Progressive and regressive vowel assimilations
In Turkic languages, sound assimilations usually operate
progressively (i.e. the preceding syllable influences the following with
respect to certain sound features.) Sound assimilations typical for
Turkic languages are palatal harmony and labial harmony. Palatal harmony
renders an opposition front vs. back [[+ or -] front]. In Uyghur for
example, there is a plural suffix with a front (palatal) variant -ler
and a back (velar) variant -lar. The appropriate variant is selected
according to the stem. Hence, the plural of oy 'house' is
oyler, the plural of oy 'thought' is oylar. Labial harmony
renders an opposition round vs. unround [[+ or -] round]. In Kirghiz,
the plural suffix has the unround variants -Ler and -Lar and the round
variants -Lor and -Lor. The plural of ay 'month' is aylar, the
plural of oy 'thought' is oylor. Although palatal and labial
harmony are often collectively referred to as "vowel harmony",
palatal harmony affects consonants as well. In Uyghur, the dative of oy
'house' is oyge; the dative of oy 'thought' is oyya.
(1) The configuration of palatal and especially labial harmony differs
considerably between the various Turkic languages. What is common to all
varieties of Turkic is that these sound harmony processes operate
However, in some Turkic varieties there are also regressive vowel
assimilations. (2) These are especially striking in the Southeastern
branch of Turkic and are often referred to as umlaut. Two different
types of umlaut phenomena, as well as two unrelated but equally striking
types of vowel reduction will be the topic of this contribution.
I-umlaut is a vowel assimilation phenomenon typical for Uyghur. It
affects the low unrounded vowels /a/ and /e/ in initial open syllables.
In this position, /a/ is raised and fronted to a closed /e/ if a primary
/i/ (3) occurs in the subsequent syllable. For example, the Turkic word
sariy 'yellow' is represented as seriq in Uyghur (UTIL 3:
650), T yarim 'half' as Uyg yerim (UTIL 6: 694), T yani
'new' as Uyg yeni (UTIL 6: 723-724) and T yasil
'green' as Uyg yesil (UTIL 6: 709). /e/ in open initial
syllables is raised to /e/ as well, but unlike /a/, there need not be a
primary /i/ in the following syllable. Hence, Uyghur has kerek
'necessary' from T kergek (with an intermediate form kerek,
Clauson 1972: 742) and keler 'will come' from the verb stem
kel- 'to come'.
The Uyghur converb form yetip may be derived from either yat-
'to lie' and yet- 'to reach', as the converb suffix
contains a primary /i/. On the other hand, Uyghur has kelidu
'comes' (with the shift /e/ > /e/) but qalidu
'remains' (without the shift /a/ > /e/), as the /i/ of the
aspect suffix idu is a secondary development from /e/ and /a/ (*keledur,
I-umlaut is fully productive in that it is not restricted to
lexical units, but also occurs under suffixation if the necessary
conditions for umlauting are fulfilled. Loans do not undergo umlauting
consistently; hence Uyghur has hell 'a moment ago, in a
moment' (UTIL 5: 581) from Arabic hali 'present' with
umlaut but hall 'its state' (Uyg hal 'state' from
Arabic hal) without (UTIL 5: 454-457). An epenthetic /i/ may trigger
umlaut, as in meyiz 'kernel, marrow' < *mayiz < P magz.
The Uyghur i-umlaut resembles the i-umlaut of Germanic and Eastern
Middle Iranian. It might have been triggered by an Indo European
Similar phenomena are also reported for the Uzbek dialects around
Namangan, see Wurm 1959: 499 and Resetov and Saabdurahmanov 1978: 38.
Sporadically, it occurs in other varieties of Turkic as well. The
Turkish lexemes yesil 'green' and yeni 'new' display
the same vowel development, (5) while the neighboring Azerbaidjani has
the original yasil for the former (but also yeni for the latter).
Standard Uzbek has sekin 'slowly' (OTIL 2: 33-34; likewise
Uyghur sekin, UTIL 3: 665) from Arabic sakin 'quiet'. Kazakh
bari 'all', originally bar 'existing' with the
possessive suffix, also displays fronting of /a/ under the influence of
a subsequent /i/. Erratic forms like these--although probably resulting
from the same process--are purely lexical, while in Uyghur the process
is productive (i.e. the mechanism remains active under speech). This is
even true within the lexical stem. For example, the intensive derivation
of yesil, which is formed by a reduplication of the first syllable, is
yapyesil 'very green' (UTIL 6: 405) in Uyghur, that is, under
the process of derivation the original form yasil is taken as the basis.
Conversely, the corresponding Turkish form is yemyesil, with the
innovative form as the basis for derivation. (6)
Although i-umlaut usually is restricted to open syllables, there
are also instances of umlaut in closed syllables: yessi 'flat'
(UTIL 6: 708) from T yasi (Clauson 1972: 973-974). One may speculate
that the umlauting occurred prior to the lengthening
("gemination") of /s/. (7) Nadzip (1968: 797) records this
word as yese, without consonant lengthening, but with an additional
vowel lowering in the second syllable (see below).
U-umlaut is the labialization of low vowels in initial syllables
under the influence of a round vowel in the subsequent syllable, i.e. a
partial regressive assimilation that adds the feature [+round] to the
vowel. This phenomenon has been noted by Pritsak (1959a: 538) for modern
Uyghur, see also Johanson 1998: 93. It is extremely frequent, though not
productive, in this language and may affect both words of Turkic origin
and loans. There is also a limited set of lexemes in which forms with
u-umlaut have spread beyond the limits of Uyghur.
An example for this is the Turkic word etuk 'boot'
(Clauson 1972: 50) which appears in Uyghur as otuk (UTIL 5: 796), in Ili
Salar as o:tux (Yakup 2002: 148), in Kirghiz as otuk (Judaxin 1965:
602), in Altay Turkic as oduk (Baskakov and Toscakova 1947: 118), in
Qumandi (8) as oduk or odek (Baskakov 1972: 238), in Tuba (Baskakov
1966: 141) and Quu as oduk and odok (Baskakov 1985: 182) and in Xakas as
odik 'shoe' (Baskakov and Inkizekova-Grekul 1953: 133). On the
other hand, Uzbek, the closest relative of Uyghur, has etik (OTIL 2:
455) with the frequent delabialization of /U/ in non-initial syllables.
The same form is attested for Kazakh (Bektaev 2001: 166), Karakalpak
(Baskakov 1967: 903, entry sapogi) and Kumyk (Bammatov 1969: 379). That
the /u/ of T etuk has been delabialized at a very early stage in some
Turkic varieties can be seen from the Koman (fourteenth century) form
etik (Gronbech 1942: 969). Further forms without umlaut are Tatar itek
(TatRS 180), Tuvan idik 'shoe' (Tenisev 1968: 201) and Turkmen
a:dik (TDS 824). Siberian variants like odek and odok display a vowel
lowering in the second syllable (see below).
Another example of a word with spread u-umlaut is Turkic eruk
'stone fruit' (Clauson 1972: 222), which appears in Uyghur as
oruk 'apricot' (UTIL 5: 807). Uzbek has umlaut as well: orik
(OTIL 2: 510-511) with later delabialization of the /u/ of the second
syllable. The fact that in this lexeme the u-umlaut is quite old is
shown by the fact that oruk <'wrwk> is attested as early as
in the Baburname (e.g. 3r13) which was finished in 1530. (9) Variants
with umlaut are also present in Kirghiz oruk (Judaxin 1965: 600), Kazakh
orik (Bektaev 2001: 374) and Ili Salar orox (Yakup 2002: 148). (10) The
variant without umlaut, erik, occurs among other things in Koman
(Gronbech 1942: 92), Karakalpak (Baskakov 1967: 14, entry abrikos),
Kumyk (Bammatov 1969: 376), Karachay-Balkar (Tenisev and Sujuncev 1989:
770, meaning: 'plum') and Turkish (meaning: 'plum').
U-umlaut is less spread in the Turkic word caruq 'a rough,
home made boot' (Clauson 1972: 428), which appears in Uyghur as
coruq 'sandal' (UTIL 2: 686) and Karachay-Balkar as curuq
'boot' (Tenisev and Sujuncev 1989: 738), the latter with total
assimilation, but in Uzbek as cariq (OTIL 2: 379) without umlaut (cf.
The Turkic verb egsu- 'to be/become
defective/deficient//lacking' (Clauson 1972: 117) displays u-umlaut
in the Baburname: oksu- 'to lack' (11) (in the word form
oksumedi <'wkswm'dy>, 14r6]). The form with umlauting
survives both in Uyghur (oksu-, UTIL 5: 837) and Uzbek (oksi-, OTIL 2:
The Turkic word gamin 'reed' (Clauson 1972: 628-629) is
gomus in Uyghur (UTIL 4: 291) and xomus in Yakut (Slepcov 1972: 496)
with an intermediate *qamus (12), while Babur Turkic has gamin
(288r7) and Uzbek gamin (OTIL 2: 547).
The Mongolic word serigun 'cool' (Khalkha seruun) appears
in Yakut as soruun (Slepcov 1972: 337) and in Uyghur as sorun (UTIL 3:
605), but in Uzbek as sarin (OTIL 2: 23), in Altay Turkic as seruun
(Baskakov and Toscakova 1947: 128), in Tuvan as serun (Tenisev 1968:
374) and in Turkish as serin. Xakas has the variants soron and siren
(Baskakov and Inkizekova-Grekul 1953: 197, 188), i.e. one variant with
and another without umlaut. The underlying form copied into Turkic from
Mongol seems to have been *seruun.
Besides these very few lexical items, u-umlaut seems to be more or
less restricted to Uyghur. Here is a list with examples for u-umlaut in
Uyghur which is not paralleled in other Turkic languages, with the
corresponding Old Turkic forms. Uzbek as the closest relative of Uyghur
is quoted as well.
bosuk 'cradle' (UTIL 1: 515) < T besuk (Clauson 1972:
380); Uzbek besik (OTIL 1: 108).
tosu- 'to carry' (UTIL 2: 248) < T tasu- (Clauson
1972: 561); Uzbek tasi- (OTIL 2: 138).
tomur 'vein, artery' (UTIL 2: 285) < tamur (13) < T
tamir (Clauson 1972: 508); Uzbek tamir (OTIL 2: 200).
toms- 'to know' (UTIL 2: 291-292) < T tams- (Clauson
1972: 516); Uzbek tani- (OTIL 2: 117-118).
tosuk 'hole' (UTIL 2: 360-361) < tesuk (Clauson 1972:
563); Uzbek tesik (OTIL 2: 169). (14)
tomur 'iron' (UTIL 2: 367) < temur < T temir
(Clauson 1972: 508); Uzbek temir (OTIL 2: 158).
xotun 'woman, wife' (UTIL 2: 855-856) < T xatun
(copied from Sogdian, Clauson 1972: 602-603); Uzbek xatin (OTIL 2: 334).
gozuq 'peg, stake' (UTIL 4: 248) < gazuq < T gazyuq
(Clauson 1972: 682); Uzbek gaziq (OTIL 2: 592-593).
gosuq 'spoon' (UTIL 4: 258) < T gasuq (Clauson 1972:
671); Uzbek gasiq (OTIL 2: 607).
goyun 'melon' (UTIL 4: 265) < T gayun (Clauson 1972:
611); Uzbek gavun (OTIL 2: 591); the Baburname has gavun
ocuq 'open, clear' (UTIL 5: 619-620) < T acuq (Clauson
1972: 22); Uzbek aciq (OTIL 1: 552).
oruq 'emaciated, skinny' (UTIL 5: 637) < T aruq
(Clauson 1972: 214); Uzbek ariq (OTIL 1: 542).
ozuq 'provisions' (UTIL 5: 645-646) < T azuq (Clauson
1972: 283); Uzbek aziq (OTIL 1: 524).
osuq 'more than' (UTIL 5: 651) < T *asuq; Uzbek asiq
(OTIL 1: 555).
yopurmaq 'Blatt' (UTIL 6: 580) < T yapuryaq (Clauson
1972: 879 880); Uzbek yapraq (OTIL 2: 484-485) (15).
yoruq 'cover' (UTIL 6: 583) < *yapuq (16) < T yapip
(Erdal 1991: 215); Uzbek yapiq (OTIL 1: 260).
yoru- 'to become bright, to shine' (UTIL 6: 591-592) <
T yaru- (Clauson 1972: 956); Uzbek yari- (OTIL 1: 261).
yoruq 'light, bright' (UTIL 6: 588) < T yaruq (Clauson
1972: 962 963, Erdal 1991: 255); Uzbek yaruy (OTIL 1: 262). In modern
Uyghur, there is a contrast between yoruq 'light, bright'
(with u umlaut) and yeriq 'gap' (UTIL 6: 693, with i-umlaut).
The earliest occurrence of the latter (eleventh century) is in the form
yaruq (Erdal 1991: 255). However, the immediate predecessor for the
Uyghur form yeriq must have been *yariq with an unrounded vowel in the
second syllable. It is not clear why and at which stage this vowel had
been delabialized in an environment in which round vowels are usually
preserved. One contributing factor might be a late reflex of the vowel
length in the Old Turkic verb stem ya:r- 'to split' (Erdal
1991: 255). Note that this stem is preserved in modern Uyghur as yar-
(UTIL 6: 438), while the verb stem underlying yoruq is yoru-. The Uzbek
form yaruy--with the round vowel (irregularly) preserved and the final
consonant voiced--seems to be a Chaghatayism.
yosur- 'to conceal' (UTIL 6: 592) < T yasur- (Clauson
1972: 979); Uzbek yasir- (OTIL 2: 491).
Some Arabic and Persian loans undergo this process as well, e.g.
tonur 'tandoor' (UTIL 2: 289) < P tanur or tandur.
zorur 'necessary' (UTIL 3: 335) < A daruri.
gobul 'acceptance' (UTIL 4: 224) < A gabul.
nomus 'honor' (UTIL 5: 393) < P namus.
Uyghur hosul 'harvest' (UTIL 5: 546) seems to go back to
an intermediate *hasul, although the Arabic model is hasil (Southern
Xinjiang: ha:sil, et cetera, Jarring 1964: 115).
U-umlaut turns out to be most common in Uyghur among all Turkic
languages. Almost all Turkic words in which u-umlaut operates display it
at least in Uyghur. There are however, erratic word forms with something
resembling u-umlaut where there is no parallel in Uyghur, e.g. Tuba
sogis 'eight' (Baskakov 1966: 149), the Uyghur cognate being
sekkiz. There is no obvious etymological motivation for a word form like
sogis (*sekuz), as the common Turkic form is sekkiz (Clauson 1972: 823).
### The rounding of Common Turkic /a/ to /a/ which is often ob served in
Uzbek is not related to the regressive labial assimilation process under
description here, as it occurs irrespective of the vowel in the
subsequent syllable. (17) Hence, Uzbek has /a/ in ayiz 'mouth'
(OTIL 1: 561-562) (T ayiz, Uyg eyiz) as well as in yariq 'gap'
(OTIL 1: 262) (T yaruq, Uyg yeriq) and yaruy 'light' (OTIL 1:
262) (T yaruq, Uyg yaruq).
As it becomes obvious from the examples quoted, u-umlaut most
frequently occurs in open syllables. Hence, Uyghur has artuq
'excess, more than' (not *ortuq) (18), teksur- 'to
check' (not *toksur-) and yastuq 'pillow' (not *yostuq).
There are however examples to the contrary, like oksu- 'to
lack' from T egsu- (see above) and ocke 'goat' from T
ecku (see below).
Vowel raising in non-initial syllables
In Uyghur, low vowels in open non-initial and non-final syllables
undergo an extreme degree of reduction, resulting in a reduced central
high vowel. This phenomenon, which is fully productive, has been
labelled raising by Hahn (1991: 52). In written Uyghur, the resulting
vowel is indicated by the letter for <i>, e.g.
'the one being in our houses' instead of
*oylerimizdeki. However, this reduced vowel has a different phonological
status than the primary Uyghur /i/, which corresponds to Common Turkic
/i/ and /i/. (E.g., the reduced vowel does not trigger i-umlaut.)
Neither the reduced vowel nor the primary /i/ affect the palatal harmony
status of the word.
In connected speech, word-final low vowels may be raised as well,
e.g. /'aniyi 'oxsas/ 'like the
mother'. On the other hand, stressed vowels are never raised, for
example, neme qilarini bilmey> 'without knowing what to do'
Vowel reduction in Eastern Turki dialects is commented on by
Jarring as follows:
"Die reduzierten Vokale Bind ein sehr charakteristisches
Element im Ostturkischen. Besonders in einem schnell gesprochenen
Wortschwall Bowie bei Vokalen in schwachtoniger Stellung tritt eine
Tendenz zur Vokalreduzierung hervor, die wohl ausserhalb einer
Turksprache ohne Gegenstuck ist."
(Jarring 1933: 41)
In modern Uyghur, precisely this wortschwall has been taken as the
phonetic basis of the standard language.
Reduced vowels in non-initial syllables are common in many Turkic
languages. In many cases, the reduced vowels are marked for some
phonological oppositions like [[+ or -] back], [[+ or -] round], [[+ or
-] high] or all. Uyghur is different in this respect.
All syllables subject to intersyllabic sound harmony must have
undergone a previous stage of reduction. In this respect, the Uyghur
vowel raising is related to the intersyllabic sound harmony phenomena
common to almost all Turkic languages.
Vowel lowering in non-initial syllables
The vowel phenomena referred to so far are strikingly concentrated
on the Uyghur area although they do occasionally occur in other
varieties of Turkic as well. The phenomenon to be considered next, by
contrast, is much more spread across the Turkic world. In some lexemes,
high vowels (/i/, /i/, /u/, /u/) in non-initial syllables become
lowered. Depending on the labial harmony rules of each language, the
lowering of the vowel is often accompanied by delabialization of former
A well-known example for vowel lowering is the Turkic word kemi
'ship, boat' (Clauson 1972: 721) which survives with a closed
vowel in Oghuz: Tur/Azr germ, Tkm ga:mi (TDS 232), Ili Salar ki:mu
(Yakup 2002: 135), but with an open vowel elsewhere: Uyghur has keme
'ship, boat' (UTIL 4: 743) with the vowel modification in the
first syllable described above under the heading "I-umlaut",
Uzb kema (OTIL 1: 378), Kaz keme (Bektaev 2001: 233), Krg keme (Judaxin
1965: 371), Krc keme (Tenisev and Sujuncev 1989: 322), Kumyk genre
(Bammatov 1969: 96), Alt keme (Baskakov and Toscakova 1947: 79), Xak
kime (Baskakov and Inkizekova-Grekul 1953: 75) and Tuv xeme (Tenisev
1968: 473). The form keme is attested in the Codex Cumanicus (14th
century, Gronbech 1942: 137), which proves that the lowered vowel in
this word is quite old.
The high vowel of Old Turkic topu 'hill, top' (Clauson
1972: 436) has been lowered all across the Turkic world: Uyghur has tope
(UTIL 2: 352), Kazakh tobe (Bektaev 2001: 446), Kirghiz dobo (Judaxin
1965: 198), Altay tobo (Baskakov and Toscakova 1947: 154) and Yakut tobo
(Slepcov 1972: 393). The Oghuz forms tepe (Turkish, Azerbaidjani) and
depe (Turkmen, TDS 248) as well as Uzbek tepa (OTIL 2: 161) have their
first vowel delabialized as a result of regressive assimilation.
The Old Turkic word yugun 'bridle' (Clauson 1972: 913)
has undergone vowel lowering everywhere, too: (19) Uyg yugen (UTIL 6:
678), Uzb yugan (OTIL 2: 462), Kaz zugen (Bektaev 2001: 211), Krg jugon
(Judaxin 1965: 272), Alt uygen (Baskakov and Toscakova 1947: 171), Xak
cugen (Baskakov and Inkizekova-Grekul 1953: 324), Tuv cugen (Tenisev
Uyg beket 'station, stop' (UTIL 1: 533), Uzb bekat
'station, stop, postal station' (OTIL 1: 97), Krg beket
'postal station' (Judaxin 1965: 126) and Kaz beket
'station, postal station' (Bektaev 2001: 99) probably
developed from Old Turkic bekut, which Clauson (1972: 325) registers as
a hapax legomenon derived from the verb stem beku-'to be
The high vowel of Old Turkic sunuk 'bone' (Clauson 1972:
838-839) survives only in Azerbaidjani sumuk. It has been lowered
elsewhere: Uyghur sonek (UTIL 3: 620), Uzbek songak (OTIL 2: 97) and
suyak (OTIL 2: 90-91), Turkmen (archaic) suyek (TDS 615), Kazakh suyek
(Bektaev 2001: 424). The contracted form sook which occurs in Kirghiz
(Judaxin 1965: 659-660) and the Siberian Turkic languages Altay
(Baskakov and Toscakova 1947: 131), Tuvan (Tenisev 1968: 386) and Xakas
(Baskakov and Inkizekova-Grekul 1953: 196) hints at an older form *sonek
with vowel lowering, as *sonuk, at least in Kirghiz (20) and (Southern)
Altay (21), should have developed to *suuk (Tuvan and Xakas are less
instructive in this respect).
Old Turkic yoyun 'thick' (Clauson 1972: 904) maintains
its high vowel only in Oghuz (Tur yogun, Tkm yoyi:n, TDS 297), while
lowering has taken place elsewhere: Uyg yoyan 'huge, massive'
(UTIL 6: 595), Uzb yoyan (OTIL 2: 356), Ili Salar yoyan (Yakup 2002:
130), Kaz zuwan (Bektaev 2001: 207). Kirghiz (Judaxin 1965: 261) and
Altay (Baskakov and Toscakova 1947: 56) have joon < *yoyan. The Tuvan
(Tenisev 1968: 540) and Xakas (Baskakov and Inkizekova-Grekul 1953: 320)
form coon could have developed from either *yoyan or *yoyan.
The Uyghur word cocek 'fairytale' (UTIL 2: 709) is
probably related to Old Turkic surcuk (Clauson 1972: 845; however,
Clauson is not aware of the connection). The elision of /r/ is quite
common in Uyghur (e.g. in Uyg tot 'four' < T tort), and the
initial /s/ has been assimilated to /c/ as in Uyg cac 'hair'
< T sac. Wherever *surcuk survives, vowel lowering has occurred: Uzb
copcak (22) 'riddle, fairytale' (OTIL 2: 387, URS 551), Alt
corcok (Baskakov and Toscakova 1947: 181), Qumandi corjek, corcok,
corcok (Baskakov 1972: 269), Tuba corcok Baskakov 1966: 166), Quu
corjok, corcok, sorjok, sorsok (Baskakov 1985: 223).
Much less uniform is the picture in the following cases: Old Turkic
toyuru 'straight, right' (Clauson 1972: 473) has become toyra
(with lowering) in Uyghur (UTIL 2: 248-249) but toyri (without lowering)
in Uzbek (OTIL 2: 260). The high vowel has been preserved in Turkmen
doyri (TDS 261) and Turkish dogru. Some Kipchak languages have preserved
the high vowel as well, as Karakalpak tuwri (Baskakov 1967: 822, entry
prjamo) and Tatar turi (TatRS 558), while others have a low vowel, as
Kazakh tuwra (Bektaev 2001: 448), Kirghiz tuura (Judaxin 1965: 772-773),
Kumyk tuwra (Bammatov 1969: 321) and Karachay-Balkar tuwra (Tenisev and
Sujuncev 1989: 657). The split within Kipchak is already present in
Koman, which has both the conservative form toyru (Gronbech 1942: 247)
and the innovative variant tuwra (Gronbech 1942: 258). The Altay lexeme
tuura 'breadth, width' (Baskakov and Toscakova 1947: 159) as
well as the Xakas word toyira 'against' (Baskakov and
Inkizekova-Grekul 1953: 229) without doubt also stem from the same Old
Old Turkic tolu 'full' (Clauson 1972: 491-492) has
correspondences with a low vowel in Uyghur tola 'much, very'
(23) (UTIL 2: 273), Uzbek tola 'full' (OTIL 2: 246), Kirghiz
tolo (Judaxin 1965: 746) and Altay tolo (Baskakov and Toscakova 1947:
152). Most languages preserve the closed vowel: Karachay-Balkar tolu
(Tenisev and Sujuncev 1989: 643), Kumyk tolu (Bammatov 1969: 319),
Kazakh toli (Bektaev 2001: 444), Karakalpak toli (Baskakov 1967: 719,
entry polnyj), Tuvan dolu (Tenisev 1968: 170), Turkish dolu, Ili Salar
do:li (Yakup 2002: 80).
The Old Turkic word sogut 'tree' (Clauson 1972: 819) is
not very widespread in the modern Turkic languages. It survives with an
open vowel in Uyghur soget 'willow' (UTIL 3: 619) and Kirghiz
sogot (Judaxin 1965: 657). Turkish sogut and Turkmen sowut (TDS 605)
preserve the closed vowel. Xakas soot (Baskakov and Inkizekova-Grekul
1953: 197) could be based both on *sogut and *soget (cf. Pritsak 1959c:
609). The word does not exist in Uzbek and Kazakh, which use tal
What is yotul 'cough' (Clauson 1972: 889-890) in Old
Turkic has a low vowel in Uyghur yotel (UTIL 6: 638), Uzbek yotal (OTIL
1: 355), Kazakh zotel (Bektaev 2001: 207), Kirghiz jotol (Judaxin 1965:
266) and Yakut sotol (Slepcov 1972: 337). Altay however, the vowels of
which often behave similar to Kirghiz, has preserved the high vowel:
jodul (Baskakov and Toscakova 1947: 57). The same is true for Tuvan
codul (Tenisev 1968: 542) and Xakas cidil (Baskakov and
Inkizekova-Grekul 1953: 316).
The word for 'brain' is mene in Uyghur (UTIL 5: 279-280)
and miya in Uzbek (OTIL 1: 468). Almost all other Turkic languages in
which this word is disyllabic have a high vowel in the second syllable
(e.g. Tur beyin, see Clauson 1972: 348-349 for further examples). The
Proto-Turkic form of this word is reconstructed as *beni by Clauson.
While both Uyghur and Uzbek have lowered the second vowel, the closed
/e/ of the first syllable in Uyghur could possibly be attributed to an
i-umlaut (meni > meni) prior to the vowel lowering (the original
Turkic closed /e/ is usually not preserved in Uyghur, e.g. T ber-
'to give' > Uyg ber-).
The Turkic word bori 'wolf' (Clauson 1972: 356) displays
vowel lowering in Uyghur bore (UTIL 1: 513) and Yakut boro (Slepcov
1972: 78), but maintains the high vowel elsewhere, e.g. Uzb bori (OTIL
1: 164), Kaz bori (Bektaev 2001: 108), Krg boru (Judaxin 1965: 153), Alt
boru (Baskakov and Toscakova 1947: 34), Tuv boru (Tenisev 1968: 118),
Ili Salar bu:ri (Yakup 2002: 65).
Uyghur tulke 'fox' (UTIL 2: 407) corresponds to Old
Turkic tilku (Clauson 1972: 498-499). (24) Virtually all languages
except Uyghur preserve the high vowel: Uzbek has tulki (OTIL 2: 222),
Kazakh tulki (Bektaev 2001: 452), Kirghiz tulku (Judaxin 1965: 780),
Altay tulku (Baskakov and Toscakova 1947: 161), Xakas tulgu (Baskakov
and Inkizekova-Grekul 1953: 242), Ili Salar tulgu (Yakup 2002: 170) and
Azerbaidjani tulku. Tuvan dilgi (Tenisev 1968: 162) as well as Turk men
(TDS 649) and Turkish tilki preserve the unround vowel of Old Turkic in
the first syllable, but have /i/ in the second syllable due to labial
harmony. Only in Qumandi, forms with low vowel (tulge, tulgo, tulke) are
recorded besides forms with high vowel (tulgi, tulku) (Baskakov 1972:
Uyghur has goza 'lamb' (UTIL 4: 246) for Old Turkic quzti
(Clauson 1972: 681). A first step of vowel lowering is visible in the
Ili Salar form Goz[??] (Yakup 2002: 109), although the process is not
yet completed in this language. All other languages preserve the high
vowel, as Uzbek gozi (OTIL 2: 630), Kazakh gozi (Bektaev 2001: 288),
Kirghiz gozu (Judaxin 1965: 392) and Turkish kuzu. The Baburname,
incidentally, has gozi (164r12) with a high vowel as well.
In the items mentioned so far, vowel lowering can be observed in at
least one language besides Uyghur. However, there are a lot of instances
in which this phenomenon is restricted to Uyghur:
burya 'auger, drill, gimlet; wind instrument made from the
horn of an animal' (UTIL 1: 494 (25)) < T buryuy (Doerfer 2:
286-288; Clauson 1972: 361). Interestingly in Uzbek, a split has
occurred between the regular bury 'auger, drill, gimlet' (OTIL
1: 154; URS 85) with delabialization of /u/ in the second syllable and
burya 'pipe, horn' (URS 85); 'ancient wind
instrument' (OTIL 1: 154) with the labial vowel preserved. The
latter form is either an archaism or has been taken into the standard
language from a different dialect. The word survives with various
meanings in Kaz bury 'drill' (Bektaev 2001: 114), Krg burya
'drill' (Judaxin 1965: 160) and Ili Salar burya 'ox
horn' (Yakup 2002: 65). In Turkish, burgu 'auger, gimlet;
corkscrew; screw; peg to tighten a string in stringed instruments;
torture screw; kind of trumpet' (Redhouse 201) exists side by side
with the etymologically related item born 'pipe, tube; horn,
trumpet' (cf. below, entry mora).
Although Doerfer (2: 288) rejects a connection to Mongolian
buriy-e/buriyen 'trumpet, horn' (Lessing 1960: 149) (26) as
"lautlich unmoglich", it is my conviction that the Mongolic
item has been copied from Turkic buryu, especially if we take the
alternative spelling burege (Lessing 1960: 148) into consideration. (27)
The Tuvan form buree 'trumpet (musical instrument)'
(Pal'mbax 1953: 616, entry truba) is without doubt a loan from
buya 'deer' (UTIL 1: 502-503) < Mongolian buyu; cf.
Babur Turkic buyu (4a10), Uzb buyu (OTIL 1: 156). There is
a striking resemblance of Uyghur buya and Khalkha Mongolian buya;
however, the word is certainly not borrowed from Khalkha; rather, the
low vowel in Uyghur buya is the result of an internal Turkic
development, while the sound written in the second syllable of
the Khalkha word reflects an internal Mongolic devel opment (28) (Posch
tuge 'millet' (UTIL 2: 399) < T tugi (Clauson 1972:
478). Does not exist in Uzbek, which uses tariq or sok instead (both
teriq and sok exist in Uyghur as well).
coqqa 'summit, top' (UTIL 2: 688); Uzb coqqi (OTIL 2:
cine 'porcelain bowl' (UTIL 2: 790) < P cini
'Chinese, porcelain'; cf. Uzb cinni 'porcelain'
(OTIL 2: 370).
dora 'medicine, remedy' (UTIL 3: 94) < P daru. This
word is an example of u-umlaut and vowel lowering occurring in one word.
It is clear, that the labialization /a/ > /o/ in the initial syllable
must have preceded the lowering /u/ > /a/ in the second syllable.
Uzbek has dari (OTIL 1: 234) with a delabialized high vowel in the
second syllable. The Kazakh form duri (Bek taev 2001: 141) obviously
results from a fronting (i-umlaut) of /a/ after delabialization of the
second vowel (daru > dari > duri).
dose 'wooden chopping board' (UTIL 3: 129; Nadzip 1968:
447) < M dosi 'anvil' (Lessing 1960: 269; cf. Yakup 2005:
gosaq 'folk song' (UTIL 4: 253) < T qosuy (Clauson
1972: 671); Uzb gosiq (OTIL 2: 643).
gosna 'neighbour' (UTIL 4: 257) < T qonsti (Clauson
1972: 640) with metathesis; Uzb qosni (OTIL 2: 644); cf. Tur komsu.
geza 'horse meat sausage' (UTIL 4: 378) < *gezi <
*gazi < T gazti 'the fat on a horse's belly, a sausage made
from such fat' (Clauson 1972: 681); Uzb gazi (OTIL 2: 536). (29)
Iii this word, both i-umlaut and vowel lowering can be observed.
keke 'adze' (UTIL 4: 541 (30)) < *kerke < T kerki
(Clauson 1972: 741). This word seems to survive in Uyghur only. (31)
kotek 'stump; tear-off slip, counterfoil' (UTIL 4: 620
(32)) < T ketuk 'notch, chip' (Clauson 1972: 702). This is
a deverbal noun from ket- 'to notch/chip/gash' (Clauson 1972:
700). ketuk has become gedik 'gap; notch' in Turkish, with the
usual phonological developments. The Uyghur form kotek displays both
u-umlaut (ketuk > *kotuk) and vowel lowering (*kotuk > kotek).
kulke 'laughter' (UTIL 4: 707-708) < T kulgu (Clauson
1972: 718); Uzb kulgi (OTIL 1: 404), Kaz kulki (Bektaev 2001: 257), Krg
kulku (Judaxin 1965: 463), Xak kulki (Baskakov and Inkizekova-Grekul
1953: 96), Ili Salar kulgu (Yakup 2002: 137).
kirpe 'hedgehog' (UTIL 4: 759) < T kirpi (Clauson
1972: 737); Uzb kirpi (33) (OTIL 1: 388; URS 214), Krg kirpi (Judaxin
1965: 389), Kaz kirpi (Bektaev 2001: 259).
mora 'chimney' (UTIL 5: 175). A loan from Persian mori
'an earthen pipe joined to an aqueduct; a watercourse, channel,
gutter' (Steingass 1892: 1343), which in turn is probably a loan
from Turkic bori/boru 'pipe, tube', which is derived from Old
Turkic buryu, see above. Steingass transcribes the Persian word as muri
but the Tadjik form mori 'chimney, smoke outlet' (TRS 250) as
well as the loaned forms in Turkic bear witness to the waw-i majhtiil
(34). Iii Uyghur, the second vowel is lowered. The word exists in Uzbek
as mori 'chimney' (35) (OTIL 1: 490) and in Kirghiz as mor
'chimney', southern dialects mooru (Judaxin 1965: 532). For
the Uyghur dialects of Southern Xinjiang, Jarring (1964: 198) notes the
variants mori and moue 'chimney'. The latter form will be
commented on below.
here 'bee' (UTIL 5: 498) < T ari; with prothetic /h/;
Uzb ari (OTIL 1: 53). For unknown reasons, this lexeme has front vowels
hoyla 'courtyard' (UTIL 5: 551); Uzb havli
'courtyard' (OTIL 2: 702); Tkm hovli (TDS 714); cf. Tur avlu
'courtyard'. The Tadjik form of this item is havli
'courtyard' (TRS 494), written in Arabic letters
(TRS 646). This lexeme is very interesting in terms of etymology. It is
very likely that all the variants noted here (all carrying the same
meaning) are derived from one common origin. Jarring (1964: 122) claims
that the Eastern Turki word hojli/hojle/hojla originates from an
Arabic/Persian word . This word is in fact attested in Farsi
(Junker and Alavi 1965: 257), Afghan Persian (DRS 310, with a variant
havili ) and Tajik (TRS 646). Steingass does not mention
this word. Arabic has no word in a meaning that makes a
direct borrowing into Persian probable, but the Persian word could be
reasonably related to Arabichaula 'around' (Wehr
1985: 309) with Persian ya-yi nisbat, originally meaning something like
'surroundings', which could have acquired a more narrow
semantics of 'surroundings of a house' >
'courtyard'. However, Redhouse (98, entry avlu) and Zenker
(397, entry hwlw/hwly) claim this word to be of Greek origin, which
would relate it to Greek [alpha]v[lambda]n 'court' (cf. Latin
aula, German Aula and the English adjective aulic). The latter
possibility seems much more plausible to me. Iii this case, the Greek
word [alpha]v[lambda]n would have been borrowed into Turkic in the form
*avli. While in Turkish, the word was first accommodated to the rules of
palatal harmony (> avli) and after that to labial harmony (rendering
avlu, with the second vowel rounded under the influence of the labial
consonant /v/), the word received a prothetic /h/ in Central Asia. Iii
Uyghur, the /i/ was lowered to /[epsilon]/ and later adapted to the
rules of palatal harmony (all three stages are noted by Jarring 1964).
Regarding the question whether the word entered South East Turkic via
Persian or was borrowed into Persian from Turkic, I would argue for the
latter possibility, as innovative forms with a prothetic /h/ are typical
for Southeastern Turkic, but not so for Persian. The orthography with
, which usually points to words of Arabic-not Persian or
Turkic-origin, could result from an erroneous etymological
identification (i.e. a contamination) with Arabichaula
'around', hawali 'vicinity', et cetera.
oya 'poison' (UTIL 5: 652) < T ayu (Clauson 1972: 78)
via an intermediary form *oyu. This is another example of u-umlaut and
lowering occurring in one word. This word survives in Uzbek as ayu (OTIL
1: 564), in Kirghiz as uu (Judaxin 1965: 809), in Kazakh as uw (Bektaev
2001: 457) and in Xakas (Baskakov and Inkizekova-Grekul 1953: 127),
Tuvan (Tenisev 1968: 321) and Quu (Baskakov 1985: 181) as oo. Ili Salar
a: yu (Yakup 2002: 34) preserves the Old Turkic state.
ocke 'goat' (UTIL 5: 798) from Old Turkic ecku (Clauson
1972: 24) also combines u-umlaut and vowel lowering. Umlaut has also
occurred in Tuv osku (Tenisev 1968: 342) and Xak oski (Baska kov and
Inkizekova-Grekul 1953: 137). Most Turkic languages have delabialized
the second syllable: Uzb ecki (OTIL 2: 456-457), Kaz eski (Bektaev 2001:
167), Kkp eski (Baskakov 1967: 345, entry koza), Krg ecki (Judaxin 1965:
969), Kumyk ecki (Bammatov 1969: 380), Krc ecki (Tenisev and Sujuncev
1989: 779), Alt ecki (Baskakov 1964: 248, entry: koza); the Oghuz form
keci (with metathesis) displays the same change. Ili Salar esgu (Yakup
2002: 90) is again closest to Old Turkic. Also in ocke, the umlauting
must have preceded the lowering (ecku > ocku > ocke).
yeza 'village, countryside' (UTIL 6: 700-701 (36)) < T
yazi (Clauson 1972: 984); Uzb O. Cognates are Krg jazi 'broad,
wide' (Judaxin 1965: 212) and Xak cazi 'field, steppe'
(Baskakov and Inkizekova-Grekul 1953: 306). This lexeme contains
i-umlaut and vowel lowering: yazi > yezi > yeza.
yeya 'disaster, mayhem' (UTIL 6: 712 (37)) < T yayi
'enemy' (Clauson 1972: 898). The Uyghur form displays both
i-umlaut (yayi > yeyi) and vowel lowering (yeyi > yeya). Babur
Turkic has preserved the original form: yayi
'foe' (e.g. 218r1). Xak and Tuv caa 'war' (Baskakov
and Inkizekova-Grekul 1953: 304; Tenisev 1968: 503) are contracted from
yayi: Uzbek has the Kipchak form yaw 'enemy, desaster' (OTIL
1: 252), which is also found in Alt juu 'war, battle'
(Baskakov and Toscakova 1947: 58), Tat yaw 'foe' (TatRS 719),
Kaz zaw 'foe' (Bektaev 2001: 189), Krc jaw 'foe'
(Tenisev and Sujuncev 1989: 235-236) and Krg joo 'foe'
(Judaxin 1965: 260). The Kipchak form is attested in modern Uyghur as
well: yaw 'foe' (UTIL 6: 521 (38)). Here, two etymologically
identical forms have assumed different meanings.
Vowel lowering in final syllables is also observed in three Uyghur
yigirme 'twenty' (UTIL 6: 770) < T yegirmi (Clauson
1972: 915), cf. Tur yirmi. Corresponding forms with lowered vowels exist
in various Turkic languages, e.g. Uzb yigirma (OTIL 1: 347), Kaz ziyirma
(Bektaev 2001: 200), Kir jiyirma (Judaxin 1965: 278) and Yak suurbe
(Slepcov 1972: 352).
yette 'seven' (UTIL 6: 537) < T yetti (Clauson 1972:
886); Uzb yetti (OTIL 1: 251); cf. Yak sette (Slepcov 1972: 370).
alte 'six' (UTIL 1: 159) < T alti (Clauson 1972: 130);
Uzb alti (OTIL 1: 534); cf. Yak alta (Slepcov 1972: 39). The
non-harmonic form of this item suggests that it represents a case of
In some Uyghur dialects, vowel lowering seems to be even more
widespread than in the standard language. For the dialect of Khotan (Uyg
Xoten), forms like konel for Standard Uyghur konul 'heart',
bugen instead of SUyg bugun 'today', toxa instead of toxu
'chicken' and oval in place of oyul 'son' are
reported (Sayim 2007: 11). (39) On the other hand, some dialects have
more conservative forms, e.g buri/bori 'wolf' (Jarring 1964:
62, cf. Yakup 2005: 443; SUyg bore), toyrti 'straight'
(Jarring 1964: 309) and tuyru (Yakup 2005: 471; SUyg toyra), tulki
'fox' (Jarring 1964: 318; SUyg tulke), da: ru; do 're
'medicine, drug, spice, spices' (Jarring 1964: 81; SUyg dora),
Qozi 'lamb' (Jarring 1964: 253; SUyg qoza), qosni
'neighbour' (Jarring 1964: 253; SUyg gosna), kerki
'adze' (Jarring 1964: 167; SUyg keke), kulgu
'laughter' (Jarring 1964: 179; SUyg kulke), kirpi
'hedgehog' (Jarring 1964: 174; SUyg kirpe), mori/mo:re
'chimney' (Jarring 1964: 198; SUyg moray, mini
'brain' (Jarring 1964: 196; SUyg meie), hojli/hojl[epsilon]
'courtyard, yard, house, palace' (Jarring 1964: 122; SUyg
hoyla), ayu 'poison' (Jarring 1964: 14; SUyg oya), Ec'xu
'goat' (Jarring 1964: 39; SUyg ocke), jazi/joezi/je:zi
'country, countryside' (Jarring 1964: 151; SUyg yeza) and
joeyi/jiyi 'enemy, rebellious' (Jarring 1964: 152; SUyg yeya).
While vowel lowering of the type described above is very common in
Uyghur words, there are cases where this sound change occurs in other
Turkic languages while it is absent in Uyghur. The Old Turkic word boyuz
'throat' (Clauson 1972: 322) has a high vowel in many Turkic
languages, e.g. Tur bogaz 'throat', Kaz buwaz
'pregnancy' (Bektaev 2001: 109). Krg boot 'pregnant'
(Judaxin 1965: 144) and Alt boos 'pregnant' (Baskakov and
Toscakova 1947: 33) testify to an underlying form *boyaz as well,
otherwise these items should appear as *buuz/buus respectively. On the
other hand, Uyghur boyuz 'throat, fodder' (UTIL 1: 468) and
Uzbek boyiz 'throat' (OTIL 1: 167-168) preserve the original
structure of the word. Uzbek additionally has a word boydz
'pregnant (of animals) ' (OTIL 1: 169) of the same Turkic
origin, but possibly borrowed from Kipchak.
The Turkic word yani 'new' displays vowel lowering in a
number of Kipchak languages, e.g. Tatar yana and Kazakh zana, and also
in Yakut saia, but not Uyghur, where it is yeni. (40)
Another word for 'courtyard' (besides Uzb havli/Uyg
hoyla, see above) in South East Turkic is Uzbek qora (OTIL 2: 639)/LTyg
qoru (UTIL 4: 242). (41) The exact etymological affiliation of this word
is not clear. One feels, of course, tempted to relate it to T qorip
'enclosure' (Clauson 1972: 652; cf. Tur koru
'grove'), which is a nominal derivation from the verb qori=
'to protect' (Uzb qori-, Uyg qoru-), but this word survives
regularly as Uyg qoruq (UTIL 4: 243, entry qoruq II) and Uzb qoriq (OTIL
2: 640) with meanings like 'field; shelter, cover, et cetera'.
Both form and meaning of qora/qoru however, make a relation to the verb
qori= obvious. We could speculate that qora/qoru are the remnants of an
otherwise unattested Turkic noun *gori 'shelter, protection',
of which qori= is a zero-derivation (as in koc 'migration'
> koc- 'to migrate', tuz 'level' > tuz-
'to level' and ya :r 'cleft' > ya :r- 'to
cleave'). Iii this case, Uzb qora would display vowel lowering in
the second syllable, while Uyg qoru would represent a more conservative
In Yakut, vowel lowering occurs in a couple of inflectional
morphemes, e.g. the 3rd person singular possessive suffix in -(t)E (T
-(s)i) and the 3rd person singular preterite in -DE (T -DI).
Origin and development of umlaut and vowel lowering
In his chapter on modern Uyghur in the first volume of the
Fundamenta, Pritsak (1959: 538) explains umlaut as an influence of the
stressed second syllable on the unstressed first syllable:
"Das ursprungliche, betonte enge i/i der zweiten Silbe
verandert das breite a/a der unbetonten offenen Silbe [...] in e. [...]
Das ursprungliche betonte enge u/u der zweiten Silbe veriidert das
breite a/u der ersten offenen Silbe [...] in o/o."
However, we need not take it for granted that the first syllable
was unstressed. While Turkic stress in general is only insufficiently
explored, almost nothing is known about stress on the historical stages
of Turkic (cf. Rasanen 1949: 32-41). Umlaut does not require unstressed
initial syllables. The Germanic umlaut, for instance, affects the
stressed syllable (German Buch-'Bucher; Norwegian
bok--'boker). If Pritsak's assumption on the former stress
configuration in Uyghur is true, the situation is completely different
today. Today, the umlauted vowel often carries an extremely emphatic
stress while the subsequent vowel which originally must have triggered
the umlaut is often unstressed and sometimes reduced to nothing. (42)
The umlaut is often accompanied by a lengthening of the umlauted vowel;
hence umlaut cannot be accounted for in the context of vowel reduction.
On the other hand, there are also cases where not the umlauted vowel,
but the vowel of one of the subsequent syllables (which need not be the
umlaut trigger), is stressed. For instance, in the phrase ['je.n[??] 'je.z[??] qu'rai|I]
'let's create a new countryside', the umlauted vowels are
both stressed and lengthened, while the umlaut trigger is extremely
reduced (In yeza, the original umlaut trigger /i/ has been lowered to
/a/. However, this change does not surface in the absence of a pause,
see above). In the phrase [b+r
'tho.[integral]k e.t[integral]+|Ipt[phi][??]] 'a hole became
open', the umlaut trigger in tosuk disappears completely, while the
umlauted vowel both carries an emphatic stress and is lengthened. The
umlauted vowel in the word eciliptu does not carry the main stress but
is slightly lengthened. However, the stress does not rest on the umlaut
trigger either, which is reduced, but on the syllable subsequent to the
umlaut trigger. Iii so far as umlaut is productive today (that is,
i-umlaut which involves an extant primary /i/), it appears to be a
process completely detached from stress questions.
Besides the natural tendency of mutual interference of sounds, the
predictable and in part obligatory umlaut patterns of Uyghur may
possibly also be motivated by substratum influence of Eastern Middle
Iranian, in which similar processes are encountered as well (Sims
Williams 1989: 168). (43) Divergent stress patterns at earlier stages
may (but need not) have contributed to the rise of umlaut in Uyghur as
As we have seen above, in front vowel words the sound process goes
even further, as /e/ in open initial syllables is raised to /e/ even if
there is no /i/ in the subsequent syllable. This is probably in analogy
to a genuine i-umlaut which was already firmly established in the
Contrary to umlaut, vowel lowering affects non-initial unstressed
high syllables. At the first stage, the vowel is reduced. This reduction
in terms of quantity is accompanied by an indifferent pronunciation in
terms of quality. This stage is reflected for example in the Ili Salar
word Goze 'lamb' < T quzi: This stage is comparable with
the Uyghur vowel raising, which is also basically a reduction in
quantity and quality (but of syllables with a low vowel). This stage of
indifference (Indifferenzstufe, Johanson 1979: 70) can function as a
starting point for a recategorization of the vowel with respect to
certain phonetic features. This recategorization evolves in two steps.
Iii Uyghur at least, maybe generally, the first step of recategorization
concerns the opposition [[+ or -]open]. Originally high vowels are
lowered. This is testified to by transitional unharmonic forms like
Eastern Turki do 'r[epsilon] 'medicine', mo:r[epsilon]
'chimney', hojl[epsilon] 'courtyard' and the
Standard Uyghur word alte 'six'. Iii a second step the forms
are harmonized according to the opposition [[+ or -] back], resulting in
forms like dora, mora and hoyla as well as Yakut alta. Languages in
which low vowels are subject to labial harmony may add the feature
[+round] in a final step, e.g. Kirg tolo 'full' < tola <
tolu. Iii Uyghur, the low vowels resulting from vowel lowering are
raised again if the conditions for raising (see above) are fulfilled.
Hence, vowel lowering starts as a merely phonetic process of
reduction of high unstressed vowels, which first bears no systematic
implications. Only on a later stage, the vowel changes become
systematicized. The result of vowel lowering may be phonologically
relevant, as it can be seen from minimal pairs as Uyg gosuq
'spoon' vs. gosaq 'folk song'. However, in minimal
pairs like this it is obviously not the opposition [[+ or -] high] which
is phonologically signifcant, but the opposition [[+ or -] round], as
the opposition [[+ or -] high] is neutralized in positions where the
syllable becomes open and non final (qosuqi 'his spoon' vs.
qosiqi 'his folk song').
While vowel assimilations usually operate progressively in Turkic
("vowel harmony"), regressive vowel assimilations do occur as
well. The Uyghur i-umlaut is especially famous and well-documented as it
is a fully productive sound process and one of the most striking
features of Uyghur phonology. Additionally, Uyghur has also u-umlaut
which has received less attention in turcological literature, probably
due to the fact that it is non-productive and affects selected lexical
Both umlaut types also occur sporadically outside the Uyghur area
as a non-productive phenomenon. While in the Turkish words yesil
'green' and yeni 'new' the preceding /y-/ may have
contributed to the change, Kazakh buri 'all' and duri
'medicine' as well as Uzbek sekin 'slowly' are
unambiguous instances of stray i-umlaut.
A few lexical items with u-umlaut have gained a comparatively broad
distribution, as otuk 'boot', oruk '*stone fruit'
and various forms of *ocku 'goat'. Even these however, have
not reached very far to the west of the Turkic world.
Turkic umlaut phenomena do not seem to require the umlauted vowel
to be unstressed.
Conversely, both vowel raising and vowel lowering result from the
reduction of an unstressed vowel. While the Uyghur raising is a fully
productive process which affects low vowels, vowel lowering obviously
affects selected lexical or grammatical items only. Lowering in lexical
items is much more spread across the Turkic world than the other sound
processes mentioned in this contribution. Naturally, vowel lowering
affects originally high vowels. Typical words displaying this process
are keme 'ship', yigirme 'twenty' and yoyan
'huge, massive'. Reduction processes like these seem to have
paved the way for the emergence of vowel harmony.
While both umlaut and vowel lowering do occur across the Turkic
world, they are strikingly concentrated in the southeasternmost corner
of the Turkic world, i.e. in Uyghur, the vowel system of which seems to
have been especially vulnerable, and which obviously is especially
innovative with respect to vowel changes.
A Arabic Alt Altay Turkic Azr Azerbaidjani Kaz Kazakh Krc
Karachay-Balkar Krg Kirghiz M Mongolian P Persian SUyg Standard Uyghur T
Turkic Tkm Turkmen Tur Turkish Tuv Tuvan Uyg Uyghur Uzb Uzbek Xak Xakas
BAMRS: Bolsoj akademiceskij mongolsko-russkij slovar'. 4 Bde.,
Bammatov, Z.Z. 1969. Kumyksko-russkij slovar'. Moscow.
Baskakov, N.A. 1964. Russko-altajskij slovar'. Moscow.
-- 1966. Dialekt cernevyx tatar (Tuba-kizi). Grammaticeskij ocerk i
-- 1967. Russko-karakalpakskij slovar'. Moscow.
-- 1972. Dialekt kumandincev (Kumandy-kizi). Grammaticeskij ocerk,
teksty, perevody i slovar'. Moscow.
Baskakov, N.A. and Inkizekova-Grekul 1953. Xakassko-russkij
Baskakov, N.A. and Toscakova, T.M. 1947. Ojrotsko-russkij
Bektaev, Kaldybaj 2001. Bol'soj kazaxsko-russkij
russko-kazaxskij slovar'. Almaty.
Clauson, Gerard 1972. An Etymological Dictionary of Pre-Thirteenth
Century Turkish. Oxford.
Doerfer, Gerhard 1963-1975. Turkische and mongolische Elemente im
Neupersischen. 4 vols., Wiesbaden.
Gronbech, Kaare 1942. Komanisches Worterbuch. Turkischer Wortindex
zum Codex Cumanicus. Kopenhagen.
Hahn, Reinhard F. 1991. Spoken Uyghur. Seattle-London: University
of Washington Press.
Jarring, Gunnar 1933. Studien zu einer ostturkischen Lautlehre.
-- 1964. An Eastern Turki-English dialect dictionary. Lund.
Johanson, Lars 1979. Die westoghusische Labialharmonie. Orientalia
Suecana 27-28, 63-107.
--1998. The history of Turkic. Iii: Johanson, Lars and Csato, Eva
A. (eds.). The Turkic Languages. London-New York: Routledge, 81-125.
Judaxin, K.K. 1965. Kirgizsko-russkij slovar'. Moscow [Reprint
Junker, Heinrich F.J. and Alavi, Bozorg 1965. Worterbuch
Persisch-Deutsch. Leipzig: Verlag Enzyklopdie.
Lessing, Ferdinand D. (ed.) 1960. Mongolian-English Dictionary.
Berkeley-Los Angeles-London [Reprint Bloomington 1995].
Menges, Karl Heinrich 1959. Die aralo-kaspische Gruppe. Iii: Deny,
Jean et al. (eds.). Philologiae Turcicae fundamenta 1. Aquis Mattiacis,
Nadzip, E. N. 1968. Ujgursko-russkij slovar'. Uyyurce-rusce
OTIL: Ozbek tilining izahli luyati. 2 vols. Moscow 1981.
Pal'mbax, A. A. 1953. Russko-tuvinskij slovar'. Moscow.
Posch, Udo 1964. Khalkha und Verwandtes. Iii: Poppe, Nikolaus et
al. Mongolistik. Handbuch der Orientalistik, Erste Abteilung, V.2,
Pritsak, Omeljan 1959a. Das Neuuigurische. Iii: Deny, Jean et al.
(eds.). Philologiae Turcicae fundamenta 1. Aquis Mattiacis, 525-563.
-- 1959b. Das Altaiturkische. Iii: Deny, Jean et al. (eds.).
Philologiae Turcicae fundamenta 1. Aquis Mattiacis, 568-598.
-- 1959c. Das Abakan- und Culymturkische und das Schorische. Iii:
Deny, Jean et al. (eds.). Philologiae Turcicae fundamenta 1. Aquis
Ramstedt, G.J. 1935. Kalmuckisches Worterbuch. Helsinki.
Rasanen, Martti 1949. Materialien zur Lautgeschichte der turkischen
Redhouse Turkce/Osmanlica-Ingilizce sozluk. Istanbul 1968.
Resetov, V.V. and Saabdurahmaiiov, S. 1978. Ozbek dialektologiasi.
Sayim, Abdugeyyum 2007. Xoten xelq qosaqlirida ucraydiyan
di'alekt terkibliri toyrisida izdinis. Til ve terjime 2/2007,
Sims-Williams, Nicholas. Eastern Middle Iranian. Schmitt, Rudiger
(Hrsg.). Compendium Linguarum Iranicarum. Wiesbaden 1989, 165-172.
Slepcov, P.A. 1972. Jakutsko-russkij slovar'. Moscow.
Steingass, Francis 1892. A Comprehensive Persian-English
Dictionary. London [Reprint: Delhi 2000].
TatRS: Tatarsko-russkij slovar'. Moscow 1966.
TDS: Turkmen dilinin sozlugi. Asyabat 1962.
Tenisev, E.R. 1968. Tuvinsko-russkij slovar'. Moscow.
Tenisev, E.R. and Sujuncev, X.I. 1989. Karacaevo-balkarsko-russkij
Thomsen, Kaare 1959. Das Kasantatarische und die westsibirischen
Dialekte. Iii: Deny, Jean et al. (eds.). Philologiae Turcicae fundamenta
1. Aquis Mattiacis, 407-421.
Todaeva, B.X. 2001. Slovar' jazyka ojratov sin'czjana.
TRS: Raximi, M.V. and Uspenskaja, L.V. 1954. Tadziksko-russkij
URS: Akobirov, S.F. and Mixajlov, G.N. 1988. Uzbeksko-russkij
UTIL: Uyyur tilinin izahliq lugyiti. 6 vols., Beijing 1990-1999.
Vasmer, Max 1953-1958. Russisches etymologisches Worterbuch. 3
Vietze, Hans-Peter 1988. Worterbuch Mongolisch-Deutsch. Leipzig.
Wehr, Hans 1985. Arabisches Worterbuch fur die Schriftsprache der
Gegenwart. Arabisch-Deutsch. 5. Auflage, unter Mitwirkung von Lorenz
Kropfitsch neu bearbeitet und erweitert. Wiesbaden: Harrassowitz.
Wurm, Stefan 1959. Das Ozbekische. Iii: Deny et al. (eds.).
Philologiae Turcicae fundamenta 1. Aquis Mattiacis, 489-563.
Yakup, Abdurishid 2002. An Ili Salar vocabulary. Introduction and a
provisional Salar-English lexicon. Tokyo.
-- 2005. The Turfan dialect of Uyghur. (Turcologica 63.) Wiesbaden.
Zenker, Julius Theodor 1866-1876. Turkisch-Arabisch-Persisches
Handworterbuch. 2 vols., Leipzig [Reprint: Hildesheim-Zurich New York
(1) The factors determining the choice of the suffix variants are
quite intricate: On the one hand, there are cases where not the stem
vowel, but the stem as a whole seems to carry the sound harmony
information, as in is 'work', which is [+back] although the
stem vowel /i/ is ambiguous in this respect (hence suffixed forms like
is'Zar, isga, et cetera). On the other hand, there are cases where
the stem vowel determines the suffix vowel even if the stem ends in a
consonant that unambiguously carries the opposite sound harmony feature:
Thus, xelq 'people' takes front vowel suffixes although the
consonant auslaut is back (xelgler, xelgte). This results in suffix
variants that are hybrid with respect to sound harmony (in other words,
in which there is no infra-syllabic sound harmony), as the dative suffix
variant -qe in xelgqe.
(2) In some Turkic languages, there are also regressive consonant
assimilations, as Yakut orommut 'our beds' from oron
'bed', or spoken Uyghur kuller 'days' from kun
'day'. Consonant assimilations will not be dealt with in this
(3) There is no systematic contrast between /i/ and ri in modern
Uyghur. The phoneme resulting from the merger of Turkic /i/ and ri is
represented by /i/ in this contribution.
(4) The conditions for this process are described exhaustively by
Hahn (1991: 51 52), while most of the other sources are less
exact.--There is a considerable dialectal variation within the Uyghur
area with respect to the question to which degree umlaut is employed. If
we compare the data presented on the southern dialects of Uyghur by
Jarring (1933: 90-94) with the situation in the modern standard
language, it becomes clear that the process is much more advanced in the
standard language. This may be both due to the situation in the dialects
the Uyghur standard language is based on, and to a diachronic
(5) To which the initial /y-/ may have contributed.
(6) Note also the diverging initial consonant of the reduplication
(7) Likewise in the word yessimuq 'lentil'. The original
Turkic form was probably yasmuq, as most modern Turkic languages
preserve a similar form, cf. Uzb yasmuq (OTIL, 2: 488), Tur yasmik.
Clauson (1972: 975) records an Old Uyghur form yasunuq, which I would
argue to derive from an older form *yasmuq with an epenthetic vowel /i/.
The Modern Uyghur form may either directly originate from Old Uyghur
yasunuq, or be a contamination from yasmuq with the etymologically
related adjective yessi.
(8) Qumandi (Kumandin), Tuba (Black Forest Tatar) and Quu
(Lebed' Tatar) are northern varieties of Altay Turkic.
(9) Thackston transcribes this word as uruk, which is not at all
impossible: After the labial assimilation eruk > oruk, /o/ could have
been raised to /u/, resulting in a total assimilation of the vowel of
the initial syllable to the vowel of the subsequent syllable. However,
taking the Uyghur and Uzbek forms into consideration, oruk seems the
more plausible reading.-The form *uruk is indirectly attested by Tatar
orek (TatRS 740). In Kazan Tatar, /u/ and /u/ have been lowered to /o/
and /o/, while /o/ and /o/ have been raised to /u/ and /u/ (Thomsen
1959: 411). It has not yet been completely understood how exactly this
development evolved. The orthographical in the second
syllable, by instance, is not the result of a vowel lowering of the type
described below, but represents a reduced high front vowel (< /i.-The
form uruk is also found in the Uyghur dialects of Southern Xinjiang
(Jarring 1964: 328).
(10) Russian urjuk 'dried apricot' is probably a loan of
the umlauted Turkic form oruk (cf. Vasmer 3: 189, where no etymology is
offered for urjuk).
(11) Zenker 127 gives meanings like 'to diminish'.
(12) Vowel labialized under the influence of /m/.
(13) With labialization of the second vowel triggered by the labial
consonant /m/. The Turkish form of this word, damar, originates from the
variant tamar which is also attested in Old Turkic (Clauson 1972: 508).
(14) This word survives also in Turkish desik 'pierced;
(15) The variant yapraq exists in Uyghur too: UTIL 6: 399.
(16) Vowel labialized under the influence of /p/.
(17) This statement is true both synchronically (where there is
rarely a high round vowel in the second syllable in Uzbek) and
(18) Interestingly, Yakut has precisely the umlauted form orduq
(from T artuq) not permitted in Uyghur (Slepcov 1972: 276).
(19) However, Yakut uun (Slepcov 1972: 459) seems to point to an
underlying *yugun rather than *yugen.
(20) C Menges 1959: 442.
(21) C Pritsak 1959b: 580.
(22) I have no idea about how the /p/ came into being.
(23) 'Full' in Uyghur is toluq (UTIL 2: 278-279) with
another deverbal noun suffix. The equivalent toliq exists in Uzbek, too
(OTIL, 2: 246-247).
(24) Labialization of the vowel of the first syllable occurs
frequently in this item. Already Babur Turkic (sixteenth century) has
tulku (71r5-6). (The word tulku occurs thrice at this
place, the third occurrence is misspelled as tuklu.)
(25) I kanciliq nefit, xiiniye sana'iti ge'ologiyilik
gidiris ve su qatarliq islarda gezis, kolas, tesis ucun islitilidiyan
uskune; perme. II kala, tay tekisi qatarliqlarnin munguzidin
yasilidiyan, bir xil qedimiy calyu esvabi.
(26) Khalkha buree(n) 'trumpet, horn' (Vietze 1988: 67;
BAMRS 307), Kalmuck buree 'Blasinstrument, Trompete' (Ramstedt
(27) The only possible objection against the assumption of a loan
of Turkic burya into Mongolian in the form burege is the change of
palatal harmony (back > front). However, variation in palatal harmony
(in both directions) is not uncommon even within Turkic; cf. Old Turkic
muyuz/munuz (< *bunuz, Clauson 1972: 352) 'Horn', Uyghur
muiguz, Kazakh muyiz, Kirghiz muyuz, Altay muus vs. Turkish boynuz,
Tuvan miyis (front > back); Old Turkic bin 'thousand',
Uyghur min (contrary to the first impression, this word is back: e.g.
the plural is minlar), Kazakh miry, Tuvan mun vs. Kirghiz min, Turkish
bin (back > front); Old Turkic yegirmi '20', Uyghur
yigirme, Uzbek yigirma, Turkish yirmi vs. Kazakh ziyuma, Kirghiz jiyirma
(front > back).
(28) On the other hand, although the sound changes in Turkic and
Mongolic have lead to slightly different results in terms of phonetics,
we may note a certain similarity in terms of sound development in the
Turkic-Mongolic contact area.
(29) Clauson (1972: 681) erroneously notes that this word does not
survive in the southeastern group of Turkic ("s.i.a.m.l.g. except
(30) tiyi destisige ornitilidiyan cepis, yonus qorali.--Nazip
(1968: 640) gives the following meanings: I rezec; II 1. kirka; 2.
toporik mestnogo izgotovlenija; 3. noz dlja rubki solomy i klevera.
(31) Russian kirka 'pick' very much looks like a loan of
Turkic kerki (possibly even the variant with lowered vowel). The word
must have been quite widespread in pre-modern Turkic and is even
recorded in Ottoman dictionaries (Redhouse 641: kerki 'large
ax'). However, Vasmer 1: 560-561 mentions Ottoman Turkish kurek
'shovel, paddle' as a possible model for the Russian word.
(32) 1. derexnin kesilgendin keyinki yeltiz bilen gelip qalyan
tovenki qismi; 2. di'ametri con, tom, gisgaraq yayac; 3. talon ve
ispatlarni yezip bergende, talon ve ispatnin yezip bergucide qelip
(33) According to the Uzbek dictionaries, kirpi is a dialect form,
the standard form for 'hedgehog' being tipratikan.
(34) Le. the long vowel phoneme /b/ that has merged with /G/ in the
western varieties of Persian.
(35) Interestingly, Uzbek has a de-nominal verb morala- 'to
peer', 'to peep' (podsmatrivat', podgljadyvat'
(cerez cto-l.); vygljadyvat', URS 277; Tosiq, tirqis, tesik arqali
qaramaq, OTIL, 1: 490), without doubt derived from mori, with the same
vowel lowering as in Uyghur mora. A very similar case of vowel lowering
in a derivation is the Ottoman Turkish word borazan
'trumpeter' (Redhouse 190), a de-nominal noun derivation from
the word boru already mentioned.
(36) 1. ahalisi asasen dehqanciliq bilen suyullinidiyan jay,
mehelle, rayon; qislaq; sehra; 2. memlikitimiznii memuriy rayon
bolunmisidiki asasiy gatlam tes'kiliy orgini. adette nahiye yaki
nahiye derijilik organnii basqurusida bolidu. unii basqurusida yene
kentler bolidu.-Nadzip (1968: 797) gives the meanings derevnja, selo
(37) bala-gaza, apet, jengi jedel, urus-yowya.
(38) urus, jendiki qarsi terep; dusmen, reqib.--Nadzip (1968: 798)
mentions only the form yeya 'bedstvie' ('disaster,
(39) What is yessi 'flat' in the Standard Uyghur of China
(< T yasi) has been standardized as yese in Soviet Uyghur (Nadzip
1968: 797), reflecting a slightly different dialectal basis.
(40) In spoken Uyghur however, also in the spoken standard
language, the /i/ of yeni can be realized as a mid central vowel /a/.
(41) The words are not totally synonymous, qora/qoru being also
used for a pen, i.e. an enclosure for animals.
(42) This is time and again even reflected in writing, e.g. in
ertmaq [sic!] 'to wipe, to clean' (UTIL, 6: 76) < T arit-
(Clauson 1972: 207); cf. Uzb art- (OTIL 1: 55), Turkish arit-. Soviet
Uyghur has yesmuq 'lentil' (Nadzip 1968: 797) < yasimuq
(Standard Uyghur of China: yessimuq).
(43) However, Mongolic languages have i-umlaut as well. This is
especially striking in Oirad (also spoken in Xinjiang), which has e.g.
adl 'equal' (Ramstedt 1935: 21; Todaeva 2001: 45) < adil,
cf. Written Mongolian adali, and morn 'horse' (Ramstedt 1935:
266; Todaeva 2001: 238) < morin. In Mongolic, a significant Iranian
substratum does not seem probable.