I had a linguistics professor who adamantly maintained English had no voiced stops and they were actually voiceless unaspirated. I guess I just imagined the vibration in my throat, then.
Linguistics is probably one of the sciences that has the most pseudo-science mixed in with it.
More that it's on the softer edge of science, which correlates with how many competing theories there are in it. The big outlier, of course, is physics, which is so hard that it's gone horseshoe into an all-out abstract circlejerk about how the universe works.
But... I mean, the professor's claim is just flat-out wrong. It's just provably false by anyone who can put their fingers against their Adam's apple and say "bagged".
He's not wrong. English really just has a strong and a weak series of stops, which manifest word-initially as an aspiration distinction and word-medially as a voicing distinction.
How can you test this? English stops are voiceless but unaspirated in word-initial clusters after /s/- so while
pie is [pʰai] (with aspiration),
spy is [spai] (with no aspiration).
If you record a native English speaker saying
spy, then mess around with the recording to delete the /s/ (this is easily done in a program called Praat, which is free to download and which linguists use all the time), and then play the result to a native English speaker, they'll hear
by. In other words, word-initially, aspiration is the salient feature- not voicing. (Weak stops
are usually at least a little voiced word-initially, but that's not what we listen for, in other words.)
This sort of situation, where a set of sounds cuts across two categories that are usually well-defined, is not actually very unusual. Most of the indigenous languages of Brazil have a system where nasal vowels are phonemic (like in French or Portuguese), and there's a series of stops that are just plain voiced stops [b d g] before non-nasal vowels, but nasals [m n ŋ] before nasal vowels. These contrast with the plain unvoiced stops, which don't vary like this. Are they nasals or voiced stops, underlyingly? There's a good case to be made for both.