Kora: A Lost Khoisan Language Chapter 3: The Sounds of Kora

From the book: Kora: A Lost Khoisan Language of the early Cape and the Gariep by Menán du Plessis

There will be many readers who would like to have a sense of how the Kora narratives would have sounded when they were originally delivered by speakers of the language. As the following notes will show, it would not be too far off the mark to assume pronunciations of individual words that are for the most part reasonably close to those of modern South African Nama and Namibian Khoekhoe – but with a few notable exceptions. Many of the examples given in this chapter are linked in the electronic version to illustrative sound files, which should help the reader to achieve a richer understanding of the Kora system of sounds. (The ‘clickability’ of a given word is indicated typographically.) This chapter, which does not assume any prior linguistic knowledge, should help readers to become familiar with the main conventions used in writing Kora, particularly where these differ from those used in the Namibian orthography for the Nama, Dama, and Hai?Om dialects. The chapter also aims to provide a basis for understanding the many variations and discrepancies in spellings that will be seen in the heritage texts.

As we have noted previously, Kora did not receive an official orthography, which is to say a formally recognised set of spelling and writing conventions. For the purposes of the present work, a set of ‘standard’ conventions has been adopted, where these are based on those used by Carl Meinhof for his Glossary of the language. [1] In the discussion that follows, any letter or letter sequence printed in bold, as for example p or kh, is intended to reflect the standardised written symbol for a given phoneme. (A phoneme, written for example as /p/ or /kh/, is a unit of sound in a given language where its phonetic distinctness is associated in that language with the marking of a difference in meaning, as in English ‘pat’ versus ‘cat’, or ‘pat’ versus ‘bat’. Any predictable phonetic variants of a sound that do not affect meaning are not treated as separate phonemes, and are written between square brackets, as for example [kx].)

It should be kept in mind that, like all Khoisan languages and indeed many other languages of Africa as well as Asia, Kora is a tone language, which is to say that it makes use of contrastive tone melodies to signal differences in meaning between two or more words that might otherwise seem identical. The tones are not marked in the examples that follow, for the simple reason that we do not have enough reliable data (either from older sources or our own material) to offer them with certainty. The Kora tone melodies are briefly discussed in the final section.

Older resources consulted include Douglas Beach’s seminal work on the phonetics of the Khoekhoe languages, [2] and the brief commentaries on Kora phonetics by scholars such as Carl Meinhof [3] and Louis Maingard. [4] A number of new observations have been made on the basis of our own recordings, guided by current insights arrived at in recent years through phonetic studies of other Khoisan languages in general, by scholars such as Anthony Traill, [5] and Hirosi Nakagawa. [6] The basic terminology of the descriptions is drawn largely from Peter Ladefoged’s A Course in Phonetics, [7] which throughout its many editions has given generations of students their fundamental knowledge of phonetics. [8] Since the present work is primarily intended, however, for a broad and non-specialist readership, technical terms will be introduced only sparingly. Where their use is unavoidable, it is hoped that their meaning will become clear from the context.

The example words given throughout this chapter include many loanwords. These will not always be specially indicated, [9] but readers from southern Africa will frequently notice them. Some of these words have become an integral part of the Khoekhoe varieties, and in a few cases, they have been reconstructed for the hypothetical ancestral language. [10]

3.1 Vowels and diphthongs

(i) Vowels

Kora has a set of five plain vowels, written i, u, e, o and a. The first two (phonetic /i/ and /u/) more or less resemble the vowels in English BEET and BOOT respectively, as pronounced by a speaker of South African English – although they may be slightly higher in Kora; while the sound represented by the letter a (phonetic /a/), more or less resembles the sound of ‘a’ in BATH as pronounced by a speaker of South African English. The mid front and back vowels e (phonetic /?/) as in BET and o (phonetic /?/) as in BOUGHT each have a higher (closer) variant (phonetic [e] and [o]), which typically occurs before a following front vowel such as i or e, or else before a nasal. [11] The low central vowel a has a similar tendency to raise in such environments. Since the occurrence of these variants (allophones) is predictable, they need not be counted as separate phonemes, or distinguished in the spelling. [12] It is not uncommon, nevertheless, to find alternate spellings in the older literature where the raised variants are specifically indicated. Carl Meinhof, for example, has ‘kx’ummi’ as a variant of kx’ommi ‘house’, while Lucy Lloyd has ‘|hummi’ for |hommi ‘cloud, sky’ and ‘surep’ (with masculine ending) for sores ‘sun’. The mid central vowel [?] (as in English ‘the’, or Afrikaans se ‘his, her, its’) was occasionally produced by our two consultants as a variant of a, as in [d?nis] for danis ‘honey’, or even as a variant of ae (phonetic /ae/), as in [|’?s?n] for |’aesen ‘be sick’. It is not a contrastive sound in its own right.

The language also has a parallel set of nasalised vowels, where the nasalisation is indicated by means of the tilde, as i, u, and ã. (In the standard spelling conventions for Namibian Khoekhoe, nasalisation is shown by the circumflex, as for example, ‘â’.) Beach did not include ? in the set of nasalised vowels for Kora, and the sole instance we have found is probably a variant of i. The nasalised õ is similarly borderline, and may simply be a variant of u. (In Namibian Khoekhoe, the vowels represented by the letters ‘e’ and ‘o’ do not occur with nasalisation.)

Following a nasal vowel, an intrusive nasal segment is sometimes heard, which may be /m/, /n/ or the velar nasal /?/, depending on whether the following sound is bilabial, alveolar, or velar. The missionary Carl Wuras indicated such sounds sporadically in his Kora Vocabulary, and they were also noted by Jan Engelbrecht. [13] Douglas Beach [14] doubted the assertions of these authors – and yet intrusive nasals of this kind are clearly audible in some of our own recordings, and the fact that they are homorganic with (that is, have the same place of articulation as) the following segment suggests that they are genuinely interpolated segments.

Examples include mu!ã(m)b   ‘head’, ?’ãi(n)dje   ‘they (3rd person feminine plural)’ and !’ã(ng)gub   ‘fight’. Since the nasal insertions are more or less predictable and not semantically significant, they need not be indicated in the spelling. (It was noted in an earlier chapter that nasal intrusions of this kind have led to a characteristic dialectal feature, involving the subsequent assimilation of the masculine singular suffix –b to the inserted nasal as –m.)

In addition to a contrast between oral and nasal production, some of the vowels in Kora (as in most Khoisan languages) may be lengthened. [15] (In Namibian Khoekhoe, long vowels are indicated by means of the macron, as ‘a’.) Several of the older authors indicate the long vowels of Kora by means of a colon (as for example, ‘a:’), but since this can lead to confusion with ordinary punctuation marks, we have chosen to follow the Namibian convention and use the macron instead. Nasalised vowels are always long, and because of this predictability, it would be redundant to indicate the feature in the spelling. [16] In our reproductions of the heritage texts and in the Dictionary, we have usually preserved any indications of vowel length that may have been given in the original sources. The reality, though, is that such indications were often only sporadically made, and were not always consistent. In the case of transcriptions made by Lucy Lloyd and Carl Wuras, it seems that their occasional use of a doubled consonant (as in the word ‘torro’ recorded by Wuras for thoro ‘scatter, sow, strew’) was intended to reflect a shortened preceding vowel. As a basic rule of thumb, the vowel in a word (other than a grammatical morpheme) that consists of just a consonant followed by a vowel (such as bi   ‘suckle’) is typically long, while the first vowel in a word that either has a medial consonant (such as koba ‘speak’), or ends in a nasal (such as xon   ‘grind’), is most often short. There are many exceptions to this basic principle, though, where these frequently seem to occur in cases where a word is reduplicated or compounded, or where it was perhaps originally so, but has subsequently undergone contraction.

Figure 3.1 Sketch of the vocal tract, showing the areas alluded to in Table 3.1

The following sets of words illustrate the different vowels. The best way to form an accurate impression of the Kora vowels will be to listen to the recorded examples in the online edition. Note that in Kora, as in Nama, a vowel that occurs at the beginning of a word (as in arib   ‘dog’) is typically produced with a glottal stop onset, as for example, [?a]. Since it is predictable, this fairly common feature, which is found also in Sotho-Tswana languages as well as many varieties of English, need not be reflected in the spelling. The endings –b and –s seen in some examples are the 3rd person singular suffixes of the masculine and feminine genders respectively. (These will be discussed in the chapter on the structures of the language.)

i, i (Phonetic /i/)

       bib        ‘milk’  

       di        ‘do’ [sõse di ‘do it quickly’]

       disi        ‘ten’

       ib        ‘father’ [s? ib ‘his father’]

       isa        ‘good, pretty, nice’

       mi        ‘say’ [tita ke n’ mi ‘I am saying’]  

       ti        ’ask’

e, (?) (Phonetic /?/)

       be        ‘go away’ [ube ‘take away’]

       he        ‘this’

       (t?)        ‘ask’ (variant of ti)

u, u (Phonetic /u/)

       gus        ‘ewe’  

       sub        ‘pot, pipe’ [?na sub ‘that pipe’]  

       tus        ‘rain’

       u        ‘take’ [ube ‘take away’]

       mu        ‘see’ [mu ti ke ‘I see’]

o, (õ) (Phonetic /?/)

       ho       ‘find, get’

       koro        ‘five’

       o        ‘if, when’

       (sõse)       ‘quickly’ [sõse di        ‘do quickly’] (Nama sûxase)

a, ã (Phonetic /a/)

       a        grammatical morpheme, hortative (‘let’)

       a        ‘be’

       ba        'tread, step’

       ha        ‘come’

       haka        ‘four’

       hã        ‘stay, remain’

       mã        ‘give’ [mã te ‘give me’]

       mã        ‘stand’

Old records show an occasional tendency for o to be pronounced as a, as in the case of gamas for gomas ‘cow’.

(ii) Diphthongs

Like other Khoekhoe varieties, the Kora dialects make use of various vowel combinations, which may be oral or nasal. (Sequences of two adjacent vowels that behave as a unit are referred to as diphthongs.) While diphthongs in some languages may have either a short onset or a short offglide, the two vowels in Khoekhoe diphthongs typically carry an equal weight, particularly in careful citations, while each may also carry a separate tone.

In the case of the nasalised diphthongs, the nasal quality spreads over both vowels in the sequence. Some older scholars occasionally indicated this by placing a tilde over both vowels, or else over just the second vowel. The convention we have chosen here is to indicate it on the first vowel, for the simple reason that this was the practice of both Meinhof and Maingard, and preserving it will enable us to stay close to the spellings used in the heritage texts. The same convention is used for Namibian Khoekhoe, where the circumflex is placed over the first vowel in a nasalised vowel sequence.

The first three sets of oral and nasal diphthongs illustrated below begin with a rounded back vowel, written u or o, and finish with i in the case of u, and either e or a in the case of o. While some of these sequences may reflect the loss of former medial segments, others have developed by a well-known process of ‘vowel breaking’, which typically affects long vowels and may be associated with labialisation and fronting, as commonly seen, for example, in Sotho-Tswana languages.

As noted earlier, the back mid vowel typically takes the form of the higher allophone [o] when it is followed by e, and [?] when followed by a. (Beach did not find a nasalised form for the sequence ending in e, namely oe.) In the cases where o is raised, it may sometimes be perceived as u, while it can also happen that the final vowels i and e in ui and oe may be heard as lower or higher respectively. These phenomena probably account for spellings in older records such as ‘kuep’ for khoeb ‘man’, ‘khoi’ for khoe ‘person’, and ‘doi’ or ‘due’ for doe ‘fly, flee, depart’. While we have preserved original spellings in the texts where possible, we have made appropriate adjustments in the case of frequently used words such as khoeb ‘man’.

In the example sets below, the letter ‘x’ stands for the fricative sound indicated by the letter ‘g’ in Afrikaans goud ‘gold’ or Tswana goa ‘cry out’. [17] (The click symbols used in some of the examples will be explained in a later section of this chapter.)

ui, ui (Phonetic /ui/)

       hui        'help’

       uib        ‘bee’

       |ui       ‘one’

       kx’uib        ‘life’

       !ui        'mountain pass’

       ?huib        ‘willow’

oe (Phonetic /oe/)

       doe        ‘fly, flee, trek’

       |’oeb        ‘curse, oath’

oa, õa  (Phonetic /?a/)

       toatoa        ‘finish’

       thoathoa        ‘begin’

       hoa        ‘all’ [hoa ?’ukua hora! ‘serve up all the food!’]

       xoa        ‘write’

       kõas        'knife’

       õab        ‘son’

In principle, these diphthongs are pronounced much as they are written. Nevertheless, there are cases in Kora, especially in the context of fluent speech, where the rounded first vowel in sequences such as ui, oe and oa is shortened and may take on the quality of a glide, so that the diphthongs are frequently heard as [wi], [we] and [wa]. These were often written down as such in older documents, giving us spellings such as ‘khwep’ or ‘kwep’ for khoeb ‘man’. Maingard consistently spelled such words with the letter ‘w’, and in these cases, we have adjusted the spellings in the texts for ease of reading and for the sake of overall consistency.

The remaining sets of oral and nasal diphthongs all begin with the sound represented by the letter a, and end with either one of the front vowels i or e, or one of the back vowels u or o. (Beach did not find nasalised forms for the sequences ending in the mid vowels e or o, namely ae and ao.) Much like the sequences discussed above, they have a number of variations in their pronunciation, and as a consequence also in some of the spellings that we encounter in the heritage texts. While they were at one time apparently spoken much as they are spelled, their pronunciation in Kora and modern Namibian Khoekhoe is nowadays a little different.

ai, ãi (Phonetic /??/)

       kai       'big’

       haib        ‘tree (tall)’

       xaib        ‘male gemsbok’

       !xaib        ‘place’

       ?ai        ‘call’

       khãi        ‘rise, wake’

       xãi         ‘swell’

       ?’ãib        ‘he’

When he was working in the 1920s, Beach [18] found that in careful enunciations of the vowel sequence ai, speakers of Nama would typically produce a sound close to the two separate vowels suggested by the spelling. In Kora, however, he found that the first vowel was often raised, so that the sound became closer to /??/, which is very roughly similar to the sound of the diphthong in South African English BAIT or Afrikaans feit ‘fact’. (This vowel sequence is represented in some older Dutch or Afrikaans-based records of Kora by the spelling ‘ei’.) Contemporary speakers of Namibian Khoekhoe dialects sometimes now use the /??/ pronunciation, [19] while our Kora consultants consistently pronounced this sequence in this way. [20] There is often a somewhat rounded quality to the sound.

ae (Phonetic /ae/)

        ?ae         ‘tell lie’

       ?kx’aeb        ‘time, season, occasion’

        !xaeb        ‘darkness, night’

       ?aeb        'smoking'

(We have not come across any instances of the nasalised form of this diphthong.)

In ordinary speech, the second vowel of the sequence can seem to vary between e and a closer form, so that it may be heard as i. This probably accounts for some of the alternate forms seen in older records, where words with this vowel sequence may be spelled with ‘ai’. As Beach heard this sound in the 1920s, it was produced approximately as a sequence of the two vowels indicated in the spelling ae, which is to say a sequence approximately similar to the sound of the diphthong in South African English BITE, but with a more open second vowel. This vowel sequence is rarely indicated in Meinhof’s Kora Glossary, [21] and we found only a few examples of it in our own data. In some cases where we expected to hear it, as for example in |’aeb ‘fire’ and |’aesen ‘be sick’, our consultants gave us [|’?b] and [|’?s?n] instead. [22] Most of the words listed above were nevertheless pronounced with the true ae by Ouma Jacoba. (In principle, the first vowel in the sequence ae should not be susceptible to raising, and it can probably be assumed that words that never reflect the variant [??] fall into this set. In Namibian Khoekhoe, the word for ‘nation, clan’ is represented with this diphthong, as ?aes. In Kora, however, the spellings used by multiple sources suggest that the sound was most often heard as ai, and it is accordingly given in the Dictionary at the end of this work as ?’ais.)

au, ãu (Phonetic /??/)

       daub        ‘male quagga, donkey’

       |’aus        ‘spring, well’

       dãu        ‘burn’

       hãukx’u        ‘seven’ (Lukas dialect of Kora hukx’u, Nama hû)

       ?ãub        ‘homestead, settlement, encampment’

       ?nãu         ‘listen, hear’

Beach [23] stated that this sound was a followed by u, and this is the way this sequence is still pronounced in careful utterances by speakers of Nama. Our two Kora consultants, however, consistently gave a pronunciation closer to /??/, with the first vowel raised, and with an overall impression of roundedness. [24] This vowel sequence, which is more or less similar to the sound of the diphthong in South African English BOAT, is represented in some older records of Kora by the spelling ‘ou’. (In some Khoekhoe varieties, this diphthong has undergone a shift and now occurs as a long vowel [u:], so that Nama, for example, has hû for ‘seven’.)

ao, (ão) (Phonetic /ao/)

       thaob        ‘ash’  

       daob        ‘path’

       saob        'winter’

       saob        ‘tail’

       gaob        ‘male gnu’

       !ao        ‘cut’  

       !’aob        ‘nape (back of neck)’  

       !haos        ‘family (extended), settlement community’

       (dãosen)        ‘burn’ (variant in Ouma Jacoba Maclear’s speech)  

This vowel sequence may be represented in some older records by the spelling ‘au’, either because this was the convention used by the author for the sound ao, or else because the pronunciation of the final vowel may vary and is sometimes sufficiently close that it is actually heard as u. The sound is approximately similar to the sound of the diphthong in South African English BOUT, but with a more open second vowel. Our consultants regularly gave us pronunciations close to the representation of the sound as ao.

3.2 The ordinary (or egressive) consonants of Kora

Kora has a fairly straightforward set of consonants, and features a range of sounds similar to those used in other languages of southern Africa. The chart in Figure 3.2 shows the set of conventional egressive consonants in Kora (where the term ‘egressive’ simply captures the idea that the airstream involved is an ordinary outward flowing one). The sounds in square brackets are either predictable variants (allophones) of certain sounds in certain environments; or, as in the case of the semi-vowels y (phonetic /j/) and w, they are simply sounds that are occasionally introduced for ease of transition (liaison) between other sounds. The sounds in parentheses are marginal, in the sense that they were only ever noted in a few words. Figure 3.2 provides a sketch of the vocal tract in which the various places of articulation referred to in the chart are identified, and the different manners are explained in the accompanying note.

The Kora inventory of consonants is slightly larger than that of other Khoekhoe varieties, mainly because, unlike Nama and other western dialects, Kora seems to have made use of voicing to create an additional series of segments, spelled b, d and g which contrast with both plain (voiceless unaspirated) and aspirated ones. (Readers familiar with Namibian Khoekhoe dialects will be aware that the orthography makes use of the letters ‘p’ and ‘b’, ‘t’ and ‘d’, and ‘k’ and ‘g’. As they are used in the Namibian system, these letters are not intended to signify any semantic contrast based on the presence or absence of voicing, but are simply used to indicate differences in tone melodies.) The eastern varieties of Kora also feature a uvular ejective affricate, represented by the spelling kx’ (phonetic /q?’/), which is not found in Nama or Dama. Both of these additional elements – that is, the feature of voicing and the ejective affricate – are carried over into the set of click consonants, as will be seen later. (These aspects of the consonant system bring Kora into line with its sister languages in the Kalahari branch of the KHOE family.) One further respect in which Kora differs from Nama is that the aspirated alveolar stop written as th (phonetic /th/) tends to become affricated to ts only when followed by the front or high vowels i, e, and u, whereas this process has gone much further in Nama, where all originally aspirated alveolar segments now occur as affricates.

Figure 3.2 Sketch of the vocal tract, showing the places of articulation referred to in the consonant chart.

Each of the consonants will be briefly discussed in turn below, with examples of words that contain them. We have chosen to group these sounds by manner rather than place, and will begin with stops (or plosives).

3.2.1 Stops

(i) The plain (voiceless unaspirated) (bilabial), dental and velar stops (p), t, k

It is not clear that the plain voiceless bilabial stop represented by the letter p ever occurred as either an initial or medial sound in any true Kora word. [25] The only examples we have of an initial p in Kora occur in two words recorded by Beach as pa ‘bite’ and pereb ‘bread’ – which Meinhof, however, transcribed as ba and bereb. We have not been able to confirm Beach’s pereb, since neither of our consultants used this word, which is in any case almost certainly a borrowing of mabele ‘maize’ from either one of the Nguni languages or a Sotho-Tswana source. (It is still found in Naro as mabele.) Ouma Jacoba pronounces the word for ‘bite’ as ba.

The plain voiceless stops t and k are pronounced much like their equivalents in Afrikaans, which is to say, without aspiration. (This frequently makes it rather difficult to tell them apart from their voiced counterparts d and g.) The voiceless alveolar stop represented by the letter t is found at the beginning of a few words and grammatical morphemes, but in general seems not to have occurred very commonly in the middle or at the end of words. The seeming counter-example of |naitab ‘baboon’ probably reflects a root |nai plus an alternative form of the diminutive suffix, which is more commonly –ra in Kora. As for the colour term xati ‘white’, which is unique to Kora, the occurrence of the medial t is not easily explained. (The word may be another borrowing.)

The voiceless velar stop indicated by the letter k occurs at the beginning of, but is not found medially or at the end of words. (Apparent counter-examples, such as haka ‘four’, may reflect an old suffix, or else are loanwords.) When the following vowel is one of the front vowels i or e, the voiceless velar stop k may be expressed as the palato-alveolar affricate [t?], which is the sound of ‘ch’ in English ‘cheep’. In Kora, there are three grammatical morphemes that only occur as [t??], and are seen spelled variously in the old texts as ‘tje’, ‘kye’, ‘kie’ or ‘gye’.

t

       ta        future tense marker

       tama        ‘not’

       tamma        ‘tongue’  

       tani        ‘carry’

       tanikua        ‘clothes’

       tarakhoes        ‘woman’

       tus        ‘rain’

k

       kai        ‘big’  

       karo        ‘hard’

       kõas       ‘knife’  

       koba        ‘speak’  

       kobab        ‘language’

       koro        ‘five’

       kunis        ‘wagon’  

       kurib        ‘year’

       –ku        3rd person masculine plural marker

(ii) The voiced bilabial, alveolar and velar stops b, d, g

There is a certain ambiguity about the voiced series of stops, and some of the tokens recorded in the past may well have been merely voiceless unaspirated segments. It is also difficult to find convincing examples, at least for initial b, where the words in question are not probable loanwords. (The scarcity of examples makes it difficult to find diagnostic ‘minimal sets’, which is to say, sets of words with different meanings that would be phonetically and tonally identical if not for a contrast based on the presence or absence of voicing alone.) As for b in the middle of a word, Kora speakers in the 1920s and 1930s are said to have pronounced this as an actual voiced stop b, unlike speakers of Nama who at this time often already used the fricative form [ß] or even the labiovelar glide [w]. Both our consultants have tended to use the Nama-like fricative form [ß] in this environment. [26] Typical stems in KHOE languages never end in b, and where such a final segment is seen, it is invariably the suffix –b of the 3rd person masculine singular. In this position, the sound is devoiced so as to sound much like p, and the suffix is frequently seen in older records with a spelling that reflects this. Ouma Jacoba often produced a fricative [ß] rather than a devoiced [b?] for the suffix – and sometimes even left the sound unexploded, so that the suffix is occasionally almost inaudible in her speech. Maingard [27] noted this tendency as a feature of the Kora dialect spoken by members of the Bloemhof community, who were mainly of the Links clan, while a similar trend seems to have been characteristic of old Cape Khoekhoe.

The voiced stop indicated by the letter d occurs only at the beginning of words, and in only a few of them at that. Any medial d is for the most part replaced by r. In the case of gaida ‘old’, where d occurs contrary to expectation in the middle of a word, –da may be an old derivational morpheme. The sound was often palatalised and affricated by Kora speakers before a front vowel to occur as [d?], which is the sound indicated by the letter ‘j’ in English ‘jeep’. An example of this is seen in the usual Kora pronunciation [d?isi] for disi  ‘ten’. We heard another instance in Ouma Jacoba’s pronunciation of the pronoun ?’ãidi   ‘they (3rd person feminine plural)’, which she produced as [?’?~ind?e], although she produced the same suffix –di without affrication in her utterance of the phrase ?na !nona khoedi   ‘those three women’. The verb di ‘make, do’seems to have been exempt from the expected changes.

The voiced velar stop written as g occurs only at the beginning of a word or grammatical morpheme. In Ouma Jacoba’s pronunciation of words like gaida ‘old’, gama ‘crooked’ and gus ‘sheep ewe’, the sound has a breathy-voiced or even slightly affricated quality. It is remarkable that Burchell seems to have observed much the same thing in 1813. His word list (shown previously in Figure 2.5) includes the word gaida in the context of a phrase for the ‘old’ (or waning) moon. Burchell represented this as ghyda’kaam (gaida ?xãb), commenting that ‘the h in this place gives a strong and guttural aspiration to the G’.

b initial

       baxab        ‘tobacco’

       bib        ‘milk’

       biris        ‘goat ewe’

b medial

       koba        ‘speak’  

       khabu        ‘blaze, flame up’ (note Tswana kgabo ‘a flame’)

       khoba        ‘open’

       subu        ‘easy, light’

d

       danis        ‘honey’  

       daob        ‘path’

       aub        ‘quagga, donkey (male)’

       dãu        burn’  

       di        ‘make, do’ [sõse di ‘do quickly’]

g

       gaida       ‘old’  

       gamasa        ‘crooked’

       gaob        ‘gnu (male)’

       gare        ‘praise, pray’

       goesi        ‘nine’ (Meinhof gives ‘guesi’, Engelbrecht ‘khoese’)

       gomas        ‘cow’  

       gorab        ‘flower’

       gus       ‘sheep ewe’

       –gu        reciprocal verb extension

(iii) The aspirated alveolar and velar stops th, kh

There does not seem to be an aspirated bilabial stop ph (phonetic /ph/) in Kora, but the language has both th (phonetic /th/) and kh (phonetic /kh/). These phonemes, which only occur at the beginning of words, may become affricated to [ts ~ t?] and [kx] respectively before high or front vowels, or may even be pronounced consistently in this way by some individual speakers. (In Nama, it is generally the norm that the variant [ts] occurs before all vowels in the corresponding words, as noted earlier.) Examples of words beginning with the aspirated alveolar and velar stops are shown below.

th (Phonetic /th/, may alternate with [ts ~ t?] before a front or high vowel)

       thã        ‘feel, taste’ (Nama tsâ)

       thamsa        ‘soft’ (Nama tsam)  

       tharab        ‘dust’ (Nama tsarab)

       thoathoa        ‘begin’ (Nama tsoatsoa)  

       thub         ‘pain’ (Nama tsûb)  

kh (Phonetic /kh/, may alternate with [kx] before a front or high vowel)

       khas        ‘bow, hunting or musical’

       khabu        ‘blaze, flame up’

       khãi        ‘go up, rise’

       khao        ‘dig’

       khob        ‘skin, hide’  

       khõa        ‘break’ [khõa xukua ‘break the things’]  

       khoba        ‘open’

       khoes        ‘woman’  

        khom        ‘speak’

3.2.2 Nasals

The bilabial and alveolar nasals m, n

The Khoekhoe languages have only the bilabial and alveolar nasals m and n. Both sounds may occur at the beginning, in the middle, and at the end of a word (where as final segments they may carry tone). Where a noun ends with the bilabial nasal m, the first segment of the 3rd person masculine singular suffix –b(i) is assimilated to it. This is the source of the allomorph –mi seen in words like xammi ‘lion’ and |hommi ‘cloud, sky, heaven’.

In a few cases, the initial alveolar nasal n participates in a set of cross-dialectal alternations involving n, l and t, as in the case of words for ‘tongue’, where Nama has nammi, Dama varieties have lammi or tammi, and Kora has tamma. The word for ‘lightning’ occurs similarly as nabab in Nama, labab or tabab in varieties of Dama, and tabab in Kora. Interestingly, the Nama word tabete ‘greet’, which is thought to reflect an early borrowing from Malay, occurs as nawede in Dama, and nabe in Kora.  

On one occasion, Ouma Jacoba Maclear produced l instead of n after a preceding n, in the phrase ?hana la |’on ke  ‘the child is crawling’, where the progressive (imperfective) aspect marker would ordinarily be na, and where the substitution of la seems to have had a dissimilating function (that is, to prevent a succession of similar sounds). Ouma Jacoba sometimes also alternated between the use of a medial n and r, as in the case where she gave us the word kx’arina for ‘chicken’ (compare kx’anis ‘bird’) but in a follow-up sentence immediately afterwards pronounced it as kx’anina, as if by way of self-correction. (These examples also illustrate cases where the common plural marker –n is used with a singular implication.)

m initial

       mi         ‘say’

       mã         ‘give’

       mã        ‘stand’

       mu         ‘see’

m medial

       gomas         ‘cow’

       !nomab         ‘root’

m final

       kx’am         ‘roast’  

       xammi        ‘lion (male)’ [xamma]  

       |am         ‘two’ [|amse]

       |hommi         ‘cloud, sky, heaven’ [|homma ‘heaven’]

       !om         ‘blow (as wind)’

       !xom         ‘break off’

       !kx’am         ‘green’ [!kx’amsa]  

n initial

       na         progressive (imperfective) aspect

       nabe        ‘greet’

       ni        obligatory mood

n medial

       danis        ‘honey’

       kx’anis        bird’

       !hanab        ‘garden’

       !nona        ‘three’

n final

       |kx’onna        ‘name’ (Nama |ons)  

       ?’an         ‘know’

       ?’anna         ‘knowledge’ (Nama ?ans)

3.2.3 Fricatives

(i) The voiceless alveolar, velar and glottal fricatives s, x, h

It is notable that there is no voiceless bilabial fricative [?] in Kora, although as we have already noted, the voiced counterpart of this sound [ß] is often heard in place of b, either in the middle of a word, or as a variant of the masculine singular suffix –b at the end of a word. The language nevertheless has the alveolar fricative s, the velar fricative x, and the glottal fricative h. [28] The velar fricative x is the sound indicated by the letter ‘g’ in some speakers’ pronunciation of Afrikaans goed ‘things’ or goud ‘gold’ and the Tswana place-name Gauteng ‘at the place of gold’. [29]

Beach commented [30] that the speakers he worked with frequently produced s as the palatal fricative [?], which is the sound of ‘sh’ in English ‘sheep’. We have not found this variant to be very frequent in the speech of our consultants, although in one or two instances we heard [x] for s in the speech of Oupa Dawid Cooper. When the fricative x occurs before a front vowel, it may be palatalised and affricated to give [t?] (as ‘ch’ English ‘cheep’). An example of this is seen in the name Tjelkausob, from Afrikaans Geelkous ‘Yellow Sock’. (A similar process occurs in the Sotho-Tswana languages, where, for example, the word for ‘money’ is itjelete, from Afrikaans geld.) 

None of the voiceless fricatives occurs in the middle or at the end of a word. The seeming counter-example of !aixab ‘doctor, sorcerer’ really reflects an incorporated derivational morpheme –xa: Nama has !gaixa as an adjective, which means roughly ‘possessed-of !gaib (‘medicine’)’. Similarly, the cases seen below where s occurs medially all involve additional morphemes, such as the reflexive verb extension –sen, or the adverbial formative –se. Where s is seen at the end of a word, this is invariably the 3rd person feminine singular suffix –s.

The alveolar fricative s occurs at the beginning of a number of grammatical morphemes, and in cross-KHOE comparisons is frequently found as h in equivalent items. The word ?naxusi that was given to us for ‘pour out, away’ by Ouma Jacoba Maclear, and which features the causative extension –si, might seem to be a mis-remembered form of the word that Meinhof recorded as ?naxudi(~ ?naxuri), except that Wuras also found the form with –si. Unless the two forms reflect different morphemes, the possibility arises that some instances of s have arisen from an original d.

s

       sao        ‘follow’

       saob        ‘tail’

       saob        ‘winter’

       sãsi        ‘cook’

       sisen        ‘work’

       sõse        ‘quickly’  

       sores        ‘sun’

       sub        ‘pot’

x (as ‘g’ in Afrikaans goed ‘things’ or Scots ‘ch’ in ‘loch’)

       xaib        ‘gemsbok (male)’

       xãi        ‘swell’

       xammi        ‘lion (male)’ [xamma]

       xati        ‘white’  

       xoa        ‘write’

       xoasaob        ‘leopard (male)’

       xon        ‘grind’  

       xub        ‘thing’

       xurub        ‘powder’

h

       hab        ‘stallion’  

       hagub        ‘boar’

       haka        ‘four’

       hãukx’u        ‘seven’

       huni        ‘mix, stir’  

(ii) The voiced uvular and glottal fricatives [?], [?]

The existence of voiced fricatives in Kora has not previously been noted, and it should be clarified that they are marginal sounds which, with the exception of the voiced glottal fricative, are likely to have been merely occasional variants of the voiceless fricatives, most often occurring between vowels. (In other words, they are not part of the phoneme inventory of the language.)

On one or two occasions, Oupa Dawid Cooper gave us pronunciations of words that seemed to feature the uvular trill [?], which is the sound of ‘r’ in the French word rose (for ‘rose’) or the burr in some northern varieties of the English spoken in England. Examples include a word guru that he twice used for ‘big’, in the expression guruxub for groot goed or ‘big things’, and in the sentence gurukhoeb i-b ke ?naba ‘that is a big man there’. (The word guru rather oddly resembles the Shona word –kuru, which occurs as –guru with nouns of Class 5, and means ‘big’, being cognate with Nguni –khulu.) Ouma Jacoba Maclear, on the other hand, gave us guxu  for ‘long, tall’, where the occurrence of the medial fricative [x] is highly unusual for Khoekhoe. The mystery surrounding this word is compounded by the fact that Andries Bitterbos used the word guxukua (AB6) in the sense of kleinvee ‘small livestock’, where xukua is the plural of xub ‘thing’.

As far as the breathy-voiced glottal fricative (phonetic [?]) is concerned, Beach never heard a speaker who used it, although he speculated [31] on theoretical grounds that it might have existed in the past, since it is a type of sound that can be associated with certain changes in tone melodies (specifically tone lowering), which he believed had taken place. We have had the great fortune to find a voiced glottal fricative still present and clearly distinct in the speech of Ouma Jacoba Maclear – in haib  ‘tree’ and huri   ‘jump’– which confirms Beach’s surmise. In the case of huri ‘jump’, there is a faintly audible bilabial stop at the beginning of Ouma Jacoba’s utterance, where the sound appears to be breathy-voiced [bh].

In the case of hoa ‘all’, our two consultants did not produce the initial fricative with any noticeable degree of voicing, although we suspect that this word may have had such a sound in the past, given the remark by Maingard [32] that in addition to the four basic tonemes in the dialect of the Bloemhof Kora speakers, ‘a low or very low tone, e.g. hoan (all) is […] very occasionally heard’. It is notable that in Lucy Lloyd’s records of Piet Links’s variety, the word hoa ‘all’ is written with a subscript ‘hook’ (see Figure 3.3), which was her convention for indicating a ‘rough, deep pronunciation’ of the vowels. [33] She sometimes used the same symbol to write hã ‘stay, remain’, and, together with a doubled ‘h’, for ho ‘take, find’. We have not specially distinguished the breathy-voiced [?] as a phoneme in the Dictionary, for the simple reason that it is so seldom (and never consistently) reflected in the sources.

3.2.4 Affricates

(i) The voiceless alveolar affricate ts

It has been noted at various places above that a number of affricated forms, including the one represented by the spelling ts, can occur as variants of certain sounds in certain predictable contexts (typically before high or front vowels). The alveolar affricate ts does not seem to have occurred as widely in Kora as it does in Nama, although Engelbrecht [34] noted that he heard it in the Kora variety spoken by the Lukas and Springbok people. Even though it is strictly speaking not a phoneme in Kora, we have chosen to list this particular affricate separately in the Dictionary, partly because of its prevalence in Nama, and partly because the process of affrication in Kora was evidently an ongoing and uneven one, so that it seems better for practical purposes, where the affricate occurs, to indicate it explicitly in the spelling. The words below illustrate the limited occurrence of the alveolar affricate in Kora varieties as spoken by members of the Links, Kats and Kraalshoek people.

ts

       tsi        coordinating conjunction ‘and, or, if’

       tseb        ‘day’ (note Tswana motshe)

       tsuguru        ‘circle’

       motsab        ‘water pool, lake’ (given only by Engelbrecht, in 1928)

The word for ‘water pool, lake’ is of course a counter-example to the rule of thumb mentioned above for the occurrence of the variant form, since the vowel in question is not front or high. However, this word also occurs in languages of the Sotho-Tswana family, typically as letsha, and that it must have been borrowed into the KHOE languages is rather clear, given that the Kora form even preserves the original Sotho-Tswana prefix.

Given the regular occurrence of aspirated stops in Kora, we might have expected to find also the aspirated form of the alveolar affricate, namely /tsh/. This sound does not seem to have occurred in Kora, however, unless it has merged with ts. (It is notable that tsha ‘water’, as opposed to (mo)tsab ‘water pool, lake’, is distributed widely across Kalahari members of the KHOE family, and is even reconstructed for Proto-Kalahari KHOE, as *tsha.) [35]

(ii) The ejective alveolar and post-velar affricates ts’ and kx’

It may be worth mentioning, simply so as to contextualise them, that sounds such as the aspirated affricate /tsh/ mentioned above, and the ejective affricate /ts’/ discussed under the present heading, are by no means unusual in a southern African context, and occur regularly, for example, in Sotho-Tswana languages, where they typically form part of a more extensive set of alveolar and palato-alveolar affricates. [36] The ejective affricate ts’ similarly occurs in the Nguni languages, where it is associated, with only a few rare exceptions, with prenasalisation. [37]

The ejective alveolar affricate also has a fairly common occurrence in languages of the KHOE family, particularly those of the Kalahari region, but is seldom recorded in languages of the Khoekhoe branch. This makes it intriguing that Meinhof found a handful of marginal words in Kora that seem to have contained it, as follows:

       Ts’oa?’ai-        ‘a clan of people half Korana, half Bushman’

       ts’units’eb        ‘small aquatic larva’ (perhaps mosquito or midge)

       ts’ururu        ‘mosquito’

       ts’uts’u        ‘wink, blink’

Beach [38] took note of Meinhof’s observation, and reported that his own consultants were certainly familiar with the last two words, adding that Wuras had recorded tsuniseb for ‘mosquito’ (though without indicating an ejective). Engelbrecht [39] confirmed the presence in the variety spoken by the Lukas people of the last three words in the list – which Meinhof [40] had gone on to suggest were perhaps loans from a ‘Bushman’ language.

If we follow up on Meinhof’s suggestion, we find that the word list for the !Ui language ?Ungkwe, which Meinhof himself documented, [41] includes only two words with the sound, namely ts’axo ‘eye’ and ts’e ‘be home, returned’. The composite dictionary compiled by Dorothea Bleek [42] gives ts’un for the |Xam language – also of the !Ui family – where it meant ‘wink at’, but only tsutau ~ tsuto for ‘mosquito’. On the other hand, the Kalahari KHOE languages Naro and Khwe have ts’om or ts’omts’om ‘close the eyes’. If the few Kora words mentioned really reflected this sound, this may be another instance where Kora seems to have preserved certain characteristics of the ancestral ‘Proto-KHOE’ language no longer found in South African Nama or Namibian varieties of Khoekhoe. [43]

Kora has another ejective affricate, typically written as kx’, which was one of the salient characteristics of eastern varieties of the language. It does not occur in Nama or Namibian Khoekhoe, but is found, like ts’ , in the Kalahari branch of the KHOE family. Cognate words in western Kora dialects, as well as Giri, Nama, and other dialects of Namibian Khoekhoe have typically lost the initial segment and simply begin with a vowel, as in Nama aib ‘face’ where Kora has kx’aib, and Nama anis ‘bird’ where Kora has kx’anis. [44] (The Nama vowels in these cases typically have a glottalised onset.) Like the other affricates, kx’ is found only at the beginning of a word. [45]

Strictly speaking, kx’ is not so much an affricate that is ejected, as an ejective that is accompanied by frication: the sound appears to be an ejected uvular stop followed by a brief uvular trilling, where this vibration is probably set off by the ejective burst, though it quickly loses its periodic character to end as mere frication. (In the eastern Kalahari KHOE variety, Tjwao, cognate words feature a straightforward ejective stop /k’/.) While this sound has been termed an ejective velar affricate in the past, it has emerged during the course of our studies [46] that the closure is typically further back, and in reality, has a uvular quality. [47] However, since there is no semantic contrast in Kora that turns on the phonetic distinction between velar and uvular places of articulation, there is no need to make any pedantic change to the established spelling of the sound. (In Tswana, the spelling ‘kg’ is used for a fricative phoneme that is basically velar, but predictably uvular when produced before back or low vowels.) [48] It should perhaps be added – in anticipation of speculations about substrate influence – that this sound is the regular Tswana reflex of Proto-Bantu *k when preceded by a nasal. (Intriguingly, some of the Kora words that feature kx’ have counterparts in Tswana that feature the sound indicated by ‘kg’.)

The ejective affricate seems to have been heard regularly by Lloyd in the speech of her consultants, who were from the Links clan (or Left-hand Korana). The pages from her 1879 notebooks reproduced in Figure 3.4 show her use of the Greek letter gamma (?) for what must have been the equivalent of this sound in the word kx’ui ‘live’. Lloyd would have adopted this usage from Wilhelm Bleek, [49] who left us a detailed note on this ‘guttural sound’ in the notebook he used for his first sustained and in-depth study of a !Ui language, which he carried out with the assistance of the multilingual (Kora-|Xam-Dutch) speaker, Adam Kleinhardt. It is interesting that in this note, Bleek described the sound as lateral, with ‘the breath passing out on the left side of the mouth’ – given that the same sound occurs in some of the Nguni languages, where it may similarly have a lateral character. [50] (In Zulu it is typically spelled ‘kl’, and is described as an ‘ejective velar affricate consonant, or ejective velar lateral affricate’.)

kx’

       kx’a        ‘drink’ (Nama a)

       kx’a        ‘cry out, shout’ (Oupa Dawid Cooper has kha)

       kx’ai        ‘on’ (Nama ai)

       kx’ãi        ‘laugh’ [kx’ãi ?na khoena ‘those people are laughing’]  

       (Oupa Dawid Cooper has xai)

       kx’aib        ‘face, surface’ [ti kx’aib ‘my face’] (Nama aib)

       kx’ãib        ‘liver’ (Nama âib)

       kx’aisi        ‘first’ (Nama aibe)

       kx’am        ‘roast’ (Nama am)  

       kx’am        ‘right (hand)’ (Nama am)

       kx’amma        ‘mouth’

       kx’anis        ‘bird’ (Nama anis)

       kx’anu        ‘correct, proper’ (Nama anu ‘be suitable, fitting’)

       kx’ao        ‘man, male’ (note Tswana lekau ‘young man’)

       kx’ausa        ‘bitter’ (Nama au)

       kx’arob        ‘young man, boy’ [kx’aro|’ob]  

       kx’oa        ‘seek’

       kx’ob        ‘meat’ (Oupa Dawid Cooper has xob)  

       kx’om        ‘build’ (Nama om)

       kx’ommi        ‘house’(Nama oms)

3.2.5 Approximants

The approximants [j], [w], (l)

The sounds [j] and [w] (as in English ‘yet’ and ‘wet’) are sometimes referred to as semi-vowels, or glides. These two sounds do not occur as independent phonemes in Kora, but are occasionally introduced by speakers between two vowels to facilitate transition (liaison) between them.

Typically, the insertion of the palatal approximant [j]) is associated with a following front vowel such as i or e. This is why it is sometimes seen before the passive verb extension, which is usually –he but may be shortened to –e. (A noteworthy implication of this is that the vowel in this case is not glottalised.) Meinhof indicated an intrusive [j] by means of the letter ‘i’, as in |iieb or |aiieb ‘jackal’.

The labiovelar approximant [w] may occasionally be inserted in speech ahead of the back or low vowels u, o and a, as suggested by the spelling ‘!Oara’ for !Ora. It is sometimes also seen written between a verb and the copula (or ‘stative marker’) a, which suggests that the vowel of the copula is similar to that of the passive –e, in the sense that it is not glottalised. The labiovelar glide [w] can also occur as a variant of the bilabial fricative [ß], which is itself a variant of b. In other cases, it can arise from a natural progression where a vowel sequence such as oa is produced as [wa]. (As noted earlier, spellings of the diphthongs in the old records frequently show ‘wi’ for ui, ‘we’ for oe and ‘wa’ for oa.) There is probably no need to indicate either of the two glides explicitly in modern spellings.

There are very few instances of the lateral approximant /l/ in Kora. One of these marginal occurrences is in the personal name Mulukab, which is perhaps equivalent to the name Mookha noted by Campbell in 1813, who said it meant ‘sharp sight’. Another rare instance occurs in the word lu ‘swear’ recorded by Meinhof, which has nu as its counterpart in Nama (but lu in Dama). [51] (In KHOE languages of the Kalahari, Naro has luu ‘swear (a solemn oath)’, while Khwe uses a different word.) Wuras recorded the word as ‘nau’, while Maingard noted [52] an occurrence of ‘laula’, which he said meant ‘I am sorry, excuse me’. [53] (Samuel Dornan [54] recorded taolo ‘a command’ in the Kalahari KHOE language, Hie Tshware, but noted specifically that it was a Tswana loanword.) A few of the old records occasionally give local names for rivers of the south-eastern Cape where the spellings reflect /l/, but these may have been mishearings for /r/, or else the names were not Khoekhoe.

3.2.6 Trill

The trilled or flapped r

In careful pronunciations, the sound spelled r in Kora is typically produced as an alveolar trill, much as it is in Nama or Afrikaans. Beach [55] reported, however, that in the speech of his consultants, it was frequently ‘a mere flap, with very weak tension and plosion, and with more or less free lateral exit for the air-current’. He added that this sound could strike the ear of a non-Kora speaker ‘as a sort of mixture of r, d and l’, and he also noted that Meinhof spelled some of the words in his short word list for Giri with ‘l’ in place of an ‘r’.  Maingard [56] identified the sound as a ‘flapped retroflex consonant which is heard as r or l or d’.

While the sound mostly occurs in the middle of words, where it seems to be an allophone of d, it may occasionally occur as the initial sound of a grammatical morpheme, such as the –re in the 1st person masculine singular pronoun tire, where the feminine form is tita. (Oupa Dawid Cooper, however, used tite when referring to himself in the 1st person. Nama does not have a separate pronoun for the feminine ‘I’, but uses tita for both.)

In a few cases the sound was an almost inaudible tap in the speech of our consultants, and on one occasion we heard Oupa Dawid Cooper produce arib ‘dog’ as [a?b]. This kind of elision probably explains variant spellings of the word for ‘jackal’ as ‘|kerap’ and ‘|keyap’ in Lloyd’s records of her work with members of the Links family. On one or two other occasions, as mentioned earlier, Oupa Dawid gave us pronunciations of words that seem to feature the uvular trill [?] for r.

r (medial)

       arib        ‘dog’

       karab        ‘shade, shadow’

       gare        ‘praise’

       huri        ‘jump’

       koro        ‘five’  

       sores        ‘sun’  

       xurub        ‘powder’

3.3 The clicks, or ingressive consonants of Kora

The remaining class of consonants consists of the clicks, or ‘lingual ingressives’. [57] This last term and some of the more technical aspects of click production are explained in Figure 3.5. For present purposes, it is enough to note that the manner of a click’s production involves what might be described in broad terms as a special type of co-articulation, and that the Khoekhoe languages make use of clicks distinguished by four different places of the forward closure. These places are dental, alveolar lateral, (post)alveolar, [58] and palato-alveolar. (The bilabial click (?) occurs only in the !Ui-Taa languages and ?’Amkoe, and there will be no need for us to make any further reference to it here.)

Fig. 3.5. Sketch of the vocal tract, with a brief explanation of the manner involved in the production of clicks, and the elaborations that may accompany them.

Much like the conventional consonants of many other languages found in southern Africa, clicks may be elaborated by features such as voicing, aspiration and ejection, while they may also be prenasalised. (All of these elaborations are phonemic, which is to say, associated with a difference in the meaning of words that are phonetically identical in every other respect.) Although the rear closure of a click is typically released inaudibly, a distinct and audible release is nevertheless possible, in which case a few further elaborations of the sound may be delivered.

While most of the Khoekhoe languages do not make use of contrastive voicing, some of the early authors recorded the presence of this feature in Kora. One further feature reported by earlier scholars was the accompaniment of a click in Kora by an ejective affricate – an elaboration that does not occur in western varieties such as Nama and other Namibian Khoekhoe dialects, though it is found in related languages of the Kalahari KHOE branch. The presence in Kora of the two extra click accompaniments parallels the presence of the extra segments in the inventory of the conventional consonants – namely the voiced series of stops, and the ejective affricate.

3.3.1 The four basic (or ‘radical’) clicks of Kora, identified by place

The dental click | (the ‘c’ click of Nguni languages) is made by squeezing the tip of the tongue firmly against the alveolar ridge, just behind the upper front teeth, and then slowly drawing it away. The release may be prolonged, and at times has a ‘fricated’ quality (although the click itself is, of course, necessarily a plosive). Ouma Jacoba Maclear sometimes prepared for this sound by making a bilabial hold and then placing the tip of her tongue so far forward that it was almost interdental. This sound is impressionistically quieter than the other clicks, and when it is produced in combination with nasalisation it may at times be almost inaudible.

The alveolar lateral click ? (the ‘x’ click of some of the Nguni languages) is made by pressing the tip of the tongue firmly against the alveolar ridge, and then releasing the hold along the side of the tongue. Some speakers may draw out the release, but our consultants tended to make it rather cleanly and sharply.

The (post)alveolar click ! (the ‘q’ click of some of the Nguni languages and Sotho) is made by pressing the tip of the tongue firmly up against the alveolar ridge, as far back as possible. Beach [59] found that the position is quite variable, even for individual speakers, but that the general target is ‘much the same position as for the lateral type’. The tongue is brought away sharply, and the auditory impression is of a strongly popping sound.

The palato-alveolar click ? is is the only one of the four Khoekhoe clicks that does not occur in Xhosa and Zulu. [60] It is made by squeezing the blade of the tongue firmly against the alveolar ridge and then pulling it sharply away. While some speakers may produce it with the tip of the tongue kept behind the upper front teeth, others anchor the tongue tip behind the lower front teeth. It is easy to mistake this click for the dental one (particularly an aspirated dental click), and the old records frequently reflect the two clicks as alternates. Some speakers substituted the dental click for it, while one of the speakers who assisted Jan Engelbrecht [61] habitually used the (post)alveolar click in its place.

3.3.2 The accompaniments of the Kora clicks

Because of the diversity of the elaborations that may characterise the different click phonemes, Traill [62] introduced the neutral cover-all term ‘accompaniments’ to refer to them, and there is no harm in continuing to use his convenient expression here. [63]

We have already noted that Kora (unlike Namibian Khoekhoe) never received an official orthography – which is to say, a standardised and formally instituted set of practical spelling rules. The different authors who described the language necessarily used various conventions of their own to represent the click phonemes, sometimes changing their own systems over time. The table in Figure 3.6 sets out some of the main conventions used for the clicks by individual scholars in the past (excluding the set of symbols devised by Wuras, and the 1928 notation of Engelbrecht). The order in which the different accompaniments are listed is somewhat arbitrary, but is based (with only a minor modification) on the ordering principle adopted by Wilfrid Haacke and Eliphas Eiseb for their Khoekhoegowab Dictionary. It is also the order in which the different click phonemes will be listed in the Kora Dictionary.

The column on the far left-hand side of the table uses the (post)alveolar click as an example and offers a tentative phonetic representation of the click described – though it should be noted that even such representations are not currently stable or universally agreed on by linguists! The column second from the right in the chart shows the symbols used by Meinhof (whose name is abbreviated in the table as ‘Mhf’). These are the ones that will be adopted here as the ‘standard’ representations for the Kora clicks and their accompaniments, both for purposes of discussion, and for the key entries in the

Fig. 3.6. Some of the different conventions used to represent click phonemes in Namibian Khoekhoe and Kora.

Perhaps inevitably, there are a few minor aspects of Meinhof’s system that are not entirely satisfactory. For example, it would perhaps have been more appropriate to write the letter ‘n’ before the click symbol in the case of the nasalised clicks, since this would have given a more accurate reflection of the articulatory sequence. However, the placing of the ‘n’ after the symbol was not only the practice of almost all the original Kora scholars, but is also the present standard for Namibian Khoekhoe. (There are some Khoisan languages where it is necessary to reflect a distinction between prenasalisation and a nasalised click, but this does not apply to the Khoekhoe varieties.)

Secondly, the click written with a following letter ‘x’ is in principle a simple aspirated click – except that the aspiration in many cases develops into a degree of frication. It might have been better to adopt the Namibian convention [64] of indicating this accompaniment by the letters ‘kh’, except that this was not generally the practice of any of the older authors for Kora, while the retention of Meinhof’s original convention is in keeping with our overall strategy of conservatism. The representation of this particular accompaniment is somewhat erratic in the original heritage texts (particularly those collected by Maingard), probably because it seems to have been produced by some speakers as aspiration, and by others as frication, with fluctuations even on the part of the same speaker within the same sentence.

The different series of clicks will be discussed next, in the order in which they are arranged in the Dictionary.

(i) The plain (voiceless unaspirated) clicks: |, ?, !, ?

The plain clicks of Kora are equivalent to the clicks written in the Namibian orthography with a following letter ‘g’, as ‘|g’, ‘?g’, ‘!g’ and ‘?g’. (For those readers who are familiar with Namibian Khoekhoe, it is important to note that in the Namibian system, the letter ‘g’ does not indicate voicing, while clicks written without any following letter in Namibian Khoekhoe are ‘glottalised’.) In the conventions used by some older authors, particularly Lloyd, the plain or voiceless clicks in Kora are those indicated by a following letter ‘k’.

The plain clicks are voiceless and unaspirated, and may be described in Traill’s terminology as clicks that have a ‘zero accompaniment’. They are made with a silent release of the rear closure and are followed immediately by the vowel. The example words below include some that were listed as featuring plain clicks by Meinhof, plus a few that are unequivocally without any audible back release in the speech of our consultants. [65] It is notable that the set of plain clicks is fairly small, and it may also be significant that at least some of these words fall into the category of frequently used words.

       |a        ‘small’  

       |am        ‘two’

       |ui        ‘one’

       ?oe        ‘lie down’

       !u         ‘go’  

       ?aob        ‘heart’

       ?om         ‘believe’

(ii) The ‘voiced’ clicks: |g, ?g, !g, ?g

The occurrence of voiced clicks in Kora was either occasionally noted or at least hypothesised to have existed at some time in the past. As it turns out, the distinction between ‘voiceless’ and ‘voiced’ is just as elusive in the case of the Kora clicks as it is in the case of the ordinary egressive consonants.

The only scholar to indicate the existence of a full series of voiced clicks was Engelbrecht, [66] in the Word List he supplied at the end of his first work on Kora. For the most part, however, the words containing Engelbrecht’s ‘clicks plus g’ pattern with words represented by other authors with plain clicks. Nevertheless, Engelbrecht’s information may one day prove valuable to linguists wishing to undertake comparative studies, and we have accordingly listed his words separately in the Dictionary.

There is an ambiguity embedded in the representation of these clicks: while the writing of a letter ‘g’ after the click symbol might be assumed to indicate voicing, it was intended by many older writers simply and quite literally to convey the auditory impression of a click with a following voiced stop. Maingard [67] referred to an ‘indistinctness’ concerning ‘the isolated velars k and g’ as the ‘velar element of the click’, and suggested that this could be part of the reason for discrepancies so often found in the spellings given by different scholars. He added:

The Bloemhof Korana were, however, more careful when this element meant a semantic difference, e.g. !garib, ‘the Orange River’ and !karib, ‘honey beer’ and !harib ‘whip’;?kaeb ‘time’ and ?kxaeb ‘law’.

We have preserved all instances in the texts obtained by Maingard where he wrote a click with a following letter ‘g’.

Beach, who wrote explicitly about voicing (as opposed to the presence of an audible segment after the click), did not find any instances of clicks in Kora that he judged to have this feature, while Meinhof before him noted only four isolated instances where he said he detected voicing before the click. (It is significant that Meinhof said he heard pre-voicing, because this would have constituted true voicing in the strict technical sense of the term – which is to say, involving ‘voice lead’, where the vibration of the vocal folds commences during the hold and ahead of the release.)

A few words in the speech of our two consultants gave the initial auditory impression that they contained voiced clicks – or at least, clicks resembling the familiar voiced clicks of the kind that occur in the Nguni languages. In a few cases, these words were among those mentioned by older authors as having been voiced, which gave us an ideal opportunity to re-visit the topic.

The results of this investigation are presented elsewhere, [68] but in summary, the clicks perceived to be ‘voiced’ in Kora are found to be clicks characterised primarily by an audible release of the rear closure (as a uvular stop). This is the reason for the neutral phonetic representation of the ‘voiced’ click in the table given earlier (Figure 3.6) simply as [!q]. Any voicing of this release is indeterminate, and if the sound is not merely a plain uvular stop [q] released immediately on to a following vowel, it seems to feature only the type of ‘short-lag’ onset of voicing that similarly characterises the voiced clicks as they are produced by many speakers of Zulu and Xhosa. [69] (In this case, the accompaniment could perhaps be legitimately represented by the symbol for the voiced uvular stop [?].) In a few cases, an audible and impressionistically voiced release was additionally characterised, at least in the speech of Ouma Jacoba Maclear, by a degree of aspiration or frication [?h], much like the voiced conventional velar (or post-velar) stop g (or [?]) in her speech. We have not found any cases of clicks featuring the kind of voice lead suggested by Meinhof’s brief notes, [70] unless these were clicks of the kind with prenasalisation followed by audible release of the rear closure as a voiced uvular stop [n!?]. [71] This last click, for which we have just a single token in our data, and which has not previously been noted in Kora, is discussed elsewhere below.

The example words below are drawn mainly from Engelbrecht’s list of words where a click is written with a following letter ‘g’. It is certainly intriguing that in cases where these words also occur in Nguni languages (or N|uu for that matter), the clicks are typically represented as voiced.

       |gib        ‘aardwolf’ (Proteles cristatus) (note Xhosa ingci, Swati singci)

       |gãb         ‘grass’ (note Xhosa ingca)

       ?gammi         ‘water’

       !gaixab         ‘sorcerer, doctor’ (note Xhosa igqira ‘sorcerer’, ugqira ‘doctor’)

       !gam         ‘kill’  

       !garib        ‘river’

       ?gaeb         'marrow (of bone)’  

(iii) The clicks with ‘delayed ejection’ (or ‘glottalisation’): ’, ?’, !’, ?’

It is important to note that this type of accompaniment is associated, like the feature of delayed aspiration discussed below, with a latent prenasalisation. As Beach described this accompaniment, it involves a closure of the glottis, silent release of the posterior click closure, and a subsequent glottal stop release. Although the symbol used by Meinhof for this type of accompaniment is the apostrophe, which is the standard international phonetic symbol for an ejective release, the actual production of these clicks involves a combination of mechanical events that do not strictly constitute a straightforward ejective release, though the outcome might be described as ‘delayed ejection’. For this type of release to be truly ejective, it would require a preliminary raising of the glottis, with a resulting compression of the air trapped between it and the closure above it – that is, the post-velar posterior closure of the click – with a consequently forceful burst of the pent-up air on release of that upper closure. What happens in reality, though, is that the prenasalisation enables a ‘venting’ of air flow, and so prevents an ejective burst as the posterior closure is released. [72] Instead, a ‘delayed’ burst comes from the release of the glottal closure, which is then simply heard as a glottal stop – which is why these clicks are often described as ‘glottalised’. (In the context of rapid speech, a speaker may sometimes produce a glottalised click simply as plain.) 

The prenasalisation associated with the glottalised (and also the delayed aspirated) clicks is especially apparent when a click word is preceded by the vowel of another word or grammatical morpheme, as in the following phrases produced by Ouma Jacoba Maclear:

       he xati (n)|’uib        ‘this white stone’ (JM)

|nika ti ke (n)?’u, t? ‘I will never eat, ever’  (JM)

Maingard [73] noticed that the nasalisation was often heard in the names of clans, where the word for ‘clan’ is ?’ais. This phenomenon was probably responsible for early spellings that reflect the presence of a nasal in some of the old clan names, such as Goringhaikua (for !Uri?’aikua). (It may also account for some spellings where the word for clan is shown with a nasalised diphthong.)

Both Beach and Engelbrecht indicated this type of click by means of the click symbol with a following glottal stop symbol [?]. We will preserve Meinhof’s convention of using the apostrophe that symbolises an ejective, partly because there is a sense in which these clicks reflect what might be termed a special type of ejection, and partly because the texts transcribed by Meinhof constitute a major component of the heritage corpus. Lloyd and Engelbrecht wrote these clicks without any additional symbol after the sign for the click, although Lloyd seems to have experimented briefly with the use of a glottal stop symbol. (The clicks with this feature are represented in the Namibian orthography without any letter after the click symbol, and are accordingly ordered first in the arrangement of the Khoekhoegowab Dictionary.)

       |’ub        ‘salt’

       |’uib        ‘stone’  

       ?’ais        ‘clan’

       ?’o        ‘die’

       !’ãb        ‘brother’ [ti !’ãb ‘my brother’]  

       !’ãb        ‘hunger’  

       !’ao        ‘fear’

       !’aub        ‘veld, hunting ground’

       !’os        ‘axe’

       ?’an        ‘know’

       ?’oab        ‘wind’  

       ?’om         ‘sew’

       ?’ukua        ‘food’

(iv) The clicks with delayed aspiration: h, ?h, !h, ?h

As Beach [74] described the delayed aspirated clicks of Nama and Kora, they are produced by first making the click, then silently releasing the back closure, and lastly ‘making the glottal fricative h’. These clicks are indicated in the Namibian orthography as they are for Kora, which is to say by means of a letter ‘h’ written after the click.

Like the glottalised clicks described above, the delayed aspirated clicks are associated with a latent prenasalisation. This normally slight degree of nasalisation re-emerges when the clicks are preceded by a vowel. Beach [75] explained the phenomenon as a kind of ‘venting’ of pulmonic airflow through the nose, which is initiated to ensure the silent release of the back closure. Beach’s idea was later developed by Traill, [76] and this type of accompaniment associated with nasal venting has come to be referred to as ‘delayed aspiration’. [77] Since it is a predictable feature, the prenasalisation is not explicitly indicated in the Namibian orthography for Khoekhoe, although some of the other Khoisan languages use an apostrophe before the letter ‘h’ to indicate the delayed nature of the aspiration (as in the name of the language Ju|’hoan). It is probably this feature that led Meinhof to give variant spellings for |hõab ‘wild cat’ and |hai ‘yellow’ as ‘nuab’ and ‘|nai’ respectively.

       |hai        ‘yellow’

       |hõab        ‘wild cat’ (note Swati ingcwa [in|ghwa] ‘cat’)

       |hobab        ‘grave’ (note Swati lingcwaba [lin|ghwaba], Xhosa ingcwaba ‘grave’)

       |homkua        ‘clouds’

       |huru        ‘play’

       ?habo         ‘dream’

       ?habob         ‘sandal, shoe’  

       !hanab        ‘garden’

       !harib        ‘large settlement, town’

       !haridas        ‘small settlement, village’

       !horob        ‘grain’ (used by Ouma Jacoba Maclear for ‘bread’)  

       ?hanu ‘right, correct, good’  

(v) The clicks with ordinary aspiration or velar frication: x, ?x, !x, ?x

This accompaniment is indicated in Namibian Khoekhoe by the letters ‘kh’ after the click symbol, and it is likely that the accompaniment of ‘velar frication’ developed out of ordinary aspiration (that is, without the complication of prenasalisation), much as the ordinary velar affricate [kx] occurs in Kora as a variant of the aspirated velar stop kh. The release of the posterior closure is silent in the case of these clicks. [78]

In the Word List that accompanies his 1928 study of Kora, Engelbrecht included a series of clicks that he represented with a following letter ‘k’. Although this was often the convention for indicating a plain click in the work of other authors, in this case the words involved for the most part match words that other authors recorded as clicks accompanied by velar frication. Since Engelbrecht used a separate symbol (the letter ‘x’) for the velar fricative when it occurred independently, we should probably not assume that the click accompaniment indicated by his ‘k’ was anything other than simple aspiration – and since the occurrence of clicks with aspiration in place of frication may have been a particular dialectal feature, the words spelled this way by Engelbrecht are listed separately in the Dictionary, as a sub-set of the clicks with frication. (The dialect of Engelbrecht’s first consultants also seems to have lacked clicks with the ejective affricate accompaniment.)

       |xa        ‘with’ (postposition)

       |xomma        ‘beg, pray’

       |xuru        ‘sour’  

       ?xãb         ‘month, moon’

       ?xaisi        ‘eight’

       ?xoba        ‘open’

       !xãkua         ‘kidneys’  

       !xaris/b        ‘honey beer’  

       !xo        ‘catch’  

       ?xonsa        ‘sweet’  

       ?xoab        ‘elephant (male)’

       ?xum        ‘shave’  

(vi) The clicks with affricated ejection: |kx’, ?kx’, !kx’, ?kx’

One of the respects in which Kora differs from Nama and Namibian Khoekhoe, and also makes it seem closer to the Kalahari varieties, is its use of clicks, at least in the eastern dialects, where the posterior closure is released audibly as an ejective affricate. In his earlier period, Maingard did not hear this particular click accompaniment in the speech of any of his consultants, but at a later date he reported [79] finally encountering it – specifically in some of the easternmost (which is to say, Upper Gariep) dialects of Kora, such as the variety spoken by consultants from the Kx’am?’õakua or Right-Hand people. [80] In western varieties of Kora as well as other dialects of Khoekhoe, where the ejective affricate release does not occur, equivalent words were sometimes recorded as having merely the glottalised accompaniment, while others were found to be plain. It is interesting to note that in varieties (whether dialectal or individual) where the ejective affricate was recorded, the simple aspirated click was also more likely to be fricated.

We were fortunate to find examples of this click in the speech of our two consultants, and it has emerged during the course of our subsequent studies [81] that, much as in the case of the ordinary ejective affricate spelled kx’, the release of the posterior closure for these clicks appears to be uvular, rather than velar as has previously been assumed. The spectrographic profile of this feature closely resembles that of the ordinary ejective affricate kx’, which, as we have already noted, is in strict terms an ejected uvular plosive [q’] followed by a brief uvular trill, where the trill seems to be precipitated by the ejection, and quickly loses its periodic character to end as noisy uvular frication [?]. (This common and even diagnostic characteristic of uvular sounds is known as ‘uvular scrape’.) It is notable that we have found a few variant pronunciations where this accompaniment is heard simply as [q] – and it seems likely that the ejective component in these cases has simply been produced too weakly to cause the characteristic trill. (Note that in the case of the Namibian counterparts, the clicks written with no additional symbols after them are in fact glottalised.)

       |kx’a       ‘be sharp’ [[|kx’asa kõas ke ‘the knife is sharp’] (Nama |a)  

       |kx’ã        ‘steal’ (Nama |â)  

       |kx’aba        ‘red’ (Nama |apa)  

       ?kx’aro       ‘thin’ (Nama ?aro ‘narrow’)

       !kx’amsa        ‘green’ (Nama !am)  

       !kx’amma        ‘porridge’

       ?kx’oa         ‘go out’ [?kx’oa na] (Nama ?oa)  

       ?kx’onib        ‘worm’ (Nama ?unib)  

(vii) The clicks with nasalisation: n, ?n, !n, ?n

The nasalised clicks in Kora are much like those of Namibian Khoekhoe, and feature the onset of nasalisation well before the click burst. The nasalisation persists throughout the production of the click, and the release of the posterior closure of the click is not audible. The nasalisation is auditorily salient, which makes it unsurprising that some of the older records for Cape Khoekhoe and the eastern dialects typically use just the letter ‘n’ to represent nasalised clicks in the original names of various rivers. (The dental click in particular may be almost unheard when it is prenasalised.)

       ||naidab        ‘baboon’  

       |nam        ‘love’

       |nub        ‘leg, foot’  

       ?nãub        ‘ear, leaf’ [?nãukha (common dual)]  

       ?nubu         ‘churn’

       !nab        ‘belly, insides’ [!nakua]  

       !naib        ‘giraffe’

       !nani        ‘six’

       !nomab        ‘root’

       !nona        ‘three’

       ?na        ‘kick’  

       ?nam        ‘whistle  

       ?namma         ‘cloak, kaross’  

       ?nau        ‘hit’

       ?nu        ‘black’  

       ?nu        ‘sit  

(viii) A solitary nasalised voiced click: n!g

Ouma Jacoba Maclear gave us a single token of a word that contained a nasalised click with audible release of the posterior closure as a voiced stop [?]: [82]

       n!gausa        ‘brown’  

The two Nama-speaking members of our fieldwork team were not familiar with this word, and we were able to find a match for it only in Leonhard Schultze’s record [83] of an old Nama coat colour term, !gãu!huni, which he described as being ‘after the grey-yellow colouring of the dassie, used of cattle, horses, goats and dogs’. (The Khoekhoegowab Dictionary has the verb !gau ‘become besmeared, of the mouth’ and !gau(a)am ‘with a black or dark muzzle’, where the equivalent of the latter in Kora would probably have been !gauxakx’am.)

This accompaniment is not typically found in Khoekhoe dialects, though it occurs regularly in languages of Kalahari KHOE. It is notable that Schultze’s transcription of the word reflects a voiced click (or at least one with an audible posterior release), as well as nasalisation of the diphthong, and it is conceivable that the Kora clicks occasionally reported in the past to feature true voicing (which is to say, voicing that commenced ahead of the click) may originally have been clicks of this type.  

3.4 The Kora system of tone melodies

Like all other Khoisan languages, Kora makes use of lexical tone. The existence of a tone system in Khoekhoe varieties was noted early on by the missionaries who worked on Nama – such as Johann Wallmann, who in the introduction to his 1857 grammar of the language, gave the following three near-identical words to demonstrate the contrastive role of tones. [84]

       !xaisbn        ‘place’ (Nama !khais (High-Low))

       !xaib        ‘cold’ (Nama !khaib (Low-SuperHigh))

       !xaib        ‘headcloth’ (Nama !khaib (SuperHigh-High))

At this early date, most linguists simply distinguished between High and Low tones, although others would later go on to identify High, Mid and Low tones. (The tones indicated for the three examples above have been modernised, and are based on those given by Haacke and Eiseb in their Khoekhoegowab Dictionary.)

Leonhard Schultze [85] was perhaps the first observer to recognise and indicate falling and rising tone patterns in addition to level ones, while the first methodical study of the tonemes of both Nama and Kora was contributed by Beach, whose seminal work was subsequently built upon by Roy Hagman [86] and later Haacke, [87] who both focused on Namibian Khoekhoe. All of these studies treat the tones as register tones (rather than contour tones), where each tone is carried individually by a vowel or a nasal.

3.4.1 The citation melodies of Kora

The tone melodies of the Khoekhoe languages typically have a basic (or ‘citation’) form, which is to say, the melody that is used when a word is spoken in isolation or at the beginning of a sentence. These melodies change systematically in certain syntactic contexts, where each citation melody has (i) its own alternative melody associated with the occurrence of the word in a compound form, and (ii) another alternative melody associated with the occurrence of the word at the end of a phrase. While these phenomena were noted in passing by both Meinhof and Maingard, the details were never fully documented for Kora, and it was only much later that the equivalent phenomenon in Namibian Khoekhoe was described at length by Haacke. [88] (All isolated forms heard in the sound files linked to the electronic version of this book are true citation forms and have not been snipped out of longer phrases. In a handful of cases where the only example of a word occurs within a short phrase, the phrase is given in full.) The alternative melodies will be discussed briefly in a later section below.

As for Kora, Wuras at least attempted to provide tone markings for some of the words in his Vocabulary as well as versions of his Catechism, typically using an acute accent (á) to indicate a High tone, and a grave accent (à) to indicate a Low tone, while leaving the remaining tones unmarked (probably on the basis that he perceived them to be neutral or median tones). Lloyd’s transcriptions of Kora a few decades later are fairly fluid, and should perhaps be seen as reflecting work that at the time was still in progress. Her manuscripts show many corrections, with frequent substitutions of a grave accent for originally acute ones, so that it is not always easy to work out their significance. (Maingard established [89] through correspondence with Lloyd’s niece Dorothea Bleek that in her aunt’s system, the acute accent was usually intended to indicate stress.) The illustration (Figure 3.7) shows a page from one of Lloyd’s Kora notebooks, where the word ‘haip’ for haib ‘tree’, which typically has a very low tone in our data, is marked with an acute accent, while the word ‘|u?ip’ for |’uib ‘stone’, which typically has a very high tone, is assigned a grave accent in her amendment.

Meinhof, whose work on Kora appeared in 1930, mentioned [90] the customary three tones, namely High, Mid and Low, and also commented that the tone patterns of words could undergo various changes within the context of a sentence, as well as in certain derived forms, such as reduplicated verbs. Beach [91] subsequently estimated, however, that in the Glossary that accompanied his grammatical sketch, Meinhof provided tone markings for only about half of the roughly 2 000 entries, adding that, in most cases, the older scholar marked only level tones and never indicated any rising or falling tones – except by implication in a handful of cases where different tones are indicated on different syllables of a word.

Writing in the 1960s, Maingard [92] identified four main tonemes in the speech of the Bloemhof Korana with whom he had worked in the 1920s and 1930s, adding that in a few rare cases there was also a fifth toneme. The five tone melodies he identified consisted of two level ones, a rising one, a falling one, and the rare very low one, for which he gave examples as follows:

· High level melody [H-H], as in ‘!kab’‘brother’

· Low rising melody [L-H], as in ‘!kam’ ‘kill’

· Mid level melody [L-L], as in ‘mu’ ‘see’

· High falling melody [H-L], as in ‘?kã’ ‘enter’

· Low or very low melody [SL], as in ‘hoan’ ‘all’

While some of these examples (those that do not end in a nasal) may seem to suggest the use of a contour tone across a monosyllable, the roots in question are analysed as having two prosodic ‘morae’, where the tone-bearing units are taken to be two vowels in sequence.

Like Meinhof three decades before him, Maingard noted [93] that the basic tone melodies of words underwent changes in certain syntactic contexts. Other changes to the tone melodies were seen to occur in reduplicated forms. Maingard also contributed the astute observation [94] that the various tones and their tone variants might ‘be bound up not with the word-tone, but with the sentence-tone, which exists very prominently in Korana’. (By ‘sentence-tone’ he probably meant the intonation contour of a sentence.)

It was Beach’s study, published in 1938, that became the definitive one, however. Working mainly with just one consultant (Benjamin Kraalshoek, from the Kimberley Korana community), Beach determined that Kora had four fundamental ‘tonemes’ (citation melodies). (He briefly noted, but did not explore, the kinds of systematic variations noticed by Maingard and Meinhof.) In order to give some idea of the relative values of the four Kora tonemes, Beach plotted their melodies on to a musical stave, using four reference words as his examples, where |ui ‘one’ illustrated what he described as a ‘High level’ melody, !xai ‘cold’ a ‘High rising’ melody, ?nae ‘sing’ a ‘Mid level’ one, and !xae ‘dark’ a ‘Low-mid falling’ one. [95] Beach’s drawing is replicated in Figure 3.8, which also provides a translation of his High, Mid and Low tones into oppositions between High and Low values (with the additional options of SuperHigh and SuperLow), along the lines suggested by Johanna Brugman. [96]

Since tonal distinctions were still clearly and consistently being made by Ouma Jacoba Maclear, we had high hopes of being able to confirm the presence of Beach’s four citation melodies. We do not have recordings for each of the four reference words used in the sketch above, but Beach fortunately provided lists of additional examples for each of the four melodies (as set out in Table 3.5). When we began to make our comparisons, however, drawing initially from Beach’s lists, we were slightly disconcerted to find that the words as pronounced by Ouma Jacoba do not always fall consistently into the same melodic groups as those determined for them by Beach. (There may be various reasons for this, not least of them the advanced age of our consultant, as well as the fact that, unlike Benjamin Kraalshoek, who came from Kimberley, Ouma Jacoba belonged to the Bloemhof Korana.) What we hear in Ouma Jacoba’s speech is nevertheless a set of four clearly distinct tone melodies – just as expected, where two are level [H-H] and [L-L], one rises [L-H] and one falls [H-L].

The table in Figure 3.9 sets out the four citation melodies as produced by Ouma Jacoba, with example words for each. As this chart shows, the High level toneme [H-H] in her speech is sometimes heard as a SuperHigh variant, while the Low level toneme [L-L] is only expressed as SuperLow. There are relatively few examples for either of these two extreme melodies in our data, and most of the words reflect citation melodies that are moving. The rising melody has the contour [L-H] for the most part, but while the falling one is occasionally the High falling melody [H-L] noted by Maingard, it more often occurs in real terms as a melody that falls from Low to SuperLow – much as implied by Beach’s musical notation.

Fig. 3.9. The four Kora citation melodies in the speech of Ouma Jacoba Maclear.

The diagrams in Figure 3.10 were made using Praat software, where the first drawing shows the pitch contours for words illustrating the High and Low level melodies, and the second shows contours for the two moving melodies, where the example words are clickable in the electronic edition.

Nama has two additional melodies, but these will be discussed in the section on tonogenesis below.

3.4.2 The two classes of alternative tone melodies used in particular contexts

The tone melodies described so far have been the basic or citation forms. As noted earlier, though, these melodies may change in certain contexts. The alternative melodies are briefly discussed next.

(i) Alternative melodies associated with compound forms

The first class of alternative melodies is assigned to words when they occur in various compounds, including verbal compounds such as uha ‘bring’, from u ‘take’, ha ‘come’, and reduplicated forms such as |kx’a|kx’a ‘sharpen’, from |kx’a ‘be sharp’. (In Namibian Khoekhoe, certain verb extensions, such as the applicative –ba, have similar effects on tone melody.) We do not have many examples in our data of words used both in isolation and in the context of such formulations, but in the case of the reduplications |kx’a|kx’a, as in kõas |kx’a|kx’a  ‘sharpen the knife’, and thoathoa  ‘begin’, the first occurrence of the word in each case has a high or rising tone, while the second takes a low or falling melody. 

(ii) Alternative melodies associated with final occurrence in an expanded phrase

A second class of alternative melodies is assigned to words when they occur in certain syntactic contexts. In reality, of course, it is normal for nouns and verbs to be used in expanded phrases, where it is the general rule in Khoekhoe languages that nouns and verbs occur at the end of the phrases that contain them. A noun, for example, may be preceded in Kora by an adjective or various specifiers, including quantifiers such as hoa ‘all’, demonstratives such as he ‘this’ or ?na ‘that’, and possessives such as ti ‘my’, while a verb may be preceded by various markers of tense or aspect.

In Namibian Khoekhoe, the tendency is very broadly for the higher tone melodies of words to be replaced by lower or falling melodies when those words appear phrase-finally in an expanded phrase (although there is one notable exception to this general principle). We would expect to find in Kora a broadly similar pattern of low or falling melodies in phrase-final positions, and while we do not have sufficient data to determine the full set of alternate melodies associated with each of the four citation melodies, we can confirm that words with generally high or rising citation melodies (such as |’uib  ‘stone’) typically occur with lower alternatives when they are used at the end of an expanded phrase (such as |kx’a |’uib  ‘sharp stone’). (This seems to confirm Maingard’s insight that the determining factor may ultimately be the overall intonation contour of the sentence, since it is precisely the falling curve of the ordinary declarative sentence that is preserved by the use of the alternate low or falling lexical melodies.) [97]

3.4.3 The theory of tonogenesis in Khoekhoe

When Beach came to describe the Kora tonemes, it was after he had completed his study of Nama, where he had found six contrastive tone melodies. It puzzled him that he could find only four tone melodies in Kora, where for each of the two lower melodies, Nama seemed to have two corresponding melodies, where the extra Nama melody in each case had a low or very low onset, followed by a rise.

The question this posed for Beach was how the additional melodies in Nama could have emerged, and why they should have the lower tone onsets. Eventually he decided that the various previous authors who had alluded to the presence in Kora of a contrast based on voicing were perhaps not wrong after all, since he began to detect an apparent correlation between the presence of a voiced (or once voiced) segment – and a lower tone on the following vowel. He concluded that a formerly voiced series [98] of both ordinary and click consonants may have had a lowering (or depressor) effect on following tones. (This is a natural process not unusual among tone languages elsewhere in the world, and is thought to be caused by certain physiological aspects of the voicing mechanism.) [99] Beach theorised that the process must have gone further in Nama and the Namibian dialects, which for the most part no longer make use of contrastive voicing – presumably because the additional tonal contrasts eventually became semantically significant in themselves. [100]

We had hoped that the data from our two speakers of Kora might provide confirmation of Beach’s theory, particularly since the speech of our consultants seemed to include a residual voiced series for both egressive and ingressive consonants (even if that voicing is ambiguous in some cases, and only marginally phonemic, as we have seen). We reasoned that if the process of tonogenesis was a gradual one, we might expect to find some occasional traces of its beginnings still occurring in association with these apparently voiced segments and other potential depressor consonants. In the end, though, we have been left with more questions than answers, partly because of uncertainties surrounding the voiced segments, and partly because the melodies in our data do not consistently match those given by Beach for the same words. Our sample size is in addition so small that it does not permit us to make statistically meaningful generalisations. All of this has led us to concede that the material we were able to salvage is perhaps too fragile to withstand the burden of any strong argument. (The investigation is written up elsewhere.) [101]

Given the vagueness of the tone markings provided by many of the earlier authors, and given the many discrepancies for particular words between the melodies recorded by Beach in Kimberley and the melodies heard in our own recordings made in Bloemhof, we have thought it best not to attempt tone markings for the entries in the Dictionary. The tables in Table 3.5 nevertheless set out for reference purposes the groups of example words given by Beach for each of the citation melodies identified by him, and readers will also have the opportunity to listen to Ouma Jacoba Maclear’s pronunciations via the electronic edition. For purposes of reading the heritage texts aloud, it should be a reasonable enough compromise in cases of uncertainty to use the Nama melody for the equivalent word – without sacrificing the recognition that the tone systems of the two languages were ultimately different.

Table 3.5 The complete list of examples given by Beach for each of the four Kora tonemes in the speech of Benjamin Kraalshoek.

This ia a chapter from the book: Kora: A Lost Khoisan Language of the early Cape and the Gariep by Menán du Plessis

Endnotes

[1] Carl Meinhof, “Wörterverzeichnis,” in Der Koranadialekt des Hottentottischen (Supplement 12 to the Zeitschrift für Eingeborenen-Sprachen) (Berlin: Reimer, 1930), 78–145. ↵

[2] Douglas Beach, The Phonetics of the Hottentot Language (Cambridge: Heffer, 1938), hereafter referred to as Bch. ↵

[3] Carl Meinhof, Der Koranadialekt des Hottentottischen (Supplement 12 to the Zeitschrift für Eingeborenen-Sprachen) (Berlin: Reimer, 1930), hereafter referred to as Mhf. ↵

[4] Louis F. Maingard, “Studies in Korana history, customs and language,” Bantu Studies 6, no. 2 (1932), 148–150; Koranna Folktales: Grammar and Texts (Johannesburg: Wits University Press, 1962), 3–10. Maingard’s “Studies in Korana history” will hereafter be referred to as Mgd1932. ↵

[5] Anthony Traill, Phonetic and Phonological Studies of !XOO Bushman (Research in Khoisan Studies 1) (Hamburg: Helmut Buske, 1985). ↵

[6] Hirosi Nakagawa, “Aspects of the Phonetic and Phonological Structure of the G|ui language,” (Johannesburg: University of Witwatersrand PhD thesis, 2006). ↵

[7] Peter Ladefoged and Keith Johnson, A Course in Phonetics, 6th ed. (Wadsworth: Cengage Learning). ↵

[8] Numerous patient tutorials from Roger Lass have been invaluable – although he is most decidedly not responsible for any errors that may have slipped through despite his best efforts. ↵

[9] The reason for not specifying sources except in the most obvious cases is that it can be difficult to pinpoint an exact donor language, even when it is clear that a particular word originates from a language of the BANTU family. The other difficulty is that the marking of some words as loans sets up a natural expectation that all borrowings will be exhaustively indicated, whereas there are probably many more loanwords in Khoekhoe that have not yet been detected. ↵

[10] Menán du Plessis, “The damaging effects of romantic mythopoeia on Khoesan Linguistics,” Critical Arts 28, no. 3 (2014), 569–592. ↵

[11] Of course, an argument could be made that it is the close-mid vowels [e] and [o] that are in some sense ‘more basic’, while it is the open forms [?] and [?] that are the conditioned variants. ↵

[12] They arise most often by harmonisation with a following high vowel, or where the final segment is a nasal. (It is likely that words ending in m or n once had high vowels i or u after them.) It is possible that the harmonisation was originally based on the respective features of advanced and retracted tongue root, rather than vowel height. ↵

[13] Jan A. Engelbrecht, “Studies oor Korannataal,” Annale van die Universiteit van Stellenbosch 6, ser. B, no. 2 (1928), 7, hereafter referred to as Ebt1928. ↵

[14] Bch, 217. ↵

[15] It is a matter of ongoing debate whether differences in vowel length are genuinely phonemic – which is to say, are used to signal an actual contrast in meaning. That the contrastive factor is possibly more a matter of tone melody than length has been suggested by Niklaas Fredericks in “A study of dialectal and inter-linguistic variations of Khoekhoegowab,” (Bellville: University of the Western Cape PhD thesis, 2013) for Namibian Khoekhoe; and by Amanda Miller, “Contrastive vowel-length and variable weight reduplicative templates in Ju|’hoansi,” in Cornell Working Papers in Linguistics 18 (2001), ed. Arthur Bell and Paul Washburn, 104–125. Fredericks finds that the high and low tones interfere with the perception of vowels so that they appear long. The high tone appears to trigger slight lengthening of the vowel but this is illusory: some sounds believed to be longer are in fact shorter, which strengthens the argument that it is not length that is phonemic, but rather tonal melody. ↵

[16] For the same reason, the need will never arise for both the tilde and the macron to be used simultaneously. ↵

[17] The Tswana fricative followed by the open o is more likely to be produced further back, and is often actually uvular. ↵

[18] Bch, 50. ↵

[19] Niklaas Fredericks, personal communication, September 2015. ↵

[20] This kind of progression reflects a natural sound shift, seen also, for example, in the difference between the pronunciation of German kleine and Afrikaans klein ‘small’. In another stage of this shift, cognate words are sometimes seen in other varieties with the vowel i, as in the case of !aixab ‘doctor, sorcerer’, which has been borrowed into both the !Ui language |Xam (as g!ixa) and the Nguni language Xhosa as ugqira [u!gi:xa]. This kind of shift seems to have been a particular characteristic of the Griqua variety. ↵

[21] Mhf, 78–119. ↵

[22] As noted earlier, this may reflect a Giri influence: Ouma Jacoba told us that her father was Griqua, while Oupa Dawid mentioned that his parents had come originally from Griquatown. ↵

[23] Bch, 50, 206. ↵

[24] Again, this kind of sound shift is a natural and frequently encountered one. A similar progression is seen in the difference between the pronunciation of Dutch blaauw and Afrikaans blou ‘blue’. There are sporadic cases cross-dialectally in Khoekhoe where [a?] has shifted to [o:] or [u:]. ↵

[25] It is an intriguing aspect of Khoisan phonetic inventories in general that they almost never feature bilabials (whether oral or nasal, stop or fricative, voiced or voiceless) as initial segments in non-borrowed lexical items, even though there are grammatical morphemes in most of the languages that begin with such sounds. ↵

[26] The narrow aperture for the voiced bilabial fricative is created by placing both lips together, and is not quite the same sound as the labiodental [v], which is articulated by placing the lower lip against the upper teeth. ↵

[27] Mgd1932, 151.  ↵

[28] The sounds h and the murmured (or ‘breathy-voiced’) ? are referred to for convenience as glottal fricatives, although producing them does not strictly involve articulators in the usual sense of the word. ↵

[29] This sound may be post-velar or even uvular for some speakers of Afrikaans, certainly ahead of a low or back vowel. ↵

[30] Bch, 221. ↵

[31] Bch, 252. ↵

[32] Maingard, Koranna Folktales, 7. ↵

[33] Lucy Lloyd, “Preface,” Specimens of Bushman Folklore (London: George Allen & Co., 1911), viii. ↵

[34] Ebt1936, 199. ↵

[35] Rainer Vossen, Die Khoe-Sprachen: Ein Beitrag zur Erforschung der Sprachgeschichte Afrikas (Cologne: Rüdiger Köppe, 1997), 523. ↵

[36] These are associated with a palatalising influence of the Class 5 prefix, while the features of aspiration and ejection in Sotho-Tswana languages pattern in much the same way as they do in the case of segments preceded by a nasal, with ejectives reflecting originally voiced segments (Denis Creissels, “Remarks on the sound correspondences between Proto-Bantu and Tswana (S.31),” in Bantu Historical Linguistics: Theoretical and Empirical Perspectives, ed. Jean-Marie Hombert and Larry M. Hyman (Leland Stanford Junior University: Centre for the Study of Language and Information, 1999), 297–334. ↵

[37] J. A. Louw, Handboek van Xhosa, with J. B. Jubase (Johannesburg: Educum Publishers, 1978), 11. ↵

[38] Bch, 224. ↵

[39] Jan A. Engelbrecht, The Korana (Cape Town: Maskew Miller, 1936), 200. This work will hereafter be referred to as Ebt1936. ↵

[40] Mhf, 25. ↵

[41] Carl Meinhof, “Versuch einer grammatischen Skizze einer Buschmannsprache,” Zeitschrift für Eingeborenen-Sprachen 19 (1928/29), 161–88. ↵

[42] Dorothea Bleek, A Bushman Dictionary (New Haven, Connecticut: American Oriental Society, 1956), 221. ↵

[43] The reason for the small note of doubt is that in her transcriptions of the Kora stories dictated by Piet Links, Lucy Lloyd also occasionally represented certain sounds as ts’. In these cases, however, the words involved are recorded elsewhere by other authors with the plain affricate. An example occurs in the story of Moon and Hare, where the word tsi (meaning ‘and’) is spelled ‘tss’?’. Lloyd also once or twice wrote ‘tss’a’ ‘you (masculine singular)’ for the more usual tsa. These few instances suggest that at least some occurrences of ts’ were perhaps simply isolated cases of variation in the ordinary rapid speech of fluent individuals, which observers happened to capture in faithful but perhaps overly narrow phonetic transcriptions. A closer look at the clan name in Meinhof’s set of words triggers further doubt. Although Engelbrecht also independently noted the name ‘Ts’oa ↵’ain’, giving it as that of the clan to which Andries Bitterbos belonged, it is clear from statements made elsewhere that Bitterbos belonged to the Kat clan. This makes us inclined to suspect that the spelling Ts’oa?’ai simply reflected a mishearing of |’Hõa ↵’ain. The word ts’uni for ‘larva’ may have been a clickless form of ?kx’onib ‘worm’. ↵

[44] Some scholars believe that the glottalisation initiating these vowels constitutes an actual glottal stop segment. ↵

[45] Distributions of this kind point to the likely former existence of pre-stem-initial morphology. ↵

[46] Menán du Plessis, “The click inventory of Kora (or !Ora) (Khoekhoe, KHOE) reappraised in the light of data from two last speakers,” paper presented at the Fifth South African Micro-Linguistics Workshop (SAMWOP5), (24–26 November 2016, Bloemfontein). ↵

[47] The Sotho-Tswana languages feature ordinary affricates such as kg (phonetic [kx]), which in certain conditioning environments (before low or back vowels) are expressed as aspirated uvular affricates (phonetic [q?h]), produced so far back as to introduce ‘uvular scrape’ (A. N. Tucker, The Comparative Phonetics of the Suto-Chuana Group of Bantu Languages (London: Longmans, Green, 1929), 55). These sounds are not generally described as ejective, although an ejective series might really have been expected to occur, so as to distinguish the uvular reflex of Proto-Bantu *g from *k in the pre-nasalised environment. (The reason for the merger may lie partly in the natural tendency for a voiced uvular stop to be expressed as a fricative.) ↵

[48] This pattern may reflect an older distinction between vowels produced with ‘advanced tongue root’ and those produced with ‘retracted tongue root’. ↵

[49] Wilhelm Bleek, (Bleek Collection, Manuscripts and Archives in the Special Collections of the University of Cape Town Libraries), |Xam Notebook 001, 5. ↵

[50] This sound is typically spelled ‘kl’, and is described as an ‘ejective velar affricate consonant, or ejective velar lateral affricate’ (Clement Doke, D. M. Malcolm, J. M. A. Sikakana and B. W. Vilakazi, English–Zulu, Zulu–English Dictionary (Johannesburg: Wits University Press, 1990), 420). Xhosa also has a small number of words spelled with ‘kr’ to indicate a similar ejective velar affricate, as in –kroba ‘look through a hole, peep’, and –krwela ‘scratch, scrape’. Both these last examples are of BANTU origin, however, and have clear cognates in Sotho-Tswana languages. Although Meinhof proposed that Nguni words featuring this sound must have been borrowed from a Khoisan source, the words usually listed are for the most part not found in any of the extant Khoisan languages, and in many cases do not resemble typical Khoisan stems. It is interesting, though, that Nakagawa (“Phonetics,” 126) reports finding some speakers of the Kalahari Khoe language G|ui who produce the uvular ejective affricate q?’, as in q?’?i ‘vulture’ (compare Kora kx’anis ‘bird), with a lateral release – which suggests at least a parallel with Nguni. ↵

[51] Wilfrid H. G. Haacke and Eliphas Eiseb, A Khoekhoegowab Dictionary, with an English-Khoekhoegowab Index (Windhoek: Gamsberg Macmillan, 2002), 99. ↵

[52] Mgd1932, 149. ↵

[53] It is possible that all or some of these forms have been borrowed. It is notable that the word –laula in the Nyanja (or Cewa) language of Malawi and eastern Zambia at one time meant ‘swear, utter things unlawful; to speak words by the spirits; of trees and voices supposed unnaturally to swear; to swear at a person, to meet with bad omen […]; also, to prophesy, tell of coming misfortune’ (David Clement Scott, Dictionary of the Nyanja Language, ed. Alexander Hetherwick (London: Lutterworth Press, 1929), 232). The equivalent word in Sotho, -laola, seems to have only the last meaning, being glossed (A. Mabille and H. Dieterlen, Southern-Sotho-English Dictionary (South African orthography), rev. R. A. Paroz (Morija: Morija Sesuto Book Depot, 1988), 263) as ‘throw the divining bones’. Tswana (J. Tom Brown, Setswana Dictionary: Setswana-English and English-Setswana (Johannesburg: Pula Press, 1982), 147) similarly has –laola, ‘decide by casting the dice’, but like Sotho, does not seem to include the meaning ‘swear’ in the semantic range of the word. Since the meaning ‘swear’ is no longer associated with these words in the South African languages, it seems that if the word lau (~ luu) in Khoe is a borrowing, the transfer must have been made either directly from a language closer to the Malawi (or eastern Zambian) region, or else perhaps occurred at a time when the Sotho-Tswana languages still preserved the older meaning. ↵

[54] Samuel S. Dornan, “The Tati Bushmen (Masarwas) and their Language,” Journal of the Royal Anthropological Institute 47 (1917), 109. ↵

[55] Bch, 225–226. ↵

[56] Mgd1932, 149. ↵

[57] Amanda Miller, Johanna Brugman, Bonny Sands, Levi Namaseb, Mats Exter and Chris Collins, “Differences in airstream and posterior place of articulation among N|uu clicks,” Journal of the International Phonetic Association 39, no. 2 (2009), 129–161. ↵

[58] Although some contemporary linguists now refer to the (post)alveolar click [!] simply as ‘alveolar’, we have chosen to retain the full (and standard IPA) form, so as to avoid confusion with older literature (with which many South Africans will be familiar), in which some scholars originally referred to this click as palatal. We also use the full form of the name for the palato-alveolar click [?], rather than the commonly used current abbreviation ‘palatal’. (This click has in the past been referred to by some scholars as ‘alveolar’.) ↵

[59] Bch, 80. ↵

[60] The only BANTU language known to use all four clicks is Yeyi, which is a language of the Okavango region. ↵

[61] Ebt1938, 202. ↵

[62] Anthony Traill, Phonetic and Phonological Studies of !XOO Bushman (Research in Khoisan Studies 1) (Hamburg: Helmut Buske, 1985), 99. ↵

[63] Some linguists have dismissed the term as ‘vacuous’, but its non-specificity was rather the point of such an all-encompassing term in the first place. ↵

[64] In the current orthography for Namibian Khoekhoe, the velar affricated accompaniment is indicated by the letters ‘kh’ after the click symbol, where this convention gives a truer indication of the phonetic reality, which is that the click may be aspirated in some varieties, but has variants where the aspiration is heard as velar frication. In the set of conventions used by Christa Kilian-Hatz in the Khwe Dictionary (Cologne: Rüdiger Köppe, 2003), the clicks with ‘voiceless velar affricate accompaniment’ are indicated by means of a letter ‘x’ after the click symbol, and the clicks with ‘affricated velar ejective accompaniment’ by means of a letter ‘x’ with following apostrophe. It seems to be a grey area whether the fricative (glottal or velar) is associated with the simple click (as for example, [!h ~ !x]), or with an audibly released posterior closure (as [[!qh ~ !q?]). ↵

[65] Beach (Bch, 82–83) described the plain (or voiceless unaspirated) clicks as involving a ‘practically silent’ release of the posterior closure. At the same time, he suggested that the release may well be heard as a ‘weak velar plosive efflux’, which, like the voiceless unaspirated k, is difficult to differentiate from g. It was this difficulty, he felt, which resulted in the inconsistent descriptions of older writers such as Henry Tindall (for Nama) and Wuras (for Kora), who sometimes wrote ‘k’ and sometimes ‘g’ after a click symbol. ↵

[66] Ebt1928, 12–41. ↵

[67] Mgd1932, 150. ↵

[68] Du Plessis, “The click inventory of Kora reappraised.” ↵

[69] Patrick Midtlyng, “The effects of speech rate on VOT for initial plosives and click accompaniments in Zulu,” Selected Proceedings of the 40th Annual Conference on African Linguistics, ed. Eyamba G. Bokamba and others (Somerville, MA: Cascadilla Proceedings Project, 2011), 105–118. See also Michael Jessen and Justus C. Roux, “Voice quality differences associated with stops and clicks in Xhosa,” Journal of Phonetics 30 (2002), 1–52. ↵

[70] Some Khoisan languages, including some belonging to the Kalahari branch of Khoe, are reported to feature voiced clicks of the kind actually characterised by voice lead, where voicing begins before the release burst of the click (Nakagawa, “Phonetics,” 161). ↵

[71] The voiced uvular stop is a rather rare segment in languages of the world. In many cases where it might be expected to occur in a particular language (because of its overall phonological system), it is expressed in reality as a velar or post-velar fricative. See Bert Vaux, “A note on pharyngeal features,” Harvard Working Papers in Linguistics 7, ed. Bert Vaux and Susumu Kuno (Cambridge, MA, 1999), 39–63. ↵

[72] Nasalisation and the glottal mechanism involved in the production of an ejective are not greatly compatible. Ian Maddieson and Peter Ladefoged (“Phonetics of partially nasal consonants,” in Phonetics and Phonology, Vol 5: Nasals, Nasalization, and the Velum, ed. Marie K. Huffmann and Rena A. Krakow (San Diego: Academic Press, Inc., 1993), 254) described the prenasalised ejectives of the Nguni languages as ‘marginal, at best’. This is because nasalisation depends on a pulmonic airflow and so requires the glottis to be open, whereas ejection requires the glottis to be closed. In the case of the Nguni ejectives with prenasalisation, the nasal element is contributed by a preceding prefix, and the nasalisation ceases before the separate production of the ejective. (In the case of the Sotho-Tswana languages, the solution has been to lose the Class 9 prefix, although there may have been other reasons for this as well.) It is interesting that the Nguni language Swati has ingcwa [in|ghwa] for a ‘wild domestic cat’ (compare Kora |hõab ‘wild cat’), where the nasal element in the Swati word comes from the Class 9 prefix. Examples like this raise the possibility that prenasalisation in the Khoisan cases is not purely an accidental side-effect of the articulatory mechanism, but may have been supplied in the first place by morphology, and has resulted in the kinds of accommodations referred to as ‘delayed aspiration’, and ‘glottalisation’. (Other comparative examples include Xhosa ingcuka [in|guka] ‘brown hyena’, Kora |hukas; and Xhosa ingcwaba [in|gwaba] ‘grave’, Swati lingcwaba, Kora |hobab. There is even a dialect of the |Gui language studied by Nakagawa (“Phonetics,” 172) where the pre-nasalisation associated with delayed aspiration and glottalisation is pre-glottalised and heard as [?n]. A similar type of accompaniment was found in !Xoon by Traill (Phonetics, 134–135).) ↵

[73] Mgd1932, 150.  ↵

[74] Mgd1932, 86. ↵

[75] Mgd1932, 86–87. ↵

[76] Anthony Traill, “Pulmonic control, nasal venting, and aspiration in Khoisan languages,” Journal of the International Phonetic Association 21, no. 1 (1991), 13–18. ↵

[77] Some other Khoisan languages, such as !Xoon (Traill, Phonetics, 135, 143), and the Kalahari Khoe language |Gui (Nakagawa, “Phonetics,” 162–165), are reported to have additional accompaniment series where nasal venting is not present, and where the clicks may be straightforwardly aspirated or ejective.  ↵

[78] In some other Khoisan languages, such as N|uu, an audible release of the uvular closure may be accompanied by phonemic aspiration or uvular frication. ↵

[79] Maingard, “The Korana dialects,” African Studies 23, no. 2 (1964), 57–66. ↵

[80] Lloyd’s notebooks also show clicks followed by the Greek letter gamma (?), which seems to have been her symbol for the ejective affricate. While Maingard at one time doubted that Lloyd could have heard this sound as a click accompaniment, since he himself had not heard it among the Bloemhof Korana, he later conceded that it might indeed have been present in the speech of Piet Links. ↵

[81] Du Plessis, “The click inventory of Kora reappraised.” ↵

[82] In his study of !Xoon, Traill (Phonetics, 130) described the ‘voiced counterpart’ of the click accompanied by a voiceless uvular stop [!q] as a prenasalised click followed by a voiced uvular stop [n!?]. The unusual click noted in n!gausa may have been the Kora equivalent of such a sound. ↵

[83] Leonhard Schultze, Aus Namaland und Kalahari (Jena: Gustav Fischer, 1907), 265. ↵

[84] Johann Christian Wallmann, Die Formenlehre der Namaquasprache (Berlin: Wilhelm Hertz, 1857), 7. ↵

[85] Schultze, Aus Namaland und Kalahari, 342–363. ↵

[86] Roy S. Hagman, Nama Hottentot Grammar (Indiana University: Research Centre for Language and Semiotic Studies, 1977). ↵

[87] Wilfrid Haacke, The Tonology of Khoekhoe (Nama/Damara) (Cologne: Rüdiger Köppe, 1999). A more recent study of Namibian Khoekhoe by Johanna Brugman, “Segments, Tones and Distribution in Khoekhoe Prosody” (Ithaca, New York: Cornell University PhD thesis, 2009) takes the work of these earlier scholars forward, and suggests some revisions to the melodies previously identified, as well as providing a transposition of Beach’s High, Mid and Low tones into High and Low values. ↵

[88] Haacke (Tonology of Khoekhoe) refers to the phrase-final melodies as ‘sandhi’ forms, using a technical term devised by early scholars of Sanskrit. ↵

[89] Mgd, “Korana names,” 312. ↵

[90] Mhf, 24. ↵

[91] Bch, 236 ↵

[92] Mgd, Koranna Folktales, 7. ↵

[93] Mgd, Korana Folktales, 8. ↵

[94] Mgd1932, 150. ↵

[95] Bch, 239. ↵

[96] Brugman, “Segments, Tones and Distribution.” ↵

[97] It is possible that the lower falling melody [L-SL] so frequently used in place of the straightforward [H-L] may originally have been the alternative phrase-final melody. If this is the case, and if the alternative melody steadily began to replace the citation form, this process must already have been underway in the 1920s, since there is no doubt from his musical transcription that this is the melody Beach heard. A shift of this kind, from [H-L] to [L-SL], may have had a further ‘knock-on’ effect, with all Low tones gradually being re-analysed as SuperLow, so that the Low level melody [L-L] was eventually expressed in turn as [SL-]. ↵

[98] He suggested that the voiced segments could have been voiced consonants and clicks, but also voiced aspirated (or ‘breathy voiced’) consonants, as well as nasals, which are voiced by default. ↵

[99] See Martha Ratliff, “Tonoexodus, tonogenesis, and tone change,” in The Oxford Handbook of Historical Phonology, ed. Patrick Honeybone and Joseph Salmons (Oxford: Oxford University Press, 2015), 245–261. ↵

[100] Since Beach’s era, many more advanced studies have been made of the link between voicing and tonal depression in various languages (particularly BANTU languages) of southern Africa, both as a possible factor in tonogenesis over time, and as a factor in ongoing tonological processes (for example, where a high tone that would ordinarily be expected in certain environments does not materialise). The picture beginning to emerge is one of far greater and even baffling complexity, as reflected in the title (“Depressing facts about Zulu”) of a famous paper by Anthony Traill, J. S. M. Khumalo and P. Fridjhon (African Studies 46, no. 2 (1987), 255–274).

It now seems that very few if any contemporary languages in the region still make use of ‘true’ voicing, at least in the strict sense of ‘voice lead’, where voicing begins ahead of the consonant release. In reality, it is only the depressed tones themselves that now reflect the distinction, while the consonants that occur before them (even while they may be phonemically voiced) are in strictly phonetic terms often merely unaspirated segments associated with the ‘short-lag’ type of voicing. (See for example, Laura Downing and Bryan Gick, “Voiceless tone depressors in Nambya and Botswana Kalang’a,” Berkeley Linguistics Society 27 (2001), 65–80.) For Khoekhoe languages, nevertheless, it remains the hypothesis that, as Haacke (“Tonology [Khoekhoe]”, 96) puts it: ‘Namibian Khoekhoe has, through tonogenesis (tonal depression caused by especially voiced consonants and subsequent devoicing of those consonants), developed an additional tone, the “double low” tone, which in turn has created two new major citation melodies.’ ↵

[101] Du Plessis, “The click inventory of Kora reappraised.” ↵