Portuguese phonology

The phonology of Portuguese varies among dialects, in extreme cases leading to some difficulties in mutual intelligibility. This article on phonology focuses on the pronunciations that are generally regarded as standard. Since Portuguese is a pluricentric language, and differences between European Portuguese (EP), Brazilian Portuguese (BP), and Angolan Portuguese (AP) can be considerable, varieties are distinguished whenever necessary.

Consonants
The consonant inventory of Portuguese is fairly conservative. The medieval Galician-Portuguese system of seven sibilants (,, , and apicoalveolar ) is still distinguished in spelling (intervocalic c/ç z, x g/j, ch, ss -s- respectively), but is reduced to the four fricatives by the merger of  into  and apicoalveolar  into either  or  (depending on dialect and syllable position), except in parts of northern Portugal (most notably in the Trás-os-Montes region). These changes are known as deaffrication. Other than this, there have been no other significant changes to the consonant phonemes since Old Portuguese. However, several consonant phonemes have special allophones at syllable boundaries (often varying quite significantly between European and Brazilian Portuguese), and a few also undergo allophonic changes at word boundaries.

Phonetic notes
 * Semivowels contrast with unstressed high vowels in verbal conjugation, as in (eu) rio 'I laugh' and (ele) riu  'he (has) laughed.' Phonologists debate whether their nature is vocalic or consonantal. In intervocalic position semivowels are ambisyllabic, they are associated to both the previous syllable and the following syllable onset.
 * In Brazil and Angola, the consonant hereafter denoted as is realized as a nasal palatal approximant, which nasalizes the vowel that precedes it: ninho ( in Brazil,  in Angola) 'nest'.
 * is often the pronunciation of a sequence of followed by  in a rising diphthong in Brazil, forming a minimal pair between sonha  and Sônia ; menina, "girl".
 * is often the pronunciation of a sequence of followed by  in a rising diphthong in Brazil; e.g. limão, "lemon" ; sandália, "sandal".
 * The consonant hereafter denoted as has a variety of realizations depending on dialect. In Europe, it is typically a voiced uvular fricative . There is also a realization as a voiceless uvular fricative, and the original pronunciation as an alveolar trill  also remains very common in various dialects. A common realization of the word-initial  in the Lisbon accent is a voiced uvular fricative trill . In Brazil,  can be velar, uvular, or glottal and may be voiceless unless between voiced sounds; it is usually pronounced as a voiceless velar fricative , a voiceless glottal fricative  or voiceless uvular fricative . See also Guttural R in Portuguese. All those variants are transcribed with ⟨ʁ⟩ in this article.
 * and are normally, as in English.  However, a number of dialects in northern Portugal pronounce  and  as apico-alveolar sibilants  which are exactly the same as the ones found in Catalan and Northern European Spanish. Those apico-alveolars sound like duller versions of  and , but they are kept apart from  (written ch/x and j/g) which are laminal postalveolar. A small number of northeastern Portugal dialects still retain the medieval distinction between apical and laminal sibilants (written s/ss and c/ç/z, respectively), a distinction also found in Mirandese and analogous to the distinción of European Spanish.
 * As phonemes, and  occur only in loanwords (e.g. tchau and dee jay), with a tendency for speakers to substitute into fricatives in Portugal. However, in most Brazilian dialects d and t are pronounced as  and  before  and.
 * In northern and central Portugal, the voiced stops, , are usually lenited to fricatives , , and  respectively, except at the beginning of words, or after nasal vowels; a similar process occurs in Spanish.
 * In large parts of northern Portugal, e.g. Trás-os-Montes, and also in East Timor and the islands of Flores, and  are merged, both pronounced, as in Spanish.

Consonant elision
There is a variation in the pronunciation of the first consonant of certain clusters, most commonly C or P in cç, ct, pç and pt. These consonants may be variably elided or conserved. For some words, this variation may exist inside a country, sometimes in all of them; for others, the variation is dialectal, with the consonant being always pronounced in one country and always elided in the other. This variation affects 0.5% of the language's vocabulary, or 575 words out of 110,000. In most cases, Brazilians variably conserve the consonant while speakers elsewhere have invariably ceased to pronounce it (for example, detector in Brazil versus detetor in Portugal). The inverse situation is rarer, occurring in words such as fa(c)to and conta(c)to (consonants never pronounced in Brazil, pronounced elsewhere). Until 2009, this reality could not be apprehended from the spelling: while Brazilians did not write consonants that were no longer pronounced, the spelling of the other countries retained them in many words as silent letters, usually when there was still a vestige of their presence in the pronunciation of the preceding vowel. This could give the false impression that European Portuguese was phonologically more conservative in this aspect, when in fact it was Brazilian Portuguese that retained more consonants in pronunciation.

Consonant phonotactics
Syllables have the maximal structure of (C)(C)V(C). The only possible codas in European Portuguese are, and  and in Brazilian Portuguese  and  (or, in a minority of dialects,  or any combination of the former with the latter).


 * The consonants and  almost always occur in the middle of a word and between vowels and rarely occur before.


 * Although nasal consonants do not normally occur at the end of syllables, syllable-final may be present in rare learned words, such as abdómen ( 'abdomen'). In Brazilian varieties, these words have a nasal diphthong (, spelled as abdômen). Word-initial  occurs in very few loanwords.
 * While the sibilant consonants contrast word-initially and intervocalically, they appear in complementary distribution in the syllable coda.  For many dialects (i.e., those of Portugal and of Rio de Janeiro and the northeast of Brazil and certain other areas in Brazil), the sibilant is a postalveolar in coda position (e.g., pasto  'pasture'; futurismo  'futurism'; paz  'peace').  In many other dialects of Brazil (e.g., some of the Southeast, Northeast, and North), the postalveolar variant occurs in some or all cases when directly preceding a consonant, including across word boundaries, but not word-finally (e.g.,, , ). In a number of Brazilian dialects, this "palatalization" is absent entirely (e.g., , , ). Voicing contrast is also neutralized, with  or  occurring before voiced consonants and  or  appearing before voiceless consonants and before a pause (e.g., pasta  or , 'paste'; Islão (or Islã)  or , 'Islam').  In the vast majority of dialects, however, word-final "s" and "z" are pronounced /z/ before vowels (e.g. os ovos , "the eggs", temos hoje , "we have today", faz isso , "do that").  In European dialects, the postalveolar fricatives are only weakly fricated in the syllable coda.
 * The consonant is velarized  in all positions in European Portuguese, even before front vowels. In Portugal, the unvelarized lateral appears only in non-standard dialects. In most Brazilian dialects,  is vocalized to  at the end of syllables, but in the dialects of the extreme south, mainly along the frontiers with other countries (especially Uruguay), it has the full pronunciation or the velarized pronunciation. In some caipira registers, there is a rhotacism of coda  to retroflex . In casual BP, unstressed il can be realized as, as in fácil  ('easy').
 * For speakers who realize as an alveolar trill, the sequence  (as in e.g., os rins) can coalesce into a voiced alveolar fricative trill.
 * Regarding Brazilian Portuguese, analyzes the phonetic clusters at the start of words like qual and guardar as separate phonemes,  and, rather than as  or  followed by  in phonemic sequence. This is because when  or  is combined with a different approximant or when  is combined with a different initial consonant, the relevant approximant can always be alternatively realized as a full vowel in Brazilian Portuguese: quiabo , guiar , suar . However, this alternation can never apply to the phonetic clusters  and : qual and guardar in Brazilian Portuguese are always  and , never * or *.
 * The semivowels and  do not occur before  and  respectively, and only contrast in some diphthongs like in pai  versus pau . Otherwise they are the non-syllabic allophones of  and  in unstressed syllables.
 * Unlike its neighbor and relative Spanish, Brazilian Portuguese lacks a tendency to elide any stop, including those that may become a continuant by lenition ( >, > ,  > ), but it has a number of allophones to it.

Rhotics
The two rhotic phonemes and  contrast only between oral vowels, similar to Spanish. Elsewhere, their occurrence is predictable by context, with dialectal variations in realization. The rhotic is "hard" (i.e., ) in the following circumstances: It is "soft" (i.e., ) when it occurs in syllable onset clusters (e.g., atributo), and written as a single 'r' between vowels (e.g., dirigir 'to drive')
 * Word-initially (e.g., rosa 'rose');
 * Syllable-initially preceded by or  (e.g., guelra 'gill', Israel);
 * Following a nasal vowel (e.g., honrar 'to honor');
 * In most Brazilian and some African dialects, syllable-finally (i.e., preceded but not followed by a vowel);
 * When written with the digraph "rr" (e.g., carro 'car').

The realization of the "hard" rhotic varies significantly across dialects.

This restricted variation has prompted several authors to postulate a single rhotic phoneme. and see the soft as the unmarked realization and that instances of intervocalic  result from gemination and a subsequent deletion rule (i.e., carro  >  > ). Similarly, argue that the hard is the unmarked realization.

Brazilian rhotics
In addition to the phonemic variation between and  between vowels, up to four allophones of the "merged" phoneme /R/ are found in other positions:
 * 1) A "soft" allophone in syllable-onset clusters, as described above;
 * 2) A default "hard" allophone in most other circumstances;
 * 3) In some dialects, a special allophone syllable-finally (i.e., preceded but not followed by a vowel);
 * 4) Commonly in all dialects, deletion of the rhotic word-finally.

The default hard allophone is some sort of voiceless fricative in most dialects, e.g.,, although other variants are also found. For example, an alveolar trill is found in certain conservative dialects down São Paulo, of Italian-speaking, Spanish-speaking, Arabic-speaking, or Slavic-speaking influence. A uvular trill is found in areas of German-speaking, French-speaking, and Portuguese-descended influence throughout coastal Brazil down Espírito Santo, most prominently Rio de Janeiro.

The syllable-final allophone shows the greatest variation:
 * Many dialects (mainly in Brasília, Minas Gerais and Brazilian North and Northeast) use the same voiceless fricative as in the default allophone. This may become voiced before a voiced consonant, esp. in its weaker variants (e.g., dormir  'to sleep').
 * The soft occurs for many speakers in Southern Brazil and São Paulo city.
 * An English-like approximant or vowel (R-colored vowel) occurs elsewhere in São Paulo as well as Mato Grosso do Sul, southern Goiás, central and southern Mato Grosso and bordering regions of Minas Gerais, as well as in the urban areas in the Sinos river valley. This pronunciation is stereotypically associated with the rural "caipira" dialect.

Throughout Brazil, deletion of the word-final rhotic is common, regardless of the "normal" pronunciation of the syllable-final allophone. This pronunciation is particularly common in lower registers, although found in most registers in some areas, e.g., Northeast Brazil, and in the more formal and standard sociolect. It occurs especially in verbs, which always end in R in their infinitive form; in words other than verbs, the deletion is rarer and seems not to occur in monosyllabic non-verb words, such as mar. Evidence of this allophone is often encountered in writing that attempts to approximate the speech of communities with this pronunciation, e.g., the rhymes in the popular poetry (cordel literature) of the Northeast and phonetic spellings (e.g., amá, sofrê in place of amar, sofrer) in Jorge Amado's novels (set in the Northeast) and Gianfrancesco Guarnieri's play Eles não usam black tie (about favela dwellers in Rio de Janeiro).

The soft realization is often maintained across word boundaries in close syntactic contexts (e.g., mar azul 'blue sea').

Vowels
Portuguese has one of the richest vowel phonologies of all Romance languages, having both oral and nasal vowels, diphthongs, and triphthongs. A phonemic distinction is made between close-mid vowels and the open-mid vowels, as in Italian, Catalan and French, though there is a certain amount of vowel alternation. European Portuguese has also two central vowels, one of which tends to be elided like the e caduc of French.

The central closed vowel only occurs in European Portuguese when e is unstressed, e.g. presidente, as well as in Angola; where unlike Portugal, it only occurs in final syllables, e.g. presidente. However, does not exist in Brazil, e.g. presidente.

In Angola, and  merge to, and  appears only in final syllables rama. The nasal becomes open.

Vowel classification
In some cases, Portuguese uses vowel height to contrast stressed syllables with unstressed syllables:
 * In Portugal, unstressed tend to be raised to, whereas  remain unchanged. In final syllables, only  appear.
 * In the Southern, Central-West and Southeast regions of Brazil, unstressed are raised to, whereas in Northern and Northeast regions, they remain open. However, stressed  remain unchanged throughout the whole territory. In final syllables, only  appear.
 * In Angola, unstressed remain unchanged. In final syllables, only  appear.

See below for details. The dialects of Portugal are characterized by reducing vowels to a greater extent than others. Falling diphthongs are composed of a vowel followed by one of the high vowels or ; although rising diphthongs occur in the language as well, they can be interpreted as hiatuses.

European Portuguese possesses quite a wide range of vowel allophones:
 * All vowels are lowered and retracted before :.
 * All vowels are raised and advanced before alveolar, palato-alveolar and palatal consonants: . Those are the vowels shown on the chart. The diphthong is an exception to that as it is often pronounced  in Lisbon, with a back onset.

Furthermore, Cruz-Ferreira gives voiceless allophones of, , in the unstressed word-final position.

The exact realization of the varies somewhat amongst dialects. In Brazil, the vowel can be as high as in any environment. It is typically closer in stressed syllables before intervocalic nasals than word-finally, reaching as open a position as  in the latter case, and open-mid  before nasals, where  can be nasalized. In European Portuguese, the general situation is similar, except that in some regions the two vowels form minimal pairs in some European dialects. In central European Portuguese this contrast occurs in a limited morphological context, namely in verb conjugation between the first person plural present and past perfect indicative forms of verbs such as pensamos ('we think') and pensámos ('we thought'; spelled $⟨pensamos⟩$ in Brazil). Spahr proposes that it is a kind of crasis rather than phonemic distinction of and. It means that in falamos 'we speak' there is the expected prenasal -raising:, while in falámos 'we spoke' there are phonologically two in crasis:  (however, in Brazil both merge, falamos . In Angola, on the other hand, both merge as well, but spelling keeps differentiated: falamos/falámos ). Close-mid vowels and open-mid vowels ( and ) contrast only when they are stressed. In unstressed syllables, they occur in complementary distribution. In Brazilian Portuguese, they are raised to close after a stressed syllable, or in some accents and in general casual speech, also before it.

According to Mateus and d'Andrade (2000:19), in European Portuguese, the stressed only occurs in the following three contexts:
 * Before a palatal consonant (such as telha )
 * Before the palatal front glide (such as lei )
 * Before a nasal consonant (such as cama )

English loanwords containing stressed or  are usually associated with pre-nasal $⟨a⟩$ as in rush,  or are influenced by orthography as in clube (club),  or both, as in surf/surfe.

European Portuguese "e caduc"
European Portuguese possesses a near-close near-back unrounded vowel, transcribed in this article. It occurs in unstressed syllables such as in pegar ('to grip').


 * Traditionally, all instances of are pronounced; e.g. verdade, perigo , estado.
 * In modern European Portuguese, the initial is fronted to ; e.g. energia  →.
 * In traditional EP, was never retracted to . In modern EP, it happens when it is surrounded by, so that ministro , príncipe  and artilhar  are usually pronounced ,  and.
 * When "e" is surrounded by another vowel, it becomes ; e.g. real.
 * However, when the e caduc is preceded by a semi-vowel, it may be given the unreduced pronunciation of the letter $⟨e⟩$, that is or : poesia, quietude.
 * Regardless of the underlying phoneme, a phonetic can be elided, affecting syllabification and sometimes even producing a syllabic consonant; e.g. verdade  →, perigo  → , estado  → , energia  → , ministro  → , príncipe  → , artilhar  → , caminhar  → , pistola  →  (here, ⟨r̩⟩ stands for a syllabic alveolar trill with one contact, the syllabic counterpart of ). This can result in complex syllable onsets that are typical of Slavic languages.
 * Whenever is elided, obstruents in the resulting consonant cluster often agree in voicing, so that the most reduced form of desistiu  '(he) gave up' surfaces as ., a phonological sonorant, behaves like an obstruent in this case and can also be devoiced in voiceless clusters, as in reconhecer  'to recognize' (phonemically ).

There are very few minimal pairs for this sound, including pregar ('to nail') vs. pregar  ('to preach'; the latter stemming from earlier preegar < Latin praedicāre) as well as jure  ('I swear', subjunctive) vs. júri  ('jury').

Oral diphthongs
Diphthongs are not considered independent phonemes in Portuguese, but knowing them can help with spelling and pronunciation.

There are also some words with two vowels occurring next to each other like in iate and sábio may be pronounced both as rising diphthongs or hiatus. In these and other cases, other diphthongs, diphthong-hiatus or hiatus-diphthong combinations might exist depending on speaker, such as or even  for suo ('I sweat'), and in BP  or even  for fatie ('slice it').

and are non-syllabic counterparts of the vowels  and, respectively. At least in European Portuguese, the diphthongs tend to have more central second elements  (as stated above, the starting point of  is typically back) – note that  is also more weakly rounded than the  monophthong.

Nasal vowels
Portuguese also has a series of nasalized vowels. analyzes European Portuguese with five monophthongs and five diphthongs, all phonemic:. Nasal diphthongs occur mostly at the end of words (or followed by a final sibilant), and in a few compounds.

As in French, the nasal consonants represented by the letters ⟨m n⟩ are deleted in coda position, and in that case the preceding vowel becomes phonemically nasal, e.g. in genro ('son-in-law'). But a nasal consonant subsists when it is followed by a plosive, e.g. in cantar ('to sing'). Vowel nasalization has also been observed non-phonemically as result of coarticulation, before heterosyllabic nasal consonants, e.g. in soma ('sum'). Hence, there is a difference between phonemic nasal vowels and those that are allophonically nasalized. Additionally, a nasal monophthong written ⟨ã⟩ exists independently of these processes, e.g. in romã  ('pomegranate'). Brazilian Portuguese is seen as being more nasal than European Portuguese due to the presence of these nasalized vowels. Some linguists consider them to be a result of external influences, including the common language spoken at Brazil's coast at time of discovery, Tupi.

The and  distinction does not happen in nasal vowels; ⟨em om⟩ are pronounced as close-mid. In BP, the vowel (which the letter ⟨a⟩ otherwise represents) is sometimes phonemically raised to  when it is nasal, and also in stressed syllables before heterosyllabic nasal consonants (even if the speaker does not nasalize vowels in this position): compare for instance dama sã  (PT) or  (BR) ('healthy lady') and dá maçã  (PT) or  (BR) ('it gives apples'). may also be raised slightly in word-final unstressed syllables.

Nasalization and height increase noticeably with time during the production of a single nasal vowel in BP in those cases that are written with nasal consonants ⟨m n⟩, so that may be realized as  or. This creates a significant difference between the realizations of ⟨am⟩ and ⟨ã⟩ for some speakers: compare for instance ranço real (PT) or  (BR) ('royal rancidness') and rã surreal  (PT) or  (BR) ('surreal frog'). (Here means a velar nasal approximant.) At the end of a word ⟨em⟩ is always pronounced  with a clear nasal palatal approximant (see below). Whenever a nasal vowel is pronounced with a nasal coda (approximant or occlusive) the (phonetic) nasalization of the vowel itself is optional.

The following examples exhaustively demonstrate the general situation for BP.
 * romã ('pomegranate') : : final vowel is (phonemically) "nasal" and nasal approximants may not be pronounced.
 * genro ('son-in-law') : or  or  : nasal consonant deleted; preceding vowel is (phonemically) "nasal" and nasal approximants may be pronounced.
 * cem ('a hundred') : : nasal approximant must be pronounced.
 * cantar ('to sing') : : nasal consonant remains because of the following plosive; preceding vowel is raised and nasalized non-phonemically. (This is traditionally considered a "nasal" vowel by textbooks.)
 * cano ('pipe') : or  : first vowel is necessarily raised, and may be nasalized non-phonemically.
 * tomo ('I take') : or  : first vowel may be nasalized non-phonemically.

It follows from these observations that the vowels of BP can be described simply in the following way. With this description, the examples from before are simply. But there is no commonly accepted transcription for Brazilian Portuguese phonology.
 * BP has eight monophthongs——whose phonetic realizations may be affected by a nasal archiphoneme . The vowel  is typically nasalized (in every position), but this is not phonemic.
 * All eight vowels are differentiated in stressed and unstressed positions. But in word-final unstressed position and not followed by, they reduce to three vowels——in most dialects. In this position, has a free variation  and this fatally impairs  distinction. (For instance: the word ímã ('magnet') is effectively pronounced as either ima or ímam, depending on speaker.)
 * Like the ん of Japanese, the archiphoneme is a nasal archiphoneme of syllabic codas and its actual place of articulation is determined by the following sound:
 * otherwise it becomes a nasal approximant (as in Japanese kan'i  [かんい], etc.). After the vowels  this approximant may also be pronounced as ; and after  as  (free variations).
 * The system of eight monophthongs reduces to five——before and also in stressed syllables before heterosyllabic nasal consonants. The grapheme ⟨a⟩ stands for  in these cases.
 * is not allowed at word-final position because ⟨em⟩ stands for in this case. (Here  means the same phoneme that ⟨nh⟩ represents; and  may be nasalized non-phonemically.) This is the only case of  in coda-position.
 * otherwise it becomes a nasal approximant (as in Japanese kan'i  [かんい], etc.). After the vowels  this approximant may also be pronounced as ; and after  as  (free variations).
 * The system of eight monophthongs reduces to five——before and also in stressed syllables before heterosyllabic nasal consonants. The grapheme ⟨a⟩ stands for  in these cases.
 * is not allowed at word-final position because ⟨em⟩ stands for in this case. (Here  means the same phoneme that ⟨nh⟩ represents; and  may be nasalized non-phonemically.) This is the only case of  in coda-position.

Vowel nasalization in some dialects of Brazilian Portuguese is very different from that of French, for example. In French, the nasalization extends uniformly through the entire vowel, whereas in the Southern-Southeastern dialects of Brazilian Portuguese, the nasalization begins almost imperceptibly and then becomes stronger toward the end of the vowel. In this respect it is more similar to the nasalization of Hindi-Urdu (see Anusvara). In some cases, the nasal archiphoneme even entails the insertion of a nasal consonant such as (compare ), as in the following examples:


 * banco
 * tempo
 * pinta
 * sombra
 * mundo
 * fã
 * bem
 * vim
 * bom
 * um
 * mãe
 * pão
 * põe
 * muito

Nasal diphthongs
Most times nasal diphthongs occur at the end of the word. They are:


 * -ãe . It occurs in mãe(s) ('mother[s]') and in the plural of some words ending in -ão, e.g., cães ('dogs'), pães ('breads'); and exceptionally non-finally in cãibra ('cramp'). In Central European Portuguese, it occurs also in all words ending in -em, like tem ('he/she/it has'), bem ('well', 'good', as a noun), mentem (they lie), etc.
 * -em . It occurs, both stressed and unstressed, in Brazilian Portuguese and in European Portuguese (both northern and southern) in word-final syllables ending in -em or -ém, like bem, sem, and além, as well as in verbs ending in -em (the 3rd person plural present indicative of verbs whose infinitives end in -er or -ir). In Greater Lisbon, has merged with ; and it occurs duplicated in têm  or  (3rd person plural present indicative of ter, originally tẽem), which in Brazilian is homophonous with tem (the 3rd person singular).
 * -õe . It occurs:
 * in the present indicative of pôr and its derivatives; in the 2nd person singular (pões, opões, compões, pressupões), in the 3rd person singular (põe , opõe etc.), and non-finally in the 3rd person plural (põem , opõem etc.).
 * in the plural of many words ending in-ão, e.g., limões ('lemons'), anões ('dwarfs'), espiões ('spies'), iões ('ions'), catiões ('cations'), aniões ('anions'), eletrões ('electrons'), neutrões ('neutrons'), protões ('protons'), fotões ('photons'), positrões ('positrons') and the plurals of all words with the suffix -ção (compare English -tion, like in communication), like comunicações ('communications'), provocações ('provocations').
 * -uim or -uin Example: pinguim ('penguin').
 * ui occurs only in the words muito  and the uncommon mui . The nasalisation here may be interpreted as allophonic, bleeding over from the previous m (compare mãe with the same bleeding of nasality).
 * -ão or -am. . Examples: pão ('bread'), cão ('dog'), estão ('they are'), vão ('they go'), limão ('lemon'), órgão ('organ'), Estêvão ('Steven'). When in the -am form (unstressed) they are always the 3rd person of the plural of a verb, like estavam ('they were'), contam ('they account'), escreveram ('they wrote'), partiram ('they left').
 * -om . It occurs in word-final syllables ending in -om like bom and som. However, it may be also monophthongized to.

and are nasalized, non-syllabic counterparts of the vowels  and, respectively. In European Portuguese, they are normally not fully close, being closer to. As with the oral, the nasal is not only more central but also more weakly rounded than the  monophthong. This is not transcribed in this article.

Vowel alternation
The stressed relatively open vowels contrast with the stressed relatively close vowels  in several kinds of grammatically meaningful alternation:


 * Between the base form of a noun or adjective and its inflected forms: ovo ('egg'), ovos  ('eggs'); novo, nova , novos , novas  ('new': masculine singular, feminine singular, masculine plural, feminine plural);
 * Between some nouns or adjectives and related verb forms: adj. seco ('dry'), v. seco  ('I dry'); n. gosto  ('taste'), v. gosto  ('I like'); n. governo  ('government') v. governo  ('I govern');
 * Between different forms of some verbs: pôde ('he could'), pode  ('he can');
 * Between some pairs of related words: avô ('grandfather'), avó  ('grandmother');
 * In regular verbs, the stressed vowel is normally low, but high before the nasal consonants , ,  (the high vowels are also nasalized, in BP);
 * Some stem-changing verbs alternate stressed high vowels with stressed low vowels in the present tense, according to a regular pattern: cedo, cedes, cede, cedem ; movo, moves, move, movem (present indicative); ceda, cedas, ceda, cedam ; mova, movas, mova, movam  (present subjunctive). (There is another class of stem-changing verbs which alternate  with  according to the same scheme);
 * In central Portugal, the 1st. person plural of verbs of the 1st. conjugation (with infinitives in -ar) has the stressed vowel in the present indicative, but  in the preterite, cf. pensamos ('we think') with pensámos ('we thought'). In BP, the stressed vowel is  in both, so they are written without accent mark.

There are also pairs of unrelated words that differ in the height of these vowels, such as besta ('beast') and besta  ('crossbow'); mexo  ('I move') and mecho  ('I highlight [hair]'); molho  ('sauce') and molho  ('bunch'); corte  ('cut') and corte  ('court'); meta  ('I put' subjunctive) and meta  ('goal'); and (especially in Portugal) para  ('for') and para  ('he stops'); forma  ('mold') and forma  ('shape').

There are several minimal pairs in which a clitic containing the vowel contrasts with a monosyllabic stressed word containing : da vs. dá, mas vs. más, a vs. à, etc. In BP, however, these words may be pronounced with  in some environments.

Unstressed vowels
Some isolated vowels (meaning those that are neither nasal nor part of a diphthong) tend to change quality in a fairly predictable way when they become unstressed. In the examples below, the stressed syllable of each word is in boldface. The term "final" should be interpreted here as at the end of a word or before word-final -s.

With a few exceptions mentioned in the previous sections, the vowels and  occur in complementary distribution when stressed, the latter before nasal consonants followed by a vowel, and the former elsewhere.

In Brazilian Portuguese, the general pattern in the southern and western accents is that the stressed vowels, , neutralize to , , , respectively, in unstressed syllables, as is common in Romance languages. In final unstressed syllables, however, they are raised to, ,. In casual BP (as well as in the fluminense dialect), unstressed and  may be raised to,  on any unstressed syllable, as long as it has no coda. However, in the dialects of Northeastern Brazilian (as spoken in the states of Bahia and Pernambuco), non-final unstressed vowels are often open-mid, , , independent of vowel harmony with surrounding lower vowels.

European Portuguese has taken this process one step further, raising, , to , ,  in almost all unstressed syllables. The vowels and  are also more centralized than their Brazilian counterparts. The three unstressed vowels are reduced and often voiceless or elided in fast speech.

However, Angolan Portuguese has been more conservative, raising, , to , ,  in unstressed syllables; and to , ,  in final unstressed syllables. Which makes it almost similar to Brazilian Portuguese (except by final, which is inherited from European Portuguese).

There are some exceptions to the rules above. For example, occurs instead of unstressed  or, word-initially or before another vowel in hiatus (teatro, reúne, peão). is often deleted entirely word-initially in the combination becoming. Also,, or  appear in some unstressed syllables in EP, being marked in the lexicon, like espetáculo (spectacle) ; these occur from deletion of the final consonant in a closed syllable and from crasis. And there is some dialectal variation in the unstressed sounds: the northern and eastern accents of BP have low vowels in unstressed syllables,, instead of the high vowels. However, the Brazilian media tends to prefer the southern pronunciation. In any event, the general paradigm is a useful guide for pronunciation and spelling.

Nasal vowels, vowels that belong to falling diphthongs, and the high vowels and  are not affected by this process, nor is the vowel  when written as the digraph $⟨ou⟩$ (pronounced  in conservative EP).

Epenthesis
In BP, an epenthetic vowel is sometimes inserted between consonants, to break up consonant clusters that are not native to Portuguese, in learned words and in borrowings. This also happens at the ends of words after consonants that cannot occur word-finally (e.g.,, , ). For example, psicologia ('psychology') may be pronounced ; adverso ('adverse') may be pronounced ; McDonald's may be pronounced. In northern Portugal, an epenthetic may be used instead,, , but in southern Portugal there is often no epenthesis, ,. Epenthesis at the end of a word does not normally occur in Portugal.

The native Portuguese consonant clusters, where there is not epenthesis, are sequences of a non-sibilant oral consonant followed by the liquids or, and the complex consonants. Some examples: flagrante, complexo , fixo (but not ficção ), latex , quatro , guaxinim ,

Further notes on the oral vowels

 * Some words with in EP have  in BP. This happens when those vowels are stressed before the nasal consonants, , followed by another vowel, in which case both types may occur in European Portuguese, but Brazilian Portuguese for the most part allows only mid or close-mid vowels. This can affect spelling: cf. EP tónico, BP tônico "tonic".
 * In most BP, stressed vowels have nasal allophones,, , , , , etc. (see below) before one of the nasal consonants , , , followed by another vowel. In São Paulo, Southern Brazil, and EP, nasalization is nearly absent in this environment, other than in compounds such as connosco, comummente (spelled conosco, comumente in BP).
 * Most BP speakers also diphthongize stressed vowels in oxytones to, , , , , , etc. (sometimes ), before a sibilant coda (written s or z). For instance, Jesus ('Jesus'), faz  ('he does'), dez  ('ten'). This has led to the use of meia (from meia dúzia 'half a dozen") instead of seis  ('six') when making enumerations, to avoid any confusion with três  ('three') on the telephone.
 * In Greater Lisbon, is pronounced  when it comes before a palatal consonant, ,  or a palato-alveolar , , followed by another vowel; as well as  is pronounced.

Sandhi
When two words belonging to the same phrase are pronounced together, or two morphemes are joined in a word, the last sound in the first may be affected by the first sound of the next (sandhi), either coalescing with it, or becoming shorter (a semivowel), or being deleted. This affects especially the sibilant consonants, , , , and the unstressed final vowels , ,.

Consonant sandhi
As was mentioned above, the dialects of Portuguese can be divided into two groups, according to whether syllable-final sibilants are pronounced as postalveolar consonants, or as alveolar ,. At the end of words, the default pronunciation for a sibilant is voiceless,, but in connected speech the sibilant is treated as though it were within a word (assimilation):


 * If the next word begins with a voiceless consonant, the final sibilant remains voiceless ; bons tempos or  ('good times').
 * If the next word begins with a voiced consonant, the final sibilant becomes voiced as well ; bons dias or  ('good days').
 * If the next word begins with a vowel, the final sibilant is treated as intervocalic, and pronounced ; bons amigos or  ('good friends').

When two identical sibilants appear in sequence within a word, they reduce to a single consonant. For example, nascer, desço, excesso, exsudar are pronounced with by speakers who use alveolar sibilants at the end of syllables, and disjuntor is pronounced with  by speakers who use postalveolars. But if the two sibilants are different they may be pronounced separately, depending on the dialect. Thus, the former speakers will pronounce the last example with, whereas the latter speakers will pronounce the first examples with if they are from Brazil or  if from Portugal (although in relaxed pronunciation one of the siblants may be dropped). This applies also to words that are pronounced together in connected speech:
 * sibilant +, e.g., as sopas: either (most of Brazil);  (Portugal, standard)
 * sibilant +, e.g., as zonas: either (mostly in Brazil);  (Portugal, standard)
 * sibilant +, e.g., as chaves: always ;
 * sibilant +, e.g., os genes: always.

Vowel sandhi
Normally, only the three vowels, (in BP) or  (in EP), and  occur in unstressed final position. If the next word begins with a similar vowel, they merge with it in connected speech, producing a single vowel, possibly long (crasis). Here, "similar" means that nasalization can be disregarded, and that the two central vowels can be identified with each other. Thus,


 * → (henceforth transcribed ); toda a noite  or  ('all night'), nessa altura  or  ('at that point').
 * → ) (henceforth transcribed ); a antiga ('the ancient one') and à antiga ('in the ancient way'), both pronounced or . The open nasalized  appears only in this environment.
 * → (henceforth transcribed ); de idade  or  ('aged').
 * → ; fila de espera ('waiting line') (EP only).
 * → (henceforth transcribed ); todo o dia  or  ('all day').

If the next word begins with a dissimilar vowel, then and  become approximants in Brazilian Portuguese (synaeresis):


 * + V → ; durante o curso ('during the course'), mais que um  ('more than one').
 * + V → ; todo este tempo ('all this time') do objeto  ('of the object').

In careful speech and in with certain function words, or in some phrase stress conditions (see Mateus and d'Andrade, for details), European Portuguese has a similar process:


 * + V → ; se a vires ('if you see her'), mais que um  ('more than one').
 * + V → ; todo este tempo ('all this time'), do objeto  ('of the object').

But in other prosodic conditions, and in relaxed pronunciation, EP simply drops final unstressed and  (elision), though this is subject to significant dialectal variation:
 * durante o curso ('during the course'), este inquilino  ('this tenant').
 * todo este tempo ('all this time'), disto há muito  ('there's a lot of this').

Aside from historical set contractions formed by prepositions plus determiners or pronouns, like à/dà, ao/do, nesse, dele, etc., on one hand and combined clitic pronouns such as mo/ma/mos/mas (it/him/her/them to/for me), and so on, on the other, Portuguese spelling does not reflect vowel sandhi. In poetry, however, an apostrophe may be used to show elision such as in d'água.

Stress
Primary stress may fall on any of the three final syllables of a word, but most commonly on the last two, as antepenultimate stress is relatively less frequent in the language. There is a partial correlation between the position of the stress and the final vowel; for example, the final syllable is usually stressed when it contains a nasal phoneme, a diphthong, or a close vowel—this is especially relevant given the influence of Indigenous languages in Brazil, as indigenous words often have final stress: urubu 'vulture', açaí 'açai', both of which originate from Tupi and have final stress. The orthography of Portuguese takes advantage of this correlation to minimize the number of diacritics, but orthographic rules vary in different regions (e.g., Brazil and Portugal), and should not be used as a reliable guide to stress, despite the existing correlations found in the grapheme-phoneme conversion of Portuguese data.

As in English, stress in verbs and non-verbs is computed differently. For verbs, stress is deterministic, as different morphemes determine the location of stress in a given word based on tense, mood, person and number. For this reason, the few examples one can find of minimal pairs of lexical stress involve either different verb forms for a given stem, such as ruíram /ʁuˈiɾɐ̃w̃/ 'they collapsed' vs. ruirão /ʁuiˈɾɐ̃w̃/ 'they will collapse', or a noun-verb pair, such as dúvida /ˈduvidɐ/ 'doubt' vs. duvida /duˈvidɐ/ 's/he doubts'. For example, regardless of which verb one considers, stress is always final for the first person singular in the future simple tense (indicative): (eu) falarei 'I will speak'. For nouns and adjectives (i.e., non-verbs), stress is mostly affected by phonological factors such as syllable weight, although morphology also plays a role, as different suffixes may affect the location of stress in a given word. In simple terms, the algorithm for stress in non-verbs is similar to that of English and Latin (modulo the extrametricality of the final syllable, which is absent in Portuguese): stress is final if the final syllable is heavy and penultimate otherwise. A heavy syllable contains a diphthong or a coda consonant. All other patterns are considered to be irregular in most approaches in the literature, even though subregularities have been determined and discussed in recent studies. Finally, syllable weight has also been shown to affect the position of secondary stress in the language, whose location can vary in words with an odd number of pre-tonic syllables. In such words, there's some experimental evidence showing that the presence of a given syllable containing a coda leads to a preference for secondary stress in said syllable.

Prosody
Tone is not lexically significant in Portuguese, but phrase- and sentence-level tones are important. As in most Romance languages, interrogation on yes–no questions is expressed mainly by sharply raising the tone at the end of the sentence. An exception to this is the word oi that is subject to meaning changes: an exclamation tone means 'hi/hello', and in an interrogative tone it means 'I didn't understand'.

As for prosodic domains, the presence of metrical feet in Portuguese has been called into question in a recent study, which argues that no strong evidence exists in the language that supports this particular domain (at least Brazilian Portuguese)—despite the presence of stress and secondary stress, both of which are typically (albeit not always) associated with feet.