ervin used media comparisons and cultural studies by anthropologists
to predict contrasts before collecting the data. Three were significant: in French the
subjects told more stories involving themes of automomy /passive aggression, and more
involving verbal aggression to peers, and the women subjects produced more achieve-
ment themes in English than in French.
An expanded study in the early sixties (Ervin-Tripp, 1964a, 1967) found female
Japanese-English bilinguals in order to enlarge the contrast of both culture and language.
There were two samples: Nisei born in the U.S., who spent part of their childhood,
often during the WWII years, with relatives in Japan, and Japanese “war brides” who
grew up in Japan, married Americans and moved to the U.S. Monolingual matched
samples were added, all subjects were female, and a variety of measures besides the
three-minute Thematic Apperception Test narratives about ambiguous pictures were
used — including chain word associations and sentence completions. Because of the
monolingual comparisons, it was possible to have a completely statistical measure of
proximity to monolingual norms, which often did not fit stereotypes. Because there was
The International Journal of Bilingualism
98 S. Ervin-Tripp and I. Reyes

a possibility that differences could be merely the result of role playing or self instruction
rather than due to change of language, some subjects were told to use the same language
but answer like a Japanese woman or like an American.
The results suggested that semantic separation varies with the task. All the
structured stimuli were chosen because cultural differences were expected. The most
consistent differences were on the word associations and sentence completions. When
speaking Japanese, the women gave word associations and sentence completions more
like Japanese monolinguals, and when speaking English, more like Americans.
When the stimulus word or phrase for chain word associations was a standard
contrastive cultural object like tea, familiar to everyone, there were strong contrasts both
by language and by instructional set. But for emotion words the associations are much
subtler, less often known, and contrasts much greater for language difference. (Similarly,
Block [1957] found that grief, guilt, love, sympathy were different for Norwegians and
Americans). For both word associations and sentence completions, the contrast with
language was larger for the Japanese who had learned English as adults than for the
Nisei. This makes sense in terms of the stronger separation of language and culture in
their life histories. The Nisei, who were childhood bilinguals, were relatively American
in both languages.
This issue has come back to life recently. There has been a resurgence of serious
study of the effects of language on concepts and memory, in widely varying languages,
using experimental methods to study observables such as spatial language (Lucy, 1997).
One version is in work on the study of concepts in language and culture, using a variety
of methods, as summarized recently by Pavlenko, who identified many of the factors
affecting these phenomena (Pavlenko, 1999, p.224).
Therapists who treat bilinguals have always been aware of the differences in
memory and sense of self in both languages. Koven (1998) found that discursive displays
of notions of the self differed in the Portuguese-French bilinguals’ languages. She argues
that this is possible because each language creates a new frame for the speaker. The
resulting display of differences in personality is very similar to the contrastive findings
in the Japanese-English research (Ervin, 1964a). These are some of the differences in
the sense of self, in values, and in concepts available to be drawn on when bilinguals
take different stances in codeswitching during dialogic discourse.
The differences repeatedly found in chain word associations involving cultural
values indicate that the conceptual domains that are accessed in lexical processing can
be reliably contrasted. However, these differences are greatest when there is contact with
relatively monolingual communities, or age difference in the learning, as in the case of
adult bilingualism. In ongoing bilingual communities there is a strong pressure to the
merger of semantic spaces and the compromise of the contrasts between the monolin-
guals not in contact. It can be argued that what happens in ongoing bilingualism is that
besides convergence, new semantic contrasts are constructed in the local group, as long
as the lexical possibilities are present and there is an interest in maintaining contrasting
varieties. There is evidence of both processes in studies of bilingual immigrant communi-
ties in which there has been maintenance without significant monolingual replenishment
(e.g., Haugen, 1950) or many generations of bilingualism (Gumperz & Wilson, 1971).
The studies on codeswitching have all found significant local conversational marking
The International Journal of Bilingualism
 Child codeswitching 99

of interactional processes with switches of code, but only some have found that the
direction of switching made a difference. Codeswitching in dialog can call on contrasts
in ideational reference, action inferences, social indexing, or emotional associations.
Each of these can be indexed in switching.
It seems likely that sociolinguistic and personal history factors that create large
contextual and cultural contrasts play an important part in producing shared differ-
ences. It is not an accident that social differentiation involving contrasting varieties
used in education was important in Gumperz’s work in India and Norway. Among adult
immigrants with similar histories, there can be shared contrasts in cultural frameworks
that are invoked in story-telling tasks or in meaningful codeswitching.
The codeswitching we see in infants grows out of a milieu with minimal differ-
ence to be linked with language; in older children and adolescents, codes are used for
locally constructed semiotic contrasts and for interactional processes. So it is only in
a special subset of bilingual lives that one can expect the carry-over of monolingual
experience to show up in either experimental studies of contrastive elicited discourse
or in spontaneous codeswitching.

7 Conclusions
From a very early age, children raised as bilinguals use the appropriate language for
addressees and for different contexts. When there are lexical gaps, borrowing occurs.
Codeswitching of longer segments can be found before age two. As children’s pragmatic
and social skills develop, codeswitching has more interactional functions, which supple-
ment their within-language, monolingual speech marking.
Some adult codeswitching makes use of semiotic differences implied by language
for speakers to signal the meaning context for interpretation. Certain bilinguals in
elicitation tasks show differentiated selves in their two languages. They give different
associative responses and sentence completions in their languages, which are more like
those in their monolingual network. Yet the normal tendency of continuing bilingualism
is to semantic merger, so we can expect that these types of separation, rather than
being universal in bilinguals, are to be found in particular conditions, such as educa-
tion in a second language, adult immigration, and frequent contact with monolingual
communities.

References
ALFONZETTI, G. (1998). The conversational dimension in code-switching between Italian and
dialect in Sicily. In P. Auer (Ed.), Code-switching in conversation. Language, interaction and
identity (pp. 180 – 214). London/New York: Routledge.
ANDERSEN, E. S., BRIZUELA, M., DUPUY, B., & GONNERMAN, L. (1999). Cross-linguistic
evidence for the early acquisition of discourse markers as register variables. Journal of
Pragmatics, 10, 1339 – 1351.
AUER, P. (1984). Bilingual conversation. Amsterdam; Philadelphia: J. Benjamins.
BERTHELE, R. (2000). Divergence and convergence in a multilectal classroom: Patterns of group
structure and linguistic conformity. In K. Mattheie (Ed.), Dialect and migration in a changing
Europe (pp. 155 – 171). Frankfurt, Germany: Peter Lang.
The International Journal of Bilingualism
100 S. Ervin-Tripp and I. Reyes

BLOCK, J. (1957). Studies in the phenomenology of emotion. Journal of Abnormal and Social
Psychology, 54, 358 – 363.
BLOM, J.-P., & GUMPERZ, J. J. (1971). Social meaning in linguistic structures: Code-switching in
Norway. In J. J. Gumperz & A. Dil (Eds.), Language in social groups (pp.174 – 310). Stanford
CA: Stanford University Press.
COMEAU, L., & GENESEE, F. (2001). Bilingual children’s repair strategies during dyadic communi-
cation. In J. Cenoz & F. Genesee (Eds.), Trends in bilingual acquisition. (231 – 256). Amsterdam:
John Benjamins.
CROMDAL, J., & ARONSSON, K. (2000). Footing in bilingual play. Journal of sociolinguistics,
4, 435 – 457.
ECKERT, P. (1989). Jocks and burnouts: Social categories and identity in the high school. New York:
Teachers College Press.
ERVIN, S. M. (1964a). An analysis of the interaction of language, topic, and listener. American
Anthropologist, 86 – 102.
ERVIN, S. M. (1964b). Language and TAT content in bilinguals. Journal of Abnormal and Social
Psychology, 68, 500 – 507.
ERVIN-TRIPP, S. M. (1967). An Issei learns English. Journal of Social Issues, 23(2), 78 – 90.
ERVIN-TRIPP, S. M. (1982). Ask and it shall be given you: Children’s requests. Georgetown
Roundtable on Languages and Linguistics, 235 – 245.
ERVIN-TRIPP, S. M. (2000). Studying conversation: How to get natural peer interaction. In
L. Menn & N. B. Ratner (Eds.), Methods for studying language production. (pp. 271 – 288).
Hillsdale, NJ: Lawrence Erlbaum Associates.
ERVIN-TRIPP, S. M. (2001). Variety, style-switching, and ideology. In P. Eckert & J. Rickford (Ed.),
Style and variation. (pp. 44 – 56). New York: Cambridge University Press.
ERVIN-TRIPP, S. M.,& GUO, J. (1992). Factors in individual mother tongue maintenance in
California. Unpublished paper.
ERVIN-TRIPP, S. M., STRAGE, A., LAMPERT, M., & BELL, N. (1987). Understanding requests.
Linguistics, 25, 107 – 143.
FANTINI, A. E. (1985). Language acquisition of a bilingual child: A sociolinguistic perspective (to
age 10). Clevedon, Avon England: Multilingual Matters.
GENESEE, F., BOIVIN, I., & NICOLADIS, E. (1996). Talking with strangers: A study of bilingual
children’s communicative competence. Applied Psycholinguistics, 17, 427 – 442.
GENESEE, F. NICOLADIS, E., & PARADIS, J. (1995). Language differentiation in early bilingual
development. Journal of Child Language, 22, 611 – 631.
GOFFMAN, E. (1974). Frame analysis. London: Harper and Row.
GOFFMAN, E. (1979). Footing. Semiotica, 25, 1 – 29.
GOODZ, N. S. (1989b). Parental language mixing in bilingual families. Infant Mental Health
Journal, 10, 25 – 44.
GROSJEAN, F. (1982). Life with two languages: An introduction to bilingualism. Cambridge, MA:
Harvard University Press.
GUMPERZ, J. (1982). Discourse strategies. Cambridge: Cambridge University Press.
GUMPERZ, J., & WILSON, R. (1971). Convergence and creolization: A case from the Indo-Aryan/
Dravidian border in India. In D. Hymes (Ed.), Pidginization and creolization of languages
(pp. 151 – 167). Cambridge: Cambridge University Press.
HAKUTA, K. (1994). Distinguishing among proficiency, choice, and attitudes in questions about
language for bilinguals. In G. Lamberty & C. Garcia Coll (Eds.), Puerto Rican women and
children (pp. 191 – 209). New York: Plenum Press.
HALMARI, H., & SMITH, W. (1994). Code-switching and register shift: Evidence from Finnish-
English child bilingual conversation. Journal of Pragmatics, 21, 427 – 445.
JISA, H. (2000). Language mixing in the weak language: Evidence from two children. Journal of
Pragmatics. Special Issue: Codeswitching, 32, 1363 – 1386.
The International Journal of Bilingualism
 Child codeswitching 101

JØRGENSEN, J. N. (1998). Children’s acqusition of codeswitching for power yielding. In P. Auer

(Ed.), Codeswitching in conversation. Language, interaction and identity (pp.237 – 258). London/
New York: Routledge.
JUAN-GARAU, M.,& PEREZ-VIDAL, C. (2001). Mixing and pragmatic parental strategies in
early bilingual acquisition. Journal of Child Langage, 28, 59 – 86.
JUN, D. (1992). Bilingualism among Korean-Americans: on factors related to the knowledge of
honorifics of the Korean language. Unpublished honors thesis. University of California,
Berkeley, Psychology Department.
KOVEN, M. E. J. (1998). Two languages in the self/the self in two languages: French-Portuguese bilin-
guals’ verbal enactments and experiences of self in narrative discourse. Ethos, 26, 410 – 455.
KWAN-TERRY, A. (1992). Code-switching and code-mixing: The case of a child learning English
and Chinese simultaneously. Journal of Multilingual and Multicultural Development, 13,
243 – 259.
KYRATZIS, A.,& ERVIN-TRIPP, S. (1999). The development of discourse markers in peer inter-
action. Journal of Pragmatics, 31, 1321 – 1338.
KYRATZIS, A., MARX, T., & WADE, E. R. (2001). Preschoolers’ communicative competence:
Register shift in the marking of power in different contexts of friendship group talk. In H.
Marcos (Ed.), Early pragmatic development [Special issue]. First Language, 21, 387 – 343.
LANVERS, U. (2001). Language alternation in infant bilinguals: A developmental approach to
codeswitching. International Journal of Bilingualism, 5, 437 – 464.
LANZA, E. (1992). Can bilingual two-year-olds code-switch? Journal of Child Language, 19,
633 – 658.
LANZA, E. (1997). Language mixing in infant bilingualism: A sociolinguistic perspective. Oxford:
Clarendon Press.
LEOPOLD, W. F. (1939 – 1949). Speech development of a bilingual child. Evanston IL: Northwestern
University Press.
LOWIE, R. H. (1945). A case of bilingualism. Word, 1, 249 – 259.
LUCY, J. A. (1997). Linguistic relativity. Annual Review of Anthropology, 26, 291 – 312.
MCCLURE, E. (1981). Formal and functional aspects of the code-switched discourse of bilingual
children. In R. P. Duran (Ed.), Latino language and communicative behavior (pp. 69 – 94).
Norwood NJ: Ablex.
MYERS-SCOTTON, C. (1993). Social motivations for codeswitching. Oxford: Clarendon Press.
MYERS-SCOTTON, C., & BOLONYAI, A (2001). Calculating speakers: Codeswitching in a
rational choice model. Language in Society, 30, 1 – 28.
PARADIS, J., NICOLADIS, E., & GENESEE, F. (2000). Early emergence of structural constraints
on code-mixing: Evidence from French-English bilingual children. Bilingualism: Language
and Cognition, 3, 245 – 261.
PAVLENKO, A. (1999). New approaches to concepts in bilingual memory. Bilingualism: Language
and Cognition, 2, 209 – 230.
PETERSEN, J. (1998). Word-internal code-switching constraints in a bilingual child’s grammar.
Linguistics, 26, 479 – 493.
QUAY, S. (1995). The bilingual lexicon — implications for studies of language choice. Journal of
Child Language, 22, 369 – 387.
REYES, I. (2004). Functions of code switching in schoolchildren’s conversations. Bilingual Research
Journal, 28(1), 77 – 98.
RONJAT, J. (1913). Le développement du langage observé chez un enfant bilingue. Paris:
Champion.
SEBBA, M., & WOOTTON, T. (1998). We, they and identity: Sequential versus identity-related
explanation in code-switching. In P. Auer (Ed). Code-switching in conversation; language,
interaction, and identity (pp. 262 – 290). London/New York: Routledge.
The International Journal of Bilingualism
102 S. Ervin-Tripp and I. Reyes

SHANTZ, C. U. (1983). Social cognition. In P. H. Mussen (Ed.), Handbook of child psychology.

(pp. 495 – 555). New York: Wiley.
VALDÉS, G. (1996). Con respeto: Bridging the distances between culturally diverse families and
schools: An ethnographic portrait. New York: Teachers College Press.
VALDÉS, G. (2003). The caser of young interpreters from immigrant communities. Hillsdale, NJ:
Lawrence Erlbaum Associates.
WEI, L. (1998). The “why” and “how” questions in the analysis of conversational code-switching.
In P. Auer (Ed.), Code-switching in conversaton; language, interaction, and identity. London/
New York: Routledge.
WONG-FILLMORE, L. (1991). When learning a second language means losing the first. Early
Childhood Research Quarterly, 6, 323 – 346.
WOOLARD, K. (1987). Codeswitching and comedy in Catalonia. IPrA Papers in Pragmatics, 1,
106 – 122.
WOOLARD, K. A. (1997). Between friends: Gender, peer group structure, and bilingualism in
urban Catalonia. Language in Society, 26, 533 – 560.
ZENTELLA, A. C. (1990). Integrating qualitative and quantitative methods in the study of bilingual
code-switching. In E. Bendix (Ed.), The use of linguistics: Annals of the New York Academy
of Science Vol. 583 (pp. 75 – 92). New York: Academy of Science.
ZENTELLA, A. C. (1997). Growing up bilingual. Malden MA: Blackwell.

The International Journal of Bilingualism