A funny thing happens in European Portuguese accents. It has been described in this question. One of the sounds the letter e represents in Portuguese is /e/ (as in vê). However when a stressed e pronounced by Brazilians and many speakers in southern Portugal as /e/ is followed by a post alveolar fricative at the start of the next syllable, /ʃ/ (as in fecho) or /ʒ/ (as in seja) most speakers in Portugal will pronounce it as /ɐj/ (as in the English word bay.)

For instance (follow the audio links to Forvo), fecho is pronounced /’fɐjʃu/ by a Portuguese, but as /’feʃu/ by a Brazilian.

And seja is pronounced /’sɐjʒɐ/ by user Flowerchild from Portugal, but as /’seʒɐ/ by at least two Brazilians. However I’d say user Sirasp from Brazil pronounces it very close to /’sɐjʒɐ/ as well.

And then when the e pronounced by Brazilians and many speakers in southern Portugal as /e/ is followed by the palatal lateral approximant /ʎ/ (as in espelho) most speakers in Portugal will pronounce it as /ɐj/, /ɐ/ or something in between. Hear espelho in Forvo pronounced by people from Portugal and Brazil. But if is followed by the palatal nasal /ɲ/ (as in tenho) it will be pronounced by most speakers in Portugal as /ɐ/. Hear tenho in Forvo.

The stressed e will remain /e/ or /ɛ/ for all speakers when followed by other consonants. Now is there a phonetic reason why an e normally pronounced as /e/ would change into /ɐj/ or /ɐ/ when followed by /ʃ/, /ʒ/, /ɲ/ or /ʎ/, but remain /e/ when followed by other consonants? Do those particular consonant sounds sort of invite or make it easier for the /e/ to turn into /ɐj/ or /ɐ/?

Answer in whatever language you prefer. As long as it is English or Portuguese, that is.

Em Português

Há uma coisa curiosa nos sotaques de Portugal. Foi descrito nesta pergunta. Um dos sons que a letra e representa no português é /e/ (como em vê). Contudo quando um e tónico pronunciado por Brasileiros e muitos falantes no sul de Portugal como /e/ é seguido por uma fricativa pós-alveolar no início da sílaba seguinte, /ʃ/ (as in fecho) or /ʒ/ (as in seja), a maioria dos falantes em Portugal pronunciam-na como /ɐj/ (como na palavra inglesa bay.)

Por exemplo (sigam os links para ouvir as palavras pronunciadas no Forvo) fecho é pronunciado /’fɐjʃu/ por uma portuguesa, mas /’feʃu/] por um brasileiro.

E seja é pronunciado /’sɐjʒɐ/ pela Flowerchild de Portugal, mas /’seʒɐ/ por pelos menos dois brasileiros. Mas parece-me que a brasileira Sirasp pronuncia seja um pouco como /’sɐjʒɐ/ também.

E o e pronunciado por Brasileiros e muitos falantes no sul de Portugal como /e/ é seguido da lateral aproximante palatal /ʎ/ (como em *espelho), a maioria dos falantes em Portugal pronunciam-na /ɐj/, /ɐ/ ou algo no meio. Ouçam espelho no Forvo pronunciado por brasileiros e portugueses. Mas se for seguida da nasal palatal ɲ/ (como em tenho), a maioria dos falantes em Portugal pronunciam-na /ɐ/. Ouçam tenho no Forvo.

O e tónico continua para todos os falantes /e/ ou /ɛ/ quando seguido por outras consoantes. Ora haverá alguma razão fonética para um e normalmente pronunciado /e/ passar a ɐj/ ou /ɐ/ quando seguido de ʃ/, /ʒ/, /ɲ/ or /ʎ/, mas permanecer /e/ se seguido por outras consoantes? Há alguma coisa nestas consoantes particulares que convide ou facilite a mudança do /e/ em /ɐj/ ou /ɐ/?

Responde na língua que preferires. Desde que seja Português ou Inglês.

  • There are several peculiarities in pt-PT pronunciation of certain morphemes. Some people I've met pronounce the first "o" in "todo" as /ou/. or they pronounce "pra ela" as something like /praiela/ or "peixe" as /paish/. Very often the Portuguese omit weak vowel sounds such as first "e" in "competentes". Escuse my not using the IPA (I would if this were an answer).
    – Centaurus
    Commented Jan 7, 2017 at 1:13
  • @Centaurus Yes, but this pronunciation of e as /ɐj/ ou /ɐ/ occurs only before those four consonant sounds; so there must be something special about them.
    – Jacinto
    Commented Jan 7, 2017 at 11:50
  • Great question, but where is English "bay" pronounced similar to [bɐj]? I've always heard English /e/ as [ej] or [ɛj], at least here in northeast USA. (My ears don't distinguish those two.) Could a better example be the vowel in like or nice (as opposed to the [aj] of I or bye)? (My sense is there's still a difference, but that might be closer?)
    – Dan Getz
    Commented Jan 11, 2017 at 1:02
  • @Dan maybe my ears are fooling me. Here's how I perceive it. Brazilians and Portuguese pronounce ei differently. Compare lei pronounced by pathgs and aimae here. I hear English lay as Portugal's lei; but maybe that's UK lay only. I don't here lei as English lie. But Infopédia renders lei as /lɐj/. I've seen it like that elsewhere too. (continues)
    – Jacinto
    Commented Jan 11, 2017 at 8:09
  • But now listening to the vowel table here, [lɐj]doesn't sound like a good representation. Perhaps the best would be [lɜj] (for Portugal), where [ɜ] is as in English /bird/. Then even if ei is [ej] in Brazil and [ɜj] in Portugal, and we tell the difference, they don't make distinguishing minimal pairs. So maybe I could just represent it /ej/. I'd like to give the phonetic representation because, in principle someone who doesn't know Portuguese might still be able to answer this question.
    – Jacinto
    Commented Jan 11, 2017 at 8:20

2 Answers 2


According to the Wikipedia article on Portuguese, in central and southern Portugal /e/ can be centralised before palatal sounds (and by context this includes post-alveolar sounds). That is, /e/ will become [ɐ] before /ʃ/, /ʒ/, /ɲ/, /ʎ/. The article about Portuguese dialects also says that

In and near Lisbon, /ei̯/ and /ẽi̯/ are pronounced [ɐi̯] and [ɐ̃i̯], respectively. Furthermore, stressed /e/ is pronounced [ɐ] or [ɐi̯] before a palato-alveolar or a palatal consonant followed by another vowel.

Consider the well-known fact that many dialects (both in Brazil and Portugal) turn syllable-final [s] to [ʃ]. Also note that in Northeastern Brazil as well as in Rio de Janeiro speakers diphthongize stressed vowels before [ʃ] when word-final (therefore mas [majʃ], fez [fejʃ], nós [nojʃ]) - and in some other contexts too, as I've heard.

Not all phonetic changes have to have a "reasonable" explanation, but in this case, inserting a yod [j], which is a palatal approximant, between a vowel and a palatal (or post-alveolar or alveolo-palatal) consonant, seems rather reasonable.

Combine both tendencies (centralization and diphthongization of stressed vowels in the vicinity of the alveolo-palatal point of articulation) and you have an explanation of why those consonants might trigger /e/ → [ɐj].

  • So the [ j ] is sort of on the way from [ɐ] to palatal, post-alveolar and alveolo-palatal consonants?
    – Jacinto
    Commented Jan 11, 2017 at 22:27
  • 1
    @Jacinto Well, /j/ is palatal, like /ɲ/ and /ʎ/, and /ʃ/ and /ʒ/ are post-alveolar, so they're all pronounced pretty close to each other inside the mouth. So a sequence like [ɐjʒ] (as in seja) means you have central vowel + palatal approximant + post-alveolar fricative. Each sound grades neatly into the next.
    – pablodf76
    Commented Jan 11, 2017 at 22:59
  • 1
    I wonder whether you have any explanation for the centralization of /e/ before /ʃ/, /ʒ/, /ɲ/, /ʎ/. That’s probably the main question. And does your explanation for the insertion of the /j/ requires the previous centralization? Because in Portugal most people centralize all /ej/ into [ɐj] or maybe [ɜj] (see comments to the question above), so I take this as a given; but we centralize /e/ only before /ʃ/, /ʒ/, /ɲ/, /ʎ/. And also before /ɲ/ the /e/ does not get the yod; it becomes just [ɐ] or [ɜ]; to some extent that’s also true before /ʎ/. (continues)
    – Jacinto
    Commented Jan 13, 2017 at 10:58
  • So, if your argument for the insertion of the yod works with [e], then we could have /ʃ/, /ʒ/ and maybe /ʎ/ triggering /e/ → [ej]; and then [ej] → [ɐj], because we do that for all /ej/. That would just leave /e/ → [ɐ], [ɜ] unexplained. I have a more comprehensive description of the pattern /e/ → [ɐj] , [ɐ] in an answer to this question; it is in Portuguese, but I see you understand it well :)
    – Jacinto
    Commented Jan 13, 2017 at 11:07
  • Also, would you like to have a go at this other question of mine; there are a few comments there that sort of make sense to me, but there's no full answer.
    – Jacinto
    Commented Jan 13, 2017 at 11:15

/e/ is a high-mid front vowel: Portuguese Vowel Chart

Notice the position of of your tongue when you pronounce the sound /e/.

Now, notice the position of your tongue when you pronounce the consonants /ʃ/, /ʒ/, /ɲ/ and /ʎ/, and the vowel /i/.

These sounds make your tongue assume a position close to the position it must have when pronouncing the /e/ sound. That's why they are special.

European Portuguese appears to dislike to pronounce a vowel followed by a sound that is to close to that vowel.

Not only the /e/ became an /ɐj/ or /ɐ/ before /ʃ/, /ʒ/, /ɲ/, /ʎ/ and /i/, but the /i/ may also change to an /ɨ/ before these sounds.

Also, notice that the diphthongs /ej/ and /ow/ don't exist anymore in standard European Portuguese.

  • Queres dizer /e/ muda para /ɐj/ ou /ɐj/? Tens /a/. Já me tinha passado pela cabeça, que o /j/ fica ali próximo do /e/ e daquelas consoantes. Já o /i/ passar a /ɨ/ , isso também pode acontecer antes de /z/ (vizinho, dizer). O /u/ também pode passar a /ɨ/ ou desaparecer antes de /ʎ/ (mulher, colher). Isso é capaz de ser o fenómeno mais geral de redução/cancelamento de certas vogais átonas.
    – Jacinto
    Commented Nov 2, 2018 at 18:56
  • Sim, era /ɐj/ ou /ɐ/. No caso de vizinho pode ser explicado por ser seguido de sílaba com outro som i, tal como em ministro, príncipe, militar, civil, etc. Mas não excluo a hipótese de também acontecer antes de /z/ visto que também é feito com a língua debaixo do palato, daí talvez resultar em pronúncias como «dezer». Commented Nov 4, 2018 at 18:27

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.