This article has an unclear citation style.Learn how and when to remove this template message)(August 2018) (
|Manners of articulation|
Click consonants, or clicks, are speech sounds that occur as consonants in many languages of Southern Africa and in three languages of East Africa. Examples familiar to English-speakers are the Tut-tut (British spelling) or Tsk! Tsk! (American spelling) used to express disapproval or pity, the tchick! used to spur on a horse, and the clip-clop! sound children make with their tongue to imitate a horse trotting.
Technically, clicks are obstruents articulated with two closures (points of contact) in the mouth, one forward and one at the back. The enclosed pocket of air is rarefied by a sucking action of the tongue (in technical terminology, clicks have a lingual ingressive airstream mechanism). The forward closure is then released,[note 1] producing what may be the loudest consonants in the language, but in some languages such as Hadza and Sandawe, clicks can be more subtle and may even be mistaken for ejectives.
|This table contains phonetic symbols, which may not display correctly in some browsers. [Help]|
|• • •|
Click consonants occur at five principal places of articulation. The International Phonetic Alphabet (IPA) represents a click by placing the assigned symbol for the place of click articulation adjacent to a symbol for a non-click sound at the rear place of articulation. The IPA symbols are used in writing most Khoisan languages, but Bantu languages such as Zulu typically use Latin ⟨c⟩, ⟨x⟩ and ⟨q⟩ for dental, lateral and alveolar clicks respectively.
The above clicks sound like affricates, in that they involve a lot of friction. The other two families are more abrupt sounds that do not have this friction.
Clicks occur in all three Khoisan language families of southern Africa, where they may be the most numerous consonants. To a lesser extent they occur in three neighbouring groups of Bantu languages--which borrowed them, directly or indirectly, from Khoisan. In the southeast, in eastern South Africa, Swaziland, Lesotho, Zimbabwe and southern Mozambique, they were adopted from a Tuu language or languages by the languages of the Nguni cluster (especially Zulu, Xhosa and Phuthi, but also to a lesser extent Swazi and Ndebele), and spread from them in a reduced fashion to the Zulu-based pidgin Fanagalo, Sesotho, Tsonga, Ronga, the Mzimba dialect of Tumbuka and more recently to Ndau and urban varieties of Pedi, where the spread of clicks continues. The second point of transfer was near the Caprivi Strip and the Okavango River where, apparently, the Yeyi language borrowed the clicks from a West Kalahari Khoe language; a separate development led to a smaller click inventory in the neighbouring Mbukushu, Kwangali, Gciriku, Kuhane and Fwe languages in Angola, Namibia, Botswana and Zambia. These sounds occur not only in borrowed vocabulary, but have spread to native Bantu words as well, in the case of Nguni at least partially due to a type of word taboo called hlonipha. Some creolised varieties of Afrikaans, such as Oorlams, retain clicks in Khoekhoe words.
Three languages in East Africa use clicks: Sandawe and Hadza of Tanzania, and Dahalo, an endangered South Cushitic language of Kenya that has clicks in only a few dozen words. It is thought the latter may remain from an episode of language shift.
The only non-African language known to have clicks as regular speech sounds is Damin, a ritual code used by speakers of Lardil in Australia. One of the clicks in Damin is actually an egressive click, using the tongue to compress the air in the mouth for an outward (egressive) "spurt".
For the most part, the Southern African Khoisan languages only use root-initial clicks.[note 2] Hadza, Sandawe and several Bantu languages also allow syllable-initial clicks within roots, but in no language does a click close a syllable or end a word. Once clicks are borrowed into a language as regular speech sounds, they may spread to native words, as has happened due to hlonipa word-taboo in the Nguni languages. In Gciriku, for example, the European loanword tomate (tomato) appears as cumáte with a click c, though it begins with a t in all neighbouring languages.
Scattered clicks are found in ideophones in other languages, such as Kongo //, Mijikenda // and Hadza // (Hadza does not otherwise have labial clicks). Ideophones often use phonemic distinctions not found in normal vocabulary.
English and many other languages may use bare clicks in interjections, without the accompaniment of vowels, such as the dental "tsk-tsk" sound used to express disapproval, or the lateral tchick used with horses. In Ningdu Chinese (a variety of Hakka), flapped nasal clicks are used in nursery rhymes. In Armenian, Bulgarian, Greek, Levantine Arabic, Maltese, Persian, Turkish, some time in French, as well as southern Italian languages such as Sicilian, a bare dental click accompanied by tipping the head upwards signifies "no". Libyan Arabic apparently has three such sounds.
Clicks occasionally turn up elsewhere, as in the special registers twins sometimes develop with each other. In West Africa, clicks have been reported allophonically, and similarly in German, faint clicks have been recorded in rapid speech where the consonants /t/ and /k/ overlap between words.
Occasionally other languages are said to have click sounds. This is usually a misnomer for ejective consonants, which are found across much of the world.
Like other consonants, clicks can be described using four parameters: place of articulation, manner of articulation, phonation (including glottalisation) and airstream mechanism. As noted above, clicks necessarily involve at least two closures, which in some cases operate partially independently: an anterior articulation traditionally represented by the special click symbol in the IPA--and a posterior articulation traditionally described as oral or nasal, voiced or voiceless, etc. The literature also describes a contrast between velar and uvular rear articulations for some languages.
However, recent work shows that in languages that make this distinction, all clicks have a uvular, or even pharyngeal, rear closure--and the clicks explicitly described as uvular are in fact clusters/contours of a click plus a pulmonic or ejective component, in which the cluster/contour has two release bursts, the forward (click) and then the rearward (uvular) component. "Velar" clicks in these languages have only a single release burst, that of the forward click release, and the release of the rear articulation isn't separately audible (Miller 2011).
Nonetheless, in most of the literature the stated place of the click is the anterior articulation (called the release or influx), whereas the manner is ascribed to the posterior articulation (called the accompaniment or efflux). The anterior articulation defines the click type and is written with the IPA letter for the click (dental ⟨?⟩, alveolar ⟨?⟩, etc.), whereas the traditional term 'accompaniment' conflates the categories of manner (nasal, affricated), phonation (voiced, aspirated, breathy voiced, glottalised), as well as any change in the airstream with the release of the posterior articulation (pulmonic, ejective), all of which are transcribed with additional letters or diacritics, as in the nasal alveolar click, ⟨⟩ or ⟨⟩ or--to take an extreme example--the voiced (uvular) ejective alveolar click, ⟨q'⟩.
The size of click inventories ranges from as few as three (in Sesotho) or four (in Dahalo), to dozens in the Kx'a and Tuu (Northern and Southern Khoisan) languages. Taa, the last vibrant language in the latter family, has 45 to 115 click phonemes, depending on analysis (clusters vs. contours), and over 70% of words in the dictionary of this language begin with a click.
Clicks appear more stop-like (sharp/abrupt) or affricate-like (noisy) depending on their place of articulation: In southern Africa, clicks involving an apical alveolar or laminal postalveolar closure are acoustically abrupt and sharp, like stops, whereas labial, dental and lateral clicks typically have longer and acoustically noisier releases that are superficially more like affricates. In East Africa, however, the alveolar clicks tend to be flapped, whereas the lateral clicks tend to be more sharp.
The five click releases with dedicated symbols in the International Phonetic Alphabet (IPA) are labial ?, dental ?, palato-alveolar or "palatal" ?, (post)alveolar or "retroflex" ? and lateral ?. In most languages, the retroflex and palatal releases are "abrupt"; that is, they are sharp popping sounds with little frication (turbulent airflow). The labial, dental and lateral releases, on the other hand, are typically "noisy": they are longer, lip- or tooth-sucking sounds with turbulent airflow, and are sometimes called affricates. (This applies to the forward articulation; both may also have either an affricate or non-affricate rear articulation as well.) The apical releases, ? and ?, are sometimes called "grave", because their pitch is dominated by low frequencies; whereas the laminal releases, ? and ?, are sometimes called "acute", because they are dominated by high frequencies. (At least in the N?ng language and Ju?'hoan, this is associated with a difference in the placement of the rear articulation: "grave" clicks are uvular, whereas "acute" clicks are pharyngeal.) Thus the alveolar click /?/ sounds something like a cork pulled from a bottle (a low-pitch pop), at least in Xhosa; whereas the dental click /?/ is like English tsk! tsk!, a high-pitched sucking on the incisors. The lateral clicks are pronounced by sucking on the molars of one or both sides. The labial click /?/ is different from what many people associate with a kiss: the lips are pressed more-or-less flat together, as they are for a [p] or an [m], not rounded as they are for a [w].
The most populous languages with clicks, Zulu and Xhosa, use the letters c, q, x, by themselves and in digraphs, to write click consonants. Most Khoisan languages, on the other hand (with the notable exceptions of Naro and Sandawe), use a more iconic system based on the pipe ⟨|⟩. (The exclamation point for the "retroflex" click was originally a pipe with a subscript dot, along the lines of ?, ?, ? used to transcribe the retroflex consonants of India.) There are also two main conventions for the second letter of the digraph as well: voicing may be written with g and uvular affrication with x, or voicing with d and affrication with g (a convention of Afrikaans). In two orthographies of Ju?'hoan, for example, // is written g! or dq, and // !x or qg. In languages without //, such as Zulu, // may be written gq.
|Doke (1926)||?||?||? ||?||?|
|Bantu||pc||c||v ç tc
There are a few less-well-attested articulations. A reported subapical retroflex release ⟨!!⟩ in Grootfontein !Kung[note 4] turns out to be alveolar with lateral release, ⟨??⟩; Ekoka !Kung has a fricated alveolar click with an s-like release, provisionally transcribed ⟨s⟩; and Hadza and Sandawe have a "slapped" alveolar click, provisionally transcribed ⟨?¡⟩ (in turn, the lateral clicks in Hadza and Sandawe are more abrupt and less noisy than in southern Africa). However, the Khoisan languages are poorly attested, and it is quite possible that, as they become better described, more click releases will be found.
Formerly when a click consonant was transcribed, two symbols were used, one for each articulation, and connected with a tie bar. This is because a click such as  was analysed as a nasal velar rear articulation [?] pronounced simultaneously with the forward ingressive release [?]. The symbols may be written in either order, depending on the analysis: ⟨⟩ or ⟨⟩. However, a tie bar was not often used in practice, and when the manner is tenuis (a simple [k]), it was often omitted as well. That is, ⟨?⟩ = ⟨k?⟩ = ⟨?k⟩ = ⟨k⟩ = ⟨k⟩. Regardless, elements that do not overlap with the release are always written according to their temporal order: Prenasalisation is always written first (⟨?⟩ = ⟨?⟩ = ⟨⟩), and the non-lingual part of a contour is always written second (⟨k'q'⟩ = ⟨k'q'⟩ = ⟨q'⟩).
However, it has become standard to analyse clicks as simplex segments, as research has shown that the front and rear articulations are not independent, and to use click symbols to cover the rear articulation as well, with diacritics rather than digraphs for the accompaniments. At first this tended to be ⟨?, , ⟩ for ⟨k, , ⟩, based on the belief that the rear articulation was velar; but as it has become clear that the rear articulation of both "velar" and "uvular" clicks is actually uvular or even pharyngeal, voicing and nasalisation diacritics more in keeping with the IPA have started to appear: ⟨?, , , ⟩ for ⟨?, , , ⟩.
|k ~ k
(k? ~ ?k)
|k ~ k?
(k ~ ?k?)
( ~ )
( ~ )
|q ~ q |
(q? ~ ?q)
In practical orthography, the voicing or nasalisation is sometimes given the anterior place of articulation: dc for and m? for , for example.
Kirshenbaum transcription uses a very different convention: clicks are denoted by ⟨!⟩ (always ⟨!⟩) added to the letter for the stop homorganic to the release, but with the manner of the accompaniment. For example, ⟨t!⟩ is a voiceless dental click, and ⟨m!⟩ is a nasal bilabial click. This convention is used in the literature on Damin, where the clicks are transcribed as ⟨m!, nh!, n!, rn!⟩.
Places of articulation are often called click types, releases, or influxes. There are seven or eight known releases, not counting slapped or egressive clicks. These are (bi)labial affricated ?, or "(bi)labial"; laminal denti-alveolar affricated ?, or "dental"; apical (post)alveolar plosive ?, or "alveolar"; laminal postalveolar (palato-alveolar) plosive ?, or "palatal"; laminal postalveolar (palato-alveolar) affricated (known only from Ekoka !Kung); subapical postalveolar (retroflex) (only known from Central !Kung and Damin); and apical postalveolar lateral ?.
Languages illustrating each of these articulations are listed below. Given the poor state of documentation of Khoisan languages, it is quite possible that additional releases will turn up. No language is known to contrast more than five places of articulation, though one publication has reconstructed Proto-Kx'a with six.
|dental ? only||Dahalo||Various nasal clicks only.|
|1 release, variable ? ~ ? ~ ?||Sotho, Gciriku, Mbukushu||In Sotho the clicks tend to be alveolar, in Gciriku and Mbukushu dental.|
|2 releases, ?, ?||Kwadi||? not found with all manners|
|2 releases, ?, ?||?Xegwi||? reacquired in loans|
|3 releases, ?, ?, ?||Sandawe, Hadza, Xhosa, Zulu||In Hadza and Sandawe, ? is often "slapped".|
|4 releases, ?, ?, ?, ?||Korana, Khoekhoe, Yeyi, Ju?'hoan|
|4 releases, ?, , ?, ?||Ekoka !Kung|
|5 releases, ?, ?, ?, ?, ?||?Hõã, N?u, ?Xam, Taa|
|5 releases, ?, ?, ?, , ?||Grootfontein !Kung|
|5 releases, ?, , ?, ?,||Damin||Aside from //, all are nasal.|
Extra-linguistically, Coatlán Zapotec of Mexico uses a linguolabial click, , as mimesis for a pig drinking water, and several languages, such as Wolof, use a velar click [?], long judged to be physically impossible, for backchanneling and to express approval. A sublingual click ("sucking-teeth") is found across West Africa, the Caribbean and into the United States.
The terms for the click releases were originally developed by Bleek in 1911. Since then there has been some conflicting variation. Here are the terms used in some of the main references.
|Click release||Bantu letters||Also known as|
|? dental||c||dental affricative/affricated/with friction; alveolar affricated; denti-alveolar; apico-lamino-dental; denti-pharyngeal|
|? palatal||ç (tc, qc)||palato-alveolar; alveolar; alveolar instantaneous; denti-alveolar implosive; palato-pharyngeal; dental|
|? alveolar||q||cerebral; (post-) alveolar implosive; palato-alveolar; palato-alveolar instantaneous; palatal; retroflex; palatal retroflex; apico-palatal; central alveo-uvular|
|? lateral||x||lateral affricative/with friction; alveolar lateral affricated; post-alveolar lateral; lateral apico-alveo-palatal; lateral alveo-uvular|
The dental, lateral and bilabial clicks are rarely confused. However, the palatal and alveolar clicks frequently have the opposite names in older literature, and they were not distinguished in the IPA until 1989. However, since Ladefoged & Traill (1984) clarified the places of articulation, the terms in the left column above have become standard.[note 5]
In several languages, including Nama and Ju?'hoan, the alveolar click types [?] and [?] only occur, or preferentially occur, before back vowels, whereas the dental and palatal clicks occur before any vowel. The effect is most noticeable with the high front vowel [i]. In Nama, for example, the diphthong [?i] is common but [i] is rare after alveolar clicks, whereas the opposite is true after dental and palatal clicks. This is a common effect of uvular or uvularised consonants on vowels in both click and non-click languages. In Taa, for example, the back-vowel constraint is triggered by both alveolar clicks and uvular stops, but not by palatal clicks or velar stops: sequences such as */?i/ and */qi/ are rare to non-existent, whereas sequences such as /?i/ and /ki/ are common. It is also triggered by labial clicks, though not by labial stops. Clicks subject to this constraint involve a sharp retraction of the tongue during release.
|ballistic tongue retraction
& back-vowel constraint
|no retraction, no constraint||?||?|
Miller and colleagues (2003) used ultrasound imaging to show that the rear articulation of the alveolar clicks ([?]) in Nama is substantially different from that of palatal and dental clicks. Specifically, the shape of the body of the tongue in palatal clicks is very similar to that of the vowel [i], and involves the same tongue muscles, so that sequences such as [?i] involved a simple and quick transition. The rear articulation of the alveolar clicks, however, is several centimetres further back, and involves a different set of muscles in the uvular region. The part of the tongue required to approach the palate for the vowel [i] is deeply retracted in [?], as it lies at the bottom of the air pocket used to create the vacuum required for click airstream. This makes the transition required for [?i] much more complex and the timing more difficult than the shallower and more forward tongue position of the palatal clicks. Consequently, [?i] takes 50 ms longer to pronounce than [?i], the same amount of time required to pronounce [i].
Languages do not all behave alike. In N?uu, the simple clicks /?, ?, ?/ trigger the [?i] and [æ] allophones of /i/ and /e/, whereas /?, ?/ do not. All of the affricated contour clicks, such as //, do as well, as do the uvular stops /q, ?/. However, the occlusive contour clicks pattern like the simple clicks, and /q/ does not trigger the back-vowel constraint. This is because they involve tongue-root raising rather than tongue-root retraction in the uvular-pharyngeal region. However, in G?wi, which is otherwise largely similar, both /q/ and // trigger the back-vowel constraint (Miller 2009).
Click manners are often called click accompaniments or effluxes, but both terms have met with objections on theoretical grounds.
There is a great variety of click manners, both simplex and complex, the latter variously analysed as consonant clusters or contours. With so few click languages, and so little study of them, it is also unclear to what extent clicks in different languages are equivalent. For example, the [?k?] of Khoekhoe, [?k? ~ k] of Sandawe and [? ~ k?] of Hadza may be essentially the same phone; no language distinguishes them, and the differences in transcription may have more to do with the approach of the linguist than with actual differences in the sounds. Such suspected allophones/allographs are listed on a common row in the table below.
Some Khoisan languages are typologically unusual in allowing mixed voicing in non-click consonant clusters/contours, such as dt?s'k?x', so it is not surprising that they would allow mixed voicing in clicks as well. This may be an effect of epiglottalised voiced consonants, because voicing is incompatible with epiglottalisation.
As do other consonants, clicks vary in phonation. Oral clicks are attested with four phonations: tenuis, aspirated, voiced and breathy voiced (murmured). Nasal clicks may also vary, with plain voiced, breathy voiced / murmured nasal, aspirated and unaspirated voiceless clicks attested (the last only in Taa). The aspirated nasal clicks are often said to have 'delayed aspiration'; there is nasal airflow throughout the click, which may become voiced between vowels, though the aspiration itself is voiceless. A few languages also have pre-glottalised nasal clicks, which have very brief prenasalisation but have not been phonetically analysed to the extent that other types of clicks have.
Clicks may be pronounced with a third place of articulation, glottal. A glottal stop is made during the hold of the click; the (necessarily voiceless) click is released, and then the glottal hold is released into the vowel. Glottalised clicks are very common, and they are generally nasalised as well. The nasalisation cannot be heard during the click release, as there is no pulmonic airflow, and generally not at all when the click occurs at the beginning of an utterance, but it has the effect of nasalising preceding vowels, to the extent that the glottalised clicks of Sandawe and Hadza are often described as prenasalised when in medial position. Two languages, G?wi and Yeyi, contrast plain and nasal glottalised clicks, but in languages without such a contrast, the glottalised click is nasal. Miller (2011) analyses the glottalisation as phonation, and so considers these to be simple clicks.
Various languages also have prenasalised clicks, which may be analysed as consonant sequences. Sotho, for example, allows a syllabic nasal before its three clicks, as in nnqane 'the other side' (prenasalised nasal) and seqhenqha 'hunk'.
There is ongoing discussion as to how the distinction between what were historically described as 'velar' and 'uvular' clicks is best described. The 'uvular' clicks are only found in some languages, and have an extended pronunciation that suggests that they are more complex than the simple ('velar') clicks, which are found in all. Nakagawa (1996) describes the extended clicks in G?wi as consonant clusters, sequences equivalent to English st or pl, whereas Miller (2011) analyses similar sounds in several languages as click-non-click contours, where a click transitions into a pulmonic or ejective articulation within a single segment, analogous to how English ch and j transition from occlusive to fricative but still behave as unitary sounds. With ejective clicks, for example, Miller finds that although the ejective release follows the click release, it is the rear closure of the click that is ejective, not an independently articulated consonant. That is, in a simple click, the release of the rear articulation is not audible, whereas in a contour click, the rear (uvular) articulation is audibly released after the front (click) articulation, resulting in a double release.
These contour clicks may be linguo-pulmonic, that is, they may transition from a click (lingual) articulation to a normal pulmonic consonant like [q] (e.g. [q]); or linguo-glottalic and transition from lingual to an ejective consonant like [q'] (e.g. [q']): that is, a sequence of ingressive (lingual) release + egressive (pulmonic or glottalic) release. In some cases there is a shift in place of articulation as well, and instead of a uvular release, the uvular click transitions to a velar or epigottal release (depending on the description, [kx'] or ). Although homorganic ['] does not contrast with heterorganic [kx'] in any known language, they are phonetically quite distinct (Miller 2011).
Apart from Dahalo, Damin and many of the Bantu languages (Yeyi and Xhosa being exceptions), 'click' languages have glottalised clicks. Contour clicks are restricted to southern Africa, but are very common there: they are found in all members of the Tuu, Kx'a and Khoe families, as well as in the Bantu language Yeyi.
In a comparative study of clicks across various languages, using her own field work as well as phonetic descriptions and data by other field researchers, Miller (2011) posits 21 types of clicks that contrast in manner or airstream.[note 6] The homorganic and heterorganic affricated ejective clicks do not contrast in any known language, but are judged dissimilar enough to keep separate. Miller's conclusions differ from those of the primary researcher of a language; see the individual languages for details.
(Zulu is like Xhosa apart from not having //)
Each language below is illustrated with alveolar clicks, apart from Dahalo, which only has dental. Under each language are the orthography (in italics, with old forms in parentheses), the researchers' transcription (in ⟨angle brackets⟩), or allophonic variation (in [brackets]). Some languages also have labialised or prenasalised clicks.
|Manner||?, ?, ?, ?, ?||?, ?, ?, ?||?, ?, ?||?||?, ?, ?||?, ?, ?, ?||?, , ?, ?|
|Tenuis||/?/||⟨!⟩*||⟨?⟩||[?]||! (q)||!g||⟨k!⟩||q||q (!)||q||⟨!⟩|
[ ~ ]
|Aspirated||//||⟨!h⟩*||⟨⟩||||!h (qh)||!kh||⟨k!h⟩||qh||qh (!h)||qh||⟨!h⟩ (= !x ?)|
[ ~ ]
|⟨⟩||||n! (nq)||!n||⟨?!⟩||nq||nq (n!)||//||nq||⟨?!⟩||⟨rn!⟩|
(prenasalised between vowels)
|⟨?⟩||[? ~ ]||!'h (q'h)||!h||⟨?!h⟩|
|Breathy-voiced||//||⟨n!hh⟩||n!h (nqh)||ngq[note 9]|
|Preglottalised nasal click||//||⟨'n?⟩*||||(in Ekoka)|
|Nasal (silent initially,
prenasalised after vowels)
|/?/||⟨?"⟩||⟨?⟩||[ ~ ]||!' (q')
(w/ nasal vowels)
[ ~ ?]
(!' ~ n!')
|Nasal (prenasalised initially)||//||⟨n?"⟩|
|Voiced (and prenasalised)||/?/||⟨g!q⟩
[ ~ ]
|[note 10]||()[note 11]||⟨?!⟩
|Voiceless fricative||//||⟨!x⟩||⟨⟩||[?q]||!x (qg)||⟨q!?⟩||⟨!x⟩ (?)|
|Voiced fricative (prenasalised)||/?/||⟨g!x⟩
[? ~ ]
|Voiced ejective stop||/q'/||⟨g?q'⟩|
|Heterorganic affricate /
affricate / epiglottalised
|Egressive[note 12]||(Voiceless "spurt"; labial only)||//||⟨p'⟩|
Yeyi also has prenasalised //. The original researchers believe that  and  are allophones.
A DoBeS (2008) study of the Western !Xoo dialect of Taa found several new manners: creaky voiced (the voiced equivalent of glottalised oral), breathy-voiced nasal, prenasalised ?lottalised (the voiced equivalent of glottalised) and a (pre)voiced ejective. These extra voiced clicks reflect Western !Xoo morphology, where many nouns form their plural by voicing their initial consonant. DoBeS analyses most Taa clicks as clusters, leaving nine basic manners (marked with asterisks in the table). This comes close to Miller's distinction between simple and contour clicks, shaded light and medium grey in the table.
Clicks are often portrayed[weasel words] as a primordial feature of human language, a romantic reflection of the primordial lifestyle imagined of the speakers of Khoisan languages. One genetic study concluded that clicks, which occur in the languages of the genetically divergent populations Hadza and !Kung, may be an ancient element of human language. However, this conclusion relies on several dubious assumptions (see Hadza language), and most linguists assume that clicks, being quite complex consonants, arose relatively late in human history. How they arose is not known, but it is generally assumed that they developed from sequences of non-click consonants, as they are found allophonically for doubly articulated consonants in West Africa (Ladefoged 1968), where /tk/ sequences overlap at word boundaries in German, and for the sequence /mw/ in Ndau and Tonga.[note 13] Such developments have also been posited in historical reconstruction. For example, the Sandawe word for 'horn', /t?ana/, with a lateral affricate, may be a cognate with the root /a:/ found throughout the Khoe family, which has a lateral click. This and other words suggests that at least some Khoe clicks may have formed from consonant clusters when the first vowel of a word was lost; in this instance *[t?ana] > *[t?na] > [a] ~ [a].
On the other side of the equation, several non-endangered languages in vigorous use demonstrate click loss. For example, the East Kalahari languages have lost clicks from a large percentage of their vocabulary, presumably due to Bantu influence. As a rule, a click is replaced by a consonant with close to the manner of articulation of the click and the place of articulation of the forward release: alveolar click releases (the [?] family) tend to mutate into a velar stop or affricate, such as [k], [?], [?], [k?x]; palatal clicks ([?] etc.) tend to mutate into a palatal stop such as [c], [ ?], [ ?], [c'], or a post-alveolar affricate [t?], [d?]; and dental clicks ([?] etc.) tend to mutate into an alveolar affricate [ts].
Clicks are often presented as difficult sounds to articulate within words. However, children acquire them readily; a two-year-old, for example, may be able to pronounce a word with a lateral click [?] with no problem, but still be unable to pronounce [s].Lucy Lloyd reported that after long contact with the Khoi and San, it was difficult for her to refrain from using clicks when speaking English.