There is a Persian word اسبابكشی asbābkeši and it is quite difficult to stick strictly to the transcription saying there is [s] before [b]. The assimilation of [s] to [z] before a voiced stop is present at my mother tongue — and, of course, that's why I have troubles with اسبابكشی — but at the same time I feel like this kind of sound accommodation is likely to be universal.

Well, is it true that the voiced obstruents inevitably become more voiced before the voiced plosive obstruents like [b]? And a Persian-specific question (I hope you don't mind) — how do I pronounce اسبابكشی? With [z]?

  • 6
    Just for perspective, it's very difficult to say any phonological process "always" happens. Compare the opposite direction of assimilation: absurd /bz/ in English, absurde /ps/ in French, in the same context. (Similarly, observe/observe.) Feb 2, 2020 at 14:42
  • @LukeSawczak thanks for your observations!
    – Aer
    Feb 2, 2020 at 15:35

4 Answers 4


No, there is no [z] sound in اسبابکشی, have never heard anyone pronounce it like that, the [s] is pronounced like s in someone or ass, but shorter.

Except when someone wants to sound funny e.g. in a video, or speak in a funny way to his friends or significant other or kids/babies either to amuse them, or sound intimate and loving, then he might convert [s] sounds to ز or ژ or ش regardless of its location (and maybe convert some other sounds to something else as well, like ش to س, or ب to ف, or ز to ژ or ج).

e.g. When you see/call your 2 or 3 years old niece, or your significant other, instead of asking her: سلام، خوبی عزیز دلم؟
you might (and it's not uncommon nowadays) to ask her:
شلام، خوفی عژیژ دلم؟.

Note that even in informal (colloquial) language, this is not the proper way of speaking and it's only used is such specific cases (and even in these cases not all the time, but it's not uncommon either, these days).

And to learn to pronounce اسبابکشی correctly, you can first try to pronounce each syllable separately like this:

as - bāb - ke - shi

Pronounce each syllable with a pause, and progressively try to shorten the pause, until you can pronounce it as one continuous word.


In Arabic, تَسْبِيح‎ [tasbi:ħ] is pronounced with s. It may well be common in human languages that sequences of obstruents agree in voicing, and the main tendency is for regressive assimilation, but such assimilation is not inevitable.

  • Thank you, that's helpful. So there's even no voiceless beginning in the articulation of [b]? (It may be a very subtle question...)
    – Aer
    Feb 2, 2020 at 15:30
  • 2
    An answer to that would need a phonetic lab and a large sample of speakers. No doubt there is variation. There always is.
    – jlawler
    Feb 2, 2020 at 16:04
  • 2
    You also need a definition of "voicing" that is nuanced enough that it encompasses the facts of language in general, not weighted in favor of "English versus French". You have to have a clear definition and justification of what you are measuring and what you are ignoring.
    – user6726
    Feb 2, 2020 at 16:18

It has been claimed by some phonological theories such as Lombardi's (1991) that (de)voicing is regressive in nature, which means that in your question we would expect /s/ to become [z] before /b/. However, this impressionistic claim is not enough to make a strong argument.

In contrast, based on experimental results that aim to quantify the laryngeal behaviour cross-linguistically, Beckman et al (2013) show that languages can be divided into voicing languages such as Russian and aspirating languages such as German. In voicing languages, the voice feature is strongly present and stop sounds contrast [voice] with [Φ]. On the other hand, in aspirating languages stop sounds contrast [spread glottis] with [Φ]. According to Beckman et al's (2013) theory, we want to establish first whether Persian is a voicing or aspirating language. If a voicing language, we would expect [voice] feature to spread in either direction regressive or progressive. What matters here is that [voice] feature of /b/ is most likely to spread to /s/ rather than [spread glottis] feature /s/, and vice versa if Persian is an aspirating language.

I know it is a little bit complicated but I hope this helps.

  • Thanks a lot! Could you please give full references to the works mentioned?
    – Aer
    Mar 2, 2020 at 22:12
  • 1
    Lombardi, Linda. 1991. Laryngeal features and laryngeal neutralization. PhD dissertation, University of Massachusetts, Amherst. Published by Garland, 1994. Beckman, J., Jessen, M. and Ringen, C., 2013. Empirical evidence for laryngeal features: Aspirating vs. true voice languages. Journal of Linguistics, 49(2), pp.259-284.
    – Dallak
    Mar 2, 2020 at 22:17

Commenting here just about English, I suspect that, though phonemically, /s/ -> /z/ before /b/, it is not so that phonetically, [s] -> [z] before [b]. After the medial vowel is dropped in a word like "possible" in casual speech, the cluster [sb] remains unassimilated. A pronunciation *[pazbl] is impossible.

  • Thanks; though it seems that after the elision of schwa in possible there is still some kind of a reduced sound (we should look at the spectrogram or something).
    – Aer
    Feb 2, 2020 at 17:52
  • I don't hear any vowel there. Perhaps you are hearing one because of the impossibility of /sb/ (since phonemics tends to govern perception).
    – Greg Lee
    Feb 2, 2020 at 18:00
  • 1
    If by "computerized investigation", you mean a recording from an instrumental device rather than a human transcriber, yes, I agree.
    – Greg Lee
    Feb 2, 2020 at 18:37
  • 6
    In some English words, we have neither /s/ -> /z/ before /b/.nor [s] -> [z] before [b]. For example "disbelief" or "disband".
    – Rosie F
    Feb 2, 2020 at 19:44
  • 3
    @RosieF, Right. In those examples, the s and b are in different morphemes and different syllables. Also "misbegotten", but "Brisbane" may be a problem (the Australian pronunciation has [zb], to my ear).
    – Greg Lee
    Feb 3, 2020 at 1:12

Your Answer

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

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