Abstract
Aims and objectives:
This study examines intra-speaker variation in code-switching (CS) by comparing the frequency and pattern of CS across two different interaction modes (interview speech and online posting) based on the same individuals. This study also explores the reasons why these speakers exploit their linguistic repertoire variably in different modes.
Methodology:
I interviewed 40 Chinese–English bilinguals based in London and collected their speech data by recording the semi-structured interviews. Written data were obtained by collecting the participants’ posts on Sinaweibo, a popular Chinese social media. A detailed questionnaire was also administered among the participants to gather information on their language attitudes and other social backgrounds.
Data and analysis:
For each participant, 30 minutes of the interview were analyzed and 50 posts were chosen for written CS analysis. Quantitative analyses were performed to describe the frequency and pattern of CS. The use of CS across interaction modes was further compared and analyzed against the specific characteristics of each mode and also against the participants’ language attitudes.
Findings:
Results of quantitative analyses show that these participants used significantly more amount and more complex pattern of CS in their online posting. The suggested reason points to an elevated processing demand which is related to the nature of bilingualism and the level of synchronicity of interaction mode. Results of qualitative analyses also suggest that the participants’ positive attitudes towards CS and English language, both overtly expressed and covertly implied, contributed to the varying degree of CS use.
Originality:
This study is among the first attempts to compare CS data across different interaction modes based on same individuals. Comparative data of this kind are quite scarce in the international literature.
Significance/implications:
The findings highlight and provide insights into the dynamic interactions between social, cognitive, and situational factors under the surface of linguistic practice.
Keywords
Introduction
Most theoretical frameworks that deal with regularity and variation in code-switching (CS) have been based on face-to-face interaction. The existing discussion from both social and grammatical perspectives has mainly focused on spoken data (e.g. Clyne, 1987; Gumperz, 1982; Myers-Scotton, 1993b; Pfaff, 1982; Poplack, 1980). However, we do observe the presence of CS in written discourse, either in historical texts of Old English charters and Middle English sermons (Schendl, 2012) or in some modern mainstream publications (Mahootian, 2012). In the past, written CS did not receive much attention, partly because of the scarcity of data and partly because the documented CS use is mostly seen in works of fiction (e.g. Montes-Alcalá, 2012; Muysken, 2005; Timm, 1978) and such use may not represent natural speech and may exaggerate natural usage for literary effects (Gullberg, Indefrey, & Muysken, 2009).
With the rapid growing of communications carried out through network-linked computers, referred to as computer mediated communication (CMC) (Danet & Herring, 2007; Herring, 1996, 2007), CS has been frequently observed across a range of modes of CMC, such as emails, bulletin board systems, instant messaging, and social networking sites. Written forms of CS observed in such digital online environments have been shown to resemble conversational CS, especially in more synchronous CMC modes (Paolillo, 2011). The prevalence of CS use online thus provides promising data for sociolinguistic study of CS and an increasing body of literature has already revealed some interesting results (e.g. Androutsopoulos, 2007a, 2007b, 2013, 2014; Hinrichs, 2006; Paolillo, 2011; Siebenhaar, 2006; Tsiplakou, 2009).
Few works, however, have linked spoken CS to CS in online environments, in particular based on same individuals. Nevertheless, this is a promising research area, as illustrated by a participant’s comment in Seargeant, Tagg, and Ngampramuan’s (2012) study of language choice in online interactions between Thai and English. This participant pointed out that she would avoid using CS in face-to-face or phone conversations whereas she felt comfortable to switch to English when interacting with friends online. This perspective is not negligible as it points to the reasons why we cannot simply duplicate the existing theories and models abstracted from spoken data in a new register such as CMC. Speakers are afforded different opportunities and constrained by different rules in different interaction modes. A comparison of same language users’ CS across interaction modes (e.g. speech and online writing) may give us a better chance to investigate how social, situational and cognitive factors interact with each other and how such interactions are manifested via linguistic means. Moreover, a wider range of CS use and abilities may be revealed in a comparative study of CS across different interaction modes. Those speakers who avoid switching or only do certain types of switching in speech may display a wider range of switching styles in writing or vice versa.
In this paper, I thus propose a comprehensive approach that incorporates both spoken and written CS of the same individuals, which is rare in the existing literature. I compare the frequency and pattern of CS use across speech and online writing to investigate the difference(s). If any difference exists, I explore the reasons why speakers exploit their linguistic repertoire variably in these modes with a focus on the interaction between social motivation, cognitive and contextual factors.
Although a comparative study of CS from different registers regardless of interaction mode (speaking or writing) can also provide insights into intra-speaker variation in CS, the characteristics of CS in online written domain, which stand in sharper contrast to those of spoken CS, highlight a number of theoretically interesting questions. For example, CS use in online environments involves a certain level of consciousness on the part of its authors. A slip of the tongue or other processing effects can be among the reasons for switching in spoken mode, but the same reason is unlikely to hold for CS use in written mode, especially when it comes to two languages whose writing and typing systems are different and switching between them involves a conscious manual act. Therefore, the conscious act of language switching is valuable for research that focuses on stylistic uses of CS, and/or social motivations of CS use. For example, Tsiplakou’s (2009) study of the emails between the members of a relatively close-knit network revealed how the group members performatively acted out their authorship as a competent bi/multilingual speaker with multifaceted linguistic abilities via the means of CS.
In addition, some interaction modes in online environments are less synchronous than speaking and thus give writers more time to prepare and plan their linguistic production. Hinrichs (2006) showed that the creative and playful use of CS for stylization purposes was found in more asynchronous written modes and such use of CS can be attributed to the planned characters of more asynchronous modes. This leads us to consider the possible interaction between the use of CS and cognitive costs of language production, especially in relation to bilinguals with a less dominant language.
Furthermore, the presence of CS in a new domain might offer some more robust methodological procedures for collecting CS data (Paolillo, 2011). It is generally held that real usages of CS are typically obtained in natural and spontaneous speech. The reliability of CS collected either by participant observation or sociolinguistic interviews might be compromised by the presence of the interviewer though (known as “observer’s paradox”; Labov, 1972). However, CS use in online environments could be immune from the effects of the investigator’s observation once access to such data is obtained. The data were often produced before the investigator starts data collection; even in the case where data collection and online writing are ongoing at the same time, it is difficult to imagine that a writer would start writing in a performance style only catering for the investigator and ignore his/her wider audience.
These characteristics of online CS will enable a fuller revealing of intra-speaker variation in CS and expose the influences from social, psychological, and contextual factors to a greater extent. In the following sections, I first define the use of key terms and review some reasons explored in the literature for intra-speaker variation in CS, in particular language attitudes. After that, I compare the two interaction modes in the present study in more detail. Then I continue to introduce my participants and the methods adopted in the study. This is followed by the analyses of the data collected. Finally, I discuss the findings and how they update our current knowledge about intra-speaker variation in CS and language use in general.
Code-switching: definition and syntactic types
What is code-switching?
Poplack defined CS as “the alternation of two languages within a single discourse, sentence or constituent” (Poplack, 1980, p. 583). In fact, many definitions widely used share the core meaning of this one. For example, in Gardner-Chloros, CS is defined as “the use of two languages or dialects in the same conversation or sentence by bilingual people” (Gardner-Chloros, 2009, p. 1). Myers-Scotton also broadly described CS as “the use of two language varieties in the same conversation” (Myers-Scotton, 2006, p. 239).
However, in the current literature on CS, the use of this term is somewhat arbitrary. A comment by Milroy and Muysken reads:
The field of CS research is replete with a confusing range of terms descriptive of various aspects of the phenomenon. Sometimes the referential scope of a set of these terms overlaps and sometimes particular terms are used in different ways by different writers. (Milroy and Muysken, 1995, p. 12)
Among those confusing terms which complicate the terminology in the field of CS, there are, for example, code-mixing and borrowing, as these terms also describe the use of elements from another language in the current interacting language. Take the difference between CS and code-mixing for example. For some researchers, code-mixing is just one manifestation of CS along with other forms of switching. McClure (1977) distinguished two types of CS: code-changing where there is a complete shift to another language; and code-mixing where the shifting to another language takes place within constituents. For other researchers, an opposite approach has been taken by putting CS under the umbrella of code-mixing. Muysken used code-mixing to refer to “all cases where lexical items and grammatical features from two languages appear in one sentence” (Muysken, 2000, p. 1) and claimed that “switching is only an appropriate term for the alternational type of mixing” (Muysken, 2000, p. 4), thus reserving CS only for successive language use. To a certain degree, however, McClure (1977) and Muysken (2000) both agreed that using elements from another language within a sentence indicates a degree of mixing whereas switching involves a shift in language at bigger boundaries, such as clausal or sentential boundaries. In this paper, I adopt the term CS in a broader sense, incorporating both the use of elements from two languages within a sentence and successive language use at clausal/sentential boundaries. The differences between CS and code-mixing as reviewed are acknowledged by differentiating different types of CS, which will be discussed in the next section.
Another debate among CS researchers, and perhaps one of the most heated ones, is how to differentiate borrowing from CS, especially insertional CS, since both involve the use of lexical items from a foreign language. As the present study involves single word switching, this issue becomes particularly relevant. Some previous studies distinguish the differences between borrowing and CS based on the level of structural integration of foreign words into the recipient language. For example, borrowing normally denotes integrating a single lexeme morphologically and phonologically into the receiving language; borrowed items are often recurrent, widespread and known to monolingual speakers of the host language (e.g. Poplack, Sankoff, & Miller, 1988). In the case of CS, on the other hand, foreign elements have not been incorporated into the recipient language in any obvious way. However, phonology and morphology criteria might not be reliable to distinguish CS from load words when it comes to either typologically similar languages in terms of phonological/morphological system (for example, English and French) or structurally distant languages (for example, Chinese and English) as not all languages can integrate foreign words with equal ease (Romaine, 1995).
Another point worth mentioning concerning the relationship between CS and borrowing is their socio-pragmatic indexical meanings. It is believed that when the use of foreign words has become borrowing, fewer indexical values are associated with such use. However, Backus (2015) points out that this is not necessarily the case. In certain bilingual contexts, some established loanwords can take on pragmatic functions and indexical values associated with the language where the loanwords originate. In this case, loanwords can also function as CS. Since it is impossible to completely differentiate the two phenomena at either theoretical or empirical level (Myers-Scotton, 1993a; Treffers-Daller, 1991), it is proposed that we should not consider borrowing and CS as two categorical concepts with static status. Instead, they fall along a continuum where we see completely conventional and unmarked use of items from the same system at one end and fully strategic use of the otherness of foreign elements at the other.
In the present paper, I treat any insertion of single English words (phrases) as CS on the condition that: (a) they are not proper names; (b) they do not have well-established status as borrowed words, that is, they are not phonologically integrated into Chinese and are not known to monolingual Chinese speakers; and (c) they are not used in exclusion of their native Chinese equivalents. Otherwise, these words might have well entrenched in individual speakers to the degree that they are becoming borrowed words (Backus, 2015). To sum up, in this study, CS describes the phenomenon of using English elements in Chinese discourse. When a speaker switches to English to express what could have been expressed in Chinese and it is observed that s/he has used the Chinese equivalents in other parts of the discourse more or less often, such English expressions are examples of CS.
Two types of code-switching
Two main types of switching have been attested in the data: switching within clauses; and switching between clauses. Among the first type (switching within clauses), there is switching for nouns, noun phrases, verbs, and adjectives. These four sub-types account for about 90% of switching within clauses. The remaining 10% of switches of this type include switching for phrasal verbs, adverbs, adjectival phrases, complex prepositions, prepositions, and pronouns (see examples 1–10). In the following examples, Chinese is in italics and the switched elements are in capitals. Free English translation follows each example. If there is more than one case of switching in the same example, the one under discussion is also underlined. The examples are also specified for interaction mode.
(1) Switching for nouns: an example from the interviews
wo xihuan zai wode juzi li fang henduo de EMPHASIS.
‘I like putting a lot of emphasis in my sentences.’
(2) Switching for noun phrases: an example from the online posts
jintian jingran shoudao laizi OMEGA de
‘I received a review invitation from Omega today. I feel a bit flattered. Shall I be a bit stricter this time?’
(3) Switching for adjectives: an example from the interviews
zai jia menkou zuo le yi ci youke, buguo tianqi NICE, zhen shi zenme dou kaixin.
‘I was a tourist today in my own city. The weather was nice, so I am happy no matter what.’
(4) Switching for verbs: an example from the interviews
qishi wo ye bu shi zhege gongsi de, xianzai dengyu shi wo zai zheli zuo shiyan, danshi PAY wo de hai shi women xi.
‘Actually, I am not working for this company. I am doing experiments here, but it is still our department that pays me.’
(5) Switching for phrasal verbs: an example from the online posts
zhezhong cuowu dou fan le duoshao ci le, zan neng bu neng
‘How many times has this mistake been made? Can you make sure that you go over everything before you send it to me every time?!’
(6) Switching for adverbs: an example from the interviews
wo zhijie gen renjia shuo yige xiaoshi, juede zheme chang a, wo zhijie shuo ban ge xiaoshi, yige xiaoshi ROUGHLY dou keyi.
‘If I say it needs one hour, they will think it is too long. So I will just say half an hour, but it should be fine if you go up to roughly an hour.’
(7) Switching for adjectival phrases: an example from the online posts
xuexiao yongyuan shi nide guoqu, jiusuan nide xuexiao shi ABSOLUTELY FANTASTIC, danshi ruguo meiyou gongzuo jingyan, zhe dou shi meiyou yiyi de.
‘Which university you attended is always your past. Even if your university is absolutely fantastic, without work experience, you still cannot go far.’
(8) Switching for complex prepositions: an example from the interviews
yinwei dajia zai zuo zhe jian shiqing de shihou, dou shi zai
‘When we were doing this, most people are doing it on top of a full pay job. So, we had to do it in our spare time.’
(9) Switching for prepositions: an example from the interviews
na ni zhege shi UNDER xinlixue ma, hai shi xingweixue?
‘So, yours (subject) is under psychology, or is it sociology?’
(10) Switching for pronouns: an example from the online posts
zhe zhong EVERYONE dou dong de daoli qishi shi zui nan zuo dao de
‘What everyone understands is the most difficult to put into practice.’
Switching of the second type in my data takes place between clauses. Speakers switch to a different language (English) at clausal boundaries and a whole clause is realized in English in contrast to other Chinese clauses. In most cases, it is a single clause in English amidst discourse in Chinese. Although the base language of interaction is still Chinese, switching of this type involves a complete shift to English in both lexicon and grammar. When examined in a micro-context where only the adjacent two or three clauses are considered, it seems as if a speaker is alternating between two languages. For this reason, I refer to switching between clauses as alternational type (cf. Auer, 1999; Muysken, 2000). Compared to alternational code-switching (ACS), the first type of switching (within clauses) is featured with English words being inserted into Chinese base language, guided by Chinese grammatical rules. Therefore, it is re-named as insertional code-switching ( ICS).
Four types of ACS are distinguished: simple independent clauses; complex independent clauses (with one or more dependent clauses); compound independent clauses (coordinated independent clauses); and dependent clauses (embedded clauses). Examples 11–14 illustrate each type.
(11) Switching for simple independent clauses: an example from the interviews
(Context: talking about the pain in birth giving) tamen hui juede zhe hen zhengchang, hui shuo zhe shi nvren de zhize, YOU ARE DESIGNED FOR IT.
‘They will think this is normal, and they will say this is women’s responsibility. You are designed for it.’
(12) Switching for complex independent clauses: an example from the interviews
ye meiyou ren jiandu ni a,
‘There is no one to supervise you. You do it because you want to do it well, and you know whether you do well or not, right?’
(13) Switching for compound independent clauses: an example from the online posts
JORDAN SPIETH GAVEＡBRILLIANT PERFORMANCE IN THE MASTERS THIS WEEK AND HE BROKE MANY RECORDS. nianqing ren qiantu wu liang a.
‘Jordan Spieth gave a brilliant performance in The Masters this week and he broke many records. A fantastic future is waiting for the young man.’
(14) Switching for dependent clauses: an example from the online posts
LIFE OF PI gaosu wo yige daoli, qianwan bie qu zuo shenme youlun, wo juedui huo bu xia lai
‘What I learnt from Life of Pi is that I will never ever go on a cruise trip. I would never survive if anything happened.’
There is a third type of switching which cannot simply be described as taking place within or between clauses as they are seen both within and outside clauses (or at the peripheral position of a clause). They are mostly discourse markers or swear words (see examples 15 and 16). This type of switching was referred to as tag/emblematic switching by Poplack, which was considered as “an emblematic part of the speaker’s monolingual style” (Poplack, 1980, p. 589). Although Poplack did not consider them as true code-switches, I still treat them as CS as their Chinese equivalents are used more or less often. Switches of this type can be freely moved to anywhere in a clause and they are less strictly governed by syntactic rules. Moreover, switching for discourse markers/swear words does not involve a shift in grammar. The syntactic rules of the clauses are still provided by Chinese. In this sense, they behave like ICS and I put them under this type.
(15) An example from the online posts:
xia ge libai xiu nianjia de jihua you paotang le, laoban de nazhang choulian baiming le shei xiujia shei gundan, daomei a, ANYWAY, wo ye dai bu liao duo jiu le, ai za za di ba.
‘My plan to take the annual leave next week is aborted. That long donkey face of my boss says it all: whoever takes leave now might as well leave forever, bad luck, anyway, I am not staying for long, whatever.’
(16) An example from the interviews:
wo bushi buxiang huiqu, YOU KNOW, zhishi you de shihou huiqu faner hui xuanze geng shao.
‘I am not saying that I don’t want to go back, you know, I just feel that I would have less choice if I go back.’
The role of language attitudes and identity construction in intra-speaker variation in code-switching
In the literature, there are various reasons accounting for the intra-speaker variation in CS, for example, the level of formality. Formal contexts supposedly inhibit switching behavior and this is one of the main reasons why researchers always aim for naturalistic and informal contexts to elicit CS data as they believe informal settings are where CS or real CS is most likely to happen (e.g. Auer, 1988; Li Wei & Milroy, 1995). The role of interlocutor and topic is also pertinent when explaining why speakers code-switch or code-switch to different extents in different contexts. For example, Blom and Gumperz (1972) found that their subjects switched language when there was a change of social situation involving different interlocutors (situational CS) or a change of topic (metaphorical CS). Myers-Scotton (1993b) noted that speakers’ language choice including CS responds to their perceived set of rights and obligations of one participant in relation to those of others in a talk exchange. Androutsopoulos (2014) further pointed out that the role of audience provides a backdrop against which we can better understand individuality in language choice on social media.
Another important factor accounting for the intra-speaker variation in language switching is related to language attitudes and identity construction. In fact, language attitudes are embodied in the choices made by speakers in response to other contextual factors, such as interlocutor and the level of formality. Speakers display their beliefs about their perceived appropriate language use and perform “acts of identity” through language switching. Their language choice in a particular occasion manifests their orientation to the desirable qualities indexed by the language(s) and their identification with certain social groups that embody the qualities and speak the language(s) (Eckert, 2008 Gumperz, 1982). In some cases, speakers switch to their ethnic language to show affiliation with their minority ethnic identity (Finnis, 2013; Poplack, 1980; Toribio, 2002). In some other cases, speakers switch to a language that is associated with some higher symbolic values in the linguistic marketplace to borrow or claim the associated identity (Bourdieu, 1977; Heller, 1995). It is worth pointing out that sometimes such use in the latter case can elicit negative comments. Some typical evaluations are termed as “showing off”, “ill-mannered” or “arrogant” (Gibbons, 1987).
Another way of approaching CS as an index of identity is to consider a mixed or hybrid identity that is indexed by switching/mixing languages. Speakers do not identify with one culture/language at the expense of the other. Instead, they claim joint membership of the social groups with which their languages associate. Their mixing styles construct their identities as living in or between two or more cultures/worlds (Bailey, 2001, 2002).
However, we need to take caution when establishing any a priori relationship between language choice and social identity. To avoid pitfalls of essentialist views of linking one language to one identity or equaling hybrid language use to hybrid social identities, we need to survey the actual interactions and local contexts (Auer, 2005). Most likely, the construction of identities by language choice is emergent in interaction and can vary from case to case. Local conditions, such as the nature of bilingual speakers, the relative status of each language, and the local life trajectory should be taken into account.
Two contrasting interaction modes: interview talk and online posting
In this study, I collected my data from two interaction modes, interview speech and posts on social media. The type of interview I did with my participants is what Labov (1984) described as sociolinguistic interview, where the structure of the interview was comparatively loose, the atmosphere was light and casual, and the participants were encouraged to talk in long stretches in their everyday natural speech style. Most interviews took place in local cafés where there were other customers present. The topics touched upon their previous life in China, present work, leisure activities, family and friends, and social news. These participants were recruited by the “friend-of-friend” method (Milroy, 1980; Milroy & Gordon, 2003); therefore, many of them did not know the interviewer before the study. However, the interviews did not take place when both parties were completely strangers to each other. The interviewer met them for at least once to build up a sense of familiarity before doing the actual interview.
Some may argue that using interview to elicit natural conversational data can be rather problematic as participants are always aware of the roles of each participant in interview and appropriate behavior to each role. On the one hand, I tried to minimize the risk of eliciting formal speech by building up familiarity with participants, as mentioned above, also by adjusting conversation roles, that is, putting myself in a position of lower authority who, as a student with short-term residence in the UK at the time of interview, was learning from my participants who have been in the UK for more than three years and thus have gained more social experiences (cf. Labov, 1984). On the other hand, interview speech still rests on the wide spectrum of a speaker’s speech styles and it represents how they adjust their speech and react to situations when they perceive any asymmetric roles in conversation. After all, self-conscious speech varieties are not uncommon in speakers’ everyday life and such varieties to some extent can display speakers’ own perceptions and evaluations of language varieties and the relationship of the different varieties to power negotiation and identity construction (Schilling, 2013).
The social media where the participants’ posts were collected is SinaWeibo, which is one of the most popular social networking websites among Chinese internet users. The site works in a way that is similar to Twitter. Users post social or personal events with a limit of 140 characters; they can also re-post other users’ posts with or without their own comments. Their posts and comments are open to the view and comments of their friends on the site as well as other unknown users, although views and comments from unwanted users can be restricted.
These two interaction modes contrast each other in many ways, for example, the audience. In interview, each participant addresses a single audience with the possibility that there might be eavesdroppers present. In online posting, the participants write for their friends online and a wider unknown readership. Another example of the differences between these two modes may be the level of formality. Despite the efforts that were made to minimize the formalness of the interview, the participants’ knowledge of interview as a well-defined speech event may still elicit quite formal speech style. Compared to the interview, they may feel more relaxed in this particular online space. This is shown in their use of local dialects, which are typically observed in in-group and informal conversations, and even some taboo words in their posts.
Among all the differences between interview talk and online posting, the level of synchronicity is perhaps among the most important. It is often assumed that spoken modes are more synchronous than written modes. Certainly, interaction modes which typically require immediate and online processing of incoming information, such as speaking, are more synchronous than those allowing a “time-out”, such as writing.
In the current study, interview talk is one-to-one conversation that is featured with rapid secessions of turn-taking. The interaction is taking place in real-time and the participants usually concurrently process a multi-fold workload. They simultaneously store and process information and make decisions about the changing environment based on their knowledge about the world. The high level of synchronicity places limitations on a speaker’s working memory, which means that the short-term storage of verbal and visual–spatial information in language production becomes limited as the interaction mode becomes more synchronous (Baddeley, 1986). For this reason, a speaker might avoid some complex constructions and less familiar expressions to reduce the possibility of making mistakes.
In online posting, a writer initiates contributions that potentially invite comments from their “friends” and other readers registered on the same website. Compared to spontaneous speech and other types of online texts, such as real-time chat-groups or forums, online posting is much less synchronous. Before their posts go online and become available for view, the users have plenty of time to edit and plan the contents of the contribution and the intended style of language use. They are not under the pressure of processing and producing information in real time. They are able to retrieve information from long-term memory and the cognitive burden on their working memory is thus eased (Parker & Coiera, 2000). The low level of synchronicity in this type of online writing and the related lower processing demand may encourage innovative and playful use of language (Hinrichs, 2006).
Second language (L2) speakers and their language processing of L2
The 40 participants of the present study are a group of Chinese–English bilingual professionals who are based in London. They fell into the age range between 25 and 40 and their length of residence in the UK varied from 3 to 12 years at the time of the interview. They are first-generation immigrants, and most of them came to the UK without any previous migratory link. In contrast to traditional Chinese immigrants in the UK, who are more concentrated in the catering business, members of this group are scattered across a range of different occupations. Among them, there are teachers, doctors, architects, bankers, software developers, and office administrators. In Li Wei (2007), the term “professional transients” was used to broadly refer to this group of immigrants. Most of them came to the UK for educational or professional purposes. After their education or short visit was finished, some of them chose to stay and even to establish their home in the UK.
This group of speakers is remarkably fluid and a majority of group members have a sense of uncertainty regarding settling down or going back to China for various reasons, including personal considerations and immigration policies. This indicates a high level of mobility and a sense of social insecurity. Examining their language practice and CS behavior will provide us with new insights into how linguistic resources are employed when speakers maneuver between two cultures and between social and individual realities.
Chinese language consists of many different dialects which may be mutually unintelligible (Chen, 2004). As a result, there are speakers who are bilingual in standard Mandarin and their local dialect in many places in China. In order to reduce complications of having effectively trilingual people (local dialect, standard Mandarin and English) who might be doing different combinations of CS, the current participants were also selected to only include speakers from a northern dialect area and who speak Mandarin Chinese as their first language (L1).
These participants acquired some English before coming to the UK, and the acquisition of English was mainly conducted in classroom settings after the age of 12 (a few had engaged in naturalistic learning). Their use of English in daily interaction only became more frequent after they came to the UK. There is not much variation in the self-reported English proficiency among the participants, ranging from 18 to 23 out of a full score of 28 (combined ratings on reading, listening comprehension, speaking and writing). The self-reported ratings seem to indicate fairly high proficiency in English. In interviews, however, a majority of participants reported that their English proficiency is very much domain sensitive, that is, more proficient when talking about some domain-specific topics, especially work-related. When approached by some unfamiliar topics, their performance in English is affected, sometimes severely. In this sense, they are not considered to be balanced bilinguals, but second language (L2) English users and/or learners.
However, different from L2 English learners in mainland China, these participants are situated in an English-speaking country where Chinese is preserved for Chinese-only domains and English has become their language of wider use and of interaction with people from other linguistic and ethnic backgrounds. Therefore, a new role might have been assigned to English among these participants.
A noticeable feature of L2 speakers’ production of their less dominant language is that it may be more subject to the effects of an elevated processing load and of limitations on working memory capacity (Payne & Whitne, 2002). Such effects can be more pronounced in L2 leaners whose control of their L2 is less automatic. For these speakers, the processing of L2 is controlled, and it involves a complex set of cognitive tasks. They need to regularly intervene the production process, that is, to comprehend language input, to retrieve L2 mental lexicon and syntactic representation that are perceived socio-pragmatically appropriate to the register, and to make decisions about the ongoing and changing situation based on their previous knowledge of the L2 and the world.
Paradis (1997) suggested that the fact that L2 speakers, especially L2 learners, rely more on explicit knowledge than on implicit competence can be one of the reasons why they need to consciously control and monitor their L2 production. Their explicit knowledge is the conscious awareness of their explicit analysis of the speech situation (the match between their L2 knowledge and the ongoing linguistic episode), whereas implicit competence can be seen as inherent and built-in procedures that prompt linguistic production. Kecskes and Papp (2000) also pointed out that the less automatic processing in L2 production can be attributed to a lack of conceptual fluency of L2 speakers. Consequently, L2 production can be demanding and involves a great deal of attention, which limits L2 speakers’, in particular L2 learners’ capacity for storing verbal and visual–spatial information in their working memory.
Methods
For each participant, the middle 30 minutes of the recorded interview were chosen for spoken CS analysis and 50 posts, 25 from before the date of the interview and 25 from after, were isolated for written CS analysis. As it is the overall rate of switching that was measured, not all of the posts were bilingual, and some were monolingual Chinese posts. Some single word/phrase posts were also included and they are usually a single word/phrase comment on other users’ posts that were in Chinese. When such a comment was in English, it was considered to be an instance of CS because it was still embedded in a Chinese-speaking context as a response to a Chinese post. However, single word/phrase comments are very rare and there are only 3 of them. No further specific selection criteria (e.g. length and content) were applied.
To examine the general frequency of CS, I quantified the occurrence of CS in percentage with finite clause as the measurement unit (cf. Stell & Parafita Couto, 2012). The percentage value represents the proportion of clauses that were realized in or with CS in the total number of clauses that were uttered or written down by a participant. The frequencies of ACS and ICS were also measured in percentages. Moreover, I also counted the number of CS instances and coded grammatical construction of each case to reveal and compare the make-up of switched elements across interaction modes.
After the interview, the participants were asked to fill out a detailed questionnaire, which was designed for a wider project (Liu, 2015) to elicit information of the participants’ language attitudes, social network and language usage. They rated Chinese and English as well as CS in term of status and solidarity. Their cultural and ethnic orientation was also investigated. At the same time, I also observed and noted down after the interview the participants’ attitudinal remarks in relation to Chinese and English, and also to different kinds of linguistic behavior that they experience in their daily life. These attitudinal remarks and their answers in the questionnaire were analyzed against their usage of CS across interaction modes.
Results
Comparison of CS frequency
The variation in the spoken CS is plotted in Figure 1, where the frequencies in overall CS, ACS and ICS are graphically represented. The participants on the horizontal axis are ordered by the increasing frequency of overall CS use in speech. Figure 1 indicates that most of the participants did not use ACS at all in their speech. This is also shown by the zero median in the percentage values of spoken ACS. As a result, there is very little difference in frequency between CS and ICS (as shown by the overlapping black and dark grey lines in Figure 1). Not surprisingly, the difference between ACS and ICS use is very significant (Z = −5.443, p < 0.01). Overall, what has emerged in the spoken mode is a pattern that CS is characterized by the almost exclusive use of ICS and the extreme rarity of ACS.
Variation in the frequency of code-switching (CS), alternational CS and insertional CS in interview speech.
On the other hand, a snapshot of CS in the online posts is shown in Figure 2. Once again, the frequencies of ACS and ICS use are presented alongside overall CS use. The ordering of the participants on the horizontal axis is by the increasing frequency of overall CS in writing. The pattern of CS use shown in Figure 2 is more difficult to decipher. CS in writing is not dominated by any subtype of CS but consists of a more mixed use of both types, as indicated by the similar mean values of ACS use (6.35%) and ICS use (6.73%) and the non-significant difference between them (Z = −0.884, p = 0.376).
Variation in the frequency of code-switching (CS), alternational CS and insertional CS in online posts.
Compared to the spoken CS, the total use of CS in the written mode was more frequent, with an average use of 11.16% in the total number of finite clauses that were analyzed. There were three participants (F2, F14 and M6) who exhibited exceptionally high CS frequency in their online posts, above 40%. Another prominent feature of CS use in the written mode is the relatively higher use of ACS, with a mean of 6.35%. However, it should be noted that the median use of written ACS among the participants was only 1.43%, indicating that half of the participants used ACS no more frequently than 1.43% and the other half used it to a much greater extent. As a result, not only was overall ACS use more frequent in the written mode, the variation displayed in its use was also wider.
Insertional code-switching use also became more frequent, with the average use increasing from 4.24% in speech to 6.73% in online posts. Compared to ACS, variation in ICS use was more evenly distributed among the participants, as shown by the smaller difference between median (5.20%) and mean (6.73%) values. To sum up, the written mode received more occurrences of CS and both types of CS. The use of ACS, in particular, seems to be reserved only in the written mode. The results of a series of Wilcoxon signed-rank tests confirmed that the differences in CS between the two interaction modes are significant.
Description of the details of code-switching in interview speech and online posts
Tables 1 and 2 compare the sub-types of ICS and ACS in speech and online posts. In both tables, the sub-types are listed in the left column. The middle column lists the types of ICS/ACS that were found in interview speech and the right one lists those found in online posts. The checkmark “✓” indicates the presence of each type in each mode. The percentage of each sub-type in the total number of ICS/ACS in each mode is also presented.
A comparison of interview speech and online posts on the make-up of insertional code-switching (ICS): the frequency of different parts of speech.
A comparison of interview speech and online posts on the makeup of alternational code-switching (ACS).
As shown in Table 1, interview speech and online posts did not differ from each other to a great extent. Switching for nouns and noun phrases were the more common types across the two modes. This is in line with the results from earlier research on other language pairs, for example Pfaff (1982) on Spanish and English, Meisel (1994) on French and German. The two interaction modes also shared almost all the types of ICS except two, prepositions and complex prepositions. In fact, these two sub-types of ICS were only used by one participant (F2) in her interview speech. It appears that this participant not only used CS with exceptionally high frequency, her range of CS types was also wider.
A noticeable difference between the two modes is the frequency of switching to discourse markers. In interview speech, they are the second most frequently switched elements (with a frequency of 21.99%), which confirms the previous research findings that discourse markers are prone to be switched in bilingual speech (e.g. Pfaff, 1982). In online posts, their frequency has dropped to 3.40%. In addition, the most frequently observed discourse markers differed in these two modes. In interviews, among the frequently used English discourse markers, there were “yeah”, “you know”, “anyway” and “like”; whereas in online posts, “anyway”, “well” were the only two English discourse markers that have been observed.
To account for the observed difference, we need to understand what role is played by discourse markers. Schiffrin (1987) defined discourse markers as units that coordinate talk. In spontaneous interaction, the real-time processing of information exchanged allows speakers little time to prepare and edit their speech. The premium on fluency in speech (Smith & Clark, 1993), on the other hand, then prompts speakers to adopt solutions that could prepare their audience with upcoming looseness and some off-record information. One of such solutions is to use discourse markers, especially those that are temporally sensitive, such as “like” or “you know”, which “are linked to the temporal pressures of communicating with a waiting addressee” (Fox Tree, 2015, p. 70).
When it comes to discourse markers in written contexts, previous studies show that they are more typical in more spontaneous written discourse, such as instant messaging (Fox Tree, 2015). In online posts, writers’ initiating contributions could lead to spontaneous discussion thereafter, but the initial posts are less spontaneous. As there is no pressure of waiting addressee, writers are afforded this opportunity to edit and plan their message more carefully and the relationships between concepts could be more worked out. That being said, it does not completely obviate the need for discourse markers. Some discourse markers, of which the main function is to relate information before and after and to provide transition through topics, are equally useful in speech as well as in written passages (either spontaneous or less spontaneous), since they help writers structure and plan the discourse (Fox Tree, 2015). The two discourse makers that have been observed in online posts, “well” and “anyway” are examples of this kind.
When it comes to ACS, interview speech and online posts showed more divergence. In Table 2 we can see that ACS in speech mainly consisted of simple independent clauses. There were only a few complex independent clauses (n = 2) and dependent clauses (n = 2), which were again only witnessed on one participant (F2). This means that most participants did not use ACS in speech. When they did, they only used simple independent clause structure. In online posts, ACS type was more varied and more complex types appeared. Although over half of ACS were still simple independent clauses, the proportions of the other three types have become bigger. Therefore, not only was ACS in online posts more frequent, it was also more varied and more complex in structure.
The participants’ language attitudes
As mentioned in the Methods section, I surveyed the participants’ attitudes in terms of the status and solidarity of English and Chinese, their ethnic orientation, the purposes of using English, and general evaluation of CS. In this sub-section, I briefly report the results of the survey before proceeding to discuss the implications of the study.
The eight box-plots in Figure 3 represent the variation found in the attitudinal variables. The participants’ ratings on these variables have been normalized into percentages. Therefore, the degree of their orientation with respect to these different dimensions can be measured on the single y-axis in Figure 3. As shown in Figure 3, most of the participants rated Chinese very high on solidarity with a mean of 82.5%. As their ratings on solidarity of English were distributed in a complementary manner with those on Chinese solidarity, this means that they did not perceive English as a language of solidarity. On the other hand, English and Chinese were rated similarly high on status (Chinese with a mean of 80.95% and English 81.13%). The slightly higher status of English than Chinese was not significant (t = 0.104, p = 0.918).
Variation in the variables of attitudes among the participants.
When it comes to cultural alignment, the ratings on both cultures were quite high. Again, the marginal difference between Chinese and English in this respect was not significant (t = −0.409, p = 0.685), though Chinese culture was viewed slightly more positively. Two female participants (F14 and F15) aligned themselves exceptionally closely with English culture. One male participant (M3) detached himself particularly away from English culture and another one (M6) away from Chinese culture instead. In terms of the attitudes towards CS, a relatively favorable and tolerant view was shared in general, indicated by an average rating of 64.52% and also the fact that almost half of the participants gave a rating of 70% or higher. Additionally, the participants agreed more upon integrative than instrumental purposes of using English with statistical significance (t = 4.16, p < 0.01).
In sum, the above description shows that the participants were positive about both languages and cultures overall. In addition to prescribing high status and high solidarity to their native language, they were also positive about the future of the Chinese language worldwide, exemplified by their positive answers to questions like “Do you think proficiency in Chinese might become a big advantage in the international job market?” Such positive evaluations of Chinese indicate a high vitality (Allard & Landry, 1992) of this language, even in a host country where Chinese is a minority language.
It is not surprising that slightly higher status was attributed to English, which is not only the mainstream language used in the host country but is also considered as a global language (Crystal, 2003). English has traditionally been linked to modernity and prosperity among many Chinese speakers (Zhao & Campbell, 1995). The current participants have acquired English as a L2 in China. It is thus reasonable to suggest that they have also acquired the social significance of English. Equally important is that these participants also considered speaking English as a means to integrate into local communities. As they are residing in an English-speaking country, English does not necessarily relate to upward social mobility. Instead, speaking English could index a desire to integrate into local communities, regardless of higher or lower socioeconomic status of speakers from the communities.
Discussion
The findings presented in the Results section regarding intra-speaker variation in CS use can be summarized as that the participants used more CS, and both patterns of CS, in their online posts. The use of ACS, in particular, seems to be reserved only in the written mode. In addition, ACS used in online posts was also more complex in structure. I now turn to my interpretations of these patterns in relation to the nature of bilingualism of the current participants, the different characteristics of interview speech and online posting, that is, what unique opportunities and constraints are present in each mode, and also in relation to the participants’ own language attitudes.
An elevated cognitive processing load: L2 leaners in synchronous modes
Frequent use of CS in the online written mode as shown in the findings is not surprising. The online space, unlike traditional written mode, does not require formal language use. Typical features of oral conversations are frequently seen in online environments (Collot & Belmore, 1996; Dorleijn & Nortier, 2009). What appears interesting is that the participants’ frequent use of CS in their online posts exceeded their CS use in oral speech. This is understandable if we take into account the fact that these participants are L2 English speakers/learners. A majority of them started learning English after the age of 12. From the point of view of the critical period hypothesis (Bialystok & Miller, 1999; Patkowski, 1980), it is very rare for these speakers to acquire the language as fully as their L1. Combined with their own comments on their English abilities, I thus treat them as L2 learners. As explained earlier, L2 production by these speakers is more subject to the effects of an elevated processing load and of limitations on working memory capacity due to the need for them to constantly and consciously monitor their L2 production (Kecskes and Papp, 2000; Paradis, 1997).
Of course, such effects can be mitigated by advanced L2 competence level and high intensity of language use (Payne & Whitne, 2002). Highly proficient L2 speakers reply more on their proceduralized knowledge and are thus able to produce L2 more automatically (Dewaele, 2002). This explains why the participants were more confident to report high proficiency in work-related domains as their repeated use of the relevant language elements for specific purposes have been routinized and proceduralized.
Another factor that may increase the burden on working memory is the level of synchronicity of the interaction mode. As mentioned earlier, more synchronous interaction modes, like speech, are featured with rapid rate of turn-taking, little time for speech preparation, and spontaneous reception and processing of information. The high pressure of multiple-tasking sometimes might result in failure in working memory and retrieving of prospective plans can be partly or completely forgotten (Parker & Coiera, 2000).
Therefore, language production in L2 or using L2 elements (as in the case of CS) in synchronous speech may present a challenge for L2 learners. The level of complexity of L2 elements and the amount of L2 production can affect the severity of the challenge. It is likely for them to fail in retrieving appropriate lexical items and to make mistakes when constructing grammatical constructions. More complex L2 constructions (as in the case of ACS) and lager amount of L2 production (as in the case of frequent CS) will perhaps pose a bigger challenge for L2 learners. One strategy to avoid situations where they might be embarrassed with influent speech and false L2 production is simply avoiding using L2 or complex L2 elements. As we can see from the results, this seems indeed to be the strategy that was taken by my participants. In the interviews, the participants’ employment of CS and of ACS in particular significantly dropped. Among the handful cases of ACS, the constructions were limited to simple independent clauses.
Therefore, CS production by these L2 speakers/leaners across interaction modes is suggested to be inversely correlated with processing burden on working memory capacities. In synchronous speech where processing demand on working memory is high, the frequency of CS was lower and higher linguistically demanding ACS was very rare; in less synchronous online posting where the burden on working memory is eased, both the frequency and complexity of CS increased.
It should be noted that the role of synchronicity in CS should be analyzed against the nature of language users’ bilingualism. For example, an earlier important study on CS in CMC modes by Paolillo (2011) concluded that more Punjabi/English CS was found in synchronous internet relay chat (IRC) than in asynchronous Usenet. Paolillo’s subjects were more balanced bilinguals in the sense that they were principally English-educated and their use of English in social life expanded across more domains. The reason proposed was not related to the participants’ cognitive processing of the languages involved. Instead, the role of synchronicity was displayed in enabling a greater degree of shared context in synchronous IRC as the participants shared message flow in real-time and were less selective of the contents. CS and the interpretations of code-switched discourse can be better understood in this shared context. Therefore, different types of bilingualism can give rise to different conclusions when considering the role of synchronicity in CS variation.
Some might also argue that the level of formality could affect the participants’ use of CS because interviews, as introduced earlier, are sometimes unavoidably perceived as a formal occasion and thus entail more formal language use. It can be the case that the formal setting of interviews may inhibit the use of CS. 1 But such perspective does not counteract the effects of synchronicity. The effects of synchronicity emphasize that it is the burden on working memory that discourages the use of CS. If the participants did consider interviews as a formal setting, they would perhaps pay more attention to proper language use. In addition, their interlocutor in the interview was a new acquaintance. According to Swann, Stein-Seroussi, and McNulty (1992), a conversation (even an informal one) with a new acquaintance places many demands on speakers, which could be an effortful and controlled process (Gilbert, Krull, & Pelham, 1988). During this effortful process, aiming for proper language use in a synchronous language production mode involves an extra burden on working memory, which further discourages CS because L2 production in a formal and synchronous setting can lead to more hesitations, lengthy pauses, self-corrections, and mistakes. From this point of view, the role of formality and that of synchronicity are similar and they both inhibit CS use as they both put extra burden on the participants’ working memory.
Arguing for the role of synchronicity and the related processing demand in affecting CS use has an implication, that is, CS is desired by the participants. It is unlikely that participants used more CS in online posts simply because the processing load is lower. More use of CS in lower demanding contexts indicates that the participants desire to use CS. This drives us to refer back to the participants’ attitudes.
They ARE Chinese, but not limited to Chinese: the role of attitudes
The survey results of the participants’ attitudes show that they were positive about CS in general and considered the use of English elements in Chinese discourse as natural and unavoidable. They reported that some English words, compared to their Chinese equivalents, are activated more frequently and thus easier to access. Some words were even acquired in English so they understand better the referential, semantic and pragmatic connotations of these words. In this sense, it is understandable and natural to use English words in Chinese contexts to fulfil lexical needs and make up temporal memory loss. No negative evaluation will be evoked as they themselves are practicing CS use of this type. Such use of English in Chinese discourse to some extent indexes the membership of a broadly defined Chinese–English bilingual/bicultural community in London.
Their positive attitudes, however, only point to the naturalness of CS and the objection of CS as a stigmatized form and a showing-off practice. Attitudes as such cannot explain why they used more CS in the written mode. If it is only natural, they should be using CS at the same frequency or even more in speech, because they might forget some Chinese words or expressions which have been activated with lower frequency, and the higher pressure of rapid turn-taking successions in spontaneous speech will motivate them to use English. This is apparently not the case, as shown by the results. Therefore, their attitudes towards CS may be more complicated than what is shown in the survey results.
As a matter of fact, my observations of their attitudinal remarks in the interviews show that these participants’ positive attitudes appear to only hold for the use of some individual English words and some idiomatic expressions. They did not approve long stretches of CS use. They seem to have a clear boundary of what is acceptable and what is not. The following quotation from participant F3 gives an illustration:
Although I said just now that it is better not to mix Chinese and English, you might have noticed that I myself occasionally mix too; because you cannot avoid it and I think it is normal to mix like I do, because you use some English words more frequently and get used to them. You just don’t remember the Chinese equivalents sometimes. Instead of trying hard to think of the Chinese words, I might as well use English. The hearer knows what I mean and it saves energy. . . . But if you just keep talking in English and insert whole English sentences, that would be then too weird, as if you cannot speak Chinese properly anymore. But we know that’s not true. That is just showing off.
Such evaluative comments are not uncommon among the participants, but the analysis results of their frequency of CS use across interaction modes apparently do not agree with such comments. If ACS is not acceptable and it was indeed very rare in speech, why was it used far more frequently in the written mode? One possible reason is that there are some other aspects of attitudes that are not revealed or overtly expressed. Let us take a look at the following example:
I would feel very uncomfortable and I would feel like he has something else to say. Is he trying to tell me that his English is good? Or is he trying to say he is awesome? I don’t feel comfortable anyway. Would it not be nice to stick to Chinese and be down-to-earth?
In this comment, despite of the negative response, long stretches of English use in Chinese discourse are marked language choice and imply a flagged piece of information, which is linked to good proficiency in English and “awesomeness”. Evaluations as such remind us of the participants’ attitudes reported in the Results section that they rated English high in terms of status and they had fairly strong motivation to integrate into local English-speaking communities. The prestige they assigned to English might relate to the acknowledged status of English being the lingua franca internationally (Crystal, 2003), and also to the association between English and good education and advantageous social position among Chinese L2 English speakers/learners in general. But perhaps, more importantly for the current participants, this prestige of speaking English is more closely linked to the desire to integrate into the local communities. Long stretches of switching to English to a certain extent imply higher proportions of English-speaking ties present in a speaker’s networks and more participation in the local English-speaking communities, which suggests that the speaker is successfully integrated into the local communities of the host country. S/he is not marginalized but is actively embracing and being embraced by the majority culture. Therefore, my opinion is that the “awesomeness” hinted at in the above example not only articulates a link between English and good education, more importantly, it highlights a degree of integration into the host country, which is shown to be desired by most of the participants.
However, public opinions usually do not approve obvious “flaunting” of one’s advantageous social position. In addition, the force of normativity to conform to traditional Chinese values, especially in an interview setting, could also be strong among most participants. Their high ratings on Chinese solidarity and Chinese cultural alignment dimensions shown in Figure 3 lend support to this possibility. Consequently, the link between English use and successful integration into local communities is not overtly expressed but only hinted at through the use in context (c.f. Gibbons, 1987). In fact, this force of normativity in face-to-face interaction could also be in force along with the higher processing demand in synchronous modes to further constrain the use of CS, especially ACS in spoken mode.
So far, it is shown that the participants’ attitudes towards CS appear to be mixed. On the one hand, CS is natural and understandable, but only so if the switched elements are individual words and some idiomatic expressions. On the other hand, the use of CS and ACS in particular is somehow “stigmatized”, but the negative comments were made in relation to showing-off and flaunting. It is instead linked to advantageous social position and successful integration into local communities. The mixed attitudes towards CS to a certain degree reflect the participants’ social positions and how they maneuver between the two cultures. They embrace their overseas experience and the culture diversity they are experiencing; at the same time, they still hold on to some traditional Chinese cultural values (e.g. being down-to earth) and consider Chinese language as having high solidarity. They do not see themselves as completely westernized; meanwhile, they are different from traditional Chinese in the resources they command, culturally and linguistically. They are neutral in between (Gibbons, 1987) and also comprise both. Linguistically, they have been “carving out a space” (Zhang, 2005) where they fit themselves in, that is, a space where they ARE Chinese, but not limited to Chinese; which is IN the U.K., but crosses the boundaries of the two countries. Their use of CS, and the meanings they attribute to the degree and types of CS use, along with other practices of life, are constructing an identity of overseas Chinese professionals.
Conclusion
In this paper, I described how the participants variably code-switched across two distinct interaction modes, that is, interview speech and online posting. They used more CS, and both patterns of CS, in their online posts. The use of ACS, in particular, seems to be reserved only in the online written mode. In addition, ACS used in online posts was also more complex in structure. Comparative data of this kind are quite scarce in the international literature and thus have contributed to a more solid empirical foundation to establish comprehensive theoretical frameworks in CS.
I also discussed how intra-speaker variation in CS responds to the change in the level of synchronicity of interaction mode and how the role of attitudes interacts with the level of synchronicity to contribute to the varying degree use of CS. The participants’ positive attitudes towards English and CS, either overtly or covertly expressed, could motivate their adoption of CS despite of some negative comments. However, the high pressure on working memory in synchronous speech, combined with L2 speakers’/learners’ sensitivity to an elevated processing demand, may discourage CS to minimize the risk of inappropriate use. In online posts, on the other hand, the participants have more time to plan, edit and re-read their own posts, so they are at more ease to produce desired language style to project the self-image that they prefer to be interpreted by their audience. In addition, the participants’ orientation to some traditional Chinese values and the force of normativity in face-to-face interaction may also be in force to constrain the use of CS, especially ACS in speech.
The two interaction modes discussed in this paper contrast each other in many different ways. There is a long list of how many other reasons might affect the use of CS, for example, the level of formality, audience (see Androutsopoulos, 2014), topics and even the language used in previous activities prior to the interview or the language used in offline environment when a speaker goes online to post (cf. Allport, Styles, & Hsieh, 1994). How speakers respond to these differences by adopting CS is beyond the scope of one single paper. But we should bear in mind these possibilities and keep the field open. In addition, the findings of this study are drawn upon the analyses of CS practices by L2 learners. How well they can be applied to other types of bilingual speakers is not known yet. Future research on this could yield useful findings to further sociolinguistic and cognitive studies of CS as well as L2 acquisition studies. For example, how an added language gains and loses social meanings as L2 performance progresses and how cognitive mechanisms respond to, interact with and perhaps are shaped by social environments.
It is fruitful to examine CS use across different interaction modes. What is revealed is not only a description of how CS frequency fluctuates as speakers converse in different modes, it highlights and draws our attention to the dynamic interactions between social, cognitive, and situational factors under the surface of linguistic practice. By doing so, we gain insights of what CS really means to individual speakers and how speakers mediate between social and individual realities via linguistic means.
Footnotes
Acknowledgements
I would like to thank Professor Devyani Sharma of the Department of Linguistics at Queen Mary University of London, and the two anonymous reviewers for their valuable comments and advice. I would also like to thank all the subjects for their time and participation in the study.
Declaration of conflicting interest
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was within the author’s PhD project: English in the Chinese discourse of Chinese professionals in London: Register and social factors, which was funded by the Chinese Scholarship Council.