<2>

For a long time ǂHoan was assumed to be an isolate language that could not be shown to be related to any of the other Khoisan languages. [2] ^,  [3] This was partly due to the lack of linguistic material and general information about the language and its structure. In 2010, however, Heine & Honken finally published an article showing that ǂHoan is related to the Ju languages (Greenberg's Northern Khoisan). They argue for a language family called the Kx'a family using the comparative method to reconstruct "some traits of phonology" (Heine & Honken 2010:6). The proposed Kx'a language family (kx'aa is a shared lexeme meaning 'ground' or 'soil') includes the languages ǃXuun and ǂHoan. Other authors, e.g. Güldemann (2008), subsume the Ju languages and ǂHoan under the label 'Ju-ǂHoan'. Map 1 shows the historical distribution of the languages belonging to the Khoisan lineages.

<3>

So far, ǂHoan is only poorly documented. In fact, according to Güldemann & Vossen (2000), it is one of the least described Khoisan languages still spoken today. Apart from a few articles looking at specific grammatical constructions (Gruber 1973, 1975a; Collins 1998, 2001, 2002 and 2003; Collins & Bell 2001), there is also a very short treatment of major aspects of ǂHoan grammar by Collins and an unpublished word list by Gruber (1975b). Neither a grammar nor anything on the complex topic of phonetics and phonology has yet been published.

<5>

The main motivation was the fact that it is highly important for the field of contact linguistics to have detailed descriptions not only of the languages themselves, but also of the sociolinguistic settings in which the languages are spoken when language contact is taking place. Furthermore, this sociolinguistic study was conducted in order to see how the situation of ǂHoan and its speakers has changed since Batibo conducted his survey in 2003. An additional aim was to include speakers of villages that Batibo did not include when he did his survey. This is the only way in which one will be able to see which factors may have had an impact on the contact situation. Eventually, a detailed sociolinguistic description is essential for interpreting linguistic data in a language contact situation.

<6>

Additionally, the transcriptions of the ǂhoan recordings by Sands (2005), which she kindly made available to us, systematically differ from our transcriptions. Since Sands recorded speakers from a different village, the question arose whether there may be some phonetic variation between the ǂHoan spoken in different villages. The most systematic difference is that Sands' sibilants and affricates are always post-alveolar while our data show the alveolar counterparts. (1) and (2) give some examples of the transcriptional differences between Sands' (2005) and our data.

<7>

Comparing Sands' recordings to our recordings suggests (at least on the basis of present state of documentation) that there is no (or rarely) an actual difference in pronunciation. It is, however, generally noticeable that there is variation in the pronunciation of certain phonemes within the utterances of one speaker as well as between speakers. These variations, which concern the manner of articulation and voicing, occur most frequently with sibilants and affricates, i.e. /z/ can be either pronounced [z], [Ʒ] or (at least partially) devoiced as [s]. Thus, the examples above display exactly such a variation. Depending on how often one or the other phone occurs, one will eventually choose one as the phoneme (which will thus be the variant used in the transcription). This is a possible explanation for the transcriptional differences between us and Sands. Although there is still very little data from Salajwe, it can already be observed that the Salajwe speaker (number 12 in Table 1 below) consistently uses [ʃ] where all other consultants tend to use [s].

<8>

Additionally, as mentioned above, Collins (1998) as well as Batibo (2005a) report the existence of a closely related language or dialect of ǂHoan called Sasi. The search for speakers was thus extended further to the north towards the Khutse Game Reserve in order to investigate potential phonetic variation in the language or differences in the sociolinguistic setting.

<9>

The survey was conducted in August and September 2010. The questionnaire was conducted with 13 of the 15 fluent speakers we met. The remaining two speakers were sick at that time and therefore not able to participate in the survey. In contrast to Batibo, we only included fluent speakers and did not conduct the questionnaire with non-fluent or non-speakers of ǂHoan  [5].

<10>

The questionnaire consists of 69 questions covering the following topics:

<11>

The basis of the questionnaire is a modified version of a sociolinguistic questionnaire developed by Brigitte Pakendorf that she kindly made available to us. Changes were made mainly on the basis of Blair (1990) and Grimes (1995) and in order to adapt the questionnaire to a southern African cultural context.

<12>

In addition to the questionnaire, a short word list of 58 items was elicited with six of the speakers. This was mainly done in order to get initial data on potential dialectal variation between the ǂHoan spoken in different places. The word lists were recorded with three speakers in Khekhenye and due to time limitations with one speaker each in Tswaane, Dutlwe and Salajwe.

The data was entered into a file maker data base.

<13>

Today ǂHoan speakers are mainly found in the Kweneng District, in particular in the area adjacent to Kang, the regional centre [6]. The villages and settlements are located along a quite recently reconstructed and now tarred road connecting the Trans Kalahari Highway and Letlhakeng. Speakers were found in the following villages along this road: Motokwe, Khekenye, Tswaane, and Dutlwe (cf. Map 2). Three more speakers were found in the villages of Mathibatsela and Salajwe. In the latter village there are supposedly some more speakers that were absent when we visited in September 2010. Furthermore, we met some non-fluent speakers in Khudumelapye. In other villages around this area it is still known that this language existed but no more speakers can be found. We can, however, not exclude that there are some more speakers in other, even more remote areas that we were not able to visit or do not know of. Most of the older speakers at least were still born inside of what is nowadays the Khutse Game Reserve or the Central Kalahari Game Reserve and only moved southwards when they had to leave the reserves from the mid 1990s on.

<14>

Collins (1998) reported that some speakers could be found in Tsia. When we visited Tsia in 2010, we found that Tsia is not a village but a cattle post, i.e. a non-permanent settlement where people stay for some nights in order to take care of their cattle. No speaker could be found and the two people that were present there did not know of anyone who speaks ǂHoan.

<15>

According to Collins (1998) and Batibo (2005a), the language (or dialect) most closely related to ǂHoan is called Sasi. This language is reported to be spoken in the south-eastern villages Lethajwe and Artesia (south of Shoshong) by Collins (1998) and in Serowe by Batibo (2005a). So far, however, no data of this language has been published and none of our consultants knew anything about this language. Additionally, Collins (1998) does not give any source for this information and Batibo (p.c.) got this information from his consultants but did not personally try to find the speakers. Thus, these villages are still to be visited. Note also, that Sasi is the Kgalagadi name for the Taa language and ǀaati (of which sasi could be a 'simplified Tswana pronunciation') is the ǂHoan word for 'Khoisan language'. Therefore, it is uncertain whether Sasi as a closely related dialect or language really exists or whether the term simply refers to Taa or other Khoisan languages in general.

<16>

The number of speakers does not amount to more than 50 who are mainly between 60 and 70 years old. The number reported here is only an approximate estimation. The estimated number of 50 speakers is made up on the basis of the following numbers: there are about 15 fluent speakers (all met in person) and about 20 non-fluent speakers. The remaining 15 were included for the speakers and semi‑speakers [7] that we have not met and that could potentially be found in some places that were not included in this survey, i.e. for example in areas inside the Khutse Game Reserve. Batibo (2005a) still reports a number of between 120 and 200 speakers. This might be interpreted as a tremendous decrease of the number of speakers during the last seven years. The source of Batibo's data is, however, not clear since he does not cite any census. Thus, neither the numbers of speakers nor the apparent rate of decrease during the last years can be considered to be precise.

<17>

ǂHoan is no longer passed on to the younger generations and is therefore no longer learnt as mother tongue (L1). The mother tongue of the majority of children is the Bantu language Kgalagadi (S.31d), which is also the lingua franca of the region. According to Batibo (2005a), not only the ǂHoan speakers but also the speakers of all the other Khoisan of this region (i.e. / ǃXoon  [8], Naro, Gǀui and Gǁana) are fluent in Kgalagadi.

<18>

Batibo cites Anderson & Janson (1997) who state that the ǂHoan speakers were present in the Kweneng district already before the first Bantu speaking people arrived approximately 1000 years ago. He also claims that at that time the only other Khoisan language spoken around that area was ǃXoon. According to Traill and Nakagawa (2000) there are also indications for more recent contact situation between ǂHoan and Gǀui speakers at a pan west of the Khutse Game Reserve until about 60 years ago. The interviews with our consultants strengthen the assumption that there has been intense contact between these groups (cf. section 'Multilingualism').

<19>

The individuals belonging to the speech community of ǂHoan live scattered over at least six villages (mentioned above) that are quite distant from each other (i.e. the minimal distance between two villages is 7 km, the maximal distance is more than 100 km). Frequent contact between ǂHoan speakers is thus only possible for those living in the same (or maybe the neighboring) village. This results in a situation where most of the communication involving ǂHoan speakers takes place between them and speakers of other languages in the village, and so communication in ǂHoan occurs only infrequently.

<20>

All consultants participating in our study are multilingual in the sense that they are fluent in at least one more language in addition to their mother tongue ǂHoan. Most of the ǂHoan consultants are trilingual, i.e. they speak Gǀui (Khoe-Kwadi) and Kgalagadi (Bantu) fluently besides ǂHoan. The consultants from Salajwe (cf. Table 1, speakers 12 and 13) form the only exception, since they do not speak Gǀui; however, one of them (speaker 13) does have some passive knowledge of Taa (Tuu) instead. Batibo (2005a) already reported that all of his ǂHoan consultants were trilingual, but besides ǂHoan and Kgalagadi he reports that his consultants also speak Tswana [9]. Batibo does not state how he judged his speakers' ability of speaking Tswana (i.e. whether he actually tested them or whether he relied on their self-estimation like we did). In the present survey, however, only three speakers clearly stated that they were able to speak Tswana, while most of them said they would understand the language at least to a certain extent (cf. Table 1) but are not able to speak it. This is due to the fact that Tswana and Kgalagadi are very closely related languages, so that speakers of these languages can adapt to each other quite easily. The observation that all ǂHoan speakers that participated in the our study also speak Gǀui has not been reported by Batibo. His tables actually imply that some of the speakers are able to speak another language (other than Kgalagadi and Tswana) since he has the category "other". There is, however, no explicit explanation of what "other” includes [10]. This is due to the fact that he was mainly interested in the ǂHoan speakers shifting to Kgalagadi (in contrast to his expectation that they would shift to the national language Tswana).

<21>

Concerning the multilingualism of the ǂHoan speakers, it is noteworthy that until approximately the early 1970s (i.e. already before they started shifting to Kgalagadi) there actually was an asymmetrical bilingual situation with ǂHoan speakers acquiring Gǀui, but only a few Gǀui speakers learning ǂHoan (Nakagawa, p.c.) [11]. This asymmetrical bilingual situation is still found with the fluent ǂHoan speakers today, the difference being that we did not meet Gǀui speakers who also speak ǂHoan.

Table 1 summarizes the languages that the consultants speak in addition to ǂHoan and their self-estimated language proficiency in speaking and understanding them.

<22>

Table 1 shows the following about the patterns of multilingualism:

<23>

Table 2 summarizes the languages spoken by the parents and grandparents of the consultants. Please keep in mind that these numbers of course do not rely on self-estimations, but on the judgments and the memory of our consultants. Additionally, at least for the generation of the grandparents it was not asked for every single person (i.e. languages of father's mother, languages of father's father etc.) but just for the languages of the father's parents. Furthermore, it might be the case that some of the consultants mentioned only the L1 of their parents and grandparents, while others clearly gave an exhaustive list of the languages spoken.

<24>

The following conclusions can be drawn from Table 2:

<25>

"A study of language use seeks to describe the choices that people make about what speech varieties to use in particular situations." (Blair 1990:107). Language use is, however, also connected to language attitudes. Language attitudes are the opinions of individuals about why a certain language or language variety is more appropriate in a particular situation or domain (Blair 1990). It also has to be kept in mind that an individual speaker can rate a certain language differently than the same language is rated by his community (for further discussion see section 'Language attitude' below).This is connected to the prestige a certain language has in a community. Thus, a speaker can have a positive attitude towards his mother tongue even though this language is of very low prestige within the society or community. Figure 1 shows the prestige of the languages spoken in the surveyed villages (growing level of prestige from left to right).

<26>

The levels of prestige of the languages in Figure 1 are based on the following observations: