In phonology, voicing (or sonorization) is a sound change where a voiceless consonant becomes voiced due to the influence of its phonological environment; shift in the opposite direction is referred to as devoicing or desonorization. Most commonly, the change is a result of sound assimilation with an adjacent sound of opposite voicing, but it can also occur word-finally or in contact with a specific vowel.
For example, the English suffix -s is pronounced [s] when it follows a voiceless phoneme (cats), and [z] when it follows a voiced phoneme (dogs). This type of assimilation is called progressive, where the second consonant assimilates to the first; regressive assimilation goes in the opposite direction, as can be seen in have to [hæft?].
English no longer has a productive process of voicing stem-final fricatives when forming noun-verb pairs or plural nouns, but there are still examples of voicing from earlier in the history of English:
The voicing alternation found in plural formation is losing ground in the modern language,. Of the alternations listed below many speakers retain only the [f-v] pattern, which is supported by the orthography. This voicing of /f/ is a relic of Old English, at a time when the unvoiced consonants between voiced vowels were 'colored' by an allophonic voicing (lenition) rule /f/ -> [v]. As the language became more analytic and less inflectional, final vowels or syllables stopped being pronounced. For example, modern knives is a one syllable word instead of a two syllable word, with the vowel e not pronounced and no longer part of the word's structure. The voicing alternation between [f] and [v] occurs now as realizations of separate phonemes /f/ and /v/. The alternation pattern is well maintained for the items listed immediately below, but its loss as a productive allophonic rule permits its abandonment for new usages of even well-established terms: while leaf~leaves in reference to 'outgrowth of plant stem' remains vigorous, the Toronto ice hockey team is uncontroversially named the Maple Leafs.
Sonorants (/l r w j/) following aspirated fortis plosives (that is, /p t k/ in the onsets of stressed syllables unless preceded by /s/) are devoiced such as in please, crack, twin, and pewter.
Several varieties of English have a productive synchronic rule of /t/-voicing whereby intervocalic /t/ not followed by a stressed vowel is realized as voiced alveolar flap [?], as in tutor, with the first /t/ pronounced as voiceless aspirated [t?] and the second as voiced [?]. Voiced phoneme /d/ can also emerge as [?], so that tutor and Tudor may be homophones, both with [?] (the voiceless identity of word-internal /t/ in tutor is manifested in tutorial, where stress shift assures [t?]).
In many languages including Polish and Russian, there is anticipatory assimilation of unvoiced obstruents immediately before voiced obstruents. For example, Russian 'request' is pronounced /'proz?b?/ (instead of */'pros?b?/) and Polish pro?ba 'request' is pronounced /'prba/ (instead of */'prba/). This process can cross word boundaries as well, for example Russian ? /'dod b?/ 'daughter would'. The opposite type of anticipatory assimilation happens to voiced obstruents before unvoiced ones: ?? /?p'sp?t?/.
In Italian, /s/ before a voiced consonant is pronounced [z] within any phonological word: sbaglio ['zbao] 'mistake', slitta ['zlitta] 'sled', snello ['zn?llo] 'slender'. The rule applies across morpheme boundaries, e.g. disdire [diz'di:re] 'cancel', but not word boundaries: lapis nero [?la:pis'ne:ro] 'black pencil'. This voicing is productive, thus it applies to borrowings as well as native lexicon: snob [zn?b], slinky (toy) ['zli?ki].
Final devoicing is a systematic phonological process occurring in languages such as German, Dutch, Polish, and Russian, among others. In these languages, voiced obstruents in the syllable coda or at the end of a word become voiceless.
Initial voicing is a process of historical sound change where voiceless consonants become voiced at the beginning of a word. For example, modern German sagen ['za:?n?], Yiddish ['zn?], and Dutch zeggen ['z] (all "say") all begin with [z], which derives from [s] in an earlier stage of Germanic, as still attested in English say, Swedish säga ['s?j:a], and Icelandic segja ['sei:ja]. Some English dialects were affected by this as well, but it is rare in Modern English. One example is fox (with the original consonant) compared to vixen (with a voiced consonant).