the immediate conversational functions to help forge a hybrid social identity
unique to this group’ (Sayahi, 2004: 377). In the case of the Arabic-French
group, however, he used conversational analysis to argue that in switching
from Arabic to French speakers ‘continuously reinforce their socioeconomic
identity by using the symbolically dominant language’ (Sayahi, 2004: 380).
However, to our knowledge, research on CS and identity have not previously
related self-perceptions of identity to CS patterns in a quantitative way,
something which we attempt to achieve here.
Another speaker-based variable which has been studied in relation to CS
patterns is that of social network. Milroy & Wei (1995: 155) find that ‘while
network interacts with a number of other social variables such as generation,
gender and occupation, it is capable of accounting more generally than any
other single variable for patterns of CS language choice’. Networks of
Chinese-English bilinguals in Newcastle, UK, were quantified to indicate the
proportion of Chinese contacts and these were linked to observed language
choices. Overall, they found that the higher the proportion of Chinese con-
tacts in a speaker’s social network, the more likely they were to use Chinese
(as opposed to Chinese and English) with bilingual interlocutors.
The role of community characteristics in CS behaviour
A community approach to CS, in contrast, is illustrated by work by
Muysken (2000) and related studies inspired by this, such as Deuchar et al.
(2007) and Stell (2009). Muysken (2000) makes no assumptions about uni-
formity in patterns of CS; on the contrary, he suggests that there are three
main CS pattern types: insertion, alternation and congruent lexicalization.
The insertion pattern is similar to that proposed by Myers-Scotton (2002: 8)
as representing ‘classic CS’ in that other-language material is inserted into a
morphosyntactic frame from a different language; alternation involves the
alternation of two morphosyntactic frames, and congruent lexicalization
involves the use by two closely related languages of what is essentially the
same morphosyntactic frame in which items from the two languages can be
inserted virtually at random. Although a Labovian approach would seek to
relate these patterns to the characteristics of individual speakers, Muysken
suggests that the selection of one pattern rather than another will be influ-
enced by characteristics of the communities in which the speakers are located
rather than of the individual speakers. He predicts, for example, that inser-
tion will be more common in colonial settings, alternation where there is a
tradition of language separation, and congruent lexicalisation where the lan-
guages are typologically similar and have roughly equal prestige (cf. Table 6.1
below). Testing this kind of prediction, Deuchar et al. (2007) find that the
dominant insertion pattern in the Welsh-English and Taiwanese Tsou-
Mandarin data arguably reflect postcolonial settings. Similarly, Stell (2009)
argues that the high quantity of alternations in the White Urban bilingual
speakers of Afrikaans and English compared with that of congruent
120 Part 3: Bilingual Language Use