Unvoiced schwa?

Question

I read about a the schwa being the laziest sound in all of human speech. This is because it just needs vocal cords and, poof, there's a uuuh sound. But, what happens if you take the vocal cords out? (I did this when whispering on purpose.) This sound seems even lazier. Is this sound found in human speech?

Answer 1

The English word potato is usually given a phonemic transcription starting with /pət/, but it can phonetically have no voicing in between the [p] and [t] sounds. I don't know enough phonetics to describe exactly whether this is best described phonetically as a "voiceless schwa", schwa deletion, or schwa replacement with a voiceless fricative.

Some linguists seem to have said that in phonetic terms, voiceless approximants do not exist as audible sounds: the audible part is supposedly the phonetic friction. I assume this would hold equally for voiceless vowels.

While it seems a bit strange to call something inaudible a "sound", it is possible for language to include articulatory motions that are not heard. However, this presents an obvious issue of how people acquire the habit of producing these articulations unless they are predictable from some other factor. (For example, in "potato", an English speaker might infer the presence of a phoneme /ə/ based on knowledge that /pt/ with no intervening vowel phoneme is not a valid word-initial cluster in English. Or the word might be produced with an at least partially voiced segment after the /p/ often enough that speakers remember that this word contains /ə/ even if it is pronounced in a way that makes the vowel sound inaudible.)

It does seems to be possible for turbulent noise to have similar phonetic characteristics to a vowel (the sound /h/ is often described as phonetically looking like a voiceless version of the adjacent vowel: see these spectrograms from Kevin Russell's web page at University of Manitoba). I guess it's a matter of terminological preference whether you call audible voiceless noise with formants resembling those of schwa "unvoiced schwa". It's also unclear to me whether a sound like this could be characterized as a "lazier" version of schwa: fricatives are commonly viewed in linguistics as "stronger" (more "fortis") than approximants, so I'm not sure whether frication of a vowel sound might also be viewed as a kind of fortition.

Presence of "devoiced" vowels in general in languages

As user6726 said, in languages where devoiced vowels occur, they generally seem not to exist as independent phonemes, but to be derived from non-voiceless vowel phonemes via some process of devoicing that applies as a rule in particular phonological contexts. In many languages, there seems to be some uncertainty about whether the phonetic outcome of a vowel devoicing process is a partially devoiced vowel, a fully voiceless but at least partially fricated vowel, a fully voiceless and perhaps inaudible frictionless vowel, or vowel deletion. Uncertainty could reflect either an actual phonetic gradience in the realization of devoiced vowels in the language, or methodological limitations (e.g. a purely acoustic study would not be able to differentiate between a deleted articulatory gesture and an inaudible articulatory gesture). Another attested phonetic realization of underlyingly devoiced vowels is lengthening of an adjacent consonant, especially a fricative; e.g. I have read that /su/ (also transcribed /sɯ/) in Japanese can be realized as [sː].

I haven't surveyed to see which languages with vowel devoicing have a schwa phoneme.

Some reading:

• "The Phonetics and Phonology of Non-Modal Vowels: A Cross-Linguistic Perspective," Matthew Gordon

• "The Phonetic Realization of Devoiced Vowels in the Southern Ute Language," Stacey Oberlya and Viktor Kharlamov, 2015

Answer 2

Laziness is an invalid value judgment when applied to language. It might be possible to scientifically describe the degree of difficulty in producing a certain sound in some context, so that you could say with scientific backing that a certain sound is "hard to produce".

For acoustic purposes, vowels are usually voiced (otherwise, they become very hard to identify). But in certain contexts (when the vocal folds are not supposed to be adducted before and after), rapidly switching the focal folds on and off can be very difficult, because they do not respond quickly. Therefore, shorter vowels may tend to devoice. Schwa, as well as the high vowels, tend to be the shortest, and in English this is especially true because schwa (as opposed to ʌ) only in unstressed syllables.

Voiceless vowels as actually-underlying sounds are probably nonexistent in known languages. There are rule-governed instances of vowel devoicing. Comanche is a case where the rules are phonologized, see this paper by Armagost, but still they derive by rule. So phonetically, there are phonetic voiceless schwas in some utterances in some languages (Berber and Maghribi Arabic are prominent examples, likewise certain Salishan languages). Or, certain kinds of consonant release that could be called "voiceless schwa". It is pretty much a question of analysis whether you say that č'ƛ'aʔ in Lushootseed is [č'ƛ'aʔ] or [č'ə̥ƛ'aʔ].