That is so far wrong it makes me want to peas. (Actually, this argument seems to arise every couple of weeks, with nothing new really being contributed. There are people who can hear minor distinctions between words, either different supposed homonyms pronounced in the "standard" fashion or the same word pronounced by people with slightly different dialects. And other people claim they cannot hear the differences, or at least that the differences are being misinterpreted, or are the result of simple mispronunciation. Until we can get in each others heads we must agree to disagree.)

@HotLicks OK, that';s fine, but presumably you have some vetted source to back that up with? Because it's a matter of science - not what you think ...

What I think is a matter of science? Those FMRIs at Mayo must have been more detailed than I thought! (There is very little "science" in this area, just various schools of "experts" who invent their nomenclatures and notations and try to jam everything into their categories. It works to a degree, but it can never capture the richness of the English language.)

@HotLicks No, what you think isn't a matter of science! But whether there is any voicing, for example, in a consonant segment, is a matter of science. It's pretty easy to do, you use a CAMERA that you put in someones throat. or you can record the WAVEFORMS Or you can use a SPECTROGRAM. Which parts of that are "very little science in this area"?

@HotLicks Perhaps you would like to nail down something that's in accurate or unproven in my post, so that readers here can benefit from your knowledge and expertise in this area?

Sigh. For the past 3 years I have worked, off and on, on an iPhone application which analyzes sound. (This is an actual paying job.) Often, when I look at the waveforms and spectra, I see details that the ear doesn't detect. But also fairly often the ear can hear things in the (recorded) sound that the waveforms/spectra do not reveal (at least not at the crude level you present). The brain is a very sophisticated sound analysis computer, formed over a million or so years, then fine-tuned by 10-20 years of training. It's asking a lot for a simple waveform and spectrum to be as discriminating.

@HotLicks We're talking about whether a vowel is 50% or sometimes more shorter, or whether a segment has vocal fold vibration. So that's pretty clear cut and not up for dispute. Your comments at the top here are totally unwarranted, and your most recent comment doesn't really seem relevant. Sorry. You are also contradicting yourself (people think they can discriminate things - but they can't/people can consciously discriminate sounds better than a machine). What exactly is the point you were trying to make. I still don't know which point(s) in my post you think were wrong.

You know, all I said is that we must agree to disagree.

@HotLicks I thought you said "That is so far wrong it makes me want to peas". But you haven't helped me out by telling me what I've said that's wrong! I don't mind agreeing to disagree, but it would be helpful to know exactly what we're disagreeing about! :-)

I await the outcome of this debate with bated, er, aspiration, so I can decide whether to +1. Seriously, @HotLicks, if you can still remember it, could you spell out your point of disagreement with this answer? I'm hoping that what you say might shed some light on something else I'm interested in where some people say that the acoustic evidencce from spectrograms settles the matter and others don't.

@BenKovitz I don't think this is one of those areas. Voicing by definition means a periodic waveform produced in the airstream. So if the egressive airstream does not have a periodic waveform, there is no voicing.

@BenKovitz I have a suggestion: I think if you repost a version of this question on Linguistics SE, you'll get a good answer - and I can post my answer there and you'll get some actual scientists either answering themselves or commenting on my/other answers. What do you think? (like your meta post, btw)

@AraucariaMan Good idea! When I come back from taking my own advice in that meta post, I'll look into this. (I don't have serious doubts about your answer, which I find utterly and even definitively clear, useful, accurate, informative, thorough, and to have already gotten my +1. :) )

OK, I'm back. Now that it's sloshed around in my head a couple days, I think I couldn't ask this question on linguistics.SE fully in good faith. I could still ask it, of course, but I'm confident you've already answered it correctly. I am satisfied and enlightened. Right now I'm mostly interested in understanding @HotLicks' objection in the hope that it will trigger some unexpected new insight—the kind that one can't ask about but can only stumble upon.

BTW, I'm further convinced partly by some personal observations that agree with what you've so excellently summarized from the research literature: (1) I do hear these aspects of English speech, without use of a spectrogram. (2) I've often gotten objections of the "Gibberish! Claptrap!" variety when I've tried to point them out to fellow native speakers. Most native speakers can't hear them. (3) But that changes when you stop and start an audio recording at just the right spot. Then even a native speaker who can't hear the aspiration or devoicing can easily hear the h or s. …

@BenKovitz - What don't you understand about agreeing to disagree?

@HotLicks I don't understand specifically what is the proposition you disagree with regarding this answer. Will you please tell what it is? I'm also curious about your reason. I've also noticed that the brain does some mind-bogglingly sophisticated signal processing, almost certainly not addressed by any existing scientific theory, maybe involving subtleties that don't show up on spectrograms. For example, what's the difference between sounding 'acty' and spontaneous speech? Nobody knows, AFAIK. It sounds like you've noticed something like that, but so far I can't tell what you're pointing to.