The Saussure Effect in Lithuanianyy/hp/documents/the_saussure_effect.pdf · The Saussure Effect in...

The Saussure Effect in Lithuanianlowast

Yoko YAMAZAKI dagger

Kyoto University

Abstract

The so-called ldquoSaussure effectrdquo describes the phenomenon of laryngealloss in the following enviromentsHRo- and-oRHC- The fact that theeffect is observed in several Indo-European branches suggests that it tookplace at the stage of Proto-Indo-European In this paper characteristic ex-amples of the phenomenon in Lithuanian are presented and an attempt ismade to provide a consistent morphological explanation of the reason whyno traces of the Saussure effect are observed in several seeming counterex-amples where its operation is expected

1 Introduction

The ldquoSaussure effectrdquo is a phenomenon where a laryngeal is lost without anytrace in specific environments in the vicinity of the vowel o Although notmany examples are found it is observed in Greek Anatolian Italic Sanskritand Armenian This fact suggests that the Saussure effect took place at the stageof Proto-Indo-European Recent studies on the Saussure effect include a surveyof the examples of this phenomenon in Latin and Italic in Nussbaum (1997) aninvestigation and a summary of the phenomenon in Greek in Beekes (1969 74ndash75 238ndash242) as well as in Anatolian in Melchert (1994 49ndash51)1 For Balto-Slavic the main study of this phenomenon is by Rasmussen (1989) who providesexamples of the phenomenon from Balto-Slavic in order to explain the unexpectedcircumflex tones of some forms Unfortunately however the evidence for theSaussure effect in Baltic does not seem to have received the attention it deserves

lowast This paper is a revised edition of my masterrsquos thesis submitted in 2007 I wish to thank themembers of my thesis committee mdash Kazuhiko Yoshida Yukinori Takubo Yutaka Yoshida andSatoko Shirai mdash as well as the students of the linguistics department of Kyoto University Specialthanks are due to my supervisor Kazuhiko Yoshida and Brent Vine for their insightful commentsand advise and to Aurelijus Vijunas for inspiring me to examine this interesting topic I also wishto thank Toshikazu Inoue for providing me with valuable advice and materials on Baltic KanehiroNishimura for his invaluable help and Alexander Hadjiivanov and Adam Catt for checking myEnglish Needless to say any remaining mistakes are exclusively mine

dagger anaconitelingbunkyoto-uacjp1 The phenomenon was originally described in Saussure (1905 511ff)

1

In this paper I will examine certain examples and counterexamples of theSaussure effect in Lithuanian thus refining the results obtained by Rasmussen(1989) Since as yet there are no good collections of Proto-Indo-European nomi-nal roots the data presented here in regard to the Saussure effect are largely basedon the collection of Proto-Indo-European verbal roots inLIV Furthermore I willattempt to clarify the relationship between the Saussure effect and each of theseeming counterexamples Most of these counterexamples are assumed to havebeen affected by legitimate analogical processes which restored the lost laryn-geals

11 The Saussure Effect

The Saussure effect constitutes the loss of a laryngeal in the environments indi-cated in (1) below

(1) a HRo- gt Ro-

b -oRHC- gt -oRC-

The examples in (2) and (3) which are examined in Beekes (1969 74ndash75238ndash242) show that laryngeals in Greek were not vocalized in the environmentsindicated in (1)2

(2) Failure of vocalization of laryngeals in the environment of Ro-

a h3loig-o- gt loigigravec lsquodecimationrsquo cf h3lig - gt aeliglETHg- lsquofew smallrsquo

b h3moildquogh-o- gt moiqigravec lsquoadultererrsquo cf h3mei

ldquogh- gt aeligmeETHqw lsquoto urinatersquo

c h2nor- gt nwreOslash acircnergeOslash lsquoto be activersquo (Hesychius) cf h2ner-gt n rlsquomanrsquo

d h2uldquo

ors- gt oIcircregravew lsquourinatersquo oTHORNron lsquourinersquo cf h2uldquo

ers- gt ersan egravershn lsquodewrsquo

(3) Failure of vocalization of laryngeals in the environment of-oR C-

a torh1-mo-gt tigravermoc lsquoholersquo cf terh1- gt tegraveretron lsquogimletrsquo

b porh2-neh2 gt pigravernh lsquoprostitutersquo cf perh2- gt pegravernhmi pegraverasa lsquotosellrsquo

2 (2) and (3) are based on the summaries in Nussbaum (1997 181ndash182)

2

c tolh2-meh2 gt tigravelma lsquohardinessrsquo cf tl˚

h2- gt tlumlnai lsquoto endurersquo

d solh2-uldquo

o- gt iacuteloc oTHORNloc lsquowholersquo cf sl˚

h2-i- gt CLuv salhˇ

itti-lsquogrowthrsquo

e (h1)ouldquo

Hdh- gt oTHORNjar lsquoudderrsquo cf (h1)uHdh- gt Skt acuteudhar lsquoudderrsquo

f ( polh1uldquo

- rarr) poluldquo

ildquoa-gt poll lsquomanyrsquo

g thematic optatives -o-ildquoh1(-t) gt -oi3 eggoi lsquo(s)he would leadrsquoOgravedoi

lsquo(s)he would seersquo etc

The examples from Latin in (4) below were examined in Nussbaum (1997)

(4) a solh2no- gt solno- gt sollo- lsquowhole allrsquo cf sl˚

h2-uldquo

o- gt salauldquo

o- gtsalu

ldquoo- gt salvusor sl

˚h2-eu

ldquoo- gt salau

ldquoo- gt salu

ldquoo- gt salvuslsquosafe and

soundrsquo4

b tolHuldquo

- gt toluldquo

- gt -toll- rarr meditull ium lsquothe interior inland part of acountryrsquo cf t l

˚H-tosgt tl a-tosgt Lat latus lsquobroad widersquotol(H)u- gt

OPru talus lsquofloorrsquo tol(H)u- or t elHu- gt Skt tacutealu lsquopalatersquo (lt lsquoroofof the mouthrsquo)

c kolH-ni- gt kolni- gt collis lsquohillrsquo cf kelH-tei gt Lith kelti Latvceltlsquoto liftrsquo without the laryngeal loss the expected Latin form would becolnis or culnislt kolani- lt kolH-ni-

3 For a detailed discussion of the examples from thematic optatives see Nussbaum (199718213)

4 According to Meiser (1998 108ndash109) Proto-Indo European syllabic resonants developed inLatin as follows PIECR

˚HC gt Lat CRaC eg gn

˚h1-to- gt (g)natus lsquobornrsquo PIE CR

˚HC gt

Lat CaRaC egp l˚

h2-meh2- gt palamagt palma lsquopalmrsquo On the other hand Schrijver (1991193ndash197) argues that it is unnecessary to set up a phonological ruleCR

˚HC gt CaRaCand shows

that most examples of the alleged changeCaRaClt CR˚

HC can be explained in different waysCRHeC is proposed by him as one of the possible sources of LatinCaRaCsequencespalmalt pl

˚h2-em- calamitas lsquocalamityrsquo lt k l

˚H-em- Therefore the former change fromsl

˚h2-u

ldquoo- to

Latsalvuscan be regarded as possible if we accept the developmentCR˚

HCgt CaRaCmentionedin Meiser (1998 108) At the same time the latter change fromsl

˚h2-eu

ldquoo- to Lat salvus is

possible As a resultCR˚

HC gt CaRaCis rejected and an alternative explanation is providedby Schrijver (1991 193ndash197) However since it is essentially irrelevant whether the preform ofLat salvusis sl

˚h2-u

ldquoo- or sl

˚h2-eu

ldquoo- this problem will not be discussed further here

3

d kwolHso-5 gt kwolso- gt collus collum lsquoneckrsquo6 cf Ved tuvi-kurmı(n)- lsquomoving stronglyrsquolt kwl

˚H-mi- Gkpl mnh lsquonave of a wheelrsquo

lt kwl˚

h1- or kwl˚

h2-

Melchert (1994 49ndash51) gives examples from Hittite in (5) as evidence of thelaryngeal loss

(5) a h2worso-gt worso-gt warsa-lsquomist steamrsquo cf h2wers-lsquoto rainrsquo gtGk egraversh lsquodewrsquo

b h2worg- lsquoturn twistrsquo gt (reduplicated)worg- rarr wawarkima-lsquodoor -hingersquo cf h2urg-gt h

ˇurki lsquowheelrsquo

c kolh2mo-ro-gt kalmara-lsquoray beamrsquo cf kl˚

h2mo-lsquostalkrsquo gt Gkkalmh kolh2mo- lsquostalkrsquo gt Latv salms lsquostubblersquo

d polh2-uldquo

eh2- rarr palwa(i)- lsquoto claprsquo cf Lat palma lsquopalmrsquo lt p l˚

h2-meh2- or pl

˚h2-em-7 without the laryngeal loss Hittpalh

ˇwa(i)- would

5 While the reconstruction ofkwolHso- (Nussbaum 1997 196) is phonologically sufficientit raises a morphological question concerning the suffix-so- kwolHso- seems to have a mor-phological boundary betweenkwolH- and-so- since a root of the shape ofkwolH- and a suffix-so- can be recognized on the basis of Gkπέλομαι lsquoto turn (into) to be(come)rsquo and Sktcaratilsquoto travel to wanderrsquo (lt kwelH-eo-) and so on

However the reconstruction of the suffix-so- is problematic at least for an early stage ofProto-Indo-European since there is no clear evidence for it Brugmann (1906 538ff) showedthat the suffix-so- is an extension of-s- or -es-by -o- analogous to-no- as the extension of-en- and-on- by -o- etc Furthermore even though aso-formation shares the same suffix andthe same root the difference in the ablaut of the root usually indicates that a shift of olds-stems too-stems occurred in the daughter languages eg Sktruks

˙a- lsquoshiningrsquo with zero grade and OIcel

li osslsquolightrsquo with e-grade suggest an olds-stem There are a fair number of examples of this kindAlso Chantraine (1933 433ndash436) classifies Greek words with-σος into three groups (i) loanwords (ii) expressive adjectives and action nouns each of which can be explained as a root witha desiderative suffix and (iii) hypocoristics As a result he avoids the reconstruction of the suffixfor Proto-Indo-European

Although it is clear that further research regarding this issue is needed it is beyond the scopeof the present paper The examples of the Saussure effect with the suffix-so- as identified byprevious studies are cited below with the exception of the forms in-so- in my own examples

6 However Nussbaum (1997 196) mentions that the possibility cannot be excluded thatcol-luscollumreflects the following changekwolHso-gt kwolaso-gt kwolso-gt kwollo- gt col-luscollum One of the anonymous referees suggested that the following place-name and its cor-responding adjective could be counterexamples to the above-mentioned sound change since theexpected assimilation of sonants following syncope does not seem to have occurred in themFa-lerii lt Falrior lt Falesioi (adj) Falernuslt Falrinoslt Falesinos The investigation of thesetwo forms will be reserved for a future time

7 See footnote 4 for the two competing possible preforms

4

be expected

In addition to the examples listed in (2) ndash (5) above there are some examplesfrom other languages which are provided in Nussbaum (1997 183) such as theones in (6) below

(6) Examples from other languages

a solh2-uldquo

o- gt Sktsarva-lsquowhole allrsquo cf (3d)

b kouldquo

H-lo- gt Arm soyl lsquocavityrsquo cf kuH-gt Av sura-lsquoholersquo Sktsacuteuna-lsquolack absencersquo

As shown above examples of the Saussure effect are observed in several Indo-European daughter languages In the next section I present possible traces of theSaussure effect in Lithuanian

12 Expected Traces of the Saussure Effect in Lithuanian

Although laryngeals do not have segmental reflexes in Lithuanian it is well knownthat they have reflexes as tonal accents8 as in (7) and (8) below

(7) a VRHC gt VRC [acute intonation (falling tone)]

b VRC gt VRC [circumflex intonation (rising tone)]

(8) a genh1-to- gt zentaslsquoson-in-lawrsquo bherHg- gt berzaslsquobirchrsquo

b gombhosgt zambaslsquosharp edgersquouldquo

l˚

kwos gt vilkas lsquowolfrsquo

Accordingly the intonation of-oRHC- sequences which have also under-gone the Saussure effect is expected to be circumflex in Lithuanian-oRHC- gt-oRC- [via the Saussure effect]gt -aRC- (7b) Based on this Rasmussen (1989181ndash184) gives the forms which seem to have undergone the Saussure effect9

Some of these examples which I consider plausible are given in (9)10

8 Lithuanian long vowels diphthongs and mixed diphthongs (vowel+ sonant between conso-nants) have two kinds of intonations when they are accented

a acute intonation (V) falling tone

b circumflex intonation (V) rising tone

9 The purpose of the analysis in Rasmussen (1989) is to utilize the Saussure effect to accountfor certain forms which exhibit unexpected circumflex tones but which cannot have undergonemetatonie doucesince they do not appear in the well-known environments necessary for its occur-rence

10 However since (9d) and (9e) are affected by the problem already mentioned in footnote 5they are less plausible in comparison to the other three

5

(9) Examples from Lithuanian

a (PIEkorH-neh2 gt) PB-Skarna gt karna (4)11 lsquobastrsquo cf k erH-on-sgt kerH-onrarr kr

˚H-on gt Lat caro lsquomeatrsquo

b (PIEsphorH-teh2 gt) PB-Ssparta gt sparta (4) lsquospeedrsquo cf Vedmacutea

apa spharıs˙

(R˙V 66114) lsquodo not spurn [us]rsquolt sp(h)erH- lsquoto kickrsquo

c (PIE kouldquo

H-no-gt) PB-Scaunasgt saunassaunus (4) lsquobraversquo12

d (PIEnorH-seh2 -so-gt) PB-Snarsa -sasgt narsa narsas(4) lsquocouragersquo cf OCSpo-nretż lsquoburiedrsquo Lith nerti lsquoto diversquo lt nerH- lsquoto diversquo

e (PIEbholH-so-gt) PB-Sbalsasgt balsas(4) lsquovoicersquo cf Lith bıltilsquoto begin speakingrsquolt bhl

˚H-tei OE bellan lsquoto ringrsquo lt bhelH-e- lsquoto

soundrsquo

2 Problems with the Lithuanian Examples of the Saussure Ef-fect

Followingsect12 the examples of the Saussure effect in Lithuanian should be formswhich have a circumflex accent on their respective root syllables and whose pre-

11 Lithuanian nominals are classified into four ldquoaccentual paradigms (AP)rdquo according to thepattern of the stress alternation between the stem and the ending throughout the nominal paradigmRegarding nouns with a monosyllabic stem the tone of the stem is connected to the AP to whichthe noun belongs AP1 denotes a barytone paradigm with the acute accent on the root whileAP2 is a historically barytone paradigm where the root is non-acute and the stress alternationbetween the stem and the ending is caused by Saussurersquos law and other morphological factorsAP3 comprises mobile paradigms with the acute root whereas AP4 denotes a mobile paradigmwith the non-acute root where further stress alternation has been brought about by Saussurersquos lawand other morphological factors

12 While Rasmussen (1989 182) considerssaunasto be related tosauti lsquoto shootrsquo which is adescendant of the Proto-Indo-European root keu

ldquoH- lsquoto throw push to moversquo (LIV 330) it could

also be related to Sktsacuteuna- lsquolackrsquo susvur (perfect form of the rootsavi- lsquoto swellrsquo Mayrhofer1986ndash96 II 623ndash624) and Gkκυέω lsquoto bear in the wombrsquo as suggested inIEW (592ndash594) TheGreek form suggests the existence ofh1 in the root for lsquoswellrsquo since it can be interpreted as areflection of a causative formkuh1-ei

ldquoe- However from those zero grade forms two kinds of

full grade forms namely keuldquo

h1- (following Mayrhofer 1986ndash1996 II 624) and kuldquo

eh1- (afterLIV 339) can be reconstructed It is difficult to judge which of the two forms represents the oldershape of the root and this problem remains open

In addition following the analysis of Mayrhofer the root on which the protoform ofsaunasis based is aset

˙root in the case of both keu

ldquoH- lsquoto throwrsquo and keu

ldquoh1- (or ku

ldquoeh1-) lsquoto swellrsquo

Thereforesaunasis a fine example of the Saussure effect

6

forms are considered to have contained-oRHC- sequences However these twocriteria are not sufficient for the proper identification of the effect since there areadditional factors which need to be taken into account Therefore I believe thatsome forms presented in Rasmussen (1989 181ndash184) and cited below should beexcluded from the list of probable examples of the Saussure effect for reasonswhich I will discuss in the following sections

21 Uncertain Reconstruction of Laryngeals

For some roots it is difficult to decide whether a root-final laryngeal can be safelyreconstructed Some researchers consider the following forms to have been de-rived from set

˙roots while others consider them to have been derived fromanit

˙roots

211 KOR(H)MEH2 gt SARMA (4) lsquoFROSTrsquo

Rasmussen (1989 181) regards Lithsarma as one of the cognates of OE OSaxhrımlsquofrostrsquo and reconstructs PIE kerH-i

ldquo- He considers that the Germanic forms re-

flect the zero-grade formkriH- and that their long vowels provide evidence forthe existence of a laryngeal However OE OSaxhrım is related tokrei- lsquototouch something contactrsquo together with Lithkrena (4) lsquocreamrsquo by IEW (618)which is modified asKrei

ldquoH- lsquoto touchrsquo in LIV (368) Yet two problems arise

from these reconstructions The first concerns the question of whether the initialconsonant of the target root was a plain velar or a palatovelar since these two con-sonants have different outcomes in Lithuanian namelys (lt k ) andk (lt k )Another problem is which shape of the root is the original k

( )

reildquoH- or k

( )

reH(ildquo)-

for only zero-grade reflexes are shown outside of the Balto-Slavic cognates mak-ing it difficult to determine the original shape of the root

Furthermore there are uncertainties regarding Lithsarmasand its relatedforms Derksen (1996 88ndash89) reconstructs a Proto-Baltic acute root for this formon the ground of the accential correspondences between Lithuanian [sarmas(13)lsquohoarfrost lyersquosarma (3) lsquoidrsquo sarma(1) lsquohoarfrostrsquo] and Latvian forms [sarmslsquolyersquo sarma lsquohoarfrostrsquo serma lsquoidrsquo] He regards the accentuation ofsarmas(4)andsarma (4) as the result of the analogical spread of AP4 He also discusses itscognates with a zero-grade (sırmas(3) lsquogrey dapple-greyrsquo Latvsirmaslsquogreyrsquo)which exhibit the original acute For these reasons he thinks the acute accent inthe cognates with a-no- suffix (sernas(3) lsquowild boarrsquo) and with-v- (sırvas (3)lsquogrey dapple greyrsquo) show the original acute accent whileserknaslsquohoarfrostrsquo andLatv sernis lsquohoarfrostrsquo do not show any trace of the old acute In contrast to thisIllich-Svitych (1978 119) reconstucts ananit

˙form kernom lsquohoar-frostrsquo since

some of the Slavic cognates such as Ruseren lsquocrust over snowrsquo and Ukrseren

7

lsquofrozen hard snowrsquo point to a non-acute root while he relates Lithsarma(1) to aset

˙root ( kerH-) in PIE Derksen states that if Illich-Svitych is right there seem

to be two almost identical roots ( ker- and kerH-) in Balto-SlavicThus if we postulate that all the forms quoted above are derived from the same

root in PIE it is extremely difficult to determine whether the original root in PIEwas ananit

˙or aset

˙root since some cognates (sarmas(13) Latvsarms sirmas

Latv sirms etc) suggest an original acute or aset˙

root while others (serksnasLatc sersnis Ruseren Ukr seren Slk srien) imply a non-acute or an anit

˙root

For this reason the problems surrounding the root ker(H)- should be reservedfor a later research

212 M OIldquo(H)NEH2 -OSgt MAINAS DIAL MAIN A (4) lsquoEXCHANGErsquo

Rasmussen (1989 199) reconstructs aset˙

rootmeildquoH- on the grounds of Latvmıt

lsquoto exchangersquo and Vedmayatelsquoto exchangersquo The Latvian correspondences toLith maınasaremaına maına andmaınalsquochangersquo which exhibit acute accentua-tion13 However the cognates of Lithmaınasinclude Vedapa-mıtya- (lt -mi- )lsquochargesrsquo which suggests ananit

˙root mei

ldquo- in Proto-Indo-European (Mayrhofer

1986ndash1996 II 314ndash315) SCrmijena lsquochange (of the moon)rsquo also points toan anit

˙root Consequently although the intonation of Latvian forms certainly

presents a problem it is nonetheless clear that the root itself is of theanit˙

type

213 Uldquo

OR(H)TA rarr VARTAI (2) lsquoGATErsquo

An anit˙

root uldquo

er- lsquoto ward offrsquo is reconstructed inLIV (684) while Rasmussen(1989 183) reconstructs aset

˙root u

ldquoerH- beside theanit

˙form (u

ldquoer-mn

˚) Ras-

mussen (1989 82) assumes two more structures for the root namelyuldquo

erHu-m0

anduldquo

ruH-m0 where the former producedecircruma lsquocoverrsquo whereas the latter gaverise toucircUumlma lsquocoverrsquo

The cognates of this root include Gkecircrumai lt uldquo

er-u- lsquoto check (from do-ing something)rsquo (LIV 684) Vedvr

˚ta- lsquoenclosedrsquolt u

ldquor

˚-to- vartar- lsquoinhibitorrsquo lt

ver-tor- -vr˙t- (eg arn

˙o-vr

˙t- lsquoobstructing the floodrsquo) andvaru0 (eg varutar- lsquode-

fenderrsquo varutha- lsquoumbrella protectionrsquo) (Mayrhofer 1986ndash1996 II 512ndash513)

13 Buga (19231924 272) thinks that Lithmaınaswould have been formed besidemainaonthe analogy ofvarnas varna on which the denominativemainyti (maıno [3p]) was formedThus he regards the acute intonation as original for the Baltic words meaning lsquoexchangersquo Onthe other hand Derksen (1996 266) proposes an alternative idea that themetatonie rudein a-stem agent nouns in Baltic was analogical to the derivation of nomina agentis from iterative (orintensive or durative) verbs with the suffix- i

ldquoa- Later this type of derivation became productive

and a-stems with fixed stress on an acute root were derived from verbal roots even if there wereno such verbs Althoughmaına maına andmaına lsquochangersquo are not agent but abstract nouns theymight be the result of the derivation mentioned above

8

Mayrhofer (1986ndash1996 II 513) thinks thatvaru0 does not necessarily indicatea set

˙root since Vedvr

˚ta- vartar- and-vr

˙t- suggest ananit

˙root contrary to the

assumption ofuldquo

erH-proposed in Rasmussen (1989)As briefly surveyed there are a number of forms which suggest ananit

˙root

uldquo

er- for PIE For this reason this form will be excluded from my discussion

214 POL(H)Uldquo

OSgt PALVAS (4) lsquoPALErsquo

Rasmussen (1989 184 306) reconstructs an unspecified laryngeal for the rootpelHu

ldquo- lsquopalersquo On the other hand Nussbaum provides a discussion on the root

relevant to Lithpalvas He distinguishes between the two forms (i) PItpallo-(lt pal-u

ldquoo-) identical in formation to Gmcfalwa- lsquopalersquo (OIcel folr OHGfalo)

and with Lith palvas and (ii) PIt pollo- lsquodark grayrsquo which is continued byLat pullus lsquodark grayrsquo These two reconstructed forms could be descendants ofone and the same stem although there are difficulties regarding this assumptionldquoone is thatpal- andpol- are not easily reconciled (ph2eol- being excludedby Gkpeliigravec lsquograyrsquo etc andpolH-p l

˚H- by eg OIrlıath lsquograyrsquo lt pl-ei

ldquo- and

not pl˚

H-eildquo-)rdquo (Nussbaum 1997 19162) the other is thatpallo- denotes a pale

color while pollo- denotes a dark color Considering these points Lithpalvasseems to be semantically and morphologically akin to a group of words whichhave descended frompallo- (lt pal-u

ldquoo-) lsquoa pale colorrsquo rather than frompollo-

lsquoa dark colorrsquo Furthermore following the view of Nussbaum that the sequence-ll- in pallo- is more likely to reflect-l u

ldquo- than a syncopated-l ˘Vu

ldquo- pallo- is

not likely to show the vocalized root-final laryngeal and the shape of its ancestralroot would beph2eol- without a root-final laryngeal However there remainsa complicated etymological problem namely whether there is a proper way toreconstruct one PIE root reconcilingpal- andpol- which Lith palvas shouldbe based on as well This is whypalvas is omitted from the following discussion

22 Etymological Problems

The forms listed in the following sections face etymological problems in the sensethat there are competing etymologies one of which allows the reconstruction of aroot-final laryngeal while the other one does not

221 SPARNAS (4) lsquoWINGrsquo

Rasmussen (1989 182) regards Lithsparnasas connected with Lithspırti Latvspertlsquoto kickrsquo which are cognate with Vedsphurati lsquoto push to kick awayrsquo [lt sphr

˚H-

e- (LIV 585)] However at the same time there is another possibility namely(s)per-lsquoto traversersquo [spor-no-gt sparnas] which is related to Sktparn

˙a- lsquowingrsquo

9

OHGfarn and OEfearnlsquofernrsquo in IEW (850)14

Both of the possible root etymologies work well with Lithsparnas while theone based onper- is preferred in respect to Sktparn

˙a since the root with as-

piration iesp(h)erH- (gt sphari- lsquoto kick awayrsquo) and the other without ieper- (gt par- lsquoto bring through passrsquo) are clearly distinguished in Sanskrit Soif sp(h)erH- is adopted as the root etymology ofsparnas it should not be de-rived from (s)per- which allegedly gave rise to the rest of the forms namelySkt parn

˙a- OHG farn lsquofernrsquo etc However it might be difficult to suppose that

sparnasis not etymologically related to Sktparn˙a- OHG farn lsquofernrsquo etc since

both their phonological shapes and their meanings are closely related As furtherdiscussion on the etymology ofsparnasis beyond the scope of this paper this casewill be left aside

222 TARNAS (2 gt 4) lsquoSERVANTrsquo

On the one hand Rasmussen (1989 183) reconstructsterHuldquo

- lsquoyoung delicatersquoas the root on which the protoform of Lithtarnasis based He further notes thatterHu

ldquo- is not necessarily identical toterh1u

ldquo- lsquoto wear awayrsquo but does not explic-

itly show any evidence for a root-final laryngeal On the other hand according toIEW (1070)tarnasis derived from an adjectivalanit

˙root ter- lsquodelicate weakrsquo

It is not certain that a morphological procedure for building ano-formation onthe basis of an adjectival root existed at an early stage of Proto-Indo-EuropeanConsequently this form will be excluded from the source list for the discussionbelow

23 A Form Derived from an S-stem

The form in question istom(H)seh2 gt tamsa (4) lsquodarknessrsquo which is derivedfrom Proto-Indo-EuropeantemH- lsquoto become darkrsquo We also have an adjectivetamsus lsquodarkrsquo in Lithuanian According toIEW (1063) there is a Vedic cognate ofthis word namelytamas-lsquodarknessrsquo Mayrhofer (1986ndash96 I 626) reconstructsan s-stem paradigmt emH-estemH-s- for the protoform of Vedtamas-andLat temerelsquoat random by chancersquo (lt lsquoin darknessrsquo= Ved tamas-i[locsg])which clearly points to ane-grade in the protolanguage as well as OHGdemarlsquodimrsquo ( lt temHso-) Thus these materials suggest ans-stem paradigm contain-

14 According toIEW (850) the origin of the initials- in the Lithuanian form could be the verbalroot sp(h)erH- I further note that both Illich-Svitych (1963 [1979] 38) and Southern (199936) mention Lithuaniansparnasas a variant withs-mobile although they set up(s)pern-for theProto-Indo-European root Since ldquo(s)+ plain voiceless stopsrdquo is the ldquohome territoryrdquo ofs-mobileas described in Southern (1999 20ndash21) the view of Illich-Svitych (1963 [1979] 38) and Southern(1999 36) is not implausible

10

ing ane-grade Furthermore except for a few examples (Gkicircqoc lsquoa chariotrsquoLat ponduslsquoa weightrsquo foeduslsquofoul uglyrsquo OCS kolo lsquoa chariotrsquo) s-stems usu-ally do not have ano-grade in their roots (Schindler 1975 265) Therefore itdoes not seem likely that theo-grade in the Lithuanian forms is primary

At the same time there are examples including several ones outside Balto-Slavic ofu-stem adjectives built ons-stems with ano-grade which is completelyparallel totamsus (lt tomH-s-u-) such as GkaeligxOcircc lsquosharprsquo whose protoformcan be reconstructed ash2ok-s-u- This suggests that there might have alreadyexisted a morphological process wheres-stems changed thee-grade of the rootvowel to ano-grade in the protolanguage However since the age of this reforma-tion is not certain this example should also be left aside

24 Forms which Underwent Metatony

241 METATONY

Metatony refers to the replacement of one intonation with another in Baltic lan-guages for example in the case ofu-stem adjectivessaldus (3) [Dauksarsquos Pos-til e (1599)15 ] (cf Latv salds) rarr saldus (4) lsquosweetrsquo which took place in therelatively recent history of Lithuanian (Stang 1966 160) Also some groups ofwords containing certain morphological suffixes frequently underwent metatonyfor examplepuodas(1) lsquopotrsquo rarr puodzius (2) lsquopotteryrsquo [nominal suffix with-ıildquou-] sveıkas(4) lsquohealthyrsquorarr sveikinti lsquoto greetrsquo [verbal suffix-inti ] As shown

metatony obfuscates the original toneMetatony is classified into two classes in accordance with the resulting intona-

tion Metatonie ruderefers to a change from circumflex to acute intonation andmetatonie doucerefers to a change from acute to circumflex intonation Com-pared tometatonie rude metatonie douceis much more frequent especially informs with suffixes containing- ıi

ldquo- (egu-stems in-ius i

ldquoo-stems etc)16 Thus

it is clear that forms with suffixes containing- ıildquo- such asbarnis (2)lsquoquarrelrsquo

(PIE bherH- lsquoto work on something with a sharp toolrsquo)kraıtis (2 4) lsquodowryrsquo

15 The accentuation of the words in Dauksa is adopted from Skardzius (1935)16 According to Stang (1966 145ndash147)- ıi

ldquous(gt -ius)- ıi

ldquoos(gt -is [i

ldquoo-stem]) or- ıi

ldquoa [or - ıi

ldquoe]

(gt ˜e) shifted the ictus on-ı- to the immediately preceding syllable at an early stage of Proto-BalticIf the syllable which received the ictus had a long vowel or a diphthong it received a circumflexintonation Derksen (1996 375) concluded that this retraction of stress took place at the EastBaltic stage This view of accent retraction which brought aboutmetatonie douce seems to havebeen accepted as a standard ldquothe sequence- ıi

ldquo- in medial stressed position lost its ictus to the

preceding syllable This caused the syllable to change an original acute tone into a circumflexie the retraction causedmetatonie douceon a preceding syllable This rule has been accepted asa major source of Baltic metatonyrdquo (Larsson 2004 162)

11

(PIE kwreildquoh2- lsquoto exchangersquo)kalnius (2) lsquomountaineerrsquo (PIEkelH- lsquoto raisersquo)

etc should be excluded from the list of proper examples However the situationis not so simple with nominals of other stems Sincemetatonie doucein otherforms such aso-stems anda-stems is not regular it is necessary to examinewhether a circumflex intonation in a particular stem is original or the result ofmetatonie douce

242 DISTINGUISHING BETWEENORIGINAL AND SECONDARY INTONATION

Endzelıns (1899 263ff) observed a correspondence between Lithuanian accen-tual paradigms and Latvian intonation patterns in word-initial syllables The cor-respondence is as follows

Lith acute (AP1)sim Latv sustained tone (V)eg Lithvyraslsquomanrsquo sim Latv virs lsquoidrsquo Lith duonalsquobreadrsquosim Latv duonalsquoidrsquo

Lith circumflex (AP2)sim Latv falling tone (V)eg Lith ranka (ranka [sgacc]) lsquoarmrsquosim Latv ruoka lsquoidrsquo Lith iesmaslsquospitrsquo sim Latv ıesmslsquoidrsquo

Lith acute (AP3)sim Latv broken tone (V)eg Lithnuogaslsquonakedrsquosim Latvnuogslsquoidrsquo Lith sirdıs lsquoheartrsquosim Lattvsirdslsquoidrsquo

Lith circumflex (AP4)sim Latv falling tone (V)eg Lithbalsaslsquovoicersquo lsquoidrsquo sim Latvbalss Lith draugaslsquofriendrsquo sim Latvdraugslsquoidrsquo

Also Illich-Svitych (1963 [1979] 52ndash53) presents the following argumentregarding the chronological relationship between Lithuanian and Latvian accen-tuations

The process of accent retraction to the initial syllable in Latvian took placeat an early date it is a proto-Latvian process () the tone contrast of ˜ andˆ in initial syllables in Latvian reflects an older distribution of nominals byaccent class than does modern Lithuanian ()

Thus Lithieva lsquobird-cherryrsquo (lt PIE oiuldquo

a or eiuldquo

a) occurs in AP1 2 and 4while Latv ievalsquoidrsquo preserves the original accentuation and Lithjega lsquothoughtmind witrsquo (lt PIE i

ldquoegwa) occurs only in the secondary accentuation AP4 while

Latv j ega exhibits the original accentuation (Illich-Svitych 1963 [1979] 53)

12

The studies by Endzelıns (1899) and Illich-Svitych (1963 [1979]) show theimportance of the intonation of Latvian forms which correspond to Lithuanianones since it preserves the older accentual patterns Consequently the followingexamples in (10) should be excluded from the possible examples but includedin the counterexamples since the accentuation of the Latvian correspondencespoints to the original acute

(10) a (PIE kolH-neh2- gt) PB-S calna gt salna (4) lsquofrostrsquo PIE kelH- lsquotobe coldrsquo

cf Latv salna

b (PIEkolH-uldquo

eh2- gt) PB-Skalva gt kalva (4) lsquohillrsquo PIE kelH- lsquoto beelevatedrsquo

cf Latv kalva

(10a) and (10b) will be discussed insect4

3 Possible Examples of the Saussure Effect

The problems with some of the examples of the Saussure effect provided in Ras-mussen (1989) were briefly discussed in the previous section The following threepoints have emerged from this discussion

bull Make sure the reconstruction of a root-final laryngeal and the etymology ofeach root is reliable and clear

bull Make sure the originalo-grade can be reconstructed for Proto-Indo-European(for example ano-grade of a reflex derived from an olds-stem should beexcluded since the age of its formation is not certain as already discussed insect23)

bull Make sure the intonation of the form in question is the original one

Taking these points into consideration I present some possible examples inthis section in addition to those listed in (9) after Rasmussen (1989) Most of themhave environments which allow for the shape of the form to be derived on the basisof either the Saussure effect or another rule for laryngeal loss with equal possibil-ity As we will see in the following sections the ruleCHCC gt CCC discussedin Hackstein (2002) could share the same environment as that which triggered theSaussure effect if the sequence occurred in the environmentoRHCC (where thecontext of the Saussure effect-oRHC- is followed by another consonant) Also

13

Pinaultrsquos law [H gt empty in internal syllables before yod (Pinault 1982 268ndash269)]is a rule that could share the environment with the Saussure effect when it oper-ated on the sequence-oRHi

ldquo- (where the consonant following the laryngeal is a

yod in the sequence of the Saussure effect)For this reason the possible examples listed below fall short of being cer-

tain The intonation of the Latvian forms is based on Mmacrulenbachs (1923ndash32) andEndzelıns (1934ndash46) whereas the accentual paradigms of the Lithuanian formsare based onLKZ

31 gaudus (4) lsquosonorousrsquo

gaudus (4) lsquosonorousrsquo geuldquo

h2- lsquoto callrsquo (cf LIV 189)rarr gouldquo

h2-dhh1- gtgou

ldquo-dhh1- gt gaudus

The cognates of Lithgaudus include Gkgow lsquoI wailrsquo and OHG kuma The longvowel inkuma(lt guh2-mo-) indicates the existence of a root-final laryngeal andgow (lt gou

ldquoh2-ei

ldquoe-) is suggestive of a root-finalh2 The accentuation of the

corresponding Latvian formgaudslsquomiserablersquo shows that the circumflex intona-tion of Lith gaudus is the original intonation rather than a secondary intonationderived bymetatonie douce Here it is necessary to assume that the laryngeal ingou

ldquoh2-d(h)- disappeared at the Proto-East-Baltic stage at the latest

However there still remains the problem with theo-grade ingaudus sinceu-stem adjectives are known to be of the proterokinetic type (Pinault 2003 162ff)whereo-grade roots do not appear In order to explain thiso-grade ingaudus averbal form related togaudus namely Lithgaudziu gausti lsquoto resoundrsquo can betaken into consideration The present paradigm of the verb can be interpreted asbeing built on the stemgou

ldquoh2-dhh1-i

ldquoo- which is extended by two suffixes ie-

dh(e)h1-17 and- ildquoeo- just as in Vedyu-dh-ya-telsquo(he) fightsrsquo and Lithsker-d-ziu

lsquoI slaughterrsquo (Brugmann 1892 1103) Since-dh(e)h1- can be added to a root ofany grade as in the case of Latverbum(ltu

ldquoerh1-dhh1-o-) lsquowordrsquo Lith vardas

(lt uldquo

orh1-dhh1-o-) lsquonamersquo and Gothwaurd (lt uldquo

r˚

h1-dhh1-o-) lsquowordrsquo the o-grade ofgou

ldquoh2-dhh1- is morphologically compatible Also as the suffix-

ildquoeo- was added directly to nominal stems to derive denominative verbs as inGk aelignomaETHnw (lt h1nh3mn

˚-ildquoeo- lsquoto namersquo18 ) gou

ldquoh2-dhh1- can be interpreted

17 Following Brugmann (1892 1045ff 1906 467) the-d(h)- element could be a suffix-dh(e)h1- which originated from a verbal rootdheh1- lsquoto put dorsquo

18 Although the Proto-Indo-European paradigm of lsquonamersquo has been disputed here I follow thesuggestion presented in Kim (2002 29) that Gkὄνομα is derived from the generalized weakstem of a proterokinetic paradigmh1neh3m-n

˚h1nh3m-en remade from the original acrostatic

paradigmh1nacuteeh3m-n˚

h1neh3m-n- beacause of phonetic resemblance tonomina actionisin acute-mn

˚--men-

14

as a nominal stem In Lithuanianu-stem adjectives are productive (Skardzius1943 55ff) and can be secondarily formed on the basis ofo-stem nouns forexamplegodus lsquocovetousrsquo godaslsquocovetousnessrsquo draugus lsquofriendlyrsquo draugaslsquofriendrsquo This suggests that the adjective might have been formed on the basis ofa nominal stemgou

ldquoh2-dhh1-o- although it is not attested as a noun

Consequentlygouldquo

h2-dhh1- the protoform on whichgaudus is based can beinterpreted as satisfying the condition of both environments of the Saussure effect(oRHC gt oRC) andCHCC gt CCC Accordingly there are two possiblecauses for the laryngeal loss Since it is difficult to set up a relative chronology forthese two rules the possibility cannot be excluded thatgou

ldquoh2-dhh1- might have

undergone the ruleCHCC gt CCC For this reason it should be regarded as apossible example

32 garb˜e (4) lsquohonorrsquo

garbe (4) lsquohonorrsquo gwerH- lsquoto express onersquos approvalrsquo (cfLIV 210ndash211)rarr gworH-bh(h2)-19 gt gwor-bh(h2)- gt garbe

The root-final laryngeal is reconstructed on the basis of Vedgurta- lsquoblessedrsquoLat gratus lsquoidrsquo lt gwr

˚H-to- Although there is no information regarding the ac-

cent in Latvian there is accentual information in Dauksarsquos ldquoPostilerdquo where mostof the instances appear with AP4 Following this Derksen (1996 208) recon-structs another formgwer-bh- for garbe (4) in addition togwerH- (gt Lith gırtiLatv dzirt lsquoto praisersquo and Vedgurtı- lsquopraisersquo) However assuming the Saussureeffect the reconstruction of a separate root is not necessary In fact if the Saus-sure effect is not considered then it is necessary to propose two almost identicalroots namelygwerH-andgwer- with the same meaning

On the other hand the etymology of the suffix-bho--eh2- should be takeninto account here in order to examine the plausibility ofgarbe as an example of theSaussure effect The suffix-bho- was first identified with a PIE rootbheh2- lsquotoshinersquo in Brugmann (1906 386) and has been widely accepted since then Thissuffix occurs in a considerable number of abstract nouns and adjectives withoutany connection to colors or animals Hyllested (2008) considers the original func-tion of this suffix to be to form verbal nouns Furthermore he expresses doubtover the traditional identification of the suffix with the rootbheh2- lsquoto shinersquo

19 According to Brugmann (1906 386)-bho- possibly belongs to Sktbhati lsquoto shinersquo where-upon Vedr

˚sa-bha- lsquobullrsquo which is now reconstructed asbheh2- (LIV 68ndash9) on the basis of

Gkϕάντα λάμποντα (Hesychius) lsquoaglowrsquo etc Later Bessemberger (1973 13110) suggested thatthis suffix belongs to a PIE rootbhu- (bhu

ldquoeh2- lsquoto bersquo) However Hyllstead (2008) presents a

counterargument which is discussed in the following part in this section

15

inasmuch as the phonetic similarity between the suffix and the root is limited toa single very frequent consonant and the modifying effect of-bho- should lie inthe modifying nature of derivation itself What follows is that the suffix-bho-might not necessarily be connected to the PIE rootbheh2-

However if the possibility of the suffix-bho- going back to the rootbheh2-cannot completely be denied the laryngeal loss could be caused by the soundchangeCHCC gt CCC making this form short of being certain

33 tarpas (1) lsquoholersquo

tarpas(1) lsquoholersquo terh1- lsquoto bore drill rubrsquo (cf LIV 632ndash633)rarr torh1-p-gt tor-p- gt tarpasrarr tarpas

Evidence for the root-final laryngeal includes the Greek formtegraveretron lsquogim-letrsquo lt terh1-tro- Latvt`arpslsquoworm (as a creature boring a hole)rsquo and dial Lithtarpas(42) lsquoholersquo suggest that the acute intonation intarpasmust be secondary whereoriginally it must have had a circumflex intonation Therefore it is possible thatthis form is the result of the Saussure effect

34 kraujas (4) lsquobloodrsquo

kraujas(4) lsquobloodrsquo kreuldquo

h2- lsquoclot bloody fleshrsquorarr krouldquo

h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas

An important piece of evidence for the specified root-final laryngeal is pro-vided by Gkkregraveac lsquofleshrsquo lt kreu

ldquoh2-s- According to Fraenkel (1962ndash65 I

290) the Latvian word corresponding to Lithkraujas is kraujs lsquobloodrsquo How-ever it disappeared for two reasons20 One point is that this form constituted ahomonym with Latvkraujs lsquosteep steep edge of shorersquo As a resultkraujs forlsquobloodrsquo might have been replaced with another item The second reason is thatthe ldquoother itemrdquo which replacedkraujs actually existed This item is Latvasinswhich is a cognate of Sktasr

˚j- Hitt esh

ˇar and Gkordfar ecircar (Hesychius) lsquobloodrsquo

asinswould have accelerated the shift fromkraujs to asinsfor lsquobloodrsquo There-fore the accentuation of Latvkraujs could provide circumstantial evidence whichsuggests that the circumflex accent of the Lithuanian form is original

The intonation of the Lithuanian form reflecting ananit˙

root is explainablein terms of both the Saussure effect and Pinaultrsquos law sincekrou

ldquoh2-i

ldquoo- meets

the structural description of both of these rules Interestingly Vedkravya- (lt

20 A more detailed discussion can be found in Endzelıns (1921)Filologu bierdrıbas raksti How-ever since I was unable to obtain a copy of this work the discussion here is based on the citationsin Fraenkel (1962ndash65 I 290)

16

kreuldquo

ildquoo- lt kreu

ldquoh2-i

ldquoo-) which is ane-grade cognate of Lithkraujas does not

show a reflex of the laryngeal either which is best explained by Pinaultrsquos lawNevertheless it might be interesting to mentionkraujas since the environmentmeets both of the two rules regarding laryngeal loss

35 kardas (4) lsquoechorsquo

kardas (4) lsquoechorsquo (cf LIV 353) kerH- lsquoto admirersquo rarr korH-d(h)- gtkor-d(h)- gt kardas

The root-final laryngeal is reconstructed on the basis of Vedkırti- lsquofamersquo ltk r

˚H-ti- ONhroDr lsquofamersquo lt kroH-tro- Unfortunately there is no corresponding

form in Latvian which is why this example falls short of being certain Howeversince there is no evidence formetatonie doucein Lith kardas the circumflexintonation of this form could be the result of the Saussure effect

4 Potential Counterexamples and Their Analysis21

In this section I present some seeming counterexamples of the Saussure effectand explain why these examples do not show any reflexes of the Saussure effectin environments where it is expected to occur There is no clear distinction inphonological environment between the examples which show the Saussure effectand the ones which do not This has led me to think that it is more feasible to seeka morphological explanation than to set up a new criterion for the Saussure effectsince many of the examples which do not show the Saussure effect are derivedfrom to- or no-formations with ano-grade as shown below

I also discuss the fate of the original forms and their paradigms in Proto-Indo-European from which the forms insect4 descended There are two main possibilitiesfor the original type of the paradigms of the counterexamples one of which is thethematic type and the other is the athematic type In the following sections bothcases will be discussed in this order

41 Forms Derived from Thematic Stems

First I will provide an explanation for each thematic counterexample Interest-ingly cognate forms which escaped the operation of the Saussure effect are foundfor each of them The cognate form is either a verb or less often a noun Thediscussions below include those cognate forms Since there is an etymologicalproblem in general with sets of words with different vowel grades which share

21 I am indebted to Kazuhiko Yoshida and Brent Vine with whom I had insightful discussionsregarding the problems treated in this section

17

the same root as in the case ofsuldquo

ep-no-suldquo

op-no-sup-no-lsquosleeprsquo uldquo

es-no-u

ldquoos-no-(u

ldquoes-no-22 ) lsquopurchasersquo andu

ldquoeg-no-u

ldquoog-no-ug-no- lsquowagonrsquo the

explanations given here are provisional

411 KALTAS (1) lsquoCHISELrsquo KALTI lsquo TO FORGErsquo

kaltas (1) lsquochiselrsquo kelh2- lsquoto beat hitrsquo (cf LIV 350)rarr kolh2-to- gtkaltas

Possible support for the specified root-final laryngeal includes the Greek formkla- lsquobreakrsquo lt k l

˚h2- There is a Latvian form corresponding to Lithkaltas

namelykaltas lsquochiselrsquo The sustained tone in the Latvian form suggests that theacute accent inkaltas(1) must be the original tone No trace of the Saussure effectis found in this form

Apparentlykaltas is a to-formation with ano-grade built on a verbal rootkelh2-23 Therefore it is helpful to look into the verbal paradigm derived fromthe root kelh2- Jasanoff (2003) has proposed a new type of present verbalparadigm in Proto-Indo-European the so-called ldquomolo-typerdquo which allegedly hadtheo-grade in strong forms and thee-grade in weak forms as well as perfect-likeendings Verbs in daughter languages which have present forms either with persis-tente-grade or with persistento-grade belong to this type for examplemolh2-melh2- lsquoto grindrsquo [Hitt mall(a)i- Lith malti Gothmalanlt molh2- in contrastto OIr melid OCSmeljo lt melh2- (Jasanoff 2003 64) ] Furthermore Jasanoff(2003 76) argues that the verbs which can be ascribed to themolo-presents showlsquoevidence for ano-grade present in more than one language but no unambiguousreflex of thee-grade weak stemrsquo Following this assumptionkelh2- is includedin this type on the basis of Lithkalu kalti and OCSkoljo klati The followingparadigm can be reconstructed forkelh2- cf Jasanoff (2003 71ff)

22 Vedvasna-m lsquopurchase pricersquo n lsquofee payrsquo could be from thee-grade or theo-grade23 Sinceto-formations with ano-grade for example Gkνόστος lsquoreturnrsquo χόρτος lsquoenclosurersquo

Lat hortus lsquogardenrsquo are known as abstract nouns or at least requiring this function as nouns(Brugmann 1906 420 Risch 1937 22 Schwyzer 1953 I 501) the instrumental meaning ofkaltascalls for a semantic explanation

A semantic parallel to the case ofkaltas is found in Balto-Slavic that isdolb-to- lsquochiselpointed ironrsquo(lt dholbh-to- larr PIE dhelbh- lsquoto burrow digrsquo) The reflexes of this form have in-strumental meaning as in OPrussdalptanlsquoa pointed tool to burrow with rsquo Bulgdlato lsquochiselrsquo(Derksen 2008 112) Also Gkκοῖτος lsquobed couch a going to bed sleeprsquo (cfκεῖμαι) could beregarded as a semantic parallel to the case discussed here for one of its meanings lsquobedrsquo is sug-gestive of the development of the original meaning as an abstract noun (lsquosleeprsquo) into a secondarymeaning lsquobedrsquo Therefore the semantic change from an abstract to an instrumental meaning canbe presupposed also for the meaning ofkaltas

18

1sg k olh2-h2e 1pl k elh2-meH ()24

2 kolh2-th2e 2 kelh2-(H)e ()3 kolh2-e 3 kelh2-r

˚(s)

In the 1st person singulark olh2-h2e is phonologically in one of the twoenvironments of the Saussure effect Consequently the root-final laryngeal mighthave been lost due to the Saussure effect Likewise 2sgk olh2-th2e might havelost the root-final laryngeal as a result of the Saussure effect25 Since the 3rdperson singular form and the plural forms are not in the environment of any rulesfor laryngeal loss they would not have lost their root-final laryngeals theoretically

strong forms (1 2sg)kolh2-C- gt kol-C-(3sg)kolh2-egt kolh2-e [vacuous operation]

weak forms kelh2-C- gt kelh2-C- [vacuous operation]

Thus in this case the paradigm would have comprised both stems one with aroot-final laryngeal (k olh2-k elh2-) and the other without a root-final laryngeal(k ol-) Then a theoretically possible morphological interaction between themmight have occurred producing the following results

phonological results (i)kol- (ii) kelh2- (iii) kolh2-(e)analogical results (iv)kolh2-(C-) (v) kel-

It is not inconceivable that one of the above forms was subsequently gener-alized in each of the daughter languages Thus Lithuanian generalized the stemkolh2- askalti on the basis of whichkolh2-to- would have been formed AsRusskolotž lsquoto break crushrsquo also points to a Proto-Slavic acute accent the gen-eralization ofkolh2- must have occurred at an early stage which involved bothBaltic and Slavic This is whykaltasdoes not seem to show any trace of the Saus-sure effect Thereforekaltas which can receive a morphological account is nota counterexample of the phenomenon

412 SALTAS (3) AND SALNA (4) SALTI lsquo TO BE COLDrsquo

salna (4) lsquochill of early morning hoarfrostrsquo kelH- lsquoto become coldrsquo (cfLIV 323)rarr kolH-n- gt salna (3)rarr salna (4)

24 These question marks are given by Jasanoff (2003 32) due to the fact that the situation withthe endings in the plural is less clear

25 Although it might seem conceivable that the laryngeal was lost due to another sound lawCHCC gt CCC the accent in this case falls on the preceding vowel Since this sound lawoperated when the accent was on the following vowel (Hackstein 2002 2) its operation in thisenvironment would have been impossible

19

saltas(3) lsquocoldrsquo kelH- lsquoto become coldrsquorarr kolH-to- gt saltas

These two forms are cognates and therefore they are treated together here Alaryngeal is reconstructed on the ground of the accents of the Baltic cognates forexample Latvsalna lsquofrostrsquo salt lsquoto be frozenrsquo Lithsalti lsquoto be coldrsquo etc andSkt sısira- lsquoearly spring cold season of the yearrsquolt ki-kl

˚H-o- The broken tone

of Latvsalna corresponds to Lithuanian AP 3 This suggests that the accentuationof Lith salna (4) is not primary but secondary and that this form has undergonemetatonie douce thereby not showing the expected trace of the Saussure effectLith saltas is a cognate ofsalna above and its corresponding form in Latvianis salts lsquocoldrsquo The broken tone of this Latvian cognate suggests that the accen-tuation of saltas (3) is not the result ofmetatonie rudebut is rather the originalaccent No trace of the Saussure effect is found here either

As shown abovesaltasandsalna do not show any trace of the Saussure effecteven though their preform had ano-grade In order to see how these forms arerelated to the Saussure effect in their prehistory it would be informative to lookinto the problem surrounding the root kelH- which is mentioned in Derksen(1996 84) The problem concerns the fact that some reflexes of kelH- meanlsquowarmrsquo such as Latcalere lsquoto be warmrsquo (lt kl

˚H-eh1-) OIcel hlaeligr OHG lao

lsquotepidrsquo (lt kleh1-uldquo

o-) etc while others mean lsquocoldrsquo as in the case of LithsaltasFor this reason Schrijver (1991 206ndash207) suggests that there was ananit

˙variant

of kelH- whereasLIV (323) reconstructs kelH- lsquoto become coldrsquo and kel- lsquotobecome warmrsquo The case can be summarized as follows26

26 Although Lith sılti lsquoto become warmrsquo and Latvsilt lsquoidrsquo certainly belong here these twoverbs are excluded from the following table Since both verbs occur with asta-present whichoften induces the acute intonation (metatonie rude see Buga 19231924 257 Derksen 199684) a speculation regarding their original intonations can be made on the basis of the intonationsof their respective adjectives (Lithsiltas and Latvsıltas) although by themselves they do notconstitute evidence for the existence of a root-final laryngeal

20

reflexes ofset˙

root anit˙

rootLith salti lsquoto be saltaslsquocoldrsquo salna lsquohoar- siltas lsquowarmrsquo

coldrsquo frostrsquoLatv salt lsquoidrsquo salts lsquoidrsquo salna lsquoidrsquo sılts lsquoidrsquoSkt sısira-

lsquoearly springrsquoLat calere lsquoto be

warmrsquoWel clyd ()27 lsquoidrsquo

(lt kl˚

-to-)Gmc OIcelhlaeligr

OHGlao lsquotepidrsquo(lt kleh1u

ldquoo-)

The table above shows a contrast between Baltic forms ino-grade for lsquocoldrsquoand forms in zero-grade for lsquowarmrsquo On the other hand the fact that the formsfor lsquowarmrsquo (Lat calere OIcelhlaeligr OHGlao Lith siltas Latv sılts Wel clyd)comprise bothanit

˙andset

˙forms suggests that at an early stage of Proto-Indo-

European there was a confusion betweenanit˙

andset˙

forms A possible reasonfor this might be that the verbal paradigm based on kelH- had both theo-gradeand thee-grade In this case the forms with theo-grade would have undergonethe Saussure effect while the forms with thee-grade would have retained thelaryngeal This in turn would have producedanit

˙andset

˙variants within a sin-

gle paradigm This speculation may allow one to assume amolo-type presentparadigm for kelH- with oe ablaut If the verb kelH- belonged to themolo-type the laryngeal of theo-grade forms once lost by means of the Saussure ef-fect would have been restored under the morphological influence of thee-gradeset

˙variant kelH- Thus Lithsaltasand the preform ofsalna [ salna (3)] can be

regarded as the descendants of the restoredset˙

root kolH-The view shown above suggests that the root for lsquowarmrsquo originally may have

been aset˙

root kelH- which is identical with the root for lsquocoldrsquo What followsthis is that kelH- may have been a common root for lsquocoldrsquo and lsquohotrsquo and it musthave undergone a semantic change from lsquocoldrsquo to lsquowarmrsquo in some languages forexample Latcalere and OIcelhlaeligr or from lsquowarmrsquo to lsquocoldrsquo in Baltic languagesAlthough this kind of semantic change in opposite directions is not very common

27 However a homonymic root kel- lsquoto bury hide coverrsquo draws attention which is recon-structed for Proto-Indo-European on the basis of OEhelanlsquoto buryrsquo (LIV 322) OIrceilid -ceillsquoto hidersquo (Schumacher 2004 394) Taking the homonym into consideration Welclyd can bederived fromkl

˚-to- lsquocovered (space)gt warmrsquo hence the question mark in this table

21

an instance can nevertheless be pointed out While some reflexes of the rootmeg-h2- mean lsquogreatrsquo (Mayrhofer 1986ndash1996 II 338) as in Gkmegravega- Ved mahi-the Baltic reflexes mean lsquosmallrsquo [for further details about their etymology seeFraenkel (1962ndash65 I 422ndash423) and Mmacrulenbachs (1923ndash32 II 574)] just as inthe case of Lithmazasand Latvmazs28

This still raises semantic problems with regard to whether the original mean-ing of kelH- was lsquocoldrsquo or lsquowarmrsquo and of how the original meaning changedinto the opposite one Some cognates which mean seasons or which are linkedto the change of seasons can provide clues for solving this problem Some ofthe descendants of kelH- have meanings linked to seasons vizsısira- lsquoearlyspringrsquo and OIcelhlanalsquoto thawrsquo [lt PGmcxl ewanojananlarr xl ewaz (Orel2003 176)] OIcelhlanacan be linked to seasons for lsquothawrsquo is an event takingplace in early spring when the weather gradually becomes milder Thus boththe Vedic and the Old Icelandic words indicate the season of the year when theweather becomes milder although it is still rather cold These items could leadone to establish the original meaning lsquoto be(come) mildrsquo for the root kelH- Thisoriginal meaning might have contributed to the development of the meaning fromlsquoto become mildrsquo into lsquowarmrsquo in Lithsiltas Latv sılts Lat calere OIcel hlaeligrand OHGlao while it developed from lsquoto be(come) mildrsquo into lsquocoldrsquo via lsquonotwarm enoughrsquo in the rest of the forms This is my present understanding of thissemantic problem although further discussion is necessary

413 KALNAS (3) lsquoMOUNTAIN rsquo KELTI lsquo TO RAISErsquo

kalnas(3) lsquomountainrsquo kelH- lsquoto soarrsquo (cf LIV 349) rarr kolH-no- gtkalnas

A root-final laryngeal is reconstructed since the intonations of some descen-dants of the root namely Lithkelti and Latvcelt lsquoto liftrsquo suggest its existenceSince no decisive evidence points to a specific laryngeal the coloring of the la-ryngeal in this case remains unknown29 The Latvian formkalns lsquomountainrsquo

28 One of the anonymous referees suggested that the opposite meaning of the Baltic forms canbe explained if it is supposed that the rootmeg(-h2)- meant lsquoto growrsquo (cf OIrmogaid lsquotogrowrsquo) and that the Baltic forms underwent a semantic change from lsquostill in growthrsquo to lsquonot bigenough smallrsquo while the cognates in other branches underwent a semantic change from lsquoto growrsquoto lsquogrown fully bigrsquo The following discussion in this section is also indebted to the insightfulcomments of the referee

29 In this section I provisionally follow the reconstruction of the unspecified laryngeal inLIVwhile I admit that the root-final laryngeal is most likelyh3 as the Greek formκολωνός lsquohillrsquosuggestsLIV (3491) mentions that this is not decisive since vowel assimilation has often takenplace in Greek which indicates that the specification of the laryngeal needs further discussion

22

which corresponds to Lithkalnas has a broken tone on the root vowel whichcorresponds to Lithuanian AP3 Hence the acute accent onkalnas(3) must beprimary and it does not show the result of the Saussure effect

At least two possibilities can be proposed for the morphological backgroundof kalnas First it can be interpreted as a thematizedn-stem Neri (2003 273904)proposed an amphikineticn-stem paradigmkelh3-o(n)kl

˚h3-n-es kl

˚h3-onplusmn i

on the basis of Gkkolcedilnh whereby two steps would be needed to derivekolh3-no- k l

˚h3-n- rarr k l

˚h3-n-o- [via thematization] andk l

˚h3-n-o- rarr kolh3-n-o-

[with change of the grade of the root vowel] Since there are no identical for-mations ofkolh3-n-o- outside the Baltic languages these processes must havetaken place in the inner history of the Baltic branch Therefore it is not likely thatkolh3-n-o- existed in the early stages of Proto-Indo-European when the Saus-sure effect was still operating Thus this form is likely to be irrelevant to thepresent discussion Furthermore thead hocchange of the grade of the root vowelassumed here presents a problem

SecondkolH-no- can also be interpreted as ano formation with ano-gradeas Skardzius (1943 217) notes Assuming this the acute intonation ofkalnascanbe explained by considering the influence of the verbkelti as suggested in Nuss-baum (1997 196) If this view is accepted then the change must have taken placeat an early stage when it was still possible for speakers to combine the verbalparadigm with the nominal paradigm Thus this scenario is possible althoughnot certain since there might be another source of the morphological influence onkolH-no- as will be mentioned below

It is noteworthy that there is a debate regarding the set of wordssuldquo

ep-no-su

ldquoop-no-sup-no-lsquodreamrsquo etc Although little is known about this it is possi-

ble that if kelH-no- or k l˚

H-no-30 existed it might have influencedkolH-no-However since no direct descendants of such forms are found this possibilitylacks convincing support

The discussion above shows that so far the second view (kolH-no-) is morelikely to be true than the first one (kolH-n-o-) However even in the case of thesecond view the prehistory of the form needs to be reexamined since the problemwith the thematic formation still remains as briefly mentioned above

However since it is not essentially relevant which kind of laryngeal could be reconstructed forthis root this problem will not be discussed further here

30 There might have been a descendant ofk l˚

H-no- in Greek Vine (2006 51038) suggests thepossibility that Gkκολώνη lsquohillrsquo as well asκολωνός lsquohillrsquo and κλωμακ- lsquopile of stonesrsquo mighthave originally represented a contamination involving thematic stems in other wordsk l

˚h3-no-

k l˚

h3-mo- andkol[h3]-no- Vine (p c) further notes that it can also be interpreted as a result ofthe morphological interaction betweenκλω-νό- (lt k l

˚h3-no-) andκόλων (lt κέλων lt k elh3-

on-) In both cases there might have existed a descendant ofk l˚

h3-no- in the prehistory of Greek

23

42 Forms Derived from Acrostatic Paradigms

In this section I will discuss a possible scenario in which a seeming counterexam-ple of the Saussure effect could be derived from an athematic paradigm althoughin fact most of the counterexamples mentioned insect41 seem to have been derivedfrom thematic formations Such a scenario can be invoked as a parallel to the casewhich will be discussed in the next section

If a counterexample of the phenomenon can be derived from an athematicparadigm the ablaut type of the paradigm would most likely be acrostatic withoe ablaut since the counterexamples of the Saussure effect naturally have theo-grade In an acrostatic paradigm withoe ablaut the strong forms which hado-grade roots would have undergone the Saussure effect while the weak formswith e-grade would have been exempt from it

strong cases CoRH-C-gt CoR-C-weak cases CeRH-C-gt CeRH-C- [vacuous operation]

As a result the paradigm would have had bothanit˙

andset˙

variants whichwould have subsequently interacted with each other and the theoretically possibleforms would be as follows

phonological (i)CoR-C- (gt Lith CaRC-) (ii) CeRH-C-(gt Lith CeRC-)analogical (iii)CoRH-C- (gt Lith CaRC-) (iv) CeR-C-(gt Lith CeRC-)

Eventually one of these forms must have been generalized in the paradigmIn the case where theo-grade variant with the restored laryngeal is generalizedthen a seeming counterexample is found Note that it is necessary to assume aconsequent thematization when the attested form has a thematic vowel

43 An Analysis ofkalva (4) lsquohillrsquo

kalva (4) lsquohillrsquo kelH- lsquoto soarrsquo (cf LIV 349)rarr kolH-uldquo

- gt kalva (1 (or3))rarr kalva (4)

This form is derived from the the same root as Latcollis in (4c) andkalnasin sect413 above Lithkalva belongs to AP4 and has a circumflex accent on itsroot vowel (as inkalva [accsg]) which seems to indicate that it has undergone alaryngeal loss in the environment-oRHC- through the Saussure effect Howeverthe accentuation of the cognate Latvian formkalva points to an original acuteintonation This suggests that the original accentuation ofkalva must have beenacute and that it has undergonemetatonie doucein the prehistory of Lithuanianrather than the Saussure effect at the Proto-Indo-European stage This possibilitymakes it plausible to regardkalva as a counterexample of the Saussure effect

24

There is a discussion regarding the etymology of the Germanic forms relatedto kelH- in Neri (2003 273ndash275) He reconstructs a neuter acrostatic paradigmkolH-u-kelH-u- for Proto-Indo-European on the basis of Proto-GmcXalluzConsidering this reconstruction Lithkalva can also be related to the acrostaticparadigm since its protoformkolH-u

ldquoeh2 contained a-u

ldquo- which was a con-

ditioned allophone of-u- in certain environments Acrostaticu-stems recon-structed in the protolanguage are well known as a neuter category Thereforethe feminine Lithkalva must have been derived secondarily by adding the fem-inine suffix-eh2- while the neuter gender would have turned into masculine inGothhalluslsquorockrsquo Thus Lithkalva might have been derived from this paradigmvia (thematization and) feminizationkolH-u- (rarr kolH-u

ldquoo-) rarr kolH-u

ldquoeh2 A

morphological parallel can be found indoru-deru-rarr deruldquo

-eh2 (gt Lith dervalsquoresinrsquo) deru

ldquo-o- (gt OCSdrevo SCrdrldquoevo lsquowoodrsquo) h2oi

ldquou-h2ei

ldquou-rarr h2oi

ldquou-

eh2h2eildquou-eh2 (gt OHG ewa lsquoage eternityrsquo)h2oi

ldquou-o-h2ei

ldquou-o- (gt Gothaiws

lsquoidrsquo) A possible factor contributing to the formation of the feminine form can bea functional thematization If the preform underwent a functional thematizationwhich created adjectival derivatives then it certainly must have undergone fem-inization at the same time for adjectives usually have feminine forms as wellThus the feminine form of the adjectival derivative should be the preform ofkalva31 Since the process of functional thematization is rather old and mighthave originated already at the Proto-Indo-European stage the preform ofkalvamight have undergone the Saussure effect as well However the Baltic items pointto an original acute intonation which means that the form prior to the Proto-Balticstage would have had a long vowel in its root as a result of compensatory length-ening following the loss of the laryngeal This discrepancy can be explained if theschema presented in the previous section is taken into account As the originalparadigm ofkolH-u-kelH-u- had botho-grade ande-grade forms there couldhave been a competition between two thematized forms namelykolH-u

ldquoo- and

kelH-uldquo

o-

31 The case of Gkὅλος οὔλος lsquowholersquo [cf (3d) also Latsalvus lsquosafe and soundrsquo (4a)Sktsarva-lsquowholersquo (6a)] might be a morphological parallel since Gkὅλος οὔλος and Sktsarva-can be explained as a functionally thematized neuteru-stem [sol(H)-u

ldquoo- larr solH-u-] and

Lat salvus as well as Latsalus lsquohailrsquo suggests an originalu-stem paradigm [sl˚

H-euldquo

o- or sl˚

H-uldquo

o- larr selH-u-] Schrijver (1991 196) notes that Latsalus is derived from a lost verbal stemsalu

ldquoeo- which was probably based on au-stemsalu- lt sl

˚Hu- Considering this an original

u-stem paradigmsolH-u-selH-u- can be reconstructed for Proto-Indo-European from whichsolH-u

ldquoo- (gt Gk ὅλος οὔλος Skt sarva-) sl

˚H-u- andsl

˚H-u

ldquoo- (gt Lat salus salvus) were

derived For the derivational process ofselH-u-rarr sl˚

H-u- a change of acrostatic ablaut to thatof R (o) R (empty) [for example Ved (doru gt) daru (der-u-rarr dr-eu-sgt) droh

˙lsquowoodrsquo] can be

recalled

25

thematizedo-grade formkolH-uldquo

o- gt kol-uldquo

o-thematizede-grade formkelH-u

ldquoo- gt kelH-u

ldquoo- [vacuous operation]

As mentioned previously a morphological interaction betweenset˙

and anit˙variants might have restored the lost laryngeal in theo-grade form where the

Saussure effect operated regularly This would explain why the Baltic forms pointto an original acute intonation32

5 Conclusion

While the investigation presented in this paper cannot be considered exhaustivesect3 andsect4 show that there are at least a few examples and counterexamples ofthe Saussure effect in Lithuanian However as there are no clear phonologicaldifferences between the environments of the examples and the counterexampleswe cannot set up a new criterion for the operation of the Saussure effect

Nevertheless for most of the counterexamples it is possible to provide mor-phological explanations as to why they do not show any traces of the Saussureeffect in Lithuanian In most cases morphological factors can be assumed to havetriggered the restoration of the lost laryngeal

These explanations suggest the possibility that although originally the Saus-sure effect itself was a regular sound change the traces of its operation becameunrecognizable in many cases due to the various morphological changes that oc-curred between the early stages of Proto-Indo-European and Lithuanian This canbe considered quite natural if the long span of time between the early Proto-Indo-European stage and the period of the earliest Lithuanian attestation (16CAD) istaken into account

32 The suffix-(e)h2- can be regarded as a collective suffix as well This assumption providesa semantic explanation of the original meaning ofkol[H]- u

ldquoeh2 (lsquoa pile of rocksrsquo) which later

came to mean lsquoa hillrsquo As the collective suffix is observed in several branches [cf Hittalpaslsquocloudrsquo alpes lsquocloudsrsquo alpa lsquogroup of cloudsrsquo Gkμηρός lsquothighrsquo μηροί lsquothigh-piecesrsquoμῆραlsquogroup of thigh-piecesrsquo Latlocuslsquoplacersquo pluralloci and collectiveloca] the collective formationis regarded as old Consequently at the stage of Proto-Indo European the paradigm ofkolH-u-kelH-u- could have had variants namelykolH-u

ldquo- (gt kolu

ldquo-) for collective forms (kolH-u

ldquo-eh2

etc) andkelH-u- for oblique forms for only strong cases had collective forms The explanationinvolving the restoration of the laryngeal inkol[H]- u

ldquoeh2 as provided above can also be applied

in this case

26

References

Bammesberger A1973 Abstraktbildungen in den baltischen SprachenGottingen Vanden-

hoekamp RuprechtBeekes R S P

1969 The development of the Proto-Indo-European laryngeals in GreekHague and Paris Mouton

Brugmann K1892 Grundriss der vergleichenden Grammatik der indogermanischen

SprachenVol II2 Strassbrug Karl J Trubner1906 Vergleichende Laut- Stammbildungs- und Flexionslehre nebst Lehre

vom Gebrauch der Wortformen der indogermanischen SprachenVolII1 Strassburg Karl J Trubner

Buga K19231924ldquoDie Metatonie im Litauisch und LettischenrdquoZeitschrift fur vergle-

ichende Sprachforschung51 109ndash142 52 250ndash302Chantraine P

1933 La formation des noms en Grec ancienParis Ancienne HonoreChampion

Derksen R1996 Metatony in BalticAmsterdam and Atlanta Rodopi2008 Etymological dictionary of the Slavic inherited lexiconLeiden and

Boston BrillEndzelıns J

1899 Uber den lettischen SilbenakzentBeitrage zur Kunde der indoger-manischen Sprachen25 259ndash274

Endzelıns Jamp Hausenberga E1934ndash46 Erganzungen und Berichtigungen zu K Mulenbachs lettisch-

deutschem WorterbuchRiga Kulturas fonda izdevumsFraenkel E

1962ndash65 Litauisches etymologisches WorterbuchVol I II Heidelberg CarlWinter

Hackstein O2002 Uridg CHCCgt CCC Historische Sprachforschung115 (1) 1ndash

22Hyllested A

2008 PIE-bh- in Nouns and Verbs Distribution Function Origin InLuhr R (ed)Protolanguage and prehistory Wiesbaden Reichert

IEW = Pokorny J

27

1959 Indogermanisches etymologisches Worterbuch Vol I II Bern andMunchen Francke Verlag

Illich-Svitych V M1963 Imennaja akcentuacija v baltijskom i slavjanskomMoscow Insti-

tute Slavjanovedenija Akadamija Nauka SSSR1963 [1979]Nominal accentuation in Baltic and SlavicCambridge and Mas-

sachusetts and London The MIT press (English translation ofIllich- Svitych (1963) by Richard L Leed and Ronald F Feldstein)

Jasanoff J2003 Hittite and the Indo-European verbNew York and Oxford Oxford

University PressKim R

2002 Topics in the reconstructioin and development of Indo-European ac-centPhD dissertation University of Pennsylvania

Larsson J2004 Length andMetatonie doucein Baltic Derivative Nouns In Clack-

son Jamp Olsen B A (eds)Indo-European word formation159ndash170 Copenhagen Museum Tusculanum Press

LIV = Rix H amp et al (eds)2001 Lexicon der indogermanischen VerbenWiesbaden DR Ludwig Re-

ichert VerlagLKZ = Lietuviu kalbos institutas (ed)

1968ndash87 LietuviukalboszodynasVilnius MintisMokslasMayrhofer M

1986ndash96 Etymologisches Worterbuch des AltindischenVol I II HeidelbergCarl Winter Universitatsverlag

Meiser G1998 Historische Laut- und Formenlehre der lateinischen SpracheDarm-

stadt Wissenschaftliche BuchgesellschaftMelchert H C

1994 Anatolian historical phonologyAmsterdam and Atlanta RodopiM macrulenbachs Kamp Endzelıns J

1923ndash32 Latviesu valodas vardnıca Vol I II III IV R ıga Kulturas fondaisdevums

Neri S2003 I sostantivi in -u del gotico morfologia e preistoriaInnsbruck In-

stitut fur Sprachwissenschaft der Universitat InnsbruckNussbaum A J

28

1997 The ldquoSaussure Effectrdquo in Latin and Italic In Lubotsky A (ed)Sound law and analogy papers in honor of Robert S P Beekes onthe occasion of his 60th birthday 181ndash203 Amsterdam and AtlantaRodopi

Orel V2003 A handbook of Germanic etymologyLeiden and Boston Brill

Pinault G-J1982 A Neglected Phonetic Law the Reduction of the Indo-European La-

ryngeals in Internal Syllables before Yod In Ahlqvist A (ed)Pa-pers from the 5th international conference on historical linguistics265ndash272 Amsterdam John Benjamins

2003 Sur les themes indo-europeens en-u- derivation etetymologie InTichy Eampet al (eds)Indogermanisches Nomen Derivation Flex-ion und AblautBremen Hempen Verlag

Rasmussen J E1989 Studien zur Morphophonemik der indogermanischen Grundsprache

Innsbruck Institute fur Sprachwissenschaft der Universitat Inns-bruck

Risch E1937 Wortbildung der homerischen Sprache2nd edition Berlin de

GruyterSaussure F de

1905 Drsquosup2m lusic aTriptigravelemoc remarquesetymologiques InMelangesNicole 503ndash514 Geneve W Kundig Fils

Schindler J1975 Zum Ablaut der neutralen s-Stamme des Indogermanischen In Rix

H (ed)Flexions und Wortbildung 259ndash267 Wiesbaden Dr Lud-wig Reichert Verlag

Schrijver P1991 The reflexes of the Proto-Indo-European laryngeals in Latin Amster-

dam and Atlanta RodopiSchumacher S

2004 Die Keltischen primarverben Ein vergleichendes etymologischesund morphologisches LexikonInnsbruck Institut fur Sprachen undLiteraturen der Universitat Innsbruck

Schwyzer E1953 Griechische Grammatik Vol I Munich Beck

Sihler A L1995 New comparative grammar of Greek and LatinNew York and Ox-

ford Oxford University PressSkardzius P

29

1935 Dauksos akcentologija Kaunas Humanitariniu˛ MoksluFakulteto lei-dinys

1943 LietuviukalboszodziudarybaVilnius Lietuvos MoksluAkademijaSouthern M R V

1999 Sub-grammatical survival Indo-European s-mobile and its regener-ation in GermanicWashington DC Institute for the Study of Man

Stang C S1966 Vergleichende Grammatik der baltischen Sprachen Oslo Univer-

sitetsforlagedtVine B

2006 Autour de sud-picenien qolofıtur etymologie et poetique In Pin-ault G-Jamp Petit D (eds)La langue poetique indo-europeenne actes du colloque de travail de la Societe desetudes indo-europeennes(Indogermanische GesellschaftSociety for Indo-European Studies)Paris 22-24 octobre 2003Leuven and Paris Peeters

30

In this paper I will examine certain examples and counterexamples of theSaussure effect in Lithuanian thus refining the results obtained by Rasmussen(1989) Since as yet there are no good collections of Proto-Indo-European nomi-nal roots the data presented here in regard to the Saussure effect are largely basedon the collection of Proto-Indo-European verbal roots inLIV Furthermore I willattempt to clarify the relationship between the Saussure effect and each of theseeming counterexamples Most of these counterexamples are assumed to havebeen affected by legitimate analogical processes which restored the lost laryn-geals

11 The Saussure Effect

The Saussure effect constitutes the loss of a laryngeal in the environments indi-cated in (1) below

(1) a HRo- gt Ro-

b -oRHC- gt -oRC-

The examples in (2) and (3) which are examined in Beekes (1969 74ndash75238ndash242) show that laryngeals in Greek were not vocalized in the environmentsindicated in (1)2

(2) Failure of vocalization of laryngeals in the environment of Ro-

a h3loig-o- gt loigigravec lsquodecimationrsquo cf h3lig - gt aeliglETHg- lsquofew smallrsquo

b h3moildquogh-o- gt moiqigravec lsquoadultererrsquo cf h3mei

ldquogh- gt aeligmeETHqw lsquoto urinatersquo

c h2nor- gt nwreOslash acircnergeOslash lsquoto be activersquo (Hesychius) cf h2ner-gt n rlsquomanrsquo

d h2uldquo

ors- gt oIcircregravew lsquourinatersquo oTHORNron lsquourinersquo cf h2uldquo

ers- gt ersan egravershn lsquodewrsquo

(3) Failure of vocalization of laryngeals in the environment of-oR C-

a torh1-mo-gt tigravermoc lsquoholersquo cf terh1- gt tegraveretron lsquogimletrsquo

b porh2-neh2 gt pigravernh lsquoprostitutersquo cf perh2- gt pegravernhmi pegraverasa lsquotosellrsquo

2 (2) and (3) are based on the summaries in Nussbaum (1997 181ndash182)

2



d solh2-uldquo




e (h1)ouldquo


f ( polh1uldquo

- rarr) poluldquo






h2-uldquo

o- gt salauldquo

o- gtsalu


˚h2-eu

ldquoo- gt salau

ldquoo- gt salu


soundrsquo4

b tolHuldquo

- gt toluldquo









˚HC gt

Lat CaRaC egp l˚







˚h2-u

ldquoo- to



˚h2-eu




˚h2-u

ldquoo- or sl

˚h2-eu


3



lt kwl˚

h1- or kwl˚

h2-







d polh2-uldquo


h2-meh2- or pl


ˇwa(i)- would








4

be expected



a solh2-uldquo


b kouldquo









l˚









5





apa spharıs˙


c (PIE kouldquo





soundrsquo














ldquoh1- (or ku



6





˙roots







( )

reildquoH- or k

( )

reH(ildquo)-





7





˙or aset




˙root






˙root mei





type

213 Uldquo


An anit˙

root uldquo


˙root u


˙form (u

ldquoer-mn

˚) Ras-


erHu-m0

anduldquo





ldquor



˙o-vr






8


˙root since Vedvr


˙t- suggest ananit


assumption ofuldquo


˙root

uldquo


214 POL(H)Uldquo








ldquo- and

not pl˚






ldquo- pallo- is






˚H-


˙a- lsquowingrsquo

9



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30



d solh2-uldquo




e (h1)ouldquo


f ( polh1uldquo

- rarr) poluldquo






h2-uldquo

o- gt salauldquo

o- gtsalu


˚h2-eu

ldquoo- gt salau

ldquoo- gt salu


soundrsquo4

b tolHuldquo

- gt toluldquo









˚HC gt

Lat CaRaC egp l˚







˚h2-u

ldquoo- to



˚h2-eu




˚h2-u

ldquoo- or sl

˚h2-eu


3



lt kwl˚

h1- or kwl˚

h2-







d polh2-uldquo


h2-meh2- or pl


ˇwa(i)- would








4

be expected



a solh2-uldquo


b kouldquo









l˚









5





apa spharıs˙


c (PIE kouldquo





soundrsquo














ldquoh1- (or ku



6





˙roots







( )

reildquoH- or k

( )

reH(ildquo)-





7





˙or aset




˙root






˙root mei





type

213 Uldquo


An anit˙

root uldquo


˙root u


˙form (u

ldquoer-mn

˚) Ras-


erHu-m0

anduldquo





ldquor



˙o-vr






8


˙root since Vedvr


˙t- suggest ananit


assumption ofuldquo


˙root

uldquo


214 POL(H)Uldquo








ldquo- and

not pl˚






ldquo- pallo- is






˚H-


˙a- lsquowingrsquo

9



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30



lt kwl˚

h1- or kwl˚

h2-







d polh2-uldquo


h2-meh2- or pl


ˇwa(i)- would








4

be expected



a solh2-uldquo


b kouldquo









l˚









5





apa spharıs˙


c (PIE kouldquo





soundrsquo














ldquoh1- (or ku



6





˙roots







( )

reildquoH- or k

( )

reH(ildquo)-





7





˙or aset




˙root






˙root mei





type

213 Uldquo


An anit˙

root uldquo


˙root u


˙form (u

ldquoer-mn

˚) Ras-


erHu-m0

anduldquo





ldquor



˙o-vr






8


˙root since Vedvr


˙t- suggest ananit


assumption ofuldquo


˙root

uldquo


214 POL(H)Uldquo








ldquo- and

not pl˚






ldquo- pallo- is






˚H-


˙a- lsquowingrsquo

9



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30

be expected



a solh2-uldquo


b kouldquo









l˚









5





apa spharıs˙


c (PIE kouldquo





soundrsquo














ldquoh1- (or ku



6





˙roots







( )

reildquoH- or k

( )

reH(ildquo)-





7





˙or aset




˙root






˙root mei





type

213 Uldquo


An anit˙

root uldquo


˙root u


˙form (u

ldquoer-mn

˚) Ras-


erHu-m0

anduldquo





ldquor



˙o-vr






8


˙root since Vedvr


˙t- suggest ananit


assumption ofuldquo


˙root

uldquo


214 POL(H)Uldquo








ldquo- and

not pl˚






ldquo- pallo- is






˚H-


˙a- lsquowingrsquo

9



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30





apa spharıs˙


c (PIE kouldquo





soundrsquo














ldquoh1- (or ku



6





˙roots







( )

reildquoH- or k

( )

reH(ildquo)-





7





˙or aset




˙root






˙root mei





type

213 Uldquo


An anit˙

root uldquo


˙root u


˙form (u

ldquoer-mn

˚) Ras-


erHu-m0

anduldquo





ldquor



˙o-vr






8


˙root since Vedvr


˙t- suggest ananit


assumption ofuldquo


˙root

uldquo


214 POL(H)Uldquo








ldquo- and

not pl˚






ldquo- pallo- is






˚H-


˙a- lsquowingrsquo

9



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30





˙roots







( )

reildquoH- or k

( )

reH(ildquo)-





7





˙or aset




˙root






˙root mei





type

213 Uldquo


An anit˙

root uldquo


˙root u


˙form (u

ldquoer-mn

˚) Ras-


erHu-m0

anduldquo





ldquor



˙o-vr






8


˙root since Vedvr


˙t- suggest ananit


assumption ofuldquo


˙root

uldquo


214 POL(H)Uldquo








ldquo- and

not pl˚






ldquo- pallo- is






˚H-


˙a- lsquowingrsquo

9



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30





˙or aset




˙root






˙root mei





type

213 Uldquo


An anit˙

root uldquo


˙root u


˙form (u

ldquoer-mn

˚) Ras-


erHu-m0

anduldquo





ldquor



˙o-vr






8


˙root since Vedvr


˙t- suggest ananit


assumption ofuldquo


˙root

uldquo


214 POL(H)Uldquo








ldquo- and

not pl˚






ldquo- pallo- is






˚H-


˙a- lsquowingrsquo

9



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30


˙root since Vedvr


˙t- suggest ananit


assumption ofuldquo


˙root

uldquo


214 POL(H)Uldquo








ldquo- and

not pl˚






ldquo- pallo- is






˚H-


˙a- lsquowingrsquo

9



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30



















10




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30




241 METATONY










ldquoos(gt -is [i


ldquoa [or - ıi

ldquoe]




11












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30












a or eiuldquo




12



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30



cf Latv salna

b (PIEkolH-uldquo


cf Latv kalva








13








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30








h2-dhh1- gtgou



ldquoh2-ei





ldquoh2-dhh1-i





(lt uldquo


r˚











˚--men-

14


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30


















15









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30









h2-ildquoo- gt krou

ldquo-

ildquoo- gt kraujas




˚j- Hitt esh





ldquoh2-i

ldquoo- meets



16

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30

kreuldquo

ildquoo- lt kreu

ldquoh2-i














17


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30


ep-no-suldquo


es-no-u

ldquoos-no-(u


ldquoeg-no-u












18



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30



˚(s)











19









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30









˙variant



20

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30

reflexes ofset˙

root anit˙





(lt kl˚



ldquoo-)


˙andset



andset˙


˙andset






been aset˙




21








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30








22



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30



˚h3-n-es kl

˚h3-onplusmn i


˚h3-n- rarr k l






ep-no-su








˚h3-no-

k l˚





23






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30






andset˙








24






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30






ldquoeh2 A




ldquou-h2ei

ldquou-rarr h2oi

ldquou-


ldquou-o-h2ei



ldquoo- and

kelH-uldquo

o-




H-euldquo

o- or sl˚

H-uldquo






˚H-u- andsl

˚H-u





recalled

25


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30


o- gt kol-uldquo


ldquoo- gt kelH-u





5 Conclusion







ldquo- (gt kolu


ldquo-eh2



in this case

26

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30

References

















22Hyllested A


IEW = Pokorny J

27




















28



















29







30




















28



















29







30



















29







30







30

The Saussure Effect in Lithuanianyy/hp/documents/the_saussure_effect.pdf · The Saussure Effect in...

Documents

Transcript of The Saussure Effect in Lithuanianyy/hp/documents/the_saussure_effect.pdf · The Saussure Effect in...