Ejectives in Zulu

[Chengjiang]

So Zulu has a mostly three-way contrast* between ejective, aspirated, and breathy voice (also known as "depressor" due to its effect on tone) in its stops and affricates. All well and good in and of itself, but there are a few things I find a bit odd:

• Zulu's orthography represents the ejectives as the base set (in parallel to, say, the tenuis clicks) and the aspirates as marked. Additionally, some documentation on the language refers to the ejectives as "plain" stops. Do these differ in some way from the ejectives of other language families?
• Voiceless fricatives become ejective affricates when prenasalized. I can understand them becoming affricates, but I don't really get why prenasalization > glottalization in this case. The language has /tʃʼ/ in non-prenasalized contexts as well so I could understand an affricate allophone of /ʃ/ becoming ejective by analogy, but /f s ɬ/ do not have corresponding ejective affricate phonemes; why are their affricate allophones ejective?

*It's marginally four-way, in that there's also a bilabial implosive that contrasts with the other bilabial stops. There's a phonetic velar implosive as well, but only as an allophone of the velar ejective.

[Salmoneus]

I've no idea, but wouldn't both of those be explained by a simple shift of plain voiceless to ejective stops and affricates?

[vokzhen]

Yea, I assume that's what it is. Eastern Armenian and, more closely/relevantly, Sotho show shifts of plain/aspirated > ejective/aspirated under the influence of languages with ejectives.

[Chengjiang]

OK, yes, upon further inspection this appears to be shift of tenuis > ejective in the Nguni languages probably caused by Khoisan influence.

[Pabappa]

Oh, thank you so very much. I've been wanting to justify a shift of /d t/ to /t t_>/ unconditionally (with similar shifts for the other stops) for a while now and I just couldn't quite convince myself that it would happen. I'm still not entirely comfortable with it, since ejectives would outnumber the pulmonics*, but I could pare it back down with other sound changes later. That said, I think there are some languages in the Caucausus, and perhaps also Zulu, where ejectives really do outnumber the pulmonics.

*Wrong word? I just mean "normal, non-ejective stops".

[malloc]

Oh, thank you so very much. I've been wanting to justify a shift of /d t/ to /t t_>/ unconditionally (with similar shifts for the other stops) for a while now and I just couldn't quite convince myself that it would happen. I'm still not entirely comfortable with it, since ejectives would outnumber the pulmonics*, but I could pare it back down with other sound changes later. That said, I think there are some languages in the Caucausus, and perhaps also Zulu, where ejectives really do outnumber the pulmonics.

Perhaps you could have something similar to Grassman's law and have ejectives dissimilate to pulmonics in the vicinity of other ejectives. Eg, /k_>at_>u/ > /k_>atu/ or something.

[Salmoneus]

Oh, thank you so very much. I've been wanting to justify a shift of /d t/ to /t t_>/ unconditionally (with similar shifts for the other stops) for a while now and I just couldn't quite convince myself that it would happen. I'm still not entirely comfortable with it, since ejectives would outnumber the pulmonics*, but I could pare it back down with other sound changes later. That said, I think there are some languages in the Caucausus, and perhaps also Zulu, where ejectives really do outnumber the pulmonics.

Perhaps you could have something similar to Grassman's law and have ejectives dissimilate to pulmonics in the vicinity of other ejectives. Eg, /k_>at_>u/ > /k_>atu/ or something.

Plausible. And in Kartvelian, the Grassman-equivalent dissimilates ejectives to voiced stops. Alternatively you could have them become voiced next to sonorants, or fricatives when initial, etc. The lots-of-ejectives stage only has to be transitional, after all.

Alternatively, with a plain voicing contrast, perhaps sound changes that resulted in voiceless stops being significantly less common than voiced ones might actually trigger the voiceless stops to become more marked through glottalisation?

Incidentally, English is another example of this shift: many British speakers have ejectives for voiceless stops immediately before a pause - they're a natural development of our pre- and post-glottalised final stops. It's quite easy to imagine them becoming standard pre-pausally, and then they can spread by analogy to be standard when final, and then they could spread by analogy to be standard throughout. It's not massively likely to happen in English, but it's not implausible either. Add in a high-prestige neighbouring language that already has ejectives...

[8Deer]

Oh, thank you so very much. I've been wanting to justify a shift of /d t/ to /t t_>/ unconditionally (with similar shifts for the other stops) for a while now and I just couldn't quite convince myself that it would happen. I'm still not entirely comfortable with it, since ejectives would outnumber the pulmonics*, but I could pare it back down with other sound changes later. That said, I think there are some languages in the Caucausus, and perhaps also Zulu, where ejectives really do outnumber the pulmonics.

Perhaps you could have something similar to Grassman's law and have ejectives dissimilate to pulmonics in the vicinity of other ejectives. Eg, /k_>at_>u/ > /k_>atu/ or something.

This exact sound change occurs in the Salish language Shuswap, only in the reverse order. So roots like *p'at' > pat'.