When do you pronounce ‹s› as /z/ in the middle of words? Is there any rule? I also saw there are some differences in articulating medial s between American and British accents. I already know the rule for pronouncing plural s and es.
Is there any case where ‹c› is pronounced /z/?
I couldn't find good references by Googling, and I don't know anything about British english. As I think it through, it is quite complicated! Sorry -- we should really get around to some spelling reform. I hope others can help edit this list if they think of exceptions.
In American English, typically
If there are any prefixes or suffixes causing an s to be in the middle of a word (either because the "s" is part of the prefix or because it is part of the root"), the "s" is always unvoiced /s/, e.g. subsist, substandard, mismatch, mistake, etc.
An s that is written next to an unvoiced consonant is always unvoiced /s/, e.g. lisp, rasp, history, etc.
When the unvoiced consonant of the above rule is [t], then the /t/ is silent if the next syllable is syllabic /n/ or /l/: listen, whistle. (Otherwise it is pronounced. See the comments for a more detailed description of this rule.)
An s before m is always voiced /z/: chasm, prism, plasma. However, the top rule takes precedence, so the s in mismatch is always voiceless /s/. I cannot think of another letter besides [m] where s occurs before or after a voiced consonant word-medially (except by prefixing as in example one), but it would be voiced in such a situation.
An s that is written doubled between vowels is also unvoiced (I guess this is a special case of the above rule): massive, missive, missile, etc. However, if the s would occur in the phonetic stream /sj/ then it assimilates to /ʃ/, e.g. in mission.
An s that is written as one single letter between vowels is usually /z/, e.g. laser, risible. In the same environment as mentioned above /zj/ will assimilate to /ʒ/ e.g. in vision.
Terrible exception to the above: in dessert, the s is voiced to /z/. Many native English speakers misspell dessert for this reason. Note also that the difference between desert and dessert is not voicing, but which syllable gets the accent (it is the first in desert and the second in dessert).
Possess and its derivatives are another exception; the middle "ss" is voiced to /z/. The terminating "ss" is not.
Other miscellaneous exceptions: The -ss- in the American state name Missouri is also exceptionally pronounced /z/. In raspberry, the p is silent and the [s] assimilates to the /b/, so is voiced to /z/.
Some additional guidelines beyond Hunter's:
Based on the immediately surrounding letters:
Word-internal -ns- (including -nse, with "silent e" after the s), is almost always pronounced /ns/ with unvoiced /s/ (e.g. in insist, tense, tinsel). This is different from the pattern for word-final -ns, which is usually pronounced /nz/ (as in pens or lens).
Exceptions:
cleanse ends in /nz/
Some words spelled with -nsy are pronounced with /nzi/, such as pansy, quinsy, tansy. The word teensy seems to be variable: I'm an American, and I've only heard it pronounced with /nsi/, but the OED and Collins indicate that it is pronounced with /nzi/ in British English.
It might make sense to compare these to words ending in /mzi/ and spelled with -msy, such as clumsy, flimsy, whimsy.
For some speakers, certain (but not necessarily all) words starting with trans- such as transit and transition have /nz/.
The place-name Kansas is pronounced with /nz/
intrinsic is most often pronounced with /nz/, but is sometimes pronounced with /ns/
Word-internal -ls-, e.g. in else, pulse, repulsive is almost always pronounced /ls/ with unvoiced /s/. This is different from the pattern for word-final -ls, which is usually pronounced /lz/ (as in eels or steals). Exception: palsy, which has /lz/.
Word-internal -rs-, e.g. in persist, verse, is almost always pronounced /rs/* with unvoiced /s/. (Or in a non-rhotic accent, there is no /r/.) This is different from the pattern for word-final -rs, which is usually pronounced /rz/ (as in stars or yours).
Exceptions:
berserk may be pronounced with /rz/ or with /rs/.
The place-name Jersey is pronounced with /rz/
ersatz, a recent borrowing from German, is often pronounced with /rz/ as in German.
* Speakers of non-rhotic accents would not pronounce a consonant /r/ in these contexts, but would just use a different value for the preceding vowel, e.g. /vɜːs/, /stɑːz/.
Based on identifying particular suffixes:
The ending -sive is usually pronounced /sɪv/ with voiceless /s/, even when there is a vowel letter immediately preceding the letter "s". For example, explosive, invasive, abusive, derisive are all pronounced with /s/. There is at least one known possible exception, but it is optional: divisive (there is also some variation in the pronunciation of the stressed vowel in this word: see the ELU question Differing pronunciations of “divisive”).
The ending -osity is always pronounced with voiceless /s/.
If you include unstressed re-, de-, and pre- as prefixes, then hunter's first rule isn't entirely correct. It is usual for "s" to be pronounced as /z/ when it comes before a vowel and after a fully unstressed version of one of these prefixes. Examples: /z/ is used in resolve, resound, resign, reserve, resume, result, resist, resent; deserve, design, desire, desert; preserve, present, presume, preside.
When these prefixes have any stress (and in newly-formed words, they do tend to have at least secondary stress) then a following "s" is pronounced as voiceless /s/. Examples: /s/ is used in resale, reset, reseat, reseal, resupply, resell; desalt; preset, presell, presoak.
Also, when "s" comes before a letter like "i", "u" or "e", it may represent a postalveolar consonant like /ʃ/ or /ʒ/ because of historical palatalization. Usually, the rules for /ʃ/ vs. /ʒ/ parallel the rules for /s/ vs. /z/, but in a few cases they differ. For example, words spelled with -rsion are usually pronounced with a postalveolar fricative, and in American English they are often pronounced with voiced /ʒ/ in contradiction to the general rule for "rs". (E.g. "immerse" and "immersive" have /rs/, but "immersion" has /rʒ/ for me; I also have /rʒ/ in "Persia(n)". But for some reason, only /rʃ/ is possible for me in "torsion".)
Note that variability between /ʃ/ or /ʒ/ in words spelled with the letter "s" (and sometimes "ss") is fairly common in other contexts too, not just after /r/: e.g. "fission" and "Asia". Usually, the /ʒ/ variant is more common in American English, and the /ʃ/ variant is more common in British English.
I can't think of any word where "c" is pronounced /z/ in a modern accent, but "ti" is usually pronounced /ʒ/ in the word "equation".
Actually, when the letter "s" appears between two vowels, the letter "s" is usually (but not always) pronounced as a /z/.
Some examples include: reason, busy, result, reserve, and laser
There is one problem. For example, a "closer", which is a pitcher from the bullpen, has a /z/ sound.
"Closer", the comparative form of the adjective "close", has an /s/ sound.
