voiced/unvoiced - the real story?

[Lorikeet]

Does anyone seriously teach their students that vowels are longer before voiced sounds, for example?

Absolutely! It removes a lot of difficulty for Cantonese speakers, for example, who tend to pronounce cap, cab, cat, and cad the same, or back and bag, for example. It is such a fundamental aspect of pronunciation I don't know how you could *not* teach it. Of course, we don't emphasis the final stop in American English. I suppose if you were teaching British English and emphasized the ending sound, you might not consider it as important.

[Lorikeet]

For example, take the minimal pair writer/rider. British English speakers would think of the distinguishing feature as being t or d. However, in American English, voiceless sounds are voiced between vowels (an alternative analysis is that the voicing isn't "switched off"), so the middle consonant is realised as /d/ both times. Nevertheless, the two words don't sound the same in AE: the /ai/ in rider is longer than writer because vowels are longer before voiced sounds. Native speakers are rarely aware of this until it's pointed out to them.

Actually, I did a paper in graduate school on this, and I disagree. I would choose the pair, "I wrote it again." and "I rode it again." however. In my dialect of American English, these two sentences are exactly the same. (I tested it at the time, in the ancient past, when we used tape spicing, by the way.) and after eliciting the sentence (John wrote the poem and then he....) spliced the ending from one on to the other. They were interchangeable. There was no lengthening before voiced or shortening before voiceless. It is not a "d" sound, but a sound that is sometimes called a "tap" or a "flap" in the literature I've seen. Apparently, the tap or flap occurs before the +length for voicing.

I come from Detroit, which is near the Canadian border, and I happen to have a slight vowel change in /ai/ before voiceless consonants, which is why "writer" and "rider" are different to me, and I can't use them as examples.

[lolwhites]

But my initial argument is that d isn't voiced, it just lacks aspiration, so I'd like to find a rephrasing of lol's argument! (and anyway, since a U.S medial t is "voiced" you would have to explain the vowel length depending on the theoretical underlying form and not the actual "voicing" )

As to the thinkers experiment, I'd like to take up the challenge, but we have to even things up. People will equate a nonsense word such as "spug" with the closest word they know. That might be "spud" (a British potato!) and thus they will claim a "p" sound.

How about if we introduce it like this though? "On the planet Blargzaz, are the towns of Sbeg, Sboodle and Sbroggin". We show these spellings to our subjects. "There is also a town called 'spambarg'". We do not show this spelling, and we ask our subjects which sound they hear after the s.

Actually if I try and say sboodle/spoodle, it is in fact perhaps the length of the s which would distinguish most. So if you are going to say that medial "voiced" sounds are sounds with longish sounds in front, instead of fiddling with your buzzy throat, then you'll have a point. However I suggest that "voicing" is thus a confusing term and should be referred to as "prior lengthening"! That is perhaps the main problem with phonemics - timing is so crucial and yet phonemics does not really deal with time very much.

By the way, I guess the reason that the sounds are shorter before aspirated consonants is that we tense up our mouths to make the explosion and cut off the previous sound while doing that. In the P in spam we possibly (and only sometimes, I would hazard) tense up our mouths and then do not release the explosion, so that the P will sound tenser than a regular B. Still - nothing to do with voicing.

When I tried saying "spoodle" and "sboodle", I found that the voicing in the /b/ switches on about halfway through. It reminded me of the researchers when I was at Uni, who had hours of fun attaching electrodes to people's throats, then getting them to say stuff and working out the Voicing Onset Time. The thing is that "voiced" consonants in English tend to have longer VOTs than they do in other languages.

If you analyse the utterances horizontally rather than segmenting vertically, the voicing in "Sbeg" starts somewhere between the end of the /s/ and the start of the /e/, but not definitely bang on at the beginning of the /b/. If there was a neighbouring town called "Speg", I would expect the voicing to kick in a few milliseconds later.

A similar thing happens to the nasalisation in "Spam". The nasal sound starts not on the /m/, but somewhere towards the end of the /ae/. The physiological reason for this is that the velum moves more slowly than the lips or tongue, so you start to move it slightly earlier to get it into position.

This is why some of my lecturers started saying phonemes didn't really exist, and that we should consider the word or morpheme as a sequence of coordinated movements of different speech organs. But that didn't come until the end of the third year, and I didn't study it any further than that.

Woody's thought experiment about "spambarg" is an interesting one. What our subjects thought would depend whether Blargzazian phonotactic rules allowed an /sp/ combination.

[woodcutter]

Actually I'm not sure it would really be a worthwhile experiment, because it is a question of which suggestive influence would be stronger, the alien planet introduction or regular spelling, and of how the researchers actually phrased "Spambarg". I suspect that "The inhabitants of Spambarg are green" might have a P sound which is more like a B than just the word said solo, for example.

I think the discussion over the p in spam etc is a bit of a side issue really, cleary it can vary from an initial B or P and not quite be either, and there will be certain arguments for making it an allophone of either, and indeed for throwing out the whole system like Lol's lecturers. Everything always seems to reach that point at higher levels.......

With stops, there can be no voicing (there's no air flow) in the build up/initial release stage, so I don't like the "halfway through" idea. You can do some voicing before that stage (which is characteristic of French for example, so I am told), but it won't have the full characteristics of the following sound, and I don't know if I believe it is ever used to distinguish sounds. The medial stops do not need to be distinguished by either voicing or aspiration.

[lolwhites]

With stops, there can be no voicing (there's no air flow) in the build up/initial release stage, so I don't like the "halfway through" idea.

It's more during the "follow through" stage that the voicing kicks in. It's a short period, but definitely measurable. In French and Spanish, as you rightly point out, voicing starts during the build up.

[woodcutter]

I don't see what you mean about "follow through". If it is after the actual stop in the flow of air, how can that be pre-voicing, or notable? As I said all along, a B in English will voice basically just as soon as you release it, and only aspiration provides an alternative to that.

I still think that a crisp "Bonjour" with no pre-voicing might sound a bit Rosbiffy, but that that kind of thing would never cause a problem in comprehension. After all as we suggested there are probably English dialects in which pre-voicing is common.

[woodcutter]

A guy on the Korean discussion forum thread that caused me to start this one said this:

It's actually possible to voice a stop. A stop means a stop in the airflow out of the oral cavity, not necessarily a stop of the airflow into the oral cavity. Air can go between the vibrating vocal folds and into the mouth with the lips closed. The cheeks or lips will puff out a bit to accommodate the extra air. The French /b/ is an example of that (the lips puff a little bit before the release).

For a rare English example, Robert Plant sometimes makes this fully voiced /b/ on the word "babe" in the first line of the song, "I Can't Quit You Baby."

You also do it when you make a fake barfing sound like you're trying to hold it in, you know, mmm mmm mmblaaaaa!! (but of course the "m" sounds are unreleased non-nasal voiced bilabial stops)

: I can't really argue with that. However he also says French makes a distinction with this pre-voicing because it has an unaspirated P. I don't buy that - the French can whisper as well as anyone I think and their P is tensed (as well as slightly aspirated?)

[lolwhites]

I recall various demonstration from the Phonetics component of my Linguistics degree where a mask was placed over a volunteer's mouth and nose to measure airflow, with a microphone or electrodes on the throat to measure voicing. This made it possible to accurately measure when plosives were released, voicing onset times etc.

By "follow through" I mean the time between the release of the stop and the beginning of the vowel in a sequence like "sbug" or "spug" (n.b. "spug" is not a nonsense word to Judge Dredd fans, but I digress...); the vowel doesn't begin immediately as the speech organs take time to get into place. Depending on whether you're saying "sbug" or "spug", the voicing kicks in at a different time during this short period. However, it begs the question "If the distinguishing feature is the VOT, on what basis do we say one sound is 'voiced' and the other 'unvoiced'?"

[woodcutter]

That makes sense, and I wonder if it is worth talking about degrees of aspiration (with the French P in mind!).

By the way, the Korean forum guy didn't like saying "stops" because that can include nasal "stops", but if non-aspirated stops don't actually stop either then that becomes a pretty confusing term.

[Lorikeet]

That makes sense, and I wonder if it is worth talking about degrees of aspiration (with the French P in mind!).

By the way, the Korean forum guy didn't like saying "stops" because that can include nasal "stops", but if non-aspirated stops don't actually stop either then that becomes a pretty confusing term.

What is a nasal stop?

[lolwhites]

During the first year of my linguistics degree we were told that /b/, /p/ etc were !"oral stops" and /m/, /n/ etc were "nasal stops". In the second year "nasal stops" became "nasals" and "oral stops" became "plosives".

Not sure why they did this, maybe it was to make us aware that the place of articulation was the same, but we knew that anyway because we already used terms like "bilabial", "labiodental", "alveolar" and so on.

[Lorikeet]

Sigh. I have always had trouble trying to make sense out of things that seem nonsensical. I thought I understood a stop. It's really easy, in my dialect, where the final "stop" ends with the articulation, and without a release I guess. But "stops" to me don't linger, like mmmmmmm can. Color me hopeless.