Is there a non-physiological way of distinguishing vowel sounds from consonant sounds?

Well, I’d suggest that it’s pretty much impossible to give a phenomenological account of how a vowel sounds different from a consonant – that would be circular. Formation of a consonant sound involves obstruction of the airflow. Formation of a vowel doesn’t.

As for the clicking languages, it’s very much a matter of what you’re used to – I’m sure for the speakers, it’s a perfectly natural way of communicating. There’s some indication that they were even among the first human languages.

Chinese uses intonation as a morpheme, so there’s no reason (apart from the total impracticality) why the sound of a bell can’t be used in communication.

Incidentally, check out Silbo Gomero and Pirahã – Silbo Gomero speaks for itself, and Pirahã has a complex prosodic system that (allegedly) allows its speakers to communicate without forming consonants.

I guess for another way to describe consonant or vowel sounds, you could do a computer-based analysis of the waveforms produced by each sound. Perhaps the vowel sounds will have more regular waveforms closer to a sine wave, while consonant sounds (maybe excluding the nasal consonant phonemes [m], [n], etc…) will have somewhat spikier waveforms. I’ve never seen such an analysis, though. Doubtless one exists somewhere that will (dis)confirm the hypothesis.

The language of Hawaii used a lot more vowels than consonants and they obstruct their throats without closing their lips. The sound is very distinct. Or they actually stop the flow of air for the space of time, I have no idea how that would be measured.
You would need a linguist, or a team of linguists working with a team of computer prograammers to run the program the way you speak of. Because you cannot have a standard english speaking computer accurately reproduce sounds as they are pronounced by native speakers of other languages. If you could teach your computer Hawaiian-speak, it could make Hawaiian sounding words of vowels without consonants, or maybe just one consonant. Your program does not have all the parts of the equation for what you want to do.

If you you look at waveforms, you can easily spot the consonants. They start with a spike.

Audacity is free audio recorder and editor which is great for messing around with this stuff.
Windows/Mac/Linux
http://audacity.sourceforge.net/download/

R is a consonant and doesn’t obstruct the airflow. That was the first consonant I thought of, and since it was a contradiction to your google I didn’t continue. Any sound or motion that can be generated could be a part of language. Animals use sounds and body English to communicate.

@evandad, good point, R is very different from K. I think in waveforms it would look more like A and E than T or K.

From Wikipedia:

Vowels are “pronounced with an open vocal tract so that there is no build-up of air pressure at any point above the glottis.”

Consonants are “articulated with complete or partial closure of the vocal tract. Examples are [p], pronounced with the lips; [t], pronounced with the front of the tongue; [k], pronounced with the back of the tongue; [h], pronounced in the throat; [f] and [s], pronounced by forcing air through a narrow channel (fricatives); and [m] and [n], which have air flowing through the nose (nasals).”

Learn the manner of articulation for vowels: Front, Back, High, Middle, Low, Tense, Lax. Look for a linguistics tutorial on line for phoneme production.

@Trillian You’re talking about the glottal stop in Hawaiian. The glottal stop is a consonant – it involves obstruction of the airflow. Not all consonants are made with the lips.

@evandad R does involve obstruction of the airflow. Not much obstruction, but more obstruction than a vowel.

In some languages (such as Sanskrit, as I recall), R and L are grouped with the vowels or classed distinctively as “liquids.” Also R can sound very different in different languages and may even have different sounds within the same language.

@Jeruba Sanskrit has two kinds of vocalic R (ऋ ॠ) but also has “regular” consonantal R (र). “Liquid” is a way of grouping /r/ and /l/ together, because some phonological rules only apply to /r/ and /l/.

Are they allophones of the same phoneme in Sanskrit, as they are, say, in Japanese?

@dpworkin no. And my sentence about liquids isn’t about Sanskrit. Some languages, like English, treat /l/ and /r/ in similar ways, so we classify them as “liquids”.

I wish I knew more on this subject.

@dpworkin There’s “A Course in Phonetics” by Peter Ladefoged.