BIROn - Birkbeck Institutional Research Online

Dewaele, Jean-Marc and Li, Wei (2014) Intra- and inter-individual variation
in self-reported code-switching patterns of adult multilinguals. International
Journal of Multilingualism 11 (2), pp. 225-246. ISSN 1747-7530.

Downloaded from:

Usage Guidelines:
Please refer to usage guidelines at or alternatively
contact lib-eprints@bbk.ac.uk.
Intra- and inter-individual variation in self-reported code-switching
patterns of adult multilinguals1
Jean-Marc Dewaele
Birkbeck, University of London
Li Wei
Birkbeck, University of London

Corresponding author:
Dr Jean-Marc Dewaele,
Birkbeck College, University of London,
26 Russell Square,
London WC1B 5DT
UK
E-mail address: j.dewaele@bbk.ac.uk
Tel +44 207 631 6399

Abstract
The present study is a large-scale quantitative analysis of intra-individual variation (linked to type
of interlocutor) and inter-individual variation (linked to multilingualism, sociobiographical
variables and three personality traits) in self-reported frequency of code-switching (CS) among
2116 multilinguals. We found a significant effect of interlocutor (friends, family, colleagues and
strangers) on self-reported CS. Participants who grew up and work in a multilingual environment,
know many different languages, which they learnt from a young age, have advanced proficiency in
various languages reported more frequent CS. Sex, Extraversion and Cognitive Empathy, but not
Tolerance of Ambiguity, are linked with higher self-reported CS. We conclude that the frequency
of self-reported CS depends not just on situational, complex sociobiographical and environmental
factors, but it is also mediated by the personality of the multilingual.

Key words: individual differences, code-switching, multilingualism, personality

Introduction

Rodriguez-Fornells, Krämer, Lorenzo-Seva, Festman and Münte (2012) observe that: ‘Although a
plethora of studies have been devoted to code-switching, an important aspect that has been neglected in
psycholinguistic, linguistic, and sociolinguistic approaches to this phenomenon is the role of individual
differences in language switching’ (2012, p. 1). The present study is intended to start filling this gap by
investigating intra- and inter-speaker variations in code-switching (hereafter CS) patterns of adult
multilinguals. We define CS as ‘changes from one language to another in the course of conversation’
(Li Wei, 2007, p. 14). There are many different structural configurations of CS (Li Wei, 2013). Whilst
individuals may indeed vary the way they code-switch from one context to another, this is a topic that is
beyond the scope of the present study. We focus instead on the likelihood of individual speakers to
engage in CS in different settings and for different purposes.
The lack of systematic research into individual differences in CS could be linked to a number of
reasons. First of all, CS can be a creative discourse strategy applied by multilingual language users in

1
Preprint version of the paper that appeared in International Journal of Multilingualism, January 2014,
published online on http://dx.doi.org/10.1080/14790718.2013.878347
real-life interactions in order to achieve effective communication (Ritchie and Bhatia, 2013). This has
two critical implications. One is that most studies of CS are focused on either the discourse processes
whereby multilinguals combine elements from different languages to create mixed-code utterances or
the constraints, linguistic and cognitive, on switching from one language to another. The study of
self-reported individual differences in CS requires quantifiable and comparative data from a large
number of participants. One way to collect such data is through self-reporting questionnaires, as in the
case of Rodriguez-Fornells et al’s study. We will also exploit this data collection method in the present
study.
The other implication is that the spontaneity of CS render the identification of CS patterns, a
pre-requisite for analysing individual differences, an almost impossible task, if patterns are to be
understood as regular usage shared by a group of speakers. Multilinguals are known to be very
inventive, creating novel structural forms with elements of different languages spontaneously and
deliberately flouting linguistic and social conventions for special effects. Rodriguez-Fornells, et al.
(2012) base their study on four types of CS: switches to L1; switches to L2; contextual switches, i.e.
switches that are triggered by a particular situation, topic, or environment; and unintended switches.
The first two types are intended to measure switching behavior related to linguistic factors such as
competence and proficiency in the target and non-target languages and semantic differences across
languages. The contextual switch is designed to measure switching patterns influenced by
sociolinguistic factors. And the unintentional switch is aimed to assess cognitive control and
performance monitoring or problems in the control of activation of the non-target language.
In the present study, we focus exclusively on situational variation in the frequency of self-reported
CS and make no assumption of either the structural configurations of the mixed-language utterances or
the level of activation and cognitive control i.e. the ‘flexible and controlled manner of voluntarily
guiding goal-directed behavior’, and more specifically ‘executive processes such as voluntary response
inhibition and working memory that allow planned responses’ (Luna, 2009, p. 233). We believe that
this is a necessary first step toward establishing individual differences in self-reported and actual CS.
Details of how exactly individuals differ in combining elements of different languages to produce CS or
in the level of activation and control are important topics for future investigations.
The scarcity of research on individual differences in CS could also be due to the confusion over what
the very notion of individual differences means to different researchers. For instance, does it refer to the
same individual’s different usage in different contexts, i.e. intra-speaker variation, or is it differences
between speakers in producing the same code-switching pattern, i.e. inter-speaker variation, or both?
Moreover, there is huge diversity in the factors researchers choose to use in studying individual
differences. Language proficiency and history of language acquisition and learning are probably the
most frequently used factors. But other factors, such as personality, sex, and interpersonal relationship,
which are the core factors in psychological research on individual differences, remain under-explored
in CS research. The present study will examine both intra- and inter-speaker variation with a particular
focus on type of interlocutor, prior and current linguistic practices, age, sex, education and personality
traits such as extraversion, tolerance of ambiguity, and cognitive empathy.
The paper is structured as follows. We first review the existing research on individual variation in CS,
focusing on sociolinguistic studies. We then introduce the personality traits which we hypothesise to
have a prima facie link with self-reported CS. Our research instruments include an online questionnaire
specially designed for the present study, the short version of the Eysenck Personality Questionnaire
(EPQr) (Eysenck, Eysenck & Barrett, 1985); the adapted measure of Cognitive Empathy (Baron-Cohen
and Wheelwright, 2004); and the adapted Tolerance for Ambiguity Scale (TAS) (Herman, Stevens,
Bird, Mendenhall & Oddou, 2010). We will describe the design of our empirical study, and our
research questions and hypotheses. Subsequently, we will test four specific research hypotheses with
our empirical data. The implications of our findings are discussed in the concluding section.

2
 Existing sociolinguistic studies of individual variations in CS

Multilingual language users code-switch for a variety of reasons that sociolinguists have termed as
‘motivations’. Gumperz (1982) and Blom and Gumperz (1972) introduced the notions of situational
versus metaphorical CS. Situational CS is triggered by a change of address, topic or setting, whereas
metaphorical CS does not involve any change of these variables but are motivated by communicative
functions such as quotation, addressee specification, interjection, reiteration, message qualification, or
personalization versus objectivation; in other words, special communicative effects. To these we can
add i) cultural specificity - certain notions or concepts may only exist in one language or are simply
better expressed in a specific language; ii) experiential association – as Grosjean says: ‘Bilinguals
usually acquire and use their languages for different purposes, in different domains of life, with
different people. Different aspects of life often require different languages’ (2010, p. 29). And iii)
identity and identification - CS can be strategically employed when the speaker wishes to show
involvement, group membership, expertise, power, and also excluding someone who does not know the
language. Myers-Scotton’s (1993) investigation of CS in conversations of multilingual Africans
(mostly Kenyans living in Nairobi) considered different types of CS, its structure, and some social
motivations. She concluded that CS is a type of skilled performance, rather than a means of
overcoming linguistic difficulties. CS expressed social import and could be interpreted as identity
negotiations (p. 151). The work on social motivations for CS, however, does not generally discuss
individual variations.
Since the 1980s, there has been a group of sociolinguists who have attempted to explore the question
whether different groups of bilingual and multilingual speakers would use CS in different ways in
different contexts. Working in the Labovian variationist sociolinguistics paradigm, Poplack (1980),
found that speakers in the East Harlem Spanish-English bilingual community in New York City
adapted their language choice between Spanish and English and their CS patterns according to the
situation they found themselves to be in, e.g. home, school and neighbourhood. On the whole, the most
significant variation existed between individual speakers, especially between children who were
receiving bilingual schooling or monolingual schooling, with the former switching much more.
Gardner-Chloros (1991), using a similar approach, surveyed the use of French and Alsatian, and also
switching between the two, in Strasbourg’s department stores. Complex language choice and CS
patterns were revealed. For example, the highest rate of CS was found among the youngest group of
shoppers, who were also the most French-speaking, when they were in the most Alsatian-speaking
store, whereas the same group of speakers used little Alsatian, and little CS also, in another, more
linguistically neutral store. Gardner-Chloros explained the variation in terms of speech accommodation
and language prestige. Li Wei (1995) examined a number of CS patterns - inter-speaker, inter-sentential
inter-clausal, and content words – by three groups of Chinese-English bilinguals: students, British-born
Chinese youth and long-term immigrants from Hong Kong, all living in the North East of England. He
found that variations existed across these groups according to the bilingual experience and social
network contacts the speakers had, as well as the topic, setting and interlocutor of the interaction.
Ritchie and Bhatia (2013) noted that CS is linked to social roles and relationships between
participants. Other factors include situational factors, message-intrinsic considerations and finally
language attitudes including social dominance and security (p. 378).
Dewaele (2010) investigated the effect of type of interlocutor and conversation topic on self-reported
frequency of CS among 1453 multilinguals who filled out the BEQ (Bilingualism and Emotions
Questionnaire - Dewaele & Pavlenko, 2001-2003) and among 20 multilinguals who were interviewed
about their CS practices. Self-reported CS was found to be significantly lower when speaking about
neutral topics compared to personal or emotional topics (Dewaele, 2010, pp. 196-197). When strong
emotions need to be verbalized quickly, and the speaker realizes that it would take too much time to
express them in the weaker output language, possibly with unwanted effects, CS might seem like an
acceptable option (p. 213). The analysis of the narratives showed that CS is strongly linked to the
perceived emotionality of the languages and that CS is deployed strategically. The typical direction of

3
the CS in situations where strong emotion had to be expressed was from the foreign language to the L1.
Some participants, often of Arab or Asian origin, reported CS in the opposite direction, especially for
the expression of anger and swearing. They said that for them, CS to English (a foreign language)
allowed them to escape the social constraint that weighs on them in their home environments, where
anger cannot be displayed as openly as in English and where swearing carries strong social stigma (p.
212).
Dewaele also found that the higher frequency of CS with known rather than less known, or unknown,
interlocutors was linked to a conscious, strategic choice of the speaker. In dealing with an unknown
interlocutor the most logical option is to stick to the language in which the interaction was started.
Conversations with known interlocutors allow CS to the shared languages (p. 219). Resnik (2012)
reported similar effects of the type of interlocutor on self-reported CS in emotional interactions with
German, Chinese and Japanese foreign language users of English.
Some significant sex effects emerged in Dewaele (2010), with female participants reporting more
frequent CS with friends and on more emotional topics, while male participants reported more CS with
strangers. These sex effects seem rather unusual considering Gardner-Chloros (2009) who states that
CS is unconnected with sex, although ‘it intersects with a large number of intervening variables which
are themselves connected with gender issues’ (2009, p. 82).
Another extralinguistic variable found to be associated with the frequency of CS is the linguistic
environment in which an individual grows up or is currently living. Family and community norms may
affect the extent of CS of the individual (Lanza, 2008).
One factor that has hitherto not been examined systematically is personality. Gardner-Chloros (2008)
points out that the way in which multilinguals use CS is part of the way in which they present
themselves as speakers, i.e. ‘styling the self’. She further observes that among the many independent
variables which have been studied in relation to CS, personality variables affecting the amount and type
of CS have on the whole been omitted.
In the present study, we hypothesise that certain personality traits contribute to a higher self-reported
frequency of CS, possibly in interaction with particular social variables such as context and language
background. In the following section, we discuss the key personality traits that we intend to investigate.

Personality traits

Psychologists generally agree that personality traits are hierarchically organized with three or fivei
broad, independent dimensions at the top - and a larger number of more specific traits lower down
(Cervone & Pervin, 2013). Personality questionnaires allow researchers to establish profiles of
participants. In the present study, we focus on one dimension out of the so-called ‘Giant three’ii,
Extraversion versus Introversion, and two ‘lower-order’ traits, Cognitive Empathy and Tolerance of
Ambiguity. All these three traits have been found to be associated with second language learning and
multilingualism (Dewaele, 2012; Dewaele & Li Wei, 2012, 2013). We hypothesize that they are also
linked to CS.

Extraversion - Introversion
Eysenck and Eysenck (1985) argue that variation on the Extraversion – Introversion dimension is
linked to the amount of cortical arousal, which leads to different behavior, also different
communicative behaviour in a foreign language (Dewaele, 2012). While extraverts are under-aroused,
introverts are over-aroused. This has behavioral consequences: extraverts compensate for their
suboptimal arousal levels by tending towards activities that involve greater sensory stimulation while
introverts will tend to avoid over-arousing situations. While the typical extravert is gregarious, ‘craves
excitement, takes chances, often sticks his neck out, acts on the spur of the moment’ (Eysenck &
Eysenck 1964, p. 8), the typical introvert ‘is a quiet, retiring sort of person, introspective, (...) tends to
plan ahead, ‘looks before he leaps,’ and (...) does not like excitement’ (p. 8). The extraverts’ inclination
to take risks seems to extend to their linguistic behavior. Extravert L2 users use more stigmatised

4
variants than their more introverted peers and more likely to talk about emotion, a particularly
challenging topic in a foreign language (Dewaele 2004; Dewaele & Pavlenko, 2002). Li, Chen and
Xiao (2009) have argued that extraverts possess superior pragmatic competence and awareness, which
allow them to engage in more risky linguistic behaviour. In his overview of individual differences and
creativity, Kaufman (2011, p. 681) reports that extraversion has been linked to creativity in some
domains but not in others. The study of Peterson, Smith and Carson (2002) found a significant positive
correlation between extraversion and creativity (p. 1143). King, Walker and Broyles (1996) also found
a significant positive correlation between extraversion and verbal creativity. It is thus plausible that
extraverts are more likely to engage in creative CS.

Cognitive empathy
Empathy has been defined as ‘the ‘glue’ of the social world’ (Baron-Cohen & Wheelright, 2004, p.
193). It refers to the ability to empathise, to understanding what other people might be thinking or how
they might be feeling (p. 193). People with high levels of empathy are better at understanding the
intentions of others, are more accurate in predicting their behaviour, and are better able to recognise the
emotion of their interlocutor. Empathy is thus crucial in social interactions, ‘drawing us to help others
and stopping us from hurting others’ (p. 193).
Dewaele and Li Wei (2012) investigated the relationship between multilingualism and cognitive
empathy among 2158 mono- and multilinguals (who also contributed the data for the present study). A
significant positive correlation emerged between multilingualism (operationalised as advanced levels
of proficiency in several foreign languages and frequent use of these languages) and cognitive empathy.
They concluded that intense multilingual practice makes multilinguals more skilful in conversations as
they learn to see the world from their interlocutor’s point of view. It is possible that the ability to
empathise with a bilingual or multilingual interlocutor might be linked to increased CS, as this is a way
of highlighting the specific links between the speaker and the interlocutor. A high level of empathy
might nudge a bilingual or multilingual to converge to a higher frequency of CS to reflect the pattern of
the interlocutor.

Tolerance of Ambiguity
According to Furnham and Ribchester (1995, p. 179), ‘The person with low tolerance of ambiguity
experiences stress, reacts prematurely, and avoids ambiguous stimuli. At the other extreme of the scale,
however, a person with high tolerance of ambiguity perceived ambiguous situations/stimuli as
desirable, challenging, and interesting and neither denies nor distorts their complexity of incongruity’.
Dewaele and Li Wei (2013a), using the database on which the present study is based, found that
monolinguals and bilinguals scored significantly lower on Tolerance of Ambiguity (TA) compared to
multilinguals. Moreover, participants with higher levels of multilingualism, as measured through total
proficiency in multiple languages, also scored significantly higher on TA. A multilingual upbringing
had no effect on TA but those who have lived abroad scored significantly higher on TA. The authors
concluded that TA is determined by individuals’ social-linguistic-cultural environment and especially
by the experience of having to survive in a foreign cultural and linguistic environment. A prolonged
stay in a new environment requires a sustained and conscious effort to acquire the new sociopragmatic
rules and cultural values. CS could be considered as a way to make the interaction more interesting. We
will consider the potential link between TA and CS.

Research questions and hypotheses

The present study aims to address the following research questions:

1) Does the amount of self-reported CS vary according to the interlocutor type? We
expect most frequent CS with friends, followed by colleagues, family, and least of all with
strangers.

5
 2) Are prior and current linguistic practices linked to the amount of self-reported CS? We
expect participants who grew up in a linguistic and ethnic diverse environment to code-switch more.
We also expect participants who know more languages, who have advanced knowledge of several
languages, who grew up with two or more languages before age 3, who learned languages early in
life to code-switch more. Finally, we expect participants who work in multilingual environments to
code-switch more.
3) Are sociobiographical variables linked to the amount of self-reported CS? We expect
female participants to code-switch more with known interlocutors. Age and education level could
also be linked to self-reported CS.
4) Are personality traits linked to the amount of self-reported CS? We expect participants
who score high on Extraversion, Tolerance of Ambiguity, Cognitive Empathy to code-switch more.

Method

Participants
A total of 2116 multilinguals (1564 females, 443 males) filled out the online questionnaire on the use
of CS that has been designed specially for the present study. The mean age was 34.5 yrs (SD = 12.1). A
closer look showed that 90 participants were in their teens, 799 were in their twenties, 583 in their
thirties, 347 in the forties, 190 in their fifties, 87 in their sixties, 8 in their seventies and one participant
aged 84. Participants were generally highly educated with only 31 having a high school diploma, 650 a
Bachelor’s degree, 772 a Master’s degree, and 657 a PhD. This majority of highly educated, female
participants is quite typical in web-based language questionnaires (Wilson & Dewaele, 2010).
The participants reported 204 different nationalities, including many participants with double
nationalities. The largest group came from the USA (n = 478), followed by British (n = 299); Dutch (n
= 145); Belgians (n = 81), Germans (n = 81), Canadians (n = 76), Polish (n = 65), French (n = 58),
Spaniards (n = 42), Chinese (n = 41)iii.
English was the most frequent L1 (n = 843), followed by Dutch (n = 195), French (n = 155), Spanish
(n = 138), German (n = 124). The most frequent L2 was English (n = 924) followed by French (n =
455), and Spanish (n = 248). The pattern was different for the L3 with French coming first (n = 422),
followed by German (n = 330) and English (n = 248). The most frequent L4s were German (n = 205),
Spanish (n = 196) and French (n = 174). The most frequent L5 was Spanish (n = 101), followed by
Italian (n = 69) and French (n = 50). Mean age of acquisition of the L2 was 10.1 years (SD = 5.4), this
increased to 15 years for the L3 (SD = 6.4), 18.3 years for the L4 (SD = 7.8) and 21.7 for the L5 (SD =
8.6).
We used participants’ information on self-perceived proficiency in their various languages to
develop a global measure of multilingualism, first presented in Dewaele and Stavans (2012). The ‘total
proficiency score’ is the sum of self-perceived proficiency scores collected on 5-point Likert scales for
oral proficiency and written proficiency in up to 6 languages (including 2 L1s) (maximal possible score
10 X 6 = 60). Dewaele and Stavans (2012) argued that such a measure is potentially useful to
distinguish pentalinguals with limited knowledge of three languages from trilinguals with advanced
knowledge of 3 languages. The trilingual might know fewer languages, but knowing them better makes
that individual more strongly multilingual. In other words, we avoid the lack of clarity inherent to
labels such as ‘bilingual, trilingual, quadrilingual etc’, where every language is included, despite the
fact that knowledge in some can be very limited. In the present sample total proficiency scores vary
between 5 and 55 with a mean of 25.5 (SD = 8.0). Participants with scores that were more than 1
standard deviation below the mean were categorised as ‘Low Proficiency’ (n = 244), those with scores
that were more than 1 standard deviation about the mean were categorised as ‘High Proficiency’ (n =
278), while the remaining participants were categorised as ‘Medium Proficiency’ (n = 1510). We feel
also that it makes sense to look for a link between this global language measure and CS, rather than
proficiency measures for particular languages (L1, L2, L3...).

6
 The sample consists of 399 bilinguals, 566 trilinguals, 557 quadrilinguals, 359 pentalinguals, 143
sextalinguals, 54 septalinguals, 21 octalinguals, 9 nonalinguals, 5 participants knew 10 languages, and
1 participant reported 12 languages. A single category was created including all participants with 6 or
more languages. A majority of participants (n = 1825) reported having been exposed to a single L1
before age 3, a minority reported growing up with two L1s (n = 274) and 17 participants grew up with
three languages (the latter two groups were aggregated).
Mean score for ethnic and linguistic diversity during the participants’ childhood was rather low (M =
1.26, SD = 1.23, on a 4-point Likert scale). However, the mean score for linguistic diversity in the
participants’ workplace was higher (M = 2.59, SD = 1.23, on a 4-point Likert scale).
Instruments
The data was collected through the snowballing technique. The anonymous online CS questionnaire
was an open-access survey, advertised through several listservs, targeted emails to teachers and
students, and informal contacts asking them to forward the link to friends. The introductory paragraph
stated: ‘The aim of this research is to help us better understand the link between personality and
linguistic behavior of people knowing one or several languages’. It remained online over a four month
period and attracted responses from mono- and multilinguals across the world. We removed the data
provided by the 41 monolinguals for the present study, as they did not code-switch between languages.
Because participants left occasional questions blank, totals for specific variables can vary. Participants
also filled out a short sociobiographical questionnaire with questions about sex, age, education,
language history and present language use. The research design and questionnaires received ethical
clearance from the research institution.
Buchanan (2007) observes that the use of on-line questionnaires in psychological research is
increasing exponentially. The main advantage is of an ‘economic’ nature: large amounts of data can be
collected automatically at a fraction of the cost and time of ‘pen and paper’ equivalents (Buchanan,
2007). This allows researchers to reach larger and more diverse samples from all over the world.
Internet samples have been found to be more diverse in terms of sex, age, race, socio-economic status
and geographical location than the pen-and-paper samples (Gosling et al., 2004). Scores of personality
dimensions and patterns of socializing are comparable in internet samples and pen-and-paper samples
(Gosling et al., 2004). Moreover, online versions of traditional questionnaires do not compromise the
psychometric properties of such measures (Denissen, Neumann & van Zalk, 2010).
Wilson and Dewaele (2010) have pointed out that in multilingualism research participants must meet
specific linguistic criteria, have high levels of metalinguistic and metacognitive awareness, and must be
able and willing to engage with the questions on language preferences and use (2010, p. 108). We feel
that multilinguals are perfectly aware of the amount of CS they use in specific interactions, and they
have no reason to lie about the frequency with which they use CS, as they perceive the phenomenon as
something positive (Dewaele & Li Wei, 2013b)iv. The reliability of the data might in fact be stronger in
linguistic internet-based research, as anonymous volunteers would not benefit in any way from
falsifying answers (p. 108). Another advantage of internet-based questionnaires is that they reduce
social desirability (the tendency of participants to answer questions in a manner that they imagine will
be viewed favourably by the researcher), which leads to increased levels of honesty (and therefore
higher validity in the case of self-report) (Joinson, Paine, Buchanan & Reips, 2008). Finally, a sample
of more than 2000 multilinguals from all over the world ensures strong ecological validity, as the
effects of local social, political and historical factors linked to particular languages or linguistic
practices are evened out (Wilson & Dewaele, 2010).
In addition to the background questionnaire, participants filled out the items for the personality
dimension Extraversion extracted from the EPQr (Eysenck, Eysenck & Barrett, 1985), also based on
self-reported behaviour. Participants filling out the EPQr are invited to tick either ‘yes’ or ‘no’ for 12
items for each dimension. One item for Extraversion is for example: ‘Are you a talkative person?’.
Barrett, Petrides, Eysenck and Eysenck (1998) demonstrated the factorial similarity of the
dimensions in data collected from 34 countries, which suggests that the Eysenck factors are strongly

7
replicable across the world. The EPQr is considered robust (Barrett, 1999). Mean scores for
extraversion was as follows: N = 1888, M = 7.6, SD = 3.4. Internal consistency, as measured by
Cronbach alpha coefficient, was high (0.84). Participants with scores that were more than 1 standard
deviation below the mean were categorised as ‘Introverts’. Those with scores ranging from 1 standard
deviation below the mean to 1 standard deviation above the mean were categorised as ‘Ambiverts’.
Finally, those with scores that were more than 1 standard deviation above the mean were categorised as
‘Extraverts’. As a result we have 399 Introverts, 1008 Ambiverts and 481 Extraverts.
Participants also filled out the Tolerance for Ambiguity Scale (TAS), a 12–item questionnaire with
5-point Likert scales, contextualized to cross-cultural contexts (Herman et al., 2010, p. 60). According
to the authors, it ‘can be used in cross-cultural research and practice to assess individual TA’ (p. 62). It
consists of four distinct dimensions: (1) valuing diverse others; (2) change; (3) challenging
perspectives; (4) unfamiliarity (p. 62). We used 11 items of out the original 12-item TAS scale and
made some minor stylistic adaptationsv. The items had to be rated on a 5-point Likert scale. A Cronbach
alpha analysis revealed modest but sufficient internal consistency reliability: 0.64. Mean score for TA
is 28.9, SD = 5.6, with a range between 4 and 44. Participants were also categorised in three groups
(low TA, medium TA, high TA), following the procedure set out for extraversion. As a result 277
participants were labelled ‘low TA’, 1337 ‘medium TA’ and 281 ‘high TA’.
Finally, participants filled out an adapted version of Baron-Cohen and Wheelwright’s (2004)
questionnaire to measure Cognitive Empathy (CE). We selected five items with the highest factor
loadings (Lawrence et al., 2004, p. 915) on the dimension of CE, which ‘measures the appreciation of
affective states’ (p. 918). One such item was ‘I am good at predicting how someone will feel’. The
items had to be rated on a 5-point Likert scale. This subscale was internally consistent (Cronbach's
Alpha = 0.84)vi. Mean score of CE is 18.5, SD = 3.5, with a range between 5 and 25. Three groups were
created (low CE, medium CE, high CE), following the procedure set out for Extraversion and TA. As a
result 230 participants were labelled ‘low CE’, 1397 were in the middle group (‘medium CE’) and 267
were labelled ‘high CE’.
Data about CS practices of the participants were elicited through the following general closed
question: ‘Do you switch between languages within a conversation with certain people?’ It then
specified four situations: ‘when speaking with some friends’, ‘with some family members’, ‘with
strangers’ and ‘with some colleagues or clients at work/school’vii. Participants were asked to choose a
response on a 5-point Likert scale, i.e. never, rarely, sometimes, frequently, always. We are therefore
more concerned with the frequency of CS with different interlocutors in different contexts rather than
with the structural patterns of switching.
A series of one-sample Kolmogorov-Smirnov tests showed that the values for self-reported
frequency of CS with the four types of interlocutors are not normally distributed
(Kolmogorov-Smirnov Z-values vary between 8.2 and 10.4, all significant at p < .0001). More than a
third of participants reported never using CS with family members nor strangers. The proportion of
non-CS users dropped to 22% in interactions with colleagues and 12% in interactions with friends
where the frequency category ‘sometimes’ is the largest (a third of participants). The complete
distribution is presented in Figure 1.
Figure 1. Distribution of frequency of self-reported CS with four types of interlocutors

8
 45

40

35
Percentage of participants

30
Friends
25
Family
Strangers
20
Colleagues
15

10

0
Never Rarely Sometimes Frequently Always
Frequency of CS

As a consequence, we have used non-parametric statistical techniquesviii.

Results

A Friedman’s ANOVA test for related samples revealed that the effect of type of interlocutor is
highly significant on the amount of self-reported CS (N = 1997, Chi2 = 1047.7, df = 3, p = .0001) (Table
1). This result confirms that multilinguals report CS most with their friends and least with strangers.
The standard deviation is also the smallest for these two situations, especially for strangers. There is
much more dispersion around the mean for self-reported CS with family members. This suggests that
there is more variation in CS practices within families, compared to the groups of friends, colleagues or
strangers. This result is somewhat surprising as family members are likely to share similar language
backgrounds.
Table 1. Mean scores on self-reported frequency of CS

Code-switching with... M SD
Friends 2.16 1.15
Colleagues 1.70 1.19
Family 1.46 1.41
Strangers 1.09 1.00

The next cluster of variables linked to participants’ linguistic history and current practice show the
extent to which it is linked to self-reported CS. The Kruskal-Wallis tests showed that participants who
grew up in an ethnically and linguistically diverse environment report more frequent CS in all situations
except with strangers (Table 2 and Figure 2).

Table 2. The effect of linguistic history and current practice on self-reported frequency of CS
(Kruskal-Wallis Chi2 values)
Friends Family Strangers Colleagues
Linguistic diversity during childhood 10.7* 29.5*** 4.3 8.3*
Number of languages known 33.9*** 51.6*** 38.8*** 35.6***

9
 Total Proficiency 93.4*** 80.9*** 51.1*** 79.2***
Multilingualism in work environment 52.8*** 3.2 11.4** 97.2***
p < .05, ** p < .001, *** p < .0001

Figure 2. Ethnic and linguistic diversity during childhood and self-reported frequency of CS
2,5

2,3

2,1

1,9
Frequency of CS

1,7 Friends
Family
1,5
Strangers
1,3 Colleagues

1,1

0,9

0,7

0,5
not at all not especially so-so quite diverse very diverse
Ethnic & linguistic diversity during childhood

The number of languages known by participants also had a highly significant effect on self-reported
frequency of CS in all situations (Table 2). Participants knowing more languages reported more
frequent CS with all interlocutors (Figure 3).
Figure 3. Number of languages known and self-reported frequency of CS

2,5

2,3

2,1

1,9
Frequency of CS

1,7 Friends
Family
1,5
Strangers
1,3 Colleagues

1,1

0,9

0,7

0,5
two three four five six+
Number of languages known

Total Proficiency was found to have a significant effect on self-reported frequency of CS in all
situations (Table 2). The High Total Proficiency group reported more frequent CS with all
interlocutors (Figure 4).

10
Figure 4. Total Proficiency and self-reported frequency of CS
3

2,5
Frequency of CS

2 Friends
Family
St rangers
1,5 Colleagues

0,5
Low Medium High
Total Profici e n cy

A Mann-Whitney test showed significant differences between early and later bi- and multilinguals in
interactions with friends and family, with a marginal difference in interactions with strangers (Table 3).
As expected early bi- and multilinguals reported using more CS than later bi- and multilinguals (Figure
5).
Table 3. The effect of early bi- or multilingualism on self-reported frequency of CS
(Mann-Whitney)
Interlocutor Z
Friends -4.3***
Family -11.9***
Strangers -1.8
Colleagues -0.7
*** p < .0001

11
Figure 5. Monolingual vs Bi- and Multilingual upbringing and self-reported frequency of CS
3

2,5

2
Frequency of CS

M onolingual
1,5
Bi-M ultilingual

0,5

0
Friends Family Strangers Colleagues
Interlocutor

A Spearman Rank correlation analysis revealed significant negative values between Age of Onset of
Acquisition (AoA) in the L2 and L3 and self-reported CS with family, friends and strangers (but not
with colleagues). This suggests that multilinguals who started learning their L2 and L3 earlier in life
reported using more CS in specific situations. The AoA of the L4 and L5 (both over the age of 18) was
unrelated to self-reported CS frequency (see Table 4).
Table 4. The relationship between Age of onset of Acquisition and CS (Spearman Rho)
Language Measure Friends Family Strangers Colleagues
L2 Rho -.124*** -.211*** -.052* -.033
N 2028 2012 2011 2009
L3 Rho -.060* -.165*** -.044 -.022
N 1633 1622 1623 1618
L4 Rho -.057 -.098** -.044 -.047
N 1062 1061 1058 1059
L5 Rho -.039 -.012 -.007 -.020
N 514 512 511 511
 p < .05, ** p < .001, *** p < .0001

The degree of multilingualism in the work environment of participants also has a significant effect on
self-reported frequency of CS in all situations except in the family sphere (Table 2). Participants in
more multilingual workplaces report more frequent CS with colleagues and clients (Figure 6).

12
Figure 6. Multilingual workplace and self-reported frequency of CS
2,5

2,3

2,1

1,9
Frequency of CS

1,7 Friends
Family
1,5
Strangers
1,3 Colleagues

1,1

0,9

0,7

0,5
not at all minority about half > half almost all

The next research question deals with the effect of sociobiographical variables such as age, sex and
education level. A Spearman correlation analysis showed that age is positively linked to self-reported
frequency of CS in interactions with family and strangers, and negatively linked to frequency of CS in
interactions with friends and colleagues. In other words, older participants report using CS more with
family members and strangers and less with friends and colleagues (Table 5).
Table 5. The relationship between age and self-reported frequency of CS (Spearman Rho)
Interlocutor N Spearman Rho
Friends 2022 -.071**
Family 2006 .133***
Strangers 2006 .081***
Colleagues 2003 -.043
* p < .05, ** p < .001, *** p < .0001

The Kruskal-Wallis analyses reveal significant differences in self-reported CS between the various
education levels for two types of interlocutors (Table 6). Participants with higher levels of education
reported more CS with colleagues and family members.
A Mann-Whitney test shows significant differences between male and female participants in
interactions with friends, family and colleagues (Table 6). Female participants report more CS in these
interactions than the males.
Table 6. The effect of education level and sex on self-reported frequency of CS (Kruskal-Wallis
Chi2 and Mann-Whitney Z)
Interlocutor A-level BA MA PhD Chi2 Female Male Z
Friends 2.0 2.2 2.2 2.2 1.1 2.2 1.9 -4.4**
Family 1.5 1.4 1.4 1.6 8.6* 1.5 1.2 -4.6**
Strangers 1.1 1.1 1.1 1.1 0.1 1.1 1.1 -0.3
Colleagues 1.2 1.7 1.6 1.8 10.2* 1.7 1.6 -2.5*
* p < .05, ** p < .001

The final research question focuses on the effect of some personality traits on self-reported CS.

13
 The Kruskal-Wallis tests did reveal a highly significant effect of Extraversion on self-reported CS in
four situations (Table 7).

Table 7 The effect of personality variables on self-reported frequency of CS (Kruskal-Wallis Chi2

values)
Trait Friends Family Strangers Colleagues
Extraversion 32.2*** 13.7** 20.0** 21.6**
Tolerance of Ambiguity 2.53 2.0 2.7 0.8
Cognitive Empathy 14.0** 5.5 0.3 4.9
* p < .05, ** p < .001, *** p < .0001

A look at the means shows that more extraverted participants report more CS with the various
interlocutors (Figure 7).
Figure 7. Extraversion and self-reported frequency of CS
2,5

2,3

2,1

1,9
Frequency of CS

1,7 Friends
Family
1,5
Strangers
1,3 Colleagues

1,1

0,9

0,7

0,5
Introverts Ambiverts Extraverts

Tolerance of Ambiguity has no effect on self-reported frequency of CS with the various interlocutors
(Table 7). Cognitive Empathy has a significant effect on self-reported frequency of CS with friends and
a marginal effect on CS with family members and colleagues or clients (Table 7). A look at the means
shows that the High CE group reports more frequent CS with friends (Figure 8). The same (but only
marginally significant) pattern exists for self-reported frequency of CS with family members and
colleagues or clients (Figure 8).

14
 Figure 8. Cognitive Empathy and self-reported frequency of CS
2,6

2,4

2,2
Frequency of CS

2
Friends
Family
1,8
Strangers
Colleagues
1,6

1,4

1,2

1
LowCE M ediumCE HighCE

Discussion

The analyses of the data show that our first hypothesis was confirmed: self-reported CS was most
frequent in interactions with friends, followed by interactions with colleagues and family members, and
it was least frequent in interactions with strangers.
Our second hypothesis was also confirmed: participants’ linguistic history - including the
environment in which they grew up, current linguistic practices and linguistic diversity in their current
workplace all have a significant effect on self-reported CS.
Our third hypothesis was partly confirmed as participants’ age was correlated with frequency of
self-reported CS (though in different directions). Sex was significantly linked to self-reported CS:
female participants reported more CS in interactions with friends, family and colleagues. The effect of
education was significant in interactions with family members and colleagues where highly educated
participants reported more CS.
Our final hypothesis was partly confirmed: the degree of Extraversion and Cognitive Empathy was
significantly linked to higher levels of self-reported CS, though only with friends for CE. However,
Tolerance of Ambiguity had no effect on self-reported CS.
These findings show that multilinguals adapt the frequency of CS to the type of interlocutor they are
interacting with. Just as in Dewaele (2010) we found that self-reported CS is less frequent with
unknown interlocutors, probably because speakers need to know which languages they have in
common with their interlocutors before starting CS. Unsurprisingly, when the interlocutor is known to
the speaker, there is a higher likelihood of CS. The highest amount of CS was reported with friends,
followed by colleagues. The amount of CS in interactions with family members is slightly lower. One
possible reason for this is that the groups of friends and colleagues are probably more heterogeneous in
linguistic terms, whereas as family members may form a more homogeneous group, with clear
language preferences. Rodriguez-Fornells (personal communication) suggested that the fact that more
CS is reported with friends than with family members might be a simple effect of sample bias, where
this population comes from, and how their familial environment was. This result might simply reflect
the fact that the participants are now not too much exposed to their families anymore. It is equally

15
possible that CS was banned within the family home of some participants in order to maintain the
minority language. However, a closer investigation confirmed that the early bi- and multilinguals
reported significantly more CS with their family members and friends compared to participants who
became bi- and multilingual after the age of 3.
The fact that participants who had grown up in highly multilingual and ethnically diverse
environments reported using more CS with known interlocutors shows that linguistic practices in
childhood continue to resonate through a multilingual’s life. Less surprising is the finding of a positive
link between multilingualism in the work environment and self-reported CS (except with family).
Indeed, one can assume that our participants conformed to the unwritten rules of their community of
practices, which involve CS. However, they did not follow these rules when interacting with family
members – which probably constitutes a different community of practice.
The relationship between age and self-reported CS was unexpected because it was positive with
some interlocutors, and negative with others. There is no obvious reason why older participants would
use more CS more with family members and strangers but on the other hand, use significantly less CS
with friends and colleagues. These results differ from those reported by Dewaele (2010, p. 199),
namely an absence of correlation between age and self-reported CS in interactions with friends and
strangers, and a significant positive correlation between self-reported CS when talking to colleagues or
speaking in public. The difference cannot be linked to a different age profile of the sample, as the mean
age in Dewaele (2010) was 34 years, identical to the one in the present study. Had the link between age
and CS with different interlocutors been consistent, it could have been associated with changes in
cognitive control, which reaches a peak in adulthood and remains stable before declining in old age
(Luna, 2009)ix.
The significant sex effect in self-reported use of CS in interactions with friends, family and
colleagues partly confirms previous research. Dewaele (2010, p. 198) found that the female
participants reported using CS significantly more with friends, but not with colleagues. Male
participants were also found to report more CS in interactions with strangers, a pattern that is absent in
the present study. It is difficult to speculate on the reason why our female multilinguals report more CS
with friends, family and colleagues. Are they conforming to rules of their speech communities? Does it
reflect a greater involvement of our female participants in conversations (cf. Tannen, 1993)?
Education level was found to have significant effect on self-reported CS with two types of
interlocutor: colleagues and family members. It is again hard to speculate on a possible cause. That
highly educated participants report using more CS with colleagues could be linked to the type of work
environment they function in. Indeed, a Pearson Chi2 analysis showed that highly educated participants
were significantly more likely to work in highly multilingual workplaces (Pearson Chi2 = 44.7, df = 12,
p < 0.0001). It is equally possible that the families of the highly educated participants were more
multilingual, but we have no information on the degree of multilingualism of the participants’ families.
One of the most striking findings of the present study is the link between two personality traits and
self-reported CS. The positive link between Extraversion and self-reported CS can be understood if we
see CS as a form of impulsive linguistic risk-taking, or verbal acrobatics, giving the speaker a certain
thrill. More introverted speakers report engaging less frequently in this verbal behaviour. The more
gregarious multilingual extraverts might be more inclined to use CS once they discover they share
another language with an interlocutor. By doing so, they know that they underline the common ground
with the interlocutor and converge towards them. The relationship between extraversion and CS could
also be linked to the higher levels of (verbal) creativity of extraverts (King et al., 1996; Peterson et al.,
2002). Metaphorically one could compare a monolingual interaction to a monochrome exchange and a
multilingual interaction to something more colourful. Creative multilinguals can insert threads in
different colours in the exchange and hence insert an element of novelty, uniqueness and surprise.
Tolerance of Ambiguity was unrelated to frequency of CS. This is surprising because we had
assumed that CS is a way of introducing and sharing ambiguity in the interaction. Indeed, the pupils
described in Li Wei and Chao-Jung Wu (2009) deliberately used CS to create ambiguity and undermine
the monolingual teacher’s authority. While CS can serve this function, the opposite might be true in

16
general, namely that CS allows multilinguals to avoid ambiguity. By switching to another shared
language, communication is more efficient and to the point, it allows the multilingual to find ‘le mot
juste’ (Grosjean, 2010, p. 53). One could even argue that multilinguals that share a language and can
fluently CS may find this speech style less ambiguous than monolingual-mode. The possibility of
substituting a word in the other language if necessary, or use a colloquial saying with the interlocutor
may introduce greater security into the interaction. Whilst our study does not aim to investigate the
cognitive control in CS, one may speculate, based on existing evidence that links language switching
with executive control (e.g. Rodriguez-Fornells, De Diego Balaguer & Münte, 2006; Soveri et al.,
2011), that CS requires more selective attention; therefore, it would be associated with precision rather
than ambiguity.
The positive link between Cognitive Empathy and self-reported CS in interactions with friends
suggests that multilinguals might resort to CS if they feel it could help their friend. In other words, their
ability to see the friend’s perspective may push them to CS if they feel it could help the flow of the
interaction. Interestingly, that relationship faded in interactions with other interlocutors.
Some of these personality traits might share more or less variance with cognitive control processes
and therefore, some of these processes could be mediating the association between personality traits
and CS (reduced inhibition, poorer vigilance, uncertainty...) (Matthews, 2009; Mushtaq et al., 2011). A
similar explanation might hold for the increased amount of CS with friends, as this is normally a more
informal environment and cognitive control and performance monitoring might be reduced.
Since CS is a defining form of bi- and multilingual behaviour (Grosjean, 2010; Li Wei, 2007, 2013),
it comes as no surprise that the more multilingual a person is, the greater the use of CS. It also shows
that CS is linked to high levels of proficiency in different languages. This confirms the finding reported
in Poplack (1980), Ritchie and Bhatia (2013), and Dewaele (2010) that a significant positive
relationship exists between levels of self-perceived competence in the L2 and L3 and self-reported
frequency of CS. CS is therefore ‘not an indication of a deficit in the LX but on the contrary a
characteristic of participants who feel proficient in their LXs’ (2010, p. 201).
To conclude, it is worth discussing the strengths and weaknesses of our research design. We relied
on participants’ self-reports concerning CS, which is not the same as actual CS by participants. We
have argued earlier that multilinguals are capable of making generalisations about their CS behaviour
in specific situations, and that our data have sufficient validity to investigate inter- and intra-individual
variation. We feel that a research design that would rely on frequencies of actual CS would be dogged
by a different and even more serious set of problems: firstly, participants would have to be wired up and
masses of data would have to be recorded in a variety of settings (this would be ethical headache and a
practical one: how to find researchers mastering all the languages to code the data?). While this is not
impossible, the sheer amount of imposition on the participants and the amount of data to be analysed
would inevitably limit the sample size to not more than a handful. Such a small sample would be
insufficient for any statistical analysis into individual differences, which was the aim of the present
study.
We are aware that some bi- or multilinguals with a negative perception of CS could have
underestimated the frequency of their CS. There are three points to counter this criticism: 1) we
reported that our participants had a generally positive attitude towards CS (Dewaele & Li Wei, 2013b);
2) the size of our sample is large enough to iron out tainted feedback from some individuals; 3) the
anonymous nature of the questionnaire meant that participants could no gain anything by choosing
what they perceived to be the socially desirable option.
The main advantage of the data collection via an on-line questionnaire is that it allowed us to tap into
a large and culturally diverse pool of participants, which increases the ecological validity of the
resulting database (Wilson & Dewaele, 2010). Our aim was to find general relationships between
independent and self-reported dependent variables. The heterogeneous nature of our sample in terms of
language profiles is a strength but it could also been seen as a limitation. The patterns that emerged
form our data could only be described as ‘generic’. We are fully aware that in some regions of the
world, the particular sociolinguistic and sociohistorical context determines language use and CS, and

17
the local attitudes towards these languages and CS, which could result in subtly different patterns of
relationships.
Another limitation is the absence of information on whether the participants had access to CS with
friends, family and co-workers. For some participants, the opportunity to CS might only exist within
the family, or within the workplace. If a participant reports never to CS with colleagues, we cannot
know whether this is out of choice or simply a necessity (because the colleagues do not share all the
languages of the speaker).
A final limitation is that we did not differentiate between self-reported contextual switching and
unintended language switching in our questionnaire. We were therefore unable to link our findings to
current studies on cognitive control and bilingualism (Soveri et al., 2011). Festman (2012), for example,
reported that among German L1-Russian L2 late multilinguals, the non-switchers exhibited stronger
language control abilities than the habitual switchers (p. 11). Finally, it has been pointed out that
frequent CS could lead to higher levels of cognitive control in bilinguals (Yim & Bialystok, 2012, p.
882). This is a fascinating perspective because it suggests that CS itself could be a source of individual
differences.
Future research could focus on the relationship between same independent variables and CS in
recordings of actual conversations. A more qualitative approach with a smaller number of participants
could also allow an exploration of possible causes behind the patterns we uncovered. It would be
particularly useful to select participants from a wide range of educational backgrounds.

Conclusion

The present study is the first systematic investigation of inter- and intra-individual variation in
self-reported CS of a very large sample of adult multilinguals from across the world. We are aware of
the need to be prudent in interpreting the present findings as we used a non-probability sampling
process, and our sample contains different types of multilinguals. The statistical analyses showed that a
range of both inter- and intra-individual variables are linked to the amount of self-reported CS. The
intra-individual variable, namely the type of interlocutor, turned out to have a strong effect on the
amount of self-reported CS: multilinguals report significantly more CS in interactions with friends,
followed by interactions with colleagues and family members, and report using significantly less CS in
interactions with strangers. This confirms earlier research findings (Dewaele, 2010; Resnik, 2012).
Interindividual variation in self-reported CS is also considerable. The environment in which
individuals grow up and live is also linked to their self-reported frequency of CS. Interestingly, CS is
clearly a phenomenon linked to higher levels of multilingualism and early bi- and multilingualism (low
age of onset of acquisition, many languages known, high levels of total proficiency in the languages),
again confirming earlier research (Dewaele, 2010; Poplack, 1980). Some sociobiographical variables
were also found to be linked with self-reported frequency of CS: female participants reported using
significantly more CS in interactions with known interlocutors. The effects of age and level of
education were less clear. Finally, the present study is the first to uncover a significant relationship
between two personality traits, Extraversion and Cognitive Empathy, and self-reported frequency of CS.
Extraverts and participants with high levels of Cognitive Empathy report to engage more frequently in
CS.
To sum up, the frequency of self-reported CS depends not just on situational, complex
sociobiographical and environmental factors, but it is also mediated by the personality of the
multilingual.

Acknowledgment
We would like to thank our friends and colleagues who forwarded the call for participation in this
study, as well as all those who participated. Many thanks also to the anonymous reviewers for their
excellent comments.

18
References

Baron-Cohen, S., Wheelwright, S. (2004). The Empathy Quotient: An investigation of adults with
Asperger syndrome or high functioning autism, and normal sex differences. Journal of Autism and
Developmental Disorders, 34, 163-175.
Barrett, P.T. (1999. Rejoinder to: The Eysenckian personality structure: A ‘Giant Three’ or ‘Big Five’
model in Hong Kong? Personality and Individual Differences, 26, 175-186.
Barrett, P.T., Petrides, K. V., Eysenck, S., Eysenck, H.J. (1998). Eysenck Personality Questionnaire: an
examination of the factorial similarity of P, E, N, and L across 34 countries. Personality and
Individual Differences, 25, 805-819.
Blom, J.P., Gumperz. J. J. (1972). Social meaning in linguistic structure: code-switching in Norway. In
J.J. Gumperz & D. Hymes, (Eds.), Directions in Sociolinguistics (pp. 407-434). New York: Holt,
Rinehart & Winston.
Buchanan, T. (2007). Personality testing on the internet: What we know, and what we do not. In A.N.
Joinson, K. McKenna, T. Postmes, & U.-D. Reips (Eds.), Oxford Handbook of Internet Psychology
(pp. 447-459). Oxford: Oxford University Press.
Cervone, D., Pervin, L. A. (2013). Personality: Theory and Research. Hoboken, NJ: John Wiley &
Sons (12th edition).
Denissen, J.J.A., Neumann, L., van Zalk, M. (2010). How the internet is changing the implementation
of traditional research methods, people’s daily lives, and the way in which developmental scientists
conduct research. International Journal of Behavioral Development, 34, 564-575.
Dewaele, J.-M. (2004). Individual differences in the use of colloquial vocabulary: the effects of
sociobiographical and psychological factors. In P. Bogaards & B. Laufer (Eds.), Learning
Vocabulary in a Second Language: Selection, Acquisition and Testing (pp. 127-153). Amsterdam:
John Benjamins.
Dewaele, J.-M. (2010). Emotions in Multiple Languages. Basingstoke: Palgrave Macmillan.
Dewaele, J.-M. (2012). Personality. Personality traits as independent and dependent variables. In
Mercer, S. Ryan, S., & Williams M. (Eds.), Psychology for Language Learning: Insights from
Research, Theory and Practice (pp. 42-58). Basingstoke: Palgrave Macmillan.
Dewaele, J.-M., & Li Wei, (2012). Multilingualism, empathy and multicompetence. International
Journal of Multilingualism, 9, 352-366.
Dewaele, J.-M., & Li Wei, (2013a). Is multilingualism linked to a higher tolerance of ambiguity?
Bilingualism: Language and Cognition, 16, 231-240.
Dewaele, J.-M., & Li Wei, (2013b). Attitudes towards code-switching among adult mono- and
multilingual language users. Journal of Multilingual and Multicultural Development.
Dewaele, J.-M., Pavlenko, A. (2001-2003). Web questionnaire ‘Bilingualism and Emotions’. London:
University of London.
Dewaele, J.-M., Pavlenko, A. (2002). Emotion vocabulary in interlanguage. Language Learning, 52,
265-324.
Dewaele, J.-M., & Stavans, A. (2012). The effect of immigration, acculturation and multicompetence
on personality profiles of Israeli multilinguals. The International Journal of Bilingualism. DOI:
10.1177/1367006912439941
Eysenck, H.J., & Eysenck, S. (1964). The Eysenck Personality Questionnaire. London: Hodder and
Stoughton.
Eysenck, H.J., & Eysenck, M. W. (1985). Personality and individual differences: A natural science
approach. New York: Plenum.
Eysenck, S., Eysenck, H.J., & Barrett, P. (1985). A revised version of the Psychoticism scale.
Personality and Individual Differences, 6, 21-29.
Festman, J. (2012). Language control abilities of late bilinguals. Bilingualism: Language and
Cognition, 15, 580-593.

19
Furnham, A., & Ribchester, T. (1995). Tolerance of ambiguity: A review of the concept, its
measurement and applications. Current Psychology, 14, 179–199.
Gardner-Chloros, P. (1991). Language selection and switching in Strasbourg. Oxford: Claredon Press.
Gardner-Chloros, P. (2008). Intra-individual variation: The cinderella of code-switching. Paper
presented at the 17th Sociolinguistics Symposium, Amsterdam.
Gardner-Chloros, P. (2009). Code-switching. Cambridge: Cambridge University Press.
Gosling S.D., Srivastava, S., Pand, O., & John, O.P. (2004). Should we trust web-based studies? A
comparative analysis of six preconceptions about internet questionnaires. American Psychologist,
59, 93–104.
Grosjean, F. (2010). Bilingual. Life and Reality. Cambridge, Mass. & London: Harvard University
Press.
Gumperz. J.J. (1982). Discourse Strategies. Cambridge: Cambridge University Press.
Herman, J. L., Stevens, M. J., Bird, A., Mendenhall, M., & Oddou, G. (2010). The Tolerance for
Ambiguity Scale: Towards a more refined measure for international management research.
International Journal of Intercultural Relations, 34, 58–65.
Joinson, A.N., Paine, C., Buchanan, T., & Reips, U-D. (2008). Measuring self-disclosure online:
Blurring and non-response to sensitive items in web-based surveys. Computers in Human Behavior,
24, 2158-2171.
Kaufman, J. C. (2011). Individual differences in creativity. In T. Chamorro-Premuzic, S. von Stumm &
A. Furnham (Eds.), The Wiley-Blackwell Handbook of Individual Differences (pp. 679-698).
Oxford: Wiley-Blackwell.
King, L., Walker, L., & Broyles, S. (1996). Creativity and the Five-Factor Model. Journal of Research
in Personality, 30, 189–203
Lanza, E. (2008). Multilingualism and the family. In Li Wei & P. Auer (Eds.), Handbook of
Multilingualism and Multilingual Communication (pp. 45-68). Berlin: Mouton de Gruyter.
Lawrence, E. J., Shaw, P., Baker, D., Baron-Cohen, S., & David, A.S. (2004). Measuring empathy:
reliability and validity of the Empathy Quotient. Psychological Medicine, 34, 911–924.
Li, M., Chen, X., and Xiao, Y. (2009). The relationship between Chinese English majors' personality
and their pragmatic and grammatical competence/awareness. Foreign Language Teaching and
Research, 41, 119-124.
Li, Wei, (1995). Variations in patterns of language choice and code-switching by three groups of
Chinese/English speakers in Newcastle upon Tyne. Multilingua, 14, 297-323.
Li, Wei, (2007). Dimensions of bilingualism. In Li Wei (Ed.), The Bilingualism Reader (pp. 3-22). 2nd
ed., London: Routledge.
Li, Wei, (2013). Code-switching. In R. Bayley, R. Cameron & C. Lucas (Eds.), The Oxford Handbook
of Sociolinguistics (pp. 360-378). New York: Oxford University Press.
Li, Wei, & Wu, C.-J. (2009). Polite Chinese children revisited: Creativity and the use of codeswitching
in the Chinese complementary school classroom. International Journal of Bilingual Education and
Bilingualism, 12, 193-212.
Luna, B. (2009). Developmental changes in cognitive control through adolescence. Advances in Child
Development and Behavior, 37, 233-278.
Matthews, G. (2009). Personality and performance: cognitive processes and models. In P.J. Corr & G.
Matthews (Eds.), The Cambridge Handbook of Personality Psychology, (pp. 400-426). Cambridge:
Cambridge University Press.
Mushtaq, F., Bland, A. R., & Schaefer, A. (2011). Uncertainty and Cognitive Control. Frontiers in
Psychology, i, 249.
Myers-Scotton, C. (1993). Social Motivations for Codeswitching: Evidence from Africa. Oxford:
Clarendon Press.
Peterson, J. B., Smith, K.W., & Carson, S. (2002). Openness and extraversion are associated with
reduced latent inhibition: replication and commentary. Personality and Individual Differences, 33,
1137–1147.

20
Poplack, S. (1980). Sometimes I'll start a sentence in Spanish y termino en español: toward a typology
of code-switching. Linguistics, 18, 581-618.
Resnik, P.A. (2012). Differences in feeling – feeling the difference: Language, gender and emotion in
bilingualism. Unpublished PhD dissertation, Karl-Franzens-Universität Graz, Austria.
Ritchie, W.C., & Bhatia, T.K. (2013). Social and psychological factors in language mixing. In T.K.
Bhatia & W.C. Ritchie (Eds.), The Handbook of Bilingualism (pp. 375–390). Chichester, UK: John
Wiley & Sons.
Rodriguez-Fornells, A., De Diego Balaguer, R., & Münte, T.F. (2006). Executive control in bilingual
language processing. Language Learning, 56, 133-190.
Rodriguez-Fornells, A., Krämer, U. M., Lorenzo-Seva, U., Festman, J., & Münte, T.F. (2012).
Self-assessment of individual differences in language switching. Frontiers in Psychology, 2, 388.
doi: 10.3389/fpsyg.2011.00388
Soveri, A., Rodriguez-Fornells, A., & Laine, M. (2011). Is there a relationship between language
switching and executive functions in bilingualism? Introducing a within group analysis approach.
Frontiers in Psychology, 2, 183: 10.3389/fpsyg.2011.00183
Tannen, D. (1993). Gender and Conversational Interaction. Oxford: Oxford University Press.
Yim, O., & Bialystok, E. (2012). Degree of conversational code-switching enhances verbal task
switching in Cantonese–English bilinguals. Bilingualism: Language and Cognition, 15, 873–883.
Wilson, R., & Dewaele, J.-M. (2010). The use of web questionnaires in second language
acquisition and bilingualism research. Second Language Research, 26, 103–123.

i
Extraversion, neuroticism, openness, conscientiousness, and agreeableness.
ii
Extraversion, neuroticism and psychoticism.
iii
We do not report the smaller groups.
iv
Dewaele and Li Wei (2013b) found that the sample of participants on which the present study is
based had a mean score of 3.85 on a 5-point scale measuring positive attitudes towards CS.
v
For a more detailed discussion of this instrument, see Dewaele and Li Wei, 2013a.
vi
For a more detailed discussion of this instrument, see Dewaele and Li Wei, 2012.
vii
We are aware that this is an abstraction, as some may use more CS with specific individuals, but
we feel that the categories reflect typical behaviour within particular language domains (Grosjean,
2010). Moreover, personality questionnaires ask even more decontextualised questions: the item
‘are you a talkative person?’ does not specify the presence of particular (types of) interlocutors or
situations. Personality psychologists do not see this as posing a problem.
viii
Unfortunately there is no non-parametric equivalent of multiple regression analysis which
would have allowed us to measure the relative impact of all independent variables together.
ix
Our sample had too few adolescents and older people (only 8 aged over 70) to carry out a
comparison. We also suspect that self-reported data lack the degree of accuracy to investigate this
phenomenon.

21