Whilst we're all familiar with voicing on an intuitive and/or phonological level, the actual acoustic phonetics are somewhat less intuitive to many of us.

The main way of formalising this intuitive notion of voicing in acoustic phonetics that I'm aware of is Voice Onset Time (VOT).

This is typically defined as the difference in time between the release of the consonant and the start of voicing.

  • A negative VOT means the voicing started before the release of the consonant and corresponds to a voiced consonant.
  • A VOT close to 0 means the voicing starts as the consonant is released and correspond to a tenuis voiceless consonant.
  • A positive VOT means the voicing starts only a while after the release of the consonant and corresponds to an aspirate consonant.

This definition works perfectly well for consonants in the syllable onset (although there are potentially issues post-pausa), but it's extremely unclear how to apply this in the syllable coda (especially pre-pausa). It also doesn't seem to have a way to describe preaspiration which according to Silverman, 2003, On the rarity of pre-aspirated stops can occur word and utterance initially in Huautla Mazatec (although no other examples seem to be known, so if this is mistaken then this issue is connected to the preceding one).

The obvious equivalent for coda consonants would be to look at the Voice Cessation Time (i.e. the time between the start of occlusion in the consonant and the end of the voicing) which would seem to give:

  • A negative VCT would then be preaspirated.
  • A VCT close to 0 would be tenuis.
  • A positive VCT would be voiced.

This seems a plausible extension, but doesn't seem to allow for final (post)aspirates, something that's frequently described.

My question then is: what's accousatically going on with preaspirates (especially word and utterance initially, if such exist), and relatedly, what's going on acoustically with (post)-aspirates word and utterance finally?

2 Answers 2


Voice onset time is a convenient measurement that can be used to compare aspirated, tenuis, and voiced stops to each other, since across different languages (or across different contexts within a language) the duration of aspiration or voicing is not always the same.

But aspiration does not consist solely of delay between release and voicing onset. Consider that we do not hear total silence between release and voicing onset: aspiration is a form of noise, and that noise can occur in other contexts than before a voiced segment. For comparison, the fricative [h] has no release phase, but it has turbulent airflow which produces audible noise. Preaspiration is when this type of noise occurs before the hold of a stop rather than after the release.

I found this web page showing what [h] and aspiration look like on a spectrogram: Identifying sounds in spectrograms

[h] is really a voiceless version of the preceding or following vowel. On a spectrogram, it looks a little like a cross between a fricative and a vowel. It will have a lot of random noise that looks like static, but through the static you can usually see the faint bands of the voiceless vowel's formants. [...] Aspiration will look like a period of [h] between the blank gap and the vowel -- specifically, a voiceless version of the following vowel.

When no vowel follows, presumably the acoustic characteristics of [h] and aspiration may depend on the identity of the preceding sounds.

In terms of how aspiration is articulated, here are some sections I found in a paper that seem relevant:

Earlier research has proposed to equate aspiration with VOT, but this approach has been shown to be problematic (Vaux and Samuels 2005:406), as it cannot account for instance for the existence of aspiration contrasts in coda position.

The most usual definition of aspiration is an articulatory one: the feature [spread glottis] (Kim 1970, Halle and Stevens 1971). Two competing theories exist concerning the exact interpretation of this feature; Kim’s original idea was that [spread glottis] depended on the size of glottal opening (the Glottal width theory), while others such as Löfqvist (1980) have argued that the aspiration contrast depended on the relative timing of laryngeal and oral gestures (the Glottal Timing theory). For stops, the aspiration contrast generally correlates with the opening of the glottis: voiced and voiceless unaspirated stops are pronounced with a smaller glottal opening (while the glottal opening gesture starts at implosion), while aspirated stops are pronounced with a spread glottis.

(p 21)


Friction and aspiration are basically two types of turbulent sound (noise), differing by the place where the noise is realized: at the glottis for aspiration, and in the supraglottal cavity for friction.

(p. 22)

(Guillaume Jacques. A panchronic study of aspirated fricatives, with new evidence from Pumi. Lingua, 2011, 121 (9), pp.1518-1538. 10.1016/j.lingua.2011.04.003 . halshs-00605893)


In the contexts where the VOT measurement of aspiration fails (absolutely or because of the wording of the measurement), you can still measure the duration of the aspiration, just as you can measure the duration of a fricative before a stop ([spɪt]) or after one ([pɪts]). Likewise pre-consonantal post-aspiration can be measured. In a vowel-adjacent context, there are additional challenges compared to measuring [s] duration given the low amplitude broad spectrum noise characterizing the aspiration. In other words, one can simply cut the cord w.r.t. voicing, then understand long-lag VOT as the overlay of a glottal fricative (aspiration) on top of the onset of a vowel, though this would be a very unconventional way of talking about aspiration.

Prepausal postaspirates would indeed be most challenging, since one expects there to be some release after a stop. If the measured final stop closure durations is e.g. 2 seconds on average, you can arguably attribute that to non-linguistic physiology and aerodynamics, but if it averages 200 msc and is followed by parseable noise, that would be aspiration in some sense. Then if there is a contrast in closure duration and post-closure fricative noise (e.g. Hindi sĩk 'twig', ãkʰ 'eye') there should be a lexical difference in one or both of those measures).

Since initial preaspiration would be very hard to detect especially outside of a lab setting, it is not surprising that this is rare. In some dialects of Arabic (e.g. Moroccan with rampant deletion of unstressed vowels leading to initial clusters), /hVC/ can end up as [hC], which is phonetically the same as preaspiration in Germanic and Saami, just phonologically different.

  • preconsontal aspiration makes sense in VOT sense when the following consonant is itself voiced, but what about if the following consonant is voiceless? In that case I struggle to see how a tenuis consonant makes sense (seeing as there is no voicing until well after the release of the consonant). Does that mean that all /atka/ ought properly to be /atʰka/?
    – Tristan
    Sep 24, 2023 at 15:59
  • Not all, for example my US English, it is [at̚ka] though /atka/. I believe Tsou is a language with an obligatory preconsonantal release (and initial stop clusters). You would not ordinarily call that "aspiration" since there is just one voiceless series. The reason for this release is that initial [tp, pt] are hard to parse without that release cue.
    – user6726
    Sep 24, 2023 at 16:32
  • @Tristan Further to what 6726 said, in English you'd almost definitely get [at’ka] if the [t] was audibly released (say in rather pedantic speech.) That is to say, it would be ejective. So it would be pretty hard to consider aspiration at all in relation to that [t], because there'd be no pulmonic air coming through the glottis at all. My intuition tells me that the pre-consonantal releases in Tsou are likely to be ejective too. Is that correct user6726? Sep 25, 2023 at 20:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.