Clicks, genetics, and “proto-world” from a linguistic perspective

Full text

LANGUAGES AND LITERATURES
ULPA
University of Leipzig Papers on Africa
No. 29
Clicks, genetics, and "proto-
world" from a linguistic
perspective
Tom Güldemann
Leipzig 2007

University of Leipzig Papers on Africa
Languages and Literatures Series No. 29
Clicks, genetics, and “proto-world” from a linguistic perspective
Tom Güldemann
Leipzig, 2007
ISBN 3-935999-55-0
Knight et al. (2003) have argued, largely from a genetic perspective, that clicks “may be
more than 40.000 years old” (p.470) and thus “are an ancient element of human
language” (p.471). This has nourished the hypothesis, expressed especially in popular
science, that clicks were a feature of the ancestral mother tongue. The claim by Knight et
al. (2003) is based on the observation that two populations in Africa speaking languages
with click phonemes, namely Hadza in eastern Africa and Ju|’hoan in southern Africa, are
maximally distinct in genetic terms: both Y chromosome and mtDNA data suggest that
the two “are separated by genetic distance as great [as] or greater than that between
any other pair of African populations” (p.464). It is also claimed that the only explanation
for the presence of clicks in the two groups is inheritance from an early common ancestor
language, hence the alleged, very great age of clicks in general. Other explanations for
the clicks of Hadza and Ju|’hoan, in particular independent development and language
contact, are explicitly excluded by the authors.
This paper seeks to demonstrate on the basis of purely linguistic evidence that this view
cannot be accepted: both independent innovation and contact-induced transmission of
clicks are attested. The click system of Hadza in particular will be shown to have a profile
which is quite compatible with an explanation in terms of language contact. The linguistic
evidence thus does not imply that clicks go back to a language spoken at the dawn of
human evolution; there is no good reason to exclude the possibility that the emergence
of clicks in Africa represents a far later episode in the diversification of human speech.
More reliable hypotheses about the early development of language can be reached only
by truly interdisciplinary research in the disciplines concerned, here genetics and
linguistics.
This paper was presented on previous occasions; namely at the Institut für Afrikanistik,
Universität zu Köln (28/11/2003); at the “Jour fixe” series of the Institut für Afrikanistik,
Universität Leipzig (17/12/2003); at the “Geneling” series of the Max Planck Institute for
Evolutionary Anthropology Leipzig (12/03/2004); at the International Conference
“Evolution of Language (EVOLANG)” at Leipzig (02/04/2004); at the “Human Genetics
Seminar” of the Faculty of Health Sciences, University of Cape Town (06/09/2005); and
at the International “Leipzig Spring School on Linguistic Diversity” (23/03/2006).

Clicks, genetics,
and “proto-world” from a
linguistic perspective
Tom Güldemann

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 1
Contents
1. Introduction..........................................................................................................................2
2. The modern distribution and function of clicks................................................................4
2.1. Clicks as non-phonemic speech sounds.......................................................................................4
2.2. Clicks as phonemic speech sounds .............................................................................................. 5
2.3. Summary.................................................................................................................................... 10
3. The origin of clicks in individual languages.....................................................................10
3.1. Independent innovation of clicks............................................................................................... 11
3.2. Contact proliferation of clicks.................................................................................................... 12
3.3. Summary.................................................................................................................................... 16
4. The historical problem of clicks in Hadza .......................................................................16
5. Historical aspects of clicks as a phoneme type ................................................................20
5.1. A cultural advantage of click phonemes? ..................................................................................21
5.2. The stability of click phonemes................................................................................................. 23
5.3. The age of click phonemes......................................................................................................... 24
6. Summary............................................................................................................................. 27
References...............................................................................................................................29
Appendix: Click inventories of selected languages .............................................................33

2 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
1. Introduction
‘Current Biology’, a leading journal of its discipline, has published an article by Knight et al.
(2003) with the title “African Y chromosome and mtDNA divergence provides insight into
the history of click languages”. This article does in fact not only deal with the history of click
languages but involves far-reaching conclusions for the early evolution of human language(s)
before the colonization of man outside Africa and thus concerns linguistics in general.
The authors discuss phylogenetically relevant genetic1 data (regarding both Y
chromosome and mtDNA) of two African populations speaking languages with clicks,
namely Hadza in eastern Africa and Ju|’hoan in southern Africa (sometimes referred to in the
paper by the generic term “San”), and conclude that the two groups are genetically maximally
distinct with respect to the modern diversity of humans:
... San and Hadzabe are among the most highly divergent of African (and therefore global)
population pairs. (p.470)
The separation of the ancestors of click-speaking Hadzabe of Tanzania and click-speaking San of
Botswana and Namibia appears to be among the earliest of human population divergences. (p.469)
From a linguistic perspective, they claim that the only explanation for the presence of
clicks in Hadza and Ju|’hoan is inheritance from an early common ancestor language and
explicitly exclude other explanations for the presence of clicks in the two groups. Following
from the supposed age of the genetic populations, they propose a second major hypothesis
relevant for linguistics:
The deep genetic divergence among click-speaking peoples of Africa and mounting linguistic
evidence suggest that click consonants date to early in the history of modern humans. (p.464)
In more concrete terms, they conjecture that clicks “may be more than 40.000 years
old” (p.470). Under the usual (though not uncontroversial) assumption that all modern
languages descend from a single common ancestor, the simplified hypothesis that clicks were
a feature of the ancestral mother tongue has been expressed especially in popular science (see,
e.g., New York Times of 20/03/2003, Die Zeit of 27/03/2003, Academic Press - Daily
inSCIght of 22/10/2003).
Knight et. al.’s analysis of the genetic data regarding Hadza and Ju|’hoan is not
unproblematic (M. Stoneking p.c.). Moreover, the general scenario for the origin and
proliferation of clicks in Africa in genetic terms must be far more complex than outlined by
1 The term “genetic” will be reserved here for biology. As soon as family relationships among languages are concerned, the
term “genealogical” will be used.

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 3
the authors because the Ju|’hoan population is biologically not representative for all southern
African click-speaking groups (Chen et al. 2000).
In any case, the present paper takes the author’s interpretation of the genetic data as a
given; it provides a more thorough discussion of the LINGUISTIC aspects of the problem.
Even with the genetic part of the argumentation intact, it shows that there is no strong case for
the above hypotheses as far as the history of early human language(s) is concerned.
Knight et al.’s paper contains a number of misinterpretations and misrepresentations of
the available linguistic data.2 A linguistically better informed analysis yields several types of
evidence contradicting their view. In particular, independent innovation and contact-induced
transmission of clicks are more important than assumed and the profile of the Hadza click
system is more compatible with such non-genealogical explanations. More generally, there is
no “mounting linguistic evidence” suggesting that clicks go back to a language spoken at the
dawn of human linguistic evolution. There is in fact a real possibility that the emergence of
phonemic clicks in Africa represents a far later episode in the diversification of human
speech.
This article will thus reiterate a general methodological point, which might appear
trivial, but in practice is not: apparent historical correlations between genetic and linguistic
data should not be addressed from either perspective only; more reliable hypotheses about the
early development of linguistic populations and human language in general can only be
reached by truly interdisciplinary research in the disciplines concerned.
Before discussing the subject matter, the terminology must be clarified, because Knight
et al.’s paper (cf., e.g., p.464, 468-9) is potentially confusing in this respect. A first remark
concerns the term “San (traditional foragers)”, which by and large has the same meaning as
the older, but derogatory Bushmen. In its standard use, it refers to a population of a particular
subsistence mode (at least until fairly recently) in a certain geographical area, namely hunter-
gatherers of southern Africa.3 While at most a convenient entity of cultural anthropology, the
concept is vacuous in terms of linguistic and genetic classification: the different San groups
speak a number of languages comprising inter alia three quite distinct language families (see
Section 2.2) and possess very different genetic profiles, on both an African and global scale.
2 It is unclear to me to what extent the authors on the one hand received expertise feedback from linguists specialized in the
relevant languages and on the other hand integrated it into their argumentation.
3 It is thus comparable to “Aborigines” referring generally to the indigenous hunter-gatherer population of Australia. The
word saa-n (-n is a gender-number suffix for common plural) was originally used by the pastoralist Khoekhoe (for this term
see below) as a generic exonym for hunter-gatherers and literally means ‘foragers’.

4 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
The term “Khwe” is potentially even more misleading, because it can be confused with
Khoe (< khoe ‘person, human being’), which refers to one of the Khoisan language families.
This purely linguistic entity comprises in cultural terms hunter-gatherer and pastoral groups
and in genetic terms population profiles which are typical for southern African and others
which are much less so.
What the authors actually mean with “Khwe (traditional herders)” are the pastoralist
Khoekhoe (= formerly Hottentots). These are a concrete anthropological entity defined by
culture, geography, language, and genetic profile; i.e. they are the pastoral groups of South
Africa and Namibia speaking language varieties of a sub-branch of the Khoe family and are
genetically of the southern African type with some non-southern African admixture.4
Pace Knight et al. (2003: 469), “Khwe” (alias Khoe) and “San” are not complementary
concepts, some ethnic groups self-identify as Khoe AND San, because they call themselves in
their own language Khoen ‘people’ and are/were culturally San, i.e. hunter-gatherers. This
also means that the terms “San”, “Khoekhoe”, and “Khoe” are semantically neither
comparable to nor exclusive of each other and hence cannot be used in a meaningful way for
referring together to what is commonly meant by “Khoisan”.
2. The modern distribution and function of clicks
It is generally assumed that clicks have a very uneven occurrence across the world’s
languages. However, this view applies only to one kind of click use. For a full understanding
of the modern distribution of clicks in human language it is necessary to make a basic
distinction between two employments of clicks, namely as PHONEMIC speech sounds which
distinguish lexical meaning on the one hand and as non-phonemic, PARALINGUISTIC
speech sounds on the other hand.
2.1. Clicks as non-phonemic speech sounds
While clicks as phonemes are indeed alien to the large majority of modern languages, the use
of clicks as a paralinguistic phenomenon is attested far more frequently -- a fact known and
discussed in science for a long time (cf., e.g., Darwin 1872).
4 Khoekhoe-n was originally the pastoralist’s autonym and means literally ‘real people, people of people’. “Khoekhoe” has
two slightly different connotations in linguistics related to the primary sense. On the one hand, it denotes one of the two
branches of the language family Khoe (cf. Voßen 1997). On the other hand, “Khoekhoegowab” refers to a concrete
Khoekhoe variety, which is an official, standardized language in Namibia. This has a more recent and complex genesis and is
thus not only spoken by (former) pastoralists.

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 5
Gil (2005) is a first attempt to a world-wide survey of the distribution of non-phonemic
clicks. He distinguishes three sub-types of such clicks which are: (a) “logical” (= click means
“yes” and/or “no”); (b) “affective” (= click expresses “positive” and/or “negative” attitude; cf.
dental click [|] in English for negative attitude); and (c) “neither (a) nor (b)” (= click is used
for turn-taking, communication with babies and animals, etc.; cf. lateral click [||] in English
for driving horses).
Since paralinguistic phenomena are virtually undocumented in average linguistic
descriptions, Gil’s study is based largely on personal communications. The results of the
survey, which are shown in Map 1, do not provide complete coverage of the globe; in
particular, an area without dots does not imply the absence of paralinguistic clicks.
However preliminary the results are, two conclusions can be drawn with respect to the
present topic: clicks per se are geographically and genealogically widespread across human
languages and, as a consequence, should not be viewed as unusual speech sounds in terms of
production and in-principle usability in language.
Map 1: Clicks in the world’s languages (phonemic clicks exhaustive, non-phonemic clicks
after Gil 2005)

6 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
2.2. Clicks as phonemic speech sounds
Clicks as phonemes, which are those concerned in Knight et al. (2003), show a highly
different distribution when compared to their non-phonemic use in that they are restricted to
just three wider geographical locations: two in Africa and one in Australia. This is also shown
in Map 1.
From all what is known about the present linguistic diversity on earth, Table 1 provides
a complete list of attested languages and language groups with click phonemes.
Language or LANGUAGE FAMILY Area Highest linguistic affiliation
1 all JU-úHÕA (includes “Northern Khoisan”) southern Africa isolate family1
2 all TUU (= “Southern K.”) southern Africa isolate family2
3 all KHOE-KWADI (includes “Central K.”) southern Africa isolate family3
4 Sandawe eastern Africa isolate language (?or to Khoe-Kwadi)4
5 Hadza eastern Africa isolate language5
6 Dahalo (CUSHITIC) eastern Africa Afroasiatic
7 some BANTU (groups K30, R40, S30, S40)6 southern Africa Niger-Congo
8 Damin (speech register of Lardil, TANGKIC) northern Australia Australian
Notes: 1 see Westphal (1974), Sands (2003), Güldemann (forthcoming b), Honken (2006)
2 see Güldemann (2005)
3 see Güldemann (2004), Güldemann and Elderkin (forthcoming)
4 see Elderkin (1986, 1989), Güldemann and Elderkin (forthcoming)
5 see Sands (1998a, b)
6 in K30: Kavango group; R40 = Yei; in S30: Southern Sotho; S40 = Nguni group
Table 1: Attested languages/LANGUAGE GROUPS with click phonemes
Table 1 displays eight independent linguistic lineages; that is, each of these units
represents a separate genealogical group in the sense that it has not (yet) been shown to have a
relative among the attested languages of the world (a lineage can be an isolate language, a
language family, or a yet larger group). This classification has been established according to
commonly accepted linguistic methodology, namely the historical-comparative method.
Greenberg (1963) and others have claimed that the units 1-5 form a genealogical
language group “Khoisan”, and the units 1-3 a lower order group “Southern African Khoisan”

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 7
(cf. Map 2). Both the wide and narrow version of this hypothesis rest on evidence that does
not conform to the standards of historical comparison and diachronic typology and are not
accepted by the majority of Khoisan linguists; they thus have to be rejected for the time being
(cf. Güldemann and Vossen 2000, Güldemann forthcoming a).5
Map 2: Non-Bantu language families with clicks in southern Africa
5 This classification does not imply that some language (group) is not related genealogically to another one. Future research
might well support or newly bring up viable hypotheses on higher order relations of one or the other unit. The notes in Table
1 indicate problematic classifications and promising hypotheses for higher-order classifications including the respective
references. A very “optimistic” guess on “Khoisan” brings down the number of independent lineages to no less than four or
three.

8 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
The above set of languages and groups is not homogeneous regarding the use of click
phonemes. There are differences between individual lineages in terms of the functional load
and the phonotactic employment of clicks.
The functional load of clicks in a language can be ascertained by two parameters: (a) the
complexity of the click phoneme system and (b) the importance of clicks for the distinction of
lexical meaning. The complexity of click systems is measured conveniently by the size of the
segment inventory. Charts of individual click systems which are representative of eight of the
nine lineages in Table 1 are given in the appendix. It can be seen that the size of click
inventories ranges from 3 clicks (in Dahalo) to 83 clicks (in East !Xõo). A summary of
inventory sizes of click phonemes across languages and language groups is given in Table 2.
Inventory size “Khoisan” Bantu Other
Great Ju,
ú
Hõa, most - -
(30 and more) Tuu, most Khoe
Intermediate //Xegwi (Tuu), Khoekhoe Nguni Bantu, Yei -
(between 10 and 30) (Khoe), Sandawe
Small Kwadi (8), Hadza (9) Southern Sotho (5) Dahalo (3), Damin (5)
(10 and less) Kavango Bantu (5)
Table 2: Inventory size of click phonemes across languages and language groups
The importance of clicks in the lexicon can be measured in a language by the relative
frequency of items with and without clicks. Although the available information on this
parameter is still insufficient, a rough cross-language comparison can be achieved on the basis
of various kinds of data such as figures in the published literature, personal communication by
language experts, and estimates from available dictionaries or vocabularies. Since
approximate values are already sufficient for the present purpose, my analysis of dictionaries
has been very crude in that I counted the pages of words with clicks vs. without clicks.6 Table
3 provides a synopsis of the available data.
6 This is possible in most click languages of southern Africa, because clicks there are word-initial (see below).

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 9
Language Language family Clicks Source
1 Ju|’hoan Ju 60% Dickens 1994
2 úHõa - ? -
3 Eastern !Xõo Tuu 69% Traill 1994
4 Standard Khoekhoe Khoe (Khoekhoe) 63% Haacke and Eiseb 2002
Naro Khoe (Kalahari) 40-50% Visser 2001, Barnard 1985
5 Sandawe - 22% Kagaya 1993, Elderkin p.c.
6 Hadza - 15-25% Elderkin 1978: 20, p.c.
7 Dahalo Cushitic <5% Tosco 1991
8 Zulu Benue-Congo ca. 15% Herbert 1990b: 296
9 Damin Tangkic 17% Hale and Nash 1997: 253
Table 3: Approximate proportion of click lexemes in individual languages
However imprecise the figures in Table 3 are, they suffice to give an idea about the
existence of considerable differences across click languages regarding click frequency. In
general, languages from the Ju, Tuu, and Khoe families, located in southern Africa, have a
high proportion of click words in the lexicon (50% and more), while all other languages have
considerably less lexemes with clicks (25% and less).
The lexical frequency of clicks does not necessarily correlate in a language with the
click inventory size. For example, Khoekhoe varieties conform to the general areal trend of a
great importance of clicks for the lexicon in spite of their relatively small segment
inventories. Nevertheless, both measurements can be conflated so that a language can be
assigned to an approximate place on a scalar continuum between high and low functional load
on clicks. Such a summary is given in Figure 1.
HIGH <--------------------------------------------------------------------------> LOW
JU,
ú
Hõa, Sandawe, Yei, Kwadi, Southern Sotho, Kavango
TUU, KHOE Nguni BANTU Hadza BANTU, Dahalo, Damin
Figure 1: Functional load of click phonemes across languages/ LANGUAGE GROUPS

10 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
Languages with click phonemes also differ with respect to the phonotactic
characteristics of clicks, that is, their syntagmatic distribution vis-à-vis other speech sounds
within words. Prototypically clicks are associated with a particular stem structure of lexical
items, namely C(C)VCV (C = consonant, V = vowel, C in parentheses indicates a consonant
cluster), in which clicks are restricted to the first consonant position. This applies to Ju-úHõa,
Tuu, and Khoe-Kwadi.7 Traces of this pattern are found in Sandawe; Damin does not have the
same stem structure but its clicks are at least word-initial. The only exceptions to this property
are Bantu languages, Hadza, and Dahalo, where clicks can also have a medial position.
2.3. Summary
The following can be summarized for the modern distribution and function of clicks:
(1) Clicks as such are common as human speech sounds.
(2) Clicks as phonemes are cross-linguistically quirky (cf. Maddieson 2005).
(3) Click phonemes have a high functional load only in Ju-úHõa, Tuu, and Khoe-Kwadi.
(While these are commonly subsumed under “Southern African Khoisan”, here they
will henceforth be referred to as “core click languages”.)
Clearly, the phenomenal space regarding clicks is limited. However, the above data are
sufficiently diverse to assess Knight et al.’s (2003) claims with more rigor. In the following,
several types of objections will be made which relate to (a) the treatment of non-genealogical
explanations for the origin of clicks (Section 3, and with particular reference to Hadza,
Section 4) and (b) the general properties of clicks as a class of phonemic speech sounds
(Section 5).
3. The origin of clicks in individual languages
There are three basic types of explanations for the question as to how a language or a
population has come to possess a certain linguistic feature. They are: (a) retention from an
ancestor language, (b) independent innovation, and (c) contact. Knight et al.’s (2003) general
linguistic hypotheses rest on the assumption that clicks in modern languages are by default
inherited, inter alia in Hadza and Ju|’hoan, and only in a few cases they are due to language
contact with “genuine” click languages. Hence they can conclude that the origin of clicks
7 Güldemann (2001: 45-7) argues that a large inventory of stem-INITIAL clicks has a considerable importance in most of
these languages for the distinction of lexical meaning.

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 11
goes back to a single historical process; that is, their “mono-genesis” in proto-world or a
similarly ancient language. In the following I will try to demonstrate that the downplaying or
even outright exclusion of explanations for clicks other than inheritance are not at all
supported by the entire range of cross-linguistic data on clicks in human languages.
3.1. Independent innovation of clicks
A first defect of Knight et al.’s (2003) approach is to ignore entirely the possibility of
independent innovation of click phonemes, because it is clearly attested in one case, namely
in Damin. This is a fully functional speech form used in an Australian Aboriginal group by
second-degree male initiates to ritually related community members. The normal linguistic
register is Lardil - a language of the Tangkic family (Non-Pama-Nyungan, Australian). Lardil
and the initiate register Damin are mutually unintelligible, so that the latter can be considered
to a certain extent to be a separate language.
While Damin is a parasitic speech form on Lardil in terms of grammatical structure, it
differs from it radically in lexicon and phonology (Hale 1973: 443). One of the special
phonological properties of Damin is a set of five click phonemes (see Table 8 of the
Appendix). It is clear that these developed as a local innovation associated with the creation
of the Damin register. McKnight (1999: 245) writes:
The Demiin [= Damin] speakers claimed that the language was developed in Dreamtime ... But I
think one can safely conclude that it was initially invented by a few initiated men who consciously
decided to invent a language that would be spoken by knowledgeable men -- that is, men who
were subincised. When that was done is open to conjecture. ...
What is remarkable about Demiin is the extraordinary linguistic insight that the inventors had
about language. They were obviously acutely aware of the sounds and grammar of Lardil. What is
more, they could imagine sounds that do not occur in Lardil or any other language that they were
familiar with in the Gulf area. They systematically used new sounds in Demiin in a logical and
coherent fashion, and they discarded some of the Lardil sounds. By these means, and by
incorporating Lardil grammar with some modifications into Demiin, they invented a language for
which the basics could be learnt in a few sessions but which the uninstructed would find confusing
and unintelligible.
According to N. Evans (p.c.), there is also linguistic evidence for the innovative status
of Damin-typical sounds, including clicks: some of its words can be shown to have been
created by replacing normal consonants of inherited words by new marked speech sounds, for
example, Damin m!ii (m! stands for a nasal labial click [má]) < proto-Tangkic *mi(y)i
‘vegetable food’ and Damin k’uu < proto-Tangkic *kuu ‘eye’.

12 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
Another important fact about Damin clicks is that they “are in all essential respects like
those of the Khoisan languages of southern Africa” (Hale 1973: 443). Among other things,
they occur only in word-initial position -- only a characteristic of the core click languages --
and they display the same basic influx types (dental, alveolar, palatal, labial; only the lateral
click is lacking). The lack of a larger set of so-called effluxes or accompaniments is due to the
restricted system; the only existing nasal accompaniment conforms in fact to typological
expectations in that a prevalence of nasalization is also typical of small click systems in
Africa (cf. Maddieson, Ladefoged and Sands (1999: 87) on the East African click languages).
The only unique feature in Damin is the re-articulated version of a click as a phonemic
segment. An important conclusion from all these facts for the present topic is that independent
click origin is not necessarily detectable by different properties of the relevant sounds.
3.2. Contact proliferation of clicks
A second type of non-genealogical origin of clicks involves contact between populations with
languages that are distinguished by the presence/absence of clicks. Contact provides four
basic scenarios for the proliferation of these sounds across languages or genetic population
types. Each scenario on its own represents an idealization, because more than one scenario
can be involved in a particular case.
First, a population with a click language can change its biological profile through heavy
gene flow; this is excluded by Knight et al. (2003: 470) on account of the genetic data and
will not be discussed any further. A second possibility is that clicks are borrowed by a
population with a click-less language from a click language.
A possibility of click proliferation totally ignored in Knight et al.’s discussion is
language shift; here two further scenarios can be distinguished. On the one hand, a population
can shift from a click language to a click-less language whereby clicks enter the target
language by substrate interference. On the other hand a population can shift from a click-less
language to a click language whereby clicks are retained.
Only click borrowing and click language substratum, but not gene flow and click
language superstratum are associated by language change regarding the presence/absence of
clicks (cf. Thomason and Kaufman 1988). With respect to genetic properties of the
population, a rough probabilistic generalization would be a cline of salience of genetic
change: heavy gene flow would involve the highest degree; a click language substratum is
also likely to leave a trace in the genetic record of the relevant population; the language shift
scenarios from and to a click language are not necessarily associated with easily detectable
genetic change. A summary of the four scenarios is given in Table 4.

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 13
Language change Population-internal
regarding clicks genetic change
(a) Gene flow into a click language NO HIGH
(b) Borrowing from a click language YES ⇑
(c) Language shift from a click language YES ⇓
(d) Language shift to a click language NO LOW
Table 4: Contact scenarios for the proliferation of clicks across languages or populations
As Knight et al. (2003) acknowledge, there are attested cases of contact-induced click
origin. They briefly mention the cases in Bantu languages of southern Africa; another case is
probably the Cushitic language Dahalo in eastern Africa, although here independent origin
cannot be safely excluded either. That these languages have acquired clicks through contact
with click languages can be discerned from several facts. Most importantly, click sounds
cannot be reconstructed to the respective proto-language. Also, language contact is
historically attested or can be assumed, because unrelated click languages are found in the
geographical vicinity. Finally, there is partially direct linguistic evidence that click words are
borrowings from one or the other attested core click language.8
The case of Dahalo is the historically least clear one; regarding a contact scenario, it is a
possible candidate for both the borrowing scenario (b) and the interference-through-shift
scenario (c). Click lexemes cannot be reconstructed to Proto-Cushitic, but there is also no
apparent source in modern click languages; there exist at least other click languages in eastern
Africa so that clicks are likely to have been a wider areal feature in the past.
For southern African Bantu with clicks, contact with core click languages is attested up
to the present and must be assumed to have occurred already in prehistoric times. The contact
scenarios (b) and (c) are both relevant, but can no longer be disentangled.
Regarding the contact proliferation of clicks in general, the evidence from the
secondarily acquired systems in Bantu provides three important conclusions to the effect that
clicks, once borrowed, have a life on their own in the borrowing language and can undergo
changes which are independent from their properties in the original donor languages.
First, click inventories of some Bantu languages are as complex as, or even more
complex than systems of such Khoisan languages as Standard Khoekhoe, Kwadi, Sandawe,
8 This does not imply that all click words in these languages can be traced back to an identifiable source language (see also
below).

14 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
and Hadza. Thus, Yei has 27 click phonemes (cf. Sommer and Voßen 1992) and Nguni
varieties have 15 such segments (Poulos and Msimang 1998).
Second, there exist click accompaniments in Bantu which are not attested in any of the
possible source languages. Such a genuinely Bantu click type is the murmured nasal series in
Zulu (see Table 7 of the Appendix).
Finally, Nguni gives clear indications that a considerable number of click words are not
due to borrowing or substrate interference, but have been innovated on the basis of the
inherited Bantu lexicon. Quite similar to the origin of certain Damin click words, this can be
discerned from two lexical patterns: (a) a click has replaced an original consonant as in -cima
= [|ima] ‘extinguish fire’ from Proto-Bantu *-dima and (b) there exists a doublet of lexical
items distinguished by a slightly different meaning and the opposition click vs. non-click as in
-chela = [|hela] ‘pour ceremonially, asperse’ vs. -thela ‘pour’.
Herbert (1990b) makes the important observation that the salience and degree of
integration of clicks in the phoneme systems of Bantu languages correlates in South Africa
with another feature, the existence and importance of a particular speech form in the linguistic
community. This concerns Nguni and to a lesser extent Southern Sotho. Isihlonipho sabafazi
(= ‘wives’ avoidance language’), as this register is called in Nguni, is part of a wider complex
of avoidance customs which primarily concerns the behavior of married women towards their
male in-laws (see, inter alia, Kunene 1958, Finlayson 1982, Herbert 1990a). Its linguistic
reflex in its most extreme form is a taboo on uttering the names of senior male in-laws
(focusing on the father-in-law) and any of the syllables of which these names are composed.
A major strategy to achieve this goal is the substitution of an original consonant by another
consonant, for example, a click. Herbert (1990b), following Faye (1923-5), argues that the
contact of Bantu speakers with click languages provided them with a welcome addition to the
available inventory of segments, which as a class was recruited in particular for the Hlonipha
register, but also entered the normal language. This is one reason why many click words
cannot be traced back to a non-Bantu source and explains why Southern Nguni (= Xhosa and
Zulu), where Hlonipha is most salient, displays the highest degree of click integration in the
linguistic system.
A general conclusion from the above observations is that in the long run the contact-
induced origin of clicks is, like their independent innovation, not necessarily detectable by the
properties of the click system or by the historical profile of the click lexicon.
A final point regarding contact-induced click origin concerns the idea of language shift.
Such a scenario can involve the proliferation of any linguistic feature into another population,

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 15
including a quirky one like clicks. Important here is that genetic changes can be virtually
absent if the shifting population maintains its distinct identity. Such language shifts seem to
be particularly relevant for hunter-gatherers where contact with other groups often involves
unilateral, socially “upward” gene flow. Classical cases where hunter-gatherer populations
certainly underwent language shift, but kept (initially) fairly separate from their contact
groups are inter alia the Negritos in the Philippines, the Wanniyala-aetto (alias Veddah) in Sri
Lanka, the Pygmies in central Africa, and the Okiek (alias Ndorobo) in eastern Africa.
The possible shift of a population TO a click language is particularly challenging in the
present context. It could confront us with a “perfect crime”, so-to-speak, because it need not
have a considerable effect on the genetic profile of the shifting population AND does not
involve a language change regarding clicks.
One might be tempted to counter that click languages are unlikely targets of language
shift, because they are mostly spoken by hunter-gatherers and, partly as a result of this fact,
are generally associated with low social prestige. However, this assumption is irrelevant in the
present context for two reasons. On the one hand, there is sufficient evidence that languages
of foraging cultures did expand and thereby were targets of language shift just like any other
language; this must have been particularly relevant before the global expansion of food
production.9 On the other hand, click languages are not necessarily associated with a cultural
profile of low prestige; some have been the target of language shift until fairly recently like
Sandawe in eastern Africa (Newman 1994) as well as languages of the pastoral Khoekhoe in
southern Africa (cf. inter alia Traill 1995).10 The modern marginalization of most click
languages seems to be the result of more recent historical processes both on a global scale
(marginalization of hunter-gatherer subsistence) and in Africa in particular (Bantu expansion,
European colonization).
9 See, for example, Ives (1990) and Golla (2000) on the Athabascan expansion in northwestern North America; Evans and
McConvell (1998) and McConvell (2001) on the Pama-Nyungan expansion in Australia; and Bahuchet (1993) on the
westward spread of a Pygmy population within the Congo Basin giving rise to such modern widely dispersed groups as the
Mbuti in northwestern Congo-Kinshasa, the Aka in the southwestern Central African Republic, and the Baka in southeastern
Cameroon and northeastern Gabon.
10 A probable language shift without a change of the genetic profile also seems to be relevant for the Damara in Namibia.
Their original language might have been from the Khoe-Kwadi family, but unlikely the Khoekhoe subbranch; today,
however, they speak the same language as the Nama who are a pastoral Khoekhoe group.

16 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
3.3. Summary
It can be summarized that both types of non-genealogical origin of clicks, i.e. independent
innovation and contact, are attested. These involve 3 of the 8 independent lineages with click
languages listed in Table 1: Bantu of Niger-Congo, Dahalo of Afroasiatic, and Damin of
Australian. The Bantu languages provide in fact more than one case of click borrowing: at
least the clicks in Yei (R40) and Kavango Bantu (K30) on the one hand and in Southern
Sotho (S30) and Nguni (S40) on the other hand are historically unrelated. Hence, there are at
least 4 instances of non-genealogical clicks. The controversial case of Hadza aside, these
account for half of all attested, reasonably independent cases of clicks. Under a more
optimistic view on genealogical relations within “Khoisan” (e.g., 4 instead of 5 groups, in
case Sandawe turned out to be related to Khoe-Kwadi), this would even rise to more than half
of the total. In view of this fact, there is no empirical ground for underestimating or even
excluding non-genealogical explanations for the presence of clicks in a language.
4. The historical problem of clicks in Hadza
In the following section, I will show that a non-genealogical origin of clicks is particularly
relevant for Hadza. For this purpose, it is useful to recapitulate the main features of its clicks.
First, the inventory of 9 segments is clearly in the lowest range of complexity (see Table 2);
within an alleged “Khoisan” group, it is the simplest or second-most simple system (the
Kwadi system with 8 clicks is uncertain). Second, clicks in Hadza have a relatively low
frequency in the lexicon unlike the majority of “Khoisan” languages (see Table 3). Finally,
the phonotactics of clicks in Hadza are not that of core click languages; Hadza is the only
“Khoisan” language without a trace of this feature. All in all, the Hadza click profile is not
reminiscent to cases where clicks are most likely of a genealogical nature.
The problem of clicks in a language should, of course, not only be evaluated in purely
structural-linguistic terms. After all, a click profile like the Hadza one can be reached via two
scenarios: the gradual elaboration of a borrowed or innovated click system or the truncation of
an inherited system that was originally more elaborate and salient. However, the historical
and areal setting of the Hadza, too, does not single out the genealogical scenario against
others.
One non-linguistic factor is the time depth of this population. According to Knight et al.
(2003), the Hadza are one of the oldest genetically distinct groups, involving several tens of
thousands of years. Such a time depth puts serious limits to any attempt to determine the
origin of a certain linguistic feature. In any case, the potential number and complexity of

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 17
historical events which may have occurred within this enormous time span to give rise to the
genetic and particularly to the linguistic profile of the modern Hadza are apparently
underestimated by the authors. Their assumption implies that Hadza had almost the entire
human linguistic history for acquiring clicks within a non-genealogical scenario. This should
be seen against the case of a Bantu language like Yei which just had a time span of ca. 2000
years to develop a click system which is three times as big as that of Hadza.
This implies that one cannot rule out that clicks emerged in Hadza independently in
space and/or time from clicks in the rest of Africa, pace Knight et al. who write (2003: 470):
Two lines of evidence, rarity of clicks in human languages and complexity of the shared repertoire
of clicks and accompaniments, suggest that independent invention of clicks in San and Hadzabe
populations is an unlikely explanation for the observed genetic pattern. With regards to complexity
of click repertoires, each click language includes a particular set of clicks and accompaniments.
Some languages include larger sets than others do, but these sets do overlap. The clicks integral to
Hadzane largely overlap with those clicks integral to Khwe and San languages. The hypothesis of
independent invention, as it applies to the languages of the Hadzabe and San, lacks linguistic
support.
Apart from the fact that click innovation in Hadza is compatible with the genetic data
and the time depth involved, the above statement is linguistically grossly inadequate. Clicks
as such are not rare in human languages (see Section 2.1). The clicks in Hadza deviate in
several ways from those in the core click languages. Finally, Damin shows that clicks DO
emerge independently and then are comparable to African clicks. Moreover, compared to
Hadza, Damin has only a slightly smaller inventory; it has in fact one more click influx
(labial); and its clicks conform with “canonical” phonotactics, while those of Hadza do not.
Entertaining here the idea of independent innovation of clicks for Hadza does only
mean that it must not be excluded offhandedly. Another non-genealogical hypothesis, namely
contact, has in fact a slightly greater probability. This not so much because the case of Damin
-- the only clear case of independent click innovation -- does not involve clicks in a “normal”
speech register, but rather because a contact scenario fits nicely with the linguistic areal
context of Hadza. That is, apart from the fact that clicks are widely available globally and
particularly in Africa as paralinguistic speech sounds, Hadza is spoken in eastern Africa
where clicks are also attested as PHONEMES elsewhere. While this area is not, and with all
likelihood never was, homogeneous in terms of such population criteria as genetic profile,
mode of subsistence, social organization, etc. as well as the genealogical affiliation of the
languages involved, some linguistic features cut across non-linguistic and linguistic
boundaries. For the present discussion it is important that eastern Africa hosts still today three
unrelated click languages which are geographically dispersed and whose click words cannot

18 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
be traced back to a single source. This suggests that the area in the past hosted more click
languages and language groups, which were obliterated with a few exceptions in the course of
later colonization by such non-click lineages as Cushitic, Bantu, and Nilotic. In other words,
clicks were a likely areal feature of eastern Africa in the past.
Knight et al. (2003: 470) also exclude explicitly the possibility that Hadza clicks are the
ultimate result of population contact when they write:
A third a priori explanation of sharing of clicks by San and Hadzabe in the context of genetic
differentiation is linguistic borrowing. Xhosa, for instance, while uncontestedly a Bantu language,
incorporates some clicks borrowed from Khwe or San languages. The extensive population contact
required for such click borrowing, however, leaves a genetic signature through gene flow, as has
been well documented. ... Finally, distortions of the tongue required to produce click consonants
inhibit borrowing of the full repertoire of clicks by adult nonnative speakers. The Nguni language,
for instance, includes a click system that is far less deeply integrated and complex than the systems
of Hadzabe and San languages.
Again, this statement contains several untenable assumptions and assertions, both in
general and for Hadza in particular. One major problem is that they entirely ignore the
possibility that clicks in the Hadza population are the result of language shift. Both shift
scenarios, i.e. clicks were either a feature of the source language or the target language, are
compatible with the genetic record and the linguistic facts regarding the Hadza click system.
For their exclusion of click borrowing, too, there are no empirical grounds. As
discussed above, Hadza is far from having “the full repertoire of clicks” and the production of
its nine distinctive segments does certainly not require any “distortions of the tongue.” Its
click use is overall different from that in core click languages. Also, clicks have been acquired
through contact in at least three independent cases, namely in Dahalo, in Bantu languages of
northern Namibia/Botswana, and in Bantu languages in the east of South Africa. The clicks in
such Bantu languages as Nguni (including Xhosa) and Yei are not “less deeply integrated and
complex” than in Hadza, rather to the contrary. Clicks do not display “canonical”
phonotactics in both Bantu and Hadza, but Bantu displays more complex click inventories
than Hadza and at least one genuine click accompaniment.
Finally, one must reckon with the possibility that the modern system and distribution of
clicks in Hadza are the result of a series of processes whereby a small click inventory
acquired by contact expanded later through purely language-internal changes; the Bantu
evidence shows that such a scenario accounts in fact for systems which are far more complex
than that of Hadza. Overall, Hadza clicks are more similar to those of languages where they
are due to contact rather than to inheritance.

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 19
To be sure, to regard Hadza clicks to be due to language contact is not uncontroversial.
However, it is not compelling when Maddieson, Ladefoged and Sands (1999: 67-8) state that
There is no evidence that clicks are a borrowed feature of the phonology of Hadza; neither the
language-internal distribution of the clicks nor the ability to identify their sources in external loans
points in this direction.
On the one hand, these authors ignore Elderkin (1978: 29-32) who does identify potential
phonotactic evidence in Hadza “for clicks belonging to a secondary [i.e. borrowed] system.”
On the other hand, their argument regarding a lacking source language is invalid, because the
clicks in the Cushitic language Dahalo are certainly secondary and there is no identifiable
click source either.
A local contact scenario for Hadza clicks in eastern Africa leaves open the question
whether click phonemes have emerged in Africa more than once, i.e. in eastern and southern
Africa independently; again, this is theoretically possible. It is also plausible, however, that all
clicks attested in African languages today are due to a single historical event, despite their
modern geographical dispersal. Given that the Bantu expansion into eastern and southern
Africa is only a few thousand years old, it is quite possible that there existed an earlier
linguistic macro-area that reached from eastern Africa to the southern end of the continent.
Under this hypothesis, the Bantu spread would have submerged a linguistic-areal connection
between eastern and southern Africa by causing the extinction of a great many languages,
which may have been of different type and genealogical affiliation, but shared at least some
linguistic features -- inter alia clicks as a common phoneme type.
This scenario is suggested by several synchronic linguistic indications: (a) typological
similarities across such unrelated lineages as South Cushitic, Sandawe, and Hadza in eastern
Africa and the various “Khoisan” language groups in southern Africa, for example, the
presence of lateral consonants, (b) an exclusive typological affinity of Khoe-Kwadi towards
languages in eastern Africa (Heine and Voßen 1981, Güldemann forthcoming c), possibly
involving even a genealogical link between it and Sandawe (Elderkin 1986, Güldemann and
Elderkin forthcoming), and (c) linguistic features in some Bantu languages of eastern and
southern Africa which can be interpreted as the result of interference from such a pre-Bantu
substrate (Güldemann 1999).
It is unwarranted under this scenario that there must be a genetic and/or linguistic
affinity between the Hadza population and those speaking the core click languages in
southern Africa. The geographical distance and the time depth involved would suggest that
the latter are not the direct source of Hadza clicks. These would rather come from a click-
speaking population that existed at a time when the wider geographical area had a higher

20 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
incidence of click languages and of which the linguistic and genetic profiles are no longer
clearly identifiable.
In general, the historical processes that have brought about the modern click distribution
in Africa involve with all likelihood a complex scenario of divergence, convergence, and
obliteration of distinct languages and populations across space and time, including language
shifts that leave no or few linguistic and genetic traces. So even under the assumption that
Knight et al.’s (2003) interpretation of the genetic data is correct, the existence of clicks in
Hadza can be reconciled with various patterns of language contact.
In conclusion, Knight et al.’s (2003: 470) claim that “current genetic and non-genetic
data are inconsistent with three of four a priori explanations for sharing of clicks without
genetic similarity” is certainly inadequate The available data are in fact CONSISTENT with
all explanations considered, i.e. independent emergence, language/population contact, and
inheritance from an early ancestor language, which -- it must be stressed -- has not been
falsified by the above discussion. While any scenario for the origin of clicks in Hadza must be
highly speculative, to the extent that the evaluation of one hypothesis against the other is a
matter of weighing degrees of plausibility, the genealogical explanation is, however, least
compatible with the linguistic evidence.
5. Historical aspects of clicks as a phoneme type
The general hypothesis by Knight et al. (2003) implies two assumptions: (a) clicks as
phonemes have been lost in the great majority of linguistic lineages and (b) clicks were
locally retained in some African languages. They do not provide, however, a plausible
scenario of how/why the status of clicks as “normal” speech sounds in early forms of human
language changed towards their highly marked status within modern linguistic diversity. This
leads to the general question of the stability of click phonemes over time. They only entertain
briefly two hypotheses:
Clicks may have persisted for tens of thousands of years, independently in multiple populations, as
a neutral trait. Alternatively, clicks may have been retained, because they confer an advantage
during hunting in certain environments. (p.464)
Apparently, they follow the general assumption that clicks are inherently instable and
can be lost fairly easily, except if there is a factor countering this tendency. The following
section will address different questions relating to the stability and age of phonemic clicks
arguing that clicks are neither inherently instable nor evidently old.

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 21
5.1. A cultural advantage of click phonemes?
Knight et al. (2003) are, of course, aware of the problem that clicks have been retained for
such a long time, despite their alleged instability, and then with a very skewed geographical
and genealogical pattern, i.e. in just a few African languages. Since the first hypothesis in the
above citation does not solve either of the two puzzles, they propose an alternative hypothesis,
namely that clicks might be advantageous to hunter-gatherers:
So far, we have discussed clicks as if assuming their cultural neutrality. We cannot rule out the
possibility, however, that clicks may have persisted because they confer, in particular
environments, an advantage. Click systems may impact hunting success. During stalking of prey,
Ju|’hoansi revert to a hushed whisper-like communication. Speech is devoiced and consists almost
entirely of clicks. ... Click density of Ju|’hoan allows devoiced communication. While there is little
precedence for phonetic elements conferring a functional advantage, we hesitate to rule out this
possibility without further study. (p.471)
The above hypothesis is also mentioned in the initial summary of the article and is the
only one tackling the above issues. In view of this fact, a wording like “hesitate to rule out
this possibility” is quite an understatement; rather, it must be concluded that another
explanation is not available to the authors. Their hypothesis is, however, highly unlikely for
reasons of both a general and specific nature.
First, it remains to be shown that hunting success is at all decisive for the survival of
foraging communities; just to take the case of the southern African San, it has been shown
that even under a traditional way of life their diet consisted predominantly of plant food.
Another problem is that not all languages/ language groups where clicks can be traced
back to an early stage of linguistic development are demonstrably associated with a hunter-
gatherer culture. For example, the speakers of Proto-Khoe, the ancestor of the majority of
modern languages subsumed under “Khoisan”, seem to have had a partially food-producing
subsistence according to the reconstructed lexicon (cf. Köhler 1986, Voßen 1997).
Also from a general perspective, their hypothesis still does not answer the question why
clicks were retained in just a few hunter-gatherer groups. Presumably, Knight et al. (2003)
imply that the human populations colonizing the world outside Africa already had lost clicks
(hence no inherited clicks outside Africa). But even these out-of-Africa colonizers without
clicks must have been hunter-gatherers. Why then did clicks cease to be advantageous already
for certain forager groups in ancient Africa before the global human expansion?
Finally and most importantly, Knight et al. (2003) build a major hypothesis on a minor
empirical phenomenon. A closer look at precisely this feature, based on a brief literature

22 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
survey and field investigation, reveals that their argument rests on shaky grounds, to say the
least.
The phenomenon they refer to with “hushed whisper-like communication” is called in
Ju|’hoan gòngòma or gùmágùmá (cf. Dickens 1994: 179). It is not only used in hunting, but
represents a fairly ordinary aspect of human languages in general; it is not just “whisper-
LIKE”, it is nothing but whispered Ju|’hoan in the canonical sense of the word, entirely
parallel to whispering in, for example, English. Its articulatory basis, auditory effect, and
efficiency for certain purposes has, pace the authors, nothing to do with the “click density of
Ju|’hoan”; any language allows “devoiced communication”, rendering the consonants to be
more important clues to speech recognition.
Ironically, this acoustic effect of whispering makes the hypothesis particularly unlikely,
because clicks are auditorily the strongest consonant type attested in human languages (cf.,
e.g., Traill 1985: 170ff). In being such high-impact sounds, they appear to be the worst option
available from the sound class of consonants for avoiding disturbance of game through noise.
That mammals do not react indifferently to clicks, as opposed to non-click speech sounds and
other non-linguistic noise (e.g., from a dry leaf creaking under a hunter’s foot), is evident
from the fact that one domain of paralinguistic clicks is in fact precisely the communication
WITH animals (see Section 2.1).
That there is no connection between clicks and hunter success is corroborated by
published information on the context and way of use of gòngòma ~ gùmágùmá in Ju|’hoan
and whispering in other click languages during hunting:
... If they speak at all, it is in muted tones. Most communication is by hand signal. ... (Silberbauer
1981: 209-10)
After reconnoitering, the hunters plan their approach, ... the attack is worked out in a series of
gestures and a whispered word or two. (Silberbauer 1981: 211)
When a few hunters work together, they communicate with hand signals. When they cannot see
one another, they may use bird calls and whistling. Once the animal has been sighted, they may
come together and discuss their strategy in soft whispers. (Liebenberg 1990: 108, see also p.55)
As opposed to one possible reading of Knight et al.’s description, whispering is not used
when the communicating parties of a hunt are separated by some distance whereby clicks
would bridge this space without disturbing the prey. Instead, it is employed as silent FACE-
TO-FACE communication to coordinate the end phase of the hunt, i.e. before actually
approaching the game for the final attack. Compared to whispering in other languages, the use
of whispering involving clicks does not have advantages for the avoidance of noise; on the

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 23
contrary, according to all what is known, one is forced to conclude that it would, if anything,
disfavor hunters.
5.2. The stability of click phonemes
As mentioned above, there is a general assumption that clicks are difficult and hence instable
as a sound class, which would favor their loss over time. This seems to be corroborated at first
glance by the observation that most of the recent, historically observable cases of language
change affecting click languages indeed attest inter alia for click loss (cf. Traill and Vossen
1997), so that one is tempted to assume that click loss is an important aspect in the dynamics
of such languages.
This hypothesis is, however, not at all conclusive from a more general perspective. The
click loss referred to above is described for northern !Xu)u varieties (Ju-úHõa family),
northeastern and eastern Kalahari Khoe languages (Khoe-Kwadi family), and the easternmost
!Ui language ||Xegwi (Tuu family); it is presumably relevant for yet other languages, for
example, Kwadi (cf. Güldemann and Elderkin forthcoming). Looking at the geographical
locations of these languages (see Map 2) it can be observed that all of them are situated at the
periphery of the core click language area in southern Africa and, as a result, share a particular
socio-linguistic setting, as recognized by Traill and Vossen (1997): they have had an intimate
contact history with click-less Bantu languages, which are sociolinguistically more
prestigious and often are the ultimate targets of language shift. Thus, the frequent and
considerable click loss in these core click languages might well be a phenomenon that is
associated with an extra-linguistic factor of language change rather than motivated by
properties inherent to the speech sounds themselves.
This seems in fact to be more likely, given the situation in click languages in relatively
“undisturbed” sociolinguistic environments where clicks are stable sounds. Traill (1974: 39-
40) writes on the so far most complex system of the !Xõo language complex (Tuu family):
It is a striking fact that the !xõ dialect area exhibits such homogeneity at all levels of linguistic
structure, phonetic, phonological, morphological and syntactic, despite its largeness (about 90,000
square miles). This may argue for a relatively recent dispersal of the dialects, but it is impossible
to give substance to this. What one can say, however, is that the homogeneity suggests that an
unexpected degree of stability is characteristic of the language. I say “unexpected” for two
reasons. Firstly, one may be led by the nature of Bushman [= San] society to expect sociolinguistic
conditions which would favour unchecked differentiation; communities are very small, often
socially isolated and there is not the linguistic self-consciousness or literacy that would lead to
standardisation. On the linguistic side the language shows amazing phonetic complexity and one
would expect - although there are not well-developed theoretical grounds for this - this to amount

24 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
to an instability in the sense that it would lead to the rapid rise of many variant pronunciations. It is
just a fact that the number of phonetic parameters a Bushman controls in speech production
represents something approaching a maximum for human linguistic behavior, and I suggest
(theories of markedness and the so-called “principle” of least effort aside) that it is reasonable to
expect such complexity to go hand in hand with instability. But this turns out not to be the case.
Far from variability or simplification being the rule, the maximum phonetic complexity is retained
and lexical items retain fairly standard pronunciations.
There are yet other indications that clicks are not inherently instable and prone to loss.
Looking at the clicks in an average core click language and comparing them with other
consonants, they in fact turn out to be exceptionally “successful” sounds. Within a language,
they normally outnumber other consonants, both in the phoneme inventory and in the lexicon
(see Section 2.2). Cross-linguistically, there is no other major sound type which is subject to
such an extensive series formation (inter alia by the unique possibility of combining with a
second, i.e. the pulmonic, air-stream mechanism) and thus can provide an enormous
multiplicity of lexically distinctive segments (!up to 83 in East !Xõo).
Last but not least, the evidence presented in the previous sections, namely that click
systems can emerge independently, be transferred to click-less languages, and expand over
time, also does not suggest that this sound class is an inherently recessive linguistic feature.
While the available data do not allow one to give a conclusive answer to the problem of
click stability, they certainly justify the null-hypothesis; that is, all things being equal, clicks
as a sound type are not more (in)stable than other infrequent speech sounds. Hence, any
theory which assumes proto-world to have been a click language has to invoke more than just
the alleged “instability” of clicks for motivating that the clicks have mostly been lost.
5.3. The age of click phonemes
Another widely held view on clicks is that there is something inherently “archaic” to these
sounds. In linguistics, too, this idea has an uninterrupted tradition, going back to the first
scientific research on click languages in southern Africa; one can notice a clear conceptual
continuity in the relevant works, for example, from Grolier (1990), over Stopa (1960, 1977)
and Ginneken (1938), back to Bleek (1862). Until fairly recently, this approach could hardly
be separated from the stereotypical idea that the southern African peoples speaking the core
click languages are themselves “archaic” and “primitive”. Compare a representative statement
like that by Jespersen (1922: 418, capitals mine):
First, as regards the purely phonetic side of language, we observe everywhere the tendency to
make pronunciation more easy so as to lessen the muscular effort; difficult combinations of sounds
are discarded, those only being retained which are pronounced with ease ... In most languages now

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 25
only such sounds are used as are produced by expiration, while inbreathed sounds and clicks or
suction-stops are not found in connected speech. In civilized languages we meet with such sounds
only in interjections ... In some VERY PRIMITIVE South African languages on the other hand,
clicks are found as integral parts of words; and Bleek has rendered it probable that in former stages
of these languages they were in more extensive use than now. We may perhaps draw the
conclusion that primitive languages in general were rich in all kinds of difficult sounds.
In modern treatments of the issue, the old age of click phonemes is simply asserted
without giving new and convincing evidence in support. Compare, for example, Kohler
(1998: 267) who states “clicks, although very rare in the world’s languages ... must be
regarded as being among the basic archetypal phonemic elements of sound systems.” The
crux of the matter in this claim is the necessary distinction between phonemic and non-
phonemic clicks. It is indeed probable that clicks are “archetypal” elements of human
communication as a non-phonemic, paralinguistic phenomenon, because this is supported by
cross-linguistic evidence (see Section 2.1). That clicks are archetypal as PHONEMES is a
possible hypothesis, but it is speculative on linguistic grounds.
The available data are also compatible with an alternative hypothesis which is not tied
to a very ancient stage of human language.11 In line with the empirical findings laid out in
Section 2.1 clicks have been widely available as a paralinguistic aspect of communication
throughout human history. It is conceivable that, before this background, they made it very
occasionally from this domain into the phoneme inventory of a language. The major reason
for this assumption is the following empirical fact: there are only two attested cases where
clicks in modern languages cannot at all or not exclusively be ascribed to inheritance or
contact, i.e. where it is certain that clicks are completely or partly independent from clicks in
the core group of modern click languages in southern Africa; they are the click innovation in
Damin and the click proliferation in Nguni Bantu. Significantly, both cases suggest that the
innovative “promotion” of clicks from non-phonemic to phonemic speech sounds involves
more than just their mere availability; that is, they are both associated with a marked
sociolinguistic phenomenon in the form of an avoidance language. Apart from the general fact
that a language with linguistic avoidance is overall more dynamic in historical terms, this
suggests the following, more general scenario for the rare, but possible INDEPENDENT
innovation of click phonemes.
11 There may, of course, be more theoretical possibilities for the origin of clicks. One linguistically relevant scenario is the
emergence of clicks by way of natural sound changes from other more frequent consonants. Since there are as yet no
empirical grounds for this idea, it will not be discussed here.

26 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
Clicks are initially recruited for phonemic purposes in a marked speech register. Here,
they have at least two potential advantages as a new and marked sound class vis-à-vis the
canonical segment inventory: on the one hand, they are emblematic and thus attractive, like
any other “alien” sound type would be for such a marked speech form; on the other hand, they
facilitate the required manipulation of the lexicon by meeting the need for new sounds to
replace sounds to be avoided. As a second step, it can be assumed that such a special register
can affect in the long run the “normal” language; it would serve, so to speak, as a permeable
mediator between the distinct sound inventories of paralinguistic and linguistic
communication. As soon as there are click phonemes in the sociolinguistically unmarked
register, an initially small click inventory can consolidate and even expand gradually, both in
the phoneme system and the lexicon.
In general, the evidently rare emergence of clicks as phonemes could have resulted from
complex, historically incidental interactions of different linguistic and non-linguistic factors.
Since this must only be assumed in an exceptionally small number of cases (possibly only
two: one in Africa and one in Australia), the cross-linguistic rarity of phonemic clicks falls
out naturally from this scenario.
Coming back to the actual age of clicks as a phoneme type, it should be reiterated that
the linguistic evidence by itself does not provide obvious support for an old-age hypothesis.
To a certain extent, this must have been felt by previous scholars tackling the issue in that
they often tied the presence of clicks to the old age of either the relevant linguistic lineage or
the population type. Both solutions are problematic, though.
Linguistically, the time depth involved, namely tens of thousands of years, is
incompatible with the possible historical evaluation of genealogical linguistic entities attested
today; the rigor of even the most ambitious linguistic method presently available fades out
from 10000 years backwards, that is chance, inheritance, and contact can no longer be
securely distinguished in the case of a similarity. In the case at issue, there is no good reason
for associating even the oldest linguistic lineages with clicks with some linguistic entity
spoken, say, 20000, 30000, or 40000 years ago.
From a genetic perspective, the major population in southern Africa associated with
click sounds is indeed also associated with old genetic markers. However, in view of the
above data on non-genealogical click origin and the relevant time depth, the emergence of
click phonemes within this population cannot be tied securely to its origin in genetic terms.
In conclusion, there is no good reason as yet to assume with any confidence that clicks
were among the earliest phonemic speech sounds. The possibility is very real that the

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 27
emergence of clicks as phonemes in Africa represents a far later episode in the diversification
of human speech. It must be kept in mind in this respect that a time depth of several tens of
thousands of years invokes highly different connotations of “old”~”early” vs. “young”~”late”.
If clicks in Africa had an age of, say, 20000 years, they would be a relatively “young”
phenomenon vis-à-vis the identifiable time depths of human genetic profiles; in linguistic
terms, they would be exceptionally “old” in the sense that available methods of this discipline
are incapable of identifying such an early date.
Viewed from a purely synchronic perspective, click phonemes simply represent a
linguistic-typological “quirk”. As such, they have a number of parallel cases in the universe of
attested linguistic features.12 These also warrant historical interpretations, which are likely to
be as complex as that for clicks, but the hypotheses would probably be less spectacular.
6. Summary
The idea that the origin of click phonemes is of the same age as the origin of language, as has
been proposed for a long time and is again entertained by Knight et al. (2003) on the
exclusive basis of genetic data, is a possible hypothesis not falsified by the present discussion.
However, against the unfounded claims of these authors, the available linguistic data do not
single out this hypothesis in favor of other hypotheses. The idea that modern click phonemes
have their ultimate origin in the linguistic feature of a very ancient human language remains
just one among several speculative hypotheses. The above paper seems to be inspired by the
outdated default assumption that linguistic, genetic, and cultural features correlate, and thus
achieves first of all to perpetuate old stereotypes about the African groups speaking the
relevant languages. Unspectacular as the conclusion of this paper may appear, whatever the
genetic relations between the different populations with click languages are, we don’t really
know much more regarding clicks than before.
There is, however, another lesson to be learned from the present problem: the desirable
integration of genetic and linguistic data is potentially confronted with a serious problem,
namely the possible incompatibility of time depths. The enormous time depths that can be
reached by modern genetic research have so far no counterpart in linguistic methodology. It is
also necessary to appreciate the different historical dynamics of the primary research objects
12 For example, there are other rare consonant types whose skewed geographical distribution is surprisingly similar to that of
clicks: labiovelar stops are only found in a large Sub-Saharan African belt and in two small pockets in East Africa and East
New Guinea (see Maddieson 2005); labial flaps are attested in one larger area in Central Africa and in a few isolated
languages in Southeast Africa and Flores (Indonesia) (see Olson and Hajek 2003).

28 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
of the two disciplines: linguistic features, languages, and language groups on the one hand
and genetic features and populations on the other hand. More reliable hypotheses about the
early development of language can be reached only by truly interdisciplinary research in the
disciplines concerned.

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 29
References
Academic Press - Daily inSCIght (22/10/2003). Ancient roots for an African language? (by Caroline
Seydel).
Bahuchet, Serge. 1993. History of the inhabitants of the Central African Rain Forest: perspectives
from comparative linguistics. In Hladik, C. M. et al. (eds.), Tropical forests, people and food:
biocultural interactions and applications to development. Paris: UNESCO and Parthenon, 37-
54.
Barnard, Alan. 1985. A Nharo wordlist with notes on grammar. Occasional Publications 2. Durban:
Department of African Studies, University of Natal.
Bleek, Wilhelm H. I. 1862. A comparative grammar of South African languages, part 1: phonology.
Cape Town/ London: J. C. Juta and Trübner.
Chen, Yu-Sheng et al. 2000. mtDNA variation in the South African Kung and Khwe - and their
genetic relationships to other African populations. American Journal of Human Genetics 66:
1362-1383.
Darwin, Charles. 1872. The expression of the emotions in man and animals. London: John Murray.
Dickens, Patrick J. 1994. English-Ju/’hoan|Ju/’hoan-English dictionary. Quellen zur Khoisan-
Forschung 8. Köln: Rüdiger Köppe.
Die Zeit (27/03/2003). Der erste Zungenschlag: Kommunizierten die Urmenschen schnalzend und
schmatzend? (by Tobias Hürter).
Elderkin, Edward D. 1978. Loans in Hadza: internal evidence from consonants. Occasional Papers 3.
Dar es Salaam: Department of Foreign Languages and Linguistics, University of Dar es Salaam.
Elderkin, Edward D. 1986. Diachronic inferences from basic sentence and noun phrase structure in
Central Khoisan and Sandawe. Sprache und Geschichte in Afrika 7,2: 131-156.
Elderkin, Edward D. 1989. The significance and origin of the use of pitch in Sandawe. Ph. D. thesis:
Department of Language and Linguistic Science, University of York.
Evans, Nicholas and Patrick McConvell. 1998. The enigma of Pama-Nyungan expansion in Australia.
In Blench, Roger and Matthew Spriggs (eds.), Archaeology and language II: correlating
archaeology and linguistic hypotheses. London: Routledge, 174-192.
Faye, C. U. 1923-5. The influence of “Hlonipa” on the Zulu clicks. Bulletin of the School of Oriental
and African Studies 3: 757-782.
Finlayson, Rosalie. 1982. Hlonipha - the women’s language of avoidance among the Xhosa. South
African Journal of African Languages, Supplement 1: 35-60.
Gil, David. 2005. Paralinguistic usages of clicks. In Haspelmath, Dryer, Gil & Comrie (eds.), 572-575.
Ginneken, Jacobus van. 1938. Les clics, les consonnes et les voyelles dans l’histoire de l’humanité. In
Proceedings of the Third International Congress of Phonetic Sciences. Ghent: University of
Ghent, 321-326.
Golla, Victor. 2000. Language history and communication strategies in aboriginal California. In
Languages of the Pacific Rim. Suita, Japan: Faculty of Informatics, Osaka Gakuin University,
vol. 5: 43-64.
Greenberg, Joseph H. 1963. The languages of Africa. Bloomington: Indiana University.
Grolier, Eric de. 1990. Toward a tentative “reconstruction” of homo sapiens sapiens language(s): an
essay in glossogenetics theory. In Koch, Walter A. (ed.), Geneses of language. Bochum
Publications in Evolutionary Cultural Semiotics 11. Bochum: Universitätsverlag Dr. Norbert
Brockmeyer, 135-163.
Güldemann, Tom. 1999. Head-initial meets head-final: nominal suffixes in eastern and southern Bantu
from a historical perspective. Studies in African Linguistics 28,1: 49-91.

30 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
Güldemann, Tom. 2001. Phonological regularities of consonant systems across Khoisan lineages.
University of Leipzig Papers on Africa, Languages and Literatures 16. Leipzig: Institut für
Afrikanistik, Universität Leipzig.
Güldemann, Tom. 2004. Reconstruction through 'de-construction': the marking of person, gender, and
number in the Khoe family and Kwadi. Diachronica 21,2: 251-306.
Güldemann, Tom. 2005. Tuu as a language family. In Güldemann, Tom, Studies in Tuu (Southern
Khoisan). University of Leipzig Papers on Africa, Languages and Literatures 23. Leipzig:
Institut für Afrikanistik, Universität Leipzig, 10-29.
Güldemann, Tom. forthcoming a. Greenberg’s “case” for Khoisan: the morphological evidence. In
Voßen, Rainer (ed.), Problems of linguistic-historical reconstruction in Africa. Sprache und
Geschichte in Afrika 19. Köln: Rüdiger Köppe.
Güldemann, Tom. forthcoming b. Pronominal noun phrases in Tuu and Ju with special reference to
their historical significance. Afrika und Übersee.
Güldemann, Tom. forthcoming c. Typology. In Voßen, Rainer (ed.), The Khoisan languages. London:
Routledge.
Güldemann, Tom and Rainer Vossen. 2000. Khoisan. In Heine, Bernd and Derek Nurse (eds.), African
languages: an introduction. Cambridge: Cambridge University Press, 99-122.
Güldemann, Tom and Edward D. Elderkin. forthcoming. On external genealogical relationships of the
Khoe family. In Brenzinger, Matthias and Christa König (eds.), Khoisan Languages and
Linguistics: the Riezlern Symposium 2003. Quellen zur Khoisan-Forschung 17. Köln: Rüdiger
Köppe.
Haacke, Wilfrid H. G. and Eliphas Eiseb. 2002. A Khoekhoegowab dictionary with an English-
Khoekhoegowab index. Windhoek: Gamsberg Macmillan.
Hale, Kenneth L. 1973. Deep-surface canonical disparities in relation to analysis and change: an
Australian example. In Thomas, Sebeok (ed.), Current trends in linguistics, vol. 11. The Hague:
Mouton, 401-458.
Hale, Kenneth L. and David Nash. 1997. Damin and Lardil phonotactics. In Tryon, Darrell and
Michael Walsh (eds.), Boundary rider: essays in honour of Geoffrey O’Grady. Pacific
Linguistics C136, 247-259.
Haspelmath, Martin, Matthew Dryer, David Gil & Bernard Comrie (eds.). 2005. The world atlas of
language structures. Oxford: Oxford University Press.
Heine, Bernd and Rainer Voßen. 1981. Sprachtypologie. In Heine, Bernd, Thilo C. Schadeberg and
Ekkehard Wolff (eds.) Die Sprachen Afrikas. Hamburg. Buske, 407-444.
Herbert, Robert K. 1990a. Hlonipha and the ambiguous woman. Anthropos 85: 455-473.
Herbert, Robert K. 1990b. The sociohistory of clicks in Southern Bantu. Anthropological Linguistics
32,3/4: 295-315.
Honken, Henry. 2006. Eastern úHoã as an NK language. Unpublished manuscript.
Ives, John W. 1990. A theory of Northern Athabascan prehistory. Boulder: Westview Press.
Jespersen, Otto. 1922. Language, its nature, developments and origin. New York: Henry Holt.
Kagaya, Ryohei. 1993. A classified vocabulary of the Sandawe language. Asian and African Lexicon
26. Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa, Tokyo
University of Foreign Studies.
Knight, Alec et al. 2003. African Y chromosome and mtDNA divergence provides insight into the
history of click languages. Current Biology 13: 464-473.
Kohler, Klaus J. 1998. The development of sound systems in human language. In Hurford, James R.,
Michael Studdert-Kennedy and Chris Knight (eds.), Approaches to the evolution of language:
social and cognitive bases. Cambridge: Cambridge University Press, 265-278.

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 31
Köhler, Oswin. 1986. Allgemeine und sprachliche Bemerkungen zum Feldbau nach Oraltexten der
Kxoe-Buschleute. Sprache und Geschichte in Afrika 7,1: 205-272.
Kunene, Daniel P. 1958. Notes on Hlonepha among the Southern Sotho. African Studies 17: 159-182.
Liebenberg, Louis. 1990. The art of tracking: the origin of science. Cape Town: David Philip.
Maddieson, Ian, Peter Ladefoged and Bonny E. Sands. 1999. Clicks in East African languages. In
Finlayson, Rosalie (ed.), African Mosaic: Festschrift for J. A. Louw. Pretoria: University of
South Africa, 59-91.
Maddieson, Ian. 2005. Presence of uncommon consonants. In Haspelmath, Dryer, Gil & Comrie
(eds.), 82-85.
McConvell, Patrick. 2001. Language shift and language spread among hunter-gatherers. In Panter-
Brick, Catherine, Robert H. Layton and Peter Rowley-Conwy (eds.), Hunter-gatherers: an
interdisciplinary perspective. Biosocial Society Symposium Series 13. Cambridge: Cambridge
University Press, 143-169.
McKnight, David. 1999. People, countries, and the rainbow serpent: systems of classification among
the Lardil of Mornigton Island. Oxford: Oxford University Press.
Nakagawa, Hirosi. 1996a. A first report on the click accompaniments of |Gui. Journal of the
International Phonetic Association 26,1: 41-54.
Nakagawa, Hirosi. 1996b. An outline of |Gui phonology. African Study Monographs, Supplement 22:
101-124.
Newman, James L. 1994. Reconfiguring the Sandawe puzzle. In Sprache und Geschichte in Afrika
12/3. Köln: Rüdiger Köppe, 159-170.
New York Times (20/03/2003). In click languages, an echo of the tongues of the ancients. (by Nicholas
Wade).
Olson, Kenneth S. and John Hajek. 2003. Crosslinguistic insights on the labial flap. Linguistic
Typology 7,2: 157-186.
Poulos, George and Christian T. Msimang. 1998. A linguistic analysis of Zulu. Cape Town: Via
Afrika.
Sands, Bonny E. 1998a. Eastern and southern African Khoisan: evaluating claims of distant linguistic
relationships. Quellen zur Khoisan-Forschung 14. Köln: Rüdiger Köppe.
Sands, Bonny E. 1998b. The linguistic relationship between Hadza and Khoesan. In Schladt, Mathias
(ed.), Language, identity, and conceptualization among the Khoisan. Quellen zur Khoisan-
Forschung 15. Köln: Rüdiger Köppe, 265-283.
Sands, Bonny E. 2003. Northern Khoisan reconstruction and subgrouping. Paper presented to the
International Symposium on Khoisan Languages and Linguistics in Memory of Jan W. Snyman
(4-6 January , 2003, Riezlern, Germany).
Sands, Bonny E., Ian Maddieson and Peter Ladefoged. 1996. The phonetic structures of Hadza.
Studies in African Linguistics 25,2: 171-204.
Silberbauer, George B. 1981. Hunter and habitat in the central Kalahari Desert. Cambridge:
Cambridge University Press.
Sommer, Gabriele and Rainer Voßen. 1992. Schnalzwörter im Yei (R.41). Afrika und Übersee 75: 1-
42.
Stopa, Roman. 1960. The evolution of click sounds in some African languages. Zeszyty Naukowe,
Rozprawy i Studia 25. Kraków: Uniwersytet Jagiellonski.
Stopa, Roman. 1977. The evolution of click sounds in West-Sudanic. In Benzing, Brigitte, Otto
Böcher and Günter Mayer (eds.), Wort und Wirklichkeit: Studien zur Afrikanistik und
Orientalistik, Band 2: Linguistik und Kulturwissenschaft (Eugen Ludwig Rapp zum 70.
Geburtstag). Meisenheim am Glan: Anton Hain, 65-88.

32 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
Thomason, Sarah G. and Terrence S. Kaufman. 1988. Language contact, creolization and genetic
linguistics. Berkeley: University of California Press.
Tosco, Mauro. 1991. A grammatical sketch of Dahalo. Kuschitische Sprachstudien 8. Hamburg:
Helmut Buske.
Traill, Anthony. 1974. The compleat guide to the Koon. Communications 1. Johannesburg: African
Studies Institute, University of the Witwatersrand.
Traill, Anthony. 1985. Phonetic and phonological studies of !Xóõ Bushman. Quellen zur Khoisan-
Forschung 1. Hamburg: Helmut Buske.
Traill, Anthony. 1994. A !Xóõ dictionary. Quellen zur Khoisan-Forschung 9. Köln: Rüdiger Köppe.
Traill, Anthony. 1995. The Khoesan languages of South Africa. In Mesthrie, Rajend (ed.), Language
and social history: studies in South African sociolinguistics. Cape Town: David Philip, 1-18.
Traill, Anthony and Rainer Voßen. 1997. Sound change in the Khoisan languages: new data on click
loss and click replacement. Journal of African Languages and Linguistics 18,1: 21-56.
Visser, Hessel. 2001. Naro dictionary: Naro-English|English-Naro. D’Kar: Naro Language Project.
Voßen, Rainer. 1997. Die Khoe-Sprachen: Ein Beitrag zur Erforschung der Sprachgeschichte Afrikas.
Quellen zur Khoisan-Forschung 12. Köln: Rüdiger Köppe.
Westphal, Ernst O. J. 1974. Notes on A. Traill: “N4 or S7”. African Studies 33,4: 243-247.Abstract

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 33
Appendix: Click inventories of selected languages
(48) Lt Dt Al Pl Lb
Plain Ñ | ! ú
Voiced gÑ g| g! gú
Plain + Gl Ñ’ !’ |’ ú’
Plain + As Ñ’h |’h !’h ú’h
Voiced + As nÑh n|h n!h núh
Plain + /x/ Ñx |x !x úx
Voiced + /x/ gÑx g|x g!x gúx
Plain + /kx’/ Ñk |k !k úk
Voiced + /kx’/ gÑk g|k g!k gúk
Plain + /kh/ Ñh |h !h úh
Voiced + /kh/ gÑh g|h g!h gúh
Plain nasal nÑ n| n! nú
Table 1: The click system of Ju|’hoan (Ju-úHõa) (after Dickens 1994)
(83) Lt Dt Al Pl Lb
Plain Ñ | ! ú á
Voiced Ñg |g !g úg ág
Plain + Gl Ñ’ |’ !’ ú’ á’
Plain + As Ñh |h !h úh áh
Voiced + As gÑqh g|qh g!qh gúqh gáqh
Plain + /x/ Ñx |x !x úx áx
Voiced + /x/ gÑx g|x g!x gúx gáx
Plain + /kx’/ Ñkx’ |kx’ !kx’ úkx’ ákx’
Voiced + /kx’/ gÑkx’ g|kx’ g!kx’ gúkx’ gákx’
Plain + /qh/ Ñqh |qh !qh úqh
Voiced + /qh/ GÑqh G|qh G!qh áqh
Plain + /q/ Ñq |q !q úq áq
Voiced + /q/ ÑG |G !G úG áG
Plain + /q’/ Ñq’ |q’ !q’ úq’ áq’
Plain nasal Ñn |n !n ún án
Voiceless nasal Ñn8 |n8 !n8 ún8 án8
Plain nasal + Gl ’Ñn ’|n ’!n ’ún ‘án
Table 2: The click system of East !Xõo (Tuu) (after Traill 1985, 1994)

34 University of Leipzig Papers on Africa, Languages and Literatures, No. 29 2007
(52) Lt Dt Al Pl Lb
Plain kÑ k| k! kú
Voiced g„ g| g! gú
Plain + Gl N•Ñ’ N•|’ N•!’ N•ú’
Plain + As N•Ñh N•|h N•!h N•úh
Plain + /x/ qÑX q|X q!X qúX
Plain + /kx’/ qÑX’ q|X’ q!X’ qúX’
Plain + /k’/ kÑ’ k|’ k!’ kú’
Plain + /kh/ kÑh k|h k!h kúh
Plain + /q/ qÑ q| q! qú
Voiced + /q/ GÑ G| G! Gú
Plain + /qh/ qÑh q|h q!h qúh
Plain + /q’/ qÑ’ q|’ q!’ qú’
Plain nasal NÑ N| N! Nú
Table 3: The click system of G|ui (Khoe-Kwadi) (after Nakagawa 1996a, b)
(15) Lt Dt Al Pl Lb
Plain Ñ | !
Voiced gÑ g| g!
Plain + Gl Ñ’ |’ !
Plain + As Ñh |h !h
Plain nasal nÑ n| n!
Table 4: The click system of Sandawe (after Elderkin 1989: 37)
(9) Lt Dt Al Pl Lb
Plain kÑ k| k!
Plain + Gl NÑ’ N|’ N!’
Plain nasal NÑ N| N!
Table 5: The click system of Hadza (after Sands, Maddieson and Ladefoged 1996: 173)

Tom Güldemann, Clicks, genetics, and “proto-world“ from a linguistic perspective 35
(3) Lt Dt Al Pl Lb
Plain N|
Voiceless N•|
Voiceless labialized N•|w
Table 6: The click system of Dahalo (after Maddieson, Ladefoged and Sands 1999: 66)
(15) Lt Dt Al Pl Lb
Plain x c q
Murmured gx gc gq
Plain + As xh ch qh
Plain nasal nx nc nq
Murmured nasal ngx ngc ngq
Table 7: The click system of Zulu (Bantu) (after Poulos and Msimang 1998: 481)
(5) Lt Dt Al Pl Lb
Plain n! rn! m!
Plain + Re nh!2 n!2
Table 8: The click system of Damin (after Hale and Nash 1997: 251)
Click symbols are taken over from the respective source. In spite of the considerable
orthographic differences involved, the clicks themselves are arguably comparable in phonetic-
phonological terms (hence the largely identical labels in the headings of columns and lines).
The abbreviations are: Al alveolar, As aspiration, Dt dental, Gl glottalization, Lb labial, Lt
lateral, Pl palatal, Re rearticulation.

ULPA
University of Leipzig Papers on Africa
Languages and Literatures Series
Edited by H. Ekkehard Wolff
No. 10
H. Ekkehard Wolff (ed.)
Contributions to Bantu Lexicography, 1999, 17 pp. (Є 4,-), ISBN 3-932632-39-7
No. 11
Zakaria Fadoul Khidir
Lexique des plantes connues des beri du Tchad, 1999, pp. 35 (Є 4,-), ISBN 3-932632-40-0
No. 12/13
Mbai-yelmia Ngabo Ndjerassem
Phonologie du Ngambai, Parler de Benoye (Tchad), 2000, pp. 74 (Є 10,-),
ISBN 3-932632-79-6
No. 14
Gerald Heusing
The Classification of Kumam within Nilotic, 2000, pp. 22 (Є 5,-), ISBN 3-932632-80-x
No. 15
Constanze Schmaling
Modalpartikeln im Hausa, 2001, pp. 56 (Є 5,-), ISBN 3-932632-92-3
No. 16
Tom Güldemann
Phonological regularities of consonant systems across Khoisan lineages, 2001, pp. 50 (Є 5,-),
ISBN 3-932632-96-6
No. 17
Zakaria Fadoul Khidir
Lexique des animaux chez les Beri du Tchad, 2002, pp. 72 (Є 5,-), ISBN 3-935999-00-3
No. 18/19 Christfried Naumann
Vergleich demonstrativer Formative ausgewählter Berbersprachen, 2002, pp. 76 (Є 10,-),
ISBN 3-935999-01-1
No. 20
Mohammed M. Munkaila
On Double Objects Constructions in Hausa, 2004, pp. 25 (Є 5,-), ISBN 3-935999-24-0
No. 21
Girma A. Demeke & Ronny Meyer
Die unauffindbare Nadel. Amharisch – deutsche Lesematerialien, 2004, pp. 51 (Є 5,-),
ISBN 3-935999-28-3
No. 22
Andrew Haruna
An Appraisal of British Colonial Language Policy and the Obstacles to the Ascendancy of Hausa in
Education, 2004, pp. 37 (Є 5,-), ISBN 3-935999-29-1
No. 23
Tom Güldemann
Studies in Tuu (Southern Khoisan), 2005, pp. 30 (Є 5,-), ISBN 3-935999-380
No. 24
Gerald Heusing (Hrsg.)
Aspekte der linguistischen und kulturellen Komplexität Ugandas, 2005, pp. 102 (Є 10,-),
ISBN 3-935999-43-7
No. 25
Joachim Crass, Girma A. Demeke, Ronny Meyer & Andreas Wetter
Copula and Focus Constructions in selected Ethiopian Languages, 2005, pp. 36 (Є 5,-),
ISBN 3-935999-46-1
No. 26
Ludwig Gerhardt
Swahili – eine Sprache, zwei Schriften, 2005, pp. 16 (Є 5,-), ISBN 3-935999-47-X
No. 27
Ludwig Gerhardt
Das Amharische, 2006, pp. 19 (Є 5,-), ISBN 3-935999-48-8
No. 28
Christina D. Schmidt
My Church – My Language? Language attitudes and language policy in a South African church, 2006,
pp. 43 (Є 5,-), ISBN 3-935999-54-2
No. 29
Tom Güldemann
Clicks, genetics, and “proto-world” from a linguistic perspective, 2007, pp. 35 (Є 5,-),
ISBN 3-935999-55-0
Orders to:
Institut für Afrikanistik, Universität Leipzig, Beethovenstr. 15, D-04107 Leipzig, Germany
www.uni-leipzig.de/afrikanistik/

ULPA
University of Leipzig Papers on Africa
Institut für Afrikanistik
Universität Leipzig
Beethovenstraße 15
D-04107 Leipzig
Germany
Tel. ++49-(0)341-9737030
Fax: ++49-(0)341-9737048
Email: mgrosze@uni-leipzig.de
Website: http://www.uni-leipzig.de/~afrika/