'Citisen'

[PVER•PVERVM•AMAT]

Where does the non-standard pronunciation of 'citizen' with /s/ in /z/'s stead come from, and how did it become so widespread? It isn't a spelling pronunciation, not even of an archaic spelling; the word was formerly 'citizein', but 'denizen' influenced its spelling. Even so, no one says 'denizen' with /s/.

[Matrix]

Maybe they're assimilating it with the voicless /c/ [s] and /t/?

[PVER•PVERVM•AMAT]

The <c> in 'citizen' does not represent a voiceless palatal stop that allophonically becomes a voiceless alveolar sibilant fricative. Both the phoneme and its realization should be transcribed as a plain s (perhaps with diacritics for its realization).

[CaesarVincens]

I'd agree with Matrix that the most likely source is assimilation toward the preceding voiceless phones. Voicing in many English dialects is usually fairly weak anyway.

Edit:
I might even transcribe the word as [ˈsɪtɪz̥ən] with a devoiced /z/. Similar to words like 'hands' [hændz̥] with final devoicing.

[Nortaneous]

It would have to be old then, since AmE merges t d / V_V as [4], and the distinction isn't maintained except after /ai/.

[CaesarVincens]

It would have to be old then, since AmE merges t d / V_V as [4], and the distinction isn't maintained except after /ai/.

That's true, but since [ɾ] isn't phonemic in English, I think the average American English speaker isn't aware that they aren't saying a /t/. But that could be a wrinkle the voicing assimilation hypothesis.

[CatDoom]

That's true, but since [ɾ] isn't phonemic in English, I think the average American English speaker isn't aware that they aren't saying a /t/. But that could be a wrinkle the voicing assimilation hypothesis.

I know I wasn't... I still have a lot of trouble pronouncing [ɾ] in other positions; I usually just end up saying [d].