Annotated Swadesh wordlists for the Kalahari Khoe group (Central Khoisan family).

Languages included: ǀǀAni [kho-tan], ǀGanda [kho-gnd], Kxoe [kho-huk], Naro [kho-nar], ǂHaba [kho-hab], ǀGwi [kho-gwi], ǀǀGana [kho-gan], Cara [kho-car], ǀXaise [kho-xai], Danisi [kho-dan], Ts'ixa [kho-tsx], Deti [kho-det], Kua [kho-kua], Tsua [kho-tsu], Hiechware [kho-hie].
Reconstruction: Preliminary version available.

DATA SOURCES

General.

Vossen 1988 = Vossen, Rainer (in collaboration with Sabine Neumann, Christina Patriarchi, Margit Rottland, Rainer Spörl, Beate Vagt). Khoe Linguistic Relationships Reconsidered: The Data. In: New Perspectives on the Study of Khoisan. Ed. by Rainer Vossen. Hamburg: Helmut Buske Verlag, pp. 67-108. // Comparative lexicostatistical data on 14 Khoe lects, personally collected by R. Vossen and his team through extensive fieldwork. The entries that represent reconstructible etymologies have been reprinted in [Vossen 1997], but many of the non-etymologized data are unique for this source.

Vossen 1997 = Vossen, Rainer. Die Khoe-Sprachen: Ein Beitrag zur Erforschung der Sprachgeschichte Afrikas. Köln: Rüdiger Köppe Verlag. // A large monograph that describes phonological and grammatical systems for most known Central Khoisan (Khoe) languages. The Proto-Khoe phonological and morphological systems are also reconstructed. In terms of lexical data, the most useful section is the lexical index, containing several hundred cognate sets with the author's reconstructions and all the actual data provided as well (much of it collected by the author himself). Inarguably the best publication on the Khoe family available so far.

Vossen 1992 = Vossen, Rainer. Q in Khoe: borrowing, substrate or innovation? In: African linguistic contributions (Festschrift Ernst Westphal). Ed. by Derek F. Gowlett. Hamburg: Helmut Buske Verlag, pp. 363-388. // The article deals with most known cases of uvular stops and click effluxes in Khoe (Central Khoisan) languages. For poorly described languages, contains some lexical data with these segments that are unavailable in other sources.

I. ǀǀAni.

Vossen 1986 = Vossen, Rainer. Zur Phonologie der ǀǀAni-Sprache. In: Contemporary studies on Khoisan (Festschrift Oswin R. A. Köhler), v. 2. Ed. by R. Vossen & Klaus Keuthmann. Helmut Buske Verlag, Hamburg, pp. 321-345. // Brief description of the phonological system of ǀǀAni, with numerous examples that make it a useful source of lexical data in the absence of a dictionary.

Vossen 2000 = Vossen, Rainer. Khoisan languages, with a grammatical sketch of ǀǀAni (Khoe). In: Areal and genetic factors in language classification and description: Africa south of the Sahara. Ed. by Petr Zima. Lincom Europa, München, pp. 129-145. // The article gives a very brief overview of Khoisan languages as a whole, plus, as an illustration, some grammatical data on ǀǀAni. Of moderate use as a lexical data source.

II. Kxoe.

Main sources

Kilian-Hatz 2003 = Kilian-Hatz, Christa. Khwe-English Dictionary. Köln: Rüdiger Köppe Verlag. // Large dictionary based on the author's own field research as well as previously available data from Oswin Köhler's materials. Includes an English-Khwe index as well.

Köhler 1981 = Köhler, Oswin. Les langues khoisan. In: Les langues dans le monde ancien et moderne. I: Les langues de l'Afrique subsaharienne. Ed. by G. Manessy. Paris: Editions du CNRS, pp. 455-615. // Detailed grammar sketch of Kxoe with special emphasis on verbal morphology. Includes a large amount of illustrative lexical material and an annotated text.

Additional sources

Köhler 1966 = Köhler, Oswin. Die Wortbeziehungen zwischen der Spra\-che der Kxoe-Buschmänner und dem Hottentottischen als geschichtliches Problem. Neue Afrikanische Studien. Ed. by J. Lukas. Hamburg: Deut\-sch\-es Institut für Afrika-Forschung, pp. 144-165. // The paper contains numerous Kxoe lexical items from the author's own field research.

Köhler 1973 = Köhler, Oswin. Neuere Ergebnisse und Hypothesen der Sprachforschung in ihrer Bedeutung für die Geschichte Afrikas. Pai\-deuma, 19-20, pp. 162-199. // The paper contains numerous Kxoe lexical items from the author's own field research.

III. Naro.

Visser 2001 = Visser, Hessel. Naro Dictionary: Naro-English, English-Naro. Ghanzi (Botswana): SIL International. // Large, generally high quality, dictionary of Naro, based on the author's own research and transcribed in the Naro alphabet, specially developed by the author himself.

Barnard 1985 = Barnard, Alan. A Nharo wordlist, with notes on grammar. Durban: University of Natal. // A classified vocabulary of Naro, highly detailed and including the results of the author's own field research as well as data compiled from earlier sources. Unfortunately, the source is not very reliable due to poor quality of transcription.

IV. ǀGui; ǀǀGana.

Nakagawa 1996 = Nakagawa, Hirosi. An Outline Of ǀGui Phonology. African Study Monographs, Suppl. 22, pp. 101-124. // Detailed sketch of the phonology of ǀGui, illustrated with numerous lexical examples. Contains some comparative data on ǀǀGana as well.
Nakagawa 2006 = Nakagawa, Hirosi. Aspects of the phonetic and phonological structure of the Gǀui language. Ph.D. thesis, University of the Witwatersrand, Johannesburg. // A detailed description of ǀGui phonetics and phonology, illustrated with numerous lexical examples. Contains some comparative data on ǀǀGana as well.
Tanaka 1978 = Tanaka, Jiro. A San Vocabulary of the Central Kalahari: Gǀǀana and Gǀwi Dialects. Institute for the Study of Languages and Cultures of Asia and Africa, 1978. // As of now, the only representative dictionary of both ǀǀGana and ǀGwi, unfortunately, marred by extremely poor quality phonetic transcription, reducing it to the status of an auxiliary source.

V. Hiechware.

Dornan 1917 = S. S. Dornan. The Tati Bushmen (Masarwas) and their language. In: Journal of the Royal Anthropological Society of Great Britain and Ireland, 47, pp. 37-112. // Includes a large, but highly inaccurate, sketch of the phonology and grammar of the Hiechware dialect of the Tsua subgroup, accompanied with an extensive vocabulary: for the moment, this is the largest vocabulary of an East Kalahari Khoe language in existence, but, unfortunately, with tremendously poor quality of notation.

NOTES

I. ǀǀAni; ǀGanda; ǂHaba; Cara; ǀXaise; Danisi; Ts'ixa; Deti; Kua; Tsua.

I.1. General.

Significant information on these languages is available only through published results of fieldwork by R. Vossen. Most of these results have been summarized in [Vossen 1988] and [Vossen 1997]; however, this monograph does not contain lexicostatistical data and is insufficient for the construction of complete Swadesh wordlists. For some of the languages (such as ǀǀAni), additional short publications by the same author are available that permit to make the picture more comprehensive. For others (such as ǀGanda), lexicostatistical results will be inevitably skewed due to significant gaps in the wordlists.

I.2. Transliteration.

R. Vossen uses a consistent system of transliteration for all the Khoe languages treated in [Vossen 1997] and various short papers. Minor differences between this system and the UTS are summarized in the following table:

Sound or sound type	Vossen's transcription	UTS representation
Clicks:
Glottalized click	ǀʔ, ǂʔ...	ǀʼ, ǂʼ...
Velar/uvular affricate release	ǀxʔ, ǂxʔ...	ǀkx, ǂkx...
Non-clicks:
Velar ejective stop	kʔ	kʼ
Velar ejective affricate	kxʔ	kx
Voiceless alveolar affricate	ts	c
Voiced alveolar affricate	dz	ʒ
Voiceless palatal affricate	c	ɕ
Voiced palatal affricate	j	ʓ

In the majority of Khoe idiolects, the intervocalic resonant -r- (or -ɽ-) is found in complementary distribution with the word-initial voiced stop d-, which permits Vossen to transcribe the words phonologically, using d as the primary allophone (e. g. write kadi instead of kari). We consistently replace Vossen's intervocalic -d- with -r-, not only for phonetic reasons, but also because the functional status of d- and -r- is significantly different, and it works better to differentiate between these two entities on a graphic level.

II. Kxoe.

II.1. General.

The orthographic spelling "Kxoe" for this language is retained from the works by O. Köhler and R. Vossen (in Kilian-Hatz's definitive vocabulary, it is more appropriately spelled as "Khwe"), in order to better distinguish the actual language from the entire "Khoe (Khwe)" group of languages to which it belongs. (O. Köhler introduced the now abandoned practice of transcribing the velar aspirated stop kʰ as a velar affricate kx).

The basic source for Khoe (Khwe) is [Kilian-Hatz 2003], but it was thought useful to include references to earlier quotations of Kxoe lexical items in works by O. Köhler, who pioneered the serious study of this language in the 1960s / 1970s. In most cases, Köhler's data only differ from the new dictionary in terms of transcription, but there may be occasional phonological discrepancies as well, most notably in the assessment of a particular word's tonal characteristics.

In addition, notes on Kxoe also contain information on the "Buga" or "Buga-Khoe" dialect, taken from [Vossen 1997]. The dialect is very close to Kxoe "proper" lexically, but there is not enough data on it to construct a separate wordlist.

II.2. Transliteration.

The main transcriptional discrepancies between the major sources on Kxoe are as follows:

Sound or sound type	Köhler's transcription	[Kilian-Hatz 2003]	UTS representation
Clicks:
Voiced click	ɡǀ, ɡǂ...	ǀg, ǂg...	ɡǀ, ɡǂ...
Prenasalized click	ɳǀ, ɳǂ...	nǀg, nǂg...	ɳǀ, ɳǂ...
"Fully nasalized" click	ɳǀn, ɳǂn...	nǀ, nǂ...	ɳǀn, ɳǂn...
Velar/uvular affricate release	ǀxʼ, ǂxʼ...	ǀxʼ, ǂxʼ...	ǀkx, ǂkx...
Aspirated release	ǀh, ǂh...	ǀh, ǂh...	ǀʰ, ǂʰ...
Non-clicks:
Velar ejective affricate	kxʼ	kxʼ	kx
Voiceless palato-alveolar affricate	tc	tc	̌č
Voiceless palatal fricative	c	c	ʆ
Glottal stop	ʼ	ʼ	ʔ

III. Naro.

III.1. General.

The basic source for Naro is [Visser 2001], a dictionary that combines representativeness with a generally high quality of phonetic transcription (including tonal notation). Visser's data are complemented by entries from [Vossen 1997] (also highly precise phonetically, but, unfortunately, limited in scope) and [Barnard 1985] (here the problem is the opposite: a very large amount of data, but frequently erroneous phonetic transcription, along with observable semantic inaccuracies from time to time).

There is no reliable information on Naro dialects, but it is clear that there is both phonetic and (very minor) lexical differentiation between the dialects described by Vossen, Visser, and Barnard. In particular, Visser frequently tends to transcribe mid vowels (e, o) where Vossen and Barnard give close vowels (i, u), and often omits nasalization present in Vossen's notes (probably because the forms are denasalized, not because Visser failed to hear the nasalization).

In Visser's dictionary, all the words are given in phonetic transcription and in the "semi-official" Naro alphabet, developed by H. Visser himself. Since the alphabet deviates significantly from the "average" transcriptions of Khoisan phonology, it may be useful to present a complete transliteration table:

Sound or sound type	Visser's orthography	[Vossen 1997]	[Barnard 1985]	UTS representation
Clicks:
Dental clicks	c	ǀ	ǀ ~ ǀk	ǀ
	cg	ǀx	ǀx	ǀx
	cgʼ	ǀxʔ	ǀkʼ	ǀkx
	ch	ǀh	ǀh	ǀʰ
	cʼ	ǀʔ	ǀʼ ~ ǀ	ǀʼ
	dc	ɡǀ	gǀ	ɡǀ
	nc	ɳǀ	nǀ	ɳǀ
Alveolar clicks	q	!	! ~ !k	!
	qg	!x	!x	!x
	qgʼ	!xʔ	!kʼ	!kx
	qh	!h	!h	!ʰ
	qʼ	!ʔ	!ʔ ~ !	!ʼ
	dq	ɡ!	g!	ɡ!
	nq	ɳ!	n!	ɳ!
Palatal clicks	tc	ǂ	ǂ ~ ǂk	ǂ
	tcg	ǂx	ǂx	ǂx
	tcgʼ	ǂxʔ	ǂkʼ	ǂkx
	tch	ǂh	ǂh	ǂʰ
	tcʼ	ǂʔ	ǂʔ ~ ǂ	ǂʼ
	dtc	ɡǂ	gǂ	ɡǂ
	ntc	ɳǂ	nǂ	ɳǂ
Lateral clicks	x	ǀǀ	ǀǀ ~ ǀǀk	ǀǀ
	xg	ǀǀx	ǀǀx	ǀǀx
	xgʼ	ǀǀxʔ	ǀǀkʼ	ǀǀkx
	xh	ǀǀh	ǀǀh	ǀǀʰ
	xʼ	ǀǀʔ	ǀǀʔ ~ ǀǀ	ǀǀʼ
	dx	ɡǀǀ	gǀǀ	ɡǀǀ
	nx	ɳǀǀ	nǀǀ	ɳǀǀ
Non-clicks:
	b	b	b	b
	d	d	d	d
	g	x	x	x
	gh	g	g	g
	h	h	h	h
	k	k	k	k
	kgʼ	kxʔ	kʼ	kx
	kh	kh	kh	kʰ
	m	m	m	m
	n	n	n	n
	s	s	s	s
	t	t	t	t
	th	th	th	tʰ
	ts	ts	ts	c
	tsh	tsh	ts(h)	cʰ
	tsʼ	tsʔ	tsʼ	cʼ
	z	z	z	z
Vowels:
	a	a	a	a
	e	e	e	e
	i ~ e	i	i	i
	o	o	o	o
	u ~ o	u	u	u
Additional phonation:
Nasalization	Ṽ	Ṽ	Ṽ	Ṽ
Pharyngealization	V_	V̰	V̰	V̰
High tone	V́ [transcr. only]	V́	V́	V́
Mid tone	V [unmarked]	[not distinguished]	V [unmarked]	V̄
Low tone	V̀ [transcr. only]	V̀	V̀	V̀

IV. ǀGwi; ǀǀGana.

IV.1. General.

The most accurate data on ǀGwi have been collected up to date by Hirosi Nakagawa; however, only a part of them have been officially published in papers, such as [Nakagawa 1996], and the author's PhD thesis [Nakagawa 2006]. These are used as default sources for the wordlist, but in many cases we had to use R. Vossen's slightly less reliable (but still quite accurate) data from [Vossen 1988] and [Vossen 1997]. For the closely related ǀǀGana language, Vossen's data always serve as the default source (although some forms can also be taken from H. Nakagawa's comparative sections on ǀGana-ǀǀGwi).

The earlier dictionary [Tanaka 1978] is a reasonably representative compendium of lexical data on both of these lects. However, it suffers from notoriously poor quality of transcription (the author himself admits to a lack of phonetic training), with inaccurate representations of click influxes and effluxes, inconsistent notation of the same phonemes by different symbols, no understandable tonal notation, and plenty of transcriptional symbols that are not given an adequate explanation in the introduction to the work. In the notes section, we always quote Tanaka's entries for Swadesh items that are represented in his dictionary, but almost completely refrain from using it as the primary source of data.

IV.2. Transliteration.

Hirosi Nakagawa's phonetic research on ǀGwi has resulted in a serious reinterpretation of some of the phonetic properties of Khoisan clicks, which is appropriately represented in his transcriptional notation, some elements of which, mainly concerning the click "efflux" or "accompaniment" system, are quite innovative. However, in the database, we had to unify his system with R. Vossen's and reduce both to the "common denominator" of the UTS conventions for clicks, so as not to confuse non-specialists in the intricacies of Khoisan phonetics. The major transliteration conventions only concern the "click accompaniment" system and are as follows:

Type of click accompaniment	Nakagawa's transcription	Vossen's transcription	UTS representation
Voiceless velar plosive	kǀ, kǂ...	ǀ, ǂ...	ǀ, ǂ...
Voiced velar plosive	gǀ, gǂ...	ɡǀ, ɡǂ...	ɡǀ, ɡǂ...
Voiceless velar ejective	kǀʼ, kǂʼ...	[no equivalent]	ǀkʼ, ǂkʼ...
Voiced velar nasal	ŋǀ, ŋǂ...	ɳǀ, ɳǂ... ~ ɳǀn, ɳǂn...	ɳǀ, ɳǂ... ~ ɳǀn, ɳǂn...
Aspirated velar nasal	ŋǀh, ŋǂh...	[no equivalent]	ɳǀʰ, ɳǂʰ...
Aspirated velar plosive	kǀh, kǂh...	ǀh, ǂh...	ǀʰ, ǂʰ...
Voiceless uvular plosive	qǀ, qǂ...	ǀq, ǂq...	ǀq, ǂq...
Voiced uvular plosive	ɢǀ, ɢǂ...	[no equivalent]	ǀɢ, ǂɢ...
Voiceless uvular ejective	qǀʼ, qǂʼ...	[no equivalent]	ǀqʼ, ǂqʼ...
Aspirated uvular plosive	qǀh, qǂh...	[no equivalent]	ǀqʰ, ǂqʰ...
Voiceless uvular affricate (fricative?)	qǀχ, qǂχ...	ǀx, ǂx...	ǀx, ǂx...
Affricated uvular ejective	qǀχʼ, qǂχʼ...	ǀxʔ, ǂxʔ...	ǀkx, ǂkx...
Glottal stop	ŋ̥ǀʔ, ŋ̥ǂʔ...	ǀʔ, ǂʔ...	ǀʼ, ǂʼ...

Notes:
1. Nakagawa's "voiceless velar ejective" clicks is a unique series that has not been previously identified by any researcher (and so has no equivalents in R. Vossen's system). UTS differentiation of it from the usual "glottal stop" series of clicks is highly provisional.
2. On the other hand, Nakagawa fails to see the difference between Vossen's "nasal" and "voiced nasal" (i. e. "pre-nasalized" and "fully nasalized") varieties of clicks, marking both as clicks with "voiced velar nasal accompaniment". We preserve Vossen's allegedly phonological distinction in the UTS notation, but one should keep in mind that it has not been independently confirmed.

Quite troublesome is the transcription system in [Tanaka 1978]. No direct transliteration to UTS standards is possible, since the notation is quite chaotic and riddled with errors. For the most part, we therefore leave Tanaka's transcriptions just as they are, with some cosmetic changes to avoid utter confusion, namely: Tanaka's ch = UTS č; Tanaka's sh = UTS š; Tanaka's j = UTS ʓ. In any case, one should always pay primary attention to the way the words are transcribed in Nakagawa's and Vossen's publications, and only use Tanaka's data as supplementary.

V. Hiechware.

V.1. General.

The so-called "Hie-chwa-re" (or, graphically, "Hie-tshwa-re") dialect is one of several East Kalahari Khoe dialects, closely related to the Cua, Kua, and Tsua dialects described in [Vossen 1997], but not necessarily completely identical with any of these. The only description, albeit a fairly extensive one, has been published as [Dornan 1917], and, among other things, includes a large glossary that is currently the single largest source of lexical data on any given East Kalahari Khoe idiom. This makes it into a source of high importance, and one that is also usable for lexicostatistic purposes. Unfortunately, the quality of data notation is very low; in numerous cases not only the phonetic representation, but also the English semantic equivalents adduced by Dornan may be placed under heavy doubt. The resulting wordlist, therefore, despite having very few explicit "gaps", does not pretend to a high degree of accuracy due to flaws that are inherent in the primary data source.

V.2. Transcription.

Much of Dornan's transcription is highly inaccurate. For the GLD, we utilize a more or less direct transliteration method from his own system into the UTS, but it should be kept in mind (especially if the data are to be used for any automatic comparison) that the symbols frequently may not accurately reflect the real phonetics and phonology of the word. For instance, Dornan uses the standard four-symbol click scheme (ǀ, ǂ, !, ǀǀ) throughout, but external comparison of his data shows that he frequently uses one click for another (e. g. alveolar ! instead of palatal ǂ, dental ǀ instead of lateral ǀǀ and vice versa, etc.) or even places click signs where, by all means, there should have not been any clicks in the first place (e. g., ǂ as an interpretation of ejective kx). Even worse is the situation with click effluxes, where the opposition between voice, voicelessness, and glottalization is marked extremely chaotically.

The rules for the transliteration itself are fairly close to the ones employed for [Barnard 1985] for Naro. In a few cases, Dornan makes a strange distinction between a "velar" and a "dental" efflux (e. g. writes ǀt or ǀd instead of ǀk or ǀg). Such an opposition is impossible in Khoisan and may only reflect certain idiosyncratic pronunciation peculiarities; we do not respect it in our transliteration.

VI. Proto-Kalahari Khoe.

The most comprehensive list of reconstructions for Proto-Kalahari Khoe to date (in his version, called "Proto-Non-Khoekhoe") has been published by Rainer Vossen in his state-of-the-art monograph on Khoe languages (Vossen 1997). Outside of a small handful of phonetic details that might require a different solution, the only reason why this list is not fully compatible with the purposes of our database is that it is strictly etymological rather than lexicostatistical - meaning, for one thing, that there is no proper semantic reconstruction involved; for another, that Vossen only includes those etyma for which ample lexical support is found throughout the entire Khoe (or at least Kalahari Khoe) family, and so, for instance, such lexical items as 'all', 'cold', 'root', etc., are not represented by proto-etyma in his work just because their reflexes are unstable in daughter languages.

Because of these reasons, even though our Proto-Kalahari Khoe wordlist is largely based on Vossen 1997, it differs from it in two respects: (a) occasional small amendments to Vossen's phonological reconstruction; (b) inclusion of several items that are at least tentatively reconstructible on the Proto-Kalahari Khoe level if their reflexes are found in at least 2-3 languages from different branches of the family.

The amendments in question are sometimes purely technical (for instance, we have refrained from marking prosody in reconstructions - although in quite a few cases, tonal reflexes are highly consistent across all languages and most likely retain the original situation, in almost just as many cases tonal reflexes are fairly chaotic and require a very thorough analysis, as well as additional checking of potentially mistranscribed data). The main substantial amendments to Vossen's reconstructions are as follows:

(a) reconstruction of a valid phonological opposition between Proto-Kalahari Khoe *e and *ɛ, both as simple vowels and as parts of diphthong codas (*oe / *oɛ). Based on internal data (the presence of such an opposition in several languages) as well as external cognates (in related Khoekhoe languages, usually *e > e, but *ɛ > a);
(b) recognition of pharyngealization (PKK *aˤ, *oˤ, *uˤ), as well as uvular consonants (*q, *qʰ, *G) and uvular click effluxes (*!q, *!qʰ) as features and phonemes possibly reconstructible on the PKK level. Although they are all confined to a select number of languages, and there is frequent unconditioned variation between their presence and absence, neither Vossen nor anybody else has been able to properly explain them as innovations; for this reason, we believe it makes sense to mark their presence within the reconstruction, even if the possibility of areal innovation cannot be ruled out either.

For a detailed list of regular phonetic correspondences between the various Kalahari Khoe languages, [Vossen 1997] should still be considered as the default source. We typically mark any irregular developments in the comment section.

Database compiled and annotated by: G. Starostin (last update: August 2020).