Melissa Wright, a linguist at Birmingham City University, specializes in “the relationship between the phonetic-linguistic details of everyday talk and the interactional structures within which (and through which) that talk is produced”; at her home page, she says of her PD thesis: “I examined the phonetic and interactional organisation of naturally-occurring British and American English conversation. I showed that there are complex and systematic mappings between clicks and interactional structures in talk, a finding which is striking given that clicks have so far been regarded by linguists as functioning only paralinguistically.” Her recent paper “On clicks in English talk-in-interaction” (Journal of the International Phonetic Association 41: 207-229; abstract) has been briefly described in this Scientific American article by Anne Pycha as follows:
Speakers, it turns out, use clicks for a previously overlooked purpose: as a form of verbal punctuation in between thoughts or phrases. Melissa Wright […] recently analyzed click sounds in six large sets of recorded English conversations. She found that speakers used clicks frequently to signal that they were ending one stretch of conversation and shifting to a new one. For example, a speaker might say, “Yeah, that was a great game,” produce a click, then say, “The reason I’m calling is to invite you to dinner tomorrow.”
This pattern, which occurred for both British and American speakers, suggests that clicks have a meaning similar to saying “anyway” or “so.” That is, clicks provide us with a phonetic resource to organize conversations and communicate our intentions to listeners. This finding had previously eluded linguists, whose research often focuses on words and sentences in isolation. Wright was able to uncover the new pattern because she analyzed clicks in the context of complete conversations, suggesting that this method could be important for making new discoveries about the nature of language.
I don’t know if the “new discoveries about the nature of language” thing is from Wright’s paper or added by Pycha to spice things up; it raised my eyebrows, but not as much as the succeeding paragraph about, yes, the origin of language, for which see this exasperated 2003 LH post. (Thanks, Paul!)
In Hebrew, the clicks may mean, additionally, a sign of impatience, or even slight exasperation with the opponent.
“Had previously eluded linguists”?
Is “shifting to a new stretch of conversation” so different from “beginning point of a conversational unit”?
I know that “origin of language” trips a lot of reflexes, including mine. Nevertheless, as it is now uncontroversial that Homo sapiens arose in Africa, the “click cline” should also be fairly uncontroversial.
[…] as it is now uncontroversial that Homo sapiens arose in Africa, the “click cline” should also be fairly uncontroversial.
I probably misunderstand you, but where’s the “click cline”? What I see is a “click cliff”. There are languages with phonemic clicks in a part of Africa, languages with ejectives scattered around the world, and a lot of languages everywhere with neither. Without a real gradation repeated across the range of linguistic features, how can we discern relics from local innovations? Correlation is not causation and all that.
speakers used clicks frequently to signal that they were ending one stretch of conversation and shifting to a new one
The click naturally comes as you’re opening your mouth for an intake of breath with which to fuel the new stretch. It gets stylised or exaggerated for the purpose of announcing the change in subject.
I wonder if it falls in the same linguistic category as the short loud intake of breath that some German & Scandinavian women (but hardly ever men) use, sometimes habitually, instead of saying “yes”.
Some people snort to signal that they are ending a stretch of conversation. The snort means: “that about wraps it up, then, I’m right and you’re wrong”. My father did that often, as I once reported here. It may have been in the same comment thread that Crown introduced his female Germanico-Scandinavian correspondents.
Damn, I meant “corespondents”.
I mentioned to LH that I was not aware of clicks in English – apart from the “tut,tut” sense. I certainly have never heard it from a CNN newsreader. I know I am rather deaf but am I really missing a very widespread phenomenon ?
It’s really the high frequencies that one loses, Paul, so it’s probably the deafness.
I’m not aware of any co-respondents, G. I thought that sort of thing (leaving a pair of brown suede brogues outside a hotel-room door etc.) went out in the 1950s.
Nevertheless, as it is now uncontroversial that Homo sapiens arose in Africa, the “click cline” should also be fairly uncontroversial.
The leap of logic here is breathtaking. Just because mankind arose in Africa doesn’t mean any damn idea that involves African languages must be correct. See here for my thoughts on the subject: “Why then it follows as the night the day that the clicks are inherited from our earliest ancestors, if you ignore inconvenient facts like the inevitability of language change, the irrelevance of genetics to linguistics, and the propensity of language communities to borrow sounds from each other (the Bantu languages of the region, for instance, have borrowed clicks from the languages that were there when they arrived).”
That Hebrew use is interesting. There’s a more generally Jewish click (you find it a lot in Yiddish, and also in the English of Orthodox Jews) that means something like: “What follows is going to be a complication or subclause of what we were talking about.” At least that’s how I understand it. Sometimes it’s accompanied by gesturing up and down with the thumb, which marks internal divisions in what’s being said. Also sometimes a sing-song tone that involves dropping the first syllable down about a fourth and then going back up a fourth on the stressed syllables of key words; this is called “shteyger” (melody). All this comes in handy in discussing complicated things, but it is disturbing to people who haven’t seen it before; the combines effect is sort of like clicking, chanting robot with a malfunctioning hand motor.
I took JC’s invocation of the ‘click cline’ to be purposefully humorous.
I took JC’s invocation of the ‘click cline’ to be purposefully humorous.
Goddammit, did another joke whiz by me? What a drag it is getting old.
What a drag it is getting old.
Wikipedia:
In other words, he was 22 when he wrote that.
Well, Mick wasn’t talking about himself, right? He was putting down the pill-popping old lady (probably at least 30 . . .) he was singing about. See also the rather nastily-edged Jam song “The Butterfly Collector” about a legendary groupie inevitably losing her sex appeal as she aged. Paul Weller wrote that he he was 20 and while there are I believe various theories as to which real-life habitue of the London music scene it was based on, one candidate I found via some quick googling would have been a near-decrepit 24 or 25 at the time.
Well, “what a drag it is getting old” is the classic pubertarian whine, so no surprise that Jagger was 22. It’s not a deep insight, but the song is great.
Ageing has never creased my brow. When I was 16 I thought I would never reach 21, which symbolized for me formal freedom from restraints that I never submitted to anyway.
If I remember correctly, I obtained legal majority (is that what it’s called ?) when I was 18 by some lawyering hook or crook. It’s all been so long ago. What a drag it was being young !
Oh fiddledesticks ! JW’s comment reminds me that Jagger was singing about pill-poppin’ mommas, not himself.
Likewise, Crown’s North Germanic she-interlocutors could have picked up the habit of sharp inbreaths of assent from anyone, maybe a non-native toad or snail.
A blind friend told me that she can tell if people are glad to see her because, when they smile broadly, the slight displacement of saliva in the mouth and the stretch of the lips produces a click.
She’s right. Go ahead and try it!
No, I wasn’t joking. I was, however, overly elliptical. The fact is that no phonologist has ever been able to come up with a mechanism whereby consonants with other airstream mechanisms can evolve into clicks in any reasonably natural way. (Ejectives and implosives are another matter, being essentially exaggerations of very late and very early voice onset times respectively.) So as far as anyone knows, if languages have phonemic clicks, it is because their ancestors had phonemic clicks too. (Yes, Damin’s an exception, but it’s an exception to a lot of things; it’s probably a sort of conlang too.)
And when I referred to a cline, I meant in all uses of clicks. As map 142A shows, the “propositional” uses of clicks (for things like agreement and disagreement) are to be found in Africa, Southern Europe, and Southern India. Affective uses are found in the rest of Africa and Europe and in Southeast Asia including the Pacific. The other continents are essentially devoid of them, as far as the evidence goes.
@John Cowan: I’m no connoisseur of the whole debate, but couldn’t clicks have appeared in expressive sounds, then phonologized and make their way into expressive words and then normal words? (I have the impression that a few high-frequency verbs in English are of expressive or onomapoetic origin so the latter part should not be problematic)
Yes, I realize clicks do not appear to evolve in any reasonably natural way, but “So as far as anyone knows” is a remarkably weak reed on which to base a sweeping and entirely hypothetical theory. I think minus273’s suggestion is entirely reasonable. One should never underestimate the capacity of languages to do unexpected things (something Chomsky has never gotten through his skull).
John Cowan: I left some comments on the May 18 2009 thread which I suspect you would find of interest. Basically, I asked the following outrageous question: could clicks as phonemes only appear in conlangs such as Damin? If so, does this not imply that their presence in Southern Africa must be due to a conlang, from whence the clicks either spread to ancestral forms of present-day click languages, or to said hypothetical conlang nativizing, i.e. turning into a natural language (Proto-Khoisan, perhaps)?
David Marjanovic claimed that phonemic clicks must arise out of consonant clusters: why consonant clusters never yield clicks outside of Southern Africa was not discussed, so that I think my outrageous question is worth asking.
Minus273: I have the same reservation about the possibility you propose (that clicks might have spread from the expressive/onomatopeic subset of the language to the general vocabulary): quite apart from the fact that I have never heard of such a process involving any phoneme, the matter of its distribution would still need to be faced: why would expressive clicks have become phonemes in Southern Africa only?
Some people snort to signal that they are ending a stretch of conversation.
And most everyone, apparently, raises their eyebrows for an eighth of a second before addressing someone else. Or was that a sixth? Anyway, that tidbit comes from one of David Crystal’s books. Steady eyebrows are often perceived as menacing — not just in Westerns.
Basically, I asked the following outrageous question: could clicks as phonemes only appear in conlangs such as Damin? If so, does this not imply that their presence in Southern Africa must be due to a conlang, from whence the clicks either spread to ancestral forms of present-day click languages, or to said hypothetical conlang nativizing, i.e. turning into a natural language (Proto-Khoisan, perhaps)?
Intriguing suggestion, which could lead to a highly speculative scenario. Perhaps created as a ritual language, deliberately intended to sound bizarre and be difficult to learn, to guard it from non-initiates. How, then, could such a thing become an everyday language? Co-opted as a power grab against high-status ritual initiates?
Aren’t traditional Khoisan communities famously flexible, with small groups reshuffled every other year? If this was the situation in a multi-lingual environment, I could easily see a neutral ceremonial language gaining ground or providing loans into the various member languages of the macro-community.
@John Cowan: The thing is, even if we accept that the presence of clicks in Khoisan languages implies the presence of clicks in Proto-World, and even if we accept that Proto-World was spoken in a place where Khoisan languages are spoken today, we still have a big ??? in explaining why the maintenance of clicks would be so strongly correlated with proximity to the World Urheimat. It’s not even a “theory”, because it has no explanatory power.
Ran: Your second assumption is not necessary. The Bantu languages migrated from the north, and probably either displaced Khoi-San language speakers, or at least replaced their languages.
And if clicks can be lost but not gained (except through borrowings), then we would indeed expect them to slowly fade out with distance from Africa.
And if clicks can be lost but not gained (except through borrowings), then we would indeed expect them to slowly fade out with distance from Africa.
No, that logic doesn’t work, unless Africa is a continually replenishing source of clicks.
@John Cowan
” “propositional” uses of clicks (for things like agreement and disagreement) are to be found in Africa, Southern Europe, and Southern India.”
I have to agree. As a North Indian, I find the abundance of clicks perplexing sometimes, and to me, it comes across as even rude. I would usually interpret it as frustration or exasperation, though the other person might only mean it to imply mild disagreement. This confusion can be quite pronounced at times.
@John Cowan: “And if clicks can be lost but not gained (except through borrowings), then we would indeed expect them to slowly fade out with distance from Africa”: That’s not how that works. Even if it were true, overall, that languages tend to be more conservative the closer you are to their Urheimat — which I doubt very much — it’s obvious that there are plenty of counterexamples. In general, when people migrate, they take their whole language with them, and have no particular reason to start slowly dropping clicks and other phonemes in direct proportion to how far they’ve traveled. (Even in the modern world, where migratory peoples generally encounter other peoples and experience language contact — which the original emigrants from Africa would not have — there’s no all-encompassing force against linguistic conservatism. Consider Romanian, which has been heavily influenced by its Slavic neighbors, but still has some elements of Latin grammar that other descendants of Latin have lost. No clicks, admittedly, but still!)
Ran: What I’m assuming (which may be wrong) is something like allopatric speciation: that it’s the group that leaves the homeland that undergoes the most change. I know this is not true for dialects, which tend change fastest in the mother country where the capital is, but it may be true on larger scales.
So it wouldn’t be about distance in kilometers as such, but distance in branching points.
@John Cowan: Allopatric speciation doesn’t necessarily cause the homeland-leaving group to change more; and when it does, it’s because of (1) founder effects and (2) adaptation to a new environment. It’s hard to see how either of those would apply to the loss of click phonemes, unless either (1) population bottlenecks cause phoneme loss or (2) clicks are somehow maladaptive outside of southern Africa (and increasingly so the further away you get). Both of these seem clearly implausible to me, but the former (as you know) made it into Science, and the latter (as you’ve probably seen) has been argued as well, so I guess your mileage may vary?
But you omit (3) random drift. And once click phonemes drift away, there is no way to get them back. The Science article was based on (1), which I agree makes no sense for languages.
In general, as was mentioned above, change occurs more rapidly in the centre of a linguistic area (especially if many people congregate in it from different points) than in the periphery (eg in large, culturally attractive cities). If a group emigrates, two things can happen: if the group remains relatively homogeneous and maintains its own language, it often maintains older features which may have disappeared in the original centre (eg Canadian French vs “Parisian” French); but if the group finds itself in a thickly inhabited region where it dominates another group which adopts the newcomers’ language (even without abandoning its own), it is likely that the resulting new speech community will modify the incoming language through interference of the dominated language (eg Afrikaans vs Dutch). If two linguistic groups that come into such contact are more or less equal in political and social status, the two languages can coexist, each adopting and adapting features of the other one, so that they end up converging (the well-studied case of Kupwar [I think[ in India). These are more or less extreme, clear-cut cases, but there are many instances where several factors are involved to complicate the analysis.
About clicks: just because human language may have originated in Africa does not mean that a single group has preserved features of the oldest human language. The difference between clicks in Khoi-San and in other languages of the world is that in KS they are used just like any other sounds, in order to form words by linear association with other sounds, while in most other language groups they are uttered singly and their communicative significance is tied to the context of the situation. Clicks (often written in English “tsk tsk!” are not alone in this second group: saying “Mmmm” or “Ahem” (imperfect orthographic renderings of sounds) is not considered to be actual language, and such utterances are not listed in an English dictionary, for instance.
As to the number of phonemes: the theory discussed here seems to assume that the original language had a large number of phonemes which got reduced with time and distance. Even if there is some slight statistical support for this assertion (which is based on averages, not on individual languages), it seems totally unrealistic. There is nothing to suggest that the original language must have had a large number of phonemes (sounds functioning as distinct units), which got whittled down as humanity spread. I had never considered the problem, but it would seem more probable that it had a very limited number of phonemes, which expanded over time! Without trying to speculate about the remotest past, the study of sound changes in languages with a long recorded history includes both merging and splitting of phonemes : two (or more) phonemes may get closer to each other and merge (eg the varied pronunciations of the first vowel in Mary-merry-marry in English, which can involve three [as I learned], two or just one phoneme) but one phoneme may develop variants (allophones) under certain conditions, which evolve into separate phonemes if conditions change, such as when the language adopts words from a different language. An example of the latter is English /v/ where the sound [v] was originally an allophone of the phoneme /f/ between vowels, but the borrowing of French words where /v/ and /f/ were distinct led to the phonemicization of the contrast [v/f] in English. Borrowing is only one of possible factors in the splitting of phonemes, which can result from other conditions affecting a single language or dialect.
A much more serious objection to this “out of Africa” phonemic decay theory is that nothing like this has been observed with the phonological history of actual language families: for example, the Romance varieties spoken around Rome today are not phonologically more complex than those spoken in (geographically) peripheral areas such as Northern France, Moldavia or Portugal. Since we actually know what the phoneme inventory of Latin was, and since we know the phonological history of various Romance languages, I would expect this “out of center” phonemic decay effect to be clearly visible in the case of a shallow, well-studied language family such as Romance.
The fact that no such effect seems *ever* to have operated in Romance (to the best of my knowledge there is nothing to indicate that the area around Rome ever preserved inherited phonemes that were lost elsewhere) seems to indicate that said effect does not exist.
I wouldn’t want hatters to think (on the basis of my previous comment) that I’m a “stick-in-the-mud” type of linguist who rejects new ideas with a look of horror. To prove this, I’d like to give some further (circumstantial) evidence in support of the hunch I wrote about above, to wit, that clicks and other odd features found in various languages might be best explained by assuming that some artificial/ceremonial languages, which had said odd features (perhaps by design), subsequently became a community’s L1.
1-Among speakers of the various Yanomaman languages (these mutually unintelligible languages are spoken in the Amazon: Venezuelan/Brazilian border area) there exists a single, common ceremonial language.
2-Let us imagine some catastrophe causing large numbers of Yanomaman speakers to flee, mix and settle and found new communities. Among these new mixed communities, is it not possible that the ceremonial language would become the new mixed communities’ daily language, becoming the L1 of the first generation born and raised there?
Speculative, say you? I would agree, except that, reading Dixon and Aikhenvald’s THE AMAZONIAN LANGUAGES (Light bedtime reading, you understand: the chapter on “Other small language families” is where I obtained the above information on Yanomaman languages and the common ceremonial language), I learned that:
A-Dixon was originally drawn to the study of Amazonian languages because, as a typologist, he found that the Amazon was where most counter-examples to his typological generalizations were found, and
B-As a result of epidemics and population displacement over the course of European colonization, it was very frequent in the Amazon for various tribes to merge with one another.
Now, call me crazy if you like. But if my hypothetical scenario involving Yanomaman speakers is possible, if in fact it happened with other groups in the Amazon, couldn’t B explain A?
m-l, Etienne: Let me distinguo as clearly as I can. I do not believe the “steadily decreasing phoneme count” theory put forth in the Science article; indeed, I consider it absurd. Because there are splits as well as mergers, I’d expect the average number of phonemes to remain the same throughout human existence, though also with a wide range, as between Hawaiian and Abkhaz.
What I am proposing, simply (and stripped of any question of center vs. periphery for the moment), is that once phonemic clicks are lost by merger or loss, they can’t return by any known phonological process (though they can do so by lexical borrowing from languages which still have them), and therefore once they happen to be absent (as a result of sound-change) from a particular area, they cannot appear again in areas settled from there. Indeed, Khoi-San territory is known to be a relic area: until the migration of Bantu-speakers from the north, people who spoke related languages probably dominated (thinly) the whole of Southern Africa. I also don’t think it’s a coincidence that the expressive use of clicks very roughly follow historical out-of-Africa settlement patterns.
m-l: I think it’s hard to say whether the second or the third of your patterns of change would be dominant among small bands of hunters and gatherers; certainly the first would tend to appear only once there were large cities.
The OED2 actually does have entries for tsk, an alveolar click expressing any of “commiseration, disapproval, or irritation”; tck, a palatal click expressing “surprise or vexation”; and ach/ag, ah, aha, ahem, arf, argh, atishoo, augh/auh, aw, to mention only the members of English’s large stock of interjections with irregular phonology that happen to begin with a. Comprehensive dictionaries should always give these words, as they are quite language-specific.
Etienne: Fear not, we would never mistake you for a stick-in-the-mud linguist!
JC, I was not criticizing you, just listing some generalities about language change as documented in many known languages, especially modern ones. Obviously one cannot infer that there were exact parallels in the remote past of humanity when conditions were so different, but given that it took many thousands of years for humans to spread from Africa to Asia, Australia and America, the last few of those spreads, involving thoroughly modern humans, may be comparable to what is known to have happened in the historical past and what is reflected in the current geographical distribution of human languages. Small bands of hunter-gatherers are now marginal in the modern world, but they were the prevalent type of society in Siberia and Canada (among others) just a few generations ago.
And you are right about those interjections being in dictionaries: as a sample, I looked up “pfft” and “pouah” in the Petit Robert and sure enough, they are there. But they don’t have a defined, translatable meaning like ordinary words do.
Etienne, interesting hypothesis. You could be right. And no, you don’t sound like a “stick-in-the-mud linguist” (we both know some of those, and they are not fond of hypotheses)!
> But you omit (3) random drift.
I omit it because it’s wrong: random drift does not cause the homeland-leaving group to change more than the staying-in-homeland group. It affects both groups equally, except insofar as it interacts with founder effects (since the homeland-leaving group starts off with a population bottleneck), which I addressed.
Ran: Right, but my revised hypothesis no longer requires that, simply that there is a mechanism for clicks to be lost and no mechanism for them to be recreated.
John Cowan: there is a known mechanism for clicks to be (re)created, namely through deliberate creation (as in the initiation language Damin). I presume you meant “no mechanism in the context of normal language change”.
However, I’d like to examine this from another direction: is there any evidence that clicks can be lost, either? To the best of my knowledge there is no known instance of a language with phonemic clicks which subsequently lost them. And while some Bantu languages have acquired clicks through contact with Khoisan languages, I do not believe any Khoisan language has lost its clicks through contact with Bantu.
If clicks as phonemes cannot be lost, this would mean of course that phonemic clicks can only have arisen in Southern Africa (through language creation/engineering, perhaps) long after the rest of the world had been colonized from Africa.
I wouldn’t pay any attention to the distribution of the expressive use of clicks in this context: to repeat a point I made earlier to Minus273, such expressive clicks cannot have given birth to phonemic clicks.
There is, indeed, no evidence of click loss: how could there be, with only a few click languages, none recorded in writing until the 19th century?
I agree that expressive clicks couldn’t get ut I do think the expressive use might be relevant. Among North American victims of the cot-caught merger, /ɑ/ and /ɔ/ are mostly in free variation, but the expressive words “Ahh!” (surprise, satisfaction) and “Aww!” (sympathy, mock-sympathy) remain distinct, and can be used to teach the difference. So clicks might be lost from all ordinary words while remaining in expressive words, first in “logical” expressive words meaning ‘yes’, ‘no’, etc., and then only in affective words.
The human vocal tract is capable of producing a wide variety of sounds, only a subset of which are used in individual languages. A few of them are close to universal (eg [n]), some of them highly restricted in their occurrence (eg clicks). In many cases people are capable of producing individual sounds that they have great difficulty producing in languages not their own. For instance, francophones speaking English have trouble with the interdentals (th sounds), which come out either as [s,z] (in France) or [t,d] (in Canada). Yet the same people have no trouble imitating children who cannot pronounce the French /s, z/ properly and use interdentals instead (most of them eventually outgrow them). When I was first exposed to instruction in English, our class burst out laughing when the prof demonstrated the th sounds to us – it seemed he wanted us to talk baby talk! I have since read about similar instances with other languages – people can indeed produce certain sounds which they would not use in normal speech because those sounds are – rightly or wrongly – associated with particular types of speech which they would not ordinarly use themselves.
JC’s example of “Ahh!” and “Aww!” is quite interesting in this respect.
The Wikipedia article on clicks gives a lot of good, detailed information. Among other things, it has a table (and audio links) comparing clicks with other types of consonants with complex articulation, including fricative and affricate ejectives (sounds pronounced with constriction of the vocal cords).
Although affricates (such as English (t)ch in witch or cheese) usually function as single phonemes in the languages that have them, historically speaking they usually derive from the evolution of stops (eg cheese ultimately from a borrowing of Latin caesum into the Germanic ancestor; German Käse, Dutch kaas have kept the initial [k] stop sound represented by the Latin letter c). Affricates can also arise from contact between two consonants which are difficult to keep apart in rapid speech, so that the resulting cluster evolves into an affricate. An English example is some people’s pronunciation of train almost as “chwain”.
Like clicks, some ejective affricates can sound like very loud “pops”. I remember hearing such a sound in my first experience with fieldwork: I had no idea what I was hearing, but it was quite a bit louder than the rest of the words I had been learning. It turned out that it was the glottalized lateral affricate (often written tl’ when not using specialized phonetic symbols).
It seems quite possible to me that the Khoi-San clicks, found not just in interjections but in regular words, are not leftovers from the earliest human language but products of a very long in-situ evolution of simpler sounds in populations which have remained isolated for a long time. The fact that such sounds are used in isolation for “paralinguistic” expressive uses in other languages is not really relevant to the “linguistic” evolution within the Khoi-San group.
Oops! My mistake(s): Latin for “cheese” is not caesum but CASEUS. I thought the word I wrote looked funny.
John Cowan: some click languages belong to language families, in which case ALL members of said language family will have clicks. Which seems to indicate that clicks, if they’re not permanent, seem not to be lost very easily. This is quite unlike other types of phonemes: plenty of languages with tones are genetically related to languages without tones, for example: the same is true of nasal vowels, front rounded vowels, phonemic length…
Marie-Lucie: I now think you’re right: clicks (as phonemes!) aren’t that old. And I *think* I’ve figured out why they only arose in Southern Africa (insert dramatic clash of cymbals here, followed by mad scientist-type “evil laugh”):
I don’t think ejectives are the best thing to compare clicks to: implosive consonants are closer to clicks, inasmuch as both (unlike other consonants) involve air flowing into the mouth, rather than out of the mouth. So: let’s assume clicks are an outcome of consonant clusters involving implosives. To repeat an earlier question of mine: why then should they have arisen in Southern Africa only?
Well, WALS chapter 7 informs us that there are two “hotbeds” of implosives on the planet: West Africa and Southeast Asia (Vietnamese is an example of a Southeast Asian language with implosives). Now, Southeast Asian languages are phonotactically simple, with few if any consonant clusters. That leaves West Africa and neighboring parts of the continent as the main “hotbed” of implosives + phonotactical complexity, i.e. consonant clusters involving implosives. My guess is that originally Southern Africa was likewise a hotbed of consonant clusters and implosives, eventually leading to the appearance of clicks. Perhaps such a diachronic change is sufficently unlikely/rare that it could only have arisen where the preconditions (clusters involving implosives) were the rule and not the exception in the languages of the area.
It’s true that all the Khoi-San languages have clicks (whether you consider the two northern outliers to be members or not), but the number of clicks varies greatly. From what I understand, the chief source of implosives is over-voicing, or perhaps I should say premature voice onset (ejectives, by contrast, are the result of highly delayed voice onset).
Still, the idea of mixed implosive and explosive consonant clusters is interesting. Damned if I can articulate them, though.
My point is that clicks and ejective affricates are comparable in complexity of articulation, which suggests that they result from a long evolution rather than being very ancient in themselves. In addition to being complex, at least some of these articulations produce sounds which are strikingly louder than most other speech sounds and which can therefore be used paralinguistically for expressive purposes in languages which do not have them in their phonological (linguistic) inventory.
It seems that implosives and ejectives are alternate, probably areal versions of how to manipulate voice onset. They can arise spontaneously with single consonants, but affrication must result from further evolution, usually through contact of consonants with certain vowels or with other consonants. Simplification of affricates is a common process, and simplification of complex consonants through borrowing is also common, so that it is not surprising to find complex affricates in relatively isolated populations rather than across wide expanses which favour language contact and convergence. But there is no reason to conclude that such complex sounds, whether implosive or ejective, have always been part of the inventory of the languages in question.
I don’t mean to suggest that all instances of clicks must derive from a long evolution: obviously, tsk tsk or pfft or similar sounds used paralinguistically for expressive purposes, communication with domesticated animals, and also simply for fun, can probably arise spontaneously in any culture. But in their linguistic use (as in Khoi San) I think that it is much more likely that they arise from long evolution.
John Cowan: actually, I was more thinking along the lines of an implosive stop + liquid or nasal consonant cluster. Imagine a cluster consisting of, say, implosive /p/ followed by an /l/, the whole in syllable-initial position. It’s not too difficult to pronounce: however, because of the quick succession of inflow and outflow of air it would be natural for this to become a wholly implosive cluster /pl/; delete the initial /p/, with a transfer of its plosive quality (delayed onset) to the /l/, and what you are then left with is close enough to a lateral click for me.
Marie-Lucie: I quite agree, and indeed I think I have a reference somewhere on the relative articulatory complexity of different phonemes, with several of the most complex ones being phonemic clicks found in some Khoisan languages…
Etienne: I can’t really swallow that (or rather, I cough when I try!) First of all, voiceless implosives are rare and doubtful; most implosives are voiced. Second, clicks have velar closure, whereas implosives have glottal closure.
OT but I found the following page after reading some older LH threads. John Hawks is trying to claim that language is a case of instantiation of a “generic” rich information processing ability of the brain, and that there shouldn’t be much genetic basis for language in Neanderthal-to-human comparisons.