I have read some time ago that [z] is the voiced counterpart of [s], as [d]-[t], [g]-[k], [b]-[p] and [v]-[f]. For all pairs except the first, I was able to consciously perceive it by starting to pronounce the unvoiced consonant, adding voice and realizing that I am now doing a voiced consonant.

However, when I start to produce a S (AFAIK a hissing sound) and add voice, I don't get a Z, but rather a hissed Z. AFAIK Z is the same sound a flying insect (eg bee) does and has no hiss on it. Or is that another phone? Am I the only one who perceives this difference between a voiced S and Z? I don't know if I'm mispronouncing [s] or [z], somehow affected by my native language (Brazilian Portuguese) or by some regionalism.

  • 1
    I wish that the person who voted negative for this question would explain why. Sep 13, 2019 at 23:48
  • By definition the only difference is voicing.
    – curiousdannii
    Sep 14, 2019 at 0:10
  • 1
    I know the theoretical definition, but I'm not able to replicate the simple voice/unvoiced difference and I'm looking for help. I also wanted to know (1) if it is valid to all languages and (2) how exactly my pronunciation deviates from the standard. Sep 14, 2019 at 2:05
  • Tried to compare my [s] and [z] (Russian of Ukraine) - except voicing, difference is quite subtle (a bit wider mouth with [s] and tongue a tiny closer to teeth with [z]), sounds (when add/remove voicing) are indistinguishable.
    – Netch
    Feb 25, 2020 at 7:05

5 Answers 5


Phonetically, the only distinction between a pure, unaltered IPA [s] and a pure, unaltered IPA [z] is voicing. However, most languages don't use pure, unaltered IPA [s] and [z]!

One phoneticist I know, for example, tends to pronounce English /s/ as [s͎]: a "whistled" sibilant with a very distinctive, high-pitched "overtone". This makes his /s/ very different from his /z/, in more ways than simple voicing.

I can't say for sure without a recording, but your /s/ may be similar, or have some other feature setting it apart from [s]. If you want to make sure, I'd suggest finding some good phonetic recordings (the ones made by Ladefoged are especially famous) and comparing your /s/ against them.

  • I haven't found the audios you mentioned, but any IPA chart with audio (eg ipachart.com) should do it, right? I have been able to reach a new sibilant sound (new for me) doing the opposite: starting with /z/ and then dropping voice. That sound is surely not how I pronounced S in any language in my whole life, and it seems too "soft" to be a S. I noticed in the mirror that, when pronouncing a S, my lips are really stretched (as when pronouncing the vowel I) and my vocal cords (or veins?) are visibly tense on my neck's skin, from the outside. Sep 13, 2019 at 18:38
  • When I pronounce Z, my lips are more relaxed, in the same position of the vowel E, and I see no alteration on the skin of my neck. I am not able to perceive any difference in tongue position, besides the one caused by the stretching of the lips. Would it be possible to guess which phonems I am really pronouncing from this physical description? I suppose one of them is not /s/ or /z/ ? Sep 13, 2019 at 18:42
  • Does that whistle sound like s͎ean conary who has been moked for it in various media? The fact alone that Sean is read "Shawn" hints at a dialectal difference. He Waelsh, from Waelsh if I remember correctly.
    – vectory
    Sep 14, 2019 at 0:23
  • I don't know, I have never noticed any whistle in Sean Connery's speech. Anyway, as I'm from Brazil, we don't share any regional speech. I suspect that this is irrelevant to this discussion. Sep 14, 2019 at 2:10

As far as I know, the symbols [s] and [z] typically refer to consonants with the same place of articulation, in terms of tongue and lip position (that exact place is different in different languages).

They are also both fricatives.

Differences in voicing often go with differences in pitch contours

It is also supposed to be common for there to be difference in the pitch transitions betwen voiced and voiceless consonants and the following vowels (as mentioned in this Language Log article by Mark Liberman: Consonant effects on F0 of following vowels, although it doesn't look at fricatives). I'm not sure how much of that (if any) is an automatic consequence of the difference in voice, but it seems that in some languages different pitches can become phonologized in syllables that originally had voiced or voiceless consonants.

Differences in voicing often go with differences in duration

I've read that in many languages, [s] and [z] tend to have a difference of duration, with [s] being longer than [z]. If I remember correctly, this is the case for the /s/ and /z/ of English, German, and Navajo (in the latter two languages, the phonetic voicing of /z/ may be weak or even absent). But either sound can be prolonged, so I don't think that a difference in duration is generally essential for the contrast.

As Araucaria mentions, in certain contexts, the duration of the preceding vowel is related to the /s/ vs. /z/ contrast in English, and I think I've read that this is also detectable in German, but I don't think this is a universal phenomenon, and I'm not sure how common it is outside of Germanic languages.

The meaning of "fortis" and "lenis"

"Fortis" is a somewhat common term in literature, but I don't know what it means when talking about the phonetic qualities of sounds.* Unlike Greg Lee, I can't directly hear or feel a difference in the amount of force when I pronounce [s] and [z]--the idea of "forceful articulation" feels more like a metaphor to me, like calling palatalized consonants "soft" or velarized consonants "hard".

Since it's come up in both Greg Lee's and Araucaria's answers, I thought I'd do more research on what "fortis" means. The preceding paragraph, which is what I wrote originally in this answer, is based on my inability to perceive this quality through personal introspection, as well as some statements in the Wikipedia article that suggest that it is difficult or controversial to identify which measurable correlates of articulatory force (aside from voicing) are connected to phonological contrasts between consonants.

I found paper with some discussion that helped me a bit. "The Phonetics of Fortis and Lenis Consonants in Itunyoso Trique", by Christian T. DiCanio, has a section titled "Defining 'strength'" (1.2, p. 244-245). DiCanio says

Malécot (1966) attempted to provide some phonetic grounding for differences in articulatory energy between consonants. He concluded that the “intra-buccal air pressure impulse” was the primary correlate of force of articulation, where stops [p] and [t] were produced with greater intra-buccal air pressure than stops [b] and [d]. [Footnote 3: Yet Malécot did not consider that differences in voicing would have significantly contributed to the lower oral pressure and amplitude values that he observed.]

(p. 244)

There are three phonetic correlates which unambiguously reflect a stronger primary articulation. First, one expects stronger articulations to involve greater constriction between two articulators (Fougeron and Keating 1997, Keating et al. 2000, and Lavoie 2001). The only articulation that can be responsible for this is greater muscular tension between the articulators. Second, one expects stops to have louder bursts if they are produced with greater muscular tension. Articulators with more muscular tension will close and release more quickly, causing burst intensity to increase (Debrock 1977 and Kohler 1984). Related to this is the fact that increased stiffness among articulators will result in more abrupt formant transitions into and out of a consonant and faster falling and rising intensity contours (Debrock 1977). Thus, one usually expects consonants with increased articulatory strength to be abruptly timed relative to the surrounding segments. While differences in the constriction degree of consonants are usually measured using articulatory measures (e.g., electropalatography [Lavoie 2001]), both burst amplitude and formant and intensity transitions can be measured acoustically

(p. 245)

I don't know whether differences in any of these articulatory qualities have been found for a [s]/[z] pair in some language. DiCanio says burst intensity is relevant specifically for plosives, but [s] and [z] are fricatives.

I think lower constriction is a fairly common characteristic of the voiced counterparts to non-sibilant voiceless fricatives. E.g., in many languages, voiced "fricatives" like [β], [v], [ð], [ʝ], [ɣ], [ʁ] fairly often can be realized more like approximants than their voiceless counterparts [ɸ], [f], [θ], [ç], [x], [χ], and few languages have a contrast between voiced approximants and voiced fricatives at the same place of articulation: e.g. [v] and [ʋ] or [ɣ] and [ɰ]. But I'm not sure whether this is true of sibilant fricatives: I can't tell whether a less constricted, more approximant-y realization of [z] is possible. Diachronically, [z] can be subject to change into a rhotic approximant (a sound change called rhotacism), but the result of that change is no longer a sibilant consonant.

  • So I assume that you do not see those nuances between the pronunciation of /s/ and /z/ that I have mentioned in Draconis' answer? Sep 13, 2019 at 19:49
  • 2
    @AlanEvangelista: I'm not familiar with the way whistled consonants are articulated. The Wikipedia article makes it look like either [s] or [z] can potentially be whistled, so if someone whistles [s] but not [z], that would be an additional difference. But I don't know if I've ever heard this. Draconis mentioned a pitch element, which is something that I've read about before, and which is worth mentioning. Sep 13, 2019 at 20:01
  • How would this be detectable in German when German has phonological obstruent 'devoicing' in codas? Sep 23, 2019 at 15:36
  • @Araucaria: Two ways. German has a detectable /s/ vs. /z/ contrast intervocalically, as in Weiße "white" vs. Weise "way". And the word-final neutralization of voice in German is supposed to be incomplete in production. Sep 23, 2019 at 21:38
  • Given a maximum onset principle, we would expect intervocalic /z/ to be syllable initial. This would also be expected precisely because of final obstruent 'devoicing'. This gives /z/ the same kind of distribution as /r/in nonrhotic English accents or clear [ l ], both of which are barred from syllable codas. So there's no glimmer of hope for decisive data there. Arguments for phonemic differences on the basis of inaudible phonetic diifferences are not likely to win people round, it seems to me. – Sep 24, 2019 at 8:21

No, voice is not the only difference between /s/ and /z/. /s/ is fortis (forcefully articulated), while /z/ is lenis (weakly articulated). When I do your experiment, saying /s/ and adding voicing, without making other adjustments, I get exactly the result you report -- the /s/ made voiced doesn't sound anything like /z/ -- instead, it's an angry hornet sound.

  • 1
    As the wikipedias on those 2 phonems do not mention this difference of strength, I can only assume that it is not standard among all languages? Those wikis mention that each of those phonems can be executed in different ways: dental-alveolar, non-retracted alveolar and retracted alveolar. I am starting to think that I pronounce the /s/ in the first way, producing the hissing effect, and /z/ in the second neutral way. Sep 13, 2019 at 19:59
  • As I mentioned in Draconis'answer, my lips are more stretched (as when pronouncing the vowel I) when pronouncing /s/ than when pronouncing /z/. I assume that unintentionally brings the tongue closer to the teeth. Sep 13, 2019 at 20:04
  • 1
    Do you mean /s z/ or [s z]?
    – Draconis
    Sep 13, 2019 at 20:45
  • 1
    I mean the latter. Sep 13, 2019 at 23:46
  • I have no information about other articulatory differences between the English fortis-voiceless series /s p t k .../ and the lenis-voiced series /z b d g .../ -- other, that is, than the fortis/lenis and voiceless/voiced differences.
    – Greg Lee
    Sep 14, 2019 at 3:05

the voiced [z] requires a harmonic vibration to create a low frequency oscilation to modulate the voice and is, in my case, liable to fricate with as much of the tongue as possible. The voiceless [s] is just a noisy turbulence; any added voice is not necessarily harmonically modulated and the place of articulation may differ slightly, if only the tip of the tongue touches the alveolar ridge from below, or the dents.

The exact forming of the tongue can differ for channelling the air stream(s) differently in either case and not necessarily in the same manner.

To my surprise, closed teeth aren't needed to produce the sounds, although the vibration between the teeth is needed for me to satisfy the feeling that I fully formed the phoneme. There is a strong prejudice against lisping in German, which might somehow explain that (I think I recall being called to order exactly once), quite different from thSpanish. German also lost th, which Proto-Germanic presumably had.


The presence or absence of vocal fold vibration is the most widely used distinction to notionally differentiate so-called fortis and lenis consonants. However, it is not the only one, and for this reason, the following transcriptions do not indicate the same thing:

  • b̥e̞d̥
  • pe̞t

Fortis (notionally unvoiced) consonants generally differ from lenis ones in the following ways:

  • They require greater exertion of the lungs, resulting in higher intra-oral pressure
  • They are generally of longer duration
  • When occurring in the coda of a syllable, they may have the effect—where a contrast between fortis and lenis pairs is avaliable—of clipping the vocalic elements of the syllable.
  • They may cause devoicing of a following vowel (known as aspiration).

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.