1. Introduction

This essay will define and critically explore code-switching. Different studies and models of code-switching will be discussed accordingly. Since both sociolinguistic and psycholinguistic approaches can be used to account for contexts and grammatical constraints on code-switching， this essay will only focus on the sociolinguistic approach.

2. Bilingualism and code-switching

Bilingualism is a phenomenon that can be drawn on a continuum. Due to its vitality， there seems to be no clear-cut archetypal bilingualism. Definitions of bilingualism range from a minimum proficiency level in a second language to symmetrical competence in two languages （Myers-Scotton， 2006）. Contrastively， Diebold （1964） proposed a minimal requirement for bilingualism： a person can be recognized as bilingual with limited knowledge in both languages； competence of uttering complete and meaningful languages is not necessary （Romaine， 1995）. However， Diebolds （1964） approach does not hold since many European countries are linguistically related， and people from different countries might to various extents “know” languages of other European countries. As such， balanced， or at least near-balanced， bilinguals tend to be chosen for code-switching studies.

3. Definition of code-switching

One of the main aspects of bilingualism is code-switching. Code-switching is the alternation between languages or varieties by bilinguals （Milroy and Muysken， 1995）. However， such definition seems to oversimplify the complex nature of code-switching. Code-switching， influenced and discussed by Blom and Gumperzs （1972） and Poplacks （1980） studies， was readapted under a scenario involving bilinguals having symmetric competence in two varieties or languages.

In spite of Blom and Gumperzs and Poplacks definition of code-switching， this phenomenon remains as fuzzy-edged as it is （Gardner and Chloros， 1995） because， after decades of research and discussion， the difference between code-switching， code-mixing and lexical borrowing are still unclear. Kachru （1978） equates code-mixing with language dependency and code-switching with language manifestation （Dabène and Moore， 1995）. In simpler terms， code-mixing happens when the speakers are not balanced bilingual （i.e. not fully competent in both languages）， while code-switching occurs when the speakers have symmetrical competence in both languages.

4. Types and patterns of code-switching

Two divisions of code switching are diglossic code-switching and bilingual code-switching （Myers-Scotton， 1993； Sayahi， 2014）. Diglossic code-switching is excluded from the sphere of code-switching by Romaine （1995）. Diglossia was defined as two varieties which are closely related， preferably two similar dialects （Romaine， 1995）. However， the diglossic code-switching was valued as part of the general code-switching in Myers-Scottons （1993） and Sayahis （2014） African studies. Diglossia was classically discussed in situations where two closely related languages are used in a speech community. One serves as High （H） variety， for example languages used by sermons and news readers broadcasting national or international news； and one is named as Low （L） variety referring to language use in the informal settings like home （Ferguson， 1959）， e.g. Kikuku and Swahili， the indigenous language and the lingua franca in Kenya （Myers-Scotton， 1993）. However， it is now extended to refer to any two languages that are not necessarily closely related （Fishman， 1967； 1970）， even typologically deviant languages that have this H/L social and functional distribution， e.g. Francophone versus Anglophone in Quebec of Canada （Heller， 1995）. Ramats （1995） discussion about code-switching was also based on Italian dialects and standard Italian. It is obvious that diglossic code-switching seems to be recognised validly as a part of code-switching. Therefore， diglossic code-switching will be discussed in respect of such in this essay.

Three types of code-switching can be identified according to Poplack （1980）： tag-switching， inter-sentential and intra-sentential switching. Tag-switching involves adding a tag in one language such as you know into a complete utterance in another language. Tags appear to have minimal syntactic constraint and can be easily inserted without violating syntactic rules （Romaine， 1995）. This type of code-switching thus can be suggested to require less symmetrical language competence. Inter-sentential switching is switching at a clause or sentence boundary， where one clause or sentence is expressed in one language or another： this requires more balanced language competence （Romaine， 1995）. Examples may be found in Panjabi/English discourse， Spanish/English speech， French/English， Chinese/English speeches （Romaine， 1995； Poplack， 1980； Heller， 1982； Li， 1998）. Intra-sentential switching is code-switching within a sentence. This arguably raises syntactic risks， as it might violate syntactic rules. Analogously， intra-sentential code-switching requires the highest proficiency level in both languages （Romaine， 1995）.

5. Factors affecting code-switching （the Macro level versus the Micro level）

Many models from linguistic， psycholinguistic and sociolinguistic points of view were proposed to account for the manifestation， constraints and motivations for code-switching. On one hand， linguistic， especially syntactic and psycholinguistic models were proposed to specify constrains and explain why certain manifestations surface （Romaine， 1995； Myers-Scotton， 2006）. On the other hand， two models associating with social motivation and pragmatic purposes - the Markedness model proposed by Myer-Scotton （1993） and the Sequential model proposed by Auer （1995） - are used to explain social and individual factors resulting in code-switching.

Two levels of influence could be observed and are discussed by several researchers： the macro level and the micro level. The macro level refers to overall social factors like migration and language policy， which affect code-switching in general code choice and shifting （Milroy and Muysken， 1995）. Micro level focuses on code-switching at conversational level between interlocutors （Ramat， 1995）.

Presumably， social factors such as language prestige， domains of conversation and personal identity would impact on both macro and micro linguistic levels. Patterns of code-switching are also predictable using the Markedness model. Code-switching is categorized as either marked or unmarked. Several domains where code-switching is necessary are proposed， such as home， neighbourhood， work and education （Fishman， Cooper and Ma， 1971； Myers-Scotton， 1993）. Within each domain， the Markedness model predicts that use of certain language within its respective domain is unmarked： for example， speaking English/Swahili in a government office between colleagues talking about formal issues in Kenya （Myers-Scotton， 1993）. On the contrary， speaking local dialects in the formal setting will be marked， because such code choice appears to be out of context. According to this model， incidents of code-switching should be predictable and the code-switching trend is seemingly obvious. However， several observations contradict the certainty of prediction by the Markedness model. Although the code-switching trend seems to be predictable at macro level， individual choice at micro linguistic level （the code-switching within each dialogue） remains subjective. It is even argued that code-switching at micro linguistic level is unpredictable （Ramat， 1995； Auer， 1995； Milroy and Li， 1995）. Code-switching can differ from individuals and it is not necessarily marked. In order to better explore and count for code-switching at micro linguistic level， readapting Gumperzs （1982） discourse analysis approach at conversational level， Auer （1995） have proposed the Sequential model， which focuses more on talk-interaction level and pragmatic purposes than on the large social scale.

6. Reasons of diglossic code-switching at macro linguistic level

Reasons for diglossic code-switching provide good examples on how general social factors such as social setting， language prestige and language policy affect code-switching. These reasons are discussed as follows to illustrate social influence on code-switching at macro level. In diglossic code-switching， H variety is presumably used more often in the formal setting whereas L variety is used more frequently in the casual setting. For example， in Singapore， English is used in education and policy implementation， while other official languages like Malay and Tamil are used elsewhere in spite of the racial harmony proposal by the Singapore government， influenced by the British colonization （Clammer， 1998）.

Language prestige also affects speakers choice of varieties. The language or variety that is believed to have overt prestige is supposedly used more in the public setting and connotes higher social identity. For example， the rhoticity pronunciation was valued as a more middle-class related preference as opposed to that of working-class in Labovs （1966） study. However， it cannot be argued that the covert variety carries a negative attitude. Despite massive use in the informal setting， Singapore Colloquial English was rated as a solidary， friendly and local-bounded variety （Cavallaro， Ng and Seilhamer， 2014）. Therefore， language prestige does affect language choice but might not deploy negative attitudes.

So far， it seems that code-switching is affected by how people or the society value one language. The consequent question then asks which factor determines this evaluation system. Language policy， the invisible hand manipulating language status， can be one of the answers. In Redingers （2010） discussion on the multilingual system of Luxemburg， Luxembourgish is favoured by the educational system and Luxembourgish nationals， because Luxembourgish symbolizes their unique identity. However， French has been still treated as a more prestigious language-local and emigrated people in Luxemburg， invariably， express positive attitudes towards the use of French beyond the educational system. This supports the notion of the power of language policy on influencing language status. However， it is rather vague to argue it can change peoples deeply built-in attitudes towards using that language. Choice of code-switching thus is partially but yet not fully accounted for by language policies.

7. Reasons of code-switching at micro linguistic level

Differing from the large influence of code-switching in the general trend， social factors seemingly impact less on the micro linguistic level. Though Milroy and Li （1995） have attempted to make more sensible connections between social factors such as age and generation to bilingual Chinese/English speakers at Tyneside， other scholars like Ramat （1995） and Gardner-Chloros （1995） argued that when and how each individual code switches is almost unpredictable and more inclined to subjective choice.

In spite of impact from social networks， code-switching is proposed as a contextualisation cue in dialogues， because code-switching serves as a process of inference， indicator of disagreement and tool of language negotiation between interlocutors in the Sequential model （Auer， 1995； Milroy and Li， 1995）. For example， the Chinese/English bilingual teenagers chose to speak English with their parents when they disliked the proposed idea （Milroy and Li， 1995）.

Moreover， code-switching is used as a language negotiation tool to gain more than one social identity in the exchange （Auer， 1995； Myers-Scotton， 1993）. For example， in a local market in a city of Kenya， bargaining in local dialect would help customers get cheaper groceries， but speaking English or Swahili would not （Myers-Scotton， 1993）.

8. Code-switching， a symbol identity

No matter the social factor， language policies and pragmatic purpose affect peoples choice of language. It can be seen as a symbol of personal and social identity. People use their distinctive language to define in-group members and further define their identity （Myers-Scotton， 1993； Milroy and Li， 1995）. This can be illustrated in Lis （1998） finding in the third generation of Chinese immigrants， the British-born Chinese. This group uses English-dominant bilingualism to express their hybrid identity. Furthermore， code-switching in commercials indicate its social identity that people may perceive in different social settings. Piller （2001） has found that nearly 70% of German TV adverts blended in another language， with English the favourite choice； probably because English to some extent represents modern， sophisticated and international orientation. These examples indicate that code-switching may portray different social identities according to social setting. Code-switching as a symbol of identity has seemingly become a relatively new trend and Auer （2005） suggests that new models must be proposed to account for this new approach in code-switching.

To conclude， code-switching is a complex phenomenon that hardly has one clear-cut definition. Two generic levels， the macro and micro linguistic level， can be proposed for where code-switching happens and how it would be influenced. Last but not the least， code-switching is a representation of how a community and a person see themselves. Despite myriad studies in code-switching， it is still a field worth further investigation.

References：

[1]Agnihotri，R.K.Crisis of identity： the Sikhs in England[J].New Delhi：Bahri，1987.

[2]Agnihotri，R.K.Code-mixing among Sikh children in Leeds，U.K.Paper presented at Sociolinguistics Symposium IX[D].University of Reading， April，1992.

[3]Auer，P.The pragmatics of code-switching：a sequential approach. In L.Milroy and P.Muyske，（Eds）.One speaker，two languages[M].UK： Cambridge University Press，1995.

[4]Auer，P.A postscript：code-switching and social identity. Journal of pragmatics，2005，37（3）：403-410.

[5]Blom，J.P.and Gumperz，J.J.Social meaning in linguistic structures： code-switching in Norway.In J.J.Gumperz，D.Hymes （eds）. Directions in sociolinguistics.The ethnography of communication[J].New York： Holt，Rinehart & Winston，1972.

[6]Bloomfield，L.Language[J].New York：Holt，1933.

[7]Cavallaro，F.，Ng，B.C.，&Seilhamer;，M.F.Singapore Colloquial English：Issues of prestige and identity[J].World Englishes，2014， 33（3）：378-397.

[8]Clyne，M.G.Constraints on code-switching： how universal are they [J].Linguistics，1987，25：739-764.

[9]Clammer，John.Race and State in Independent Singapore 1965-1990[J].Brookfield：Ashgate，1998.

[10]Dabène，L.and Moore，D.Bilingual speech of migrant people.In L. Milroy and P. Muyske，（Eds）.One speaker，two languages[M].UK：Cambridge University Press，1995.

[11]Diebold，A.R.Incipient bilingualism.In Hymes，D.（eds）.Language in culture and society[J].New York：Harper & Row，1964.

[12]Fehlen，F.Luxembourg，a multilingual society at the Romance/Germanic language border[J].Journal of multilingual and multicultural development，2002，23：80-97.

[13]Ferguson，C.Diglossia[J].Word，1959，15：325-40.

[14]Fishman，J.A.Bilingualism with and without diglossia； diglossia with and without bilingualism. Journal of Social，1967，23：29-38.

[15]Fishman，J.A.Sociolinguistics： A Brief Introduction[J].Rowley， MA：Newbury House，1970.

[16]Fishman，J.A.，Cooper，R.L.and Ma，R.（eds）.Bilingualism in the Barrio[M].Bloomington：Indiana University Press，1971.

[17]Gal，S.Language Shift： Social determinants of linguistic change in bilingual Austria. New York：Academic Press，1979.

[18]Gardner-Chloros，P.Code-switching in community[J].regional and national repertoires：the myth of the discreteness of linguistic systems， 1995.

[19]In L.Milroy and P.Gumperz，J.J.Discourse Strategies[M].Cambridge： Cambridge University Press，1982.

[20]Heller，M.Negotiations of language choice in Montreal.In J. Gumperz （eds）.Language and social identity[D].Cambridge and New York：Cambridge University，1982.

[21]Heller，M.Code-switching and the politics of language.In L. Milroy and P.Kachru， B. B. Code-mixing as a communicative strategy in India[J].International dimensions of bilingual education， 1995，1978，107-124.

[22]Labov，W.The Social Stratification of English in New York City[J]. Washington：Center for Applied Linguistics，1966.

[23]Li，W.Banana split？Variations in language choice and code-switching patterns of two groups of British-born Chinese in Tyneside.In R. Jacobson.（eds）.Codeswitching worldwide.Trends in linguistics：studies and monographs 106[J].Berlin and New York：Mouton de Gruyter， 1998.

[24]Milroy，L.and Li.W.A social network approach to code-switching： the example of a bilingual community in Britain.In L.Milroy and P. Muyske，（Eds）.One speaker， two languages[M].UK：Cambridge University Press，1995.

[25]Muyske，（Eds）.One speaker，two languages[M].UK：Cambridge University Press.

[26]Muyske，（Eds）.One speaker， two languages[M].UK：Cambridge University Press.

[27]Haugen，E.The Norwegian language in America：a study in bilingual behavior[M].Philadelphia：University of Pennsylvania Press，1953.

[28]Milroy，L.and Muysken，P.One speaker， two languages[M].UK：Cambridge University Press，1995.

[29]Myers-Scotton，C.Social motivations for codeswitching：evidence from Africa[M].Oxford：Clarendon Press，1993.

[30]Myers-Scotton，C.Multiple voices：an introduction to bilingualism [J].Hong Kong：Blackwell，2006.

[31]Piller，I.Identity constructions in multilingual advertising[J]. Language in Society，2001，30：153-186.

[32]Poplack，S.Sometimes I will start a sentence in Spanish Y TERMINO EN ESPAN?L：toward a typology of code-switching[J]. Linguistics，1980，18：581-618.

[33]Ramat，A.G.Code-switching in the context of dialect/standard language relations.In L. Milroy and P. Muyske，（Eds）.One speaker， two languages[M].UK：Cambridge University Press，1995.

[34]Redinger，D.Language Planning and Policy on Linguistic Boundaries：the case of Luxembourgish.In Miller，R.M.（ed.）2010. Marginal dialects：Scotland，Ireland and Beyond[J].Aberdeen：Forum for research on the languages of Scotland and Ireland，2010：90-106.

[35]Romaine，S.Pidgin，creoles，immigrant and dying languages[J].In Dorian（eds），1989：369-383.

[36]Sayahi，L.Diglossia and Language Contact：Language Variation and Change in North Africa[M].Cambridge University Press，2014.

【作者簡介】曾熙然，昆明医科大学海源学院。