Is it possible to identify click sounds like [‖ ʘ !] by formant transitions in the surrounding vowels? I know stops and fricatives have that feature. I'm just wondering how the five (main) click sounds differ from each other acoustically.
6
-
2For the record, there are a lot more than five click sounds! There are five (maybe six) basic places of articulation, but just as with non-clicks, you can modify them in all kinds of ways - indeed, they're rather more likely to be modified somehow. isiXhosa has 18 phonemic clicks, and that's on the low end for languages that have them - Taa potentially has as many as 83 phonemic clicks.– SjiveruCommented Aug 22, 2013 at 23:57
-
As a general rule, the more consonant phonemes there are, the more conditioned vowel allophones there are. This is certainly true in the Caucusus (with ejectives instead of clicks), with the bounding case of Kabardian.– jlawlerCommented Aug 23, 2013 at 0:03
-
1My guess would be that clicks have formant transitions relatively similar to plosives/stops at the same place of articulation. Whether air is sucked in (clicks) or pushed out (plosives) doesn't affect the frequencies at which there is resonance in the oral cavity. And resonance is the basis of formants and formant transitions.– robertCommented Aug 23, 2013 at 14:14
-
2@ Robert: Not quite. A click involves a complete closure both at the velum and somewhere further forward. When the forward closure is released to suck in air, the velar closure still exists. So the resonating chamber isn't the whole oral cavity, but the much shorter part between the two closures. I'd predict higher formants for an alveolar click than for an alveolar plosive, and so on. And I doubt you'd get useful place-related formant transitions into the following vowel, since the vowel won't start till after the velar closure releases and its transition will be dominated by the velar.– Kevin RussellCommented Aug 26, 2013 at 1:57
-
@KevinRussell, good points!– robertCommented Aug 26, 2013 at 9:34
|
Show 1 more comment
1 Answer
I tenatively propose that you can not, and that click identity depends very heavily on the burst. I don't know if anyone has run the experimenent, but you can sort of start with recordings of Nama, Xhosa or !Xóõ, edit out the burst, and play the result. Ideally you'd want minimal pairs (and actual speakers of the language for your subjects), but you can get a quick DIY answer from these materials. My opinion is that click place cannot be identified at all, not just limiting oneself to formant transitions, but some of the fancy features like epiglottalization might have an audible carry-over past the burst though not something realized in the pattern of formant transitions.
EDIT: I'll be less tenative. Tony Traill's 1997 paper Linguistic phonetic features for clicks ran the experiment with !Xóõ speakers, and found this result. He notes that the anterior constriction does not coarticulate with the following vowel and the posterior constriction coarticulates with the preceding only to the point of indicating some velar constriction.
