Synchronic evidence for historical hypothesis, vietnamese palatals

SYNCHRONIC EVIDENCE FOR HISTORICAL HYPOTHESIS –
VIETNAMESE PALATALS 
Andʃea Hòa Phạm
University of Florida
Researchers report that synchronic and diachronic changes parallel each other in various
languages. For example, labialized velars in Indo-European become labials in Classical
Greek, labialized velars in Proto-Bantu become labials in West Teke, and labialized
velars in Proto-Zapotec become labials in Isthmus Zapotec (Ohala 1992). The final
palatal and alveolar consonants in various dialects of Vietnamese are another interesting
case of the diachronic-synchronic relationship in sound change and of sound change in
pʃogʃess since we can document the Pʃoto Việt-M ng palatals and alveolaʃs aʃe
gradually being lost in the coda position as they merge with velars.
In the literature on Vietnamese linguistics palatals are said to be allophones of velars
in Northern dialects, based on various synchronic evidence, such as patterns in
reduplication, symmetrical distribution of final consonants, and mutual phonetic effects
between the vowel and final consonant. However, new evidence shows the gradual loss
of palatals and alveolars in various dialects of Vietnamese. Data from under-examined
dialects not only further support the hypothesis that palatals once occurred distinctively in
Vietnamese but also pʃovide the ‘missing link’ in synchʃonic accounts that argues for
palatals being allophones of velars.
Noʃtheʃn Vietnamese dialects, ʃegaʃded as standaʃd, include the speech of the Hà N i
city and eight surrounding provinces in the Red River Delta (Map 1).



Paper given at the 39th annual conference of the Linguistic Association of Canada and United States, York

University, Toronto, August 8 – 11, 2012, to appear in LACUS, Vol. 39. I would like to thank two
anonymous reviewers, and to William Sullivan and Marc Brunelle for useful comments. My gratitude also
goes to the consultants in my 2012 field trip in Vietnam. Any shortcomings are mine.

1


Map 1. Regions of Vietnam (source: Wikimedia Commons).
Phonetically, northern dialects have five places of articulation in the final position:
labial, (apico-) alveolar, palatal, labio-velar, and plain velar. Labio-velars occur after
back rounded vowels and are in complementary distribution with plain velars. Palatals
occur only after [i], [e], and [a],1 and are in complementary distribution with velars,
except after [a] palatals contrast with velars. Moreover, neither palatals nor velars follow
the shoʃt vowel [ ]. This situation cʃeates two asymmetʃical distʃibutions shown in Table
1: i) although palatals occuʃ afteʃ /i/ and /e/, they do not occuʃ afteʃ [ ], and ii) [a] is the
only vowel that occurs before both palatals and velars.

i
e
a

c, ɲ
+
+
+

k, ŋ
+

Table 1. Distribution of final palatals after vowels in Northern dialects.
1

In this paper when the phonemic status of a sound is not yet clear, its IPA symbol has
brackets around until after its phonemic status is discussed. Examples are phonetically

transcribed and placed in the brackets. Tones are omitted.
2


Examples are given in (1).
(1) lính
chênh
anh
măng

[liɲ]
[ceɲ]
[ʔaɲ]
[maŋ]

‘soldieʃ’
‘tilted/diffeʃent’
‘eldeʃ bʃotheʃ’
‘bamboo shoot’

In 1967, Thompson published an article on the history of Vietnamese palatals. Based on
compaʃisons of Vietnamese with the M ng language, the closest sisteʃ of Vietnamese in
the Việt-M ng subfamily, and of dialects spoken in North Vietnam (the Red River
Delta) with those spoken in South Vietnam (Southeast and Mekong River Delta), he
suggested that the oʃiginal palatals fʃom the Pʃoto Việt-M ng weʃe pʃeseʃved in many
M ng dialects, but in Vietnamese only afteʃ /i/, /e/, and /a/. Elsewhere palatals
developed to alveolars in much the same way that the original palatals developed to
alveolars after /i/, /e/, and /a/ in southern Vietnamese dialects while original alveolars
became velars elsewhere. Contrastive palatals of ancient Vietnamese, if there were any,
aʃe not documented. Thompson’s suggestion makes a few pʃedictions: fiʃst, if palatals

used to occur after most vowels in Vietnamese, it is unlikely that they disappeared all
together and at the same time in all dialects. Second, in southern dialects, although
alveolars merge with velars after certain vowels, alveolars might still occur and contrast
with other consonants.
Data from understudied dialects spoken in South Central Coast of Vietnam confirm
these predictions and offer a picture of sound change in progress. Field work in Vietnam
in 2012 revealed a fascinating pattern of occurrence of the final palatals in the speech of
Hồi Nhơn Distʃict in the Bình Định pʃovince, and the speech of Hòa Quang Bắc village
in the Phú Yên pʃovince. In the Bình Định and Phú Yên dialects, palatals aʃe found afteʃ
the shoʃt vowel [ ]. In the Quảng Nam and Bình Định dialects palatals occuʃ afteʃ /i/ and
/e/, and contrast with alveolars; elsewhere alveolars merge with velars. Such data clearly
show that the historical palatal is disappearing.
This paper makes the following claims: first, there are historical palatal consonants in
the Vietnamese coda, and they are disappearing. Second, historical alveolar consonants
are also disappearing, merging with velars syllable-finally in Central and Southern
dialects. Finally, mass immigration from the area of north central to central and south
central Vietnam could be a factor in the change.
Section 1 presents a discussion of the phonemic status of final palatals. Section 2
shows sound change in progress: the historical palatals and alveolars are disappearing.
ɤection 3 shows that the ‘hypothetical’ foʃm [ ɲ] in Noʃtheʃn dialects is indeed found in
the speech of Bình Định and Phú Yên pʃovinces. Section 4, using patterns of
immigration, explains the old linguistic features found in the new land. Section 5 is the
conclusion.
1. The Phonemic Status of Final Palatals and Their Historical Development
This section presents background information about the debate and discusses the
historical view of the development of final palatals in Vietnamese.

3

Table 2 shows the phonemic vowel inventoʃy of the Hà N i dialect. In this papeʃ, Hà
N i, the standaʃd dialect, ʃepʃesents Noʃtheʃn dialects spoken in Hà N i and suʃʃounding
provinces in the Red River Delta. There are eleven vowels and two diphthongs. Two
vowels contʃast in length: /a/ and /ɘ/. Long vowels can occuʃ in eitheʃ open oʃ closed
syllables. Short vowels occur only in closed syllables. All back vowels are rounded.
Three diphthongs can occupy the nucleus.

high
mid
low
diphthong

Front

Central

Back

i
e

ɨ

u
o
ɔ
uə

ɘ
a

iə

ɨə

ɘː
aː

Table 2. The phonemic vowel inventoʃy of the Hà N i dialect.
Eight consonants, [p, t, c, k͡p, m, n, k], and [ŋ͡m], and two glides, [w] and [j], can occuʃ
syllable-finally. Palatals occur only after [i], [e], and [a], and contʃast with velaʃs afteʃ the
shoʃt [a]. Plain velaʃs occuʃ elsewheʃe, except they contʃast with palatals afteʃ the shoʃt
[a]. Labio-velaʃs, [k͡p] and [ŋ͡m], occuʃ only afteʃ back ʃounded, shoʃt vowels. It is
important to note that vowels preceding palatals and labio-velars are always short. Table
3 shows the phonetic distribution of palatals and velars. Labial place of articulation is
omitted. In Table 3 only alveolaʃs occuʃ afteʃ the shoʃt vowel [ ], e.g., đen [d n] ‘black’,
and only plain velaʃs occuʃ afteʃ the long vowel [ :], e.g., xẻng [s ːŋ] ‘shovel’. Because
theʃe is no length contʃast in [ ], the long [ :] is not ʃegaʃded as phonemic in the
literature.

i
e
:
ɨ
ɘː
ɘ
a:
a
u
o

ɔ
iə
ɨə
uə

t/n
+
+
+
+
+
+
+
+
+
+
+
+
+
+

k͡p / ŋ͡m
+
+
+
-

c/ ɲ
+
+

+
-

k/ŋ
+
+
+
+
+
+
+
+
+

Table 3. Phonetic distʃibution of alveolaʃs, palatals and velaʃs in the Hà N i dialect.
4


Table 1 and Table 3 show the two asymmetrical patterns: a) [a] is the only vowel that
contʃasts palatals and velaʃs, and b) neitheʃ palatals noʃ velaʃs follow the shoʃt vowel [ ].
The asymmetrical pattern in the distribution of palatals created a long debate about
the underlying forms of final palatals and the relevant vowels: should the final palatal be
included in the final inventory or it is an allophone of a velar? The absence of palatals
afteʃ [ ] raises the issue of whether anh [ʔaɲ] ‘eldeʃ bʃotheʃ’ is phonemically /ʔaɲ/ oʃ
/ʔ ŋ/. Theʃe aʃe two major views with some variations. The first view argues that palatals
are phonemic and should be included in the final inventory, i.e., anh [ʔaɲ] is /ʔaɲ/. The
second view argues that palatals are allophones of velars in the final position, i.e., anh
[ʔaɲ] is /ʔ ŋ/. Each view affects the numbeʃ of vowels with length contʃast in the
inventoʃy (see Cao 1998 and Đoàn 1977 foʃ detailed ʃeviews). The latteʃ view is
oveʃwhelmingly adopted (e.g., Đoàn 1977, V ơng & Hoàng 1994, Nguyễn 1997, Cao

1998, Hoàng 1989) because it produces a symmetrical pattern of distribution: palatals
occur after all three front vowels and velars occur elsewhere.
The first view, although no longer in favor nor adopted in the literature, is worthy of
mention because of the historical hypothesis discussed later. This view is reflected in the
official orthography. The letters ch and nh are used to write the palatals, as in sách [sac]
‘book’, anh [ʔaɲ] ‘eldeʃ bʃotheʃ’, and c and ng are used for velars, as in khác [xa:k]
‘diffeʃent’ and mang [ma:ŋ] ‘to caʃʃy’. Authoʃs holding this view maintain that the palatal
in these syllables is phonemic; however, they disagree among themselves whether the
vowel is phonemically /a/ oʃ / /. ɤome authoʃs identify it as undeʃlyingly /a/, e.g., [ʔaɲ] is
/ʔaɲ/, (Lê Văn Lý 1948, Thompson 1965). This view cʃeates the two asymmetʃical
patterns mentioned above. To solve the distributional problem, other authors identify the
vowel as undeʃlyingly / /, e.g., [ʔaɲ] is /ʔ ɲ/ (Emeneau 1951). Anotheʃ pʃoblem occuʃs:
palatals are said to have a phonemic status but they are in complementary distribution as
seen in Table 4. In Table 4 palatals follow three front vowels and velars follow non-front
vowels.

i
ɨ
u

e
ɘ
o

[aɲ]
a
ɔ

Finals
/c, ɲ/

/k, ŋ/
/k, ŋ/

Table 4. Distribution of phonemic palatals syllable-finally in the first view.
The second view argues for final palatals to be allophones of velars (e.g., Haudricourt
1951, Đoàn 1977, Cao 1998, Cù et al 1977, Hữu & V ơng 1980, Hoàng 1989, Nguyễn
1997). This view is widely adopted in the modern literature. Supporters argue for a
system with a symmetrical distribution of final consonants based on a phonological
process such as ʃeduplication (Đoàn 1977:210). As shown in Table 5, this view produces
a symmetrical distribution of final palatals in which palatals surface after all three
vowels, and /a/, like any other non-fʃont vowel, occuʃs only befoʃe velaʃs, i.e., [aɲ] is
undeʃlying / ŋ/. Palatals aʃe in complementaʃy distʃibution with velaʃs. Examples in (1)
aʃe ʃepeated in (2) with the undeʃlying foʃms (see Đoàn 1977 foʃ details).
5


[c, ɲ]
+
+
+
-

/i/
/e/
/ /
/a/

/k, ŋ/
+

Table 5. Phonemic distribution of palatals and velaʃs in the Hà N i dialect.
(2) lính
chênh
anh
măng

[liɲ]
[ceɲ]
[ʔaɲ]
[maŋ]

/liŋ/
/ceŋ/
/ʔ ŋ/
/maŋ/

‘soldieʃ’
‘tilted/diffeʃent’
‘oldeʃ bʃotheʃ’
‘bamboo shoot’

Ouʃ focus is the low fʃont / / being centʃalized befoʃe palatals. Phonetically, the
general assumption is short vowels show a strong effect on final velars. For example,
afteʃ the thʃee back ʃounded vowels /u/, /o/, and /ɔ/, the velaʃ becomes labialized, e.g.,
chung [cuŋ͡m] ‘shaʃed’, ốc [ʔok͡p] ‘snail’, and ong [ʔɔŋ͡m] ‘bee’, and the vowel is short,
less ʃounded. ɤimilaʃly, the fʃont vowels /i/, /e/, and / / befoʃe a palatal aʃe shoʃt, pulled
toward the central area. The mutual relationship between the vowel and final consonant is
shown in (3) based on Cao 1998’s illustʃation.
(3) Mutual effects between front vowels and final velars
lính *[liŋ] ֜

chênh *[ceŋ] ֜
anh
*[ʔ ŋ] ֜

*[lijɲ] ֜ [lɨɲ] ‘soldieʃ’
*[ceʲɲ] ֜ [cəʲɲ] ‘uneʂual’
*[ʔ ʲɲ] ֜ [ʔ ʲɲ] ‘eldeʃ bʃotheʃ’

The asterisk illustrates the reconstructed forms. The first stage is the input. In stage 2, a
velar following a front vowel is palatalized and surfaces as a palatal. In Stage 3, the front
vowel and palatalized consonant are adjacent with similar features, and the front vowel
dissimilates and is centralized. ([ ʲɲ] is tʃanscʃibed as [aɲ] in otheʃ accounts, be it
phonetic oʃ phonemic.) The centʃalization of / / is stʃongeʃ than that of /i/ and /e/ befoʃe
palatals. In the ɤài Gòn dialect, a southeʃn dialect spoken in Hồ Chí Minh city and
surrounding areas, centralization strongly changes vowel quality in all front vowels, e.g.,
Northern lính [liɲ] ‘soldieʃ’ and ếch [ʔec] ‘fʃog’ aʃe pʃonounced as [lɨn] and [ʔ :t],
ʃespectively. Phạm 2006 offeʃs a similaʃ account using featuʃe geometʃy. In these
analyses, [ ɲ] oʃ [ c], is one of the ʃeconstʃucted foʃms of / ŋ/ oʃ / k/. Neitheʃ [ ɲ] noʃ
[ c] occuʃ in any majoʃ dialects. Howeveʃ, as we will see below, in the Bình Định and
Phú Yên dialects [ ɲ] and [ c] do exist, seʃving as a link to connect the undeʃlying foʃm
/ ŋ/ with the suʃface [aɲ] in noʃtheʃn dialects in a synchʃonic explanation. Moʃeoveʃ, this
foʃm, [ ɲ], also seʃves as evidence foʃ the loss of the final palatals in Vietnamese.
Cao (1998:95) makes an interesting remark that the relationship between front or
back vowels and velar consonants, although synchronic, could have originated from a
diachronic process in which “only some stages aʃe confiʃmed in dialects”. These stages
6


are indeed found in many dialects in South Central Coast which show the loss of final
palatals and alveolars in Vietnamese.

Compaʃative data fʃom the M ng speakeʃs, who live in mountainous pʃovinces in
Noʃth Vietnam, and fʃom the Quảng Nam, Bình Định, and Phú Yên speakeʃs living along
the coastline in South Central Vietnam support the process of sound change: final palatals
are disappearing along with final alveolars.
The following sections will show preliminary results of production and perception
tests which demonstrate that in some South Central dialects, palatals are still preserved
and contrast with alveolars after three front vowels. Which vowel(s) occur(s) depends on
the dialect. Inteʃestingly in these dialects the pʃocess of Hà N i alveolaʃs becoming
velars, seen in southern dialects, also takes place. This fact tells us that both palatals and
alveolars are disappearing.
2. Remnants from the Past: Final Alveolars and Palatals in the Quảng Nam, Bình
Định, and Phú Yên dialects
The field woʃk in 2012 in Quảng Nam, Bình Định, and Phú Yên in ɤouth Centʃal Coast
of Vietnam shows a very interesting phenomenon (map 2): the dialects aʃe ‘mixed.’ Fiʃst,
in otheʃ dialects in Centʃal and ɤouth Centʃal Vietnam the Hà N i palatals completely
disappeaʃ, becoming alveolaʃs. Howeveʃ, in the Quảng Nam, Bình Định, and Phú Yên
dialects, the Hà N i palatal becomes an alveolaʃ only afteʃ vowel [a], but is retained after
[i] and [e] in the Quảng Nam and Bình Định dialects, and afteʃ [ ] in the Bình Định and
Phú Yên dialects. Second, like other dialects of Central (from Hue southwards) and South
Vietnam, Hà N i alveolaʃs become velaʃs afteʃ non-front vowels. The surviving palatals
make Quảng Nam, Bình Định, and Phú Yên conseʃvative dialects, like noʃtheʃn dialects.
The loss of alveolars after non-fʃont vowels is a new development, and gʃoups Quảng
Nam, Bình Định, and Phú Yên among innovative dialects, those that were formed during
the last few centuʃies in the new teʃʃitoʃies. This ‘mixed’ distʃibution of palatals and
alveolaʃs in the Quảng Nam and Bình Định dialects documents a sound change in
progress in Vietnamese: palatals and alveolars in the coda are disappearing.

7

Map 2. Provinces of Vietnam
ɤouʃce: Haughton, Domiʂue and Nguyễn Phong 2004
Foʃ the pʃoduction test, the data fʃom Quảng Nam dialects were collected from field
woʃk conducted in Vietnam in 2004 and 2012. The data fʃom the speech of Hoài Nhơn
and Phù Cát distʃicts of the Bình Định pʃovince, and of Hịa Quang Bắc of the Phú Yên
pʃovince, weʃe collected fʃom field woʃk in Bình Định and Phú Yên in 2012. The
production test used pictures to elicit the relevant syllables in monosyllabic and disyllabic
words and also in interviews.

8


The production test used a list of 178 words consisting of all possible Vietnamese
rhymes. The majority of words are monosyllabic. The recorded tokens were transcribed
by the author.
Theʃe weʃe thʃee Bình Định speakeʃs: a female and a male fʃom Hoài Nhơn distʃict,
and a male from Phú Cát district, all students in their early twenties. The speech of the
Phú Cát male sounds slightly diffeʃent fʃom the two Hoài Nhơn speakeʃs, showing
variations with the ʃhymes pʃonounced as [ ɲ] by the two Hoài Nhơn speakeʃs. This
section ʃepoʃts the ʃesults of pʃoduction test fʃom the two speakeʃs of Hoài Nhơn as they
show consistency, although each subject comes fʃom a diffeʃent village. Hoài Nhơn
district is located near the coastline; residents live in isolated fishing villages. The Phú
Cát distʃict is located closeʃ to Quy Nhơn city and on the National Highway 1, which
could be a factor for residents here adopting a more urban pronunciation. The perception
test, however, includes results from listening to the Phú Cát speaker. For the Phú Yên
dialect, two females in their early forties were used for data elicitation. They are small
business owneʃs in the Hòa Quang Bắc village, and theiʃ speech also shows consistency
within and across speakers.
In order to avoid the reading effect, pictures were used to elicit these 178 syllables.
The subjects were asked to say the name of the objects on the screen, one object at a time.

Ambiguous pictures were avoided in order to reduce unwanted synonyms. The majority
of the testing words were nouns, referring to familiar objects in daily life: tools, flowers,
plants, numbers, furniture, colors, nature events, etc. This method made it difficult to
have available minimal pairs, but easy for the subjects to identify the picture with just one
quick choice of words. The words were put in two lists in a random order, i.e., two
repetitions for each syllable. The same syllable/word was represented with a different
picture in each list. The pictures in two lists were displayed on a laptop screen as slide
shows to elicit the testing words. When ambiguity happened or the subject offered more
than one word to name a picture but still failed, the correct word appeared on the screen
and the subject was asked to read it out loud for the recording. This method was used to
get the most natural and quickest reaction from the subject without any orthographic bias.
Sounds were recorded using Audacity program with an iMic microphone connected to a
Macbook Pro laptop. The subjects were told what to do, then looked at the pictures on the
laptop screen, and named the picture. The sample rate was set at 44kHz.
In addition to naming the objects, each subject was interviewed for a few minutes and
recorded. Relevant syllables from the interviews were isolated and transcribed to check
against tokens elicited in the picture naming task.
Although the list includes all rhymes, we will discuss only syllables with final
alveolars and palatals after front vowels and [a]. The description of the entire sound
system of these dialects is outside the scope of this paper.
Let us first look at how alveolars are disappearing in dialects of central and south
Vietnam. Compaʃative data of Hà N i (a ʃepʃesentative northern dialect) and Sài Gòn (a
representative southern dialect) are given below in (4). Transcription is phonetic. The Hà
N i velaʃs coʃʃespond to ɤài Gòn velaʃs in all enviʃonments, which aʃe not included in
(4). The Hà N i alveolaʃs coʃʃespond with the Sài Gòn velars (4f – n). After [i], [e] and
[a] both Hà N i alveolaʃs and palatals coʃʃespond with the ɤài Gòn alveolaʃs (4a-e).

9

(4) Compaʃative data of the Hà N i and ɤài Gịn dialects
a.
b.
c.
d.
e.
f.
g.
h.
i.
j.
k.
l.
m
n.

Orthography
xin
lính
lên
ếch
ăn
đen
mứt
ớt
mất
cát
cắt
hút
tốt

ngọt

HÀ N I
sin
liɲ
len
ʔec
ʔaɲ
d n
mɨt
ɘ:t
mɘt
ka:t
kat
hut
tot
ŋɔt

SÀI GỊN
sɨn
lɨn
lɘ:n
ʔɘ:t
ʔan
d :ŋ
mɨk
ɘ:k
mɘk / mak
ka:k
kak

huk͡p
to:k͡p
ŋɔ:k

Translation
to ask for
soldier
up
frog
to eat
black
jam
pepper
to lose
sand
to cut
to suck
good
sweet

Thompson 1967 mentions a similaʃ patteʃn between Vietnamese and the M ng
Khen dialect, where palatals contrast with alveolars and velars after most vowels. He says
that wheʃe the M ng ɜhen dialect has final palatals, northern Vietnamese dialects have
alveolaʃs, much like the Hà N i palatals ʃelationship to the ɤài Gòn alveolaʃs. The data in
(5) are from Thompson 1967, supplied with (5h – m) from Baker 1970. (5a-m) show that
afteʃ [ɘ:, a:, a, e, , u, o], and [ɔ] M ng final palatals coʃʃespond to Vietnamese
alveolaʃs. Inteʃestingly, M ng [ ɲ] and [ c] coʃʃespond to (noʃtheʃn) Vietnamese [aɲ]
and [ac], as in (5n-p).
(5) Compaʃasion of final palatals and alveolaʃs in M
a.

b.
c.
d.
e.
f.
g.
h.
i.
k.
l.
m.
n.
o.
p.

Viet orthography
bán
sàn
cát
bắn
cắn
lên
đến
rút
bớt
một
mọt
ruột
anh
rách

cành

M

NG
pa:ɲ
xa:ɲ
ka:c
paɲ
kaɲ
tleɲ
teɲ
ruc
pɘ:c
moc
mɔc
ʃɔc
ʔ ɲ
ʃ c
k ɲ

ng and Vietnamese

VIETNAMESE
ba:n
sa:n
ka:t
ban
kan
len

den
zut
bɘ:t
mot
mɔt
zuət
ʔaɲ
zac
kaɲ
10

Translation
to sell
floor
sand
to shoot
to bite
up
to arrive
to withdraw
to diminish
one
termite
intestine
elder brother
be torn
branch


From this correspondence, Thompson concludes that the original palatals (in the ancestor

of Viet-M ng) that occuʃʃed afteʃ many vowels aʃe still pʃeseʃved in M ng, but in
Vietnamese, palatals remain only after front vowels. Elsewhere they developed to
alveolaʃs (i.e., “apicals” in Thompson’s), in the Hà N i dialect.2 Thompson’s suggestions
have not received any further discussion. However, his comparison of palatals with
alveolars, not with velars, is interesting.
Like otheʃ dialects spoken in Centʃal and ɤouth Vietnam, the Quảng Nam, Bình Định,
and Phú Yên dialects also lost alveolars in the final position. Alveolars, which correspond
to the Hà N i palatals, occuʃ only afteʃ thʃee shoʃt vowels [i], [e], and [a], shown in
Table 6, examples given in (6). However, unlike other dialects, the palatals are still found
after short vowels, creating a contrast with alveolars, but varying depending on dialects:
both alveolaʃs and palatals occuʃ afteʃ the fʃont vowels [i] and [e] in Quảng Nam and
Bình Định dialects, and afteʃ [a] in Phú Yên dialect. Alveolaʃs do not follow [ ].

t, n

c, ɲ

i

+

+ (QN, BD)

e

+

+ (QN, BD)

k, ŋ

+ (BD, PY)
a

+ (QN, BD, PY)

+

Table 6. Distʃibution of final alveolaʃs and palatals in the Quảng Nam, Bình Định and
Phú Yên dialects.
(6) đinh [dɨn]
chín [ciɲ]
sên [sen]
lên [leɲ]
ăn [ ɲ]
cánh [kan]
nhẫn [ɲaŋ]

‘nail’
‘nine’
‘snail’
‘go up’
(Bình Định, Phú Yên) ‘to eat’
‘wing’
‘ʃing’

The fact that alveolaʃs occuʃ afteʃ shoʃt [a] but not shoʃt [ ] cʃeates a similaʃ
asymmetʃical pʃoblem as seen in the Hà N i dialect. If, as in the ɤài Gòn dialect,
assuming [a] (phonetically moʃe like [ ] oʃ [ɐ]) befoʃe alveolaʃs is undeʃlyinly / /, the
alveolaʃs occuʃ only afteʃ 3 fʃont vowels: /i/, /e/, and / /. Elsewheʃe velaʃs/palatals occuʃ.

Table 7 shows this symmetrical distribution. Palatals are retained after all three front
vowels in the Bình Định dialects.
2

In (5f) and (5g) the alveolars must be original alveolars in the Hanoi dialect, where
alveolars contrast with palatals.
11


t, n

c, ɲ

i

+

+ (QN, BD)

e

+

+ (QN, BD)

+ (QN, BN, PY)

+ (BD, PY)

a

k, ŋ

+

Table 7. Phonemic distʃibution of final alveolaʃs and palatals in the Quảng Nam, Bình
Định, and Phú Yên dialects.
Because both alveolars and palatals occur with restriction and are in complementary
distribution with velars, it cannot be the case that both alveolars and palatals are separate
phonemes in these dialects. Either alveolar or palatal can be said to be an allophone of
velars while the other is a phoneme. If the alveolar consonant is allophonic and the
palatal consonant is a phoneme, we can assume palatals are remnants of the historical
palatals, ʃemaining afteʃ /i/ and /e/ in the Quảng Nam and Bình Định dialects, and after
/ / in the Bình Định and Phú Yên dialects.
Alternatively, one can say the final palatal is an allophone of velars and the final
alveolaʃ is a phoneme. In the Quảng Nam and Bình Định dialects palatals occuʃ only afteʃ
/i/ and /e/ and are in complementary distribution with velars, so we can assume that
palatals are allophones of velars and that the alveolar after [i], [e], and [a] is a phoneme, a
historical remnant from Northern dialects. However, it would be hard to explain why
alveolars are lost after most vowels. Moreover, in Southern dialects from Hue southward
alveolars are consistently found after [i], [e], and [a], but palatals are completely lost in
these dialects.3 The palatals that survive in South Central are found after various front
vowels and in different sub-dialects. It is unlikely that alveolars are historical remnants
and palatals are allophones in these dialects. Palatals seem best treated as remnants from
the past.
The fact that in the Quảng Nam, Bình Định, and Phú Yên dialects Northern palatals
aʃe still ʃetained but alveolaʃs aʃe lost also suggests that these dialects aʃe a ‘hybʃid’ type,
falling in a ‘peʃipheʃal’ aʃea wheʃe the old featuʃes aʃe still seen but innovative featuʃes
have already emerged. This phenomenon further supports a hypothesis of gradual
disappearance of final palatals in modern Vietnamese.

Besides final palatals, anotheʃ ‘dinosauʃ’ featuʃe we aʃe able to tʃace is the suʃfacing
of the shoʃt vowel [ ] befoʃe a palatal, i.e., [ ɲ]. Histoʃically, if the final palatal in modern
3

Among children born to Northern parents living in south Vietnam, or children who left
north Vietnam during their childhood to live in South Central and South Vietnam and still
speak with a northern accent, the first feature that is lost in their speech is final palatals
when all other features might still be preserved, e.g., final alveolars after all vowels, or
tone ngã (glottalized, rise-fall).
12


Vietnamese is a ʃelic of palatals fʃom the ancestoʃ of the Việt-M ng, then we might be
able to find some Vietnamese dialects in which palatals still occuʃ afteʃ [ ]. Indeed, the
ʃhyme [ ɲ] and [ c] aʃe found in the dialects spoken in Bình Định and Phú Yên
provinces, South Central Vietnam, which are discussed in the next section.
3. The ‘Missing link’ [ɛɲ] and [ɛc] - Palatals after [ɛ]
Although researchers today adopt the view that northern anh [ʔaɲ] ‘eldeʃ bʃotheʃ’ is
undeʃlyingly /ʔ ŋ/ based on both phonetic and phonemic gʃounds, not eveʃyone agʃees on
how the process works when both the vowel and consonant on the surface are different
from the input in the underlying form. In northern dialects it is said that the vowel does
not cleaʃly sound like [a] oʃ [ ] but is something in between, and that the final palatal is
produced a little more further back than when it occurs in the onset, e.g., compare lính
[liɲ] ‘soldieʃ’ and nhìn [ɲin] ‘to look at’. This is because the palatal is indeed a
palatalized velar. Neither the vowel nor the final consonants in this rhyme is phonetically
clear. The rate of confusion from the perception test confirms this observation.
Two tests, production and perception, were conducted to confirm the findings of [ ɲ]
and [ c]. The ʃesults show that speakeʃs of both the Hoài Nhơn (Bình Định) and Hịa
Quang Bắc (Phú n) dialects have the ʃhymes [ ɲ] and [ c] in theiʃ speech, something
not seen in any of the well-studied dialects.

3.1. Production test
Woʃds containing [ ɲ]/ [ec] elicited in the pʃoduction test and used in the peʃception test
are shown in (7).
(7) Woʃds ʃecoʃded with [ ɲ]/ [ec] in the Hồi Nhơn (Bình Định) and Hòa Quang Bắc
(Phú Yên) dialects.
a. chén [c ɲ] ‘small bowl’
b. đèn [d ɲ] ‘lamp’
c. sen [s ɲ]
‘lotus’
d. xẻng [s ɲ] ‘shovel’
e. sét [s c]
‘lightning’
f. kẹt [k c]
‘to jam’
g. ăn [ ɲ]
‘to eat’
h. tʃăn [ʈ ɲ] ‘python’
e. ʃắn [ɻ ɲ]
‘snake’
g. măng [m ɲ] ‘bamboo shoot’
k. lăng [l ɲ] ‘mausoleum’
l. tʃăng [ʈ ɲ] ‘moon’
m. ʃăng [ɻ ɲ] ‘tooth’
n. tʃắng [ʈ ɲ] ‘white’
o. xăng [s ɲ] ‘gas’
p. mắt [m c] ‘eye’
In the production test, the final consonant in words in (7) produced by speakers of the
Bình Định and Phú Yên dialects sometimes sounds like a sound between a palatal and
13

velar. 4 To be consistent, because palatals and velaʃs do not contʃast afteʃ [ ] in any
dialect, the IPA symbols for palatals are used here to represent the final consonant in this
rhyme. However, the vowel in (7) is cleaʃly [ ].
In oʃdeʃ to confiʃm that the ʃhyme the speakeʃs of Bình Định and Phú Yên dialects
pʃoduces is [ ŋ] oʃ [ ɲ] but not [aɲ], which exists in the Hà N i dialect, a peʃception test
was used as a follow up for the production test.
3.2. Perception test
The word file used for the perception test was excerpted from the recordings of the
production test. The perception test was designed to see how speakers of other dialects
peʃceive woʃds in (7) spoken by the Phú Cát speakeʃ (Bình Định pʃovince) and of the two
Hịa Quang Bắc speakeʃs (Phú Yên pʃovince). The ʃecoʃded speech of these speakeʃs was
chosen for the experiment because the recordings are very clear, with minimal
backgʃound noise. The speech of the Hoài Nhơn speakeʃ (Bình Định) was also used, but
with only four listeners, and the results were diverse. Therefore, the discussion focuses
only on the ʃesults of the speech of the Phú Cát and Hòa Quang Bắc speakeʃs. (Due to the
limited time, the perception test was not given to speakers of the same dialects.)
Only three to five words in (7) from each speaker were selected, segmented, and
saved in individual files, then played to listeners. Each token was played three or four
times. The listener was asked to write down what he/she heard, using the official
oʃthogʃaphy. The peʃception test was peʃfoʃmed in two cities: Đà Nẵng (ɤouth Centʃal
Coast) and Hồ Chí Minh (ɤoutheasteʃn ʃegion). Beside natives of these cities, listeneʃs
included those who come from various provinces to work there. The results show the
confusion of speakers when listening to these words.
Before looking at the results in Table 8 below, we should note some points about the
writing system. The Vietnamese orthography, using a Roman script, was created by
Western missionaries in the 16th century for the purpose of learning the local language in
order to preach. The script was not officially used until the late 19th century and from the
mid 20th centuʃy became the only wʃiting system of the nation, called “the National
ɤcʃipt”. The oʃthogʃaphy represents the sound system of the northern dialects plus three

initial ʃetʃoflexes, /ʈ/, /ɽ/, and / /, found in otheʃ dialects. ɣecall that letteʃs ch and nh are
used to write the palatal stop and palatal nasal, respectively, i.e., inh foʃ [iɲ], ênh for [eɲ],
anh foʃ [aɲ], ic for [ic], êch for [ec], and ach for [ac]. A long vowel before a velar is
written with two identical letters, but this convention is not always consistent. For
example, ôông is used foʃ [o:ŋ], oong foʃ [ɔ:ŋ], but [e:ŋ] and [ :ŋ] aʃe written with only
one letter, ê and e, respectively, as in (côông) kêng [ke:ŋ] ‘to caʃʃy someone on
shouldeʃs’, and xẻng [s :ŋ] ‘shovel’. This inconsistency cʃeates some pʃoblem with the
test’s design: the woʃds in (7) aʃe pʃonounced with [ ] befoʃe [ɲ] oʃ [ŋ] in the Bình Định
and Hòa Quang Bắc dialects, but theʃe is no official combination in the oʃthogʃaphy to
write this rhyme, which does not exist in other dialects. The listeners were told to use any
letters that best represent the sounds they hear, even if the sounds were nonsense and the
Ladefoged (2003:160) states that “the place of aʃticulation of consonants as tʃaditionally
defined is not readily apparent through acoustic analysis”. ɣatheʃ, he advises “you can do
much better with the simple palatography technique ..or even by just looking at the
speakeʃ’s mouth”.
4

14


combination of letteʃs did not look ‘familiaʃ’. When the subject heaʃd [ ŋ] oʃ [ ɲ], the
most appropriate combinations of letters to choose are eng or enh, respectively; however,
in the orthography eng is used to write a ʃhyme with the long [ :] befoʃe a velaʃ, as in
xẻng [s :ŋ] ‘shovel’, and the combination enh is not used in the orthography although it
would be the most logical orthographic representation for the combination of the vowel
and consonant in this rhyme. Only one token was written with enh. The orthographic
convention was a drawback of the test, as it would have been unreasonable to ask a
subject to write down a combination that has no equivalent representation in the
orthography. (Future research should possibly use a better design for a test with a similar
purpose.)

Even with the orthographic problem, we can still predict that if the sound they heard
is not familiar to them the listeners would be confused and try various existent
combinations to write down what they hear. If inconsistency is found among subjects, it
shows that the sound is not how it is pronounced in the dialects the subjects are
acquainted with. This prediction turned out to be correct.
Let us now look at the results when the speech of the Phú Cát speaker was played.
Ten listeneʃs, age ʃanged fʃom eaʃly 20s to late 50s, speakeʃs of the Quảng Nam dialect,
participated in the test. The subjects were not told which dialect they were going to listen
to. Five monosyllabic words selected from the Phú Cát speaker transcribed based on the
northern dialects are: răng [zaŋ] ‘tooth’, ăn [an] ‘to eat’, trắng [caŋ] ‘white’, sắn [san]
‘yucca’, and mắt [mat] ‘eye’. These woʃds aʃe pʃonounced with the shoʃt vowel [a] in
noʃtheʃn dialects, but the shoʃt [ ] before a palatal in the Phú Cát dialect. Ten subjects
produced 50 tokens. Table 8 shows this results. The first column gives the orthography
used by the subjects to record how they perceived the sounds. The phonetic transcription
of the first column is given in the second column. The last column shows the number of
tokens for each combination, e.g., 39 tokens from any of the five syllables were perceived
as anh [aɲ].
Orthography
ên
anh
eng
ăng
Other

Transcription
[en]
[aɲ]
[ :ŋ]
[aŋ]

%
6
76
6
6
4

Total: 50
3
39
3
3
2

Table 8. Results of 10 subjects listening to răng, ăn, trắng, sắn, and mắt in the Phú Cát
speech.
Oveʃall, the Quảng Nam listeneʃs used many ways to wʃite down what they heaʃd:
combinations inh foʃ [iɲ], anh foʃ [aɲ] and eng foʃ [ :ŋ] weʃe used most fʃeʂuently. The
combinations in [in], ên [en], or en [ n] weʃe also used but much less fʃeʂuently.
Table 8 shows 76% of occurrences of anh [aɲ] heaʃd as [ ɲ] oʃ [ ŋ]. Note that all the
syllables in (7) aʃe pʃonounced with the long [ :] in the Quảng Nam speech, which is the
dialect of the listeneʃs. Inteʃestingly, even though the Hà N i shoʃt [a] in (7) coʃʃesponds
to the Quảng Nam long [ :], the Quảng Nam listeneʃs still used the shoʃt [a] foʃ what
sounds moʃe like [ ] than [a]. This could be an orthographic influence.
15


Nine subjects, ages ranging from the early twenties to late fifties, listened to the data
pʃoduced by the Hòa Quang Bắc speakeʃ. Among these, fouʃ aʃe fʃom Đà Nẵng city
(whose dialect is similaʃ to the Quảng Nam dialect), and five aʃe fʃom Hồ Chí Minh city

(Sài Gịn dialect). Four monosyllabic words selected from (7) were played. These
syllables again are pronounced with the short [a] in Northern dialects: trăng [caŋ]
‘moon’, măng [maŋ] ‘bamboo shot’, sắn [san] ‘yucca’, and khăn [xan] ‘scaʃf’. ɣesults
are shown in Table 9. Nine listeners with four words produced 36 tokens. We see again
that [aɲ] has the highest scoʃe, 31%, although not oveʃwhelmingly as high as in the case
of Phú Cát speaker.
Orthography
ăn
anh
eng
inh
i
other

Transcription
[an]
[aɲ]
[ :ŋ]
[iɲ]
[i]

%
3
31
19
19
14
14

Total 36

1
11
7
7
5
5

Table 9. ɣesults of 9 subjects listening to 4 woʃds in Hòa Quang Bắc : trăng, măng, sắn,
and khăn.
The listeneʃs to the Hòa Quang Bắc speakeʃ seemed to be moʃe puzzled than those
listening to the Phú Cát speaker. They also used [i] in a relatively high number of tokens,
eitheʃ with a final palatal (19%), oʃ in open syllables (14%). Peʃhaps the Hòa Quang Bắc
speakeʃ pʃonounced the vowel in [ ŋ] with a much naʃʃoweʃ openness and a higher
tongue position than foʃ the usual [ ].
In both cases, it is interesting that anh [aɲ] was oveʃwhelmingly the choice. Note that
the Northern anh [aɲ] is pʃonounced with an alveolaʃ, [an], in ɤouth Centʃal and ɤoutheʃn
dialects. In order to determine whether when the subject wrote anh, they really meant the
noʃtheʃn pʃonunciation [aɲ], a numbeʃ of monosyllabic woʃds pʃonounced with the
noʃtheʃn [aɲ] was included in the test, e.g., cánh [kaɲ] ‘wing’, gạch [ ac] ‘bʃick’, hành
[haɲ] ‘onion’. All these syllables weʃe coʃʃectly wʃitten with the ʃhymes [an] and [at] just
as they aʃe pʃonounced in the subjects’ dialect. Theʃefoʃe, the final consonant in tokens
written as anh in Table 8 and Table 9 is certainly not an alveolar, as perceived by these
listeners.
It is also interesting to note that when the subjects used the combination anh to record
most of these words they probably indicated that the vowel is short, because nh occurs
only after short vowel. On the other hand, because the vowel sounds moʃe like [ ] which
can only precede ng (a velar) but not nh, a palatal, in the official orthography, some
speakers used eng although eng is used to wʃite ʃhyme with a long [ :] befoʃe a velaʃ in
the orthography. The fact that listeners used various spelling to write down what they
heard suggested that they were not satisfied with any of these spellings.

In summaʃy, the bewildeʃment of listeneʃs in finding the ‘coʃʃect’ spelling to wʃite the
ʃhymes [ ɲ] and [ c] cleaʃly shows that this is not what they are familiar with in their
dialects or in any other dialects they know.

16


Table 10 summarizes the distribution of final palatals after front vowels in the
dialects discussed so far, in the order from north to south. We see that the further south
the dialect, the less chance for final palatals to survive. Whereas palatals occur after three
vowels in northern dialects, they are completely lost in Sài Gịn and all other southern
dialects.

NORTH
֝
SOUTH

i
X
X
X

Hà N i
Quảng Nam
Bình Định
Phú Yên
Sài Gòn

e
X

X

X
X
X

Table 10. Distribution of palatals after front vowels
Palatals occuʃ afteʃ all fʃont vowels in the Hà N i dialect. In otheʃ dialects palatals aʃe
either retained afteʃ two fʃont vowels, oʃ only afteʃ / / in the Phú Yên dialect. Those
surviving palatals become alveolars in most dialects of central and southern Vietnam.
The occuʃʃence of the palatal afteʃ / / in the Bình Định and Phú Yên dialects could
result from patterns of mass migration from north central to central and south central
Vietnam, as we will see in the next section.
4. Historical Palatals and Immigration to Central and South Central from North
Central Vietnam
This section examines the pattern of immigration during five centuries, from 14th to 19th
centuries, which might be responsible for the palatals after front vowels in the dialects
discussed in this paper. The migration of the Vietnamese people from north central to
central and south central Vietnam carried with it archaic features of the language to the
new lands.
The territory of current central Vietnam belonged to the Champa kingdom for
seventeen centuries, from 192 to 1832. In 1306 Cham King Jaya Sinhavarman III offered
the Vietnamese king, as a marriage gift, two northern provinces, roughly corresponding
to the cuʃʃent Quảng Bình, Quảng Tʃị, and Thừa Thiên pʃovinces (southeʃn paʃt of Noʃth
Centʃal). People fʃom Thanh Hoá and Nghệ An (noʃtheʃn paʃt of noʃth centʃal) weʃe sent
to these provinces to live. Later, through war with Champa in 1402 Vietnam took two
pʃovinces, Quảng Nam and Quảng Ngưi: the fiʃst big wave of immigʃants, again, was
fʃom Thanh Hoá, Nghệ An, and Hà Tĩnh pʃovinces (Nguyễn 2007). In 1446 Cham lost
Vijaya (current Qui Nhon city) for Vietnam. This piece of land corresponds to current
Quảng Nam, Quảng Ngưi, and Bình Định pʃovinces (noʃtheʃn paʃt of south centʃal

coast). In 1545, in its “Nam Tiến” (ɤouthwaʃd Expansions) movement, Vietnam took
over Phú Yên province: the second big wave of settlers came here from Ninh Bình,
Thanh Hóa, Nghệ An, and Hà Tĩnh (Bʃunelle 2005). Vietnam was divided: the south was
ʃuled by Nguyen Loʃd, the noʃth by Tʃinh Loʃd. In 1648, 30,000 Quảng Bình soldieʃs
were captured by the Nguyen army and brought to Quảng Nam to live. A few years later,
in 1657, afteʃ winning a big battle, Nguyễn Loʃd bʃought home with him people fʃom 7
distʃicts of Nghệ An pʃovince to Quảng Nam to live.
17


The vast majority of Vietnamese immigrants living in the central and south central of
Vietnam, theʃefoʃe, came fʃom the noʃtheʃn paʃt of noʃth centʃal coast. Thanh Hoá, Nghệ
An, and Hà Tĩnh pʃovinces in noʃth centʃal Vietnam aʃe known as the most conseʃvative
language area. Alves 2012 examines the grammatical vocabulary of north central
Vietnam and compares it to other Vietic languages to show that the dialect spoken in
north central Vietnam still keeps many old words that it does not share with either
northern or southern dialects.
The noʃth centʃal aʃea is also wheʃe the M ng people live, although theiʃ settlement
is mainly in mountainous provinces in North Vietnam, especially in Hồ Bình, Thanh
Hố, and Ninh Bình pʃovinces, south of Hà N i. Bakeʃ’s 1970 list shows that befoʃe
alveolaʃs oʃ velaʃs, the Pʃoto Việt-M ng * coʃʃesponds to in both Việt and M ng.
When the vowel precedes alveolars or velars or is in open syllables, there is no
centʃalization. Foʃ example, PVM *l n > l n (Việt and M ng) ‘secʃetly’, *d > d (Viet)
and t (M ng) ‘give biʃth’; PVM * ŋ > :ŋ (Việt and M ng) ‘shovel’ (Bakeʃ
1970:280). Howeveʃ, when * pʃecedes palatals, the vowel is centʃalized in Việt, but not
in M ng. It seems final palatals cause centʃalization in Vietnamese. ɤome examples foʃ
this case are given in (6n - p), additional examples are provided below in (8).
(8)
PV-M
*c ɲ

*s ɲ
*l ɲ

M
c
s
l

ng
ɲ
ɲ
ɲ

Việt
caɲ
saɲ
laɲ

Việt oʃthogʃaphy
tranh
xanh
lanh

Translation
to argue
blue, green
cold

Duʃing the Việt-M ng Ancient peʃiod, fʃom the 1st to 8th or 9th centuries, and the
Common Việt-M ng Peʃiod, fʃom the 10th to 14th centuries, palatals were contrastive in

the coda (Tʃần 2005), and aʃe still contʃastive in the final position in modeʃn M ng.
Tʃần (2005:189) pʃovides a list of compaʃative vocabulaʃy of the modeʃn M ng,
northern and north central Vietnamese dialects to show that the dialects of the speakers in
noʃth centʃal aʃe veʃy close to the M ng language.
The ʃhymes [ ɲ] and [ c] ʃetained in the speech of people in Quảng Nam, Bình Định,
and Phú Yên of south central Vietnam may well be a trace from the Viet-M ng
Common peʃiod, spoken by people in Thanh Hoá, Nghệ An, and Hà Tĩnh, lateʃ going
south. The final palatals afteʃ /i/ and /e/ in the Quảng Nam and Bình Định dialects could
be a trace of immigrants from Thanh Hoá and Ninh Bình whose dialects still have final
palatals.
There is an obvious geographic gap in the distribution of final palatals in modern
Vietnamese. Palatals occur in northern and north central dialects. They completely
disappeaʃ in Thừa Thiên-Huế, the province between north and north central Vietnam.
Palatals then aʃe found in the Quảng Nam, Bình Định, and peʃhaps the Quảng Ngưi
dialects too. ɤouth of Bình Định, the final palatals again completely disappeaʃ. Theʃe aʃe
no reports on these finals anywheʃe. Also, the palatals/velaʃs afteʃ [ ] aʃe not seen
anywheʃe except in the Bình Định and Phú Yên dialects, and in some sub-dialects of
noʃth centʃal as in Hà Tĩnh and Quảng Bình (Cao 1998). This is an obseʃvation needing
further investigation. If it is correct it would provide further evidence that the rhyme
migrated with the north central speakers.
18


5. Conclusion
This paper examines the asymmetrical distribution of final consonants in various dialects
of Vietnamese, from the standard dialect of northern Vietnam to sub-dialects in south
central Vietnam. The distribution of these consonants presents a picture of sound change
in progress in Vietnamese. It shows that the historical palatals are disappearing. The
histoʃical alveolaʃ in the Hà N i dialect is also disappearing in most dialects of central
and south Vietnam. The hypothetical ʃhymes [ ŋ] and [ c] in the Hà N i dialect aʃe

actually found in the Bình Định and Hịa Quang Bắc dialects, a fact that suppoʃts the
hypothesis that palatals once occurred in Vietnamese afteʃ most vowels including / /, as
they still do in the modeʃn M ng language. The suʃviving palatals in Vietnamese lateʃ
became alveolars after front vowels in most dialects of central and southern Vietnam.
In some dialects, the sound change in progress is clearly shown in the disappearance
of palatals and alveolars (merging with velars) after most vowels, but at the same time
these historical palatals and alveolars are still preserved after short front vowels.
The pattern of immigration of Vietnamese settlers during the 14th through 17th
centuries also explains why in the new lands we find features of the old Vietnamese. The
pattern of immigration also explains why these features are found in dialects that are
remote and isolated from their ultimate sources.

References
Alves, Marc. 2012. Notes on grammatical vocabulary in Central Vietnamese. In Journal
of the Southeast Asian Linguistics Society 5:1-11.
Baker, A. Muriel and Milton E. Baker. 1970. Proto-Vietnamuong (Annamuong) Final
Consonants and Vowels. Lingua 24, 268-285.
Brunelle, Marc. 2005. Register in Eastern Cham: phonological, phonetic and
sociolinguistic approaches. Ph. D. dissertation, Cornelle University.
Cao, Xuân Hạo. 1998. Tiếng Việt - Mấy vấn đề ngữ âm, ngữ pháp, ngữ nghĩa (Some
issues on phonetics, syntax and semantics in Vietnamese). Hà N i: Giáo dục.
Cù, Đình Tú, Hồng Văn Thung & Nguyễn Nguyên Tʃứ. 1977. Ngữ Âm Học Tiếng Việt
Hiện Đại (Phonology of the Modeʃn Vietnamese). Hà N i: Publisheʃ of Education.
Đoàn, Thiện Thuật. 1977. Ngữ Âm Tiếng Việt (Vietnamese Phonology). Hà N i: Đại học
và Tʃung học chuyên nghiệp.
Emeneau, M.B. 1951. Studies In Vietnamese (Annamese) Grammar. Berkeley: University
of California Publications in Linguistics.
Haudʃicouʃt, Andʃé. 1954. De l’oʃigine des tons en viêtnamien. Journal Asiatique
242:69-82.
Haughton, Dominiʂue and Nguyễn Phong. 2004. Gʃaphical and Numeʃical Descʃiptive

Analysis: Exploratory Tools Applied to Vietnamese Data. Journal of Statistics
Education Volume 12, Number 2.
/>Hoàng, Thị Châu. 1989. Tiếng Việt trên các miền đất nước (Vietnamese dialectology).
Hà N i: Nhà Xuất Bản ɜhoa Học Xư H i.

19


Hữu, Quỳnh and V ơng L c. 1980. Khái Quát Về Lịch Sử Tiếng Việt Và Ngữ Âm Tiếng
Việt Hiện Đại (Introduction to the History and Phonology of Contemporary
Vietnamese). Hà N i: Nhà xuất bản Giáo dục.
Ladefoged, Peter. 2003. Phonetic Data Analysis: an Introduction to Fieldwork and
Instrumental Techniques. Malden – Oxford – Carlton: Willey-Blackwell.
Lê, Văn Lý. 1948. Sơ Thảo Ngữ Pháp Việt Nam (Introduction to Vietnamese Grammar).
Paʃis: H ơng ɤơn.
Nguyễn, Đình Hồ. 1997. Vietnamese. Amsterdam/Philadelphia: John Benjamins
Publishing Company.
Nguyễn, Đăng Vũ. 2007. Quảng Ngãi – Một số vấn đề lịch sử văn hóa (Quảng Ngưi –
some issues on cultuʃal histoʃy). Hà N i: Nhà xuất bản ɜhoa học xư h i.
Ohala, ɛohn. 1992. What’s cognitive, what’s not, in sound change. In Diachrony within
Synchrony: Language History and Cognition. Gunter Kellermann/Michael D.
Phạm, Andʃea Hoà. 2006. Vietnamese ɣhyme. Southwest Journal of Linguistics, Vol 25,
107–142.
Thompson, Laurence E. 1965. A Vietnamese Grammar. Seattle: University of
Washington Press.
Thompson, Laurence E. 1967. The History of Vietnamese Final Palatals. Language, Vol.
43, No 1, pp. 362-371.
Tʃần, Tʃí Dõi. 2005. Giáo trình lịch sử tiếng Việt (Lectures on History of Vietnamese
language). Hà N i: Nhà Xuất Bản Đại học Quốc gia Hà N i.
V ơng, Hữu Lễ and Hoàng Dũng. 1994. Ngữ Âm Tiếng Việt (Vietnamese phonology).

Hà N i: Giáo dục.

20