PRE-GREEK LOANWORDS IN GREEK (Beekes)

A. Introduction The substrate language of Greek will be called Pre-Greek in this dictionary; this is a translation of the German term das Vorgriechische. No written te ts e ist in this language! but it is known from a considerable number of loanwords in Greek. The study of Pre-Greek has had an unfortunate history. "n the #ast century! it was called Pelasgian and considered a dialect of "ndo-$uro#ean. This idea fascinated scholars! and research concentrated on this #ro#osal. %ut the whole idea was clearly wrong! and by now! it is generally agreed that the substrate was non-"ndo-$uro#ean. Therefore! the term Pelasgian can no longer be used. &risk already had strong doubts about the Pelasgian theory! but ne'ertheless! he often mentioned the #ro#osals of its adherents. (ince all work following this line has turned out to be useless! " decided to make no mention of the theory anymore in the dictionary. )hen &risk com#leted his dictionary in *+,-! &urn.es book /ie wichtigsten konsonantischen $rscheinungen des Vorgriechischen! which was his dissertation written under the su#er'ision of &.%.0. 1ui#er! had 2ust a##eared. "t was an elaboration of 1ui#ers *+34 study on Greek substrate words! which o#ened a new cha#ter in the research of the field. &urn.e re2ected the Pelasgian theory! too 5see es#ecially o#. cit. ##. 67-338. &urn.es book met with fierce criticism and was largely neglected. "n my 'iew! this was a ma2or mistake in Greek scholarshi#. True! some of his identifications are im#robable! and his re#eated claim that certain forms were e #ressi'e leads nowhere. )hat remains! howe'er! is that he studied a great number of rele'ant forms and drew ob'ious conclusions from them. Pre-Greek words often show a ty#e 7f 'ariation which is not found in inherited words. "t is self-e'ident that this 'ariation must be studied! and this is what &urn.e did. "t has turned out 5as 1ui#er had already shown8 that this 'ariation shows certain recurrent #atterns and can be used to recogni9e Pre-Greek elements. &urn.es book is not easy to use: e'ery form is discussed at three or four #laces! each time in a different conte t! so that it may be difficult to find out what his #oint really is. ;n the other hand! his treatment is 'ery careful! and there hardly any ob'ious mistakes. " found a number of cases which he had not recogni9ed 5e.g. <=>?@8! but this does not change the fact that his book was the best collection at the time. &urn.e worked on it for twenty years! and e'en now it is the only handbook on the sub2ect. The short o'er'iew which follows below is based on &urn.es material and on my own research of more than thirty years.A*B &urn.e went astray in two res#ects. &irst! he considered almost all 'ariation to be of an e #ressi'e character! which is certainly wrong: it is e'ident that the 'ariation found is due to the ada#tation of words 5or #honemes8 of a foreign language to Greek. )e shall see below that many 'ariants can be understood in this way. (econdly! &urn.e was sometimes o'er9ealous in his search for inner-Greek corres#ondences. Cany of &urn.es disco'eries are brilliant 5see s.'. DEFGHIEH for an e am#le8! but sometimes he went too far: not e'ery alternation necessarily #oints to Pre-Greek origin. The author can hardly be blamed for his enthusiasm. Je was e #loring new ground! and it can only be e #ected that he sometimes o'er#layed his hand. (e'eral scholars were baffled by &urn.es #ro#osals and hence re2ected the whole book altogether. Jis method! howe'er! was correct and " ha'e only filtered out the im#robable suggestions. "n many cases! of course! we cannot be absolutely certain! but this cannot be an ob2ection. $ ce#t for a 'ery small number of cases! &urn.es material does consist of Pre-Greek words. Jis inde contains 6677 words! and taking into account that many of these words concern

deri'ati'es and 'ariants! as well as a few "ndo-$uro#ean words! " estimate that &urn.es book discusses some *777 Pre-Greek etyma.A-B "n general! " ha'e gi'en only a few #ersonal names and to#onyms! and no material of this kind from outside Greece and Ksia Cinor. The com#arison with %asLue or Maucasian languages has not been considered in this dictionary! as this is not my com#etence; it is likely that there are such connections! but this must be left to other scholars. Cy suggested reconstructions are not essential. ;ne may ignore them and 2ust consider the 'ariation itself. These 'ariants are often e #lained as incidental #henomena 5assimilation! influence of other words! etc.8! and such e #lanations may be sometimes correct! but if we know that some 'ariants freLuently occur! we will ha'e to consider Pre-Greek origin. $ isting etymological dictionaries often seem to a'oid the conclusion that a word is a substrate element. "t is remarkable that Mhantraine was Luite aware of the #roblem in his Formation! but in his dictionary he often withdrew his earlier e'aluation 5which in my 'iew was correct8. "t looks as if substrate elements were not welcome there. The relationshi# with Knatolian languages is a se#arate #roblem. K Greek word is often called a loan from an Knatolian language! while it may 2ust as well be borrowed from the Pre-Greek substrate. "t is generally acce#ted! on the basis of to#onyms! that there was a language which was once s#oken both in Greece and in western Ksia Cinor.ANB "n most cases! howe'er! it is im#ossible to distinguish between substrate words and loans from Ksia Cinor 5the latter are from a later date8. K word may ha'e been ado#ted through commerce! as often ha##ens between two neighboring countries! or starting from the time when Greeks settled in Ksia Cinor! #robably as early as the *3th century. &rom a methodological #oint of 'iew! " think it is better to consider such words as Pre-Greek! and to define them as loanwords from an Knatolian language only when there is reason to do so. (till! it is clear that we may often make mistakes here. K case in #oint is =EO<P clew! ball of wool ready for s#inning. The word is clearly related to Quwian and Jitt. taluppa/i- lum#! clod. The Greek word is ty#ical of Pre-Greek words: the structure CaC- up- 5with a a##earing as o before u8 and the absence of an "ndo-$uro#ean etymology 5Celchert Orpheus R 5*++R8: 6,-3* is not con'incing8 im#ly that the word is Pre-Greek or Pre-Knatolian. ;n the other hand! clew is not a word that is easily brought from o'erseas; it is an e'eryday word that the s#eakers of Greek and Knatolian must ha'e #icked u# not far from home. " com#letely agree with &urn.es inter#retation 5N3NN8 that the word was brought to Greece by settlers from Knatolia who s#oke the language! which! from another #ers#ecti'e! we call Pre-Greek. "n other words! =EO<P is a loan from an Knatolian language! but this 5#robably non-"ndo-$uro#ean8 language was also s#oken in large #arts of Greece before the Greeks 5s#eaking an "ndo-$uro#ean language8 arri'ed there. "t is essential to reali9e that substrate words are a freLuent #henomenon. ;ne may regret this 5for instance! from the "ndo-$uro#eanist #oint of 'iew8! but this is irrele'ant; the e istence of Pre-Greek words is sim#ly a fact that has to be acce#ted. To me! it is fascinating that in this way we can learn something about the oldest language of $uro#e 5including Knatolia8! of which we otherwise ha'e no e'idence. The Pelasgian theory has done much harm! and it is time to forget it. The latest attem#t was Jeubecks Cinoisch-Cykenisch 5discussed by &urn.e 33-448! where the material was reduced to some ten words; the theory has by now been tacitly abandoned. B. Phono o!" #. $he %hone&ic s"ste& o' Pre-Greek

Voiceless! 'oiced and as#irated sto#s may interchange in Pre-Greek words! without any a##arent conditioning factors. This fact shows that 'oice and as#iration were not distincti'e features in Pre-Greek.A6B ;n the other hand! the Qinear % signs 5gra#hemes8 for rjo! rja and tja show that palatalization #robably was distincti'e. This is confirmed by the sign pte 5e.g. in ra-ptere Shra#teresS with the agent suffi -ter- )! which must go back to an earlier pye. "n the Pre-Greek material! such a #honeme may underlie e am#les like T<=U. ;ne may wonder whether GFEVVWTEH #oints to py X pt! which was reali9ed with as#iration. &urther! the signs two! twe! dwo! dwe! nwa! swa! swi! #oint to labialization as a distincti'e feature! i.e. two! twe! dwo! dwe! nwa! swa! swi. Note that #alatal and labial forms of gra#hemes are found both with resonants and sto#s! which is a #henomenon alien to "ndo-$uro#ean languages. The e istence of labio'elars is confirmed by qa-sire-u Y ZUVIO[@! etc. 5see further %eekes Glotta ,N 5*++3S48: *-f.8. )e may thus #osit the following systemA3B:

# t k s r l m n

#y ty ky sy ry ly my ny

#w tw kw sw rw lw mw nw

;f course! it is #ossible that one or more of the #osited #honemes did not occur in Pre-Greek 5e.g.! my is a rare sound in the languages of the world8. )e can now use this insight in e #laining the surfacing Greek forms. Thus! DWHP S DU\?H5U8can now be e #lained from a Pre-Greek form ]dakwn- .A4B "n the former form! the labio'elar yields a labial sto# W. "n the latter! it is rendered by -\?-! with antici#ation of the labial feature! while the labio'elar turns u# as a 'elar! #ossib-ly by dissimilation from ukw. Kgain! note that as#iration is not #honemic in Pre-Greek. "t is 'ery im#ortant to note that we cannot #redict how a Pre-Greek form will surface in Greek: sometimes a sto# turns u# as an as#irate! sometimes as a 'oiced sto# 5e.g. U<@ S WUF! see % 3.*. below8. Ks a conseLuence! it may ha##en that there is a large number of 'ariants! but it may also be that there are no 'ariants at all. Ks a second e am#le! we may also understand U?H S Qesb. ^WPH from a #re-form ]ankwn. The latter form is directly understandable! with W from the labio'elar. The first form went through ]anwkn or ]awnkn! gi'ing U?H with loss of the nasal 5a de'elo#ment known from Krmenian8. Perha#s! a scenario ]akwn X U?H is also #ossible! with a #renasali9ed form ]ankwn 5X ^WPH8 beside ]akwn.A,B (uch inter#retations may be wrong in indi'idual cases! but this is no reason not to try. ;n the other hand! 'ariation that is strange from an e clusi'ely "ndo-$uro#ean #oint of 'iew becomes understandable in this way! starting as we do from a limited set of assum#tions.

The e istence of #alatali9ed #honemes in Pre-Greek may e #lain a number of other de'elo#ments. Thus! " assume that a geminate OO may continue Pre-Greek ]ly. )e know that "$ ]ly ga'e OO in Greek! but if a 'ariant with single O coe ists! we are warned. &or e am#le! the name _?IOO[@ has a 'ariant _?IO[@ with one O. Knd although the latter only occurs in Jomer! this fact #oints to Pre-Greek origin. The 'ariant was preser ed because it was metrically con'enient! it was not !reated for metrical #ur#oses. ;f course! the fact that there was more 'ariation at an earlier date is what we e #ect. Ks far as the other #alatali9ed resonants are concerned! any may ha'e gi'en UIH! ary may ha'e gi'en UIF 5or also [IF with coloring of the 'owel! see section M- below on the suffi es8! etc. )e ha'e -UIF-! -UIH- but no ]-UIO- in Pre-Greek words. This is confirmed by the fact that geminate OO is 'ery freLuent 5&ur. NR,8! whereas geminate FF! HH and ^^ are much less freLuent! or e'en rare. "n a similar fashion! ]asy may ha'e yielded either -UIV- or -UV-! cf. GZUIVE@! which has a '.l. GZUVE@. "n rendering such a foreign word! the #alatali9ation may ha'e been re#resented at one time! and may ha'e been neglected at another. This #henomenon was the main cause of 'ariation in Pre-Greek forms. The inter#retation is further confirmed by the #arallel de'elo#ment of labiali9ed consonants. Thus! " su##ose that arw resulted in -U5\8F- 5see the section on the suffi es8. "n this way! we may understand GUOUFE` beside GEOFEZEH from a #re-form kalarw- op-. Knother form which shows the remarkable interchange USU\ is FUV?D[@ S UFEV?@. Jere one might assume a #reform ]arwask-at-. Note that the labial element would at the same time e #lain the o as a 'ariant of a in both cases. K similar mechanism must be at the basis of the etymon OEa! UOUa! OUa! [OG! which is ho#eless from an "ndo-$uro#ean #oint of 'iew. " assume that all forms go back on Pre-Greek ]alw- ak- . "t gi'es UOUG- through antici#ation! OEG- through coloring. "n this way! the first two forms! which are best attested! are directly clear. &urther! U\S[\S> interchange freLuently! which e #lains OUa and [OGU; OEG- is not #roblematic either! as both SaSs were colored to AoB by the labiali9ed resonant. ;nly the Jomeric accusati'e OGU is ho#eless: it is the only form that has no 'owel between O and G! and therefore may be due to some accident of the tradition. This is a #roblem that has not been sol'ed yet. " do not know whether a di#hthong is allowed in suffi es of the structure VM! cf. the forms in -UIE@. (tructurally! one could think of -ayw-" or e'en -awy-" but such sounds are rather rare in the languages of the world. Kn instance of -UI- due to a #alatali9ed consonant is aUWHP@ S aU<HP@ S WH> 5a brilliant combination by &ur. *3R! etc.8! which must contain -apy- 5the #alatali9ation was ignored in the last form8. Mom#arable to the de'elo#ment in aU<HP@ is GH` S GIH<[=EH! from kyn- with I re#resenting #alatali9ation! cf. %eekes -77R. Qikewise! " assume that <IH\=@ beside <H\=@ #oints to ]pynut-. Perha#s! we must inter#ret VI><> as ]sy#p- because of [V><U. Kn interesting case is O^IHT[@b O^IHT[@! for which " assume ]lym- beside ]alym- with #rothetic a 5see %N below on the #rothetic 'owel8. K #alatali9ed consonant could color a to e. K good e am#le is G<UFE@! G<UIFE@! but also G<[FE@! G<[IFE@! where we ha'e all #ossible 'ariants due to the #alatali9ed consonant. Mom#are further cZUFHEI ne t to cZ[IFEI. Qikewise! we ha'e dUG[O=@ ne t to d[G[O=@ GEOG\H=UI! where the interchange occurs after d from earlier #alatali9ed ty. DIWTFU beside DI`FU may ha'e had -pty-; O5O8E` ne t to O5O8ZP@ goes back to ]alyap- " with the common 'ariation a / o before a labial. K clear e am#le is OUVI=@ with! ne t to it! O[VI=@ and OV=UI! OV=U\FE@. "t may be inter#reted as re#resenting PG ]lasyt-. 1ui#er $in%ua -* 5*+4R8: -4+--,, #ointed out that the substrate language had labio'elars. Je es#ecially #ointed to TUO\GF@ ne t to TO\`U! TO<>. " added a few remarks in %eekes Glotta ,N 5*++3S48: *-f. &rom Cycenaean! we ha'e a-to-ro-qo 5HTF><E@8 and qe-to 5<TE@8! &o-qo-so 5e`E@8! qi-si-pe-e 5the dual of aWE@8. &urther there is '-i-ti-jo-qo 5gen. _TE<E@8! WTUO^@ with the 'ariants G=UOOE@ and <=O5O8E@! which cannot be e #lained from "ndo-$uro#ean.

"nstead of aWE@! we would #erha#s e #ect ]]`WE@. (o the de'elo#ments are largely as those of Greek! but not com#letely. Pre-Greek #robably had a SyS and a SwS. "nitial ya- #resumably often lost its y- " but it may sometimes be re#resented by U- as in U^ZE@! V f >H. The ending -\IU may ha'e been -uy- a 5a Pre-Greek y may ha'e had a different de'elo#ment from y in inherited words8. "n the same way! -UIU may deri'e from PG ]- ay- a with a 'ariant -[IU! cf. gPH[O<[IU. Perha#s! the y disa##eared in some cases! gi'ing hUU beside h 5see below on the suffi -UI- S -[5I8-8. "nitial w- was often lost 5HUa8! but wa- may also ha'e been rendered by EU-! as in Ua@ beside Mret. Ua@. The same holds for O[@! which has been considered to be identical with the root of OE@8. )e find \U- 5which became U-8 in GIHTE@! Mret. GIHTE@. &ur. N,, assumes a #rothetic \- in the latter word! but this seems im#robable to me. Knother e am#le may be US[OE@. The differences are #robably due to the date at which the word was borrowed and de#end on whether the Greek dialect concerned still had a at that time. Knother treatment can be found in the word for truffle! for which we find EI=EH! EDHEH 5also -=H-8! DHEH 5also -=H-8! or =EH. These are #robably all renderings of ]wit-. 5&ur. *R6 again assumes a #rothetic 'owel! I=- S I=-! which does not seem to be the right solution. Je further assumes a 'ariation ]wit- S wut-" which also seems im#robable to me! though the 'ariation I S \ is attested.8 iather! \- is a form of EI-! with the -ochanged under influence of the -I- 5cf. Qe2eune *+,-: *,6! and note that Greek did not allow -\Ibefore consonants; of course! EI became \ in %oeotian in the Nrd c. %M; 'ariation EI S \ is found in more Pre-Greek words8. This case nicely shows that 'ariation in Pre-Greek words is due to different rendering of the sounds of a foreign language! and therefore has to be taken seriously. ZFGUOEHb <UOEH 5J.8 #robably attests a de'elo#ment ]wrak- X ZFUG- 5as &ur. *6, remarks on GUOUFE`: j/ie landlkufige $tymologie lconnectingX <> ... ist wohl ohne weiteres auf9ugeben.m8. VEFUb <UOIEFE\ [DE@ sorb-a##le 5J.8 continues ]sorw- 5cf. Qat. sorbus! &r. sorbier! &ur. -N78. "t seems that there was no initial as#iration in Pre-Greek. &urn.e has a few words with !-! "5one or two with #-; none with $-! %-! &-8. (e'eral of these are doubtful; best is U#^UVI 5U#^E8. ;ne might conclude that the language had no h. This would agree with the fact that as#iration is not a distincti'e feature in the sto#s. Jowe'er! this conclusion is remarkable for 'F>@! (OOPH[@ and )WUIV=E@! which we e #ect to be Pre-Greek words 5but note that Cyc. a-pa-i-ti-jo does not ha'e a(- ). ;f course! as#iration may ha'e been added secondarily in Greek in indi'idual cases! cf. the 'ariation in WTU S *WTU and O[DHP S "O[DHP! which is a 'ariant of D[O[DHP. Jowe'er! Prof. iui2gh #ointed out to me that Cycenaean had to#onyms 5a(-ra-tu-wa8 and #ersonal names 5a(-kumi-jo8 with initial h- ; it also occurs in inlaut 5pi-a(-la! ko-ri-a(-da-na8; cf. further e-ma-a( 5SJermnhnsS Jermes8. ;riginally! " thought that Pre-Greek only had three 'owels: a! i! u. The Greek words concerned often ha'e [ and E! but this would not be sur#rising! as the three 'owels ha'e a wide #honetic range! and the #honeme SaS may ha'e sounded like AeB or AoB in many en'ironments. The main reason for me to assume this sim#le three-'owel system was the fact that the system of suffi es has a! i! u! but not e! o. )e ha'e -Uh-! -Ih-! -\h-; #renasali9ed -Uhh-! -Ihh-! -\hh-; likewise -UT-! -IT-! -\T-; and #renasali9ed -UHT-! -IHT-! -\HT-! but no forms with -[h5h8-! -Eh5h8-! etc. The only cases " noticed are +VGEHTE@ and OEHTE@ 5but as a 'ariant of O\HTE@8! and ^POEOHTP with a 'ariant ^PO5EO8HTP. iecently! " ha'e become more inclined to assume a system with the usual fi'e 'owels! because there seems to be a distinction between the two 'ariations U S [ and U S E! on the one hand! and a stable! not interchanging U! on the other. This would #oint to a system with a! e and o. ;n the other hand! it is difficult to e #lain why the suffi es do not show the same 'ariation that we find in the root 'owels.

"t is essential that the #alatali9ed and labiali9ed consonants colored an ad2acent U to [ and E! res#ecti'ely. ;n the effects of #alatali9ed consonants see %eekes -77R: 64-33. &ur. N67 has a rule U X E before E! >! \ 5e.g. GUO\Z@ S GEO\Z@8; this can now be understood as the o-like reali9ation of SaS before high rounded 'owels in the following syllable 5see *3.N.-8. (o! e and o originally were 'ariants of the #honeme SaS. "t is difficult to establish whether they had already become full #honemes in Pre-Greek. K good illustration of the case is the name of K#ollo. "n Jittite! 'ppaliunas renders 'pollon- 5see %eekes )'*+, N! -77N8. )e know that Greek originally had ,<[OO-! with -[- arising from -a- before the #alatali9ed ly. The -o- de'elo#ed only later in Greek! but " assume that the Jittite form still shows the -a- . The Pre-Greek form was 'palyun-. " ha'e long doubted 5and still doubt8 whether there was #honemic 'owel length in Pre-Greek. Greek substrate words Luite often only ha'e a form with a long 'owel. Vacillation is sometimes found! as in TFIHGP beside TFHUa 5see o 4.-8! and note ZFI^E@ beside ZF^@! ZFf^P. puite a different argument is the following: ?\FEH and <=-FEH both mean chaff; it is therefore #robable that they contain the same suffi -\F-; but in the first word the u is short! while it is long in the second. Note that P often re#resents 5hT\OO@ S hPT-8! and as our knowledge of the rele'ant dialects is rather limited! we often sim#ly do not know whether P re#resents an older a or e. "f we had not had /or. VDFE@! we would not ha'e known that it contains an old -. Klso! q.^HE@ re#resents q^HE@. There are well-known Pre-Greek words with P l ]! like V<OUIEH. " assume two di#hthongs! ai and au. "f there were no e and o! we do not e #ect other di#hthongs. K di#hthong [\ is rare 5&ur. N3N Knm. 3; " found some *- instances in the whole of &urn.es material8; it interchanges with U\. &ur. NN+ Knm. -8 calls [I r5in mehreren &kllen8 nur eine Nebenform 'on UIr. Klso! EI is rather rare! and we may find E\ more often! but mostly interchanging with other 'owels 5see the remark on the suffi -E\F-8. (ee further section %4.* on 'owel 'ariation. iegarding the accentuation! " noted 'acillation in: ZFU^I@ S -^@; UhOIE@ S -I@; ?\FE@ S -@; ?>F S ?F; GF\DE@ S GEF\D@; GEFDUOE@ S GEF\DUOO@; ^DI^HE@ S ^[DI^H@; VG\E@ S VIG\@; FIV?E@ S FIVV@. Note also the almost identical forms such as O\GU`@ S OGE`E@. This does not im#ly that the language had no clear stress: the Greeks who ado#ted a word could sim#ly ha'e been uncertain about it. The #henomenon may! howe'er! be im#ortant heuristically: such 'ariation is 'ery rare in inherited words. (). *h)r)cteristic sounds )nd sound !rou%s "n Pre-Greek words! we find some sounds or clusters that are rare in P"$ words. "n brackets! " gi'e the 'ariants. #. +,- ;f course! U\ does occur in P"$ words! but only when it deri'es from ]h(eu 5mostly in initial #osition8 or eh(u. $ am#les: ZOUD[@! ZFUGU@! hFUGUOU@! GHU\V=FEH! GUVUFU! =FU\aHU; qUZFU\HD@. (. .- Ks is well known! ]b was rare in P"$. "n Pre-Greek words! it seems to occur relati'ely often. $ am#les: ZOUFEI! ZFZPOE@! FZOP! =FZUG=E@! ZFZIOE@! TF\ZE@! GZUOE@. "t is freLuently found word-initially. ;f course! Z may also go back to a Pre-Greek labio'elar 5i.e. labiali9ed 'elar8: e.g. ZUVIO[@! Cyc. qa-si-re-u.

/. .0- The cluster is #ossible in P"$ words! but it is rare 5see on Z sub -. abo'e8. $ am#les: ZD[OOEH! ZDPFU! ZDP@! ZDP@! U=E-GZDUOE@! GZDPOE@! G\ZZDU; cE^ZDIOI<IU. 1. 20- Mf. &ur. N*R3. There is nothing against P"$ ]%d! but it is infreLuent. ;f course! the grou# is reminiscent of ZD. $ am#les: hD\@! ^\hDOP! hDE\<> 5cf. G=\<>8! hDP! GFhDUHEH! OhDP. 3. 24- $ am#le: hH@ 5GH@8. ;n ?H! WH! see the section on the suffi es. 5. 04- The seLuence is rare in "$ words. $ am#les: GIDH@! OU<UDH@! F?IDHU! O[<UDH@ 5OU-8! V<\DHE@; _FIDHP. 6. 78- The grou# is regular in P"$! but in Pre-Greek it is found with 'ariants; see %3.3. $ am#les: Z\G=EH! ZG=UI! DG=\. 9. 7:- The grou# can hardly be of "$ origin! but it is not freLuent. " noted ZG?UF! OG?U! VG?UF! V\G?@; oG?E@! oFUG?E@! oG?I@. The grou# -G?- is the geminate of ?. Mf. on <W! =T. ;. <4- The grou# is certainly #ossible in P"$ words! but it is also freLuent in Pre-Greek. $ am#les: ^WI-GO[^HEH! [DI^HE@! U^HE@! ZUV\^HI=P@! GFP^H@! O^HU! O>F\^HH! ^FI^HU! DU^HE@! Vh\^HEH! VDFI^HEH; ,=^HIE@. #=. >,- The di#hthong is #erfectly "$! but it is found se'eral times in Pre-Greek. " do not think that Pre-Greek had a di#hthong -ou-" but it may ha'e arisen from e.g. -arw- " which often surfaces as -E\F-. $ am#les: V[HDEGP! VGE\FE@! V=FE\T@! =h?E\FE@! =EWE@! WHDE\FE@! WEVGE@! ?OEHP@. ##. ?@- The grou# can hardly be of P"$ origin! but it is rare in Pre-Greek words! too. Qike in the case of G?! it is the geminate of W. $ am#les: F?Id<WP@ 5s8; tU<W 5uU<W8. #(. A0- ;n a mor#heme boundary! the grou# is #ossible in P"$. $ am#les from Pre-Greek: hFDU! GU<UFD[VUI! GUFDU^OP. #/. A74- K rare grou#! #erha#s there is e'en no reason to s#eak of a grou#. $ am#les: ZUFGHU! Z[FGH@. #1. A4 5'ariants FD! HD8: $ am#les: GVIFHI@ 5-HD-8! ?FDU 5-HU8! VGU<FDU. (ee the section on the suffi es. #3. K V occurs both word-initially and between 'owels! where it has disa##eared in most inherited words. "nitial: VF\==U! VUhFIEH! VHU<=IH! VHDUOEH! VUFF\WT[H! V[GEU! VIZHP! Vh\^HEH. "nter'ocalic: hUV\OO@! h?E\VU 5h?-8! UTE\V5V8U! U#^UVI! UVUGE@! O[IVEH! DFVE@. Kfter resonant: OVE@! ZOVU^EH! h[OVH! hHVI^E@! ^FVI<<E@ 5-\<<E@8. #5. B.- The grou# is hardly known from inherited words 5VZHH\^I is #roblematic8. $ am#les: VZEOE@! TVZP! /VZ[=E@. -VZ- may continue Pre-Greek -s%w- : Cyc. ti-qa-jo may stand for SthisgwaiosS vIVZUE@. #6. B2- Kgain! this grou# is hardly known from "$ words. "t may sometimes continue -ty%- " as in ^\VhO! ,Vh[O=U@ 5see 3.38. $ am#les: OIVh>! VhHP! WVhUHEH! VhHDP@! <IVh@. #9. B7C B8- These grou#s are well known from "$! but mostly in word initial #osition. (ee section %3.3. $ am#les: ZVG[FEI! ZV=Ua! GV=[FEI! OUV=Uh[.

#;. B8D- Though the cluster contains nothing that could not be "$! it occurs more often in substrate words. $ am#les: V=OIhh[@! V=O[hh@. (=. 8E- The grou# can hardly be of P"$ origin. "n Pre-Greek! it is a 'ariant of == and VV 5see 3.38. (ometimes! it is clearly the geminate of T: ,=T@ beside ,THP. &urther e am#les: =TOU! G=T\ZE@! g[=TUOE. (#. @E- The cluster is #ossible in inherited words. $ am#le: HVGUWTEH. ((. :<C :4- iather rare in "$; &ur. **7 assumes that the nasal caused the as#iration. $ am#les: DU\?^@! DU\?HU-! VU\?^H. (/. &risk gi'es some se'enty lemmas with F-. Cany words are clearly Pre-Greek! and there are no con'incing "ndo-$uro#ean etymologies. That many of these words are of substrate origin is also clear from the fact that there are 'ariants with V-. K##arently! Pre-Greek did not ha'e any difficulty with ps- " as Greek has so many words with `-. ;riginally! " thought that all words with `- were Pre-Greek! but this thesis cannot be maintained. Kmong the non-substrate words! `OOU originally did not ha'e ]ps-" and `- for WT- is secondary 5see Qe2eune *+,-: N+8; the 'erb `> may well be non-"$. (1. G- ;f course! > is #erfectly "$! but it also occurs in Pre-Greek words. $ am#les: ^UGF0=I@! HTF><E@! H>H@! <EWOIE@! FF>D>! VGUO<U@! UVGHDU@! V^>O[H! ZUOO>=! GUVV>F@! O>F\^HH. (3. Ge&in)tes 5see also %3.R on single S geminated consonants8: "ndo-$uro#ean had no geminates. ;f course! geminates arose in Greek! but they are not 'ery freLuent. " doubt whether Pre-Greek had geminates! but se'eral occur in Pre-Greek words 5%ri he *+,4: +3 states that there were no geminates in this language8. Ks Pre-Greek had #alatali9ed #honemes! " wonder whether ly was 5often8 re#resented by OO in Greek. "n a similar 'ein! #erha#s ny might be re#resented as HH! and ry as FF! but this needs further in'estigation. &or VV and == see %3.3. wnclear are DD! GG! <<! and ^^ 5a #alatali9ed my is a rare sound8. (ome further e am#les:

(to#sARB: DD: DDUI! DDIa GG: GGUOE@! ZO[GGE@! OGGE@5s8 <<: hFI<<E@! OE<<I@ ==: Z==UGE@! O==U! G==E@! ^==UZE@! ^==[@! ^==E@; gI==UG@.

QiLuids:

OO: OOZP@! OOE<P@! ^OOUGUH! ZUOO>=! ZDOOU! ZOOIH! <=[OOU ^^: GO[^^@ HH: hHHU! ZOHHE@! hHHE@! OU?HHU; xG=\HHU FF: FFZUGU! ZFFP! ZFFEa! GUFFH

(ibilant VV: UTE\V5V8U! WFIVVU! hVVU. (H. IoJ to reco!niKe Jords )s Pre-GreekL This a##ears to be relati'ely easy. K first indication is that a gi'en word has no "$ etymology. ;ften! there is 'ariation which is im#ossible to e #lain in "ndo-$uro#ean terms. Therefore! the discussion of these 'ariants is essential. Then! there are numerous suffi es that are ty#ical for PreGreek 5see the list below8. The meaning may also #ro'ide an indication. The words concerned are often names of #lants or animals! or #art of 'iticulture. &reLuently! the words are se ual terms. "f we ha'e some of the abo'e features! it is Luite clear that we are dealing with a Pre-Greek word. The origin of the word is then indicated y#g in the dictionary. "n many cases! we do not ha'e enough data and can only sus#ect that the word might be Pre-Greek 5the origin is then indicated as y#gs8. /. Prothetic MoJe Pre-Greek had a #rothetic 'owel! e.g. !VGOUWE@ beside GOUWE@. "n most cases! the 'owel is -. The numbers 5&ur. N4Rff.8 are as follows: U z +7! E *7! [ 3! I N! \ {! P 4! UI -. Note that! generally s#eaking! U may interchange with E! [! and UI. "ndeed! we ha'e cases where #rothetic E interchanges with U! and the same holds for [ 5e.g. [GO- S UGO-! "`U S `U8. Klthough not all other cases can be e #lained away! it seems that the #henomenon originally only concerned U. $ am#les: hUV\OO@ S hPT\OO@; GIF@ S GFFI@; GEFHE S GFHE`; ?FUDU^OU S ?FU^UDEOUI; HUF=P@ S HPF=P@; VGOUZE@ S 5V8GUOUZ=P@; ?H>` S GH>`. 1. s-&oHi e K large number of words shows an initial V- before a consonant! which is absent in #ractically identical 'ariants. "t occurs before a sto# or m 5so not before r! l! n8; the sto# is mostly 'oiceless! sometimes as#irated; see &ur. N+7f. $ am#les: hO[HE@ S V?OIHE@! 5V8GIDWP! GG[FE@ S VGhGE@! 5V8GEFD-fOP! Z=UOE@ S V<-! <O[TE@ S V<-! WU==hP@ S V<-! TFIhG@ 5=FIh?@8 S V=Fh?@! =E<[EH S V=\<<[EH! 5V8^FIHTE@! 5V8^FUIHU. K #rothetic 'owel may a##ear before an s-mobile 5&ur. N+7R8: VGOUZE@ S VGUOUZ=P@ S GUOUZ@! VWFUhE@ S VWFUhE@ S WFUha! VGOUWE@ S GOUWE@.

3. *onson)nt M)ri)tion 3.# Noice ess O Moiced O )s%ir)ted sto% &urn.es conclusion was that Pre-Greek was a non-"ndo-$uro#ean language! with no recogni9able cognates. This im#lies that the #honemic system may ha'e been different from that of "ndo$uro#ean. Thus! he found that the sto#s show 'ariation between 'oiced! 'oiceless and as#irated! so that there #resumably was no #honemic distinction between 'oice and as#iration in the language. Ks there is no reason to assume that this is a recent #henomenon! it strongly suggests that the language was non-"ndo-$uro#ean. &or e am#le! <=>?@ belongs to a root pt-k- S pt#k- also seen in <=a! -G@. (ince such a 'ariation is hardly understandable in "ndo-$uro#ean terms! the word must be Pre-Greek. &urn.es discussion of this 'ariation runs from #. **3 till #. -77. $'en if we allow for some mistakes! it is clear that there is abundant e'idence for this #henomenon. 3.( Pren)s) iK)tion

%efore a sto#! a nasal may be #resent or not in Pre-Greek words. $.g. G?F\@ S GUh?F\@! GEF\W S GF\^ZE@! VUOZP S VUO^ZP! etc. The #henomenon is e tremely freLuent! but its #recise origin is not known 5#renasali9ed consonantss8. 3./ N)s) iK)tion K consonant is re#laced by a homorganic nasal: GIDUW[[IH S GIHUW[[IH! WOPD0H=U S WOHUWE@. 3.1. L)Hi) sto%s O m O u There are three interchanges: labial sto# S ^! labial sto# S and ^ S . L)Hi) sto% O < 5&ur. -7N---,8. $ am#les: FZOP S F^\OU n.#l.; ZFZI=E@ S ZF^I=E@; G^IHDI@ S GZIHDI@; O\GZU@ S O\G^U@; ^V=Ua S ZV=Ua; VGO\^E@ S VGO\ZE@; WF^UGEH S WFZUH=U; VWFUhE@ S V^FUhE@. L)Hi) sto% O 5&ur. --R--6-8. $ am#les: =TP<U! T<E@ S TU^U; GZUOE@ S GU\UO@; GUVVUZ@ S GUVUFU; GF^ZE@ S GFUFE@.

< O 5&ur. -6---6,8. K difficulty here is that Greek did not #reser'e a in most cases! so that we often 2ust find 9ero! and the can only be reconstructed. This gi'es rise to a certain degree of uncertainty. Perha#s! we ha'e to reckon with the #ossibility of a de'elo#ment u X b. $ am#les: ZUV\^HI-=P@ S ZUV\H-U@; GF^HEH S GFHEH; ^DI^HE@ S [DI^HE@; Vh\^HE@ S Vh-HE@ 5also Vh\HHE@8. The e'idence com#rises R or + words in -^HE@. "t is found si times word-initially: e.g. ^.OEH S 1OEH; ^EHT\O[> S HT\O[>; note ^FE` S FE` 5[FE`8! where the latter forms could continue ]-[FE` S ]-[FE` with a #rothetic 'owel. Note further GU^E@ S G^P?U! which #erha#s continues ]G\-U^-! ]G\^-PG-. 3.3 (to#s interchanging with V5V8! with sto# | VS= or with V | sto# This kind of 'ariation is Luite com#licated. " distinguished no less than *7 5or e'en *38 different ty#esA+B. They may be re#resented as follows 5M Y consonant8:

a. labials *. M S Mt -. M S Ms N. M S sM 6. Mt S Ms 3. Mt S sM 4. Ms S sM 5` S V<8 < S <= <S` 5< S V<8 <= S `

b. 'elars G S G=

G S VG G= S a G= S VG 5a S VG8

,. Ms S ss R. sM S ss Rc. M S ss

a S VV VG S VV G S VV dentals

+. t S ss *7. t S st

= S VV = S V=

The analysis of these 'ariants is not easy! and " mainly #resent the data here. K Luestion that needs to be e #lained is why e actly s or t are in'ol'ed in the gi'en 'ariation. The most com#licated instance is 3b! where we find G=SVG. "n fact! the most com#licated #henomenon contains most information! and can be sol'ed best. "n this case! one e #ects a cluster with k! i.e. a consonant before or after the k. ;ne of the two e #ected clusters must ha'e undergone metathesis. Ks Greek did undergo a metathesis =G X G= 5and no metathesis of VG or a8! we may assume that #recisely this #henomenon was o#erati'e here. Thus! for an earlier stage we may reconstruct an interchange VGS=G. This interchange can be easily e #lained by assuming a consonant! #robably unknown to Greek! which resulted either in V or in =. "n my inter#retation! this must ha'e been a #alatali9ed dental! i.e. StyS. &or instance! ^\VhOU S ^\hDOP was #robably ]amuty%ala! re#resented first as ]amus%ala or ]amud%ala! the latter yielding ]amu%dala. K less clear e am#le is Kskle#ios! who was called ,5I8VGOU<I@ or ,5I8hOU<I@. "t could be that the name was ]'tyklap-! gi'ing ]'.i)sklap- or ]'.i)d%lap- . "n the latter form! metathesis did not o#erate because ]]'%dlap- was not tolerated in Greek; the dental was then sim#ly lost. Needless to say! it often ha##ens that only one 'ariant is found. The strange feature or #honeme may also be dismissed altogether! as in DIG[H ne t to DVGE@ and DG=\EH. ;ne might su##ose that all 'ariants in this grou# are due to a #alatali9ed dental! but this is not e'ident! as consonant clusters are rather rare! and as there are no suffi es beginning with a consonant 5e ce#t n! r! etc.8. )e may be unable to determine what e actly ha##ened in each case. Ty#e 6 is treated by &ur. -4NN. (ince Pre-Greek did not distinguish 'oice and as#iration in sto#s! these often 'ary; so if we s#eak of kt or G=! this also includes reali9ation as ?T! such as in ^FE?TE@ below. "f we consider the 'ariation with labials! as in ptSps! it is clear that we are dealing with a labial followed by a dental. The dental could also a##ear as s! so it is clear that the #honeme concerned was a #alatali9ed dental! which " note StyS. This means that we are dealing with a grou# pty. "n the same way! with a 'elar we ha'e kty. The e am#le DIWTFU ne t to DI`FU is well-known and clear. &urn.e further gi'es hHU^<=E@b ?UOIHE@ 5J.8 beside hOU^`Eb ?UOIHE2 V=^U=E@ 5J.8 and com#ares <=OEH with /or. `OEH. Jis e am#le <=@ cooked ne t to `EH is less e'ident. Kmong the forms with a 'elar! there is no #roblem with ^FE?TE@ S ^FEaE@. The best known e am#le is 3F[?T[@ 5also 3FI?T[@8 ne t to }F[?V[@ on Kttic 'ases. " ha'e no o#inion on 3FI?THIE@; it may be a Graecisi9ed form! and in this case it is unim#ortant for Pre-Greek. (ee further the ethnonyms xU=O[-<=EI! xPO-<=P@! ~UOP-`E! qUD-`E and FUHI-`E. ;ther forms are less clear.

There may ha'e been series with three forms! with kt S ks! pt S ps and also k or p. " can only mention /FU?TE@ S ,FaP@ ne t to /FUhE@! and #erha#s! ne t to DIWTFU S DI`FU! the 'erb DW> 5together with D[`-8! for both cf. &ur. -4N. Kbo'e! we assumed that a labial or a 'elar could be followed by a #alatali9ed dental StyS. "f this is right! we can also #ostulate that this consonant 5labial or 'elar8 was followed by a normal dental. ;f course! this yielded pt and kt. " assume that the second consonant of this grou# 5the dental8 could ha'e been dro##ed! which yielded single p or k. This e #lains the ty#e <5=8EO[^E@ 5&ur. 378 and ZFh?E@ 5with #renasali9ation8 beside ZF?TE@ 5&ur. 3*8. " will shortly re'iew the *7 5*38 ty#es 5" call the labials *a! etc.! the 'elars *b! etc.8. #). ?8 may re#resent a single #honeme py! as we saw in %*. $ am#les: 5&ur. N*3ff.8: hH\<- S hH\<=- 5hH\<[=-8; GEO^ZUIHU S GEOZDUIHU; GZUOE@ S GZDP@; O<P S O<=U; without 'ariants note GFVVEWTEH! VUFF\WT[H. #H. 78 is most #robably e #lained like 3b! discussed abo'e 5so *b is a #art of 3b8. $ am#les 5&ur. N*+ff.8: FUGI@ S FG=PH; ^Eh> S ^E?T>; <[O[GH S V<O[G=E@; GUGU S GG=E@. (). F may result from ]pty. "t is remarkable that there is no -b. G S a! as a is un#roblematic in Greek.A*7B /). ? O B?C H. 7 O B7- %oth may re#resent ]typ! tyk. $ am#les: TVZP S TZI@ 5&ur. -+--8! ZGE@ S ZVG[FEI; ?OU S VGOUI; ^G[OOU S ^VGP 5ZVGP8; ^GE@ S ^VGE@; WG[OEH S WVG>OE@ 5&ur. -+3ff.8. 1). ?8 O FC H. 78 O P were discussed abo'e and may continue ]pty! kty; they may belong together with -a. $ am#les: DIWTFU S DI`FU 5&ur. -4N Knm. N8; ?UO\ZDIG@ S O\`E@ 5&ur. N*R! N-68; ^FE?TE@ S ^FEaE@ 5&ur. -4N Knm. N8. 3H. 78 O B7 was discussed abo'e. $ am#les: ^\VhOU S ^\hDOP 5&ur. N7* Knm. -8; DVGE@ S DG=\5EH8 5&ur. -,+! N*+8. 5). F O B?C H. P O B7. &ur. N+N sim#ly considered the interchange as due to metathesis! which! of course! is #ossible. ]sp! ]sk may re#resent ]typ! tyk. $ am#les 5&ur. N+N8: V<HTIEH S `HTIEH; VW@ S `UI; V?EH S a@; WEVGE@ S WEa@. 6H. P O BB. "f a re#resents ]kty! the k may ha'e disa##eared in other cases 5which did not gi'e a8 after which ]ty became VV. $ am#les: GFIa@ S GFIVV@ 5&ur. *N73+8; VZDU S a^ZU 5&ur. N*,8; =FUaUHU! =FaUHEH S =FUVUHEH 5&ur. -R4,-8; aOP S VOP 5VVOU! =TOU8; OaP@ S D\VV[@. 9H. B7 O BB can be e #lained #arallel to ,b: ]tyk X VG or! with loss of the k! ]ty X VV. $ am#le 5&ur. N778: FIV?E@ S FVVE@. ;). 8 O BB. This is the well-known element that yielded VV S ==. &urn.e does not discuss it under this heading! because he gi'es only one #honeme 5letter8 and its 'ariants; for instance! he discusses VG S G= under G S G=. The situation is also different here! as we are able to discern a distribution among the Greek dialects! and attribute the different renderings of these loanwords to dialectal de'elo#ments. (till! the fact remains that a foreign element was rendered in different ways! as with all other #henomena discussed here. &ur. -3N has the heading =! D! T S V5V8! d. " think this should be reformulated as = 5D! T8! == 5=T8 S V 5d8! VV! i.e. = with its usual 'ariants D! T; or the

geminated == 5with its e #ected 'ariant =T! which is the Greek form of geminated TT8! interchanging with V or VV. "f the d was AsdB! it does not fit in well. Ks to its inter#retation! it could re#resent single ]ty! which was rendered == or VV! or single V! = 5the 'ariant d would then fit in! but one would also e #ect a 'ariant V=8. $ am#les 5&ur. -3Nff.8: GI==@ S GIVV@! GF=IEH S GFVVEWTEH! ^F=HP S ^\FVHP! =[=OEH S V[=OEH! =OWP S VOWP! hDE@ S hdU@! V^>O[H S D^>O. " think that the #honeme rendered by VV! Ktt. == 5called the foreign #honeme or Fremdphonem8 was a #alatali9ed 'elar! which " write as ky! cf. %eekes )/+0 N, 5-77+8: *+*-*+,. This would be #arallel to the de'elo#ment of inherited 'elar | yod! which ga'e VV! Ktt. ==! as in W\OVV>! W\O==>. This inter#retation is confirmed by TOUVVU! TOU==U! where we ha'e a 'ariant DUOh?UHb TOUVVUH 5J.8. Jere we see that after the nasal 5#renasali9ation is well known in PreGreek8! the #alatal feature of the consonant was dro##ed. This resulted in a 'elar 5here reali9ed as an as#irate8. The 'ariant shows that we may be dealing with a 'elar in cases of VV S ==. )e can also com#are GEO^ZUIHU beside GEOZDUIHU! which had py; again we see that the #alatal feature was lost after the inserted nasal. There is a third re#resentation. )e know that the name of ;dysseus was O\VV[\-! O\==[\-. This means that it #robably had a #alatali9ed 'elar! ]ky. %ut we also find OIa[@ 5"byc. apud /iom. Gr. #. N-* 1! Jdn. Gr.! Plut.8! a form which was at the basis of Qatin 1li2es. This form was taken from a )estern Greek dialect! #robably /oric. Therefore! a third re#resentation of the foreign #honeme is -a-. #=). 8 O B8 may be from ]tyt gi'ing V= or! with loss of the t! ]ty X VV. $ am#les 5&ur. N7*ff.8: ZUOO>= S ZUOOUV=IEH; ^=F\OOE@ S ^V=F\OOE@; ^=I@ S ^V=Ua; <U=OP S <UV=OP. Ks we saw! it is 'ery difficult to determine what e actly ha##ened in each case; on the other hand! it is clear that almost all 'ariation can be understood if we start from 2ust a few assum#tions. 3.5 Ne )r O )Hi) O dent) sto%s- )HioMe )rs There is limited e'idence for 'ariation between 'elar and labial! between 'elar and dental! and between labial and dental! and between all the three classes 5&ur. NRRff.8. )e find: G S <! Z G S =! D < S = hSZ ?SW hSD WST ZSD ?SWST hSZSD

"t is remarkable that the 'ariants mostly agree in 'oice S as#iration. (ince e am#les of this #henomenon are not #articularly numerous! this may be an indication that the words concerned are not of Pre-Greek origin! but due to borrowing from a different substrate! for instance. $ am#les: G S <: GOHIEH S <OUH@ h S Z: ZFGUOEH S <UOEH; hO<> S ZO<>; ?OI@ S WUOIGFH G S =: VGHDP@ S V=HDP@ h S D: hOU=^EH S DUO=^EH < S =: ZU<UH[I S ZU=UH[I

Z S D: V^ZUOEH S VHDUOEH W S T: hH\WU S hHTE@ h S Z S D: hW\FU S ZW\FU S DW\FU. "t is tem#ting to assume labio'elars to e #lain these cases! but some cases may ha'e a different origin 5thus! ZFGUOEH S <UOEH could be due to dissimilation in the first 'ariant8. ;n the e istence of labio'elars in Pre-Greek! see abo'e on the #honemic system. 3.6. Dent) s O iQuids There are some instances of 'ariation between dentals 5including n8 and liLuids 5l! r8. This 'ariation is incidental. $ am#les 5&ur. NR,f.8: ). D S O: ZOUFE@ S ZDUFE 5&ur. NN7-,8! DWHP S OWHP! D\VV[@ S O\VV[@. Mf. Cyc. gen. dapu(-ri-to-jo Sda#hurinthoioS S OUZFIHTE@! GUO^IHTU S Cyc. ka-da-mi-ta. The interchange D S O and the fact that Qinear % has signs for da! de! di! etc. 5which Qe2eune e #lained by assuming a s#ecific! unusual sound 38 might #oint to a dental fricati'e . T S O: T<=U S O==U H S O: H=FEH S O=FEH H. D S F: VZDU S a^ZFUI H S F: ZO.?HEH S ZO.?FEH c. O S F: dPF@ S dPO@! GFZUHE@ S GOZUHE@! GF0^Ua S GO0^Ua. 3.9. Si&% e O !e&in)te $ ce#t for a few isolated cases! we find this interchange in H S HH! but more notably in O S OO. ;n = S == and V S VV see abo'e sub 3.3. Mf. &ur. NR4f. $ am#les: H S HH: HPTEH 5also =8 S HHPTEH 5also =8; =P^[H@ S =Z[HHU. "n this conte t! note the suffi -\HH-. O S OO: ZUO5O8H; T\O@ S T\OO@; V<O[TE@ S V<OOPaI dat.#l.; ^UGOP S ^G[OOU 5this #robably deri'es from PG ]-alya-). Note h[V5V8EH! VFVU S VFIVVU! and the case of ,THP S ,=T@ S ,==IG@. 3.;. B- O Kero )e discussed V S 9ero before consonant under s-mobile abo'e! section %6. Kn -s- from Pre-Greek is normally maintained. The only instances that " know of! where it may ha'e disa##eared! are 5cf. &ur. -6*8: VFI?E@! VFIVVE@ S FI?@ 5also -VGE@! -V?E@! -VVE@8; V\ZOOU@ S ZOOP@; VUhHP S My#r. hHU; VI<P S <U. Perha#s 4OO@ beside t[OOE belongs here! too. Knother instance could be <IEH! which is cognate with Qat. pirum which #oints to -pis- . 3.#=. K-C $- O Kero

There are instances where a 'elar or a dental may be absent in initial #osition 5&ur. N+*! and *N*3+8. /entals may also be absent in inlaut. $ am#les: G S 9ero: GHDUFE@ S HTFUa! GUOIHDE^UI S OHD>! Gh?HUI S h?HUI! GUHTOIEH S HTOIEH. h S 9ero: hHHE@ S HH@! but this form may be a late de'elo#ment. Ks an e #lana-tion! one could think of a u'ular q. = S 9ero: =h?E\FE@ S h?E\FE@! =hUHEH S 5hUHEH! =W\EH S W\EH 5with in Q(08; D S 9ero: D[O[DHP S "O[DHP 5also -8. Qoss of a dental in inlaut: H=><EH S H><EH! TE\O@ S E\O@! VDUFE@ S VUFE@. 3.##. 4-C D- O Kero H- and O- can also be absent 5&ur. N+*f8: HWTU S WTU 5also *-8. OUI`PF@ S U`PF@! OU^<HP S <HP! OU=^[H[U S =^H. Perha#s! it concerns #alatali9ed ny! ly! which are #ronounced 'ery light. 3.#(. Ret)thesisC shi't o' )s%ir)tion There are instances of metathesis. "t mostly concerns F! sometimes O. The consonant 2um#s to the other side of the 'owel or the consonant: GIFV@ S GFIVV@! GFIa@; =F^IHTE@ S =F^ITE@. Mf. [F^OUI S F[^OUI; F<Ia S <Fa; GDFE<U S GFDE<U; HTFUa S HFTPa. "n most cases! it cannot be determined what the original configur-ation was. "n a case like FZ>@ S [F>@! where Z may stand for 5or continue8 ! " would think that the was antici#ated. "t may concern an original rw. The cases of V< S ` and VG S a are discussed in 3.3 abo'e. (hift of as#iration is found in some cases: TFIhG@ S =FIh?@! TFUhHP S HDF?HP. "n the case of W=HP S <THP the metathesis seems to ha'e occurred in the later history of Greek 5%eekes -77N8. 3.#/ Second)r" %honetic deMe o%&ents #. )e may assume secondary #honetic de'elo#ments! either in Greek or #erha#s already in the original language. ;ne might consider:

ZD- X ZO-: ZDUFE S ZOUFEI. &or this case! cf. 3.,b D S O. ZD X ZF: ZDOOIEH S ZFOOIEH 5&ur. N7R8 hD- X D-: hDE<E@ S DE<E@ DH- X hH-: DHWE@ S hHWE@ G^- X ^-: G^O[TFEH S ^OUTFEH `- X V<-: `[HD\O- S V<EHDOPs (ee 3.3.4 abo'e.

`- X V-: `WU@ S V[WU; `==UGE@ S V==UGE@; cf. uU<W! tU<W.

(. U X E before \ in the following syllable. The a was #robably #ronounced a little higher before the u! and was reali9ed as AB! which resulted in E. $ am#les: aE\hhU X ahhIEH! GUOZP X GO\ZE@! ]VGUFUZ- 5GFUZE@8 X VGEFZ\OE@! DEFGHIEH for ]D5U8F\GH-. 3.#1 Other M)ri)tion There are a few instances of isolated and #u99ling 'ariation. " mention 2ust one! the word for night! where we ha'e `WU@! GHWU@! DHWE@! dWE@. " think that in some of these cases! the solution may be found in a cluster. Marian! for e am#le! allows an initial cluster kbd- . (uch clusters would ha'e been sim#lified in Greek. "n an inherited word! we ha'e the #arallel of Qat. pe!ten! Gr. G=[@! which is su##osed to continue ]pkt- . "f we assume a cluster ]kdn- in our e am#le! it may ha'e been reduced to kn- or! with loss of the first consonant! to dn- . Thus! the #rocess is the same as the reduction hD- X D-! see 3.*N abo'e. (uch 'ariant sim#lifications are ty#ical for loanwords. "n this way! we could connect two of the words; but " see no way to connect the other two. 5. NoJe M)ri)tion 5.# Sin! e MoJe s (ti&Hre) The 'owels show many 'ariants. " will discuss them in the following order: first a! then e and o; and within each of these grou#s first the short 'owel! then the di#hthongs! then the long 'owel 5and the long di#hthongs! but these hardly occur8. Note that a 'ariation S y is not re#eated under y.

#. the MoJe +. #). U S [ has R7 occurrences in &urn.es material 5N6,8. $ am#les: h?E\VU S h?E\VU! F\VE@ S F\VE@! hOIHTEI S hOIHTEI! dUG[O=@ S d[G[O=@! GUU=U S GUI=U@! G^<E@ S G^<EF! G?F\@ S Gh?FE@! VHD\a S V[HDEGP. #H. U S E. This interchange also occurs freLuently. &ur. NN+ mentions that he found R7 instances. $ am#les: aE\hhU S ahhIEH! FF>D> S FF>D>! hFZIEH S hEZFUI! 6<UOE@ S 6<EOE@! GZUa S GZUG=FU! GUOZP S GO\ZE@! O\GU`@ S OGE`E@. #c. U S UI 5&ur. NN4ff.8. $ am#les: GFUIWH@ S GFU<H@! VWPOE@ S UVWIE@! Oh^U=U S OUh^U=U. The I here is due to the following #alatali9ed consonant. #d. U S U\ 5&ur. N7-N,8. $ am#les: GUHUV=FEH S GHUV=FEH! ^HVIEH S ^HUVIEH; OEa S UOUa. "n the last e am#le! the \ is #robably due to the following labiali9ed #honeme lw. #e. U S >: GODE@ S GO0HUa. #'. UI S [I 5&ur. N3- Knm. 6! NN+ Knm. -8. $ am#les: GUIFU S G[IFU! G<UIFE@ S G<[IFE@! OUIU S O[UI. %oth UI and [I are due to the following #alatali9ed consonant. #!. U\ S [\ 5&ur. N3N Knm. 38. $ am#les: OU\GUHP S O[\GUHP! <=U\FEH S <=[\FEH; UOUa S [OGU.

#h. U\ S >! E 5&ur. N7*N-8. $ am#les: GUVUFU5@8 S GUV>F@! TU^U S T0^U! VUVUa S VVIG[@! ZUGUOEH S Z0GE@! GUOUFE` S GEOOFEZEH S GEOOFEZEH. #i. S UI 5&ur. NNR8. $ am#les: OTUFhE@ S OUTUFhE@! OPG> S OUIGd>! <hUHEH S WUGUHEH. #S. 7 S . $ am#les: O7DE@ 5O8DIEH8 S ODE@ 5ODIEH8.

(. the MoJe T. (). [ S U: see under U. (H. [ S I 5&ur. N33ff.8. $ am#les: ZO=\a S ZO=\[@! ZVGE@ S ZVGE@! D<U@ S Cyc. dipa! H=\ZEH S H=\ZE@! G[OOH S GOOIa! GIOOZU@ S G[OOZU@! G=[VE@ S G=IVE@! OVWE@ S OV<E@ 5W8. The e was not #honologically distinguished from i! and they were #honetically close. (c. [ S I S \ 5&ur. N36338. $ am#le: G[?F^E@ S G?FU^E@ 5GIhGF^U@8 S G?FU^E@. (d. [ S [\ 5&ur. **38. $ am#le: Fh[=E@ S FG[\TE@. (e. [I S UI: see UI. ('. [I S P 5&ur. NN+ Knm. -8. $ am#les: G[TIEH 5?[=IEH8 S GTIEH! ?[IFU^@ S ?PFU^@. (!. [\ S [: see [ S [\. (h. [\ S U\: see U\. (i. [ S P 5&ur. N3R6-8. $ am#les: H\V=FEH S 5H\V=FEH! ^F^[FE@ S ^F^PFU! `G[OEH S ZGPOEH! ^D[U S ^D[U 5^d[U8; gPOUhH[@ S g[OUhH[@. (S. P S I 5&ur. *,***68. $ am#les: ZO.=EH S ZO=EH! VG.HE@ S VGHUF! `P^TIEH S `I^TIEH.

/. the 'owel E. /). E S U: see U. /H. E S I 5&ur. *+*N,8. $ am#les: GEHE@ S GIHE@! ZFGUOEI S ZFGUOU! 9HEhOIH S HIhOIH. /c. E S \ 5&ur. N3Rff.8. $ am#les: OEHTE@ S O\HTE@! VGEOEZF> S VGEO\ZF@! VG=P S -G==U! G\DHIEH S GEDH[U! G\FVU@ S GEFV@! <F=UHI@ S <F=UHI@! =E<[EH S V=\<<[EH. E and \ were #honetically 'ery close! and not distinguished #honologic-ally 5cf. on [ S I8. /d. E S E\ 5&ur. N3+8. $ am#les: ZFGE@ S ZFEGE@! GEOE=U S GEOE=[U 5also -O\-! -O>-8. /e. E S > 5&ur. -,+8. $ am#les: hHE=FU S hH>=FU! GEOOFEZEH S GEOOFEZEH! WUV>OE@ S WUVEOE@ 5also -E\OE@8! :F\hh[@ S F\a! -hE@; ;V?E S V?P. /'. EI S \ 5&ur. *-,8. $ am#le: ?FU^UDEOUI S ?FUDU^OU 5GFU^OU8.

/!. EI S E\ 5&ur. N3R8. $ am#les: GEOE\=U S GEOEI=U 5GEOE=U8! `EDIEH S `ETP@s /h. E\ S \ 5&ur. *-7-+8. $ am#les: G=<E@ S hDE\<>! GFEHUI S hF-H@. /i. E\ S > 5&ur. *NN8. $ am#les: ^>GE^UI S ^E\Gd[I; OE<P@ S OZPa 5&ur. *6R8. /S. > S P. $ am#le: TF0HUa S HTFHP. /k. > S \ 5&ur. N7-N38. $ am#les: dhIE@ S dhhIE@! VV><E@ S #VV<E@! O>Z[> S O\Zd[IH. / . E S [. $ am#le: hEFh\FU S h[Fh\FU

1. I S \. There is some 'ariation between I and \! but " do not know how to inter#ret it. $ am#les 5&ur. N46ff.8: UV\^H> S UVI^H>; HTFVGE@ S HTF\VGEH; ZDPH S Z\DE; ZFIG@ S ZF\G@; dhUV=FEH S VhIV=FEH; GIH<[=EH S G\HE<[@; GZ[VI@ S GZIVI@; ^FVI<<E@ S ^F\<<E@.

3. \ S [. $ am#le: h\FhUT@ S h[FhUTE@.

The beha'ior of the di#hthongs may be summari9ed as follows: UI S [I U\ S [\! > EI S \! E\ E\ S \! EI! > Kll this 'ariation is understandable in terms of ada#tation of a three-'owel system. 4.-. Qong S short: ;ne may doubt whether Pre-Greek had a distinction of long and short 'owels 5see o*8. )e do find P and >! howe'er! but not 'ery often! and the latter has se'eral 'ariants. ;n the other hand! the 'ariations > S E and P S [ are not 'ery freLuent 5although in this case also the difference in timbre may ha'e been im#ortant! de#ending on the Greek dialect8. Variation between long en short I and \ is freLuent! es#ecially in suffi es: hT\EH S hTIU! GZ[VI@ S GZPVI@! TZI@ S TZI@! GF^HEH S GF^HEH! TFHUa S TFIHGP; `P^TIEH S `I^-fTIEH! V<EHD-fOP S V<EHDOIEH. Mf. GfFUZE@ S GUF^ZIE@ 5cf. GPFUW@8! W[HGd> S <PHPGd> decei'e; <hPH5-8 S <h[H5-8; hHE=FU S hH>=FU. There is some e'idence for short 'owel | MM alternating with long 'owel | M: e.g. ^GE@ S ^VGE@; qFVU S qFIVVU; see % * on -Ia! -\a. 4.N. (ingle 'owel S di#hthong: and 5'ice 'ersa8 [I S UI [\ S U\

There are se'eral instances where a di#hthong 'aries with a single 'owel. They can be found abo'e 54.*8. Cost freLuent is U S UI! but this is due to the effect of a following #alatali9ed consonant. )e further find U S U\! [ S [\! and E\ S \ and EI S \. "n two cases we find di#hthong alternating with a long 'owel: UI S ! [I S P. $ am#les were gi'en abo'e. 4.6. iising di#hthongss ielati'ely freLuent in Pre-Greek words are seLuences of a more closed 'owel followed by a more o#en one! seLuences that are not found in "$. They would be rising di#hthongs if they formed one syllable! but in fact we may ha'e to do with two syllables. $ am#les are: -[U-: V[hH 5VI-! V\-8 -IU-: ZU=IGP! TUVE@! TFU^ZE@! VUOEH! WIOP! WIUF@. Note VIhH 5V[-! V\-8 -I\-: \h -\U-: ZF\UOd>H! hUOEH 5h\[-8! GU^E@! <UOE@! <UHEH! V\UhF@ -\[-: h\OIEH 5hUOEH8! <[OE@ 5<\U-8

iemarkable! too! is the seLuence ->\- in <0\5h8a! ^>@. 4.3. (econdary 'owels 5or elision8 (ometimes! words show a 'owel that is absent in nearly identical forms. "t mostly concerns 'owels between a sto# and a resonant. "t is often not clear whether the #resence or the absence of a 'owel is secondary. (ee &ur. N,R-NR3. $ am#les: ZFh?IU S ZUFh?IU; DEFGHIEH for ]DF\GH- in V=F\?H-; VGFZEOE@ S VGFUWE@; GHdU S 5V8GH\dU; VGFEDEH S VGFDEH; =EHTEFd> S =EHTFd>; ,F[<\U S =F<\IU; GH` S GIH<[=EH S G\HE<[@; cEFZUH=[@ S cFZUH=[@. *. Ror%ho o!" #. Redu% ic)tion (ome forms seem to ha'e redu#lication! though we often cannot demonstrate this. Cost freLuent is #artial redu#lication! where only the first consonant and a 'owel are re#eated. The 'owel is mostly [ or I. $ am#les: ZZFUa; Z5^8ZFE@; hhhU^EH; hUhhOEH; hUhhFUHU; hhUF=EH; hhhO\^E@; GG\ZE@; VV\WE@ S tV\WE@ 5cf. VWE@8; ^[^UG\OEH 5also ^I-8; H[HPOE@; VV[OI5@8; VV\FU 5also -\FHU8; ^[^ZF@5s8; #erha#s GG\^E@; G?FU^E@ 5also G[-! G\-! GIhG-8; D[HDF>. Klso the names cGFE`; g[<FPTE@; I=UFVIE@; qO[h[@. )ith #renasali9ation we find =[HTFPDH! =[HTFHP 5cf. HTFPDH! TF0HUa8. "n these e am#les! " neglect the fact that there may 5or may not8 be #renasali9ation. ;ther redu#lication 'owels are found in: OOU^I@ 5cf. OUOU`8! GEGF@! #erha#s also hPh.OIa. "ntensi'e redu#lication in: ^F^EFE@ 5^EF^\FUU8! ^UF^UF\hP.

Core difficult to 2udge are hOhI@ ne t to hO@ 5#erha#s from ]h[-hO-! -hO-8! GFGU ne t to GFI@ 5if from ]G[-GF-! -GF-8. Klso e[^ZOUFE@ beside oOUFE@ 5cf. ^[^ZF@8! e[^ZO@ Y eOE@! also eI^UOO@. K com#letely different ty#e is #erha#s found in ^^Ua\@ 5cf. ^Ua@8! and #erha#s also ^U^ITD[@. (. Su''iUes (.# Introduction "t a##ears that most suffi es ha'e the same structure. They contain a consonant; if this is a sto#! it can be #renasali9ed! i.e. -Z- or -^Z-! -T- or -HT-! etc. The sto# has its usual 'ariants! like Z S < S W! etc.! although mostly one of these is #redominant. The suffi usually starts with one of the 'owels of the language! mostly U! I! \ 5we find [ or E only rarely! e.g. OEHTE@ beside O\HTE@8. Thus! we may find e.g. Uhh Ihh \hh; UHT IHT \HT! etc. K different structure is #resent in suffi es containing -H- 5mostly followed by a 'owel8 directly after the root-final consonant: e.g. GDHE@! <IVGHU! ^O\?HEH! W[HUGH@! VU=UFH@. "n this way! the grou#s -FH-! -DH-! -GH-! -^H- in Pre-Greek words #robably originated. "n the case of -^H-! we often find a 'owel again: -U^H-! -I^H-! -\^H-. The grou#s -^H- and -FH- are es#ecially freLuent. They are 'ery im#ortant! as they are found in $truscan! which for the rest shows little agreement with PreGreek; -^H- is found as far as in Ma##adocian 5see %eekes 4iOr 3+ 5-77-8: 66*f.8. Perha#s! the grou#s -UHH-! -IHH-! -\HH- arose in this way! too. ;ther consonants are found in suffi -initial #osition! too: e.g.! -F-! -D-! -h-! rarely -O-. $ am#les: `\DF@! GFITFU! <UHUhF@! WOUGFE@; VGU<FDU; OTUFhE@; HIhOIH. "t is often #ossible to determine to which series the Pre-Greek consonant belonged. Thus! -UIHcould render -any- " while -aly- seems to ha'e resulted in -UOO- 5or -[OO- with coloring of the 'owel8. Qikewise! -[IF- could re#resent -ary-. This thesis would be nicely su##orted by the segment -U\F-! if this re#resents -arw- 5e.g. UFEV?D[@ beside FUV?D[@! if this form had ]-arw- ). Mf. %* abo'e. Knother ty#e of suffi has V followed by a dental: GHUVTEH 5-V=FEH8! OUIV=FH or another sto# HTF\VGEH! UFEV?@! GUHHUZVGU; these forms may ha'e been #artly ada#ted to Greek suffi es 5-=FEH8. (ee below on the suffi -V=-. K form such as -[\=- is de'iating; we do not often find a di#hthong before the consonant. /oes it stand for ]-aut- from -ats Mf. -aiu- in OUIEH! where we may sus#ect ay or awy 5but it may be #art of the root8. (ee further section %*. Not seldom do we find a long and a short 'owel with a suffi 5Y consonant8! e.g. IT T! \G -G. "n the case of -F! one might again think of ury X uir! although ry is a rare #honeme 5like my8. (.( SurMe" o' the su''iUes "n #rinci#le! we find one of the three 'owels of the language followed by a 5#renasali9ed8 consonant: a! i! u | 5m8P! 5n8T! 5n81. The grou#s actually found are! in Greek letters 5forms in brackets are rare or less freLuent8:

*. aNM Uhh iNM uNM

-. U^Z

N.

6. 5UH=8

3. UHD IHD

4. UHT IHT 5\HT8

,.

R.

+.

Ihh \hh

\^Z

5\H=8

\HD

(o! we do not find: *. VN< and N. VNW! ,. VNG! +. VN? 5e ce#t for DUOUh?UH8.

"n the same way! we find 'owel | M. The consonant may ha'e the normal 'ariation: #lain! 'oiced! as#irated. K #alatali9ed consonant could color a #receding andSor a following SaS to A[B! which may also a##ear as [I. This #henomenon is often seen in languages with #alatali9ed consonants! such as iussian and "rish. Thus! we find -ary- re#resented as -UIF- 5-[IF- is also #ossible8. K #alatali9ed -ly- may be rendered as a geminate -OO-. "f a labiali9ed consonant followed or #receded an U! this 'owel may ha'e been #ercei'ed as 5an allo#hone of8 SoS. &or e am#le! -arw- may be re#resented as -U\F-! with antici#ation of the labial element! but also as -E\F-! in which case the U was colored. The suffi al consonant may be geminated; as there is freLuent 'ariation between single and geminated consonants in the language! there #ossibly was no o##osition. Vowels could be either short or long; in suffi es! a long 'owel was Luite freLuent. K long 5 was sometimes re#resented as >. -.N The material The e am#les are mostly taken from &urn.e! to whom " refer for details. )ords can also be checked in the #resent dictionary. Variants are gi'en in brackets. " added geogra#hical names 5TN8 from &ick *+73! and some more material! with references.

*. -UZ-5E-8 5&ur. *7,8: hFGUZE@! O5O8ZP@! VGOUZE@! ==O[SUZE@! GHHUZE@! GOOUZE@! ^V5V8UZEH! OU=FUZ@! ^==UZE@. TN cU==ZIE@ 5ihodes! &ick 6,8! cV=UZE@ 5Maria8. -. -Uh-: !F<Uh- 5cf. Mhantraine *+NN: N+,ff.8! O=Ua! EFUa. N. -Uhh-E-: VWFUhhE@. Na. -Uh?-: DUOh?UH. 6. -UD-: FUV?D[@! ^[^ZF@! V<\FD-. 3. -UT-E-: V<OUTE@! h\FhUT@! V<FUTEI. TN >FHTIEH 5$#idauros8.

4. -UI-S-[5I8- before a 'owel: There are words in -UIU S -[5I8U! such as hF\^U S hF\^[U 5also hF\^[U8 S hF\^UU 5note the hesitation in the accentuation8. " suggest that the suffi was -ay-.a)! which was #ronounced as A-6yaB or A-eyaB 5we saw that [I often 'aries with UI8. The s#eakers of Greek identified the suffi with Gr. -UI- or -[I-! but the -y- could also be lost. "n this way the three 'ariant forms can be e #lained. &urther e am#les are GEOEI=U S GEO5E8\=U! GEF?\FU 5GEFGDF\U in J. is #robably an error8; GD[IU S GD[U 5note the short U8! beside GD\IU S G>DU 5these are not entirely clear to me! but cf. ,^T\IU S ,^T[IU8. &urthermore! ]-ay-a is likely to be the same suffi as -[I? which makes feminine names! e.g. ,^OT[IU! gPH[O<[IU! WI^D[IU 5note that in Cyc. /pemedeja! the -j- is #reser'ed! cf. iui2gh *+3,: *33N8. ;f course! many #lace names end in -[IU: cUD^[U! cUOUF[IU! c[FH[IU! eD[IU! tG[O[FD[U! q[ZD[IU! etc. The final was often ada#ted to -U after the dominant ty#e! which is deri'ed from the ad2ecti'es in -UE@ 5see Mhantraine! Form. +*8: ty#e HUhGUP; cf. ZF\G=UU! DIFGUU! VIZUU. )e also find -[U used in nouns: DU\F[U! dUO[U! GE\O\Z=[IU. Nouns with -[E- are 'ery rare; we find: h>O[@! [O[@! GEO[H! HIGO[EH! V\W[@5s8! W>O[@. "t may further be found in <GUHE@ l ]-kay-an- " note the by-forms <hPH! <h[H-. %eside -UIU! -[IU! we may e #ect thematic -UI-E-; we find it e.g. in DFGUIEH! V<OUIEH! `IWUEH; hFU`UE@! ]VGUFUZUIE@ 5reconstructed by &ur. *4+8. ,. -UI58-E- 5see &ur. -NN--! -33N-8: Partly from -UIE-; it is often im#ossible to establish whether a form had a -- or not. (ee also 4. abo'e. $ am#les: G\OUEH! FUI@! ZUhUE@! ZUOUIH! DFGUIEH! OUIEH 5Cyc. era7 / rawo8! ^=UIE@! ^[VVUEH! VFUIEH; ,?UI@. TN ,V=\<OUIU 5&ick: 3R8. R. -UIZ-E-: TN g[FFUIZE 5Thess.8. +. -UIT-: TN t^UITU 5Thess.8! g[FUIT[@ 5Krc. deme8! c[OUIT[@ 5Thess. deme8! c\HUIT[@ 5Krc. deme8. *7. -UIH- 5&ur. *,***,8: GUIHU! -EH! ZEO=UIHU! hhhFUIHU! GEOZDUIHU 5also -\^Z-8! GEFWUIHU! ^FUIHU! V^FUIHU! =FUIHU. **. -UIF-5E-8 re#resents -ary-: G<UIFE@ 5also -[IFEH! -PFI@! -[FE@8! ^?UIFU. *-. -UG- 5&ur. *3R468: Z\F=GP! UOUa! ZU=IGP! DHUa S D0HUa S DEHUa! TFIHGP 5TFHUa8! TFUa 5also -Pa! -GE@8! GU\HGP! T-OGP! <IV=GP! WF^UGEH. TN Fa! -Pa 5Qac.8. *N. -UO5O8-E- 5&ur. -36-R! %eekes -77R8: FZUOOE@! UhTUO5O8E@! GEFDUO5O8E@ 5also -@8! <FDUOE@. TN cUV=UOU 5Phoc. source8! FVOE@! t=^WOE@ 5Krc.8. *6. -U^Z-E- 5&ur. *R68: DTFU^ZE@! TFU^ZE@! U^ZE@! GUF^ZU@! VFU^ZE@. *3. -U^H-E-: DG=U^HEH! DU^HE@! VWHDU^HE@. TN tDU^HE@ 5Mrete8. *3. -U^-E-: F=U^E@. TN cV5V8U^E@ 5Mos8! gFhU^EH! chU^E@ 5Qydia8! cU^EH 51ydon.8! >DFU^E@ 51ydon.8.

*4. -UH-E-: TN FDUHE@ 5JN Mrete! $lis8! ,<IDUH@ 5JN Thess.8! @FIDUH@ 5JN8! ,HDUHE@ 5&ick: *R8. *,. -H-: h[H=IUH. *R. -UHD-: VhHDP@. *+. -UHDF-: h[OUHDFH. TN OUHDFE@ 5&ick 3*8! ^UHD5F8E@ 5Pam#h.8! e\UH-D5F8E@ 5&ick: 3N8! EOhUHDFE@. -7. -UHTS=- 5&ur. *+*N3; -*4,*! -UH=- unless otherwise stated8: OZU@! VGHTP@ 5VGHTU@8! GIOOZU@ 5but G[OOZU=-8! GFZU@! WOUHTS=E@! /ZUH=[@ 5&ick: 4+! etc.8! eOUHTS=-! g[FUHTS=-! ~hUH=[@! cEFZUH=[@. TN oUZFH=IEH 5Mhios8. -*. -UHH-: =FUHHE@. --. -Ua- 5cf. -Ia-! -Ea-8: *^UaU! =FWUa\@! VUZaU@ 5also -G=-8! TN c\F=F<UaEH 5Mrete8. -N. -U<-E- 5&ur. -N3N*8: FHU<E@! hU\VU<@! ^HU<E@. TN e[VVU<IE@ 5Mrete! &ick: -68. -6. -UF 5&ur. *N6,38! mostly neuters: G=UF! GDUF! HG=UF! VGHUF! VWUF; ad2. ^GUF; anim. UF! D^UF 5gen. -F=E@8! cf. Cyc. dama beside duma. -3. -UF- 5&ur. -3,N48: VGUFE@! ZUVVFU! hDUFE@! hhhOUFE@! GVTUFE@! GVVUFE@! O[V?FU! WOUFU. Klso VDFE@s TN /<=UFU 5Mrete! Qycia8! g=UFU 5Qycia8! ehUFU 5&ick: ,38! ,OOUFU 5Mrete8. -4. -UV-USE- 5&ur. *3,3,8: GF<UVE@! GU^UV@ 5GZUIVE@8! <hUVU. TN cFZUVU 5Mrete8! gDUVU 5Cess.8! gUhUVU 5Thess.8. -,. -UVV-E-: TN +\=IUVV@ 5Mrete8! cF\UVV@ 5Mrete8! e\GOPVV@! UWIUVV@ 5&ick: N-8. -R. -U=-: <=P! 6OUG=P. TN cUFU=E@ 5Mrete8! eOU=E@ 5Mrete! &ick: -,8. -+. -U\F-USE- 5this may continue -arw-8: 58WUFE@! WOUFE@! 58^U\F@! hOU\FE@! TPVU\F@! GUVUFU 5-U@8! OV=U\FE@! <=U\FEH 5[\8. TN 3<DU\FE@. N7. -U?-: Z=FU?E@! G^ZU?E@! V[O?E@. N7a. -U`-: O\GU`@! VGIHDU`@. N*. -hD-: #erha#s <FIhDU Aad'.B. N-. -hF- 5cf. on -F-8: <UHUhF@! V\UhF@. NN. -[D-: TN H[DE@! qZ[DE@! #erha#s in qUG[DU^>H. N6. -[d-U: see below sub ,N. on -IV-. N3. -[IF-E- 5may continue -ery-" -ary-): Uh[IFE@! G<[IFEH! VZ[IFE@ 5VU<FDP@8; cZ[IFEI.

N4. -[O-USE- 5cf. the ne t8: ^\VhOU! VWD[OE@! ZFG[OE@! DF`[OU! 5<I8dW[OE@! =TOU! GZ[OU! #erha#s D\V<[^<[OE@. N,. -[OO-USE- 5cf. N4.8: GFV<[OOE@! Z=[OOU! ZDOOU! <=[OOU! <OOU. NR. -[^-E- 5&ur. *3*6-8: O[^E@! GEO[^E@! <5=8O[^E@ 5if not "$8! T[O[^H. N+. -[^H-5E-8 5&ur. *3*668: !^WI-GO[^HEH! cUF=[^HD[@. TN tO[^HE@ 5&ick: +38. 67. -[HH-U 5" wonder whether ny could gi'e HH8: =Z[HHU. Mf. ZOHHE@. Mf. Qat. 5from $truscan8 .doss-)ennus! 8orsenna. 6*. -[F-USE-: DIWTFU! VGFU 5also -PFU8! GUVV=[FE@. TN <O[FE@ 5Mrete8. 6-. -[=-E- 5&ur. **368: GUI[=@! GU<[=E@! ^V<[=EH! H<[=E@! =POh[=E@s TN \<=P 5Ktt.8! U\ fh[=E@. 6-a. -[\- as in nom. -[@: ZUVIO[@; se'eral PNs like ,?IO5O8[@! D\VV[@. 6N. -[\F-: see -U\F-. 66. -[\=- 5&ur. *,N! *R*,8: ZUVG[\=U! GFU=[\=U. 63. -PZ-USE-: GFPZE@. TN cHDPZU! HDPZU! EFFPZ@ 5all in Qydia8. 64. -PT-5E-8: TN g[<FPTE@! t[VFPTE@ 5&ick 4,8! cIGHPTE@ 5Pagas.8! gFHP@! -PT- 5Ktt.8. Mf. -UT-. 6,. -PG-! -P?- 5&ur. *++! -63,78: ZFPa! DHDPa! OZPa! <OPa! =FWPa! WOPa; G^Pa S G^P?U. 6R. -PO-E- 5&ur. **338: ZFGPOE@! ZFZPOE@! VWPOE@! ZGPOE@! GZPOE@! GZDPOE@! H[HPOE@! =F?POE@! WVPOE@. 6+. -H 5&ur. *,-**R8: =^H! ==UhH! U?H 5^WPH8! ZUO5O8H! DETIH! VVH! GU^UVH 5-UV@8! G\WH! V[IFH! V>OH! =UhH! =IZH! #erha#s DH. TN vPFH 5Mrete! &ick: -38! FEIdH! ,FUDH 5Mrete8. 37. -PH-: hOUZFHP. 3*. -PF: V<HTPF; _hOPFs TN gETPF[@ 5Mrete8! cTPFU. 3-. -PF- 5&ur. -76*78: ZDPFU! HDPFU! U`PF@! VGOPFEH! VGPF 5-FU8; ^ZPFI@! G<PFI@! O[ZPF@. 3N. -PV5V8-USE- 5cf. -UVV-8: TN eF<PVVU 5C Paros8! e\GUOPVV@ 5%oeotian! &ick R78; ,FDP==@ 5Ktt.8! >^P==@ 5Ktt.8. 36. -P=-5E-8 5&ur. *,-**R8: O5O8ZP@! GHP@! OZP@! ^VTOP@! =<P@ 5Cyc. tepa8. Perha#s also H5H8P=EH 5also -TEH! -VEH8s TN CV f P=U 5&ick ,*8. 33. -P==-: see -PVV-.

34. -P`-E-: TN _DP`@ 5$uboea8! ~UOP`@ 5Thracia8. 3,. -T-E- 5see Mhantraine *+NN: N4R! and cf. -HTE@8: ZFHTE@! GUHT@! V<O[TE@! ^HTE@. 3R. -TF-USE-: GFITFU! ^FUTFEH; JN qZPTFU. ;n -UTFEH see &ur. N7NN+: GH5H8UTFEH! V<OUTFEH. Mf. on -V=F-EH. 3+. -Z-: F\TZP. 47. -h-: ^UV=h-! <[^WIh-. 4*. -IhhSGS?-: V=OIhhU@! ZF\=IhhE! O^IhhE@! TFhhE@ 5also -IhG-! -Ih?-8! T0^Iha! -hhE@! OIhhE@. 4-. -D- 5cf. -IHD-! &ur. N-6,8: ZUOZ@! h[Oh@! GPO@! GFP<@! VWFh@. 4N. -ID-HU 5#robably a combination of two suffi es! cf. on -H-8: F?IDHU 5cf. FUGE@! -?-8. 46. -T-! -IT- 5cf. -IHD-8: hOI@! -T-! UhITUOE@! UhITE@! hOITEI! 6OTIE@! GOITE@. 43. -IG- 5cf. -I?-! &ur. --4*7-8: GOIGEH! GFHIGU! O58GFIGU! ^\FGP 5later 8! HFIGEH! VVIG[@. 43a -G-: EHIG[@. 44. -IO-! -O-: UhO>`! GEHOP! ^V<IOEH! ^UFOP! ^\V=OP! 5V8<U=OP 5-O-8 Y <UV=OP! V=FZOE@. TN tGUHDOP 5Mos8. 4,. -IOO-USE-: FhIOOE@! *^IOOU! FIOOU! VIOOU! ZIOOE@. 4R. -I^H-USE- 5&ur. -64,*8: ^DI^HE@ 5also --8! ^FI^HU! VDFI^HEH. 4+. -IH-USE-: GIHE@! <OIHEH! 5ZU8ZGIHEH! hEVV<IHEH! G=IHE@! aHU. TN eFIHU 5Qemnos8! tGIHE@ 5Myclades8. ,7. -H-5E-8: G^IHEH! <\=HP! P=HP! VOIHEH! WEaHE@! WEFHP. TN tUOU^fH-. ,*. -IHD- 5cf. -IHT- and -ID-! -I=-8: GZIHDI@! OIHDEH. TN cFU\VHD>H 5i8! gFIHDE@ 5Maria8. ,-. -IHT-5E-8 5cf. -IHD-8: UhIHTE@! V<HTIEH! OUZFIHTE@! O^IHT[@. TN cFIHTE@ 5$uboea8! cFIHTE@ 5&ick ,68. ,-a. -Ia-: GE=aI@! G\Ha[I@! VFHIaU. ,-b. -<-E-: TN }F<E@. ,N. -IV-USE-: F<IVU 5-[dU8! G=IVE@. TN qFIVU! c[DFIV@! cPWIV@ 5-VE@ Y -IVVE@! &ick -3! 4*8. ,Na. -IVG-E-: OTVGEH! ZIVGE@! ^UFVGE@! FVGE@ 5and 'ariants8. ,6. -I=-USE- 5cf. -ID-! -IT-! &ur. *4N8: ZFZI=E@! ZO5Z8I=EH! <FWI=EH. TN tZFI=U 5Mrete8. ,3. -I?- 5cf. -IG-8: FVI?E@.

,4. -GH- 5#robably a combination of -H- with a #receding consonant; see sub ,R on -H-8: ZUFGHU! DEFGHIEH! <IVGHU! WIDGHP. ,,. -^-: TN q=^E@ 5Maria8! g=^E@. ,R. -H- 5&ur. *N-438! where a #receding 'elar may become as#irated: F?HP! DU\?HU-! GFGHE@! GDHE@! G\O?HIEH! <[O?HP! VU=UFH@! =HEH S DHEH! `DHE@; cZUFHE@. TN cTHE@ 5Myclades8. ,+ -Ea- 5cf. -Ia-8: ^EFEa@ 5also -?T-8. R7. -E<- 5&ur. *7,8! often there is a 'ariant with -UZ-: O5O8E`! GUOUFE`! -<E@ 5-WI@8! GOOE`! VGOE`. TN cEF<P 5Thess.8! cUVVI<P 5Morc.8. R*. -EF- 5see also the section on word end8: ?EFU 5-\FU8! O<EFI@. R-. -EVV-U! -E==-U: TN }F^HEVVU 5Mhios8! ,dIE==PHE@ 5Qydia8. RN. -E\O-: WUVE\OE@ 5->OE@8s R6. -E\F- 5may contine -arw-): HDE\FE@! GZE\FE@! OIhhEFIEH 5also OE-! O\-8! <UOE\FE@! <HDE\FU! VhE\FEH! =h?E\FE@. TN q\GVE\FU 5Krc.! the oldest town of all; &ick: +N8. R3. -E\V5V8- 5&ur. *+,338: h?E\VU 5also -8! UTE\V5V8U 5also UD>VVU8! GDE\VU. TN ,GDE\VU! cPOEVU 5C cO>VVU8. R4. -<H- 5this may rather be a suffi -H- after a root8: T[F<HP! ^<HP. R,. -<=- 5this suffi #robably consisted of one #honeme py8: ^UF\<=H! <VV\5^8<=EH! VHU<=IH. RR. -F- 5&ur. *-6N,; -*34-8: ZOUhFE@! hOIhFE@! VhFUI; DUhFE@ 5Y Qyc. id9kres8. (ee also the suffi es -FH-! -Fh- and -hF-. R+. -Fh-: OTUFhE@ 5also -UI-! -P-8. +7. -FD-: TN cUFDU 5Maria8. +*. -FH- 5&ur. 6R*-4! -*34-8: GUFHH 5GFHUa8! G\Z[FH>! OI<[FH> 5also OIW-8! VG<UFHE@. )e also find 'ariants without -H-: VV\FHU S VV\FU! G\Z[FH> S G\^[F.HUI! VU=UFHD[@ S VU=UFD[@! GVIFHI@ S GVVIFI@. Therefore! the cluster #robably arose by addition of the suffi -H-. ote that -rnis found in $truscan and already in Ma##adocian 5&ur. 6R*-48. (ee also the suffi -F-. TN UOVUFHU 5Mrete8! qFHU! AOG\FHU 5Ket.8. +-. -VU: There are se'eral words in -VU: D`U! D`U! G`U 5GE`U8! G`U 5G^`U8! #erha#s O`U. +N. -VG-: FVGE@ 5-?-! -VV-8. +6. -V- 5&ur. -36-,! in se'eral cases this does not seem to be a suffi ! but rather the end of a root; cf. on -UV-! -IV-! -\V-8: OVE@! GZUIVE@ 5also -UV-8! ^VE@! <VE@! WFVE@. TN gFUHVE@ 5Mrete8. +3. -VV-: G<UVVI@! G\<FIVVE@! VFIVVU.

+4. -V=-: OZUV=E@! T[^IV=- 5cf. Cyc. temitija S timitija8! O[<UV=! <OU=HIV=E@. TN cF\V=E@! UIV=@. +,. -V=F- 5cf. -TF-8: OZUV=FEH! D<UV=FEH 5also O-8! H\V=FEH 5also 5-8! dhUV=FEH! OUI5V8=FH! VhIV=FEH. +R. -=-: VWUO=E@! =FUG=E@! WOUV=EH. ++. -==- 5see 3.3 on == S VV8: G\FI==E! <FEG==U; F[WU==U. *77. -\Z-: H=\ZEH! TF\ZE@! H=\ZE@ 5also -E\Z-8! VOO\ZE@! VGO\ZE@ 5also -^-8! VGEOZFU 5-EZ-8; cf. ?TEIZ@. *7*. -\hh-: OF\ha! <V\hhE@! V<.O\hh-! WF\ha. *7-. -\D-: ^@! -DE@! ^@! GEF\D@! <POU^@! ?OU^@. *7N. -\DHU: TN cUO\DH 5Mos8. *76. -\T-! --T-: hH-@ f ! OG\TE@. *73. -\I-U: h\IU! GD\IU; =F<\IU. TN cIHD\U 5Mrete! also cIHDP! &ick *R! -68. *74. -\G-: ^<\a! D\a! VGUFD^\G=E@. TN F\a 5Qocris8. *7,. --G-: DED\a! G.F\a! -\GE@! GUFG5G8P! VHD\a. *7R. -\O- 5&ur. -73*68: FZOP! DG=\OE@! GUHTOP! GF>ZOP! ^I^UG\OEH 5also ^[-8! VWHD\OE@ 5also V<-8. *7+. --O-: 5V8GEFDOP! VWEHDOP 5also V<-8. TN cUFDU^OP 5Cess.8. **7. -\OO-: tZ\OOU. ***. -\^-: hFV\^EH! hhhO\^E@! -TO\^HE@. TN cF\^UI 5Mrete8. **-. -\^Z-: T\^ZE@! GO\^ZE@. **N. -\^H- 5cf. &ur. -6N44 on -umn- in $truscan and Ma##adocian8: UV\^H>! Vh\^HE@. TN +=ST\^HU 5Mrete8! qF\^HU 5Qocr.8. **6. --H- 5see also the suffi -\HH-8: ZT\HE@! VIhHP 5cf. -\HH-8! Oh\HE@. TN ~F=\H 5Mrete8. **3. -\HH-: Vh\HHE@! xG=\HHU. Mf. on --H-. **4. -\HD- 5cf. -\HT S =-8: o[F[GHDUI. TN cU^HDIE@ 5ihodes8. **,. -\HTS=-: ZO\HTEH! O\HTE@; o[F[GHTS=UI. TN G\HTE@ 5|RR8! F\H@. **R. -\a-: TN OW\aE@ 5Kthos8. **+. -\<-: BVV\<E@ 5older VV><E@8! ^FV\<<E@! EV<P! =EO<P.

*-7. -\F-: V\FE@! ?\FU 5also -EFU8! dW\FE@! ^U\G\FH! OT\FE@! H\Fd[=UI! VU=FE@. TN CO\FE@ 5Mrete8! h\FU 5%oeotia8! NV\FE@ 5Mos8. *-*. --F-: hG\FU! Hh\FE@ 5also H-8! hW-FU! OW\FEH! <OP^\F@! <=\FEH. *--. -\V- 5on -\V=FEH see -V=FEH8: F\VE@. *-N. -\=-: <IH\=@! HP<=IE@. TN qUhIH<\=EH 5Mrete8! cEOO\=@ 5Mrete8. *-6. -\W-: GO-WE@. *-3. -\?-: Z5V8=F\?E@. TN eV\?OEH 5Qemnos8. *-4. -WT-: GFEVVWTEH! OGUWTEH! ^OEWTE@! H5V8GUWTEH! VUFF\WT[H. *-,. -W- 5on -UW-E- see Mhantraine Form. -4N8: FhOEWEI 5also -IO-8! ^UV=FEW@ 5also -<@8! VFIWE@! VV\WE@. *-R. ->G-: TN cET>GDUI 5Ktt.! &ick ,78. *-+. ->O-: <EWOIE@ 5s8! WVG>OE@. TN c^>OE@ 5Myclades8. *N7. ->^-: ZF>^E@. *N*. ->H- 5&ur. N7NN+8: OG\H! 6IH! VUHDH! VIHDH! V?UDH. *N-. -><- 5a 'ariant is -E\<-8: T\^UO>`! UhO><-! GIH<[=EH 5G\HE<[@8. TN }F><@ S U 5Mrete8! cUVV<P 5$#irus8. *NN. ->F- 5&ur. -**378: ?F! ^H>F! ZI=>F! ?F! O[=>F. TN gO>FE@ 5Mhalc.! &ick --8. *N6. ->VV 5see -E\V5V8-8: TN xIFW>VV@ 5$uboea8! gID>VV@ 5Maria! &ick -48. *N3. ->=- 5&ur. -RNRN; NR6*N-8: VGUOUZ=P@! -GU\D>=H! GZ>=@! GF^Z>=EH! EV<=P. TN v[V<F>=E. N. )ord end )ord end #ro'ides an interesting situation! as some original finals of the Pre-Greek language may ha'e been #reser'ed. ;f course! Greek endings must be remo'ed! notably -E@! -EH. Thus! -IEH! -\EH may often continue original -I! -\: cf. Cyc. dunijo ne t to duni. The words in -HTE@ ha'e re#laced almost all of those in -HT- 5as in F\HT-8. N.*. words ending in a 'owel ). -+. K short -U can only come from ]-ya l ]-ih( in inherited Greek words. "n all other cases! we may be dealing with a Pre-Greek ending -a that was originally short. "t is often difficult to see whether -U is short or long; the material reLuires further study. $ am#les: ZUFGHU! hHHU! hFDU! DUO=U 5s8! UGE\DU! ZDPFU! GUFU! GEF-HU! GV=IOU! G=FU! OUFU! ^E\?F! FDU! V=UhHU! WTU! ZFEGU! hDU! hOU! DaU! DFDU! DUOh?U! T<=U! ^DU! ^IaU! V==U! VFHIaU! VEU! GDFE<U 5also ?-8! etc. Note forms in -\U! like F\U! and in -[HHU. Note! further! VUOU^HDFU! VGEO<[HDFU.

&or words ending in -VU! see the list of suffi es.

H. -V. "$ words 5neuters8 in -I are 'ery rare in Greek. $ am#les of Pre-Greek words in -I: dG=I! GFI! VHU<I! =h\FI 5=UhFIEH8! GUF. )e may assume that many words ending in -IEH! -\EH originally ended in -I! -\. &inal -I@ is freLuent! too.

c. -,. ZUF! GHD\! ^0O\. &or -\EH! see the foregoing. &inal -\@ is also found se'eral times: hD\@! F<\@! =FWUa\@! ZT\H! ZO=\@! -^HD\@! ^^UFG\@! ^>@! <POU^@! <\@.

d. -T,W. Though the ending may also be inherited from "$! in many words it is clearly of Pre-Greek origin! e.g. ZUVIO[@ 5Cyc. qa-si-re-u8! ,?IO5O8[@. " withdraw my considerations in &( 1ortlandt on this #oint.

e. -G. GET! GFT>! ^E=! ==! ~[OO. The suffi also makes feminine names in -: qP=! tU<W. "t is usually assumed that the original inflection of all words in - deri'es from stems in ]oi-; " assume that Pre-Greek words secondarily 2oined this inflection. )ords in ->@ are masculine: ?UFH5@8! 'F>@; eH>@! O>@. /.(. Jords endin! in -A ). -+A. VUF5EH8s! G?H=UF! ZGUF! O^UF! GDUF! HG=UF! H0GUF! V[OUF.

H. ->A. $ am#les: DIhF! GGF! GUGGF! G^^EF 5also -^<-8! <VEF.

c. -,A. $ am#les: `T\F! 5/or.8 ^F=\F.

d. -GA. $ am#les: ?>F! ?F! GO>F! `T>F. N.N. words with a nom. in -a or -`. ). -P 5stem in -G-8 is found Luite often: -Ua: ZUa! GFHUa! ^ZUa! HDFWUa! HTFUa! ZV=Ua! ^OZUa! VUVUa. HUa has a stem in -G=-. -a: WHUa! a! VWa -Pa: DHDPa! ZFPa

-Ia: DDIa! HTFIa! <Fa! GOIa! F<Ia -Ea: ZZFEa! ZFFEa -E\a: ZFEa -\a: ZO=\a! hF=\a; gHa! t=a.

Note acc. ZOO[GU; acc. ZFhGU.

H. -F: O=FU`! OUOU`! OI`! GFI`! Uh<E`! GOOE`! ^FE`! h`! ^H>`. Conosyllabic: ?F^`.

1. Jords in -4- ZUOOH! GUFZH; hH. GHD\H! ^VV\H! T\H[@.

3. Jords endin! in -+W (-ste&s)- Z@! ZOU@! TFU@5s8! ^WU@! VGUO<U@! VG>HDU@! ==Uh@! ZUD@ 5ZU=@8! ZUVG@! ZU@! GUOUZ@! GUV@; ,T^U@! ,Vh[OU=U@. )ith a stem in -UH=-: OZU@ 5-H=-8! O\GZU@ 5-H=-8 etc.; see the suffi section. )ith stem in UD-: ?F@! ZE\HI@! <FP^H@; see the suffi section. D. $he unit" o' Pre-Greek The material itself shows that we are largely dealing with one language! or a grou# of closely related dialects or languages. ;f course! we cannot demonstrate in each and e'ery case that the words that are non-Greek belong to this same language. The bulk of the known non-Greek words! howe'er! seem to fit the general #icture of the Pre-Greek substrate. &or e am#le! G=T\ZE@ S GV\^ZE@ does not only show the element VV S =T! well-known from geogra#hical names! but also the suffi -\Z- with #renasali9ation. The #air GFVVIEH S GF=IEH also shows the element VV S =! but GFVVEWTEH has a suffi added that is also ty#ical for this language. The word DUOh?UH ne t to TOUVVU 5-==U8 again has the suffi VV S ==! but also #renasali9ation. V=OIha S V=OIha has both the ty#ical 5#renasali9ed8 suffi -Ihh- and 'ariation U S E. "n ^FIHTE@ S V^FIHTE@ we ha'e the s- mobile and the well known suffi ! while ^F^I@! -ITE@ has the 'ariant without #renasali9ation! and V^FIhh[@ has a different Pre-Greek suffi . "n 5^8ZF\==EI S ZF==E@ 5ZFVVE@8 we ha'e a combination of a #rothetic 'owel and #renasali9ation. ;ther languages may well ha'e e isted in the area. Thus! it is not certain that Jierogly#hic Cinoan re#roduces the same language as Qinear K. &urther! $teocretan has not yet been connected with other elements and seems isolated. Knother matter is that 5non-"ndo-$uro#ean8 loanwords from old $uro#e may ha'e entered Greece! cf. %eekes -777: -*-N*. Coreo'er! these may ha'e already been ado#ted in Pre-Greek! as is suggested by FZIHTE@! which has a Pre-Greek suffi ! but a root which is attested 5with some

'ariation8 in other $uro#ean languages. (ometimes! elements from other "$ languages may also ha'e been ado#ted at a 'ery early date! such as <O[G\@. Jowe'er! " think that it is methodologically more sound to start from the assum#tion that non-Greek words are Pre-Greek. ;nly when there is reason to assume that they ha'e a different origin! should we consider this o#tion. E. Pre-Greek is non-Indo-Euro%e)n ;ur knowledge of "ndo-$uro#ean has e #anded so much! es#ecially in the last thirty years 5notably because of the laryngeal theory8 that in some cases we can say almost with certainty that an "ndo$uro#ean reconstruction is im#ossible. K good e am#le is the word hHTE@. "n order to e #lain the -a- of this word! we need to introduce a h(. Jowe'er! a #reform ]%nh(dh- would ha'e gi'en Gr. ]hHT-. ;ne might think that assuming ]h(e would remedy the #roblem! but ]%nh(edh- would yield ]hUHUT-. The conclusion is that no "ndo-$uro#ean #roto-form can be reconstructed! and that the word cannot be of "ndo-$uro#ean origin. Knother e am#le is the word GFP^H@ o'erhanging bank! for which a connection with GF^U^UI to hang 5u#8 used to be e'ident. Jowe'er! we now know that most long 'owels go back to a short 'owel #lus a laryngeal! and that long 'owels cannot be #ostulated at random. "n this #articular case! there are sim#ly no concei'able formations that would contain a long root 'owel. This mor#hological ob2ection is strengthened by the fact that there is no trace of the e #ected root-final -U- l ]-h(- 5as in GF[^U^UI l ]kremh(-). Positi'ely! one can say that landsca#e terms are freLuently borrowed from a substrate language. The ine'itable conclusion is that the word is Pre-Greek.
A*B (ince 1ui#er was my su#er'isor as well! " was acLuainted with the book from the 'ery beginning 5see my re'iew in $in%ua N4! *+,38. A-B Note that &urn.e often adduces new material that is not mentioned in the current etymological dictionaries! mostly glosses from Jesychius. ANB K #oint for further study is to establish how far to the east such related names can be found. "t is my im#ression that these names can be found as far south as Milicia. A6B ;f course! it could be due to the fact that a different distinction was #resent in Pre-Greek 5like fortis S lenis! found in most Knatolian languages8! but no ob'ious distribution #ointing in this direction can be discerned in the material. A3B Note that " distinguish between #alatals of Pre-Greek origin! which are indicated by a su#erscri#t y 5e.g. ky8! and #alato'elars of "ndo-$uro#ean origin. A4B Klthough " assume that 'oice was not distincti'e in Pre-Greek! " do write d- in this case! because only D- surfaces in Greek. )e must a'oid losing information #resent in the Greek forms. Thus! my notation of Pre-Greek forms is heuristic to a certain degree! and not always consistent with the #honemic system " tentati'ely reconstruct here. A,B ;n #renasali9ation! see %3.-. below. Ks an alternati'e! an "ndo-$uro#ean etymology starting with the root ] h(em%hto tie! betroth! can be offered; see the dictionary 5although " #refer the analysis gi'en here8. ARB )e also ha'e to recall the instances of G?! <W! =T 5see abo'e8. A+B (ince the word `I==GIEH S <IV=GIEH #istachio is #robably an oriental loanword! there are no good e am#les for an interchange VV S V=. A*7B " ha'e some difficulty with &urn.es section " 5&ur. N-N-N-+8. Cy conclusion is that a 'ariation M S MV cannot be #ro'en! although some instances remain difficult to e #lain otherwise.