Vous êtes sur la page 1sur 68

An overview over recent research

1 September 1983
Foreword. This essay surveys recent work on the
phonology of PrototndoEuropean, with a focus on the theoretical
assumptions underlying it, Other participants in the project are
writing essays on the phonoloQy of its subbranches, so that this
1s not treated here,The version produced nere is preliminary,
In the course of preparation its aims often seemed out ot reach,
Publication is enormous, making a comPlete survey impossible in
the space available. The diversity of views adds to the feeling
of inadequacy. In spite ot shortcomings, the sketch may assist
others in identifying problems ot interest and in locating
efforts to deal with those Problems, as well as in providing a
guide to the scholarship, I will be grateful tor comments which
will make the essay more useful in adding scope to tne
materials cited, by correcting inadequacies and shortcomings of
representation or interpretation.
Typographic note: Because a sizable number is being
run off, copies of this survey are being produced as standard
computer output. Diacritics and underlines available on such
printout obscure characters. Accordin11Y the devices used to
signal special matter, such as titles, citations and so on are
2
--- indicated with especially chosen devices: angled are
used for citations; CAPS for titles, out not in the
bibliography, where however characters are
transcribed or otherwise modified; diacritics, when not
essential to illustrate a point, have been omitted, It
may be useful to point out here that the representation, Panini,
has the authorization of even small desk dictionaries like
webster's Collegiate; other wellknown names like Kurvlowtcz
(not yet in the Collegiate) are inclujed nere on a similar
basts, Only a ew pages With many citations are produced
with the more expensive laser printer and the character set
developed for the Gothic etymological dictionary project,
1, Principal areas of advance, Major contributions of
the past five decades nave modified eKtens1vely the views on
ProtolndoEuropean phonology presented in the standard handbooks
bV arugmann (18971916, 1904), Hirt (192137, 1939) and Meillet
(1937), These contributions result on the one hand from a
different approach to the languaqe, on the other from two
tarreaching theories, the laryngeal tneory and the glottaliC
theory, of which esoecially the first has implications for the
treatment of suorasegmentals.
The change in aoproach took hold slowly but now is
widely evident, tndoEuropeanists nave come to assume that they
can propose several chronological stages ot ProtolndnEuropean,
The Phonological system of tne Protolanguage is no longer held
to be an inventnrv of elements determined largely by the
comparative method CCM) from tne proposed for the various
subbranches, Instead, a series of stages of ProtoIndoEuropean
are prooosed, stages show!nq modifications which are treated
......-------------------------------------- ------- -- ----------------------------------- --
3
/- primarily by the laryngeal and 9lottalic theories,
The advances have led to massive Publication. Measured
against the total amount Of publication in linguistics dnd
philology it may seem minute. Nevertheless, so much has been
published that a brief survey cannot nope to encompass all of it.
Total accounting would require a larQer than Bechtel's
of 1892, If we set 1935 as the start of our attention, we are
dealing with a period more than twice as long as Bechtel's, and
of much more massive support for scnolarship. Bechtel's major
references are central European, This area may still be the
heartland for lndoEuropean studies, though it has brought about
research elsewhere, as in southern and eastern Europe, in India.
Japan and the country sponsoring the present such a broa1
extent of IndoEuropean studies precludes of a
comprehensive survey like Bechtel's, much as the scholarshiP ot
his day precluded preparation of an elegant tool for research
like von Bander's of 1883, Instead, a survey today may best be
characterized as an overview with references to selected work,
which in turn provides further references to the huoe amount Of
publication,
Among these works are the organs of scholarship wnich
list current publications. Even mention of them miQht extend the
patience of the editor ot this votume, thougn one can scarcely
,.....--------------- ------------ --- -----------
4
avoid reference to journals like DIE SPRACHE and KRAfYLOS, or
bibliographical sources liKe 8IBLIOGRAPHIE and the
Modern Languaqe Association Biblio1raphy. In addition, the
reports of conferences sponsored by the Indogermanische
Gesellschaft and of others less frequent, include bibliographical
references in their statements on advances. Of the conference
reports, that edited by Winter, with its extensive bibliograPhy
bY Polome, remains one of the most imPortant for JndoEuropean
phonology (1965). Seebold's book on the resonants provides a
comParable survey of scholarshiP while illustrating tnrouqh its
restriction to the two resonants /w yl the lncreasinQ magnitude
of Phonological data and Problems its mastery (1972). In
his publications, Gamkrelidze includes references of further
interest, as in the translation by Boeder 1982. Merlinqen's
bibliography must be determined from reference$ in the text and
the footnotes (1983). Bomhard includes a comprehensive
bibliography (1983:293315),
The work listed tn these and many other references has
left virtually no segments of the system in the standard
handbooks untouched, Modifications in the vowel and resonant
sets of that system are required by the laryngeal theory, in
addition to the added phonemes; the glottallc theory requires
modifications of the obstruents, and With the laryngeal theory
brings about reconsideration ot the supraseqmental system
............--------------------"----- -- -"- _" _________ ----------------- --- -""- --------"---
5
The renewed interest in the Nostratic theory a1so
requires attention. This interest largely from
publications of Ill1cSvityc C19o8, 1971, 1975); see also
R as rn us sen ( 1 9 7 4 ) and J, amp r e c h t ( 1 9 7 6 > Co :n par i son of
ProtolndoEuropean with Oralie and ultimately Nostratic leads
proPonents of the theory to assume an obstruent set of
voiceless stop, voiced stop and glottaltzed stoP as well as three
laryngeals, To the extent permitted oy the scanty materials,
assumption of a Nostratic lanryuaqe familY permits application of
the CM to ProtoIndoEuropean itself, The Phonological system
proposed for ProtoIndoEuropean under the assumption resembles
that posited unrter the laryngeal and glottallc theories,
2, Theoretical views underlying the conclusions tn the
standard handbooks and subsequent wor(, While the distance in
years may have leveled differences in the theoretical views of
Brugmann, Hirt we regard tnese as virtuallY
equivalent. I characterize their theoretical position as
natural-systematic. In support of the characterization 'natural'
we may observe their treatment Of resonants as separate entities
when consonantal, vocalic and also vocalic, They maintain
this view even though Sievers had observed tl\e alternation
between consonantal and vocalic realization of the resonants
according to the environment. In support of tne label
'systematic' we may recall the Positing of six long vocalic
-------------.----
6
r resonants including long vocalic <l> or of four (voiceless and
voiced, aspirated and unaspirated) interdentals, regardless of
the lack of evidence for them. This theoretical approach led to
the assumption of a larqe number of phonoloq1cal elements (e.g.
Brugmann 1897:9293).
r-
The Phonemic approach, which involves classification by
function as well as phonetic properties, brought about a strong
reduction in number of elements. Meillet already assumed onlY two
sets of palatovelars, a position forcefullY maintained bY
Steensland (1973). And in two powerfUllY reasoned papers,
Edgerton Posited six resonants, with vocalic and consonantal
allOPhones, as well as a third alloPhone which in his notation
consists of a vocalic plus a constituent (1934,
1943), This third allophone removed of the hasts for schwa
secundum, Jt also led to dnalyses, such as
with its maintenance of schwa secundum + consonantal
resonant corresronding to Edgerton's third allophone
(1942:3132>. It should also be noted that some scholars
interpret the Sievers' law phenomena morphophonemtcally without
setting up the compact system of ProtoIndoEuropean resonants
proposed by Edgerton. Whatever the reaction of individual
scholars to details, no analysts of tne PrototndoEuropean
phonological system after 1943 can posit the arrays of resonants
tound in Brugmann and Hirt.
7
Of the methods applied in historical studies, the
comparative method was adequate to propose the modifications in
the resonant system. It was also an adequate basis for the
laryngeal theory as proposed by Kurvlotcz. His decisive
argument rested on of tne <h>elements of Hittite with
distinctive patterns in other dialects, as in Hittite <hanti>,
Greek <anti>. But the comparative method in a restricted sense
favored bY many scholars does not permit the assumption of more
than one laryngeal. such scholars firmly maintain a phonological
svstem for ProtoIndoEuropean witn one laryngeal cszemereny1
1970, 1980, though he posits for late ProtolndoEuropean
virtuallY the entire panoPlY Of Bruqmann 1897, even omitting the
grudgingly acknowledged single laryngeal).
Many scholars on the other nand place equal credence in
the method of internal reconstruction (tR). Their we!Qhtiest
example of its reliability derives directly from the evidence in
Hittite supporting Saussure's conclusions of 1879. Strong
reliance on lR is also evident in the 1935 monographs of
Kurylowicz and aenven!ste, which toqether with Edgerton's
articles, introduced a new Phase in tne study of IndoEuropean
phonology.
Other evidence was taken penetrating analysis Of
data in the welleXPlored dialects, d!ta we may characterize as
-
8
/- residues after explanation of the major phonological and
morphological patterns. Such analysis was carried out by careful
and wellinformed IndoEuropeanists. Examples are wilhelm
Schulze's fundamental paper on tne long diphthongs (1885, 1966),
KUiPer's detailed analysts ot the nasalintlxed verbs (1937), the
many treatments of Germanic long <e2> (Streitberg 1936:369),
Arnold's not yet surpassed treatment of Vedic meter (1905).
Benveniste, Kurylowicz and others drew heavily on authoritative
treatments of fundamental problems In restating the patterns ot
the ProtolndoEuropean sound system. Subsequent scnolars who
deal with the problems must acquaint themselves with those
earlier studies if they wish to achieve recngnition for their own
tormulations. IndoEuropeanists often assume those studies to be
selfevident fundamental works, so tnat current scholars may
receive little assistance tn determining them, as a glance at the
scanty references in any ot Kurvlowtcz's publications will
indicate.
Having based their innovative on respected
earlier worK, Kurylow!cz and Benveniste were wellreeeived.
Debrunner, for example, published a very revtew of
the proposed revisions in the phonological system (1929, 1937,
1938). The list of other reviewers indicates that by the of
world war IT there was strong support for the revised system.
9
world war II however scholarlY worK generally,
1n some more than in others. When that worK resumed,
the views of scholars like Debrunner were often disregarded, and
a phonological system was again proPosed that differs little from
srugmann's. Proponents of the superseded views will not be
listed here.
Yet assumption ot the laryngeal tneory requires
abandonment of Brugmann's svstem, The contrast between the
Anatolian languaQes, with written attestation of <h>, and the
other dialects with no overt attestation, demands positing
various staqes of ProtoIndoEuropean, An earlY effort at a
simPle solution led to the IndoH!ttite hypothesis (Sturtevant
1942, and now aqain cowgill, eg 1979), But this hYPOthesis
tailed to gain widespread acceptance, in part because otner
features of Hittite make that early language seem innovative. In
thiS way it is comparaole to the previously known dialects, Which
are conservative in some respects, .innovative in others, Xhe
onlY credible solution requires the assumption that the
protolanguage is itself a historical product which underwent
changes as one segment ot grammar affected one period,
anotner segment durinq another Period. All tne techniques of
historical linguistics must be to flesh out its
successive systems.
r
10
The techniques applied in the past be supplemented
with conclusions obtained in typoloq!cal studies. These combine
results from reasoned application of the observations achieved in
linguistic study. Rather tnan random selection among
observations made in indiVidual through typological
study all observations are coordinated in the attempt to
determine the statistically probable situation for any
linguistic event or process. The considerable attention in
recent years to phonological tyPology obviates any need to cite
details here.
In applying theoretical approaches we invite the
participation of. generative phonologists, as Dressler nas
proPosed, while indicating taSkS that Phonologists expect of
Indo!uroPeanists (1972, 1980), Yet nuch of generative phonology
has been taxonomic, recasting known facts in accordance with
generative ohonology Procedures. rne jata, both in Phonology and
morphology, have often been described, and analyzed through
theoretical procedures of the time, If application of generative
phonological techniques to ablaut and other morPhopnonemic
problems provides new insights, they would indeed be welcome.
But it must be noted that these involve the assumotion
of successive of ProtoIndo!urooean in cnanqe.
Generative Phonologists mlgnt also contribute to
11
InooEuropean studies by examining the of Brugmann's
earlY years to interpret phonology a
morPhological point of view. 8oth Brugmann and H!rt in their
earlY work, tor example, distinguished the <o> alternating
<e> from "original" <o>. It woUld be far more useful to have
that work and the reasons tor its abandonment examined by
specialists in generative phonology tnan to have wellknown data
reformulated in an idiosyncratic notation.
It is also important that such undertakings make use of
the most accurate earlier understanding of the data. Dressler
and Gross, in illustrating the approach with attention to tne
palatovelar series (1972), limit the value of their
demonstration by positing three members, in contrast with
widespread assumption of two.
The force of proposed conclusions depends equa11v on
application of all methods developed in 11nquistic study and
control of the data. In dealing with the limited data of early
texts the use of every available method is essential. But methods
are of little use if the data aPPlied bV its proponents are
inadequate or inaccurate.
3. The laryngeal theory. ot the larynQeal
theory, 1n wnich much was accomPlished tnltlallY to clarifY
unsolved Problems in IndoEuropean followed for a
--
.....
12
time by disappointments. some resulted from inadequacies of
data; Mycenaean Greek, especially with its tantalizing tlnsolVed,
residual symbols, seemed to promise a second subbranch with
written evidence for laryngeals. So far the hoPe is unfulfilled.
Other disappointments relate to inajequacies of theory.
The unambiguous data for laryngeals are confined to cuneiform
symbols represented with <h>, Palate <g> in transliteration.
Even these are not entirely unambiguous; some instances of <h>
represent reflexes of entities other than the laryngeals, rhe
problems are compounded bY continued lack of information on the
precise source of the Hittite script, The uncertainties block
advance in interpreting single versus dual representations of
<h>; the problem is sufficientlY imPortant to permit a brief
summary.
By interPretation (Unfortunately labeled
in an atavistic retention ot a dubious 19th century use of the
label the treatment of <h> was equated with that of <p t
k>, When medial, these frequently contrast single verstts dual
representation with the single representing voiced obstruent, the
dUal a voiceless; see, however, Jucquois (1972:86128), and also
(1975:37576), who supports the position of Sturtevant.
If applicable to <h>, this practice provides written evidence tor
two different laryngeals. Support or rejection of
13
interpretation requires thorougn application of the philological
techniques developed in the 19th century and all available
evidence on the background of the early contacts between
Anatolian& and the users ot the script from they
adopted it. To iate the evidence is inadequate to resolve the
issue (Ketler 1970:9295 and
Further, no one has provided a definitive
interpretation of scriPtio Plene for inttlal vowels, as in
<eeszi> (Sturtevant 195112324), If such representation
reflects glottal stop, Anatolian provides written evidence for a
third larvn9eal,The scantY Palate materials may provide sUPPort
tor assuming scriptio plene as notation for larynqeals.
has POinted out alternation <aa> with <aga> in several
morPhological categories (1975: 36061): the scriptio plene
writing here may well have been introduced to represent the
laryngeal. watkins himself interprets one such pattern as
1ndicat1n9 accent (367), A comPrehensive of the
evidence by Hart (1980) has convinced some that plene writinQ
indicates accent; the argumentation that Hart uses in opposition
to the earlier interpretation may however be equallY applied to
his own. No evidence has yet been provided to demonstrate that
the sequences in question do not provide support for a larynQeal,
The and problematic Anatolian tntormatlon nas
--------- - - -- - -- -- - - - ---
14
- brought about heavy reliance on the otner subbranches, where
evidence is not unambiguous, Moeover, since the evidence is
taken from patterns which include thP. greatest
difficulties in IndoEuropean such as the original long
vowels or the Germanic strong Class VI! verbs with long <e2>,
alternate are also offered bY scholars who do not
accePt the theory, Even reference to such Proposals is
impossible within the scope of this essay, To resolve such
problems the evidence must be scrutinized thoroughlY by rigorous
aPPlication of linguistic methods, The choice of metnods used to
the evidence therefore eniovs great importance in
evaluating the data apposite for the laryngeal theory, Methods
and their application therefore provide the basis tor reviewing
here briefly work on the laryngeal tneory and its impact on
lndoEuropean grammar.
3,1 Some scholars maintain sole application of the
comparative method, positing at best only one laryngeal.
Presentation of this Position is available in publications of
szemerenyi (1967; 1970:11423, 13031), szemerenyi's
presentation ts especiallY Vdluable because of his explicit
argumentation for his position.
A strong statement in favor of the preeminence of the
comparative method is available in the publications of Lightfoot,
15
such as 1983. It is ironic Lhat Liqhtfoot selects the Lachmann
1aw phenomena to demonstrate the predominant usefulness of the
comparative method, inasmuch as he accepts a 1uestionable
interpretation of the data. The other examples given in his
study provide illustrations of the danger inherent in putting
stock in a selected method rather than acquiring a ot
the data.
3.2 Whlle relying heavilY on the comparative method,
other scholars observe the systematic approach to the
phonological system noted in Brugmann. For tnese scholars the
laryngeals must be outfitted wtth an array of members, much as
are Brugmann's interdentals. The possibilities are enormous, as
tn the ventures of Martinet (1955:2t234l. Puhvel and Lindeman
are somewhat more restrained. Lindeman's array of laryngeals
consists of labeled wtth the sub-numerals <1 2 3> in
two sets, one of which is subdotted (1970:101).
3.3 A third group incorPorates the findings of the
comparative method and internal reconstruction with analysis Of
various IndoEuropean phenomena that had been well explored,
Following the tradition of Saussure anj Meilletf this group
includes Benveniste as one of its rigorous representatives, We
recall briefly some of those longstud1ed phenomena. They include
morPhological, syntactic and lexical as well as phonoloQ!Cal
-----.--.--- ------- ---
,...- ..
16
data.
3.3.1 Amonq phonological data the most important
well be those involving long vowels not or1;1nat1nq in ablaut. In
addition to analysis of tnem, they were intensively
scrutinized, as in the monograph of (1879), in successive
pUblications of Hlrt (1900, 192137 11:3136), and 1n treatments
bV the best and most reliable IndoEuropean1sts, such as
J. Schmidt C1889J, Schulze (1885) and (1896:8092).
Found in monophthongal heavy bases, such as
<(s)taa> <doo> as well as in diphthongal
bases, such as PIE <dheey> <taav> 'steal' and <poow>
their treatment under varytn; accentual and
morphological conditions sets them clearly apart from long vowels
resulting from ablaut. By aPPlication of internal reconstruction,
the original vowels whether standing before obstruents or
resonantsin the so-called long d!Phtnongsmav be reconstructed
as <eX eY eZ> (using here labels symbolic logic for
variables>.
The variants of oriqinal long vowels under lack of
stress, of which the most prominent ts traditionallY symboltzea
bY schwa, are generally accepted as additional evidence tor
1aryngea1s. Yet the place of schwa in the phonological has
remained a problem solved differently according to the
--------- --- --- --------- - - --
17
theoretical views of the scholars concerned. some assume
vocalized variants o laryngeal&, comparable to vocalized
resonants. Others assume a + laryngeal. Whatever
the analysis of the SYllabic nucleus itself, its relation to the
threefold reflexes in Greek, as in <thetos, statos, dotos>,
provides further Some scholars ascribe tne three
contrasting vowels to the three contrasting concerned
Ccf especially aeekes 1969). Others assume one reflex,
subsequently modified by the vowel under stress, Gk <thetos> by
<titheemi> and so on. Others totallY reject schwa, assuming
instead merqer of the unstressed nucleus with a short vowel,
typically <1> in IndoIrantan, <a> in the other dlalectSJ see
Burrow 1979.
Another pattern associated With research on laryngeals,
the prothetic vowel of Greek, has been reexamined at length by
Beekes (1969>. rlis scrutiny the assumption of two
laryngeals, plus the possibilitY of a third. Although one would
prefer more mature presentation ot opposed positions, such as
those of szemereny1 (Beekes 1969:27174), such monographs
detailed examination ot data cited in support of posited
laryngeal& are highly welcome.
3.3.2 Of morpholoQlcal data the sanskrit
verbs are amo"g the most imPortant, both because of the
18
interPretation provided by the laryngeal theory and the evidence
tor the position of laryngeals in the ProtoIndoEuropean system.
By the theory the three classes of <n>infixe1 verbs for tne
synchronically based Indic grammarians are reduced to one.
v. <no>, e.g. csunoti> 'press out'
VII. <naC>, e.g. <rinaKti> 'frees'
IX. <naa>, e.g. <punaati> 'cleanses'
The laryngeal theory simPlifies the morphology by
clarifying the basis for this transparently complementary
distribution. The tnree Sanskrit classes resulted from
phonological realignments of one earlier class. Clarification of
such morphological PrOblems rather th!n purely Phonological
discussion occupied the primary attention ot Kurylowicz in 1927.
such Clarification continues as supoort tor the
laryngeal theory. The support is especially impressive when
assumption of laryngeals on the basis of one pattern, e.g. 4 in
Table I, accounts also for another pattern, e.g. s. The tneory
accounts neatlY for patterning like that illustrated in Table I.
Many problems nave remained intractable, such as tne
number of larynqeals. To account tor the lack of <h>retlex in
Hittite cognates with <a> vocalism in some words as well as its
presence in others, assume1 two <a>eoloring
19
1arvngea1s for ProtoIndoEuropean (1935:25455), Usinq tne
established convention of representing these with H the tour are
as follows, with phonetic identification proposed oy Sapir: <Ht>,
glottal stop; <H2>, voiceless velar fricative; <H3>, voiced velar
fricative, with labialization; <H4>, glottal fricative (h), Many
scholars fail to include <H4>, in disregard of the Hittite
patterning.
Other patterning is even more tenuous. since SansKrit
Class VII verbs have an obstruent suffix, and Class IX verbs an
<aa> from <h>laryngeal, it seemed attractive to derlve Class V
verbs from laryngeals associated with <o> vowels, that is, a
laryngeal with labial coarttculation. My Linguistic Society
paper of 1952 sugqest!ng such a source, as Proposed earlier by
sweet, was never published. SubsequentlY Martinet assumed
several such laryngeals (1955), followed by Adrados (1961) and
Lindeman (1970). Convincinq evidence for that laryngeal has
still not been available; see
(1965:7879); subsequent Publications vary in their analysis of
that laryngeal, its allophones and reflexes.
Strong support for three laryngeals (H1, H2, H3l was
asserted by Beekes 1969 on the basis of GreeK reflexes Ce a oJ in
unstressed position. Kortlandt asserts flatly (19B4): "Since the
publication of R. s. P. Beekes' monograph on the development ot
r--- --------
-------- --
20
the ProtolndoEuroPean laryngeal& in GreeK (1969), the triple
representation of 'schwa' is a solved problem," adding:
"Lindeman's little book C19B2) has convince; me that nothing can
be regained from reopening the issue." In this way Kort1an1t
represents a firm espousal of three 1arynqea1s1 unfortunatelY he
does not on the evidence led Kurylowtcz, Sapir and
sturtevant among others to assume four,
Still other patterns of the dialects were explained on
the basis of the laryngeal theory, them is the elegant
theory of Germanic strong verbs presented in ProKosch's
posthumous grammar Of 1939, 6y this the early Germanic system
was arranged from parallel bases, here given with Roman numerals
indicating tne later classes:
Monophthongal simple: xv-v
+ laryngeal: VI
DlPhthonJal simple lIII
DiPhthongal + laryngeal VII
supporting such a syste are limited, Very few
"laryngeal bases" with reflexes in Classes VI and VII can be
determined. Even a clear laryngeal base like <steH4> > <staa>
was restructured, presumablY in preGermanic to <stad>, wttn
newly created <n> present. The tew ascertainable lonq
diphthongal bases of Class VII are equallY Problematic, thouQh
some have support!nq coqnates, liKe <leHYd> in Lith <leid>.
And !ndoEuropeanists nave not been deeply sympathetic to
21
- proPosals of semantic shift in the early dialects, including that
from aspect to tense in Germanic verbal inflection, Though
Prokosch's tneory provides reasonej explanation of the Germanic
strong verb system, the laryngeal theory nas never achieved
adequate authority to uphold his explanation coupled with its
bold semantic proposal. In recent treatments of the
redUPlicating verbs (Voyles 1981:12045 and van 1983,
with references to his earlier PUblications> Prokosch's proposal
iS disregarded and the preterite vowel "lonq <e2>" derived bY
"reanalysis" with no attention to the semantic Problems.
----
Otner applications of the laryngeal theory to difficult
problems may be found in Watkins 1975 for Kortlandt
1978 and in numerous later publications for Baltic and Slavic,
ConnallY 1979, 1980 tor Germanic, Kuiper 1978 and Beekes t981 for
Iranian, Peters 1981 for Greek. may well be the most
surPrising feature of all these recent is the acceptance
of laryngeal reflexes into the dialects, for Kortlanjt well into
earlY Baltic, Slavic and other dialects. After the period
extending 1927 to roughlY 1960, it was widely assumed that
many unexplained phenomena treated as reflexes ot laryngeals
should be accounted tor in other ways; some of course maintain
this position.
3,3.3 The major pattern in the laryngeal theory
22
has been applied to syntax nave to 1o with the neuter plural, In
1989 J, Schmidt accounted tor the use of neuter plural subjects
having singular verbs in Greek by equating its <a> with tnat of
the feminine singular, identifying its earlier use as collective;
see amonQ others watKins 1975:36268, who sets forth explicitly
the evidence for the laryngeal collective suffix, especially in
Anatolian, By the laryngeal theory tne earlier collective
was <n>, the fourth laryngeal ot Kurvlowicz,
A similar explanation solves a congruence pattern in
Germanic, specificallY in Old saxon. Here congruence moditers of
two or more individuals of different sex are placed in tne neuter
plural. the old syntactic pattern extenied from use of <h>
tor subjects to modifiers tn collective constructions generally
(Lehmann 1957>. By the laryngeal theory a semantic as well as a
phonological explanation is provided. A recent study has once
again attempted a phonological exolanatton of the Old Saxon
syntactic pattern with no consideration of semantics (Htersche
1980),
3,3.4 Some of the most impressive evidence for
Saussure's positing of coefficients' came from lexical
evidence. Such evidence is also prominent in the early articles
of Kurylow1cz and in his 1935 monograPh. But Benveniste's
monograPh of the same year has enjoyed greater receptivitY for
----------------------------- - - ----- --- --------------------
t
23
its presentation of the lexical now generally accePted
in IndoEuropean studies. Based on from the
earliest attention to lexical elements in 19th century,
subsequently bY Schleicher and Hirt tnrougn 1ncreas1nQ refinement
as in the works of Schr!jnen (1921) Meillet (1937), the
Indo-European theory of the root received its most elegant
formulation in Benveniste 1935.
Canonical root structure provides the possibilitY of
positing laryngeal& initially, as in PIE <a1> < <heq> 'lead',
and finallY as in <Cs)taa> < <teh> 'stand'. In roots with
obstruent as the other consonant, the posited laryngeal may be
identified, either by vowel colOring or bY aspiration of a
voiceless stop, as in the root for 'stan1'.
In roots with resonant, however, such as SKt <r>
'move', the laryngeal is dlfticult to identify; of the possible
initial consonants we propose <HJ> on the basis of G< <ornumi>.
it is often difficult to 11ent1fV a laryngeal
to a root. Here the problem results from the possibility of
suffixing a number of elements to an IndoEuropean root, as to
PIE <gher> 'rub', PokornY 1959:4394n: <qhrem> 'rub
apart'C458), <ghren> 'rub apart', further <ghrend> 'crush',
<ghrendh> (459), <ghreei> < <ghreH1Y> 'anoint,
smear' (457), <ghreeu> < 'rub hard, damage'
24
(46062).
Just as different resonants or obstruents be
affixed to a root, so may different laryngeals. Amon1 examples
are Sanskrit verbs inflecting in both the 7th and 9th classes,
such as <lunott>, <lunaati> 'cut'. tne suffixes are
1aryngea1s, they can only be diStinguished in some ot their
reflexes, such as the normal grade, an1 then only in some
dialects, are given 1n Table II.
In many patterns on the otner han1 the reflexes cannot
be distinguished. Examples may be pursued in handbooks like
Pokorny 1959, as tor <ker> 'toP' (1959:57477), <ker> 'grow'
(577>, <mel> 'qrtno' (71619), <men> 'think' with <mneH2> as
r- in Doric Gk <mnama> 'memorial' and <mneH1> as in OHG <firmonen>
'despise' (72628). There is no restriction on use of any
larynqeal as suffix to a root. Since of the reflexes of
larynqeals merge, the laryngeal suffiXed in any given form may
not be identifiable.
3,4 Requirements for acceptance of further
aPPlications. Unless uncover furtner texts containing
unambiguous evidence for laryngeals, provide the best
hope of advances in determining the esr11er phonological system
of ProtoIndoEuropean. In the nature ot scientific study such
residues have been examined earlier: in examination scholars
25
nave Proposed explanations tor and continue to Propose
explanations, 1th or without the of laryngeals. Any
proPoSed exPlanation accordingly requires from all
possible sources: related phonological evidence, morphological
evidence, evidence from patterning, both in the dialect concerned
and also in other dialects.
An Illustration May be taKen from attempts to account
for Greek <z> in words like <zugon> 'VoKe'. Ry the most
plausible hypothesis before knowledge of laryngeal&, they were
derived from an earlier palatal glide, differing in some way from
the consonantal variant of Cil. such hypotheses are empty,
simPlY referring the problem to an earlter Period, use ot sucn a
hYPOthesis is also a source of the many phonological elements in
the standard handbooks.
In his very brief attention to the laryngeals, Sapir
related them and their reflexes to other Phenomena of the
phonological system, (1938:24874), rne approach 1s admiraole,
bOth for solving the Problem ot initial <z> and tor examining any
otner problems. For in the absence of direct cognate elements,
such as the initial consonants of Skt <trayas>, Gk <treis>,
<tres>, NE <three>, our best support for proposing laryngeals in
the treatment of any Phenomena lies in the relationship of those
phenomena with others in the lanquaqe and in related languages.
---------- ---------------
26
see Beekes 1969:9395.
Three such bits of evidence are pertinent for the Greek
words with <z> = PIF. <y>, one in Greek, the others in Germanic
and Inct1c. The evidence consists in other sources of <z>,
notably voiced stops+ [y), as in <Zeus>< [dyJ as in Skt
<dyaus>, and in <mezon> < [qyJ, cf. 'Qreat', If the
pertinent words had etyma with Voiced larynqeal + [yJ, their
development is parallel; see Schwyzer 193953 1:33032. The
assumption on the basis of the Greek Patterning that this
laryngeal was voiced does not seem excessive.
Sanskrit sandhi phenomena provide evidence for initial
laryngeal in <abhiiyujas aayuk>. Anj the Alemannic contrast
between retention of a palatal glide in reflexes of PIE <H3y>,
such as <joh> 'yoke' and its loss in reflexes with only <y>,
such as cener> 'that one', provides further evidence for the
assumption of a more extensive element in the <Hly>
words as opoosed to the <v> words, Moreover, the Alemannic
pattern of retention of [j) supPorts assumption of incremental
voiced elements,
Derivation of the pertinent <z> <H3y> then has
various kinds of support. The earlier explanations, cited in
Schwyzer, lack such support; yet subsequently other such
explanations have been attempted, even through ascribing the
27
Greek <z>words to borrowing from a sPecific social group,
As noted earlier, a complete review of proposals
concerning laryngeal-related data m1Jht virtually be endless.
Beekes (1969:9697) is concerned at lack of rebuttal for some
such Proposals, examples of which are noted briefly here.
Cowgill (1960:16466, 161) doubts the value of Vedic <aayukta>
tor assuming an initial laryngeal on the root <yug>,, He also
provides his own interpretation of Skt <yasta>. Further, he
points to OHG <jaar> 'year' = Gk <hoora> 'year' as not in
accordance with my proposed rules, Since the loss of initial
<j> is found only in one Old High German dialect, the occurrence
of a form with <j> in that dialect aay be the result of spread
trom another dialect, The adverb <niuru> < <*hlu jaru> 'ln this
year' may be cited to illustrate the relatively weak initial.
See also Frisk 196072 111:191 on the disputed origin of Gk
<hoora>; data of uncertain explanation is scarcely ot great note
when applied to account tor other data,
Rather than countereVidence, such statements merely
indicate the views and favored methods of the proponent, Hitt
<ishiya> 'bind' may well be related to Skt <syatt>, though the
derivatives like <ishuzzi> 'qtrdle' are semantically closer to
Gk <zoostos> gtrt', <zoonee> 'girdle' than to derivatives of
csyat1>. And if Gk ceela> is not related to <zelai> 'spelt', no
~ ~ ------------- ~ ~
28
better exPlanation has been (Chantraine 196880:408).
Tnese brief comments are not onlY to
illUstrate the exte"t of treatment for relatively tew items or
the length required tor discussing fully the views ot otners.
TheY also once again how tenuous our evidence is. One
may consider the situation hoPeless and turn ones attention
elsewhere. If one does attemot to reconstruct earlier stages of
ProtolndoEurooean, including with laryngeals, evidence
must be assembled from all dialects and that evidence be
impeccable and it must be treated witn explicit use of available
methods,
3,5 A specific hypothesis for the origin of some GreeK
<Z> illustrates the advantage Of identifying the laryngeal&
phoneticallY. readily accept Saussures use of symools for
unidentified coefficients, since his assumptions were based
solely on use of internal reconstruction. rn view ot the state
of Hittite studies in the 1930s we also understand the use of
symbols without pnonetic content by Benveniste and Kurvlowicz.
aut continued use of such sYmbols nampers understandlnQ of the
IndoEuroPean Phonological system. lt is surelY preferable to
follow the lead of a linguist like SaPir, who had investigated
other languages whiCh included elements with
phonetic members articulated in the back of tne vocal tract. As
--------------------------------- --------- - ----------- --------
,-...,
29
Sapir noted, such elements undergo Phonological
developments t.n other language groups than the tndoEuropean,
Scholars positing laryngeals well follow his lead, wnetner or
not they accept the tour symbolS he proposed, given in Table
II,B, Those hesitant to use Phonetic labels for laryngeals may
find comparable objections to phonetic labels for the
IndoEuropean obstruents after proposal of the glottalic theory,
The resulting reconstructions may come to resemble the formulae
of chemists rather tnan the representations of languages today,
3,7 In sum, acceptance of the laryngeal theory depends
heavily on views concerning linguistic methodology, Scholars who
restrict their acceptance of methods to accePt at best one
laryngeal. such scholars must also restrict their concern for
the parent language, BV the comparative method alone one cannot
reconstruct anything but tne latest stage of
ProtoIndo European,
On the ntner hand, scholars receptive to IR as well as
to CM, and to conclusions based on general linguistic study,
accept both the laryngeal theory and tne posslbilitY of
reconstructing earlier stages of ProtoIndo European,
Acceptance of that possibilitY has lej to the proposal of tne
glottalic theory.
4, the glottalic theory, this term has been
---- - - -.- -- - - - - - - -- . - - - - - - .
30
introduced for hypotheses on the early lndoEuropean obstruent
sYstem, These hypotheses prooose that the series of voiced stops
in that develooed from earlier glottalic consonants.
Moreover, the series represented with the symbols <bh dh Qh 9Wh>
iS taken by some scholars to be a set of voiced stops at the time
when the obstruent system included glottalic elements.
The primary scholars Proposing and developing the
qlottalic theory are Tamas Gamkrelidze and Vyacheslav Ivanov on
the one nand, Paul Hopperwho has returned to his interest in
sYntactic problems on the otner nan1, Their views differ in
details. But tney agree in positing glottallc consonants.
In considering tne qlOttaliC theory it is to
recall that some earlier scholars interpreted PIE <hh dh gh gwh>
as fricatives, not as voiced stops with glottal increment,
Statements of this alternate system are accessible through the
references in Prokosch 1939:30304. That interPretation, which
has not been prominently mentioned by proponents of the glottalic
theory, would deprive the theory ot of its appeal. For it
was the unusual system of:
t d
"dh"
whiCh led to the to account in a more systematic manner
31
tor the early phonoloqical system.
Treatlnq the three-member system for its distinctive
features, others attempted to for it by examlninq the
distribution of the members within ProtoIndoEuropean roots and
their development ln Armenian and (Lehmann 1963, with
reference to a broadbased series published in Voprosy
Jazykoznanija 1961). My 1963 article viewed the entity <d> as
distinguished from <t dh> primarily through lenls articulation,
and the two fortis members <t dh> distinguished trom each other
bY aspiration, with voicing as a secondary correlation.
4.1 The traditional obstruent system. The proposed
system consisting of voiceless stops, voice1 stops and
fricatives, which corresponos closely to the obstruent systems ot
many wellstudied lanquages, never was widely accepted. The set
of obstruents posited by Bopp and other early IndoEuropean1sts
on the basis of seemed to be well-supported by the data
in Indic, and was also a Plausible source tor the obstruent
systems of the dialects. For its unusual members, <th> and <dh>,
were precisely those lost over a broad span of dialects. And
phonologists seeking balance could scarcely find a more
symmetrical set than:
-
32
t d
th dh
Dialects like Baltic, Slavic, Iranian, CelticwhiCh
maintained only the two top wembers, also had symmetrical sets;
these could be derived from the posited set in Proto-Indo
European by the simple processes: loss and And the
systems of Armenian, Greek, Italic and Germanic, which maintained
one representative besides reflexes of the two top memoers, were
readily accepted after linguists in the 19th century came to
understand Phonological relationshiPs and phonological change.
Hence the set based on Indic was posited generallY for
ProtolndoEuropean in the standard handbooks.
But the symmetry of the set was destroyed by the
laryngeal theory. By it the voiceless aspirates ot Indic are
derived through a merger ot voiceless aspirates + laryngeals ith
voiced asPirates which in some became voiceless
CKurylowicz see however Kiersche 1964. This imPact
of tne laryngeal theory was not widely noted by Indo
Europeanists, having been made after the major handbooks by Hirt
(192137) and (1937) had apPeared. Even Prokosch, who
had always assumeri the set written as voiced asPirates to be
voiceless fricatives, failed to use Kurylowlczs observation in
support of his view (1939:3841, 5759).
/"""'
33
Most lndoEuropeanists devoted little attention to the
obstruents, much as they had accepted the lack of evidence for
<b> since the days of Schleicher, rypoloQ!cal observations,
however, prompted closer attention. This achieved broad notice
through Roman JaKobson's Presentation at the Eightn Congress Of
Linguists, Oslo, 1954; Jakobson in turn may have directed hiS
attention to the problem because ot an observation made
personallY by Haudricourt, In his article of 1965 Haudrtcourt
states that he proposed the new interPretation to linguists in
New York on the basis of the Vietnamese consonant system,
Additional attention came only later, in virtually
simultaneous publications tn two different quarters, av 1973,
typological studies in ononology had assembled considerable
information on patterning of PhOnological systems, This
information provided evidence that a qap in a glottalic series is
often found tor the labial order, starting from this observation
both GamKrelidze/Ivanov and Hopper proposed that the absence of
<b> in ProtoJndoEuropean can be accounted tor by assuming it
was the expected gap .in a glottaltc sertes,, The gap is typically
found, however, in a voiceless rather than a voiced, series, The
ser!es<(b) d q qw>was therefore assumed to have dev@loped
from earlier <CP') t' K' k'w> <' here indicates glottalic
articulation),
34
This assumption requires a modification in the
obstruent system of ProtoJndoEurooean. For if the set <t d dh>
was odd, <t t' dh> is scarcely an improvement, Hopper Proposed
as his solution <t t' d>, the 1ast accompanied by murmur.
as theirs, <t'> and <t d>, the last two with
redundant aspiration. By each solution the troublesome labial
gap was now accounted for.
But the cost was not small. While in the past onlY
Germanic and Armenian required the assumption of extensive
consonant shifts, now comprenensive modifications had to be
assumed for atl of the dialects. The extent of these may be
observed in an article by Kortland Cl978a), wnich assumes a
system without voicing: lenis aspirate <d/dh>r lenis qlottalized
<t'>; fortis plain <t>. In support of a system lacking voice as
a distinctive feature in the obstruent system Kortlandt cites
voiceless <s> which lacks a voiced counterpart. wnile an
impressive argument in his favor, the continued lack of voiced
sibilant Phoneme tn dialects like Sanskrit and with voice
as a distinctive feature poses a problem. Kortlandt also taKes
no note of the absence in ProtoIndouropean of roots like
<tedh dhet>, in which by his system there is a constraint
against sequences of lenes aspirates and fortes plain.
4,2 Efforts to provide suPport tor the new system, The
35
glottalic theory was proposed without application of the
established methods of historical linguistics. Neither the CM
nor IR was used in arriving at it. Its sole basis was
typological. subsequently proPonents have sought support for the
new system phonological Phenomena that have resisted
explanation, in snort, by analysis of residues. At least three
such phenomena have been cited.
Before we note tnese, a oriet on terminology
may be useful. ro avoid the connotdtions of tne <d t dh>
symbols, several devices have been used. Hopper returns to
Grimm's old labels, mediae, tenues, asptratae.
Gamkrelidze/lvanov use the term 'series,' referring to former <d>
etc. as series t, to <t> etc. as Series II, to <dh> etc., as
series Ili. Individuals will no doubt choose their symbolization
bY attempting to balance the awkwar1ness of those new devices
against the danger of misinterpretation of conventional <t d dh>.
4.2a One support is Winter's proposal that ProtoIndo
European short vowels preceding glottalics were and
articulated with acute accent in Baltic and Slavic
1979:43144). The change did not taKe place before Series 111
stops, providing contrasts as exemplified in Table III. Some
upholders Of the theory, Kortlandt (e.g. 197Ra:117),
reter to the formulation as Winter's law.
-------
36
The proposed change unfortunately tails to tal<e place
,-
1n all forms of the specified environment. Winter himself listed
six notable exceptions (1979) and to account for these.
He suggested, for example, that a word liKe OSl <voda>
water' was a borrowing from Gothic, <bogu> 'god', a borrowing
trom Iranian, an1 that Lith. <padas> 'sole of the toot' 1s not
related to PI! <Ped> toot'. The relatively large number of
exceptions is and requires explanation, whicn Kortlandt
attemPted to flesh out <1979:6061).
4.2b a second set of reflexes supporting the
assumption of glottalics Kortlandt proposed tne forms noted by
Lachmann in Latin (1978:117). These nave lengthened vowel before
series I velars, e.q. ra:J as in <actus> beside <ago> 'lead' in
contrast with short vowel before Series II velars, e,g. <factus>
beide <facio> 'make'. But this set too is not without
exceptions, as has been noted in the many attemots to account tor
the Latin forms. A still greater difficulty results from the
sequence of changes: devoicing Cdeglottalization) of velars;
lengthening of vowel. Devo1cin9 is found other dialects as
well and this is ascribed to ProtolndoEuropean under CM,
lengthening is found only tn It ts scarcely credible that
Italic escaped the devotcing assimilation, Unless it did, Italic
and its dialects possessed no trigger to produce the
lengthening,
------- ---
37
4.2e A third set ot data Kortlandt finds in
IndoIranian, with reference to the forms treated oy
law C1978a:11718). In to Skt, <dugdhan>
from PIE <dhewgh>, with Series ! consonants do
not have a voiced cluster, e.g. Skt <vuktah>, Av. <yuxta> from
PIE <yug> to Kortlandt, the "initial
glottalization" in and the fOllowing tortis stop
prevented voicing. Yet in this same section Kortlandt ascribes
the change of <p> to <b> in Skt <upabjdh> Av
<frabda> 'forefeet' also to glottalization. For this third set it
iS difficult to escape the conclusion that the argumentation is
strained.
4.2d The proponents nave reinterPreted wellknown
data in accordance with their theorv, e.g. the absence o roots
with voiceless stops and voiced aspirated stops, sucn as
*tebh and with two voiced stops, such as *deg. Here again tney
fail to mention the earlier assumption of dissimilation, as in
Lehmann 1963, which accounts more simPlY tnan does the QlOttalic
theory for the constraints on root structure by positing a single
cause.
4.2e Another wellknown situation is the scarcitY of
voiced Cglottalic) stops (Series Il in affixes and the lower
frequency overall of these consonants as opposed especially to
....---------------- --.--- -- -- - -- - -- -- --- - - -----
3B
series III and II. ret similar frequency relationships are noted
tor Middle Malayalam <b> as oPposed to <p, bh> with no suggestion
of glottalics (Puthusseri 1973:147).
4,2 In a recent unpublished paper GamKrelidze also
cites the relative great frequency of the series I labio-velar
<gw>, indicating that typological research supports the
expectation of finding more labioVelars than velars in a
glottalic set.
4,2g In addition to arguments based on general
linguistic findings and on distribution of elements in the system
evidence has also been proposed from surv1v8ls. In Latvian and
zemaitian dialects of Lithuanian Kortlandt has accounted for tne
glottalic tone of vowels which were originally pretonic and
preceded Proto-tndoEuropean glottalic stops as reflex ot the
ProtoIndoEuropean glottalic teature (1977:31930, 1984u:45>.
Moreover, he ascribes Baltic acute accent to glottal stop,
resulted from merqer of that feature and reflexes of laryn;eals.
The merged glottal stop also has cnaracter!stic reflexes in
Slavic C1984u:S6). In addition he accounts for the Armenian
consonant shift and especiallY the phonological systems of
several Armenian dialects through assumption of glottalics
(1978c, 1984u:612). Further, he the ot Sindh1
as reflexes of qlottal1cs1 similarlY, supraseqmental phonemes of
39
Panjabi (1984u:13).
Proposal of such late retention ot reflexes of
laryngeals and glottalics is in sharp contrast to the position
held bY most IndoEuropeanists the last two decades.
Further support for the views propounded especially bY Kortlandt
1s provided by tne attribution ot Nellknown ohenomena in modern
dialects, like glottal effects in Danish and other Germanic
languages to articulatory patterns as early as ProtoGermantc
(see especiallY
In this way the proponents of the gtottalic theory seek
supporting evijence from residues. They must also account for
the far more massive set of obstruent changes into the dialects
than is required tor the earlier system; see Kortlandt
1978a, with rejection of longaccepte1 developments, such as
Grimm's law, and vennemann's striking reinterpretation of the
Germanic events (t9A4>.
5. Pre-IndoEuropean suprasegmentals. Accounting for
the accentual system of the early language one ot the
major problems of IndoEuropean phonology. To explain most
simPlY IndoEuropean ablaut we must assume an early period
stress accent, followed bY a subsequent period with pitch accent,
and thereupon in most dialects the introduction of stress accent,
as in the modern subbranches of IndoEuropean.
L _________ ------- -
40
.-- 5,1 Pitch accent and <e : o> ablaut. When we examine
,,--..
the position of Pitch accent and its effects on the phonological
system, we find that its imPact was not great. The major effect
proposed is <o>ablaut, brought about in accordance the most
widely accepted assumPtion as an etfect of accent shift. In
short, pitch accent seems like an intrusion of a tonal system
into IndoEuropean.
Assumption of glottallcs oermtts us to account tor its
introduction, Recent attention to phonology has led to improved
understanding of the origin of tones in lanquage, the process
known as tonoqenests. A readilY accessible essay on tne process
ts that of Hombert et al (1979), My own acquaintance with
tonogenesis its possible apPlication to PrelndoEurooean I
owe to Kenneth Gregerson, at the time of the Linguistic Institute
of 1976J his late student, Marvin Matners, was explorinq the
effects in Vietnamese at the time of his death.
Of interest is tne proposed source of tone distinctions
from postvoca!ic glottal consonants, including glottal stop,
When these are lost, a high tone or rising tone may result on
the preceding vowel. on the other hand, with following
[hl are accompanied bY a lower fundamental frequency, which may
result in a low tone on loss of the Cnl. Since the
ProtoIndo European laryngeals <Ht> (glottal stop) and <H4>
---------- ------------ ------ ------------------------
41
C[h]) were lost at the same time, a contrast between hign pitch
and low pitch could nave arisen in roots PIE <dheH1> as in
Gk <titheemt> and PIE <Cs>tehH4> in Gk cntstaami>, The loss of
postvocal1c glottaltcs, as in roots like <ed> 'eat', <ped>
'foot*, csed> 'sit', <weg> 'weave', provided many more
possibilities for introduction ot tone. onefittn of
the roots in Whitney's handbooK on Sanskrit roots end in voicea
stops, many ot them reflexes ot the proposed series 1 elements,
There is then an ample basts for the origin of ProtoIndo
European h1Qh pitch,
In accordance with the accepted theory, ablauttng <o>
developed from <e> when the prtnctpal (Pitch) accent shifted, and
the earlier <e> came under low Pitch accent.
The process by which the <e : o> contrast came to be
used for morphological Purposes is inaccessible to us, The <o>
was available in roots like <sed> in derived formations, e,g,
Gk, <hodein> 'sell*, Oir. <adsuidi> 'postpone', Gothic <satjan>
'set'. Similarly, in <o>stems, like Gk <logos> word beside
<leg> 'assemble', in <a>stems like uat <toga> 'garment' beSide
<teg> 'cover'. Further, in Perfects like <eiloxa> beside
<leq> 'assemble'. Yet the process bY which such co>
generalized has not been
In all of these morphological patterns the use of <o>
--------------------------
-
42
ts not general. Tne incidence Of its occurrence may be compared
with that of umlaut vowels to maKe PlUrals in German. Both Point
to a situation in which features were spread outside a focal area
and never became thoroughly dominant.
such a between the introduction ot pitch accent in
late ProtoIndoEuropean and the loss of glottalics may provide
greater credence for the Qlottalic theory. As with earlier
illustrations I do not pursue this hYPothesis further here. I
must Point out however that the time of loss proposed here is
earlter than that assumed bY Kortlanjt. Rather than elaboration
of such differences my primary interest here has to do wtth the
effect of theoretical views in regard to recent theories
concerning the reconstruction of the ProtoJndoEuropean
phonological system.
5.2 Scholars agree in general on the presence in early
ProtoIndoEuropean of a stress accent, to which zero grade is
attributed as well as lengthened grade CHirt 192137 ti:3651,
IV:34955, V:2829, 199224>. Vowels affected by accentual
shifts merged at various times wttn vowels affected by
laryngeals. The several stages of snift and merger require
extensive treatment. No thorough statement is available, though
pertinent volumes of grammar contain much of the data and
can be interpreted in sympathetic reacting. successive stages
43
were sketched in PrototndoEuropean Phonology (1952:11214).
Lindeman has proposed stages of development for some of the
phenomena which correspond in general to the 1952 arranqement
(1982:6970),
6, Current situation. In reviewinq the results of
recent proposals concerning IndoEuropean phonology we must note
that acceptance of the views sketched above requires major
modification of the system presented in the handbooks,
The views of the resonant system were modified through
application of phonemic theory, ijestdes reducing to six Phonemes
the traditionally assumed consonantal tr 1 n w yJ, their
vocalic allophones, and these with an increment, e.g, Edgerton's
[1yl etc., the new formulation results in a short vowel
consisting of three members: /e a ol and a central vowel
characteristicallY found in unaccented syllables AS well as five
long vowels: 11: e: a: o: u:/,
The laryngeal theory requires positing a kind of
phonological element not recognized tn the traditional handbooks,
of which the number varies among scholars; most assume three
disregarding Kurylowtcz's observation ot the need to assume a
fourth on the basis of Anatolian coqnates lacking <h> oeside <a>
in other dialects. ses!des bringlnq an increase in of
phonological elements, the theory removed the long vocalic
-------
44
resonants, including [1 ul ln earlier stages of ProtoInao
European; also, [e a ol, wnen or1q1nally long, and [ph th kh
kWh].
The removal of the asPirated voiceless stops is largelv
responsible for the further scrutinY that led to the glottalic
theory, which requires revision of the previously best
established members of the system, the obstruents. what had
seemed one of the most solid achievements of 19th century
linguistics is now modified in every section,
The modifications had a turtner consequence. The
svstem is now reconstructed in a series of changes, For example,
laryngeal& are not proposed tor every environment in the last
predialect stageJ gradual change is assumed for other elements
as well. It is one of the important tasks of current study to
determine these successive stages of tne phonological system, and
to link them with stages in the syntactic and the morphological
svstems.
The new proPosals are by no means Pverywnere accepted.
To gain acceptance, specialists must fulfill a number of
expectations,
1. Procedures and methojs, the importance of rigorous
phtloloq1cal procedures. In view of the increasing tenuousness
45
of the data, as earlier and earlier stages are reconstructed,
strict attention must be given to evaluation ot the data. one of
the earliest syntagms singled out tor reconstruction may be cited
as example, the expression for glory',
The complex cluster in the lends support to
Kuhn's assumption that the Phrase is archaic, one that could be
reconstructed for the parent language. Yet in the 19tn century
little attention was given to the order of its constituents, or
to their formation !n the changing structure of the language.
The few Vedic and Homeric occurrences point to tne sequence:
adjective noun. In the majority of occurrences in the earliest
texts the adjective precedes, as we expect of ov languages.
Further, the tndic text twice gives the adjective as an !stem,
rather than in the late thematic form. Yet scholars persist in
reconstructinq the syntaym on the basis of GK <kleos aphthiton>,
Schmitt even attempted to argue in favor of this transparently
later form and order (1967:6180),
In much the same way scholars may fail to discriminate
between Old and New Hittite, between older segments of the
Homeric poems and newer, between archaic and innovative texts of
dialects attested only late. Procedures for identification of
older texts are not always simple, Late poets may arcnatze, But
the data of texts must be evaluated tor reliao111ty and
1---- -- - --- -- - - - - - - - - -- --
-
46
chronological sequence before analysts with linguistic methods.
Just as linguists dealing with lanquages spoken today must
control the requisite anthropological procedures, so linguists
concerned with texts must control Philological procedures.
7.2 APProPriate use of metnojs. Reconstructed
languages have been pos!tea as dialectfree; only one
form has been assumed for a lexical item or morPhological
category. Recently this assumption nas been modified, so that
even reconstructed languages like Proto-InooEuropean are viewed
as "natural languages," consisting ot a "group of dialects"
(Palmaitis 1982>. This view comPlicates teatment of tne data,
for it requires analysis for registers, like poetic language, as
well as for social and geographical dialects.
such an approach also of the data as
speech, not merely an abstract system. Detailed analysis in this
sense may provide inferences phonetic content. As an
examPle I cite Kortlandt's analYsis of Slndhi <jiu> 'living
being' and other forms from the root. Kortlandt Recounts
for the nonglottalic initial stop in this Sindhi set through
loss by dissimilation before the glottal closure "which had
develoPed from the laryngeal ot PI 'live'" (1981>.
BY Sapir's analysis the laryngeal in this base was a glottal stop
<H1>. It Kortlandt's proposal is accepted, it provides evidence
47
in favor ot the reconstruction with <Hl>, and also for Qlottal
stoP whevever the "original long" vowel <e> occurs. Continued
scrutiny of such patterns roay provide even greater precision in
determining the phonetic descriPtion ot the laryn;eals.
Similar scrutiny is necessary of evidence proposed in
favor of the new theories, Since some of this is typological
evidence, we must draw on patterning in all available languages.
In dealing with the conclusions based on absence ot <b> in
ProtoIndoEuropean, we must note tne loss in celtic of the
voiceless labial, as in Otr <athir> = Lat <pater>, By the
argumentation proposed for Celtic <p> may
also have been lost because it was qlottalic; but the lack of
glottalic dentals and velars dampens such an assumption.
Finally, it may seem that major departures from the
standard formulations in the major handbooks are radical
innovations, foreign to the solid achievements of the 19th
century. on the contrary, tney merelY continue the earlier
activities. Some of the new proPosals will no doubt be rejected,
as were "theories" proposed bY scholars of tne past. Others Will
no doubt become part ot the new standard formulation. The
increasing acceptance of the theory suggests that the
phonoloq1cal system will be reformulated to include consonants
comparable to those attested only in the Anatolian languages.
r------- ---
1
,.--.,
48
a. ProtoIndoEuropean as a natural language. rne
reformulation will follow in great part because of a departure
trom the restricted aims announced by for his day
(18971916 I:ixX), He envisioned a time when linguists would
treat ProtoIndoEuropean as a natural language, But for him the
time was not yet at nand; the data were still too unmanageable
tor such a step, All that was possible in his eyes was a
collection of the data, with no reconstruction even to the extent
of Schleicher's fable. the subsequent grammars of Hirt, Meillet
and KurylowiczWatKins observe Srugmann's self-imposed
restriction.
Attempting to carry out Srugmann's ideal aim raises
manv questions. Should Indo-European phonology be tne major
guide in this aim? How can morPhology anJ be correlated
with it? Can we correlate further the conjectures on the
lexicon, as for example in Specht 1944:1114? will
archeologists contribute to the new effort?
9, f:ar lier stages of ProtoIndoEuropean and poss 1 ble
relationship wltn otner protolanguages. Archeology has
demonstrated how ricn was the cultural diversitY of early periods
which linguists have largely dtsregarjed. know that a
succession of cultures preceded that of each oranch of
IndoEuropean, whatever its location. These cultures centered
49
around social groups which in accordance with precepts
must have corresponded to linguistic groups. This social and
cultural situation suggests one possibility tor varied
developments !n the various subgroups of ?rotoIndoEurooean.
For if the Greek set of gods owes its special features to earlier
inhabitants of the Hellenic peninsula, the Injic set to gods ot
the area occuoied by the IndoAryans, language also may have been
affected and may reflect characteristics of the preceding group.
Linguists now investigate such influences without
proposing substrata and superstrata effects, often were
aPplied with a heavy hand. Linguists must also oe more flexible
in their interpretation of as illustrated by
brilliant essay l1984); discussion of this must be
lett to the chAPter on Germanic.
After long disregard of stages of Indo
European, other than the widely discredited IndoHittite, some
scholars have ventured opinions on relatively large periods.
Meld (1975) dealt primarily with successive branching oft of
subgroups from the parent stock; see also Kuhn 1978 and Joseph
1980. Neu (1976) on the other hand, Proposing stages of the
verbal system, recommends the of models which permit
detailed scrutiny of the evidence on Which hypotheses are baSed.
That evidence must include the PhonoloQtcal data underlYing the
------ ----------------------
-
50
stages noted in 5.2 above,
Efforts to determine areal relationships for Proto
IndoEuropean and also to proPose relationships with other
reconstructed languages should frame their hyPotheses with
reference to the earlY stages, and also take archeological
evidence into consideration,
Since IndoEuropeanists now agree that the speakers of
ProtoIndoEuropean expanded from an area north of tne Caspian
and Black Seas, and since it is likelY that they adopted
innovations in the production of metals from the caucasus, the
most likely areal influence of ProtolndoEuroean came from
Kartvelian, Recent attention to its etfects is accessible
through Gamkreltdze and Macavariani 1982 and Schmidt 1983b,
Indo-European and otner reconstructed languages are
also being widely explored as noted in section 1 above, As early
as 1973 Jokl, in an important monogra9h tor its careful treatment
of data and its calm presentation of hypotheses, accompanied by
central references, acKnowledaed the efforts to establish
proto-relationships to have become a Kind of epidemic
(1973:35374), He also pointed out the problems of
distinguishing the tenuous evidence from chance similarities as
well as the difficulties caused bY onomatopoeia, If there ts any
one solid result of glottochronoloqy, it is the demonstration
----------------------------- ------ -------------------------
51
that eight millennia of separation the evidence tor
establishing relationships virtuallY to that obtainable from
chance similarities. And since it is scarcely Possible to set a
date later than sooo e.c. tor an earlv stage ot
ProtolndoEuropean, proponents of further relationships must
recognize that they are dealing with very slight evidence.
Yet work in this hazardous field continues. Bomnard
(1983) directs his Primary attention at possible relationshiP
with AfroAsiatic, as do Makki Cms> and others. The
modifications proposed for the pnonologiCal
system by proponents of the laryngeal and glottalic theories
it difficult for such efforts to establish credence for
themselves. It would be useful tor all concerned to revtew the
work of predecessors, like Cuny, ro solve the increasingly
difficult problems without textual evidence but only inferences
from linguistic and archeological data we need cooperation as
manifested in the undertaking we are now completing, and
willingness to draw on theoretical advances of all kinds in
dealing with the tenuous evidence.
--------- ------------
52
BibliograPhY
Adrados, Francisco Rodr!Juez. 1961, soore
las lar!nqales indoeuropeas. Madrid: Instituto "Antonio de
Nebrija,"
-.------
1981. Further considerations on the
phonetics and morphophonemics of <HY> and <Hw> in Indoeuropean.
Emerita 49:23171,
Allen, William s. 1976, The PlE aspirates: phonetic
and typological factors, PP. 2l7t9 in Lln1uist1c Studies
offered to Joseph Greenberg, It. A. Juillard, Saratoga:
Anma Libri.
Arnold, E. Vernon. 1905, Vedic in its
historical development. Cambridge: University Press.
Bahder, Karl von, 1883, Die deutsche Philoloqie im
Grundriss, Paderborn: Schoentngh,
Bechtel, Fritz. 1892. Die Hauptorobleme der
!ndogermanischen Lautlehre seit Schleicher.
vandenhoeck & [Reviewed by H, Moellerz ZDP
25(1893):36694,)
Beekes, R,s,P. 1969, The development of the
ProtoIndoEuropean larynqeals in The Hague: Mouton,
----------------------------------------------------- --
53
------- 1981, The neuter plural and the vocalization
of the laryngeals in Avestan, IIJ 23:27586.

1982/83. on larvngeals ana pronouns. zvs
Benveniste, Emile. 1935. Origines de la formation des
noms en indoeuropeen, Paris:
Birnbaum, Henrik, 1975, Genet!sche, typoloQ!sche und
untversale Linguistik, Folia Linquistica 7:22124,
Bomhard, Allan R, 1983, Toward ProtoNostratic: A new
beginning, Amsterdam: Benjamins,
Bruqmann, Karl, 18971916, vergleichende Laut,
stammbildungs- und Flexionslehre der indogermanischen SPracnen,
2nd ed, Strassburg: Truebner. 2 vols,
-------
1904, Kurze vergle!chende der
1ndogerman1scnen Spracnen. Strassburg: rruebner,
Burrow, Thomas. 1979. rne problem of shwa in
Sanskrit, Oxford: Clarendon,
Chantraine, Pierre. 196880, Dictionnaire
etymologique de la langue grecque: Histoire des mots, Parts:
Klincksieck,

54
Collinder, Bjoern. 1974, IndnUratischeoder gar
Nostratisch? Vierziq Jahre auf rauhen Pfajen. PP. 36375 tn
Ant!quitates Indoqerman1cae: fuer Hermann Guentert,
ed, by Manfred Mayrhofer et al, lnnsbruck:
Sprachwissenschaftliches Institut der Universitaet,
Connolly, Leo A, 1977, IndoEuropean 1 > e: An
explanation by the laryngeal theory, BGDSL CT) 99:175205,
33350.
............
--------
laryngeal theorv.
1979. <ee2> and the laryngeal theory. BGDSL
1980, "Grammatiscner wecnsel" and the
lF 85:96123.
Cowgill, warren. 1963. of Jaan Puhvel:
Laryngeal& and the IndoEuropean vero, Lg 39:24870,
-------
1965, Evidence in Greek, Po, 14280 in
evidence for laryngeal&, eo, by werner winter. the Hague:
Mouton.
------ 1979, Anatolian h1con1ugat1on and
IndoEuropean perfect: Installment II. Pp. 2540 in HethitiSCh
und Indogermanisch, ed, bY Erich Neu and Wolfgang keid,
Innsbruck: tnstitut tuer SprachWissenschaft,
55
Cox, Jerry L. 1977. The laryngeal theory its
application to problems in Germanic linqu!stics,
Indiana University diss, 373 pp.
Debrunner, Albert. 1929, Review of J. Kurylowicz:
Schwa indoeuropeen et <h> hittite. IrJb 13:6&67,
------ 1937. Review of E. Benveniste: Origines de
1a formation des noms en IF 55:31418.
-------
1938. Review of J, Kurvlowicz: tudes
1ndoeuropeennes. IF 56:5558,
Deroy, Louts, 1949. La racine indo-europeenne
<H3egw> se nourrtr, SL 3:1831,
Dressler, wolfgang w,, and Alexander Grosu, 1972.
Generative Pnonologie und tndogermanische Lautgesch1chte. IF
77:1972.
Dressler, Wolfgang w. 1980. was erwarten Phonolog1e
Theorien von der kann die Indogerman1stik
bieten? Pp, 102-19 in Lautgescntchte und Etymologie, ea. bY
M, Mayrhofer et Wiesbaden: Harrassowitz.
Edgerton, Franklin. 1934. law IE
weakgrade vocalism. Lg 10;23565.
-------- ----------------------------- -
56
-------
1943, The IndoEuropean semivowels. LQ
19:83124.
Eichner, Heiner. 1973. Die Etymologte von heth.
<mehur>. MSS 31:53107,
FistaK, JaceK, ed. 1978. Recent developments in
historical Phonology. The Hague: Mouton.
Frisk, Hjalmar. 196072. Etymologtsches
Woerterbuch. Heidelberg: winter.
GamKrelidze, Thomas v. 198.. The IndoF.uropean
glottal1c theory. A new paradigm in the Indo-European
comparative llngulstics, unpublished paper.
GamKrelidze, Thomas v., and v. Ivanov. 1973.
Sprachtypolog!e und die Rekonstruction der
verschluesse, vorlaeutiger Bericht, Phonetica 27:15056,
Hamp, Erie P, 1982. Mycenaean <daae> 'they
contributed(?)# IF 86:190 [1981 Issue),
Hart, Gillian R, 1980, Some observations on
plenewriting in Hittite, DSOAS 43:117,
Haudricourt, Andre G. 1975. Les mutations
consonant1Ques Cocclusives) en indoeuropeen,, PP. 26772 in:
. ---
57
Melanges lingu1st1ques offerts a Benveniste, Louvain:
Peeters.
Hiersche, Rolf. 1964, zur rrage der
Tenues Aspiratae 1m tndogermanischen, Harrassowttz.
-------
<*too(u)> Masc, Du,
1980, Nocnmals allnora. tnau Ntr. Pl. < idq,
zvs 94:20201.
Hirt, Hermann. 1900. Oer indoqerman!sche Ablaut.
Strassburg: Truebner.
--------
192137. Indogermanische Grammatik, I
VII, Heidelberg: winter.
--------
1939, Die Hauptprobleme 1er indoqermaniscnen
Spracnw!ssenschaft. Ed, by H. Arntz. Niemeyer,
Jean-Marie, J,J. Ohala and w.G. Ewan, 1979.
Phonetic explanations for tne development of tones, lJg 55:3758,
Hopper, Paul J, 1973, Glottatized and murmured
occlusives in IndoEuropean. Glossa 7:14166,
------- 1981, "uecem" "Taihun" An
IndoEuropean isogloss, Pp, 13342 in Bono homini donum: Essays
in historical linquistics in memory of v. Alexander Kerns, ed, bY
Yoel L, Arbeitmann and R, Amsterdam: Benjamins
------------------.-------------- --
i
i
58
lllicSvityc, v.M. 1959. un some reflexes of the
IndoEuropean "laryngeals" in ProtoSlavonlc. Voprosy
Jazykoznanija 1959, 2.318.
-------
1971. Opyt sravnentja nostraticeskich
jazvkov, Moskva,
Izut, Hisanosuke. 1981, Skr <Ca)bhuut>, Gr
<Ce)phuultl>, The Bulletin of the International Institute tor
Linguistic Sciences, Kyoto Sangyo University. !1,2:1&784.
Joki, Aults J, 1973. Uralier una Indogermanen.
Helsinki: SuomalaisUgrilatnen Seura.
Joseph, Lionel Saint, 1980, Problems in the
development of the Indo-European 1aryn9eals in Celtic, Diss.
Harvard,
Jucquois, Guy, 1972, Aspects du consonant!sme
hittite. Pp. 59128 in Hethitica. 1, Travaux edites par idem.
Louvain: Bibliotheque de l'Untvers!te.
Kammenhuber, Annelies. 1980. Zum 1ndogerman1scnen
Erbe im Hethith1schen. zvs 94;3344.
Ketler, Allan R, 1970, A phonoloq1cal study ot tne
IndoEuropean laryngeals, The Hague: Mouton.
59
Kortlandt, Freder1k. 1978a. ProtoIndoEuropean
obstruents. If 83110718,
------ 1979o, on the of Slavic
accentuation, zvs 92:26981,
--------
t978c, Indo-European palatovelars before
resonants in BaltoSlavic, PP. 23843 in Recent develoments in
historical Phonology, ed. bY Jacek Fisiak, The Hague: Mouton.
------- 1979, Three problems of BaltoSlavic
phonology, Zbornik za filologiju u 22:57&2.
-------
1980. Albanian anj Armenian. zvs 94:24351.
------.....
1981, Glottalic consonants in Sindhi and
ProtolndoEuropean, IIJ 23:1519.
--------
-------
Slavic accentuation,
-------
1st sg, middle <H2>. IF 8&:12336,
1983, Linguistic theory, universals, and
Folia Linguistica Historica 4:2143.
t984u. Notes on Armenian historical
phonology IV, To appear in: Stuaia caucasica 6.
------
19 , ProtoIn1ouropean glottalic stoPs:
The comparative evidence, To be PUblished,
---------------------- ---
60
Kuhn, Hans, 1978, oas letzte Indoqerman1sch, AAWLM
1978.4, Wiesbaden: steiner.
Kuiper, Franciscus B, J, 1937, Die
Nasalpraesentta, Ein Versucn zu einer morphologischen Analyse,
Amsterdam: North Holland,
-------
1978, Zarathustra's language. MKNAL 41.4,
Amsterdam.
Kurylowtcz, Jerzy. 1927. Schwa et <h>
h1tt1te, Pp, 95104 in Symbolae in Ioannis
Rozwadowski, vol I. cracow,
-------
5162 in Etrennes de Linqutstique, offertes par quelques amis a
Emile Benveniste. Paris: Geuthner.
-------
1935, Etudes indoeuropeennes, Cracow.
-------
1967, The Germanic "Verscnaertunq", Lg
43:44551.
1975. Probleme der indoqermantscnen
Lautlehre. Pp, 14351 in Esquisses linguist!ques II,
Fink.
Lamprecht, Arnost. 1976, Oer 1ndoeuropae1sche
-
61
Konsonantismus 1m Licht der nostratlschen rneorie, Sborn1k
Brnenske Univerzlty A24:172&,
Lehmann, P. 1952, ProtoIndoEuropean
phonology, University of Texas Press.
-----.. -
1957, A syntactic reflex of the
IndoEuroPean laryngeals, Pp. 14547 in studies presented to
Joshua Whatmougn, ed, by Ernst Pulgram, The Hague: Mouton.
------- 1963, Some Phonological observations oased
on the Germanic consonant shift. Monatshefte 55:22935.
Liberman, Anatoly. 1982, Germanic accentoloqy, Vol,
I, The Scandinavian Languages, Minneapolis: University of
Minnesota Press,
Lindeman, Fredrik o. 1964. Les origines indo
europeennes de la "Verschaertunq" germanique. Oslo:
Un1versttetsforlaqet,
------- 1970, Eintuehrung in die Laryngaltheorte.
(Sammlung Goeschen 1247/47a,) Berlin: de Gruyter,
1982, The triple representation of schwa in
Greek and some related of Indo-European phonology,
Oslo: Universitetsforlaqet,
62
-.. -----
1983. Zu altheth. aas-suu. Die Sprache
Lubotsky, A. 1981. Gr : SKt <pajra-> and
loss of laryngeals before mediae In lndolranlan, MSS 40:11338,
Mahtow, Georg Heinrich, ta79, Die langen vocale <aa
ee oo> In den europaeischen Spracnen. Diss. Berlin:
Hermann.
Martinet, Andre, 1955, Economle des changements
phonettque, (Tratte de phonoloQie diacnronlque,) aern: rrancke.
Meld, Wolfgang, 1975, der raeumlichen und
zeitlichen Gllederung des Indogermaniscnen, Pp. 204-- tn
Flexion und Wortblldung, Akten der v. Fachtaqung ct. Idq. Gr.
Regenby 1973, Wiesbaden: Reichert.
Metllet, Antoine, 1937. Introduction a
comparative des lanques indoeuropeennes. Paris: Hacnette.
Merlingen, 1983. Larvngattheorie und Laryngale,
Ablauttneorien und Ablaut. Wien: (orivatelv ouolishedl.
Moeller, H, 1880. Zur Declination germanisch caa ee
oo> in den F.ndungen des Nomens und die ntstenung des <o(a2l>,
BGOSL 7:482547,
---- -------------------------------- --------------- ---- -----------------
63
Neu, Erich. 1976, Zur des
indogermaniscnen verbalsystems. Pp, 23854 in Studies in Greek,
Italic and Indo-European Linguistics. to Leonard
R. Palmer, ed, by Anna Davies and Wolfgang Meia.
Innsbruck: Institut tuer SprachWissenscnatt.
Palmaltis, Mykolas L, 1979[801. ProtoIndoEuropean
vocalism and the development ot the IndoEurooean declensional
models. IF 84:1748,
-------
1982. The new look of tnaoEuropean
declension. IF 86:7195.
Pedersen, Holger. 1951, Die gemeinindoeuropaeischen
und die vorlndeuropaeischen Verscnlusslaute. Copennaqen.
Peeters, Christian. 1971. phonetic definition of
IE <bh dh Qh>. ZVS 85:14.
Peters, Martin. 1980, Untersuchungen zur vertretung
der indogermanischen Laryngale 1m Griecnischen. Vienna: Austrian
Academy of Sciences.
Pokorny, Julius. 1959. Indogermanisches
etymologisches woerterbuch I. Bern: Francke.
Polome, Edqar. 1965, The laryngeal theory so far. A
critical bibliographical survey, Po. 979 in Evidence for
------------------- ----------- ---
. -
64
Laryngeal&, ed, by Werner Winter. The Hague: Mouton.
Prokosch, Eduard, 1939, A comparative Germanic
grammar. Philadelphia: University of Press.
Puhvel, Jaan, 1960, and the Indo-EuroPean
verb. Berkeley: University of California press.
Puthusser1, Ramachandran. 1973, Language of Middle
Malayalam, Trivandrum: Dravidian Linguistics Association of
India,
Rasmussen, I,E, 1974, Haeretica Indogermanica.
Copenhagen,
Ravnaes, Erling, 1981, The development of
schwa/Interconsonantal laryngeal in Iranian, IIJ 23:24773,
Rix, Helmut, 1970, Anlautender Laryngal vor Liquida
oder Nasalis sonans tm Grieehischen. MSS 27:79110,
Sapir, Edward, 1938, Glottalized continuants in
Navaho, Nootka, and Kwakiutl. Lg 14:2474,
Saussure, rerdinand de. 1979. sur le systeme
prtmitif des voyelles dans les langues !ndoeuropeennes,
LeiPZig, Pp, 1268 of Recueil des publications scientlfiques de
Ferdinand de saussure. Geneva: Payot
---------- ------------ -- -- -----------------
-----
65
Schmidt, Gernot. 1973, O!e 1ran1scnen woerter tuer
'Tochter' und 'Vater' und die Retlexe des tnterKonsonanttscnen
(schwa) <HH> in den 1dg, Sprachen, ZVS e7:3ti83,
Schmidt, Johannes, 1889. Die Pluralblldungen der
1ndogerman1scnen Neutra, weimar: soenlau.
Schmidt, Karl Horst, 198la, rypusrelevanter
sprachwandelFlektierend zu Agglutin!erend und seine
Korrelationen, Etudes 15:335-46,
------
1983b, KauKas1sche TypoloQ!e als H!lfsm1ttel
tuer die Rekonstruktion des Vorindoqermaniscnen. Innsbruck:
Institut fuer Spr.achwlssenschaft det Universitaet,
Schmitt, Ruediger, 1967, Oichtunq und Oichter In
indogermanischer zeit. Wiesbaden: Harrassowitz,
Schrijnen, Josef, 1921, Einfuehrunq in das Studium
der indogermanischen Spracnwissenschaft, Trans, by walther
Fischer, Heidelberg: Winter,
Schulze, Wilhelm, 1966 (1885), Indogermaniscne
<aai>Wurzeln. Pp, 4956 tn his Kleine Schriften, ed, by
Wissmann, Goetttngen: VandenhoecK Ruprecht. Also zvs
27(1885):42029,)
Schwyzer, Eduard, 193953, Griecn!sche
66
Muenchen: Becl<,
Seebold, Elmar, 1972. Oas System der lndogermaniscnen
HalbVol<ale. Heidelberg: Winter,
Specht, Franz, 1944/47, oer ursprung der
indogerman!schen Deklination. Goettingen: Vandenhoeck &
Ruprecht,
Steensland, Lars, 1973, Die Distribution der
urtndogermanischen soQenannten Gutturle, Uppsala:
Streitberg, Wilhelm, completed by Victor Michels and
M.H, Jellinel<. 1936. Germanlscn. Berlin: de Gruyter.
Sturtevant, Edgar H. 1942, rne rndoHittite

.. -
laryngeals. Baltimore: Linguistic Society of
-------
1951, A comparative grammar of the Hittite
language. New Haven: Yale University Press,
szemerenyi, Oswald, 1967, Tne new look of
IndoEuropean: reconstruction and tyPology, Phonet!ca 17:6599,
-------
1970, 2nd ed. 1980, Einfuehrunq in die
Vergleichende Sprachwissenschatt, Oarmstadt: W1ssenschattl1Che
Buchgesellschaft.
Tischler, Johann, 1980, Hetnitlsch <h> und die
,---
67
Rekonstruktion des 1ndoqermantschen PhonPminventars. Po. 495522
in Lautqeschichte una Etymologte, ed. ov M et al.
Wtesbaden: Reichert.
Vennemann, Theo. 1984. und
Niedergermanisch, BGDSL 106:145,
Voyles, Joseph B, 1981. Gothic, Germanic and Northwest
Germanic. CZOL Heft 39.) Wiesbaden: Steiner.
wackernagel, Jacob. 1896 (1957), Altindische
Grammatik:. [2nd ed. with snpplements by A. Oebrunner, 1957,1
Goettinqen: Vandenhoeck & Ruprecht.
Watkins, Calvert. 1975, Die Vertretung der LarynQale
in gewtssen morPhologischen Kategorien in den indogermantschen
SPrachen Anatoliens. Pp. 35878 in Flexion und wortbildung.
wtesbaden: Reichert.
Winter, werner, ed, 1965. Evidence for larynqeals.
The Hague: Mouton.
-------
1978. The distribution of short and long
vowels in stems of the type Lith. <est!> : <vest!> : <mesti> and
ocs <jasti> : <vest!> : <mest1> in Baltic and Slavic languages.
PP, 43146, followed by comment of rrederlk Kortlandt,
p, 447, in Recent developments in historical phonology, ed. by
--------------------- -
68
.----- Jacel<. f'isiak. The Hague: Mouton
.,,--..

Vous aimerez peut-être aussi