Code switching and code mixing in Internet chatting: between

the
jaltcalljournal
ISSN 1832-4215
Vol. 5, No. 3 Pages 67–78
©2009 jalt Call SIg
Code switching
This case study examined the occurrences of
and code mixing
code switching and code mixing in a chatroom
based environment. In Fall 2004, the chat
in Internet
room conversations of 12 non-native speakers
of English from Spanish and Indonesian back-
chatting: between grounds were collected during a two month
period and analyzed to identify: 1) frequency
‘yes’, ‘ya’, and ‘si’ a of code switching and code mixing for both
cultures; 2) topics that triggered code switch-
case study
ing and code mixing in each culture; and 3)
topics common to both cultures and topics
less likely to occur within both cultures. The
findings suggest that technology-related terms,
along with introductory terms, triggered more
Mónica Stella Cárdenas-Claros
instances of code switching and code mixing
The University of Melbourne
regardless of the linguistic background of the
participants. Conclusions and suggestions for
further research are provided.
Neny Isharyanti
Satya Wacana Christian University,
Introduction
Salatiga, Indonesia
Developing communicative competence in
two or more languages gives individuals
opportunities to express their feelings and
thoughts and shape their identity. It also
helps them satisfy their individual and so-
cial needs in the different contexts of the
languages used. The phenomena of code
switching and code mixing of languages
have long intrigued scholars who have
examined what triggers such occurrences
(Muysken, 2000; Wei, 2005). However, most
research has been in face-to-face communi-
cation and in bilingual communities (See
Chan, 2004; Muysken, 2000; Myer-Scotton,
1992; Wei, 1998) with few studies in the
context of computer mediated communica-
tion (Danet & Herring, 2003; Durham, 2003;
Goldbarg, 2009; Ho, 2006; Huang, 2004; ).
Forum
Such studies suggest that research needs 67

---

The jalt call Journal 2009: Forums
to examine the different facets of code switching and code mixing in CMC contexts, while
keeping in mind cultural differences.
This case study examines the occurrences of code mixing and code switching produced
during interactions in a chat room environment by advanced users of English from Spanish
and Indonesian backgrounds. The paper starts by defining key terms and reviewing litera-
ture that covers the study. Then, it provides a rich description of the participants, data collec-
tion, and data analysis. Next, the paper presents the results in two sections. The first section
identifies the key topics that trigger code switching and code mixing more frequently. The
second section compares the topics based on the cultural traits and classifies code mixing
occurrences under the headings of alternation, insertion, and congruent lexicalization.
The paper concludes with a discussion of the findings and suggestions for further research.
Situating the study
Examining topics that trigger code switching and code mixing in Internet chatting requires
an understanding of the main concepts that frame this study. In the first part of the litera-
ture review, we discuss the definitions of code switching and code mixing and use examples
from our data set to illustrate each. Following a brief description of CMC, we also discuss
how the traditional distinction between spoken and written language is blurred in com-
puter mediated communication. In the second part, we examine studies that investigate
code switching in computer mediated environments.
Code alternation
The distinction between code switching and code mixing is one of the most puzzling debates
in the study of code alternation. Clyne (1991) argues that code switching and code mixing
refer to the same phenomena in “which the speaker stops using language ‘A’ and employs
language ‘B’ ” (p. 161). Romaine (1995) views code switching as a phenomenon that occurs in
a continuum where both inter-sentential and intra-sentential code alternation takes place.
Other researchers make the distinction between code switching and code mixing based on
the place where the alternation occurs. Wei (1998) notes that if code alternation occurs at
or above clause level, it is considered code switching, but if it occurs below clause level then
it is considered code mixing. These are the definitions that we adopt for the current study.
Code switching
Code switching or inter-sentential code-alternation occurs when a bilingual speaker uses
more than one language in a single utterance above the clause level to appropriately con-
vey his/her intents. Fischer (1972) suggests that language or code choice in communities
where bilingualism or multilingualism is the norm should be analyzed in the context where
the speech is produced. Fischer notes that three contextual factors should be taken into
account: 1) the relationship amongst speakers; 2) the setting where the talk takes place
and; 3) the topic being discussed. In this respect, Myers-Scotton (1992) notes that not only
contextual factors play a role in the code choice, but factors such as social identity and
educational background also affect the speaker’s choice of code. Huang (2004) adds up the
medium used with a number of factors believed to trigger code alternation. Thus, conver-
68

---

Cárdenas-Claros & Isharyanti: Code switching and code mixing in Internet chat
sations taking place in CMC environments trigger the change of code. Excerpt 1 illustrates
an instance of code switching.
Excerpt 1: Example of code switching English/Spanish
A: The picture looks so cool.
B: Which picture?
A: The one you have in your messenger.
B: Ah…Si, me gusto mucho. (Ah…Yes, I liked it a lot.)
 
Excerpt 1 shows how participant B interacted in English during most of the conversation
and suddenly switched into Spanish.
Code mixing
Code mixing also called intra-sentential code switching or intra-sentential code-alternation
occurs when speakers use two or more languages below clause level within one social situ-
ation. Muysken (2000) defines three types of code mixing: insertion, alternation, and con-
gruent lexicalization. In his view, insertion occurs when lexical items from one language
are incorporated into another. The notion of insertion, according to Muysken (2000), cor-
responds to what Clyne (1991) terms as “transference” and Myer-Scotton as “embedding”.
Figure 1 illustrates a graphic representation of insertion.
 
A
B
A
...a...
...b...
...a...
Taken from Muysken (2000, p.7)
Figure 1. Example of insertion in code mixing  
In the diagram “a” represents lexical items of the first language and “b” stands for the
lexical item of the second language that has been inserted in the utterance by the speaker.
Instances of this category of code mixing found in the data can be seen in excerpts 2a and
2b. Excerpt 2a occurs when a Spanish speaking participant discusses a paper for one of his
classes. It seems that the shared professional background with the researchers’ and the
participants’ specific language are the reasons that induce him to unconsciously insert an
English lexical item into a Spanish conversation.
Excerpt 2b occurs when the participants talks about the factors that affect the price of
a ticket of a live NBA game.
69

---

The jalt call Journal 2009: Forums
Excerpt 2a: Example of insertion (Spanish/English)
B: Pero bueno creo que basta con que incluya la pregunta de enhanced output más
todas las demás.
(Well, I think it is enough if I just include the question of enhanced output.)
Excerpt 2b: Example of insertion (Indonesian/English)
B: Tergantung team, terus juga tergantung event.
(It depends on the team and on the event.)
 
The second category proposed by Muysken (2000), alternation, occurs when structures of
two languages are alternated indistinctively both at the grammatical and lexical level. This
definition is illustrated in Figure 2.

A
B
...a...
...b...
Taken from Muysken (2000, p.7)
Figure 2. Example of alternation in code mixing
 
In the diagram, A & B represent structures of the two languages that reflect the alterna-
tion that takes places in the utterances produced by the speakers. Excerpts 3a and 3b are
examples of alternation. Excerpt 3a occurs when the researcher praises the picture of the
new bicycle of the participant. In this excerpt the participant uses English in the first part of
the sentence, but when it comes to finding a sentence to translate “pretty girl” he switches
code into Spanish. Excerpt 3b occurs when the Indonesian participant talks about a paper
about sentence choice that he needs to submit. He uses the expression “I mean” to introduce
the rest of his utterance in his first language.
 
Excerpt 3a: Example of alternation (English/Spanish)
B: I just have it in my room like a niña bonita como debe ser.
(I just have it in my room like a girl pretty as it should be.)
Excerpt 3b: Example of alternation (English/Indonesian)
B: I mean, ganti ke kalimat laen.
(I mean, change it to another sentence.)
 
70 The third and last category in code mixing is congruent lexicalization, which refers to the

---

Cárdenas-Claros & Isharyanti: Code switching and code mixing in Internet chat
situation where two languages share grammatical structures which can be filled lexically
with elements from either language (Muysken, 2000). Figure 3 illustrates this phenomenon.
A/B
...a...
...b...
...a...
...b...
Taken from Muysken (2000, p.8)
Figure 3. Example of congruent lexicalization in code mixing
Excerpts 4a and 4b are examples of congruent lexicalization. Excerpt 4a occurs when the
participant describes the way he uses his cell phone to connect to the Internet.
In Except 4b the participant states that the software that he usually uses to convert his
music file has expired.
Excerpt 4a: Congruent lexicalization (Indonesian/English)
B: Gw konek pake cellp gw.
(I connected using my cell phone.)
Excerpt 4b: Congruent lexicalization (Indonesian/English)
B: Software gua buat convert file wav jadi mp3 gua uda expired.
(My software for converting wav files to mp3s has expired.)
Computer mediated communication and characteristics of language in
Internet chatting
CMC, defined as “communication that takes place between human beings via the instrumen-
tality of computers” (Herring, 1996, p. 1), is seen as an alternative that provides computer
users with a number of options (chat rooms, blogs, instantly delivered messages, etc.) to
be used according to the communication needs of each individual. In terms of delay, there
are two types of CMC: synchronous and asynchronous. Synchronous communication or in-
teraction that takes place in real time via relay chats, chat rooms, instant messaging, voips
and twits; asynchronous communication, or interaction that allows CMC users to access the
media at a different time includes emails, blogs, and wikis among others.
Crystal (2001) asserts that these new forms of communication have triggered an evolu-
tion in concepts such as a spoken and written language. Traditionally, language is con-
ceived into two categories: spoken and written genres. While written language tends to-
wards structural complexity, formality and abstraction, spoken language is more context 71

---

The jalt call Journal 2009: Forums
dependant and structurally simpler. In CMC environments, one of the most striking fea-
tures of language is the blurring of the spoken and written distinction (Herring, 2001).
Thus, written communication in CMC environments, especially in synchronous communica-
tion, resembles most of the features of dialogues produced in face-to-face communication.
Although Internet chatting, in general, uses the written medium, the language used is less
correct, less complex, and less coherent than standard written language, making it closer
to speech (Danet & Herring, 2007). Crystal (2001) contends that many of the colloquial fea-
tures of spoken language such as short constructions, phrasal repetition, looser sentence
construction, and the use of reaction signals (you know, you see) are present in the texts
produced in synchronous communication in CMC contexts. In his view, users simplify lan-
guage to meet the goal of interactive communication.
Research on code switching and code mixing in computer mediated
environments
Most studies in code switching in CMC environments have examined asynchronous com-
munication. One of the earliest studies to examine the code switching phenomena is that
of Warschauer, El Said & Zohry (2002). They investigated English and Arabic language use
in email communication by a group of young professionals. They found that English was
more frequently used both when searching the Internet and in formal (business related)
email communications. They also found that from the two varieties of Arabic used in Egypt,
classical Arabic and the Egyptian Arabic, a Romanized version of Egyptian Arabic was most
frequently used in informal email messages and in online chats. Additionally, this choice
of code was also preferred by participants when they wanted to express highly personal
content. In a similar study but with a quite sophisticated methodology, Goldbarg (2009)
examined the Spanish-English code switching in email exchange of five Latin American
participants. In her study, she found that English was mostly associated with communi-
cations that dealt with professional and work related issues. Spanish, the participants L1,
was mostly used in communications that reflected a degree of intimacy, informality, and
group identification.
Durham (2003) investigated language choice in a Swiss mailing list by examining a
corpus of 996 emails collected from 1999 to 2002. The participants were students attending
various medical schools in Switzerland where instruction is imparted in German. She found
the use of English in the email list increased dramatically over time (from 10% to 80% in a
four year period). She explained that this trend was due to the fact that English acted as a
non-native language for all participants. Additionally, she found that the use of more than
one language was impractical and confusing in the email exchange, so participants tended
to rely on English as the main language.
Huang (2004) examined code choice and language use in the emails used for interper-
sonal communication written by 8 Chinese-English bilinguals in Taiwan. He reported that
from the analysis of a corpus of 223 emails supplemented with interviews and question-
naires, participants adopted three modes of email communication: Chinese/English bilin-
gual mode, Chinese monolingual mode, and English monolingual mode. The Chinese mono-
lingual mode was used when participants wanted to express their most personal thoughts
and feelings. Additionally, this mode was preferred when portraying their local identities.
The English monolingual mode was used by participants to display “an embrace of inter-
72 national and Internet identity and of younger generation identity” (p.307). Huang (2004)

---

Cárdenas-Claros & Isharyanti: Code switching and code mixing in Internet chat
also reported that contrary to the language used in face to face communication in which
Chinese is mostly used, a Chinese monolingual mode was the least preferred mode choice in
email writing. Additionally, Huang reported that topics related to movies, shopping, sports,
computers, along with food triggered most of the instances of code switching in his data.
Ho (2006) investigated the bilingual practices of 21 tertiary students in Hong Kong
when using ICQ – an instant messaging computer program. Upon analysis of 40 ICQ his-
tories, she suggests that English and Chinese languages were used to complement each
other, helping participants deal with the pressure of instant communication present in
synchronous CMC. In addition, she found that while English use was mostly associated
with technology and academic matters, Chinese use was associated with traditions. She also
notes that the complementary use of the two languages is a clear reflection of the identity
of new generations in Hong Kong whose culture has been shaped by the blend of Chinese
and Anglophone cultures.
The above studies point to the tendency of bilingual speakers to use the language they
identify the most with to express personal thoughts and topics that convey a degree of
intimacy. Computer-related topics along with Anglophone culture-related topics are most
likely to be used in English. Given the limited body of research on code switching and code
mixing in synchronous communication we find it pertinent to observe and analyze how
these phenomena actually occur in Internet chatting conversations taking into account
Internet users from two distinct linguistic backgrounds.  This case study strives to find
answers to the following research questions:
1. What is the number of occurrences of code switching and code mixing between Spanish
speaking and Indonesian speaking participants?
2. Does researcher-initiated code switching prompt participants from both linguistic back-
grounds to shift code?
3. On what topics and functions of the language do speakers seem to switch the
most? Additionally, what topics are common and/or different regardless of cultural
background?
4. What is the most frequent code mixing category in which participants mixed languages?
Method
Participants
The participants in this study were 12 male participants. Six participants were from an
Indonesian background and six participants were from a Latin American background
representing five nations: Colombia, Chile, Uruguay, Mexico, and Argentina. The partici-
pants were advanced speakers of English enrolled in different graduate school programs at
American universities. The decision to exclude female participants was due to the limited
number of female graduate students with whom the researchers conversed.
In contrast to research on code switching and code mixing that examined letters, notes,
homework, and to-do lists, the data for this study is a compilation of 84 transcripts of
original, typed conversations collected through the MSN Messenger. Data was collected
for two months and no topics were chosen systematically. Instead, they emerged as the
conversations developed with each of the researchers in a natural setting. To make the
data comparable and consistent between the two languages, we examined 20 instances of
conversations in Indonesian and 20 in Spanish. To decide which conversations to use as 73

---

The jalt call Journal 2009: Forums
the basis for our data, each data set was complied with three criteria: gender, number of
conversations, and occupation.
Participants were invited to participate in the study based on the frequency they used
the MSN Messenger to communicate with the two female researchers of this study. The
ethics forms sent did not include detailed explanations on the kind of linguistic feature
that was to be analyzed. MSN Messenger conversations initiated either by the researchers
or by the participants were saved.
Data analysis
Upon collecting the transcripts of the chatting conversations, the instances of code switch-
ing and code mixing in each language were identified and counted.
For the analysis of code switching, occurrences of three variables were taken into ac-
count: a) researcher-initiated switching followed by the participant; b) researcher-initiated
switching not followed by the participant and, c) participant initiated. The occurrences of
code switching were then counted and classified into the three variables and later in each
category, the topics and the frequency of the occurrences were identified. Similarly, the oc-
currences of code mixing were counted and classified into three categories: a) insertion, b)
alternation, and c) congruent lexicalization. Upon classification, the results were analyzed
and compared to draw conclusions.
Results
Although 20 conversations were analyzed for each language, the Spanish researcher-partic-
ipant chatting exchanges resulted in longer interactions in terms of word count than those
between the Indonesian-speaking researcher and her counterparts. The chatting exchanges
of Spanish speaking participants resulted in 1935 lines (9113 words), while the Indonesian
exchanges resulted in 1035 lines (4119 words). Not surprisingly, this affects the number of
code switching occurrences between the two groups.
What is the number of occurrences of both code switching and code mixing between linguistic
backgrounds?
There are marked differences between the number of code switching and code mixing occur-
rences between linguistic backgrounds. In the 20 conversations analyzed for the Spanish-
speaking participants a total of 119 alternations took place. 81 correspond to code switching
and 35 to code mixing. Indonesian speakers exhibited a higher number of alternations
adding up to a total of 174 shifts in code choice. 110 correspond to code switching and 64
to code mixing.
Does researcher-initiated code switching prompt participants to shift code?
For Spanish speakers, 23 (28%) occurrences fit into the category of researcher-initiated
code switching and 58 (72%) occurrences fit into the category of participant-initiated code
switching. Of the 23 instances of researcher-initiated code switching, change of code was
triggered 17 times (78%), one turn immediately after the researcher had switched from
74 Spanish to English. Additionally, change of code was followed after two turns on one

---

Cárdenas-Claros & Isharyanti: Code switching and code mixing in Internet chat
occasion. On 5 (22%) occasions participants did not follow the code imposed by the re-
searcher. Indonesian speakers displayed a similar trend. Of 110 occurrences of code switch-
ing, 70 (64%) happened when the researcher switched her chatting conversation to the
other language and 40 (36%) were initiated by the participants. 
Out of 70 instances of researcher-initiated code switching for Indonesians, on 36 (51%)
occasions the participants followed the code switching and switched to English, whereas in
34 (49%) occurrences, the participants remained in their L1 as illustrated in Table 1.
Table 1. Occurrences of code switching
Indonesians
Latin Americans
Researcher-initiated
Participant-
Researcher-initiated
Participant-
switching
initiated
switching
initiated
Switch
Non-switch switching
Total Switch
Non-switch switching
Total
36
34
40
110
18
5
58
81
Number of
Number of
turns to switch
turns to switch
1–2 turns 30
1–2 turns 17
3–4 turns
4
3–4 turns
1
over 5 turns 2
over 5 turns 0
On what topics and functions of the language do speakers seem to switch the most? Additionally,
what topics are common and/or different regardless of culture?
As expected, the function of the language that triggered the most code switching occur-
rences is confirming (21%). This fact is explained as the relevant English words (i.e. yeah, ok,
yes, etc.) are commonly used in colloquial Spanish and English. In addition to confirming,
the top-three topics and functions of the language that triggered the most code switching
and code mixing occurrences are: farewells (11%), computer-related terms (10%), academics
(9%), and sports (6%) regardless of language background. While Spanish-speaking par-
ticipants tended to shift code when a discussion on academics (17%) and computer-related
terms (14%) took place as well as when saying goodbye (9%), Indonesians shifted code when
functions of the language such as saying goodbye (23%) and topics such as sports (19%),
feelings, and computer-related terms (13%) were triggered. Some of the topics not shared
by the participants in both languages include money, music, and sports.
What is the most frequent code mixing category in which participants mixed languages?
Results for code mixing for Spanish-speaking participants show that from the 35 instances
identified, 32 instances (92%) correspond to the category of insertion, 3 (8%) fit into the
description of alternation and none of the occurrences fulfils the definition of congruent
lexicalization. In terms of code mixing for Indonesian participants, of the 66 instances 53
occurrences fit the classification of insertion which correspond to 83% of the total turns
where code mixing was present, 6 (9%) of occurrences correspond to alternation and 5 (8%)
correspond to congruent lexicalization.
Regarding code mixing, both Indonesian and Spanish participants used more insertion 75

---

The jalt call Journal 2009: Forums
than alternation or congruent lexicalization. In fact, the number of insertion occurrences
is significantly higher than the two other types, 83% and 92% respectively.  (Table 2)
Table 2. Occurrences of code mixing classification
Type of code mixing
Spanish subjects
Indonesian subjects
Insertion
35
53
Alternation
3
6
Congruent lexicalization
0
5
Total
38
64
Discussion
Indonesian participants shifted code more often than Spanish speaking participants. While
Spanish speakers switched code 116 times, Indonesians switched code 174 times. This re-
sult is interesting, considering that the number of exchanges of the Spanish-speaking
participants is longer than the Indonesian ones (1935 lines, 9113 words vs. 1035 lines, 4119
words). A closer look at the data indicates that most code alternation was triggered when
participants used the function of the language “confirming”. This can be explained by the
phonetic similarity between Indonesian and English “yes”, “yeah”, and “ya” if compared to
“yes” and “Si” in Spanish.
The personalities and individual interests of the researchers can be considered as fac-
tors that may have influenced the results of the study. While the Indonesian researcher
tends to be more inquisitive and asks many questions at once, the Spanish-speaking re-
searcher tends to follow turn taking conventions. This fact could explain the difference
between the percentage of researcher-initiated code switching occurrences that were not
followed by the participants in both languages (22 % for Spanish and 49% for Indonesians).
Also, while the Spanish speaking researcher is academically-oriented and has a passion
for reading and swimming – two highly individual activities – the Indonesian researcher
has interests in more group oriented activities like band playing, sports, and computer
social networking. These individual interests are somehow reflected in the topics selected
from participants of each cultural background. Spanish-speaking participants talked about
academic-related topics and used the function of the language “asking questions” during
their chatting conversations. In contrast, Indonesian participants preferred to talk about
non-academic topics, such as sports and sharing their feelings in the chatting activities.
Topics such as sports and academic matters that were not shared by the participants from
both linguistic backgrounds also reflect the individual interests the participants have.
Interestingly, participant-initiated code switching is higher for Latin Americans (72%
compared to 36% for Indonesians). Most of the code switching was done from Spanish
to English when academic topics were discussed. It seems that the shared academic back-
ground with two of the participants with the researcher can account for this.
Most words used for saying goodbye such as “bye”, “see you”, and “talk to you later” as
well as computer-related terms such as “email”, “PC”, and “attachment” have been bor-
rowed from English by both languages and their use has spread rapidly. Evidence of this
is that Spanish and Indonesian speaking participants whose linguistic and cultural back-
grounds are different use them indistinctly. These findings support Crystal’s (2001), Danet
& Herring’s (2007), and Ho's (2006) assumptions about how the emergence of computer-re-
76 lated terms in the Anglophone culture transfer easily regardless of the language where they

---