Khoekhoe Lexical Borrowing in Regionalised Afrikaans

This brief paper aims to draw attention to the widespread and ongoing phenomenon of lexical borrowing from Khoekhoe-branch languages into regional Afrikaans. A case study of one Afrikaans plant name loaned from Khoekhoe-branch languages, karee ( Searsia lancea, Searsia spp.), is used to demonstrate how post-shift phonological attrition can lead to lexical conflation, and hence to semantic extension. This paper strongly recommends that public-facing scientific organisations take greater care when providing linguistic information to lay communities, and also motivates increased study of the behaviour of click consonants in a post-shift context in order to develop a clearer understanding of Khoekhoe-branch language history.


Introduction
The plant family Anacardiaceae is represented in South Africa by a number of genera including Searsia. A broadly distributed genus, several of its species are called karee in Afrikaans. The Stellenbosch University Botanical Garden (SUBot) offers its visitors the etymology of the word karee in the form of a small informative sign near to the entrance to the garden, pictured in Figure 1 below. This etymology, taken verbatim from the PlantZAfrica (PZA) web page on Searsia lancea, does not explicitly source this claim further, although it is best traced back to Jackson 1990. Neither of these organisations specifies which "Khoi" language they refer to, and neither provides the precise word they claim signifies "mead" for the sake of linguistic comparison. The KHOE language family (Westphal 1971), also referred to as KHOE-KWADI when incorporating the disjunctively distributed Kwadi language (Guldemann 2004;Guldemann & Elderkin 2010), comprises two branches: Kalahari and Khoekhoe (Voßen 1997;Dimmendaal 2011;Güldemann 2014). The precise number of languages in the KHOE family is difficult to specify, because of the rapid rate of language differentiation and language extinction over the past three centuries, and because of the fluidity of dialect boundaries both past and present. In the Kalahari branch, distributed across contemporary Namibia, Botswana, and Zimbabwe, there historically existed at least sixteen lects (Voßen 1997:386). Within the contemporary boundaries of South Africa, there existed at least four discrete languages in the Khoekhoe branch, possibly more. These would have included Nama (today standardised together with Damara as Khoekhoegowab, well-provisioned with lexicographical and grammatical resources, and still used by at least 200 000 L1 speakers within Namibia); Kora (extinct but welldocumented in Engelbrecht 1936;Killian 2009; Du Plessis 2019); and Cape Khoekhoe and Eastern Khoekhoe (both extinct and virtually undocumented).
Sourcing the loanword karee in Afrikaans to "a Khoi word" is therefore roughly equivalent to sourcing the loanword komāsharu in Japanese to "a GERMANIC word" instead of to the US English commercial. An informative sign in a well-frequented botanical garden provides an ideal opportunity to expose visitors to Khoekhoe-branch lexical items, and to familiarise them with standardised Khoekhoegowab orthography. Failure to provide the exact lexical item robs the reader of the opportunity to compare the loan with its proposed target on their own terms and assess its validity for themselves. As will be discussed, the claim that the Afrikaans item karee originates in "a Khoi word for mead" would be imprecise and insufficiently presented even if it were accuratewhich, this paper argues, it is not.
Encouraging public-facing scientific organisations to offer accurate etymological material when discussing language history is not merely an act of pedantry. This brief discussion of karee aims to illustrate how the close and careful treatment of even one Khoekhoe-branch loanword in Afrikaans can illuminate a number of contact linguistic phenomena surrounding loan, including phonological attrition, semantic conflation, and semantic reanalysis.

Phonological Attrition of Click Consonants
Over the course of the past three centuries, Khoekhoe-branch speech communities within South Africa have slowly shifted from a number of related but distinct languages to Afrikaans (for an overview see Traill 1995:29-36;Berzborn 2003:328-330). The final stages of language shift typically involve lexical remembering. Speakers are typically exposed to the old L1 during childhood, but fail to acquire it to fluency; alternately, lexical rememberers may acquire the old L1 to fluency, but then shift away from it completely later in life. In both instances, speakers are functionally monolingual in the new L1, but do retain lexical items or chunked phrases in the old L1 (see further Sasse 1992 passim;Janse 2003:xiii). The precise ways in which shiftinduced phonological attrition applies to click consonants are still unclear (see discussion at Traill & Vossen 1997;Fehn 2020a;Fehn 2020b), and very little academic attention has been paid to the phonological ramifications of the shift away from Khoekhoe-branch languages as it has occurred within South Africa.
While it is unclear how many L1 Nama speakers remain in the Northern Cape, the language is certainly present and still fluently spoken by older community members across a number of small towns in the Namaqualand region, particularly Kuboes, Eksteenfontein, and Lekkersing (Witzlack-Makaverich 2006:12 offers an outdated assessment of the size of the remaining speech communities in the early 2000s). Further unknown is the extent to which phonological attrition may have affected Nama within Namaqualand (see observations on distinctions between Nama as spoken within South Africa and standardised Khoekhoegowab at Berzborn 2003:342-347).
Interviewing elderly monolingual Afrikaans-speakers in Namaqualand, Links (1989:61-62) recorded at least 60 Nama loanwords. Of the click consonants they used, he said the following: Since the phonemic inventories of Afrikaans and the Khoe language differ, and since Afrikaans-speakers often encounter difficulties in when articulating certain Khoe phonemes, it can be accepted that there has been a great deal of phonological adaptation. …It is clearly apparent in the speech of the consultants that the subtle distinctions in the production of the click consonants are no longer maintained in the Khoe language. When pronouncing a word like [n!gu], for example, a certain consultant will produce the click consonant [!], but then, in another use of the same word, use [ǁ]. As a result, there is a great deal of uncertainty in this area. (Links 1989:62) Unfortunately, Links chose to indicate loaned click consonants using only the IPA character [!] in all of the loan words he encountered, rather than notating each variant form individually. However, there is some evidence in these loaned items that a failure to perceive and adequately interpret crucial phonemic contrast in loaned click consonants had already begun to contribute to the development of homophony. For example, there are three distinct semantic use-cases for the loanword tkam, each of which is transcribed identically as [!αm].
a) Vasknyp, vassit. 'Ek vat die bokooi aan die agterbeen en tkam die poot tussen my knieë.' b) Aanstoot. "Ek het tkam gekry." c) Lig. "Vat 'n vuurhoutjie en tkam hier.' (Links 1989:65) I etymologise these three definitions in Table 1 Links (1989) deliberately chose to transcribe any and all click consonants he elicited during Afrikaans speech using the single uniform designation [!] in IPA and tk in Afrikaans; this implies that the elicitations transcribed in this paragraph exhibited at least some degree of variation. Even without exact transcriptions of the individual consultants' pronunciation, this set illustrates how, following lexical borrowing, the reduction or outright loss of contrast across a particular phonetic category may potentially give rise to extreme homonymy. In the Karoo Highlands, Kora is completely extinct, but monolingual speakers of Afrikaans with some lexical recall of Kora do still occur, and will loan Kora items into everyday Afrikaans conversation. The best documentation of this is to be found in Killian (2009), whose findings on phonological attrition precisely mirror those recorded in Links 1989 with the observation that, "clicks are at times used indiscriminately; switching from one click to another, adding clicks, and deleting clicks are all possible situations" (Killian 2009:84).
Thus, when lexical items hosting click consonants are introduced into a language without an established click inventory, whether via lexical remembering or via loan, the general result in a post-shift context would appear to be the complete destabilisation of the original phonemic inventory. However, it should also be noted that the failure by observers such as Links 1989 to accurately transcribe variation gives rise to the possibility that this apparent phonemic destabilisation is an artefact of poor documentation. For a preliminary effort to capture variation in loaned clicks, see Christie, 2019:82-83, but further elicitation of Khoekhoe-branch lexical retention in Afrikaans is urgently required.
Additionally, detailed commentary on how and why this destabilisation occurs, whether any contrast is retained at all, and whether this then opens up the possibility that the variable loaned clicks will eventually level and perhaps even begin to develop new systematic contrasts does not yet exist.

One Word, Four Origins
The sharp reduction of a stable contrastive inventory during loan has the potential to introduce homophony, following which semantic conflation and semantic reanalysis may develop. I argue that this process occurs at least intermittently in Khoekhoe-branch loans in Afrikaans, and that it has driven the development of the modern plant name karee.
As has been amply documented by a variety of lexicographical and ethnobotanical resources over the last seventy years, the diffusion of loan words containing click consonants into regionalised is a widespread and ongoing phenomenon (see among others Boshoff & Nienaber 1967;Le Roux 1981;Links 1989;Archer 1994;Prinsloo 2008;Nortje & Van Wyk 2015). Loaned plant names especially offer prime scope for an introductory discussion of semantic reanalysis following the collapse of phonemic contrast. In Common Names of South African Plants, Smith (1966:279,312) notes that the original target of karee, ostensibly a Khoekhoebranch item, applies to both Searsia and Lycium, but that karee underwent contraction via syncope, yielding krie.

karee.
A name originally derived from Karoo and also heard in such variants as kir(r)ie, kree, krie. In these forms, but principally as karee,  (Smith 1966:279) krie-. A common contraction by phonetic abrasion of karee-.
kriedoring. Another rendering, or contraction of kareedoring, and confined to almost all species of Lycium. (Smith 1966:312) I suggest that karee = Searsia spp. and krie = Lycium spp. are in fact loans of two wholly distinct targets. The genus Searsia occurs in the cashew family Anacardiaceae of the order Sapindales, while the genus Lycium occurs in the nightshade family Solanaceae of the order Solanales, meaning that these two genera are not closely related. Given the regularity with which kriedoring is applied to Lycium, and never to Searsia, it is worth scrutinising extant Nama plant names for separate targets. However, on being borrowed into Afrikaans, both items lost a significant degree of phonological and grammatical information. Since Afrikaans does not mark either case or grammatical gender in the morphology, both items lost their inflectional endings in an instance of morphological simplification. Afrikaans does not support phonemic tone, and so both items lost their contrastive tone melodies. Finally, both items also underwent click deletion, and the lateral click and the alveolar click were replaced by the same phoneme, the voiceless velar stop [k]. Following this extreme loss of phonemic contrast, the two historically distinct targets ǁaris = Lycium sp. and !àréb = Searsia sp. items underwent lexical conflation, and were reanalysed as the same lexical item, karee.
A further complication arises in t'narra, another Afrikaans common name of Khoekhoe-branch origin for genus Searsia. This form may stand alone, or it may serve as a prefix to -bessie, but always refers to Searsia sp., as shown in Table 2 As is common practice in lay and ethnobotanical records of Khoekhoe-branch loans into Afrikaans, the ad hoc orthographies t' and ! both serve to indicate the presence of a click consonant. While the precise quality of the click cannot be extracted from this orthography, it can be taken as evidence that some form of click consonant is retained in at least some variants on this loan. I would further like to draw attention to an underutilised resource on Nama as it was historically spoken within South Africa, the monograph Aus Namaland und Kalahari (Schultze 1907). In addition to providing an excellent and understudied corpus of Nama in the form of a number of short stories, Schultze also offers observations from 'Klein Namaland', modern-day Namaqualand, including a number of localised plant names that have not been brought forward into the standardised Namibian lexicographical resources. Among these is the plant name |gàra- [|ara]. Schultze records this item both in the feminine singular form '|gàra.s-Baum', for [Searsia] lancea itself (Schultze 1907: 144), and in the masucline singular form |gàra.b, for 'die beerenartige Frucht vom [Searsia] lancea' (Schultze 1907:346). This record locates the usage of this term within modern-day Namaqualand, and renders it a likely target for the loan generally spelled t'narra.
Finally, in addition to having been conflated with each other, the historically distinct names kareebos, kriedoring, and t'narrabessie have undergone further confusion with a series of names applied to plants traditionally used as fermentation agents by a number of indigenous southern African communities.

kareemoer.
A name applied to several species of which portions were used as a ferment or yeast (Afr.: moer) in the making of honey beer. …The "karee" part of the vernacular name is the original Khoekhoe and is said to mean "honey beer". (Smith 1966:280) Smith goes on to identify a diverse array of plants that serve as moerwortels or fermentation agents in the brewing of honey beer, ranging from Trichodiadema stellatum (a small succulent mesemb in Aizoaceae), to Euphorbia decussata (a glaucous, much-branched succulent shrub in Euphorbiaceae), to two species of Anacampseros (dwarf succulents, typically covered in distinctive leaf scales, in Anacampserotaceae). This suggests that the application of this specifier karee-to plant names is predicated on use, rather than on genetic relatedness.
In this context, the item karee-straightforwardly targets the Namibian Khoekhoe word !kharib [ǃ͡ χarib] = 'honeybeer' (Haacke & Eiseb 2002). Within the Khoekhoe branch of the KHOE family, cf. also Kora !xaris = 'honeybeer', Xri !xarip = 'honeybeer' (Haacke & Snyman 2019:304); within the Kalahari branch of the KHOE family, cf. also Naro qgari, where the orthography qg indicates the affricate alveolar click [͡ !χ] (Visser 2000:128). Crucial to note is that Smith himself does not conflate karee in its sense of "honey beer" with karee in its sense of Searsia sp., or with krie in its sense of Lycium sp. However, several subsequent authors do. Coetzee, for example, derives the loan !ari for Lycium ferocissimum from !kharib because of 'its ability to ferment honey beer known as !kharri-i' (Coetzee 2010:104), while PZA claims that the loan karee for Searsia lancea "is said to be the original Khoi word for mead" (Stern 2008 However, it might also represent a loan of the more general Nama item !àréb = Searsia sp., conflated with !kharib = 'honey beer'. The precise sequence of conflation and reanalysis is impossible to explicate; all that is clear is that lexical extension has occurred.

Conclusion and Recommendations
Four wholly distinct lexical items, ǁaris (= Lycium sp.), !àréb (= Searsia sp.), |gàrab (= berries of Searsia sp.), and !kharib (= beer), have undergone conflation following their loan into Afrikaans. This is the result primarily of the deletion of phonological and grammatical features that Afrikaans does not support, particularly phonemic tone and click consonants. Further contributing to the confusion is the failure by local linguists and ethnobotanists to familiarise themselves with the appropriate transcription protocols for Khoekhoe-branch languages, or to adequately capture and transcribe the variable click consonants encountered during loan. In other words, the collapse of phonemic contrast has resulted in the loss of lexical contrast, and hence in semantic extension.
Subsequently, karee and its primary referents in genus Searsia have become strongly associated with fermentation and with homemade beer in the minds of speakers, yielding a fascinating example of folk etymology. This illustrates also the futility of attempting to prescribe sources for well-established loanwords. Rather than artificially forcing the modern Afrikaans word karee and all its many regional variants to separate out into their discrete historical targets, it should be acknowledged as a lexical entity in its own right, with its own particular set of referents and connotations.
I close by urging that, when sourcing Afrikaans items to their original Khoekhoe-branch targets, all South African linguists adopt a protocol of specifying the source language rather than referring only to the general language family. Furthermore, the suggested target should always be provided in the appropriate standardised orthography, as would be the norm when drawing comparisons between loaned items and their targets in any other languages. The inclusion of standardised Khoekhoegowab orthography on public-facing signage could help to increase public awareness of and appreciation for the history of the Khoekhoe-branch languages, while also demonstrating the rich interplay of phonological and sociohistorical linguistic phenomena that have developed in both the Khoekhoe-branch languages and Afrikaans over centuries of contact.