Various shortcomings of the symbolic approaches, employed in both Gumperz’s and Myers-Scotton’s work, have led other researchers towards a shift in focus in their analysis of bilingual interactions. Auer (1984: 3) argues against the use of extra-linguistic macro-social categories such as speaker identity (e.g. age, gender, ethnicity etc.) in the interpretation of bilingual conversations. In fact, Auer (1984:
4) emphasises that language alternation must be investigated from an interactional perspective and, therefore, employs a Conversation Analysis (CA) approach for the interpretation of conversational code-switching. CA consists of an analytical framework focusing on the ways in which speakers locally express and interpret meanings on a turn-by-turn basis; particular attention is paid to the sequential positioning of conversational moves in the discourse (Milroy and Gordon, 2006:
218). Auer (1984: 3) strongly advocates a sequential turn-by-turn analysis of code-switching as he believes that each individual code-switch is closely tied to the specific context in which it appears. Context-bound factors are defined as and restricted to information which can be shown to be used by the conversationalists themselves (Auer, 1984: 6). In other words, Auer argues against symbolic approaches such as Myers-Scotton’s Markedness Model (2.4.3) or a ‘we code/they code’ distinction (2.4.2), as these rely on the social connotations attached to different codes. Researchers working within the CA paradigm refrain
engage in a turn-by-turn analysis of language choices in order to throw light onto the mechanisms which conversation participants themselves employ in the interpretation of the meanings conveyed by language choices (Li Wei, 2005b:
380). The CA approach is ‘sequential’ in nature due to its emphasis on the construction of meanings in the sequential development of a conversation (Auer, 1984: 78).
In his study of code-switching among Italian guestworkers in Germany, Auer (1984: 12) employs a CA approach and largely distinguishes between discourse-related transfer and code-switching on the one hand, and participant-discourse-related transfer and code-switching on the other hand. Auer (1984: 12) defines the different types of language alternation (transfer and code-switching) and their functions (discourse-related and participant-related) through the following questions:
i) Is the language alternation in question tied to a particular conversational structure for instance, a word, a sentence, or a larger unit (transfer), or is it tied to a particular point in conversation (code-switching)?
ii) Is the language alternation in question providing cues for the organisation of the ongoing interaction (i.e., is it discourse related), or about attributes of the speaker (participant related)?
Transfer typically happens on the word level and the return into the first language can be easily predicted by participants. Code-switching, on the other hand, consists of a renegotiation of the language of the interaction, and the return into the first language cannot be predicted as the language alternation is not tied to a particular unit such as a word (Auer, 1984: 78).
Discourse-related language alternation acts as an organisational tool and structures the conversation in relation to procedures such as turn-taking and topical cohesion (Shin & Milroy, 2000: 370). Discourse-related language
alternation can change the footing of an interaction (Auer, 1984: 32) which has been described as ‘the conversationalists’ alignments toward people, topics and actions’ (Cromdal & Aronsson, 2000: 435). A change of footing can be achieved, for example, through a re-definition of the participant constellation where the language alternation functions as a mechanism for next-speaker-selection in a conversation. Alternatively, the footing of an interaction can be altered through
‘sequential hierarchisation’. Auer (1984: 32) defines this process as the
‘sequential chaining of utterances’ which consists, for example, of marking the beginning or the end of a side-sequence. Participant-related language alternation, on the other hand, is largely influenced by the language preferences (preference-related language alternation) and linguistic competence of participants (competence-related language alternation) (Auer, 1984: 49-50). Whereas the former consists of a speaker taking into account other participants’ individual language preferences, the latter is characterised by the speakers adapting their own code choice to the language abilities of their interlocutors (Shin and Milroy, 2000: 362).
Auer’s CA approach and its emphasis on a turn-by-turn analysis of the sequential development of bilingual interactions constitutes the basic analytical tool for much code-switching research. In his study of bilingual groupwork among Danish-Turkish pupils, Cromdal (2003: 58) describes his analytical approach as
‘grounded in the theoretical conception of social interaction as an inherently dialogic endeavour, in which mutually recognisable interactional projects are accomplished through sequentially organised trajectories of action’. A turn-by-turn analysis reveals the ways in which bilingual children (Turkish-Danish) employ different codes in order to demarcate social relationships. The participants establish the use of different codes as a means of creating social distance as well as alliances among themselves. During a task focusing on the creation of a storyline, Selma rejects Erol’s suggestion for how the story should unfold. In reaction to Selma’s rejection, Erol code-switches from Danish into Turkish.
himself from Selma and to signal his disagreement. In addition, Cromdal (2005:
344) demonstrates how two girls in an English-Swedish bilingual school create a
‘bilingual order of conduct’ in which English is employed when direct references are made to the written text through procedures such as quoting, dictating and spelling, and Swedish functions as a medium of communication for the remainder of the interactions. Only a turn-by-turn interactional analysis can reveal how language alternation functions as a structuring device in conversations. By adapting Auer’s CA approach to bilingual conversation, Cromdal (2003, 2005) is able to demonstrate how speakers build social relations as well as structure their interactions on a local level.
The CA approach has, however, been seriously criticised due to its preoccupation with transcription details (Li Wei, 2002: 174). A detailed focus on transcription conventions can be regarded as unnecessary as all transcripts are subjective and, therefore, favour the application of some theories over others (Li Wei, 2002: 174).
During the transcription of recorded interactions, researchers take decisions regarding the amount of detail (e.g. nonverbal cues, prosody, silence) that will be included in a transcript (Li Wei, 2002: 174). Consequently, transcripts do not constitute exact representations of interactions. The decisions taken during the transcription can have an impact on the analysis of the data at a later stage. This introduces a layer of subjectivity into CA approaches which claim to base their interpretations of language alternation practices solely on context-bound cues.
Moreover, Myers-Scotton and Bolonyai (2001: 4) claim that CA offers a ‘flat explanation of choice’ due to its neglect of both the social background of speakers (age, gender, ethnicity) and the ‘socio-psychological associations and, therefore, the social messages carried by one linguistic choice rather than another’. Auer (1998: 13), however, argues that CA does not disregard the possibility of a connection between macro-social variables and specific linguistic choices in conversation. Conversation analysts simply refuse to base their interpretations on an assumption that certain codes are closely associated with certain identities (Li Wei, 2005b: 382). If, however, the analyst can demonstrate how speakers
actively construct these identities through a turn-by-turn analysis of a bilingual conversation, such extra-linguistic social explanations are supported by the CA approach. Such an analysis combines both elements from the symbolic and sequential approaches to code-switching.
In order to be able to incorporate both approaches into a single model the distinction between ‘brought-along’ and ‘brought-about’ meaning in interaction needs to be clarified. The symbolic approaches to code-switching (2.4.2 and 2.4.3) focus on ‘brought-along’ meaning in interaction as the languages which are spoken in a given interaction are seen to bring along social meaning (Mendoza-Denton & Osborne, 2010). ‘Brought-along’ or symbolic approaches to code-switching are based on the indexicality of linguistic varieties. Bucholtz and Hall (2005: 594) claim that ‘in identity formation, indexicality relies heavily on ideological structures, for associations between language and identity are rooted in cultural beliefs and values […] about the sorts of speakers who (can or should) produce particular sorts of language’. In other words, ‘brought-along’ approaches are based on the assumption that languages are associated with particular identities or social connotations and that speakers exploit these associations in order to convey meaning in their interactions.
‘Brought-about’ approaches, on the other hand, do not rely on any social or symbolic meanings attached to languages but focus on the interactional nature of code-switching. Auer’s sequential analysis of language alternation with its focus on the creation of local meaning constitutes an example of the ‘brought-about’
approach. Li Wei (2005b: 387) draws attention to the benefits of combining symbolic and sequential analyses as a ‘dual-level approach […] would help to extract factors that can deepen our understanding of the way bilingual speakers use their linguistic and interactional resources and achieve a richer, more interesting, but still relevant (to the participants of the interaction) explanation.’ A combined analysis can account for the ways in which bilingual speakers create
meaning through their code-switching practices on both local and social levels.
An example of a dual-level approach will be discussed below.
Bani-Shoraka’s (2008) analysis of Persian-Azerbaijani code-switching in Iran incorporates elements from both ‘brought-along’ and ‘brought-about’ approaches to code-switching. She supplements her sequential analysis of conversational code-switching with ethnographic information on the linguistic ideologies of the society. In Iran, Persian constitutes the sole official language and benefits from high national prestige. Azerbaijani is associated with local in-group prestige but is stigmatized in Iranian society on a national and official level (Bani-Shoraka, 2008: 17). In her analysis of an interaction between two speakers, Sara and Darya, Bani-Shoraka (2008: 25) demonstrates how Darya switches from Azerbaijani to Persian in order to convey meaning both locally and globally. Both speakers show a preference for Azerbaijani throughout the conversation. Darya has repeatedly asked Sara for permission to get involved in her business investment. After Sara has avoided giving an answer over several turns, Darya switches the medium of interaction and repeats her request in Persian. In this instance, the use of Persian sequentially contrasts with the previous turn and therefore functions as a signalling device on a local level; simultaneously, the use of Persian helps the speaker to convey a sense of authority and competence as Persian is widely associated with seriousness and competence (Bani-Shoraka, 2008: 25). By switching to Persian Darya temporarily gains increased attention as she diverges from the established medium of communication. This switch can be locally interpreted as her attempt to show her irritation with Sara’s behaviour. On a global level, she has switched to the language which is widely associated with authority and power. The local and global effects of Darya’s code-switch both contribute to the forcefulness of Darya’s request to be allowed to cooperate in Sara’s business investment. A dual-level approach can, therefore, provide a richer understanding of a speaker’s motivation for code-switching.