Saturday, June 5, 2010

Ejectives

I gave my students a brief introduction to IPA last week, and this coming week we're spending every class on a different aspect of phonetics, including IPA transcription. So I'm getting ready for a fun week of making silly sounds and writing with symbols that most people have never seen. One sound class that people seem to enjoy encountering for the first time is ejectives. They occur only in about 15% of languages, and most of those languages are minority languages that mainstream people never hear. In fact, they're exotic enough that Paul Frommer chose them as an element of the Na'vi language used in the movie Avatar.

Most inexpert descriptions of ejectives are mindbogglingly useless. If I tell you to pronounce /t'/ by pronouncing a more forceful "t" sound, I'll wager just about any amount of money or hat-eating that you're not going to come up with an ejective. The best way I've found to describe ejectives is to pronounce a consonant while holding your breath. After all, this is essentially what distinguishes ejective stops from regular stops: regular stops use the pulmonary airstream, while ejective stops use the glottalic airstream, without accessing air from the lungs. Ejective stops are the most common ejectives, but languages also have ejective fricatives such as /f'/ and /s'/, and Tlingit even has an ejective lateral fricative (it's a pretty tough one the first couple times). Though no language has ever come up with one of my favorite possible sounds: an ejective alevolar trill. Give it a try and see how many contacts you get before your tongue runs out of air.

5 comments:

James Crippen said...

I tried /r̥ʼ/ and got approximately five contacts out of it with a glottal articulation similar to what I use for Tlingit ejective fricatives. I would need to check it on a spectrogram to be sure, but actually I think it’s essentially the same as an ordinary alveolar trill. But of course this requires a speaker to have a well developed ability to do ejective fricatives. In particular, they are produced with a tighter articulatory closure than regular fricatives.

BTW, there are two different kinds of ejective stops. They are called variously “stiff”/“slack”, ”strong”/“weak”, etc. The stiff ejectives are the classical example, as in Navajo, where the glottal closure is timed well before the articulator release so that the burst is loud, long, and noisy. The slack ejectives are like those in Tsimshianic languages, where the glottal closure nearly overlaps with the articulator release and the burst is quiet, short, and not nearly as noisy. A linguist accustomed to stiff ejectives will have a hard time perceiving the slack ejectives, they sound more like a sequence of [ʔC] than they do a unit like [Cʼ], although they are not actually [ʔC] at all.

To teach people to do ejectives I usually start with [pʼ] or [tʼ]. First I get them to do [ʔC] for a while to get used to the sequence of glottal closure and then articulator release. Then I teach them to start the articulator gesture before the glottal closure. After that the glottal release is held through the articulator release. Finally timing the glottal release just during the articulator release is the last step. After that it’s just practice and muscle memory.

Ejective fricatives are another thing entirely. Teaching people them is hard, and the stops must absolutely come first. The ejective uvular fricative is the hardest of all, many people never really get it separated from just [qʼ]. Heck, I still produce [qʼ] by accident occasionally when I’m aiming for [χʼ], and even native speakers of Tlingit do too. It’s just a really hard sound, which is almost certainly why it’s in the word /χʼé/ ‘mouth’ and hence anything to do with language like /juːχʼatánk/ ‘language, speech’, /χʼatʰuːw/ ‘meaning’, /χʼagáːxʼ/ ‘prayer’, /juːχʼaɬaʔátk/ ‘talk, conversation’, and so forth.

Sandro Abramishvili said...

hi, a was wikislipping around and discovered myself at your blog. I am happy to find someone who also pronounces alveolar ejective trill :D. from my childhood i'm inventing some languages (one of them turned out quite cool :P) and of course besides lexical inventions I was inventing more phonemes. For now, in my one of the languages (which i mentioned above) there is 42 phonemes with more than 50 sounds. when i discovered my ability of pronouncing strange sounds, i went on, and for now i am up to about 150 sounds.
And yes, ejectives are my favorite ones. Recently i discovered that in my native language (Georgian) there are many sounds that are not phonemes. one of them is alveolar ejective trill. that happens when alveolar trill is standing before ejective. But in georgian more frequently alveolar trill is shifted into alveolar tap. therefore, it will sound as alveolar ejective tap before ejectives. Try to pronounce it, if not yet :).

Anonymous said...

I was complaining on Twitter (@dialect, if you're on Twitter) about not being about produce ejectives (for an ultrasound & MRI experiment I'm participating in at Queen Margaret University), and someone responded to me with this blog post. Nice post and nice discussion, it's definitely helped me!

Zort said...

How does an ejective alveolar trill work? I was under the impression that trills don't sound like much of anything when they're unvoiced and all ejective consonants have to be unvoiced.

Ryan Denzer-King said...

While trills (and other sonorants) are usually voiced, there are a number of languages that have phonemic voiceless trills. Icelandic and Welsh have contrastive voiceless alveolar trills, and Ubykh and Wari' have (allophonic?) bilabial trills (or more precisely, a voiceless bilabial "affricate" trill that begins with an alveolar stop). Many other languages have non-contrastive voiceless trills in certain positions.

Ultimately a trill is just a phoneme with rapid, usually irregular contacts created by airflow (rather than volitional control of the articulators). Why are trills usually voiced? I'm not sure, but I would guess it has something to do with the need for a certain amount of air pressure delivered somewhat consistently, and voiced sounds tend to have less airflow than voiceless sounds, meaning you'd have more time to do the trill. On the other hand, it's probably easier to get a voiceless trill going, since the suboral pressure is higher.