Vietnamese phonology

• d, gi and r are all pronounced /z/, but r is pronounced [ɹ] in loanwords, for example, cà rốt 'carrot' is pronounced [ka˩ ɹot̚˧˥].
• ch and tr are both pronounced /c/, while x and s are both pronounced /s/.
• Some rural speakers merge /l/ and /n/ into /l/, although this is not considered standard.[1]

• d and gi are both pronounced /j/, but gi is pronounced /z/ in careful speech by some speakers.
• Historically, /v/ is pronounced [j] in common speech, merging with d and gi. However, it is becoming distinct and pronounced as [v], especially in careful speech or when reading a text. In traditional performance including Cải lương, Đờn ca tài tử, Hát bội and some old speakers of Overseas Vietnamese, it is pronounced as consonant cluster [bj], [βj] or [vj].[2] In loanwords, it is always pronounced [v]: va li [vaː˧ lɪi̯˧].
• Historically, a distinction is made between ch /c/ and tr /ʈ/, as well as between x /s/ and s /ʂ/. However, in many speakers, these two pairs are becoming merged as /c/ and /s/ respectively.[3]
• In informal speech, /kw/, /hw/, /ʔw/, and sometimes /ŋw/ are pronounced [w]. However, it is becoming distinct and pronounced as [ɡw], [hw], [w], [ŋw] respectively, especially in formal speech or when reading a text.
• Many speakers pronounce /x/ as [kʰ], and it was pronounced like this in Middle Vietnamese.
• In southern speech, the phoneme /ɹ/, generally represented in Vietnamese linguistics by the letter ⟨r⟩, has a number of variant pronunciations that depend on the speaker. More than one pronunciation may even be found within a single speaker. It may occur as a retroflex fricative [ʐ], an alveolar approximant [ɹ] (unrounded), a flap [ɾ] or a trill [r]. Some rural speakers from Mekong Delta pronounced /ɹ/ as [ɣ] or [j], but this is not considered formal.

The table below summarizes these sound correspondences:

Diaphoneme	Hanoi	Saigon	Example
			word	Hanoi	Saigon
/v/	/v/	/j/ or /v/	vợ   'wife'	[və˨˩ˀ]	[jə˨˧] or [və˨˧]
/z/	/z/	/j/	da   'skin'	[za˧]	[ja˧]
			gia   'to add'
/r/		/r/	ra   'to go out'		[ɹa˧]
/c/	/c/	/c/	chẻ   'split'	[cɛ˧˩]	[cɛ˩˥]
/ʈ/		/ʈ/ or /c/	trẻ   'young'		[ʈɛ˩˥] or [cɛ˩˥]
/s/	/s/	/s/	xinh   'beautiful'	[sɪŋ̟˧]	[sɪ̈n˧]
/ʂ/		/ʂ/ or /s/	sinh   'born'		[ʂɪ̈n˧] or [sɪ̈n˧]

Vowel chart of Hanoi monophthongs according to Kirby (2011:384)

Vowel chart of Hanoi diphthongs according to Kirby (2011:384)

	Front	Central	Back
Close	/i/ ⟨i, y⟩	/ɨ/ ⟨ư⟩	/u/ ⟨u⟩
Close-mid/ Mid	/e/ ⟨ê⟩	/ə/ ⟨ơ⟩ /ə̆/ ⟨â⟩	/o/ ⟨ô⟩
Centering	/iə̯/ ⟨ia~iê⟩	/ɨə̯/ ⟨ưa~ươ⟩	/uə̯/ ⟨ua~uô⟩
Open-mid/ Open	/ɛ/ ⟨e⟩	/a/ ⟨a⟩ /ă/ ⟨ă⟩	/ɔ/ ⟨o⟩

The IPA chart of vowel nuclei above is based on the sounds in Hanoi Vietnamese; other regions may have slightly different inventories. Vowel nuclei consist of monophthongs (simple vowels) and three centering diphthongs.

• All vowels are unrounded except for the four back rounded vowels: /u, o, ɔ, uə̯/.
• In the South, the high vowels /i, ɨ, u/ are all diphthongized in open syllables: [ɪi̯, ɯ̽ɯ̯, ʊu̯], Ba Vì [baː˧ vɪi̯˩] (listen).[4]
• /ə̆/ and /ă/ are pronounced short — shorter than the other vowels.
• While there are small consistent spectral differences between /ə̆/ and /ə/, it has not been established that they are perceptually significant.[5]
• /ɨ/: Many descriptions, such as Thompson,[6] Nguyễn (1970), Nguyễn (1997), consider this vowel to be close back unrounded: [ɯ]. However, Han's[7] instrumental analysis indicates that it is more central than back. Hoang (1965), Brunelle (2003) and Phạm (2006) also transcribe this vowel as central.

In Vietnamese, vowel nuclei are able to combine with offglides /j/ or /w/ to form closing diphthongs and triphthongs. Below is a chart[8] listing the closing sequences of general northern speech.

	/w/ offglide		/j/ offglide
	Front	Central		Back
Centering	/iə̯w/ ⟨iêu⟩	/ɨə̯w/ ⟨ươu⟩	/ɨə̯j/ ⟨ươi⟩	/uə̯j/ ⟨uôi⟩
Close	/iw/ ⟨iu⟩	/ɨw/ ⟨ưu⟩	/ɨj/ ⟨ưi⟩	/uj/ ⟨ui⟩
Close-mid/ Mid	/ew/ ⟨êu⟩	– /ə̆w/ ⟨âu⟩	/əj/ ⟨ơi⟩ /ə̆j/ ⟨ây⟩	/oj/ ⟨ôi⟩
Open-mid/ Open	/ɛw/ ⟨eo⟩	/aw/ ⟨ao⟩ /ăw/ ⟨au⟩	/aj/ ⟨ai⟩ /ăj/ ⟨ay⟩	/ɔj/ ⟨oi⟩

Thompson (1965) says that in Hanoi, words spelled with ưu and ươu are pronounced /iw, iəw/, respectively, whereas other dialects in the Tonkin delta pronounce them as /ɨw/ and /ɨəw/. This observation is also made by Phạm (2008) and Kirby (2011).

The pronunciation of syllable-final ch and nh in Hanoi Vietnamese has had different analyses. One analysis, that of Thompson (1965) has them as being phonemes /c, ɲ/, where /c/ contrasts with both syllable-final t /t/ and c /k/ and /ɲ/ contrasts with syllable-final n /n/ and ng /ŋ/. Final /c, ɲ/ is, then, identified with syllable-initial /c, ɲ/.

Another analysis has final ⟨ch⟩ and ⟨nh⟩ as representing different spellings of the velar phonemes /k/ and /ŋ/ that occur after upper front vowels /i/ (orthographic ⟨i⟩) and /e/ (orthographic ⟨ê⟩). This analysis interprets orthographic ⟨ach⟩ and ⟨anh⟩ as an underlying /ɛ/, which becomes phonetically open and diphthongized: /ɛk/ → [ăjk̟̚], /ɛŋ/ → [ăjŋ̟].[9] This diphthongization also affects ⟨êch⟩ and ⟨ênh⟩: /ek/ → [ə̆jk̟̚], /eŋ/ → [ə̆jŋ̟].

Arguments for the second analysis include the limited distribution of final [c] and [ɲ], the gap in the distribution of [k] and [ŋ] which do not occur after [i] and [e], the pronunciation of ⟨ach⟩ and ⟨anh⟩ as [ɛc] and [ɛɲ] in certain conservative central dialects,[10] and the patterning of [k]~[c] and [ŋ]~[ɲ] in certain reduplicated words. Additionally, final [c] is not articulated as far forward as the initial [c]: [c] and [ɲ] are pre-velar [k̟, ŋ̟] with no alveolar contact.[11]

The first analysis closely follows the surface pronunciation of a slightly different Hanoi dialect than the second. In this dialect, the /a/ in /ac/ and /aɲ/ is not diphthongized but is actually articulated more forward, approaching a front vowel [æ]. This results in a three-way contrast between the rimes ăn [æ̈n] vs. anh [æ̈ɲ] vs. ăng [æ̈ŋ]. For this reason, a separate phonemic /ɲ/ is posited.

The following rimes ending with velar consonants have been diphthongized in the Hanoi dialect, but /i/, /u/ and /ɨ/ are more open:[10]

ong, oc	/awŋ/, /awk/	→	[ăwŋ͡m], [ăwk͡p̚]
ông, ôc	/əwŋ/, /əwk/	→	[ə̆wŋ͡m], [ə̆wk͡p̚]
ung, uc	/uŋ/, /uk/	→	[ʊŋ͡m], [ʊk͡p̚]
ưng, ưc, ưn, ưt	/ɨŋ/, /ɨk/, /ɨn/, /ɨt/	→	[ɯ̽ŋ], [ɯ̽k̟̚], [ɯ̽n], [ɯ̽t̚]
anh, ach	/ɛŋ/, /ɛk/	→	[ăjŋ̟], [ăjk̟̚]
ênh, êch	/eŋ/, /ek/	→	[ə̆jŋ̟], [ə̆jk̟̚]
inh, ich	/iŋ/, /ik/	→	[ɪŋ̟], [ɪk̟̚]

With the above phonemic analyses, the following is a table of rimes ending in /n, t, ŋ, k/ in the Hanoi dialect:

While the variety of Vietnamese spoken in Hanoi has retained finals faithfully from Middle Vietnamese, the variety spoken in Saigon has drastically changed its finals. Rimes ending in /k, ŋ/ merged with those ending in /t, n/, respectively, so they are always pronounced /t, n/, respectively, after the short front vowels /i, e, a/ (only when /a/ is before "nh"). However, they are always pronounced /k, ŋ/ after the other vowels /u, o, ɔ, iː, ɨ, aw, a, aː, ɛ, ə, əː/. After rounded vowels /aw, u, o/, many speakers close their lips, i.e. they pronounce /k, ŋ/ as [k͡p, ŋ͡m].[10] Subsequently, vowels of rimes ending in labiovelars have been diphthongized, while vowels of rimes ending in alveolar have been centralized.[12] Otherwise, some Southern speakers distinguish /k, ŋ/ and /t, n/ after /u, o, ɔ, iː, ɨ, aw, a, aː, ɛ, ə, əː/ in formal speech, but there are no Southern speakers who pronounce "ch" and "nh" at the end of syllables as /k, ŋ/.

The short back vowels in the rimes have been diphthongized and centralized, meanwhile, the consonants have been labialized. Similarly, the short front vowels have been centralized which are realized as central vowels /ă, ə, ɨ/ and the "unspecified" consonants have been affected by Coronal Spreading from the preceding front vowels which are surfaced as coronals (alveolar) /n, t/.[10]

ung, uc	/uŋ/, /uk/	→	[ʊwŋ͡m], [ʊwk͡p̚]
ông, ôc	/oŋ/, /ok/	→	[ăwŋ͡m], [ăwk͡p̚]
ong, oc	/ɔŋ/, /ɔk/	→
anh, ach	/an/, /at/	→	[ăn], [ăt̚]
ênh, êch	/en/, /et/	→	[ɤn], [ɤt̚]
in ~ inh, it ~ ich	/in/, /it/	→	[ɪ̈n], [ɪ̈t̚]
um, up	/um/, /up/	→	[ʊm], [ʊp̚]
ưng ~ ưn, ưc ~ ưt	/ɨŋ/, /ɨk/	→	[ɯ̽ŋ], [ɯ̽k̟̚]

The other closed dialects (Hue, Quang Nam, Binh Dinh) which have also been merged in codas, but some vowels are pronounced differently in some dialects:

The ông, ôc rimes are merged into ong, oc as [ăwŋ͡m], [ăwk͡p̚] in many Southern speakers, but not with ôn, ôt as pronounced [oːŋ͡m], [oːk͡p̚]. The oong, ooc and eng, ec rimes are few and are mostly loanwords or onomatopoeia. The ôông, ôôc (oong, ooc, eng, ec, êng, êc as well) rimes are the "archaric" form before become ông, ôc by diphthongization and still exist in North Central dialect in many placenames. The articulation of these rimes in North Central dialect are [oːŋ], [oːk̚] without a simultaneous bilabial closure or labialization.[16]

on, ot	/ɔn/, /ɔt/	→	[ɔːŋ], [ɔːk]
oong, ooc	/ɔŋ/, /ɔk/	→
ôn, ôt	/on/, /ot/	→	[oːŋ͡m], [oːk͡p̚].
ôông, ôôc	/oŋ/, /ok/	→
ong, oc	/awŋ/, /awk/	→	[ăwŋ͡m], [ăwk͡p̚]
ông, ôc	/əwŋ/, /əwk/	→

With the above phonemic analyses, the following is a table of rimes ending in /n, t, ŋ, k, ŋ͡m, k͡p/ in the Saigon dialect:

There is much variation among speakers concerning how tone is realized phonetically. There are differences between varieties of Vietnamese spoken in the major geographic areas (northern, central, southern) and smaller differences within the major areas (e.g. Hanoi vs. other northern varieties). In addition, there seems to be variation among individuals. More research is needed to determine the remaining details of tone realization and the variation among speakers.

The six tones in the Hanoi and other northern varieties are:

Tone name	Tone ID	Description	Chao Tone Contour	Diacritic	Example
ngang "flat"	A1	mid level	˧ (33)	◌	ba ('three')
huyền "deep"	A2	low falling (breathy)	˨˩ (21) or (31)	◌̀	bà ('grandmother')
sắc "sharp"	B1	mid rising, tense	˧˥ (35)	◌́	bá ('aunt')
nặng "heavy"	B2	mid falling, glottalized, heavy	˧ˀ˨ʔ (3ˀ2ʔ) or ˧ˀ˩ʔ (3ˀ1ʔ)	़	bạ ('to strengthen')
hỏi "asking"	C1	mid falling(-rising), emphasis	˧˩˧ (313) or (323) or (31)	◌̉	bả ('bait')
ngã "tumbling"	C2	mid rising, glottalized	˧ˀ˥ (3ˀ5) or (4ˀ5)	◌̃	bã ('residue')

Northern Vietnamese (non-Hanoi) tones as uttered by a male speaker in isolation. From Nguyễn & Edmondson (1998)

Hanoi tones as uttered by a female speaker in isolation. From Nguyễn & Edmondson (1998)

Hanoi tones as uttered by a different female speaker in isolation. From Nguyễn & Edmondson (1998)

Ngang tone:

• The ngang tone is level at around the mid level (33) and is produced with modal voice phonation (i.e. with "normal" phonation). Alexandre de Rhodes (1651) describes this as "level"; Nguyễn (1997) describes it as "high (or mid) level".

Huyền tone:

• The huyền tone starts low-mid and falls (21). Some Hanoi speakers start at a somewhat higher point (31). It is sometimes accompanied by breathy voice (or lax) phonation in some speakers, but this is lacking in other speakers: bà = [ɓa˨˩].[18] Alexandre de Rhodes (1651) describes this as "grave-lowering"; Nguyễn (1997) describes it as "low falling".

Hỏi tone:

• The hỏi tone starts a mid level and falls. It starts with modal voice phonation, which moves increasingly toward tense voice with accompanying harsh voice (although the harsh voice seems to vary according to speaker). In Hanoi, the tone is mid falling (31). In other northern speakers, the tone is mid falling and then rises back to the mid level (313 or 323). This characteristic gives this tone its traditional description as "dipping". However, the falling-rising contour is most obvious in citation forms or when syllable-final; in other positions and when in fast speech, the rising contour is negligible. The hỏi also is relatively short compared with the other tones, but not as short as the nặng tone. Alexandre de Rhodes (1651) describes this as "smooth-rising"; Nguyễn (1997) describes it as "dipping-rising".

Ngã tone:

• The ngã tone is mid rising (35). Many speakers begin the vowel with modal voice, followed by strong creaky voice starting toward the middle of the vowel, which is then lessening as the end of the syllable is approached. Some speakers with more dramatic glottalization have a glottal stop closure in the middle of the vowel (i.e. as [VʔV]). In Hanoi Vietnamese, the tone starts at a higher pitch (45) than other northern speakers. Alexandre de Rhodes (1651) describes this as "chesty-raised"; Nguyễn (1997) describes it as "creaking-rising".

Sắc tone:

• The sắc tone starts as mid and then rises (35) in much the same way as the ngã tone. It is accompanied by tense voice phonation throughout the duration of the vowel. In some Hanoi speakers, the ngã tone is noticeably higher than the sắc tone, for example: sắc = ˧˦ (34); ngã = ˦ˀ˥ (45). Alexandre de Rhodes (1651) describes this as "acute-angry"; Nguyễn (1997) describes it as "high (or mid) rising".

Nặng tone:

• The nặng tone starts mid or low-mid and rapidly falls in pitch (32 or 21). It starts with tense voice that becomes increasingly tense until the vowel ends in a glottal stop closure. This tone is noticeably shorter than the other tones. Alexandre de Rhodes (1651) describes this as "chesty-heavy"; Nguyễn (1997) describes it as "constricted".

The Southern tones contour of ngang, sắc, huyền is similar as Northern tones, however, these tones are produced with normal voice instead of breathy voice.

The nặng tone are pronounced as low rising tone (12) [˩˨] in fast speech or low falling-rising tone (212) [˨˩˨] in more careful utterance.

The ngã and hỏi tone are merged into a mid falling-rising (214) [˨˩˦] which is somewhat similar hỏi tone of non-Hanoi Northern accent mentioned above.

Southern Vietnamese tone system from female native speaker. From Jessica Bauman et al. (2009)[21]

North-central and Central Vietnamese varieties are fairly similar with respect to tone although within the North-central dialect region there is considerable internal variation.

It is sometimes said (by people from other provinces) that people from Nghệ An pronounce every tone as a nặng tone.

An older analysis assumes eight tones rather than six.[22] This follows the lead of traditional Chinese phonology. In Middle Chinese, syllables ending in a vowel or nasal allowed for three tonal distinctions, but syllables ending with /p/, /t/ or /k/ had no tonal distinctions. Rather, they were consistently pronounced with a short high tone, which was called the entering tone and considered a fourth tone. Similar considerations lead to the identification of two additional tones in Vietnamese for syllables ending in /p/, /t/, /c/ and /k/. These are not phonemically distinct from the sắc and nặng tones, however, and hence not considered as separate tones by modern linguists and are not distinguished in the orthography.