How to ejective

[Tropylium]

I got inspired to do a small survey of these, in both nat- and conlangs.
http://www.frathwiki.com/Ejective/Inventories

Seems that there's bit of a shortage of conlangs with /tsʼ tʃʼ/, while /cʼ/ is somewhat over-represented compared to natlangs.

Have you used ejectivs in a conlang? What kind of a system do they come in?

(Battlax and East Persian' there are my projects, and IIRC I was also the one to add them in the Round Robin Conlang./)

[Aurora Rossa]

My main conlang project Terpish uses ejectives. It has the series /p’ t’ ts’ tl’ k’/ paralleling voiceless, voiced, and aspirated series.

[Travis B.]

I got inspired to do a small survey of these, in both nat- and conlangs.
http://www.frathwiki.com/Ejective/Inventories

I cannot help but notice that one natlang is missing from there - English! Okay, sure, English lacks dedicated ejective phonemes, and its ejectives are rather constrained position-wise, but it has ejectives!

[Aurora Rossa]

I cannot help but notice that one natlang is missing from there - English! Okay, sure, English lacks dedicated ejective phonemes, and its ejectives are rather constrained position-wise, but it has ejectives!

I think they are only counting languages which feature ejectives as phonemic.

[Bob Johnson]

English lacks dedicated ejective phonemes

This would be why it's missing. Note that the page uses // everywhere, which means phonemes.

[Travis B.]

English lacks dedicated ejective phonemes

This would be why it's missing. Note that the page uses // everywhere, which means phonemes.

And do you think I was being entirely serious? Do you?

[Nortaneous]

I'm not entirely set on this yet, but Hoanu (a descendant of Kannow) will probably have something like /tʼ tθʼ tɬʼ tʂʼ kʼ kʷʼ qʼ qʷʼ/, contrasting with a voiced series of /b d dʐ gʷ ɢ ɢʷ/, although later the ejectives become plain stops and the voiced series either merges into plain or fricates.

[Tropylium]

Relatedly: in the natlang data, if there's a language missing /pʼ/, it seems to be in most cases inheritage from Proto-Na-Dene (which had hardly any labials at all), Proto-Mayan (which rather has *ɓ), or Proto-Afrasian. The only instance on there where a language seems to lack that while some related language has it, is Tillamook (which has done away with all the other presumably proto-Salish labials too). So where'd Hoanu misplace its?

[Risla]

South Eresian has /tsʼ tʃʼ tɬʼ/. It's also got /pʼ tʼ kʼ kʷʼ qʼ/.

[Nortaneous]

Relatedly: in the natlang data, if there's a language missing /pʼ/, it seems to be in most cases inheritage from Proto-Na-Dene (which had hardly any labials at all), Proto-Mayan (which rather has *ɓ), or Proto-Afrasian. The only instance on there where a language seems to lack that while some related language has it, is Tillamook (which has done away with all the other presumably proto-Salish labials too). So where'd Hoanu misplace its?

Folds into /b/, which either drops or turns to /v/, so Enzielu ends up with less <p>. Ugly letter, that. (Not that it'd make much difference, since only the eastern dialects borrow extensively from Hoanu...)

[Soap]

I think it's also possible to lack /pʼ/ if the process that generates ejectives in the first place just didnt apply to labials. Suppose ejectives come from voiceless stop + /k/. /kk/ > /kʼ/ is easy to imagine, /tk/ > /tʼ/ probably even more so, but /pk/ seems such an awkward cluster that it could easily just be ironed out before the ejectives arise. Or perhaps it would change to /kk/ first and then on to /kʼ/ again.

I've used ejectives in the proto-langs of my conlangs, but not much in the conlangs I show. In the very early stages of post-proto-Human, glottal stops began to disappear, but created ejectives if there was a voiceless stop before the vowel in the preceding syllable. These stayed in the language for a few thousand years, leaving some visible effects in some languages, but in all cases they disappeared eventually. I was thinking mostly of Semitic when I came up with these changes, so /kʼ/ > /q/ even when the other ejectives just changed to normal (homorganic) stops: either voiced (as in hypothetical glottalic PIE) or voiceless (as in Hebrew, I think?).

[Imralu]

I had a conlang with /pʼ tʼ kʼ tsʼ tʃʼ/.

I find [pʼ] the hardest the pronounce of all of them, and [kʼ] the easiest.

[TaylorS]

It took me a long time to "get" ejectives until I learned that English final pre-glottalized /p/ and /k/ are often realized as ejectives, and so now I can do them very easily.

[Travis B.]

It took me a long time to "get" ejectives until I learned that English final pre-glottalized /p/ and /k/ are often realized as ejectives, and so now I can do them very easily.

Very much the same here.

[Radius Solis]

In languages that have any ejectives, the rearmost stop POA (not counting pharyngeals etc) virtually always has an ejective and the frontmost is least likely, with a gradient in between. So a system of /p t k q t' k' q'/ is well behaved, whereas /p t k q p' t'/ is unlikely to exist in nature.

I played with this in Tlaliolz when it developed ejectives by fortition of what was presumably an intermediate aspirated series: it did so only for *k *kw *q *qw. The remaining two aspirates, *p *t, fortified instead by affricating to [pɸ tɬ], with the labial one undergoing subsequent deaffrication to [ɸ]. (Western dialects, which had [tθ] for the coronal affricate, also deaffricated this to just [θ].)

[Shihali]

It took me a long time to "get" ejectives until I learned that English final pre-glottalized /p/ and /k/ are often realized as ejectives, and so now I can do them very easily.

For me, the main breakthrough was the notion of a /k/ and a glottal stop said at the same time; even if it's not phonologically accurate, it produced "ejectives" that sounded like ejectives so I concluded they're ejectives. But the pre-glottalized /p/ idea just made /p'/ much easier to say!

[Whimemsz]

One natlang you need to add to your list is Huamelultec (aka Lowland Oaxaca Chontal), whose only ejectives/glottalized obstruents are: /fʼ sʼ ɬʼ xʼ/. The phonetic realizations vary pretty widely, but in part because they alternate morphophonologically with corresponding plain fricatives, Ian Maddieson, Heriberto Avelino, and Loretta O'Connor analyze them as basically glottalized/ejective fricatives. (Based on the frequency of the realizations of each one, it would also be reasonable to analyze them as respectively /fʼ tsʼ tɬʼ kʼ/).

NE: Source: Maddieson et al, 2009, "The Phonetic Structure of Oaxaca Chontal", IJAL vol. 75(1): 69-101

[Tropylium]

In languages that have any ejectives, the rearmost stop POA (not counting pharyngeals etc) virtually always has an ejective and the frontmost is least likely, with a gradient in between. So a system of /p t k q t' k' q'/ is well behaved, whereas /p t k q p' t'/ is unlikely to exist in nature.

A few Mayan varieties have /ɓ ʛ/ vs. /tʼ Cʼ kʼ/ (using C as a cover symbol for affricates) tho, which would seem to point to velars being the "stability peak".

As for pronouncing them, I used to think they require extremely strong glottalization which made actually incorporating them into words difficult, esp. before a vowel; but unsurprizingly (as they are a reasonably frequent sound after all!), after some practice it turns out it's possible to make do with a much "softer" pronunciation while still keeping them distinct.