You are on page 1of 30

SPECULATIONS ON THE PREHISTORIC DEVELOPMENT

OF THE PROTO-INDO-EUROPEAN VOWEL SYSTEM


1. INTRODUCTION:
ALLAN R, BOMHARD
Boston, Massachusetts
The first attempt to reconstruct the Proto-Indo-Euro-
pean vowel system was made by Schleicher, whose Compendium
der vergleichenden Grammatik der indogermanischen Sprachen
first appeared in 1861-62 (fourth and final edition, 1876).
Schleicher's (1876:11) system was as follows:
Original First Second
Vowel Increment Increment
A. a-grade a
a+aaa a + aa == aa
B. i-grade i a+ i = ai a + ai iii
c. u-grade u a+uau a+au=au
Even though Schleicher's system, which was based upon that
of Old Indic, was able to account for many of the develop-
ments found in the daughter languages, there remained many
unsolved problems, and his system did not endure the on-
slaughts of a series of brilliant discoveries made in the
seventies of the last century by a new generation of schol-
ars, the so-called "Neogrammarians".
Perhaps the most important discovery of the Neogram-
marian period was the "Law of Palatals", according to which
an original k, for example, developed into c in Old Indic
under the influence of a following * ~ . *r, or *y. This dis-
covery firmly established the primacy of the vowel systems
found in the European daughter languages and proved that the
Indo-Iranian system had resulted from an innovation in which
original * ~ , *a, and *a had merged into *a. Also important
was the demonstration that the Indo-European parent language
had syllabic liquids and nasals.
According to the Neogrammarians (see especially Brug-
mann 1904:66-90 and Hubschmann 1885), Proto-Indo-European
had the following vowel system:
GENERAL LINGUISTICS, vol. 21, No. 3. Published by The Pennsylvania
State University Press, University Park and London.
164
A. Vowels: e o a u a
e o a i u
B. Diphthongs:
c. Syllabic Liquids
and Nasals:
ei oi al ai
ei ol ai
I
eu ou au au
eu ou au
During the past half century or so, the Neogrammarian
view has been steadily attacked. It was dealt its first
major blow in 1927 with Kurylowicz's demonstration that one
of de Saussure's "coefficients sonantiques" was preserved
in Hittite. In one fell swoop, the so-called "original"
long vowels (as well as the long syllabic liquids and nasals)
were eliminated as were *a, which was taken to result from
*e when next to an "a-coloring" laryngeal, and nonapophonic
*o, which was taken to result from *e when next to an "a-
coloring" laryngeal. The next to go were the diphthongs,
which were reanalyzed as clusters of vowel plus resonant
and resonant plus vowel (cf. Lehmann 1952:10-14). The in-
dependent status of *i and *u had early been questioned by
Meillet (1964:118-22), who regarded them as the syllabic
forms of *y and *w respectively. Finally, a strict adher-
ence to Hirt's ablaut and accentuation theories made it
possible to eliminate apophonic *o, which was taken to re-
sult from an earlier *e when the accent was shifted from
the *e to another syllable. By applying all of these theo-
ries, it became possible to reduce the Proto-Indo-European
vowel system to a single member: *e.
It should be made clear that this extreme view was nev-
er universally adopted. In fact, it was vigorously attacked
by several scholars, including Jakobsen (1971:528) who so-
berly noted: "The one-vowel picture of Proto-Indo-European
finds no support in the recorded languages of the world".
In 1967, relying heavily on typological
data to support his arguments, reinstated all of the vowels
reconstructed by the Neogrammarians:
a e o u a
a e o i c
165
166
Szemerenyi (1967:97, fn. 91), however, ignores the diph-
thongs, "whose phonemic status is disputed". I fully sup-
port Szemerenyi's views on the vowels -- though not on the
diphthongs -- and would reconstruct an identical system
for the Indo-European antecedent of the non-Anatolian daugh-
ter languages (cf. Bernhard 1979b:72). The following corres-
pondences provide the basis for this reconstruction (cf.
Brugmann 1904:67; Meillet 1964:98-126):
Proto-Indo-European *e *o *a *i *u *a
Gothic i a i ai u au a
Old Irish e 0 a i u 0 a
Latin e 0 a i u a
Greek
e;
0 a L u a e:
Tocharian B ya- a a e a a i u a a
Indo-Iranian a i u
i a
Albanian je i e a li i u a li
Armenian e 0 a i u a
Old Church Slavic e 0 n :C 0
Lithuanian e a i u a
Proto-Indo-European *e *5 *a *T *u
Gothic e 5 el u
Old Irish
j
a
j
u
Latin e 5 a T u
Greek
n
w a i: ii
Tocharian B ye- e a a o a a I u 0
Indo-Iranian a T u
Albanian 0 e 0 i y
Armenian I u a I u
Old Church Slavic e a i y
Lithuanian e uo 5 5 y u
2. Vowel Gradation:
Proto-Indo-European, as also, for example, Proto-Kart-
velian, Northwest Caucasian, and Proto-Semitic, was charac-
terized by an interchange of vocalic elements that could
occur in any syllable. This interchange, which is commonly
called "ablaut" or "vowel gradation", was partially corre-
0
lated with the position of the accent and with distinc.tions
between grammatical categories (cf. Burrow 1973:108-17).
Several gradation series are traditionally distinguished,
and the general scheme may be represented as follows (cf.
Brugmann 1904:138-50; Meillet 1964:153-68):
I.
SHORT VOWEL GRADATION:
Lengthened-Grade
A.
e'\,o
B. ei "'51
eu "'5u
er "' 5r
el
"'51
em"'5m
en "' 5n
c.
D.
II. LONG VOWEL GRADATION:
E.
F.
G.
Normal-Grade
9"-0
el
"' ol
eu "'ou
er"' or
el "'ol
em"-orn
en "' on
a"' o
al
au
e"-5
5
ii"' 5
Reduced-Grade
l1l
u
r;
!
~
g
l1l
u
e
e
e
The most common vowel was Ae, and the most common grad-
ation pattern was the *e "' Ao contrast. The vowel Aa was of
relatively low statistical frequency and, at least according
to Meillet (1964:154), did not take part in the regular grad-
ation patterning. It should be pointed out, however, that
several rare examples of an *a "' Ao contrast are attested
in the non-Anatolian daughter languages, one probable exam-
ple being:
Gk. ciyw "to lead, carry, convey, bring"
O.,uos "any straight line: a furrow, path, etc."
3. THE LARYNGEALS AND VOWEL GRADATION:
In 1878, the young de Saussure reinterpreted the origi-
nal long vowels of Proto-Indo-European as sequences of short
167
168
vowel plus a following "coefficient sonantique". Even though
de Saussure's theories were overwhelmingly rejected at first,
the later discovery of one of his coefficients sonantiques
in Hittite proved that the essence of his theory was correct.
Briefly stated, the original long vowels came into being
when a laryngeal was lost between an immediately preceding
short vowel and an immediately following nonsyllabic, that
is to say, the preceding vowel was lengthened to compensate
for the loss of the laryngeal (cf. Kurylowicz 1935:28; Stur-
tevant 1942:66-71):

> ec
> ac
> ac
> oc
Examples:
A. Hitt. 1 sg. 3 sg. da-a-i "to put, place", Skt.
d&.-dhii-ti "to place", Gk. "to place"< PIE
(cf. Pokorny 1959:235-39; Sturtevant 1942:55 and 66).
B. Hitt. 1 sg. ti-it-ta-mi "to cause to stand, place", Dor.
"to cause to stand", Lat. stO -are "to stand" <
PIE (cf. Pokorny 1959:1004-10; Sturtevant 1942:
44 and 67).
C. Hitt. 1 sg. "to protect", Skt. pati "to pro-
tect", Lat. pasao "to feed, lead to pasture"< PIE
(cf. Pokorny 1959:787; Sturtevant 1942:36),
D. Hitt. "time, point of time, occasion", Goth. mel
"time'', Skt. mciti
11
to measure", Lat. metior ''to measure
11
<PIE (cf. Pokorny 1959:703-04; Sturtevant 1942:47).
We should note here that and are lost in Hittite, while
and are preserved (cf. Barnhard 1976:222-31 and 1979a:
133-34; Sturtevant 1942:35, 42, 46, and 53 and 1951:47-55).
Another basic assumption of the Laryngeal Theory is
that one or more of the laryngeals influenced the quality
of contiguous vowels. According to Kurylowicz (1935:28-30),
had no effect on vowel quality, but and changed
a contiguous *e to *a, and changed a contiguous *e to
*o. Sturtevant (1942:20 and 46-53), however, rejects the
theory that (= Kurylowicz's changed a contiguous *e
to *o.
Examples:
A. Hitt. 3 sg. e-es-ai "to be", Skt. asti "to be", Gk. E:oTC
"to be", Lat. est "to be" <PIE (cf. Pokorny 1959:
340-42; Sturtevant 1942:72).
B. Hitt. a-ap-pa "afterwards, back, again", Skt. 6.pa "away,
forth, back", Gk. Cino, an6 "away, from, after, back"<
PIE (cf. Pokorny 1959:53; Sturtevant 1942:45).
C. Hitt. !!a-an-ti "in front of, before, opposite", Skt. anti
"opposite, before, near", Gk. &vTL "opposite, over, against",
Lat. ante "before" < PIE (cf. Pokorny 1959:48-49;
Sturtevant 1942:40 and 62).
D. Hitt. !!a-as-ta-i "bone", Luw. !!a-(a-)as-sa "bone", Skt.
asthi "bone", Gk. oon'ov "bone", Lat. os "bone" < PIE
(cf. Pokorny 1959:783; Sturtevant 1942:57).
According to Benveniste (1935:149), the patterning of
the laryngeals was the same as that of the resonants, and,
consequently, they had syllabic variants when between two
nonsyllabics. The syllabic variants are commonly associ-
ated with the schwa primum reconstructed for Proto-Indo-
European by the Neogrammarians.
What have just been presented are the basic tenets of
the Laryngeal Theory as traditionally conceived. That is
not to say that all shades of opinion have been included --
there are simply too many conflicting viewpoints, and this
is not the place to discuss the merits or demerits of each.
For information on the current state of laryngeal studies,
see Jonsson 1978, Keiler 1970, and Winter 1965a.
One of the most difficult riddles to solve has been
and continues to be the determination of the probable pho-
netic values of the various laryngeals. I have dealt with
this problem at length in two preceding papers (Bernhard
1979a and 1981:392-96), and I would like to summarize my
169
170
views here. Following that, I would like to outline the
possible prehistoric development of the laryngeals as I
see it.
I will begin by mentioning that I would reconstruct at
least four laryngeals for pre-Anatolian Proto-Indo-European.
There is no other convincing way to account for (1) post-
Anatolian PIE *e without a corresponding Anatolian laryngeal
reflex (this is Kurylowicz's *11), (2) post-Anatolian PIE
*a without a corresponding Anatolian laryngeal reflex (this
is Kurylowicz's ~ ~ ) , (3) post-Anatolian PIE *e (Kurylowicz
1935:28-29 posits *o here) with a corresponding Anatolian
laryngeal reflex (this is Kurylowicz's * ~ s ) , and (4) post-
Antolian PIE *a with a corresponding Anatolian laryngeal
reflex (this is Kurylowicz's *t2). I follow Sturtevant
(1942:20 and 46 and 1951:49-51) in assuming that *ts like
*tl did not color contiguous vowels.
According to Colarusso (1981:512), Couvreur (1937:264),
and Sturtevant (1942:19 and 1951:54), *11 was a glottal
stop, *I?/. The interpretation of *11 as */?/ explains
why this laryngeal did not color contiguous vowels since
glottal stops do not affect vowel quality in either the
Semitic languages or the Northwest Caucasian languages (cf.
Bombard 1979a:l32; Colarusso 1981:511). Moreover, loss of
a glottal stop betwee.n an immediately preceding vowel and
an immediately following nonsyllabic causes compensatory
lengthening of the vowel in Akkadian and Arabic (cf. Bom-
bard 1979a:l30-31; Couvreur 1937:288-89; Moscati 1964:61-
64), and this is precisely the development assumed for Pro-
to-Indo-European. I thus see no reason to challenge the
interpretation of *11 as */?/. This laryngeal is not di-
rectly attested in any of the daughter languages, inctuding
(apparently) Hittite (cf. Bombard 1976:230; Sturtevant 1942:
53 and 1951:54).
Kurylowicz (1935:29-30) sets up *t- to account for
those cases in which an a in the non-Anatolian daughter lan-
guages corresponds to an a in Hittite, and Hittite lacks a
contiguous laryngeal reflex. That is to say t.hat ~ ~ is
not directly attested in Hittite or in any of the other
daughter languages (cf. Bombard 1976:230; Sturtevant 1942:
42 and 1951:51-52), though its former presence can be de-
termined by the fact that it changed a contiguous *e to *a
and by the fact that it caused compensatory vowel lengthen-
ing when lost between an immediately preceding short vowel
and an immediately following nonsyllabic. According to
Hopper (1977:49-50), typological evidence implies that the
voiceless laryngeal fricative */h/ should be added to the
Proto-Indo-European phonemic inventory, and this coincides
with the value assigned by Colarusso (1981:512),
Lehmann (1952:108), and (apparently) Sturtevant (1951:52).
In terms of distinctive feature theory, */h/ would be [+cons,
+low, -voice, +cont, +grave]. As far as we are concerned,
the most important feature is the feature [low]. According
to Chomsky and Halle (1968:305), the articulatory gesture
behind the feature [low] is a "lowering [of] the body of
the tongue below the level that it occupies in the neutral
position", while Colarusso (1981:509) defines it as "an
opening of the oral cavity to enhance resonance". It was
the presence of this feature that was responsible for the
lowering of *e to *a. Finally, we may note that develop-
ments similar to those assumed for in Proto-Indo-Euro-
pean are found in Ubykh and in the Circassian languages,
where /h/ (and /hW/) lowers and colors contiguous vowels
and also causes compensatory vowel lengthening when lost
(cf. Colarusso 1975:396).
Reflexes of are found in Hittite and the other
older Anatolian daughter languages, where they are written
a<al (cf. Bomhard 1976:222-27; Sturtevant 1942:35 and 1951:
47). Like changes a contiguous *e to *a. On this
basis, we would expect also to be characterized by the
feature [low]. Good candidates to assign as the phonetic
values of are the multiply-articulated pharyngeal/lar-
yngeals Not only are these sounds marked
by the presence of the feature [low], but they also make it
easy to account for the fact that initial appears in
Armenian as h before full-grade vowels (cf. Bombard 1976:
231-32). We can envision a change into */h/ and
of *lt91 into and then into */h/ similar to that found
in the Ashkharwa dialect of Abkhaz cited by Colarusso (1981:
516). The resulting */h/ would have been lost in all of
the non-Anatolian daughter languages except pre-Proto-Ar-
menian. As in Ashkharwa, we can venture a guess that
and */t9/ developed from the earlier pharyngeals and
respectively in early pre-Anatolian Proto-Indo-European.
It is more difficult to determine the phonetic value
of than of any of the other laryngeals. Reflexes of
are also found in the older Anatolian daughter languages
(cf. Bernhard 1976:228-30; Sturtevant 1942:44 and 1951:49-
171
172
51). Forms such as Hitt. "time" vs. Goth. miJl. "time"
and Lat. metior "to measure", for example, make it seem
likely that did not alter the quality of contiguous
vowels. Possible candidates for could be the voiceless
velar fricative */x/ and the voiced velar fricative */y/.
I have chosen velar fricatives for several reasons: First,
these are among the types of sounds that we would expect as
"laryngeals". Next, there is circumstantial evidence from
the Anatolian languages that velar fricatives may underlie
at least some of the Anatolian laryngeal reflexes. The
often-cited r alternation in Hitt.
points in that direction as do loanwords from Hittite into
other languages and orthographic variation between k and a
(for details, cf. Puhvel 1965:80-86). Finally, Colarusso
(1981:540-46) has discussed at length the possibility that
the laryngeals may have had velar allophones, and he has
noted that such sounds do not cause vowel-lowering.
We may summarize our findings by setting up the follow-
ing matrix:
? h

\'fi X
y
......
e-coloring + + +
a-coloring
+ + +
preserved in Anatolian
+ + + +
Now that we have determined the probable phonetic val-
ues of the laryngeals, we can turn to the question of their
probable prehistoric development.
I would set up four laryngeals for pre-Indo-European:
*!?/, */h/, */\'/. The earliest development to take
place was the split of into and */x/ and of */\'/
into*/\'/ and */y/. A similar development seems to have
occurred in pre-Proto-Semitic (cf. Cohen and in
several of the Northwest Caucasian languages (cf. Colarusso
1981:545-46). We may assume that, as in Proto-Semitic, the
velar fricatives */x/ and */y/ were of a lower frequency of
occurrence than the pharyngeal and*/\'/.
According to Colarusso (1981:545), pharyngeal fricatives
can develop into velar fricatives through the following pro-
gression: lbl + /xn/ (voiceless pharyngealized velar frica-
tive) + /x/. I equate the velar fricatives with Kurylowicz's
but follow Sturtevant (1942:20) in assuming that this
laryngeal did not color contiguous vowels.
The next change to take place was the shift of
and to the multiply-articulated pharyngeal/laryngeals
and respectively. Colarusso (1981:516) cites a
similar development in the Ashkharwa dialect of Abkhaz.
These pharyngeal/laryngeal fricatives, as also the voice-
less laryngeal fricative */h/, were characterized by the
presence of the feature [low]. These were the so-called
"a-coloring" laryngeals. I equate and */'ij/ with
Kurylowicz's and */h/ with his It was at this
stage that the Anatolian languages became separated from
the main speech community.
In post-Anatolian Proto-Indo-European, */?/, */h/,
*/x/, and */y/ were lost initially before vowels, while
> */h/ and *ItS! > > */h/ in the same environ-
ment. In later Proto-Indo-European ("Disintegrating" Indo-
European), all laryngeals first merged into */h/. */h/
(from earlier and was then lost initially
before vowels (except in pre-Proto-Armenian) and medially
between an immediately preceding vowel and an immediately
following nonsyllabic. This latter change occasioned the
compensatory lengthening of preceding short vowels. I can
imagine another possible scenario here as well. Since
*/?/ and */h/ are (apparently) not directly attested in
the Anatolian languages, these laryngeals may have been
lost before the others and before the separation of the
Anatolian branch from the main speech community,
I assume that the single remaining laryngeal, */h/,
was preserved in all other positions and that it had a syl-
labic allophone when between two nonsyllabics, At least
one laryngeal must be reconstructed for late Disintegrating
Indo-European to account for developments in the non-Anato-
lian daughter languages such as (cf. Barnhard 1976:233 and
1981:368; Lehmann 1952: 85): (1) the Indo-Aryan voiceless
aspirates; (2) the Greek prothetic vowels; (3) the Greek
rough breathing, in part; (4) Armenian initial h, in part;
(5) the Balta-Slavic intonations; and (6) the Germanic
Verscharfung.
This rather long discussion was necessary since the
laryngeals had profound effects on the development of the
vowel system in Proto-Indo-European. Consequently, it is
impossible to discuss one without also discussing the other.
173
174
4. NOTES ON THE CONSONANTS:
Since I do not accept the traditional reconstruction
of the Proto-Indo-European consonant system, it is necessary
to clarify my views. However, since the primary purpose
of this paper is to discuss the prehistoric development of
the vowels and since I have devoted a previous, rather leng-
thy paper to a discussion of the consonants (Bernhard 1979b),
I will present my views in only the briefest terms here.
I follow Hopper (1973:141-66) and Gamkrelidze-Ivanov
(1973:150-56) in their reinterpretation of the traditional
plain voiced stops as glottalics (cf. Bernhard 1979b:77-BO).
In addition, I would reinterpret the traditional voiced as-
pirates as plain voiced stops (cf. Bernhard 1975:378-80 and
1979b:80). The revised Proto-Indo-European consonant sys-
tem may be reconstructed as follows (cf. Bernhard 1979:460):
Stops:
Voiceless: p
Voiced: b
Glottalized: (p')
Fricatives:
Voiceless:
Voiced:
Resonants:
Nasals:
Rolled:
Lateral:
Glides: w/u
t
d
t'
s
n/g
r/t
1/l
y/i
k ~
g
gw
~
k'W
X
y
My reasons for supporting Gamkrelidze-Hopper-Ivanov
?
h
are as follows: (1) I adhere to the widely-held view that
the Proto-Indo-European stop system was characterized by a
three-way contrast, that is, that the voiceless aspirates
were not phonemic in Proto-Indo-European (cf. Bernhard 1979b:
73-74). In traditional terms, the contrast would be between
plain voiceless, plain voiced, and voiced aspirated stops,
thus (using the dental series for illustrationj:
t ~ d ~ dh
Such a system, however, has no parallels among the known
languages of the world (cf. Jakobsen 1971:528; Martinet
1970:115) and must, therefore, be considered as highly im-
probable typologically. By substituting ejectives for the
traditional plain voiced stops, the resulting system becomes
completely natural, and numerous typological parallels ex-
ist -- similar systems are found, for instance, in the Cau-
casian languages, many American Indian languages, and seve-
ral sub-Saharan African languages (for examples, cf. the
phonological systems given in Ruhlen 1976). (2) Additional
evidence that the traditional plain voiced stops were really
ejectives is found in the fact that the phoneme tradition-
ally reconstructed as *b was highly marked in the system,
being characterized by an extremely low frequency of occur-
rence (if it even existed at all). Such a frequency distri-
bution is highly uncharacteristic of the voiced bilabial
stop, but it is fully characteristic of the glottalized bi-
labial stop (cf. 1981:605-06; Greenberg 1970:
127). Therefore, the reinterpretation of *bas *p' easily
accounts for its high degree of markedness. (3) For the
first time, the root structure constraint laws can be cred-
ibly explained. These constraints turn out to be a simple
voicing agreement rule, with the corollary that two glotta-
lics cannot cooccur in a root (cf. Gamkrelidze 1976:404-05
and 1981:608-09; Hopper 1973:158-61). (4) Finally, the Ger-
manic and Armenian consonant shifts, which can only be ac-
counted for very awkwardly within the traditional framework
(cf. Emends 1972:108-22), turn out to be mirages. In fact,
the consonant systems of Armenian and Germanic preserve
many archaic features not found elsewhere (except, probably,
in the poorly-attested Thracian and Phrygian).
5. CRITIQUE OF THE TRADITIONAL RECONSTRUCTION:
Colarusso (1981:499) has astutely observed: " ... the
PIE vowel system *e *o is typologically utterly bizarre.
Even adding *a to this system does not change this fact".
Perhaps the most typologically unusual thing about the
Proto-Indo-European vowel system as traditionally recon-
structed is the great importance of the *e *o ablaut and
the concomitant marginality of *a. Adding laryngeals only
makes the system even more unusual since *a then becomes
mostly (but not in every case!) a positional variant of *e.
175
176
Rather, we would expect the relationship to be reversed.
All languages surveyed by Crothers (1978:93-152) have the
vowel /a/, and this vowel is consistently characterized by
a high frequency of occurrence (cf., e.g., the frequency
counts given by Greenberg 1966:18-19). Moreover, in the
Northwest Caucasian languages and in the Semitic languages,
which also exhibit ablaut either as an active process or
as a relic of an earlier, fully functioning ablaut process,
the vowel a is an integral part of the ablaut system (cf.
Colarusso 1981:499-502; Kurylowicz 1962). Clearly, if typo-
logical evidence is to have any meaning, there is something
wrong with the traditional reconstruction of the Proto-Indo-
European vowel system. Yet, if the Comparative Method is
to have any validity, there must be some truth to that re-
construction.
This seeming conflict can be resolved quite easily, I
believe. We can consider the traditional reconstruction to
be mainly correct but only for that form of Proto-Indo-Euro-
pean existing immediately prior to the emergence of the non-
Anatolian daughter languages. The vowel system of this form
of Proto-Indo-European is by no means ancient and is the end
product of a long, complicated evolution. In the following
sections of this paper, I will attempt to reconstruct the
prehistoric development of the Proto-Indo-European vowel
system. I do not claim to have solved every riddle or an-
swered every question -- that is simply impossible given
the nature of the problem. What I hope to have done is to
have caught glimpses of what may have been.
I have relied heavily on the works of Colarusso (1975
and 1981), Schmalstieg (1980), Schmitt-Brandt (1967), and
Wyatt (1970). Colarusso's works, in particular, have been
a constant source of inspiration.
6. EARLIEST PROTO-INDO-EUROPEAN VOWEL SYSTEM:
The earliest Proto-Indo-European vowel system may have
been as follows:
T u
~
a
Also the sequences: ay ay iy uy
ew aw lw uw
This system is identical to that proposed by Schmalstieg
(1980:22). I follow Pulleyblank (1965:86-101) in reinter-
preting the *e ~ *o ablaut of traditional Proto-Indo-Euro-
pean as a *a ~ *a ablaut. Pulleyblank mentions that a sim-
ilar ablaut pattern exists in Kabardian. Colarusso (1981:
499-501) proposes a similar reinterpretation and also dis-
cusses possible typological parallels with the Northwest
Caucasian languages.
According to Hirt (1921:172-99) and those who follow
his theories (Burrow, Lehmann, and, formerly, the author,
for example), the oldest ablaut alternation was the full-
grade ~ zero-grade contrast. This alternation arose at a
time when the Proto-Indo-European phonological system was
characterized by the presence of a strong stress accent.
This accent caused the weakening and loss of unstressed syl-
lables. This period may be called the Phonemic Stress Stage
of Proto-Indo-European (cf. Lehmann 1952:111-12). At a lat-
er date, stress became nondistinctive, and Proto-Indo-Euro-
pean was characterized by an accent system based upon pitch.
This period may be called the Phonemic Pitch Stage of Proto-
Indo-European (cf. Lehmann 1952:109-10). It was during this
period that the *e ~ *o contrast came into being. *e was
changed to *o when the accent was shifted from the *e to
another syllable. These theories find support in the fact
that the position of the accent is partially correlated with
ablaut patterning in both Greek and Old Indic. Counter-ex-
amples are usually explained as due to analogical develop-
ments or as later forms that came into being after the ac-
cent had lost its ability to influence the vowels (cf. Bur-
row 1973:112).
Though Kurylowicz originally adhered to Hirt's theories
as well, he later (1956:36-96 and 1964:52) tried to show
that the *e ~ *o contrast existed prior to the development
of the full-grade ~ zero-grade contrast. Kurylowicz argues
that the numerous counter-examples with accented *6 indi-
cate that qualitative ablaut was a morphological device in
its own right and only superficially connected with the po-
sitioning of the accent. Moreover, he notes that, while
vowel weakening and loss are closely tied to the accent, a
change in vowel quality is primarily due to the environment
-- in other words, there is no cause-and-effect relation-
ship between qualitative ablaut and accentuation. These
are convincing arguments and are the basis for my inclusion
177
178
of qualitative ablaut in the earliest reconstructed period
of Proto-Indo-European.
Vowel systems similar to the one given above -- a four-
vowel system with contrasting long and short members -- are
common enough among the languages of the world, and Crothers
(1978:109-10), in his study of vowel system typology, lists
and discusses a number of them. One important observation
made by Crothers (1978:109) is that such systems "often
show considerable subphonemic variation" (he is speaking
specifically about tbree-vowel systems in the quote given,
but on the next page, he says of four-vowel systems: "As
in the three-vowel systems, there is considerable variation,
especially in the position of the back vowel, and the gene-
ral statements made with regard to the specific phonetic
character of vowels in a three vowel system apply here too").
As an example, he cites Greenlandic Eskimo (a three-vowel
system):
Phoneme
/a/
/i/
/u/
Subphonemic Variants
[m], [a.]
[ i ], [e], [a]
[y], [u], [o], [:>]
Hockett (1955:85), speaking about the following vowel sys-
tem type (found, for example, in Ilocano and Dibabaon):
u
a
a,
notes that "the mid vowel /a/ has fronted variants, and also
higher variants (central or back without rounding)". He
also says about the Potawatomi system
0
a
e a
-- that "the mid central vowel /a/ varies more widely from
one environment to another than the other four ... ".
For the Proto-Indo-European vowels, we can tentatively
assume the following subphonemic variants:
Phoneme
/e/
/a/
/I/
/u/
Subphonemic Variant
/a/ when next to "a-coloring" laryngeal&
/e/ under stress
!+/before /1/
/ttl before /u/
/a/ elsewhere
lo./ when next to "a-coloring" laryngeal&
/o/ when next to labiovelars or /w/ (?)
/a/ elsewhere
/e/ when preceded by "a-coloring" laryngeal& or (?)
before /r/
/1/ elsewhere
/o/ when preceded by "a-coloring" laryngeals
or (?) before /r/
/u/ elsewhere
These subphonemic variants are posited on the basis of vari-
ous assumptions about the prehistoric development of the
Proto-Indo-European vowel system that will be discussed in
the remainder of this paper.
7. ANATOLIAN:
The older Anatolian languages -- Hittite, Palaic, and
Cuneiform and Hieroglyphic Luwian -- directly attest the
vowel system reconstructed for Proto-Indo-European in the
preceding section. The Proto-Anatolian vowel system may
be reconstructed as follows (cf. Bombard 1976:202 and 208-
13):
i
ii
In Proto-Anatolian, PIE *a appeared as *e under stress and
as *a (1) when next to an "a-coloring" laryngeal and (2)
when unstressed (= reduced-grade or schwa secundum of tra-
ditional grammar). *e and *a were in an ablaut relation-
ship in Proto-Antolian, and this was partially preserved in
Hittite. In the other older Anatolian languages, however,
179
180
the ablaut patterning was mostly eliminated by the leveling
out of the vocalism in favor of a (cf. Bomhard 1976:208-09;
Puhvel 1966:239-40).
Examples (cf. Bernhard 1976):
A. PIE > PAn. *e:
Hitt. 3 sg. pres. e-es-zi "is"; Luw. 3 sg. pret. a-as-ta
"was" (z Hitt. e-es-ta); Hier. 3 sg. pres. asti "is" < PAn.
3 sg. pres. *esti < PIE 3 sg. pres. Cf. Skt. asti
"is"; Gk. EoTC; Lat. est.
B. PIE *a> PAn. *a (when next to an "a-coloring" laryngeal):
Hitt. nom. sg. !;a-an-za "front, front side", !;a-an-ti "in
front of, before"; Pal. !;a-an-ti-UU-) "first"; Luw. nom.
sg. (2-an-te-Z.e-es "first" < PAn. *Aant- < PIE Cf.
Skt. anta-s "end", anti "before"; Gk. chnC "opposite, be-
fore"; Lat. ante.
C. PIE *a> PAn. *a (when unstressed):
Hitt. 3 pl. pres. a-sa-an-zi "they are"; Pal. 3 pl. impv.
a-sa-an-du; Luw. 3 pl. impv. a-8a-an-du < PAn. *asant- <
PIE *Eas&nt-. Cf. Skt. s&nti "they are"; Lat. sunt.
D. PIE *a > PAn. *a:
Hitt. nom. sg. "grandfather"; Hier. huhas; Lye.
xuga- < PAn. *AuRas < PIE *AuHas. Cf. Arm. hav "grandfather";
Lat. avus.
Hitt. nom. sg. zi-ik "you"; Pal. nom. sg. ti-i; Hier. nom.
sg. ti <PAn. *te <PIE *til. Cf. Lat. ace. sg. te "you";
Umbr. ace. sg. tiom.
F. PIE a >PAn. *a:
Hitt. dat. sg. ending -a; Pal. dat. sg. ending -ai; Hier.
dat. sg. ending -a < PAn. dat. sg. ending *-a(i) < PIE dat.
sg. ending *-a(i). Cf. Lat. dat. sg. ending -o.
G. PIE *i > PAn. *i:
Hitt. ku-is "who?", ku-it "what?"; Pal. kuis; Luw. ku-is;
Lye. ti-; Lyd. qis, qid < PAn. *kwis, *kwit < PIE *kwis,
*kWit. Cf. Skt. ki-s; Lat. quis, quid; Gk.
H. PIE *i > PAn. *i:
Hitt. dat.-loc. sg. "shaft"< PAn. *His-< PIE
*His-. Cf. Skt. il!ii "pole, shaft of a carriage or plow".
I. PIE *u > PAn. *u:
Hitt. nom.-acc. sg. gi-e-nu, gi-nu "knee" < PAn. *kenu
(see the note following the examples) <PIE *k'anu. Cf.
Lat. genu "knee"; Gk. y6vu (< *k'anu); Skt. jfrnu.
J. PIE *u > PAn. *u:
Hitt. nom.-acc. sg. ku-us-sa-an, ku-u-san "pay, price" <
PAn. *kUs- < PIE *kUs-. Cf. OE. hyrian (< Gmc. *xuzya-)
"to hire", hyr "hire, payment".
There is no positive evidence, as far as I can see, that the older
Anatolian languages had a voicing contrast in stops (for details,
cf. Bombard 1976:221-22; see also my commentary following Arbeit-
man and Ayala 1981).
8. THE PHONEMICIZATION OF *e AND *o:
The phonemicization of *e, which must have occurred
very early to judge by the Hittite evidence, is relatively
easy to explain: *e was the normal allophone of *a under
stress. John Colarusso (personal correspondence) has in-
formed me of a similar development in Ubykh and Circassian,
where
We may assume that *a had a rounded allophone in cer-
tain phonetic environments (cf. Colarusso 1981:500), per-
haps when next to labiovelars as well as when next to *w.
In late Disintegrating Indo-European, these allophones were
reapportioned, and apophonic *a was rephonemicized as *o.
That this was an extremely late development is shown by the
widespread tendency of *a and *a to have identical reflexes
in several of the non-Anatolian daughter languages.
181
182
9. THE DIPHTHONGS:
According to the traditional reconstruction, Proto-In-
do-European had the following diphthongs:
e i o i a i a i
eu ou au au
In the reduced-grade, the semivowels alone appear:
u
Szemernyi (1970:132), notes that, while this looks
good on paper, it is difficult to imagine the process that
would have led to *t and *u in the reduced-grade. He claims
that.it most certainly could not have been due to a simple
loss of *e, *o, and *a.
As a general rule, when diphthongs are "reduced", _long
monophthongs result. This observation led Schmitt-Brandt
(1967:8-31) to conclude that it is necessary to assume a
fundamental form in *i and *u for certain kinds of roots in
Proto-Indo-European and to view the forms with *ei, *eu,
etc. as secondary. Support for this conclusion is to be
found in the root-nouns, which appear in the reduced-grade
(traditional *dik-, *wik-, *duk-, etc. [I would write *t'ik-,
*wik-, *t'uk-]), this being their original form. To have it
the other way around, with *ei, *eu, etc. as the original
forms, would lead, in Schmitt-Brandt's opinion, to reduced-
grade forms with *i and *u: *deik- > *dik-, *weik- > *wik-,
*deuk- > *duk-, etc. Schmitt-Brandt thus posits *i and *u
as independent vowels in Proto-Indo-European and explains
the full-grade forms in *ei, *eu, etc. as due to analogy.
Finally, Schmitt-Brandt (1967:79-91) maintains that, in an
earlier period of Proto-Indo-European, *i and ** were con-
sonants in their own right and differed from the indepen-
dent vowels *i and *u. Somewhat similar views are expressed
by Wyatt (1970:58 and fn. 24).
The parts of Schmitt-Brandt's theories outlined in the
preceding paragraph seem to make a lot of sense. Other parts
of his theories, however, have purposely been left out of
the discussion since, at least in my opinion, they are not
convincing (see here the review of Schmitt-Brandt's book by
Kurylowicz 1969:41-49). On the basis of Schmitt-Brandt's
arguments, we can thus consider the traditional explanation
as being extremely unlikely:
*ei, *oi, *ai reduced to *t
*eu, *au, *au reduced to *U
The alternative would be to consider the possibility that
the development was reversed:
r "incremented" to *ei, *oi, *ai
*U "incremented" to *eu, *ou, *au
The problem is further complicated by the question of
what to do with the so-called "long diphthong" stems such
as, for example:
A. *dhiii(i)-, *dh-ei-, *dhai-, *dh'i-, *dhiii-, *dha- "to suckle,
suck" (cf. Pokorny 1959: 241-42)
B. *kzeu-, *kl.o(u)-, *kl.1- "to wash, cleanse" (cf. Pokorny
1959:607).
In any theory attempting to explain the origin and develop-
ment of the diphthongs within Proto-Indo-European, the long.
diphthongs must be taken into consideration as well.
It has traditionally been assumed that pre-Anatolian
Proto-Indo-European had diphthongs and that these were eli-
minated in Proto-Anatolian (cf. Bomhard 1976:202 and 207-08;
Kronasser 1956:35-36; Sturtevant 1951:35-36 and 39):
Examples:
PIE *ei, *oi, *ai > Hitt. eli
PIE *eu, *au, *au > Hitt. u
A. Hitt. eli *ei, etc. in the non-Anatolian daughter languages:
a. Hitt. 3 sg. mid. ki-it-ta(-ri) "lies"; Pal. 3 sg. mid.
ki-i-ta-ar < PAn. *kita(r) < PIE *kita. But Gk. 3 sg.
x t m ~ "lies"; Skt. sete < PIE *keitoi.
b. Hitt. 2 sg. impv. e-!!u "come!"; Pal. 2 sg. impv. i-u
183
184
"come here!"; Luw. 3 sg. pret. i-i-ta "went" < PAn.
*i- "to come, go" < PIE *Ei-. But Dor. ELTL "goes";
OLith. etti; Skt. eti < PIE *Eei-.
B. Hitt. u = *eu, etc. in the non-Anatolian daughter languages:
a. Hitt. 3 sg. pres. Zu-uk-zi "kindles"; Luw. Zu-u-!Ja-a8
(lJ < k [cf. Bombard 1976:204, 5.3F]) "light" < PAn.
*Zuk- < PIE *Zuk-. But Gk. Auxcis "white" < PIE *Zeuk-.
b. Hitt. nom. sg. i-u-ga-a8 "yoked, one year old" < PAn.
*yukas < PIE *yuk'as. But Gk. "yoke, team" <
PIE *yeuk'os.
Furthermore, it has been assumed that reflexes of ear-
lier long diphthongs are found in Hittite in forms such as
the following (cf. Sturtevant 1951:36-37 and 39-40):
.A. Hitt. nom.-acc. sg. !Ja-a8-ta-(a-)i "bone, skeleton" (gen. sg.
!Ja-a8-ti-ya-a8).
B. Hitt. nom. sg. "curse" (gen. sg.
nom. pl.
C. Hitt. nom. sg. "birth-chair" (gen. sg.
U-1Ja-a8)
A simpler explanation than the traditional one would
be to posit "original" diphthongs and "secondary" diphthongs.
The original diphthongs existed in pre-Anatolian Proto-Indo-
European. The secondary diphthongs, on the other hand, de-
veloped mostly in post-Anatolian Proto-Indo-European, and
the Anatolian forms with i and u corresponding to diphthongs
in the non-Anatolian daughter languages represent the origi-
nal patterning. In forms such as Hitt. and
the like, the -ai- may represent either an original diph-
thong or an early secondary diphthong. The most important
point is that these stems do not come from earlier long
diphthongs as has been traditionally assumed.
We can explain the origin of the long diphthong stems
of traditional grammar quite easily if we follow Schmitt-
Brandt (1967:79-91) in assuming that *y and *w were conson-
ants like any other in early Proto-Indo-European and were
not connected in any way with the independent vowels *i and
*u. Early Proto-Indo-European would have had stems such as
*CtC- and * C ~ C - as well as stems such as *Cay-/*Cay- and
*Caw-/*Caw-, and these two types were totally distinct from
each other. In post-Anatolian Proto-Indo-European, the
latter type developed thus:
Stage I Stage II
A.
aye > Te
B. aye > ee
c. ayV > ayV
D. ayV > ayV
A. awe > ue
B. awe > oe
c. awV > awV
D. awV > awV
The change of preconsonantal *ay to *i: and that of *aw
to *u have typological parallels with developments in the
Northwest Caucasian languages (cf. Colarusso 1981:504-05).
In the case of *ay, the */e/ was raised and fronted to *Ill,
which then combined with the following *y as length, while,
in the case of *aw, the */a/ was raised, backed, and rounded
to */u/, which then combined with the following *was length.
The change of preconsonantal *ay to *e and that of *aw
to *o also have typological parallels with Northwest Cauca-
sian developments (cf. Colarusso 1981:504-05) as well as
with developments in modern Arabic dialects (cf. O'Leary
1923:113 and 115) and, within Indo-European itself, in Slav-
ic (*oi and *ai > ~ ) and Old Indic (*ei, *oi, *ai > *e and
*eu, *ou, *au> *o).
In later post-Anatolian Proto-Indo-European, the pre-
vocalic forms (types [c] and [d] in the chart) were analo-
gically extended to preconsonantal positions. Type (a) was
then reinterpreted as a reduced-grade. The fact that there
is a great deal of variation both among the various non-Ana-
tolian daughter languages as well as within each daughter
language seems to indicate that there was some confusion
between competing forms in late Proto-Indo-European -- the
newer, analogical forms had not yet completely succeeded in
ousting the older forms at the time when the individual non-
Anatolian daughter languages began to emerge.
185
186
It should be clear that there is no need to reconstruct
long diphthong stems for Proto-Indo-European. Stems such as
*dh-ei-, *dhai-, *dht-, *dha- "to suckle,
suck" and *kZeu-, *kZo(u)-, *kZii- "to wash, cleanse", cited
above, go back to *day-/*day- and respectively
(in traditional terms *dhei-/*dhoi- and *kZeu-/*kZou-). It
should also be clear that laryngeals are not needed. Just
because some original long vowels can plausibly be derived
from earlier sequences of short vowel plus preconsonantal
laryngeal does not necessarily mean that aZZ original long
vowels had to have such a source. As we have seen, some
original long vowels can convincingly be derived from ear-
lier diphthongs.
The *a (> *e) *a (> *o) ablaut patterning must have
assumed an important role in grammatical categorization in
the emerging morphological system of post-Anatolian Proto-
Indo-European -- all of the older non-Anatolian daughter
languages attest to this (see, for example, the many com-
ments about the role of ablaut in Old Indic mentioned in
Burrow 1973). In order to bring stems such as *t'ik-,
*t'uk- (traditional *duk-), and the like into
line with the *a ablaut patterning, *a and *a were in-
serted before accented *t and
This development was restricted to certain specific gramma-
tical categories (such as, for example, the singular indica-
tive active verbal forms) -- that is to say, not every ac-
cented *t and *I was affected (cf., for example, forms such
as Skt. nom. sg. agnis "fire" and sunus "son" or Hitt. nom.
sg. 8aZ-Zi-i8 "glorious" and a-a8-8u-u8 ''good", which must
always have had i
We cannot rule out the possibility that the creation
of secondary diphthongs remained an active process over an
extended period of time. This would mean that some secon-
dary diphthongs would have been formed at an earlier period
and others at a later period. Moreover, it is entirely pos-
sible that some of the older secondary diphthongs might
have developed along the same lines as the original diph-
thongs:
Stage I Stage II Stage III
A. CTc CalC
CiC
calC
CeC
B. c5c
Ca!!C cue
Ca!!C
C5C
Indeed, there is evidence that just such a development did
take place.
In three roots in Old Indic, there is an apparent al-
ternation between -a- and -t-: khad-/khrd-, and
sadh-/stdh- (these examples are from Burrow 1979:69-80).
The following forms are recorded:
A. (s)khtd-
khed-
khad-
(s)khad-
B.
Bas-
sas-
c. si:dh-
sedh-
sadh-
*s'idh-
"to apply force to, crush, press"
in various forms, including Ved. kheda
"hammer"
"to eat, chew"
in the action noun khadana-, for example
"to order, COIIDII.and, instruct"
"to direct, bid, order, connnand"
in the past pte. sast&-, for example
"to attain one's aim, succeed, be successful"
in conjugation
"to go straight to any goal or aim; to be suc-
cessful, succeed, prosper"
in Gk. C<lus "straight"
The relationship between each of the various forms of these
three roots becomes clear if we posit original roots of the
type *etc-, which were first "incremented" by the insertion
of *a and *a before the *t, and which then developed as in-
dicated above. All possible forms are attested:
A. *C'i.C-
Skt. (s)khrd-,
- stdh- St.ij-,
B. *Caf:.C-/*Caf:.C-
Skt. khed-, sedh-
c. *C'iC- Gk. c.aus
D. *CeC- Skt. khad-, - sadh- sas-,
Two final notes may be added here: (1) I regard the
forms with -a- (Skt. [ s l khad- and sas-) to be analogical,
187
188
and (2) khid- and its variants may have contained a laryn-
geal.
The picture is still not complete, however, for we must
also consider how the laryngeals fit into this scheme. I
have relied heavily on Greenberg (1969:183-84) for guidance
in determining the probable developments here.
The loss of laryngeals in sequences such as *eHiC-,
*eHuC-, and the like resulted in short diphthongs (*eiC-,
*euC-, etc.) -- the preceding vowel definitely was not leng-
thened. However, the loss of laryngeals in sequences such
as *eHaxc- and the like did give rise to long vowels: first,
the loss of the intervocalic laryngeal produced a sequence
of two short vowels. Then, these vowels were contracted to
form a long vowel.
There is strong evidence that laryngeals underwent meta-
thesis in several cases. A good example is Hitt. 3 sg. pres.
"to pour" from PIE *ZeHw-, with
laryngeal preceding the *w, vs. the post-Anatolian PIE *ZewH-
!*ZowH-, with laryngeal following the *w, that is needed to
explain the forms in the non-Anatolian daughter languages:
Myc. adj. re-wo-te-re-yo /'AEFOTPELor;/ "for bathing", Hom.
'AoETp6v "bath", Att. 'Ao6w "to wash, bathe"; Lat. Zavo "to
wash, bathe"; Arm. Zoganam (< *ZowHye/o-) "to wash". Accor-
ding to Winter (196Sb:l91-92), the sequences *HiC and *Hue
were metathesized to *iHC and *uHC respectively. Winter
sets up this metathesis to account for the fact that the
reduced-grades of *eHi and *eHu are (< *iH) and *u (< *uH)
respectively. While I agree with the concept of metathesis,
I cannot follow Winter's theory that it took place in the
reduced-grade forms. Rather, I think that it would be se-
quences of the type *eHyV, *eHwV, and the like that would
tend toward metathesis.
The following chart illustrates the major developments
involving laryngeals:
A. eHiC > eiC eHuC > euC
oHiC > oiC oHuC > ouC
aHiC > aiC aHuC > auC
B. eHyV > eyHV eHwV > ewHV
oHyV > oyHV oHwV > owHV
aHyV > ayHV aHwV > awHV
c. iHC > uBC > uc
D. iHV > iyV uHV > uwV
E.
eHa:rC
> eaXc > ec
oHaXC > oaXc > oc
aHaXC > aaXc > ii:C
I will close this section by mentioning one last point.
The numerous cases of vrddhied stems in Old Indic represent
a contamination of types *CeC- and *CoC- with types *CeyV-/
*CoyV- and *CewV-/*CowV-, producing the new types Cay- and
cav-. These new types were an Indo-Iranian innovation and
do not go back to Proto-Indo-European.
10. NONAPOPHONIC *e AND *o:
Nonapophonic *e may be defined as an *e that does not
alternate with an *o in an ablaut relationship, and nonapo-
phonic *o as an *o that does not alternate with an *e. No
doubt some examples of nonapophonic *e and *o are due to
the fact that the corresponding ablaut forms, which once
existed, have simply not been preserved. In several cases,
Hittite even provides the missing ablaut forms for what is
found in the non-Anatolian daughter languages. Probable
examples are:
A. Hitt. ne-ku-U11 "bed-time, evening" vs. Lat. no:r, noatis
"night", Goth. nahts, Lith. naktis, etc. (cf. Pokorny 1959:
762-63).
B. Hitt. ne-ku-ma-an-za "naked" vs. Goth. naqaps "naked", Skt.
nagna-s, etc. (cf. Pokorny 1959:769).
c. Hitt. E (= E) "bone's house, sepul-
chre", ha-as-ta-(a-)i "bone, skeleton" vs. Gk. oon!ov "bone",
Lat. os, Skt. &sthi, etc. (cf. Pokorny 1959:783).
In each case, Hittite preserves the e-grade forms that are
missing in the non-Anatolian daughter languages. Even after
forms such as these are removed, however, there remain many
examples which are unaccounted for.
At least some examples of nonapophonic *e and *o may
have arisen from the influence of a contiguous laryngeal,
specifically a contiguous "a-coloring" laryngeal. The most
189
190
important point to remember is that all "a-coloring" laryn-
geals were characterized by the feature [low]. It was the
presence of this feature that was responsible for the low-
ering of *a to *a (cf. Colarusso 1981:504-15 for a discus-
sion of typological parallels with Northwest Caucasian de-
velopments). We should also expect *i and *u to be lowered
when next to these laryngeals, and indeed there seems to be
some evidence that this has in fact taken place. A few pos-
sible examples are:
A. *Aink- > *Aenk- "to reach, come to, arrive at" (Pokorny 1959:
316-18 reconstructs *enek-, *nek-, *enk-, Hitt. 3 sg.
"to present, deliver, offer, allot", Gk. i:vEyMELV
"to bear, convey", Skt. almoti "to reach, come to, arrive at,
get, gain, obtain; to master, become master of; to offer".
Hitt. may directly attest *Aink-.
B. *AUl- > *Aol- "to destroy" (Pokorny 1959:777 reconstructs
*ol-[e]-): Hitt. 3 sg. !Ju-ul-la-a-i "to smite, destroy",
Gk. li>.>.iilJL "to destroy", Lat. ab-oleo "to destroy".
C. *Aum- > *AOm- "all, whole": Hitt. nom. sg. !Ju-u-ma-an-za
"all, whole
11
,, Lat. omnis "all, every, whole".
At a later date, secondary e- or a-grade forms may have
developed in accordance with the regular *e *o ablaut pat-
tern (cf. Gk. OYKO(; "bulk, size, mass" from the same root
as that in tveyKEi:v "to bear, convey"). Where secondary e-
or a-grade forms did not develop, we would have examples of
nonapophonic *e or *o, as the case may be.
As in Gothic, *i and *u may also have been lowered to
*e and *o respectively before *r (cf. Colarusso 1975:392-93
for a brief discussion of the Gothic developments). Exam-
ples are difficult to come by, and there are even several
good counter-examples (such as, for instance,
*dhur-, "door" [Pokorny 1959:278-279]). In
spite of the fact that there are several counter-examples,
however, the sequences *ir and *ur were extremely rare at
best, and this may indicate an early shift of *ir to *er
and of *ur to *or. Since the resulting *er and *or could
quite easily have been absorbed into the regular *e *o
ablaut pattern, it may be next to impossible to sort them
out from *er and *or derived from other sources -- namely,
from *ar and *ar respectively.
REFERENCES
Arbei tman, YolH and Ayala, Gilbert-James. 1981. "Rhota-
cism in Hieroglyphic Luwian", in Yoel Arbeitman and
Allan R. Bernhard, eds., Bono Homini Donum: Essays in
Linguistics in Memory of J. Alexander Kerns.
Amsterdam.
Benveniste, Emile. 1935. Origines de formation des noms
en indo-europen. Paris.
Bernhard, Allan R. 1975. "An Outline of the Historical Pho-
nology of Indo-European", Orbis XXIV/2.
Bernhard, Allan R. 1976. "The Placing of the Anatolian Lan-
guages", orbis XXV/2.
Bernhard, Allan R. 1979a. "Typological Studies and the Iden-
tification of the Indo-European Laryngeals", in Bela
Brogyanyi, ed., Studies in Diachronic, Synchronic, and
Linguistics: Festschrift for Oswald Szem-
erenyi. Amsterdam.
Bernhard, Allan R. 1979b. "The Indo-European Phonological
System: New Thoughts about its Reconstruction and De-
velopment", Orbis XXVIII/I.
Bernhard, Allan R. 1981. "Indo-European and Afroasiatic:
New Evidence for the Connection", in Yoel Arbeitman
and Allan R. Bernhard, eds., Bono Homini Donum: Essays
in Linguistics in Memory of J.
Kerns. Amsterdam.
Brugmann, Karl. 1904. Kurze Grammatik der
indogermanischen Sprachen. Reprinted 1970. Berlin.
Burrow, Thomas. 1973. The Sanskrit Language. 3rd ed.
London.
Burrow, Thomas. 1979. The Problem of Shwa in Sanskrit.
Oxford.
Chomsky, Noam and Halle, Morris. 1968. The Sound Pattern
of New York.
Cohen, David. 1968. "Langues chamito-semitiques", in Le
Encyclopedie de la Pleiade.
Colarusso, John. 1975. The Northwest Caucasian Languages:
A Survey. Ph.D. Dissertation, Harvard
University.
Colarusso, John. 1981. "Typological Parallels between
Proto-Indo-European and the Northwest Caucasian Lan-
guages", in Yoel Arbeitman and Allan R. Bernhard, eds.,
Bono Homini Donum: Essays in Linguistics
in Memory of J. Kerns. Amsterdam.
Couvreur, Walter. 1937. De Hettitische g: Een Bijdrage
tot den Studie van het Indo-Europeesche
Louvain.
Crothers, John. 1978. "Typology and Universals of Vowel
Systems", in Joseph H. Greenberg, ed., Universals of
Human Language. Vol. 2: Phonology. Stanford.
Emonds, Joseph. 1972. "A Reformulation of Grimm's Law",
in Michael K. Brame, ed., Contributions to Generative
Austin.
Gamkrelidze, Thomas. 1976. "Linguistic Typology and Indo-
European Reconstruction", in Alphonse Juilland, ed.,
Linguistic Studies Offered to Joseph Greenberg. Sara-
toga.
Gamkrelidze, Thomas. 1981. "Language Typology and Language
Universals and their Implications for the Reconstruction
of the Indo-European Stop System", in Yoel Arbeitman and
Allan R. Bernhard, eds., Bono Homini Donum: Essays in
Linguistics in Memory of J. Kerns.
Amsterdam.
191
192
Gamkrelidze, T. V. and Ivanov, V. V. 1973. "Sprachtypologie
und die Rekonstruktion der gemeinindogermanischen Ver-
schliisse", Phonetiaa 27.
Greenberg, Joseph. 1966. Language Universals. The Hague.
Greenberg, Joseph. 1969. "Some Methods of Dynamic Compari-
son in Linguistics", in Jaan Puhvel, ed., Substanae and
Struature of Language. Berkeley and Los Angeles.
Greenberg, Joseph. 1970. "Some Generalizations concerning
Glottalics especially Implosives", IJAL 36.
Hirt, Hermann. 1921. Indogermanisahe Grammatik II: Der
indogermanisahe Vokalismus. Heidelberg.
Hockett, Charles F. 1955. A Manual of Phonology. Chicago.
Hopper, Paul J. 1973. "Glottalized and Murmured Occlusives
in Indo-European", Glossa 7.
Hopper, Paul J. 1977. "The Typology of the Proto-Indo-Euro-
pean Segmental Inventory", JIES 5/1.
Hiibschmann, Heinrich. 1885. Das indogermanisahe Voaalsys-
tem. Reprinted 1975. Amsterdam.
Jakobsen, Roman. 1971. "Typological Studies and their Con-
tribution to Historical Comparative Linguistics", in
Roman Jakobsen, Seleated Writings. Vol. I, 2nd ed.
The Hague. Report in the First Plenary Session of the
Eighth International Congress of Linguists, Oslo, Au-
gust, 1957.
Jonsson, Hans. 1978. The Laryngeal Theory: A Critiaal Sur-
vey. Lund.
Keiler, Allan R. 1970. A
European LaryngeaZs.
Kronasser, Heinz. 1956.
re des Hethitisahen.
Kurylowicz, Jerzy. 1935.
Kurylowicz, Jerzy. 1956.
Wroclaw.
Phonologiaal Study of the Indo-
The Hague.
Vergleiahende Laut- und FormenJeh-
Heidelberg.
Etudes indoeuropeennes. Krak6w.
L'apophonie en indo-europenne.
Kurylowicz, Jerzy. 1962. L'apophonie en stmitique. Wro-
claw.
Kurylowicz, Jerzy. 1964. The Infleational Categories of
Indo-European. Heidelberg.
Kurylowicz, Jerzy. 1969. Review of Robert Schmitt-Brandt,
Die Entwiaklung des indogermanisahen Vokalsystems,
BSL 63.
Lehmann, Winfred P. 1952. Proto-Indo-European Phonology.
Austin.
Martinet, Andre. 1970.
3rd ed. Bern.
Eaonomie des ahangements phontiques.
Meillet, Antoine. 1964. Introduation d l'ttude aomparative
des Zangues indo-europtennes. Reprint of 8th edition
(1937). University.
Moscati, Sabatino, ed. 1964. An Introduation to the Com-
parative Grammar of the Semitia Languages.
O'Leary, de Lacy. 1923. Comparative Grammar of the Semitia
Languages. Reprinted 1969. Amsterdam.
Pokorny, Julius. 1959. Indogermanisahes etymoZogisahes
W8rterbuah. Vol. I. Bern.
Puhvel, Jaan. 1965. "Evidence in Anatolian", in Werner Win-
ter, ed., Evidenae for Laryngeals. The Hague.
Puhvel, Jaan. 1966. "Dialectal Aspects of the Anatolian
Branch of Indo-European", in Henkrik Birnbaum and Jaan
Puhvel, eds., Anaient Indo-European Dialeats. Berkeley
and Los Angeles.
Pulleyblank, Edwin G. 1965.
tative Ablaut", Word 21.
"The IE Vowel System and Quali-
Ruhlen, Merritt. 1976. A Guide to the Languages of the World.
Stanford.
Saussure, Ferdinand de. 1878. Mmoire sur Ze systbme primi-
tif des voyeZZes dans Zes Zangues Re-
printed 1968. Hildesheim.
Schleicher, August. 1876. Compendium der vergleiahenden
Grammatik der indogermanisahen Spraahen. 4th ed., re-
printed 1974. Hildesheim.
Schmalstieg, William R. 1980. Indo-European Linguistias:
A New Synthesis. University Park.
Schmitt-Brandt, Robert. 1967. Die EntwiakZung des indoger-
manisahen VokaZsystems. Heidelberg.
Sturtevant, Edgar H. 1942. Indo-Hittite LaryngeaZs. Balti-
more.
Sturtevant, Edgar H. 1951. A Comparative Grammar of the
Hittite Language. Vol. I, 2nd ed. New Haven.
Szemernyi, Oswald. 1967. "The New Look of Indo-European:
Reconstruction and Typology", Phonetiaa 17.
Szemerenyi, Oswald. 1970. EinfUhrung in die vergZeiahende
Spraahwissensahaft. Darmstadt.
Winter, Werner, ed. 196Sa. Evidenae for LaryngeaZs. The
Hague.
Winter, Werner. 196Sb. "Tocharian Evidence", in Werner
Winter, ed., Evidenae for LaryngeaZs.
Wyatt, William F. 1970. Indo-European /a/. Philadelphia.
193

You might also like