Bilingualism and language shift in Western Cape communities

This paper considers a number of pertinent sociolinguistic aspects of a distinct process of language shift recently noted in some historically Afrikaans first language (L1) communities established in the Cape Metropolitan area. Particularly, it considers qualitatively how a number of families made deliberate choices to change the family language from Afrikaans L1 to English L1. It elaborates on an exploratory study undertaken in 2003, adding data collected in 2008 and 2009, investigating linguistic repertoire and language choice in a number of families where there has been contact between English and Afrikaans over a number of generations. The aim, eventually, is to characterise the nature of the perceived process of language shift. The paper considers how widespread use of both English and Afrikaans in communities that until recently were predominantly Afrikaans, impacts on linguistic identities. It reports on structured interviews with members of three generations of families who currently exhibit English-Afrikaans bilingualism where members of the younger generation are more fluent in English. It finds that there is evidence of language shift, it reports on the circumstances that motivate such shift, and concludes that the third generation presents either a monolingual English identity where Afrikaans has a decidedly second language status, or a strong English-dominant


Introduction
This paper elaborates on a research project of limited scope undertaken in 2003 1 where a process of language shift was traced across three generations in three families residing in various parts of the Cape Metropolitan area.It has already been noted that South Africa has, since the early 1990s, been witnessing dramatic social transformation which also has an effect on linguistic identities of individuals and communities.In the South African context, languages are markers not only of personal, but also of social and political identities.This can be related to the particular distribution of 11 official languages, two of which are former colonial languages, and nine of which are indigenous African languages that were afforded no official recognition until 1994.Here, considering more recent data collected in 2008 and 2009 2 , I shall report on the linguistic repertoires and language choices of members of historically Afrikaans "coloured" communities in the Western Cape.I will consider what motivates the perceived processes of language shift from Afrikaans as first language (L1) to English as L1 in a relatively short time, and I will attempt to determine what the characterising features of such a process of language shift may be.In investigating what motivates the language shift, I shall also consider what makes this process unique in comparison to recorded features of language shift elsewhere.I shall present an analysis of a number of recorded interviews where respondents were asked specifically to reflect on and try to motivate family language choices and patterns of language use of the past ten to twenty years.In all instances the informants in this study were from families where, across the last three generations, there has been a considered shift from Kaapse Afrikaans to English as the home language.The families reside in six different communities around the Cape Metropolitan area, where the majority of the residents would be classified as "coloured" in terms of the contentious racial classification that, in spite of much well argued protestation, is still evident in texts and in societal structures across the country.As it is particularly in these communities that the process of language shift is most notably evidenced, the interest here is in tracing patterns of language choice and language use across three generations in a number of social settings.Much work has been done elsewhere on language shift in displaced communities and across generations (cf.Le Page and Tabouret-Keller 1985;Wei 1994;Siegel 1987).Such work provides a framework for considering how to characterise the process of language shift and to suggest what the implications are for linguistic identities in the particular language communities.Recently collected data will be compared to the data collected in 2003 and will assist in checking for the possibility of stabilisation or change in the patterns recognised in the first study.
Using ethnographic methodology, the 2003 data were collected by means of recording interviews with members of Afrikaans-English bilingual families.This method was chosen because, in qualitative research of this kind, no other approach would give more direct insight into the choices of parents regarding the first language of their children.At the time, three women of three different families, each woman representing one of three generations (grandparents aged 60+, parents aged 35+, children aged between 10 and 23+), were invited to focused family group discussions where their language choices in various domains were topicalised.The women were asked to motivate such choices, to describe communicative practices in the particular domains and to disclose their personal linguistic preferences.Eventually the preferences and practices of more than only the three identified members of each family were recorded as it became clear during the discussions that not only interactions in the nuclear family were important; in fact, the networks in the extended families were of apparent significance in patterns of language use that emerged.On every occasion where an interview had been scheduled with the women, their husbands, fathers, brothers, sisters and in-laws gradually joined in and took part in the discussion animatedly and without prompting.All of this was captured in the recordings and, with due consent from the participants, was accepted as authentic, useful data.
The 2008 and 2009 data to which this paper refers, were collected in a similar ethnographic manner.During 2008, in research conducted for a Master's thesis investigating the registration of secondary school learners from Afrikaans L1 families in the English L1 classes of an historically Afrikaans school (cf.Farmer 2008), a number of interviews were recorded of which some excerpts have been informative to this study.In 2009, in order to elaborate and verify (or falsify) earlier findings, structured interviews with members of 10 more families were recorded.Here the invited interviewees were 2 nd generation mothers and one of their young adult children; however, as in the 2003 data collection, from time to time other family members also took an interest in the discussion and if they volunteered, were encouraged to join in.
During 2003 and again in 2009, the interviews were informally structured.Discussion was prompted and directed by means of questions (see Appendix A) intended to elicit information that would indicate specific language choice and general patterns of language use in families where there is a repertoire of at least Afrikaans and English.In 2009, the questions as to what motivated the obvious transfer from Afrikaans to English in the family were perhaps more directly articulated than in the 2003 data collection.The aims throughout were to determine -which language members of three generations of each family regard as their first language; -which language each generation had (or currently has) as their language of schooling; -which language each generation preferred (or would prefer) as the language of schooling for their children; -which language(s) are used in close family interactions; -what patterns of code-switching between Afrikaans and English are apparent; -what reasons are given for language shift from one generation to the next at the point where evidence of such shift becomes apparent; and -whether perceived language shift from one generation to the next is accompanied by improved levels of bilingual proficiency.

Field of inquiry
The research is situated in the field of sociolinguistic inquiry into language maintenance and shift.According to Fishman (1972:76) the study of language maintenance and shift responds to our curiosity and concern about the kinds of language behaviour in which "languages (or language varieties) sometimes displace each other, among some speakers, particularly in certain interpersonal or system-wide interactions".Fishman finds that scholars in this field take an interest in "the relationship between change (or stability) in language usage patterns, on the one hand, and ongoing psychological … or cultural processes, on the other hand, in populations that utilize more than one speech variety for intra-group or for inter-group purposes".To date, no calculation has been made of how many Western Cape families or what numbers of speakers can be counted in as ones where there has been a decided shift from Afrikaans to English.The data used here probe what is perceived to be a very strong and largely irreversible trend.However, in personal communication with a colleague resident in a relatively affluent, now racially mixed neighbourhood, it was pointed out that this shift is not as marked in the poorer Afrikaans L1 "coloured" communities where there are lower levels of formal education and higher levels of unemployment in the adult population.A follow-up study that will systematically check the changes that are taking place in habitual language use of the speech community in focus here, is envisaged as a next step in this research project.The intention in this paper is to investigate how members of the community themselves articulate their own choice for displacing one language with another in the intimate sphere of the family.This will assist in answering the question as to how language shift of such a rapid kind is possible in a community where there is no recent history of migration.

1.2
Definitions and terms of reference This section introduces central concepts, gives the terms to be used in referring to such concepts and, for the purposes of this study, defines each concept.Gumperz (1971:101) defines a linguistic community as "a social group which may be either monolingual or multilingual, held together by frequency of social interaction patterns and set off from the surrounding areas by weaknesses in the lines of communication" with other social groups in those areas.The linguistic community whose patterns of language use are of interest here is one that currently can be characterised as an Afrikaans/English bilingual one, where there appears to be displacement of Afrikaans as the dominant language in favour of English.In the Western Cape this is a community where "weaknesses in the lines of communication" with other social groups in the area can largely be ascribed to social isolation resulting from racial differentiation originating as far back as the 17 th century when distinctions between indigenous peoples, imported slave workers and Dutch colonists were already pronounced.Social divisions of race and class that developed during the 19 th and early part of the 20 th century, became more explicitly formalised and entrenched after the 1948 government legislated apartheid.Such social divisions were decisive in articulating the L1 speakers of Kaapse Afrikaans as a linguistic community.Language shift, language maintenance, language death and intergenerational switching as they are defined in Fasold (1984) are pertinent concepts that require explanation.
Language shift occurs where "a community gives up a language completely in favour of another one" (Fasold 1984:213).In a study that focuses on the shift from indigenous Australian languages to English, this process is referred to as one where the possibility exists that one language will give way "partially or completely" to another.The chosen working definition of the NWT Literacy Council (1999:3,10) identifies language shift as a process of change across generations: "If elders still speak their traditional language but their grandchildren do not, then language shift has occurred."Many studies of language shift coincide with histories of geographical displacement such as Scottish emigration to Nova Scotia in the 18 th and 19 th century or Indian emigration to South Africa in the late 19 th and early 20 th century.There are, however, also histories of language shift where other social conditions had a greater determining effect.Consider, for example, the status and power of the coloniser language (as among Australian Aborigines) or embarrassment at association with the community of origin (as among German communities in the USA during and after World War II) that resulted in shift towards English in each context.
Language maintenance, in contrast to language shift, occurs where "the community collectively decides to continue using the language or languages it has traditionally used" (Fasold 1984:213).Language death, a possible result of complete language shift, occurs "when a community shifts to a new language totally so that the old language is no longer used" (Fasold 1984:213).Related to questions of maintenance and shift is the occurrence of intergenerational switching where "bilingual parents pass on only one language to their children" (Fasold 1984:238).
Different definitions of bilingualism abound, ranging from ones that determine bilingual speakers as ones that have complete command of two languages to ones that determine bilinguals as speakers that have L1 proficiency in one language and minimal proficiency in a second language (L2).Adler (1977) captures this spectrum in his understanding of community bilingualism as "the complete or less complete, command of at least two languages, speaking, hearing, writing and reading them."The linguistic community in focus here is one where the majority of its members are in command of at least two languages, notably Afrikaans and English, with speaking, listening, reading and writing proficiency, even if not in equal measure, in both.It is possible that the literacy skills of some of the older generation are less advanced, particularly in the language identified as their L2.Achievement of a particular level of literacy was not directly relevant to our inquiry and was, therefore, not checked.
As is typical in bilingual communities, the data exhibit high incidence of code-mixing, "a switching of codes within an utterance without a topic change" or other easily observable social prompts.Code-switching, "a conversational strategy used to establish, cross or destroy group boundaries; to create, evoke or change interpersonal relations with their rights and obligations" (Gal 1988:247), is similarly prevalent.Examples of such code-mixing and codeswitching are given in (1) to ( 4) below (where English translations of Afrikaans words/phrases/utterances are provided in square brackets, and code-mixing and codeswitching are indicated by italics): (1) 'Cause I felt that with my, with my daughter-in-law, Doreen, when they were courting, en / dan praat sy nie 'n woord nie.Seeing that the focus here is not on micro-linguistic structures, I did not calculate how often and at which junctures in the conversation such code-switching occurred.Participants were, however, asked to reflect on situational code-switching, i.e. to report on where they are aware of changing the language they use according to the situations in which they find themselves, e.g. in speaking to different family members in different circumstances, or at home, at school, at work, or in church gatherings.Many indicated sensitivity to metaphorical code-switching where a change in topic or in positioning of participants, prompts a change of code (cf.Wardhaugh 1998:103).At least three interviewees indicated that mothers who systematically speak English to their children, would switch to Afrikaans in scolding the children or in expressing feelings of anger or irritation -as in ( 5) and ( 6).
( Indications of situational code-switching in this linguistic community give information on the domains in which various languages are used.Metaphorical code-switching can indicate language attitudes and language preferences of speakers.

Data collection
As has been mentioned, data were obtained by means of interviews, where interviewers met with each family for between one and three hours (the 2003 interviews were on average longer than the 2008 and 2009 ones).The interviews took the form of structured conversations that were recorded with the consent of all participants.Generally, the length of conversations assisted in assuring authenticity and overcoming the effects of the observer's paradox: after a while participants became engrossed in the conversation and seemed to forget that the recorder was there at all.In all cases at least one of the interviewers was a member of the particular community herself; often interviewer and interviewee knew each other well enough for pretences to be dropped -one can hardly fabricate a scenario that is patently out of line with what your friends or acquaintances know.Older family members who at first preferred only to "sit by" became drawn in, while younger children who were both keen and self-conscious in participating at first, sometimes grew tired and left the conversation altogether.
The families who participated were selected for the specific fact that the parents identify themselves as Afrikaans L1 while they habitually, and from a very early age, use English in communicating with their children to the extent that the children emphatically assert English as their L1.This project is a response to curiosity about the nature of these observed patterns of language choice and language use, where adults deliberately choose to raise their children in the parents' L2.This paper analyses and interprets the information a small number of families presented first in 2003, and then again five to six years later in 2008 3 and 2009, as indicative of trends extant on a larger scale and as pointers to a relatively exceptional occurrence of community language shift.
In 2003 members of only three families were interviewed -also with a view to determining whether the concept 'language shift' is in fact an accurate one to apply in the given constellation of competences and uses of the two languages.From the responses we gained general information on the interviewees' language identity, language preferences, and which languages were used in which settings.In 2009 members of eleven more families were interviewed; specifically, in each case members of Generation 2 and Generation 3 took part, reporting on what they know of Generation 1's language competences and practices.Only in two cases were members of Generation 1 themselves present to take part in the interview.

1.4
The research questions more precisely This project set out to provide answers to the following research questions: 1. Are the language choices apparent in the families that were interviewed, indicative of language shift? 2. Which features of the patterns of language use that participants reported, can be taken as markers of language shift? 3. Which features of the patterns of language use that emerged, appear to be unique to this particular context (thus are a-typical of language shift elsewhere)? 4. What explanations do respondents provide for the perceived language shift? 5. What is the relation between the perceived language shift and bilingualism/multilingualism?
1.5 Background of respondents This section will give a brief exposition of the social setting of the families that were interviewed.Such details assist in explaining the interconnectivity of language and society, language choice and community histories, language preference and personal histories -all aspects that are significant for a proper understanding of processes of language maintenance or shift.Notably, the language shift in focus here is in a community classified as "coloured" according to the racial classification introduced by the government from 1948 to 1994.This community which is diverse in terms of religious, social and educational cultures 4 , widely presents evidence of language shift on a scale that is considerably less evident than in comparable historically Afrikaans, white communities.The ethnic dimension to the shift can therefore not be ignored.
During 2003 the families interviewed were from three different residential areas within a 60 km radius of Cape Town, namely Kraaifontein, Blomtuin, and Firgrove/Macassar near Somerset West.During 2009 the families interviewed were from Strand, Somerset-West, Macassar and Paarl.Three different generations are identified on the basis of the age of participants and their family relationships (grandparents, parents, children).In all of these families the first generation (Generation 1) were people with a working class background, many of whom had not been in a position to complete a secondary school education.The second generation (Generation 2), who in the 2009 interviews were the ones who contributed most, had all been afforded better opportunities and, in spite of restricted choices under the apartheid government, most have a tertiary qualification from a college or university (or both).Thus the second generation participants have decidedly middle class positions working, for example, as teachers or in middle management of big companies.The third generation (Generation 3) participants were all younger than 25, all of them either still at school or studying at a tertiary institution.Thus the respondents represent a group in the community who has in the past 30 to 40 years experienced upward mobility in social and socio-economic terms.
One of the 2003 participants from Macassar related how they had been moved from an established Cape Town suburb to this windswept and underdeveloped area as part of the 1960s forced removals.In 2009 similar references were made to how forced removals affected educational choices and opportunities that to them are not unrelated to language attitudes and language choices.One participant articulated the development from such circumstances, indicating that her parents had moved from a rural community in the Northern Cape to Cape Town "vir beter werksgeleenthede en so" (for better job opportunities and so on), and remembering the community set-up as follows: (7) Macassar self was 'n slaapdorp / Afrikaans oorwegend gewees, 'n slaapdorp, mense was daarheen geskuif / eh, forced removals / en eh daar was niks, geen infrastruktuur nie / mense die.. die / niks skole nie / ons moes reis / um geen winkels nie / ehh / ja die die / dit dit was maar lae sosiale klas /en / lae behuising / so.
[Macassar itself was a "sleep-town" / was overwhelmingly Afrikaans, a town where people came home to sleep, people were moved there / eh forced removals / and eh there was nothing, no infrastructure / people the .. the / no schools / we had to travel / um no shops / ehh / yes the the / it it was quite a low social class / and / low housing / like that.] The language in which the interviews were conducted was largely decided by the interviewees.They were asked at the outset which language they would prefer.Also, the interviewers deliberately or intuitively code-switched from time to time in the course of the interview to accommodate participants in whichever language they were most at ease with.Most participants spoke in Afrikaans, although some switched to English from time to time, and one or two code-switched quite elaborately, as in (1) to (4) above.

Overview: patterns of language use in families
In this section I shall comparatively summarise findings as to language identity, languages of learning and bilingual proficiencies across the three generations in the various families.As more families were interviewed in 2009, the analysis gives a summary of findings, not a separate description of each family.

Language identity
In 2003, concerning language identity, Generation 1 of each of the families identified Afrikaans as their L1 and as the language used in their childhood homes.The grandmother of Family 1 did additionally indicate that English was used in the home in virtually equal measure to Afrikaans and she could not say she had an Afrikaans-only identity.Generation 2 of Family 2 and Family 3, similarly indicated Afrikaans as L1 and as the language of their childhood; Family 1 indicated a clear Afrikaans/English bilingual identity and English as the predominant home language.In contrast, Generation 3 unambiguously indicated an English L1 identity, and English only (Family 1) or predominantly English (Families 2 and 3) as home language.
In 2009, all of the families identified Afrikaans as the L1 of Generation 1.In two cases, however, their children (Generation 2) indicated that one of their parents (Generation 1) had been to a good school and had spoken English very well.Similarly, all Generation 2 participants said that they regard their L1 to be Afrikaans.Some acknowledged that even though they used English with their children, they found it considerably easier to express themselves in Afrikaans.Even so, they had chosen to raise their children in English.They did this by talking English, their own L2, to their children virtually from birth and by enrolling them in English schools.With only one exception 5 , all the Generation 3 participants -even those with very good Afrikaans proficiency 6 -were adamant that they were "definitely English".

2.2
Language of learning In 2003, concerning language of learning, the Generation 1 member of Family 1 indicated that her schooling in the 1930s had been in English and that she had chosen the same for her children.The Generation 1 members of Family 2 and Family 3 had themselves been schooled in Afrikaans during the late 1940s and had, similarly, considered nothing different for their children.This pattern, however, changed with Generation 2 whose schooling started in the late 1950s and 1960s.Regardless of whether they had themselves been schooled in Afrikaans or English, they elected English as language of learning for their children.In all cases Generation 2 was living within virtually the same area, no more than 20 km away from where they had been raised -thus they chose for a language change without any co-occurring geographical shift.The Generation 3 participants were mostly too young to make a choice on preferred language of learning for themselves.The 23-year old participant in this category indicated that for her own children she would prefer English, though her husband's choice would also be considered and if he preferred Afrikaans she could perhaps be swayed.
In 2009, Generation 1 and Generation 2 participants, excepting one 7 , had been in Afrikaans medium schools.Throughout, all Generation 2 participants had elected to enrol their children (Generation 3) in English medium schools.Two mothers mentioned that they had placed their first children in Afrikaans classes, but within the first three years had made a considered choice to move them to the English class or to an English school.None had regrets as they felt they had given their children better opportunities than they themselves had had.

2.3
Bilingual proficiencies In 2003, when asked to assess their own bilingual proficiencies, Generation 1 respondents disclosed rather interestingly that over time only one of the three had maintained Afrikaansdominant bilingual skills.Of the remaining two, one indicated that she was more proficient in English while the other claimed equal levels of proficiency in English and Afrikaans.Generation 2 representatives in two families claimed equal levels of proficiency in the two languages from an early age, while the third person, similar to her mother, claimed a development from Afrikaans-dominant bilingualism to balanced bilingualism during her career.Generation 3 of all three families were unambiguous about their position: they had stronger proficiency in English than in Afrikaans, even to the extent that they were often reluctant to speak in Afrikaans at all.
In 2009, Generation 2 representatives all asserted that their parents (Generation 1) had been Afrikaans.Nevertheless, many members of Generation 1 appreciated, or were reported to have appreciated, the value of English in the workplace, in the globalising world and (in some cases) even as a language associated with higher status than their L1.Such appreciation is often, though not in all cases, associated with relatively high levels of English proficiency.Generation 2 participants mostly claimed and, based on researcher analysis of the recordings, were found to have relatively strong bilingual proficiencies; some would claim equal competence in both languages, others indicated that in spite of pervasive use of English "ek voel baie meer confident in Afrikaans" (I feel much more confident in Afrikaans).Generation 3, in line with being very assertive about claiming L1 English identities, proved largely to be English-dominant bilinguals.Most would answer in English even when they were addressed in and had no difficulty understanding Afrikaans.Nevertheless, a number of the youngsters expressed pride in their ability to use Afrikaans well, as in ( 8) and ( 9). ( 8) I think I'm very bilingual 'cause my mother often communicates to me in Afrikaans and / I chose Afrikaans as a first language at school and / I actually enjoy the language I try to speak it a bit more / so it's more out of my own choice.(J1.3) (9) By Luisa vind ek nogal 'n 'n 'n eagerness om vir haar uit te druk in Afrikaans.(J4.3) [With Luisa I find sort-of a a an eagerness to express herself in Afrikaans.] Although respondents' bilingual skills were not tested and there is no necessary correspondence between high self-rating and high levels of proficiency, their self-assessment is taken as an indication of the confidence with which they engage in communication in either of the languages they know and, thus, also as indication of how often they use the two languages.

Indicators of language shift
This project emerged from an assumption that there is already, at least to some extent, a process of language shift underway in many historically Afrikaans "coloured" communities.
To motivate such an assumption, the following features typical of language shift elsewhere are reported in our data and thus also characterise the process in this community: 1.In identifying their own L1, there is a marked shift among individuals across three generations.2. Regarding family language or home language, there is a shift from Generation 1 through to Generation 3. 3.There is a shift in language of schooling, reflected in what parents have chosen for their children.4.There is a change in patterns of bilingualism from Generation 1 through to Generation 3.
English dominance correlates with the expressed language preference of the third generation in all of the interviewed families.5. Changes in patterns of language use in religious practices 8 are reported in these communities.Often, besides referring to patterns of use at home, at school and in the workplace, respondents introduced the church as a social setting that was formative in one way or another.Some Generation 1 and Generation 2 participants had themselves attended church schools (Roman Catholic, Methodist) and a few Generation 2 parents had chosen to send their children to private church schools as well.In 2003 it emerged that churches that used to be Afrikaans only, recently introduced regular services in English to accommodate specifically their younger members.In 2009 similar patterns were reported -many indicated that even where two languages were used in the family, the church services they regularly attended were exclusively English.This is significant considering that religion, even as an admittedly private domain, is important in identity construction.

Reasons for language shift
Respondents of Generation 2 were frank about the fact that they had often explicitly chosen to transfer to using English in spaces where they had previously used Afrikaans.Particularly they were articulate in justifying their choices to send their children to English schools.Mostly the choice to raise their children in English was taken at the birth of the eldest, which resulted in a practice of parents speaking Afrikaans to one another, but English to their children.The children mostly understand Afrikaans well, but are reluctant to speak anything but English.This section summarises the reasons members of the community themselves give for the language shift they have been instrumental in bringing about by choosing English for the education of their children.
Fluency in English is perceived to be an advantage, an asset that will improve social mobility and employment opportunities.Generation 2 participants were acutely aware of this, and where Generation 1 participants were interviewed, this was reiterated.Generation 3 participants shared this perception.
(10) … die status van Engels / Engels is beskou / was beskou as 'n elitetaal (J4.2) [the status of English / English is seen / was seen as an elite language] (11) … Engels is iets wat jy oral mee kan gaan (E6.1) [English is something that you can take with you everywhere you go] On the one hand, the second generation had experienced their own limited proficiency in English simply as socially embarrassing and frustrating; on the other hand, they had found that during their studies they had been disadvantaged by the difficulties they had using English academic literature.
(12) Ek het gevoel ek willie hê my kinders moet deur daai / daai um trauma gaan nie (lag) sal ek it maar nou so sê want dit was vir my nogal baie erg / want op kollege het ons / ons notas alles Afrikaans gekry maar dit was vir my 'n baie groot sprong op universiteit toe't ek nou moes die Engels vertaal het en s-/ dat allie boeke hoofsaaklik in Engels was en ek wou nie gehad het my kinders moes daardeur gaan nie (J1.2) [I felt that I don't want my children to go through that / that um trauma (laugh) shall I put it that way because it was really quite bad for me / because at college we had / our notes all in Afrikaans but it was a big step to university when I had to translate the English and s-/ that all the books were mainly in English and I did not want my children to go through the same thing] Many mentioned that they simply wanted the best schools for their children in terms of academic standards and in terms of discipline.In their context these were English schools.
Not too far removed from the argument for better schools and better educational opportunities, is one of political awareness.For many, even though Afrikaans was their L1 it was also the language of the former oppressive government 9 , and speaking English to their children was an act of defiance and a way of taking distance.
(13) … ook as ons liewer hulle in Engels grootmaak want /destyds/ was daar mos nou nie sprake van Stellenbosch Universiteit wat Afrikaans sou wees waar my kinders welkom sou wees nie, verstaan jy? (J2.2) [also if we rather raised them in English because / at the time / there was no thought of Stellenbosch University that would be Afrikaans where my children would be welcome, you know?] (14) … en dan ook as gevolg van die land se beleid / eh die die apartheidskwessie dat hulle jou nie wou erken het as mede-Afrikaner of as mede-Afrikaanssprekende / eh jy't nie trots gevoel saam oor jou taal nie / alhoewel jy baie lief was vir jou taal (J4.2) [… and then also as a result of the country's policy / eh the the apartheid issue that they didn't want to acknowledge one as a fellow-Afrikaner or as fellow speaker of Afrikaans / eh one didn't feel joint pride in your language / although you really liked your language ] An interesting, if unexpected, perspective came in reference to dialectal difference.Women whose fathers spoke Afrikaans proudly and were sensitive to the differences between "standard Afrikaans" and "Kaapse Afrikaans" checked the usage of their children, correcting "errors" in pronunciation and grammar, thus trying to educate them into what to them was the better choice.To one Generation 3 respondent in 2003 this was disconcerting, as she was never quite sure whether her father would be satisfied.She resorted to speaking English only to her father.In addition, she experienced conflict in speaking standard Afrikaans to please an older generation, and then being taunted by her peers for trying to be "smart".English provided a way out of the dilemma she had in choosing between two varieties of Afrikaans.This sentiment was also apparent in the 2009 data where at least three parents indicated that they preferred and actually used a variety closer to the standard, were sometimes even singled out and frowned upon for that, but would not succumb to what they held to be lower standards.They believed that the Afrikaans of their home community was not of an acceptable standard and, for this reason, they did not want their children to pick it up.
(15) … ek dink nie hulle sal die Afrikaans gepraat het as 'n standaardtaal nie / dit was een van my primêre redes (J3.2) [I don't think they would have spoken the Afrikaans as a standard language / that was one of my primary reasons ] Many referred to the value English has in a globalising world 10 and how they wanted their children to be in an ideal position to profit from this.
(16) … die kinders is ek bly voor / hulle't nie nodig om om die hoekie te staan / as daar Engels gepraat word (E6.1) [… I am pleased for the children's sake / they don't need to hide around the corner / when English is being spoken] Finally, at least two parents, in recalling their own difficulties with learning and confidently using English, felt that they live in Afrikaans communities where the children hear Afrikaans all the time and so would easily pick it up in the neighbourhood.Therefore, to ensure better bilingual skills, they had to give their children direct exposure to English in the home.

Unique features of Afrikaans-English shift in Western Cape metropolitan communities
In conclusion, the question one has is whether the language shift that appears to be quite advanced in some Western Cape communities is similar to recorded processes of language shift elsewhere -recall, for example, the histories of shift in places such as Nova Scotia (from Gaelic to English), and Australia (from indigenous languages or immigrant languages such as Greek, Polish, Arabic or Tagalog to English) referred to in section 1.2 above.The following list summarises certain patterns that appear to be unique and typical of the local circumstances. 1.
The shift in the Cape Metropolitan area is not related to geographical movement of a linguistic community such as is found when large numbers of people immigrate to a new country.Movement due to the forced removals of the 1960s and 1970s had an effect on language choice -but these shifts did not entail contact with new language communities, and in this particular case the movement remained within, at most, a 100km radius of where people had started out.

2.
An earlier assumption that the shift was strongly related to the need to take political distance from a former oppressive government, was only partially verified.In some cases the shift started long before the advent of Apartheid in 1948, and many continued the trend even after the introduction of an inclusive government in 1994.

3.
In a globalising world where the power of English has become evident, many made a pragmatic choice for English as a language that would be the key to better opportunities for their progeny.4.
A number of scholars writing on language shift have noted the "strong desire to maintain one's own language, one's own perspective and interpretation of the world" (Edwards 1995:53).Our data give minimal evidence of the "powerful protective sentiments for one's own group" in linguistic terms.Even if there was pride in the second generation's own bilingual proficiency, these respondents rarely voiced a sense of loss or nostalgia in reference to their children's lower levels of Afrikaans proficiency.5. Clyne and Kipp (1997:465), in tracing trends of home language shift in Australia, remarked that in the first generation females tended to be more monolingual, and that men shifted more to English in the home than women did.Considering gender patterns, we found the opposite.In the older generations the men were often Afrikaans monolinguals, or at least strongly Afrikaans-dominant bilinguals who were reluctant to use English at home or in public settings.Concerning choices for home language and for language of education, we found that the women mostly gave more direction and were more articulate.Nevertheless, a few second generation women mentioned that as their children's L1 English proficiency began to exceed their own, they felt less confident and would revert to Afrikaans more than the fathers did.This tendency was confirmed by teenage children.6.
There is a marked decline in bilingual and multilingual proficiency from Generation 2 to Generation 3. Nevertheless, even where youngsters were reluctant to show Afrikaans proficiency, they can certainly not be characterised as monolinguals.7.
Language shift from a non-prestigious and less powerful language to one more prestigious and more powerful, is widely recorded.In the linguistic community investigated here, the transfer is from a non-standard variety of Afrikaans to a variety of English that many would still characterise as non-standard.Even so, the third generation's L1 English apparently has greater prestige than either non-standard or standard varieties of L1 Afrikaans.

Acknowledgement
This paper was written during my stay in Bangor, Wales while the Centre for Research on Bilingualism hosted me as a Visiting Researcher.I would like to thank them for their support during this time.

1.
See Anthonissen and George (2003), paper presented at FIPLV annual conference and published in the conference proceedings.2.
I would like to recognise specifically the valuable assistance of Jean Farmer and Erica George in data collection for the extension of the 2003 project.See also the unpublished Master's thesis of Jean Farmer (2008).

3.
The 2008 data was not calculated into the overall assessment set out in section 2 as it was collected in the context of a Master's project where interviews with members of generation 2 and 3 made out only a very small part of the project.It is, however, used in considering features of language shift and reasons speakers give for having decided on changing their family language.

4.
In fact, it has been noted that classification as "coloured" was to a large extent a default category in the racial classification of citizens who were of mixed descent (e.g.Dutch and Malaysian, African and European) or of descent for which a separate category had not been designed in the government's four-group system (e.g.Malaysian, Indonesian, Central-Asian, Chinese).Thus, even ethnically this community can hardly be characterized as homogenous. 5.
This was a 17-year old youngster who had said that he did not pay attention to which language he used; his mother had indicated that currently most of his friends were Afrikaans.6.
Proficiency here was indicated by school marks in Afrikaans L1, by their performance at university where much of the work is presented in Afrikaans, as well as by the Afrikaans interjections they would make in the interview itself.7.
This was a mother of a 5-year-old whose own mother is a teacher and had in the 1970s already decided to place her children in an English medium class -similar to the way in which Generation 2 parents had taken the decision at a later stage.8.
For an elaboration on language shift in religious practices, see Edwards (1995:110ff) on the shift from French to English in religious confessions in Vermont.9.
See Arel (2002:92ff) for an elaboration on the association between language and state.10.
See Clyne (2004:24-27) on the "myths and paradoxes" of electing to raise children in the dominant language of a multilingual community.
Engels -of word die twee tale mooi uitmekaar gehou en elkeen redelik suiwer gepraat? - Mens skel ook lekkerder in Afrikaans -nou nie dat ons so baie vloek nie, maar … [One can also scold/talk badly more easily in Afrikaans -not that we swear such a lot, but] (J3.2 2009) 5) My mother, when she scolds us, she uses Afrikaans.(J4.3 2009) (6) If parents and children have different first languages, did this primarily come about as a result of the parents' choice?What reasons are given for the shift from one L1 (Afrikaans) in the first/second generation to another L1 (English) in the next?As die ouers en kinders nie dieselfde eerstetaal het nie, is dit omdat die ouers dit so gekies het?Watter rede(s)word gegee vir die skuifvan een T1 (Afrikaans)in die eerste/tweede generasie na 'n ander T1 (Engels) in die volgende generasie?-How would you rate your own Afrikaans-English bilingual abilities: equal, Afrikaans dominant or English dominant?How would you rate the Afrikaans-English bilingual abilities of your children?As jy moet sê hoe tweetalig jy is, is jy ewe goed in Afrikaans en Engels, of beter in Engels / beter in Afrikaans?As jy moet sê hoe tweetalig jou kinders is, is hulle ewe goed in Afrikaans en Engels, of beter in Engels / beter in Afrikaans?-Would you say that since your children have been in school their levels of bilingual proficiency have markedly improved?Do you think your own level of bilingual proficiency has improved since, through your children, more English is used than before in the family?Sou jy sê dat jou kinders se tweetaligheid merkbaar verbeter het sedert hulle in die skool is? Dink jy dat jou eie tweetaligheid verbeter het sedert julle, deur die kinders, meer Engels as tevore in die gesin/familie gebruik?note Christine Anthonissen is Associate Professor and Chair of the Department of General Linguistics, Stellenbosch University.Her research focuses on social aspects of Bilingualism and Multilingualism, as well as Discourse Studies and Critical Discourse Analysis.She is a member of the editorial board of the Journal of Language and Politics (Benjamins, editors Ruth Wodak, Paul Chilton) and of Language Assessment Quarterly (Taylor and Francis, editor Anthony Kunnan).