Sussex R., Cubberley P. The Slavic Languages
Подождите немного. Документ загружается.
this scenario, and in particular soft and palatal consonants have a
fronting and/or raising effect on vowels. If the preceding consonant is
also soft or palatal, the effect is maximal. Typical effects are:
Phonemic (distributional limitations):
/a/ > /e/ Czech: after soft or palatal consonant s and not before
hard consonants (¼ pr
ˇ
ehla
´
ska ‘umlaut’)
dus
ˇ
e ‘soul’ (PSl/Rus dus
ˇ
a
´
)
jevit ‘to show’ (PSl javiti,Rusjavı
´
t
0
-sja ‘to seem’)
be
ˇ
zˇet ‘to run’ (PSl be
ˇ
zˇati – from earlier *be
ˇ
ge
ˇ
ti,
Rus bezˇa
´
t
0
)
USorb: between soft consonants
jejo ‘egg’ (PSl jaje, Rus jajco
´
)
rjen
´
s
ˇ
i ‘earlier’ (PSl ran{s
ˇ
{j{, Rus ra
´
n
0
s
ˇ
ij)
rjad ‘row’ NomSg w rjedz
´
e LocSg
/o/ > /e/ USorb: between a hard and a soft consonant
sel ‘salt’ (PSl sol{, Rus sol
0
)
zerja ‘dawn’ (PSl zor{ja, Rus zo
´
rja)
/u/ > /i/ Czech: after soft or palatal consonants and not before
hard consonants (¼ pr
ˇ
ehla
´
ska)
lide
´
‘people’ Pl (PSl ljud{je, Rus lju
´
di)
cizı
´
‘foreign’ (PSl tjudj{j{, Rus c
ˇuz
ˇo
´
j)
klı
´
c
ˇ
‘key’ (PSl kljuc
ˇ
{, Rus kljuc
ˇ
)
lı
´
bit se ‘to admire’ (PSl ljubiti, Rus ljubı
´
t
0
sja)
Allophonic In the languages with extensive phonemic palatalization,
raising of front vowels before soft consonants is usually noticeable at
least in stressed syllables. In Russian, Belarusian and Polish /e/ is
realized here as raised [e] (vs [f] otherwise):
(15) Rus e
`
tot ‘this’ SgMasc [¨ft
e
t] e
`
ti ‘this’ Pl [¨et
j
I]
When a stressed non-front vowel in Russian is surrounded by soft
consonants, it is noticeably fronted:
/a/ > [a
¨
]: Rus pja
´
tyj ‘fifth’ [¨p
j
ati-j] pjat
0
‘five’ [p
j
æt
j
]
/o/ > [o
¨
]: Rus tot ‘that’ [t
c
t] vs te
¨
tja ‘aunt’ [¨t
j
o
¨
t
j
e
]
/u/ > [u
¨
]: Rus blju
´
do ‘dish’ [¨bl
j
ud
e
] vs lju
´
di ‘people’ Pl [¨l
j
u
¨
d
j
I]
ii. Backing caused by a following hard consonant: /e/ > /o/
The reverse phenomenon of backing occurs in Russian, Belarusian
and Sorbian, where /e/ has become /o/ when followed by a hard
consonant, or rather, when not followed by a soft consonant (thus
3.2 Historical evolution and modern equivalences 129
also in final position). In the case of Sorbian there is no other
context, and the affected /e/ may be of any source (PSl /
{
/, /
y
/
and /e˛/ as well as /e/):
(16a) Sorb zˇona ‘wife’ (PSl/Rus zˇ e n a
´
) zˇenic
´
so ‘to get married’;
polo ‘field’ (PSl/Rus po
´
l( j)e);
pos ‘dog’ (PSl /{/) c
´
elo ‘calf’ (PSl /e˛/)
(zeleny ‘green’ is inconsistent, see Rus/Pol below)
In the case of Russian and Belarusian this effect occurs only under
stress:
(16b) Rus se
¨
la ‘villages’ Pl se
´
l
0
skij ‘rural’ selo
´
‘village’
(PSl sel-)
zele
¨
nyj ‘green’ ze
´
len
0
‘greenery’ (PSl zelen-)
lico
´
‘face’ (PSl lice)
(16c) Bel se
¨
ly ‘villages’ Pl se
´
l
0
ski ‘rural’ sjalo
´
‘village’
zjale
¨
ny ‘green’ ze
´
len
0
‘greenery’
In these examples the source of /e/ may also be /
{
/: Rus/Bel ove
¨
s/ave
¨
s
‘oats’ (PSl ov{sy), but not /e
ˇ
/, which always remains /e/: Rus/Bel les
‘forest’.
For Polish the additional restriction is that only following hard
dentals have produced the change, so that backing does not occur
in word-final position either:
(16d) Pol
_
zona ‘wife’ (PSl/Rus zˇena)
_
zen
´
ski ‘feminine’
zielony ‘green’ zieleni ‘green’ PlMasc
But: pole ‘field’
In Polish the only source of the backed /o/ is /e/, not /
y
/or/
{
/,
which remain /e/: sen ‘sleep’ (PSl syny), pies ‘dog’ (PSl p{sy).
iii. Effects caused by closed syllables and word boundaries
Phonemic In Ukrainian, apart from /e
ˇ
/, the other sources of the
new high front /i/ were /o/ and /e/ in (new) closed syllables. This
change is usually interpreted as the result of compensatory lengthen-
ing on this vowel for the loss of the jer in the next syllable. The interim
stages were diphthongs of the [u
˘
c
] type, seen in some texts and still
in some dialects (10.3.2). In addition to providing new sources of /i/,
this development produced new alternations of /i/ /o/ or /e/:
130 3. Phonology
(17) Ukr nis no
´
sa ‘nose’ NomSg GenSg (PSl nos-);
sim semı
´
‘seven’ NomSg GenSg (PSl sedm-)
Proto-Slavic /{/ did not participate in this change, remaining /e/:
pes ‘dog’, den
0
‘day’; nor did bookish or foreign words participate.
In Polish, with the loss of phonemic vowel length (3.2.4.2), long
/
o/ became /u/ in closed syllables, mainly where the following
consonant was a voiced phoneme, even though, if this consonant
was an obstruent, it would be phonetically devoiced in final posi-
tion. The etymology and alternation are preserved orthographi-
cally by the letter o
´
:
(18a) Pol gro
´
d (/grud/, [grut] ) grodu ‘castle’ NomSg GenSg
pole po
´
l ‘field’ NomSg GenPl
mo
´
c moge˛ ‘be able’ Inf 1Sg
Later the new /u/ would be extended to open syllables by analogy:
(18b) Pol kro
´
l kro
´
la ‘king’ NomSg GenSg
pio
´
ro pio
´
r ‘pen’ NomSg GenPl
And the same change has occurred in Upper Sorbian, where long
/o¯/ has become the falling diphthong [¨uo] (orth. o
´
):
(19a) USorb hro
´
d hroda ‘castle’ NomSg GenSg
And similarly long /
e/ has become [¨ie] (orth. e
ˇ
):
(19b) USorb pe
ˇ
c pjecy ‘stove’ NomSg GenSg.
Allophonic In Ukra inian an d Belarusian the high vowels are re-
duced to semivowels when they occur initially after a word-final
vowel within the same syntagma, especially where there is a clitic
involved, including a prefix. /i/ and /y/ become [i 9], and /u/ becomes
[u
˘
]. Both languages indicate [u
˘
] in the orthography, Ukrainian with
the letter B (v, normally for /v-w/), Belarusian with the special letter
yÆ (uÆ). Ukrainian also indicates [i 9] with the letter u
˘
( j, normally
for /j/ ), while Belarusian leaves i (i.e. with no orthographic marker
of reduction):
(20a) Ukr Iva
´
na do Jva
´
na ‘(to) Ivan’ [GenSg];
ide
´
vona
´
jde
´
‘(she is) going’
uc
ˇ
y
´
telja do vc
ˇ
y
´
telja ‘(to) the teacher’ [GenSg];
umı
´
ju ne vmı
´
ju ‘I can(not)’
3.2 Historical evolution and modern equivalences 131
(20b) Bel Iva
´
na do Iva
´
na ‘(to) Ivan’ [GenSg];
idze
´
jana
´
idze
´
‘(she is) going’
vho
´
radze zˇyli u
˘
ho
´
radze ‘(they lived) in the town’;
uva
´
zˇyc
0
zau
˘
va
´
zˇyc
0
‘respect’ [Imprfv Prfv]
iv. Backing caused by the preceding consonant
The effect noted above (b(ii) ) of /e/ > /o/ appears in Ukrainian in a
different context, namely following mainly a palatal consonant.
This effect is extended also to some positionally palatalized con-
sonants, regardless of stress or the following consonant. It is
believed that this was the initial Common East Slavic (‘‘Old
Russian’’) step, related to the unpaired status of the palatals and
of /e/-/o/. Belarusian and Russian then moved beyond to different
rules as their vowel systems changed in relation to stress:
(21) Ukr c
ˇ
olovı
´
k ‘man’ (Rus c
ˇ
elove
´
k);
zˇ ona
´
tyj ‘married’ (Rus zˇ ena
´
tyj);
joho
´
‘he’ [GenSg] (Rus jego
´
);
s
0
oho
´
dni ‘today’ (Rus sego
´
dnja)
c. The effects on vowels of suprasegmental developments, especially stress
i. Stress
Russian and Belarusian are the most unusual within the group in
respect of stress, for they developed such an intense degree of stress
that most unstressed vowels have become markedly reduced in qua-
lity. This is the phenomenon known as a
´
kan
0
e in Russian, meaning
specifically the non-distinction of the unstressed low back vowels
/a/ and /o/. The unstressed front vowels /e/ and /i/ are also not
distinguished, a phenomenon known in Russian as ı
´
kan
0
e. The high
/u/ and [y] are not affected. There is a very large range of varieties of
this phenomenon across the Russian and Belarusian dialects (10.3),
so we shall restrict ourselves to the two standard languages. In both
languages, unstressed /o/ and /a/ merge after hard consonants into [
e
]
or [ˆ] (by position) in Russian, and into [a] in Belarusian (‘‘strong
akan
0
e’’). Belarusian reflects this effect in its orthography. Russian
does not, and its orthography shows the (morpho)phoneme:
(22) Rus golova
´
‘head’ [g
e
lˆ¨va]
trava
´
‘grass’ [trˆ¨va]
ko
´
rotko ‘short’ Adv [¨k
c
r
e
tk
e
]
Bel galava
´
[˜ala¨va] trava
´
[tra¨va] ko
´
ratka [¨k
c
ratka]
132 3. Phonology
After soft consonants /o/ and /a/, and also /e/, become [I] in both
languages in post-tonic position. This also happens in Russian in all
pretonic syllables, but Belarusian has [a] only in the first pretonic
syllable:
(23) Rus /e/: zele
¨
nyj ‘green’ [z
j
I
"
l
j
c
ni-j]
/o/: selo
´
‘village’ [s
j
I
"
l
c
]
/a/: pjatı
´
‘five’ [GenSg] [p
j
I
"
t
j
i];
Bel zjale
¨
ny [z
j
a
"
l
j
c
ni-] sjalo
´
[s
j
a
"
l
c
] pjacı
´
[p
j
a
"
t
s
j
i]
In Bulgarian there is a degree of change to unstressed vowels,
mainly in informal style, involving a loss of quality distinction
towards the narrower of a pair. Unstressed /o/ tends to be raised
to /u/ and /a/ to /
e
/, and to a lesser extent also /e/ to /i/. This last is
considered non-standard.
(24) Blg voda
´
‘water’ [vU
"
da]; blagodarja
´
‘thank you’
[blagUdar¨j
e
]
kraka
´
‘foot’ [Pl] [kr
e
¨ka], ı
´
mam ‘I have’ [¨im
e
m];
bera
´
‘I take’ (Colloq) [bI¨r
e
] (Standard [bf¨r
e
])
In Slovenian dialects there is a similar raising of unstressed /o/ to /u/,
/e/ to /i/ and /a/ to /
e
/, but the standard language retains the quality
of the underlying stressed vowel.
Shifts of stress position have sometimes produced quality
changes in the newly stressed vowel, initially prosodic changes
such as a new rising pitch, which is then resolved into a new quality.
In Proto-Slavic retraction of stress from a weak stressed jer pro-
duced just such a new rising pitch, known as the ‘neo-acute’ (see
also below, 3.2.4). This is reflected in various ways, among them
quality changes, e.g. NRus /o/ and /e/ became high-mid [o], [e], and
then the rising diphthongs [u
˘
c
], [i
˘
f], all still in evidence in the
dialects (e.g. stol ‘table’ from PSl *stoly
´
is [stu
˘
c
l] ). The unexpected
prothetic /v/ in Rus vo
´
sem
0
‘eight’ (with [u
˘
c
] > [v
c
] ) (see above) is a
rare reflection of the neo-acute in the standard language. And the
Ukrainian development of /o/ and /e/ > /i/ noted above is probably
a continuation of the same process, extended to any context before
a weak jer regardless of stress, so that we are dealing simply with
compensatory lengthening. However, when the source was indeed
the neo-acute, there is an additional effect: the inserted /o/, /e/ in the
polnoglasie forms are lengthened to /i/ only where the lost jer had
3.2 Historical evolution and modern equivalences 133
been stressed, e.g. Ukr horo
´
d ‘town’, from PSl *go
&
rdy, but holı
´
v
‘head’ [GenPl] from PSl *golvy
´
.
Later changes of this sort are discussed below (3.2.4) as essen-
tially suprasegmental in result as well as motivation.
ii. Vowel quantity/length
The common quality changes related to vowel length are in evidence
in several areas, of which the most interesting are in West Slavic,
where all but Sorbian have gone through processes of diphthongiza-
tion and change of monophthongal quality. West Slavic redeveloped
quantity as a phonemic feature after its near loss in Proto-Slavic (see
below, 3.2.4). Quantity remains a prominent feature in Czech and
Slovak, but has been lost in Polish and Sorbian. In Czech and Slovak
we observe the conversion of some long vowels into diphthongs
which function within the system as long vowels, e.g. in Czech long
/
u/ (from either /u/ or /o˛ /) > /ou/ [
c
u
˘
]:
(25) PSl dubyky ‘oak’ [Dim] > Cz dubek > doubek
PSl neso˛ t{ ‘they carry’ > Cz nesu > nesou
In the Bohemian dialects, including Common Czech (2.4.5, 10.4.3),
the front vowel has done the same, i.e. long /ı¯ / > /ei/ [ei
˘
]:
(26) PSl byti ‘to be’ > Common Cz bei
˘
t (but Standard Cz by
´
t)
There have also been shifts in quality, partly of a push–pull nature,
e.g. in Czech long /
o/ has taken the place of the older long /
u/ (cf. (25) ):
(27) PSl stoly ‘table’ > Cz st
ol >stu
˚
l
PSl mozˇes
ˇ
i ‘you [Sg] can’ > Cz m
ozˇes
ˇ
>mu
˚
zˇes
ˇ
Long /
e/ has similarly become long /ı¯ /:
(28) PSl kve
ˇt
yky ‘flower’ [Dim] > Cz kv
etek > kvı
´
tek
The literary standard shows many exceptions to the last develop-
ment, with retained /
e/:
(29) PSl nesti ‘to carry’ > Cz ne
´
st
PSl le
ˇ
tati ‘to fly’ > Cz le
´
tat
PSl le
ˇ
to ‘year’ > Cz le
´
to
Phonological contexts appear to be mainly responsible, notably
after /l/. In the colloquial language and Bohemian dialects, such
forms have more regularly given /ı¯ /: ny
´
st, lı
´
tat, lı
´
to.
134 3. Phonology
Slovak has made even more use of diphthongs. The normal (but not
the only) long partner of /a/, /e/ and /o/ is a diphthong, and a diphthong
is also a possible partner of /u/. Only /i/ has no diphthongal partner:
(30) Slk pa
¨
t’ ‘five’ piaty ‘fifth’ (Cz pa
´
ty
´
)
nesu
´
‘they carry’ niest’ ‘to carry’ (Cz ne
´
st)
moja ‘my’ [FemSgNom] mo
ˆ
j (Cz mu
˚
j) [MascSgNom]
See further below (3.2.1.7) on diphthongs.
iii. Tone/pitch
The effect of tonal changes on vowels – other than that of the
neo-acute mentioned above – is always still connected to the new
prosodic shape, and so we leave it until discussion of the supraseg-
mental features (3.2.4).
3.2.1.6 Contraction
While the status of /j/ changed with the syllable restructure (see 3.2.1.7 below), it
remained – in articulatory terms – in many areas a glide or semivowel rather than a
full consonant, at least in the intervocalic context. In this context it could in some
systems be so weak that it could disappear, and the area of such developments is an
interesting one for the Slavic group as a whole. It appears that the centre for the loss
of this /j/ was the Czech area, with Russian and Bulgarian at the outer extremes, so
that it may be regarded as starting in the former Czech area and weakening as one
moves away from this centre (Marvan, 1979). When /j/ disappeared, the typical
result was a long vowel rather than hiatus, though the latter would have been the
direct interim result. The simplest statement of the process is that the new vowel
quality was that of the vowel after the /j/, which could then be affected by the
(morpho)phonemics of a given word. Examples from Czech, Russian and
Bulgarian illustrate the effects (31a) (Russian written ‘vowel þ e’ ¼ /Vje/):
(31a)
PSl Cz Ru Blg
moja ‘my’ [NomSgFem] ma
´
moja
´
mo
´
ja
moje [NomSgNeut] me
´
moe
¨
mo
´
e
mojego [GenSgMasc] me
´
ho moego
´
–
dobraja ‘new’ [NomSgFem] dobra
´
do
´
braja (dobra
´
¼ short)
zˇenojo˛ ‘woman’ [InstrSg] zˇenou (< -u) zˇeno
´
j (< -o
´
ju)–
znajet{ ‘he knows’ zna
´
zna
´
et zna
´
e
znajo˛ t{ ‘they know’ znajı
´
zna
´
jut zna
´
jat
pojasy ‘belt, waist’ pa
´
spo
´
jas po
´
jas
stojati ‘stand’ Inf sta
´
t stoja
´
t
0
stoja
´
[1Sg]
3.2 Historical evolution and modern equivalences 135
Between these two stand mixtures ( (31b–e), with contracted forms on the left of the
colon, uncontracted on the right):
(31b) Pol dobra,
_
zona˛ , pas, stac
´
: moja, moje, mojego/mego, znaje
(also Sorb)
Ukr moho
´
, do
´
bra: the rest as Rus (incl. mojemu
´
[DatSgMasc] )
B/C/S mo
&
ga, do%bra, zˇ e
`
n
om, zna
&
(m), pa
&
s: mo
`
ja, mo
`
je, zna%j
u, sta
`
jati
Sln as B/C/S, except: sta
´
ti
Interesting are Macedonian and Belarusian. In Macedonian the /j/ has gone, but
the vowels have often not fused, a result found in Bulgarian only in 3Sg verbs,
e.g. zna
´
e:
(31c) Mac moe, znae, znaat, ovaa ‘that’ [NomSgFem]: moja, pojas, stojat
Belarusian is normally like Russian, but has sometimes retained the /j/ and lost the
second vowel:
(31d) Bel
maja
´
, maje
¨
, majho
´
, majmu
´
Similarly, Slovak, normally like Czech, has for the same grammatical forms:
(31e) Slk moja, moje, mo
ˆ
jho, m
ˆ
ojmu
The Macedonian type of hiatus occurs also in North Russian dialects, where it
represents a relatively later loss of intervocalic /j/. All the above stages may be seen
in these dialects: hiatus ( [¨znaft] ), hiatus with assimilation ( [¨znaat] ), fusion into
long ( [znaø t] ), and ultimate short ( [znat] ) (10.3.3).
3.2.1.7 Diphthongs
Proto-Slavic /j/ became freer in distribution after the syllable restructure. In
the open-syllable period it could occur only before vowels, having derived from
earlier ‘/i/ þ vowel’. After the reappearance of closed syllables, /j/ could appear in
(syllable-)final position:
1. as a weak jer disappeared:
(32) PSl moj{ ‘my’ > moj
2. as unstressed /i/ (especially post-tonic) after a vowel was reduced:
(33) PSl zna
´
i(te) ‘know’ [2Sg/Pl Imper] > zna
´
j(te)
Thus /j/ became a phoneme as free as any other consonant and was able to form
large numbers of diphthongs, since structurally a final post-vocalic /j/ was not
distinguished from [i 9].
136 3. Phonology
This structural possibility was easily extended to the back position in those areas
where /w-v/ was phonetically bilabial (see below, 3.2.2.4), giving new diphthongs
of ‘vowel þ [u
˘
]’. Where /w-v/ was phonetically labio-dental, such diphthongs
were unlikely to develop, though they could arise later from other sources (e.g.
Cz /
u/ > /ou/). These falling diphthongs are therefore common throughout the
Slavic area, and with any nuclear vowel.
The sequences /ie/ and /uo/ may occur as either falling or rising diphthongs.
The falling type – /ı
´
e/ and /u´ o/ – occur only in Sorbian, where they reflect long /e/
(or /e
ˇ
/) and /o/, respectively. The rising type – /i 9e/ and /u
˘
o/ – are also the result of
the reallocation of vowel length, as noted in 3.2.1.6.
3.2.2 Development of the Proto-Slavic consonant system
The late Proto-Slavic consonant system is shown in table 3.4 (see also table 1.3).
Forms in brackets may or may not have been still present at the late Proto-Slavic
stage. In other words, they may already have been dialectal: the labial forms with
/l’/ (/bl’/ etc.), where they existed, were morphophonemes in alternation with the
simple labial, the palatal /l’/ being in effect the reflex of a following /j/ (1.3.1.4). The
palatal stops were most likely unchanged, their further development being a major
feature in the typology of the sound systems. The quality of the voiced velar and the
labial fricative are almost certainly old dialectal features, so that they are marked
here (bracketed) as variants. And the voiced affricate and soft (palatalized) dental
fricatives probably also existed as such only in some areas. All of these points will
be taken up below.
Table 3.4. The late Proto-Slavic consonant system
Labial Dental Palatal Velar Glottal
Plosive p (pl’) t (t’) k
b (bl’) d (d’)
– (g)
Nasal m (ml’) n n’
Fricative s (s
0
)s
ˇ
’x
– (v) (vl’) z (z
0
)zˇ ’–(˜)
– (w) j
Affricate c
0
c
ˇ
’
(dz
0
) (dzˇ ’)
Liquid r r’
ll’
3.2 Historical evolution and modern equivalences 137
In this section we focus on the development of:
1. palatal consonants (the result of fusion with /j/ or of velar palatalization);
2. positionally palatalized consonants and the status of the feature
[Palatalized];
3. the voiced velar stop/fricative;
4. the labial fricative;
5. the quality of /l/; and
6. the syllabicity of the liquids /r/ and /l/.
3.2.2.1 Palatals
Velars The process of the appearance of the palatal consonants was relatively
early Proto-Slavic (for details see 1.3.1.4–5). Throughout the Slavic area /s
ˇ
/ and /zˇ /
had resulted from (1) /sj/-/zj/ and (2) /x/-/g/, followed by either /j/ or a front vowel,
and /c
ˇ
/ from /k/ in the same contexts (the ‘1st Palatalization of the Velars’, PV1).
Also throughout, the dental affricate /c
0
/ belongs to the same process, deriving
primarily from /k/ followed by the new front vowels formed during the mono-
phthongization of the diphthong /oi/ as syllables were opened (the ‘2nd
Palatalization of the Velars’, PV2). This also occurred in some other contexts
in which the preceding vowel was operative (the ‘3rd Palatalization of the
Velars’, PV3).
More uncertain as to its areal distribution is /dz
0
/, which derived from /g/ in the
last two contexts (PV2 and PV3). The end reflexes are often /z
0
/, even though the
voiceless partner never ends up as /s
0
/. This suggests that the quality of /g/ varied
(see below).
West Slavic differed from East and South Slavic in the realization of PV2 and
PV3 for /x/ only, the reflex in West Slavic being palatal /s
ˇ
/ as opposed to the dental
/s
0
/ elsewhere. A common example of PV3 is the Proto-Slavic root *v{x- ‘all’ (the jer
was strong in the NomSgMasc, weak elsewhere) (see 1.3.1.5 for other examples):
(34) PSl v{x- ‘all’
SSl: Blg vsı
´
-c
ˇ
k- Mac si- (loss of /v/) B/C/S s(a)v- (inversion) Sln v([
e
])s-
ESl: Rus, Ukr v(e)s(
0
)- Bel uv( e) s (
0
)-
WSl: Cz vs
ˇ
e-ch(e)n- Slk vs
ˇ
e-t(o)k- Pol wszy-st(o)k- Sorb ws
ˇ
-(o
´
n-)
This is the source of what structurally should be a palatalized /s
0
/ in East and South
Slavic, which would match /c
0
/ and either /dz
0
/or/z
0
/ as the earliest palatalized
phonemes. Like the velar area, this is a set of three phonemes which might pair up in
either way (hard soft was not yet a possibility): /s
0
/-/z
0
/, with /c
0
/alone,or/c
0
/-/dz
0
/,
with /s
0
/ alone. In either case there was a systemic imbalance which might lead
138 3. Phonology