1

I am a native speaker of Portuguese and for a long time I've wanted to know why some of the rules of the language are the way they are. I am very fond of linguistics and of learning languages, but I am not of the field: it's only an interest of mine. So, there is something I'd like to understand better about Portuguese:

Why does the letter "s" sound like /z/ between vowels? I know the same happens in other Romance languages such as Italian and French, but the opposite happens in Spanish, where "z" sounds like /s/ between vowels (Spanish, in general, sounds a lot more "soft/lose-tongued" then its siblings). I believe there is a name for this phenomenon.

9
  • 2
    The first is one kind of lenition, which is very common in many languages. I don't know about the second, but note that writing is an invented technology, which changes in a quite different way from the organic way in which languages develop.
    – Colin Fine
    Commented May 4, 2023 at 17:20
  • 3
    It’s very common for /s/ to become [z] intervocalically – not just in Romance languages, but throughout the world. The opposite doesn’t actually happen in Spanish, it just sort of looks like that now. Spanish ⟨z⟩ is not the same as Portuguese ⟨z⟩; it’s more comparable to Portuguese ⟨ç⟩. It derives from /ts/, which became /θ/ (still so in most of Spain). In Latin America, /θ/ merged with /s/, so now ⟨z⟩ is always pronounced /θ/ in most of Spain and /s/ in Latin American. But it’s completely unrelated to the intervocalic voicing of /s/ to /z/ that happens in Portuguese. Commented May 4, 2023 at 17:21
  • 1
    Welcome! Will you consider splitting this into two questions? We generally want to handle one topic per question on this site.
    – jogloran
    Commented May 4, 2023 at 18:19
  • 1
    @jogloran, hello, thanks! Probably I'm going to do that :) Commented May 4, 2023 at 18:23
  • 1
    A similar phenomenon exists in English, as well, which leads something very interesting about how foreign names are written, specifically the name "Hussain". We see that generally in other Arabic names being transcribed into English orthography that although double consonants do not relate to a geminate pronunciation in English itself (as gemination does not exist in English with some exceptions), they are commonly used to parallel the geminate Arabic pronunciation when transcribing names, for example in "Ammar" or "Muhammad". Commented May 10, 2023 at 16:04

2 Answers 2

1

In trying to understand the connection between letters and pronunciation, one should always look at the history of the language, which means "looking" at the pre-modern form of the various Romance languages and at Latin, and also understanding the history of the alphabet. Starting with classical Latin, there was a consonant phoneme /s/ but not /z/. The letter <z> is ancient, going back to earliest Semitic writing. It is used in Ancient Greek (ζ) where it was probably pronounced [dz], and was imported into the Etruscan alphabet and sparsely in Latin (it was banned), but ends up being used in Vulgar Latin to represent an affricate. Consequently, the letter z is a problem case across languages, that unlike m which overwhelming represents /m/, z can be used for many different phonemes, and the phoneme /z/ can be spelled many ways (especially <s>).

This is encouraged by the sound changes of various languages, where contemporary [z] historically derives from earlier /s/. There is a tendency for orthographies to be very conservative, so that the s-spelling of the Latin word casa is realized in many Romance languages with [z]. This is due to a process of intervocalic voicing. One would expect /kasa/ to be pronounced [kasa], unless there is a rule in the language changing /s/ to [z] somewhere – between vowels. This is a physically-motivated sound change based on the fact that vowels usually are produced with vibrating vocal folds (voicing), so when /kasa/ becomes [kaza], we can understand that as a physical simplification where /voiced - voiceless - voiced / becomes just [voiced] (maintain the same laryngeal configuration throughout).

1
  • 1
    It also happens in dialects with final [ʃ], which becomes [ʒ] in the same voiced context (vowel [s] vowel/voiced consonant) Commented May 9, 2023 at 11:51
2

This is very common among languages, voiceless sounds (such as [s]) tend to become voiced between other voiced sounds such as voiced consonants and vowels. And that is it what happened historically in Portuguese (asa is pronounced [ˈazɐ] and mesmo is pronounced [ˈmezmʊ]/[ˈmeʒmu]).

However, in Portuguese something called sandhi happens. In this case, voiceless sounds such as [s] and [ʃ] become voiced at the end of a word if the next one starts with a voiced sound!

So "as águas" is [ɐz‿ˈaɡwɐs] or [ɐz‿ ˈaɡwɐʃ], but "os carros" has a voiceless /s/, [ʊs ˈkaχʊs] or [uʃ ˈkaχuʃ].

In each example I gave two "main pronunciations", a somewhat more "general" brazilian and then a more general european/Rio pronunciation to show these final consonants. As you may notice the final /s/ in these latter varieties is a [ʃ] at the end of a syllable (we call it the coda of the syllable) and this sound is no exception, it becomes voiced if the next sound is voiced, it becomes [ʒ] as in "mesmo"

So the thing is: in any dialect of Portuguese, every time these two voiceless sound (in a coda) is between two voiced sounds (in this case a vowel before and a vowel or a voiced consonant after) it will become voiced as well.

4
  • The last paragraph is not correct: voicing does not apply “every time a voiceless sound is between two voiced sounds”. It applies when a voiceless sibilant is between voiced sounds (if intervocalical, more specifically only the morphophoneme |s|, not |ʃ|). Sequences like |apa aka ata aʃa| do not become [abɐ aɡɐ adɐ aʒɐ], for example, and the rare instances of /VCG/ (where C = any unvoiced consonant except |s|) generally do not exhibit voicing either (e.g., un tablet branco doesn’t become [ũ ˈtabɫɨd ˈbrɐ̃ku]). Commented May 9, 2023 at 12:18
  • Yes! You're right, by sound I meant only these two, I will fix it! Thanks! Commented May 9, 2023 at 14:57
  • Actually, part of my comment above was too restrictive, I think: it’s not only intervocalically, but always that voicing only applies when the sound is underlyingly |s|, not |ʃ|. Intervocalic examples are easy to find (achar, paixão, etc.). It’s harder to come up with examples before a voiced consonant since |ʃ| is very rare in coda position, but something like o lavash não… would retain [lɐˈvaʃ], not *[lɐˈvaʒ]. Commented May 9, 2023 at 15:10
  • Yes, in this case I am refering to sindhi and I used the phones [s] and [ʃ] that are very common in coda position to make the explanation easier. I just specified again hahahah. Commented May 9, 2023 at 15:17

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.