Sie sind auf Seite 1von 77

American Philological Association

The Prehistory of the Greek Dialects Author(s): William F. Wyatt, Jr. Source: Transactions and Proceedings of the American Philological Association, Vol. 101 (1970), pp. 557-632 Published by: The Johns Hopkins University Press Stable URL: http://www.jstor.org/stable/2936072 . Accessed: 02/03/2014 12:52
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Philological Association and The Johns Hopkins University Press are collaborating with JSTOR to digitize, preserve and extend access to Transactions and Proceedings of the American Philological Association.

http://www.jstor.org

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

THE

PREHISTORY

OF THE

GREEK
Jr.

DIALECTS

WILLIAM

F. WYATT,

Brown University

oftheGreek anddevelopment in theinterrelationship Thoseinterested importance.The of cardinal withtwo facts dialects are confronted intelsystem, one linguistic first factis thatGreekwas one language, of how it regardless "EAArwvEs ligibleto all who called themselves
* The works in this I havemost referred to which frequently paperare: C. D. Buck, TheGreek Dialects (Chicago, ofAncient I955); R. Coleman, "The DialectGeography " Transactions Greece, ofthe Philological Society I963: 58-I26; E. Risch," Die Gliederung I2 (I955) 6I-76 (reprinted dergriechischen Dialekte inneuer Sicht," Museum Helveticum in G. S. Kirk, Ed., TheLanguage andBackground of Homer I964]). 90-io5 [Cambridge References to these works preceded by a #refer to their tables ofdialect features (Buck are to page number. facing page 374, ColemanI07-I3, Risch75): all other references of include: Other important works on thedialects J.Chadwick I963, "The Prehistory theGreekLanguage," Cambridge Ancient History 2.39 (I963); Chadwick I956, "The Greek Dialects and Greek Prehistory," Greece andRome N.S. 3 (I956) 38-50 (reprinted in Kirk I06-I8); W. Porzig, "Sprachgeographische Untersuchungen zu den altgriechischen Dialekten",Indogermanische Forschungen 6i (I954) I47-69. For general information on the dialects:F. Bechtel, Die Griechischen Dialekte, 3 vols., (Berlin I921-1924); Thumb-Scherer, A. Thumb,Handbuch der griechischen Dialekte, Vol. 2, I havein general revised by A. Scherer (Heidelberg, citedinscriptions after E. I959). Schwyzer 1923, Dialectorum Graecarum Epigraphica Potiora(LeipzigI923); Exempla IG, Inscriptiones Graecae; SGDI, H. Collitzand F. Bechtel, Samnmlung der griechischien (Gottingen I884-19I5); Lesbianlyric texts are after PLF, E. Lobel Dialekt-Inschrifte and D. Page (Eds.), Poetarum Lesbiorum Fragmenta (OxfordI955). For grammars,I havereferred to E. Schwyzer1939, Griechische Vol. I (Handbuch der AlterGranmatik Greekand to Docs., M. Ventris tumswissenschaft II.i.i, Munich I939) for classical andJ.Chadwick, Documents in Mycenaean Greek (Cambridge I956) andE. Vilborg,A Tentative Grammar ofMycenaeant Graeca etLatina Greek (Studia Gothoburgensia 9, [G6teto GriechischesEtymologisches borgI960]) for Mycenaean Greek. H. Friskrefers Worterbuch (Heidelberg etymologique de la i96o-), and P. Chantrainei968 to Dictionnaire to refer to someother langue grecque (ParisI968-). I havealso had occasion worksof mine: Wyatt i964, "Arcado-Cypriote I7o-82; Wyatt 1971, Kas," Glotta42 (I964) TheGreek Prothetic Vowel, tobe published as a monograph bytheAmerican Philological and GreekDialectology," to appearin Studi Association; Wyatt 197Ia, "Sonant/r/ micenei edegeo-anatolici.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

558

WILLIAM

F. WYATT,

JR.

[I970

Thessalians Laconians, or ofAthenians, in themouths was pronounced used-REV,another-HEsone Greek ofwhether and regardless Cretans; in thefirst person plural active oftheverb;or whether for"take" one fact isthat different brands or The second saida(pE'O, another JypE'o. circumgeographically varieties of Greekwere spokenin various thesame areasor within in various andnotrandomly localities scribable of different by one and thesamepersonor by members area either onlyimperceptibly, differs, sometimes of society. Each dialect strata forms of Greekthat local thevarious and it is from itsneighbor, from of the theinterrelationship concerning impressions we form our first stagesof the dialects. We have no reasonto supposethatearlier probably differences dialectal regard: later inthis from differed language localandslight. werealways all hypothesis: of thesefactswe deriveour first From the first of form non-attested from an earlier, (G) derive ofGreek attested forms (PG). This G whichwe are in the habitof labelingproto-Greek probably uniform-it mayor maynothavebeendialectally language wereto variations dialectal assume that whatever wasnot-but we must and were not it arosewithin be foundwithin it, thatis, in Greece, of a laterformof Proto-Indointo Greeceby speakers imported andnotimagine, be veryclearaboutthis (PIE).' We must European to do, a tempted were sometimes of scholars as older generations intoGreece:theGreekdialects (IE) incursions ofIndo-European series Since we are aroseon the soil of Greece(ChadwickI963: I0-I7). to or right a PG, we haveno reason and reconstruct forced to assume or only G aroseas an amalgam ofunrelated as somehave,that suppose, related languages.2 distantly
I A corollaryto thisobservation be carefulto is thatwe must in our reconstructions *hepta"seven" forPG and PG and not PIE. Thus we want to reconstruct reconstruct "so much" from not *septm;and *pansa "each," not *pantya;and (probably) *totsos PIE *totyos. On this last cf. my article "Greek Names in -uosoorrosg-" (Glotta46 and *petsj "I [I968] 8-I4) in which I try to show that PG already had both *totsos had experiencedpalatalizationbeforethe Greeks cook," and hence that *tyand *k(w)y of PIE variety distinct of Greekwas a dialectally arrivedin Greece. That is, theancestor proleptically, before its arrivalin what was later to become itshome. We may refer, to thislinguisticsystemas the Hellenic dialect of PIE, or more simplyHellenic PIE. sources, 2 Such an amalgam, and thepossibility thatlanguagescan arisefromdisparate is not usuallyassumedin modernlinguistics. NonethelesstheItalianschool,particularly V. Pisani, has argued along theselines. Cf. Pisani in RhM 98 (i955) i-i6 (= Saggi di

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. TOI]

THE PREHISTORY

OF THE GREEK DIALECTS

559

The second fact causes ustohypothesize, andthis isonly anhypothG dialects esis, that thevarious inGreece, notonly buton developed, thevery spotin which werespoken in classical times. Thatis they to say, a more or lessuniform PG developed localvariants, someof intime which extended influence toneighboring areas andwhich their in theattested resulted dialectal informal diversity. We can derive forthis from support thefact hypothesis that Coleman's map(125) interms ofa mere shows that ofdialectal features weighting linguisticalsoto be geographically allysimilar dialects tend close as well. We shall ofcourse havetomake several Dorian obvious exceptions: speech isclearly inthePeloponnese, a relatively recent event andtheDorians tothe north oftheGulf must be removed ofCorinth the during early of dialectal stages and cannot at that timehavebeen development to Arcadian; Boeotians contiguous havebeenin maynot always Boeotia;Cypriotes certain mayhavedeveloped traits while linguistic stillresident on the mainland; Lesbians linmayhave innovated in Northeast guistically somewhere Greece than on theisland rather of Lesbos;and Jonians all their maynothavedeveloped linguistic traits in theCyclades andon thecoastof AsiaMinor. We know, that the seainantiquity however, united rather than so these divided, lastpopulation shifts-if suchthey were-however important to the people involved, mayhavebeendialectally To be on insignificant. the safe itisbest torephrase side, though, ouroriginal andto statement assume that thegeographical relation oftheG dialects to eachother was muchthesamein classical times as it was in pre-Mycenaean antiquity, whether ornotthe dialects atthat time intheir were spoken classical locations. is no need,though There is in general there no linguistic harmin so doing,to assume prehistoric migrations of Greeks within Greece. The usual question oneasks in G dialectology is: howdidG split intotheattested dialects, andwhich dialects aremore closely related tooneanother? In other words, howdo we go about constructing a tree likethat in Chadwick family given (I956: 40), which shows a first into split Eastvs.WestGreek, then "Achaean" vs. Attic-Ionic, etc? Theusual method ofanswering these questions istomake a list
linguistica storica l8i-98 [Turin i959]) and PP I4 (I959) 8T-86; and V. Gusmani,Glotta 44 (I966) I9-25.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

56o

WILLIAM

F. WYATT,

JR.

[I970

in Risch(75) or Coleman(107-I3) and see which likethat of features in withthemostfeatures whichfeatures.Thosedialects share dialects which,incicommonwill be more closelyrelated. This method, grave encounters form, simplified I have givenin unfairly dentally, Atticwith shares dialect, Lesbian(L), an Aeolic (Ae) difficulties. of *-ti in L O'potat, I ObE'povat, butwith Ionic (Al) the assibilation as in before of *kw (T) thelabialization Boeotian(B) and Thessalian /e/ related? Or is L moreclosely < *penkwe.Withwhichdialect 7TEU7rTE thoughordi(ArC) dialect, (C) an Arcado-Cypriote again,Cypriote for (e.g., Arcadian as changes same the showing narily Kacg rKai), shares of (7TlcEL instead before of *kw thelabialization withtheAe dialects /e/ Ae? Arorwith with Does C belong TE'7-El-Schwyzer I923: 679.I2). withAe, and C belongs onefeature in this is that ofcourse The answer of thatL is closerto AI thanto T and B in theassibilation similarly of *k-. The traditional *-ti, but to Ae and C in the labialization does not allow for though, relations, of lookingat dialectal method splitirrevthatdialects only,but demands in one feature similarity thenone non-Ae characteristics, shows If one say L, dialect, ocably. as Risch(71) for dialects, ofneighboring influence to assume is forced L *-ti> -si to I influence. doeswhenhe attributes instance and and unnecessary, is in general unrealistic Such an assumption approachto the of the traditional pointsout the deficiencies simply with someshared offeatures, collections aremerely problem. Dialects e.g., to I, is to indulgein a others not, and to refer, otherdialects, showsd > -1evenafter ofG which that form wayofsaying: shorthand "four"is rE'uuEpEs!;inwhich*-ti develops E t p; inwhich thenumber to occuronlyin I, the feature thefirst happens to -si, etc. Of these not also in ArC, A, L. Hencewe should secondalso in Ar,thethird which how the features but rather developed, ask how the dialects arose. And ifwe wantto knowhow thedialects thedialects define by are shared features whatinnovative ascertain arose,we mustfirst and in what temporal sequencetheyarose. In other what dialects the whichdivided those (features) isoglosses wordswe wishto establish attested G languageinto its many historically uniform originally ourtask ofisoglosses, areidentified bybundles dialects. Sincedialects dialects suchthat developed how thevarious isoglosses is to determine can,though innovations linguistic search onlyshared resulted. In this

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE

PREHISTORY

OF

THE

GREEK

DIALECTS

56i

have value. Shared retention ofan archaism they need evidential not, isno evidence whatsoever. toreplace From allthat I have itisclear I propose saidthus that far, theStammbaum infavor model ofG dialectology ofthe (family tree) oflinguistic Wellentheorie (wavetheory) first diversification proposed in I872 (Die Verwandtschaftsverhdltnisse byJ.Schmidt derindogermanischenSprachen, Weimar). Accordingto this theorya linguistic innovation, whatever it maybe, usually in a givenareaofa originates speechcommunity, and if thatarea is linguistically the prestigious, innovation is likely to spread to other contiguous areasuntilit either embraces the entire or meets speechcommunity resistance, linguistic and/or in an (several) geographic, Numerarea(s) andceasesto spread. ous casesof thisphenomenon are documented in thelinguistic literature. Familiar and accessible Bloomfield's examples (from Language) include (322): Eng. vixen beside fox directly confox. It is clearthat tinues theOld English andthat form vixen doesnot. Somehowvixen made itsway intostandard English from in which/f/ a dialect regularlypassedto /v/:foxwas unaffected. a better Perhaps exampleis thatof Netherlands "house" and "mouse" fromearlyGermanic [hu:s] [mu:si (Bloomfield 328-3I). Thereare threedialectal areas, in two of whichthe wordsare pronounced the same,whilein the third they arepronounced differently (I havesimplified thedata): western [my: [hy: s] s]
central
eastern

[mu:s] [mu:s]

[hy:s] [hu:s]

The innovation clearlyconsists in fronting the back vowel and, accordingto Bloomfield,may have originated in Flanders. It succeeded in embracing all of western HollandandBelgium, butin a central area affected only older [hu:s]. It did not reachtheeastern regions at all, and the earlierstageof thelanguagewas there preserved.3 I do not,in adopting thewave-theory modelof linguistic change, wishto denythat changes mayaffect one branch alone- the Boeotianvowel system (Buck i i53-54) is sufficient to disorove suchan
3 For discussionof the "Wellentheorie" cf. Porzig I950: 23-27 (W. Porzig, Die Gliederung des indogermanisches Sprachgebiets, Heidelberg), Lehmann I40-4I (W. Lehmann, Historical Linguistics: an Introductiotn [New York I962]), Pedersen 314-I8 (H. Pedersen,Linguistic Sciencein theNineteenth Century [trans. J. W. Spargo, Cambridge I93I], Bloomfield 3II-19 (L. Bloomfield, Language [New York '933]). And for

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

562

WILLIAM

F. WYATT,

JR.

[I970

butI do feelthat smalllinguistic comwithin one rather assertion-, suchas was ancient likely thata change Greece,4 it is highly munity in perhapsone smalllocale will spreadto neighboring originating world.5 to thewholeGreek areas as wellandperhaps mentioned: *-ti > -si. already To takeup againone oftheisoglosses
dialect geographyin general: Bloomfield 32I-45, Lehmann II5-35, Hockett 47I-84 Linguistics (C. Hockett,A Coursein Modern [New York I958]). 4 There should be no need to argue that Greece was a single linguisticcommunity. seem to assume oftenenough thatGreeksof different Nonethelesslinguisticdiscussions and enjoyment dialectsdid not conversewith mutual understanding areas and different unitymightwell be adduced here. The two most thatevidence forHellenic linguistic importantbits of evidence are: I) the Greek dialects,however widely and wildly they and morphology,are very similarin the deeper level of may divergein pronunciation of vocabulary,one of the in matters language,namelyin syntax(Buck I36); 2) similarly more unstableareas of language, Greekwas uniform to a veryhigh degree. These two between any two Greek speakers. On a nonfactorsensured mutual intelligibility linguistic level we may point to the factthat,at leastfrom776, the Greeks-all Greeksand artistic eristic contests: was one aspectoftheGreekway. Greeks indulgedin athletic at pan-Hellenicshrines suchas Delphi and Olympia to carriedon thesecontests generally which all Greeks could go. Alliances between Greek states, though perhaps the exceptionrather thanthe rule,do at leastargue forHellenic unity. Finally,and perhaps the Greeksthoughtof themselves as a singlepeople as opposed to the most importantly, 3apfapot, as is proved by the popularity of the Homeric poems and perhapsmost impressively by the genealogicalbit in Hesiod Fr. 9 (Merkelbachand West):
"'EAXpvos o' Eye'vovTo 0Ao7rToAE',ovflaocAXyosZIcpos, TE Z-oQoVs TE Kac AXoAos oT7rtoXdp,tq,s

5Examples of diffusionand purely areal extension of linguistic innovations are frequent within the ancient Greek dialectal world. The following-convincing probably in varyingdegrees-may serve as examples. Phonology: I) in a geographically but not dialectally isolable area (Attic, W. Ionic, Phocian, Arcadian,Locrian, was assimilatedto /rr/ (Coleman 67-68); 2) Megarian plus Rhodian and Theran) /rs/ palatalized A (Buck 64) occurs only in island D dialects(Coan, Melian, Thasian) plus in Cos or Cnidos; 3) ATand AApass Asia Minor Cnidian: thechangeprobablyoriginated and Ar (Buck 64-65); 4) LtN] (< *ty*ky*tw)developsa stop to vTand v0 onlyin D dialects > -ot in much the same area as thatin articulationonly in A B and Euboean I; 5) -otn dialects(Buck 88) which ATAA > vT v@,i.e., in Ar B Elean, and in general,in northern > u-T (Buck 72). Morphology: I) onlyin theDodethisis in partthe area in which Or& canese hexapolis (and Sicily) was the perfectindicative inflectedthematically(Buck II8); 2) somewhat the same area, but this time includingThera and Central Crete, is of thefuture embracedby theactiveinflection passive(Buck IIi7); 3) themiddleparticiple
of -E'W verb ends in -ElfLEvos (- ?lxEvos) only in Northwest Greek and B (Buck 124). Syntax: I) E7T plus dative with names of the dead occurs only in Phocian, Locrian, B and L (Buck 109); 2) -rapa with the accusativeinsteadof thedative,occasionalelsewhere, Onomasticon: I) occurs regularlyin Northwest Greek and T, B (Buck i08-9). elementin Tt-qa- (Ttri--) occur only in I, Cnidian, Rhodian personalnames with first (Buck I33); 2) names in -KAE'as (-KAI'as) appear in T, B, Locrian, Phocian, Aetolian,

Megarian. In all of these cases areas ratherthan dialectsare involved.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

563

It is quiteclearthat -si is thelater forwe knowbothfrom form, other G dialects andfrom withother IE languages, that themark comparison ofthethird in theverbofprimary tenses was *-ti. Hence-siis person an innovation. It isalsoclearthat innovation affected this the Cyprus, pre-Dorian and theAsia Minorcoast. It is not Peloponnese, Attica, certain thatit affected all theseareasat thesame time,but themost reasonable hypothesis isthat itdid. Ifso,L speech wasalready localized on Lesboswhenthis and that, from tookplace, theevidence change of havebeenbefore must likein all cases Mycenaean, I200.6 We should to be able to be as precise as this, but we cannot. Ideallyour goal should a relative be to provide ofchanges chronology showing stepby stephow theGreek worldsplit up dialectally.We willfallsomewhat short ofthat goal. We can approach I shallassumethat it, though. In whatfollows linguistic changes occur, for reason, ina given whatever geographically restricted locale and thentendto spreadfromthere to neighboring areas. I shallnot,therefore, demand that all Ae dialects all and share only Ae changes, Doric (D) dialects D shareall and only changes. Therewillbe many casesofoverlap from one areato another, though probably not,exceptin thecase of archaisms, of noncontiguous areas with one another. Our taskwill be to determine what happened (e.g., assibilation of *-ti); where it happened(Peloponnese, etc.); whenit happened (after settlement of Asia Minor?);and,ifpossible, how it happened (probably [ti]> [tsi] > [si]). Thisparticular instance is straightforward and easyto follow. In othercasesit may not be clear whichis the innovation, usuallybecauseof our ignorance of whatthePG form was. One such(relatively unimportant) caseis that ofAI o0vo/a vs. ovv/a of(apparently all-Buck 27, Coleman74) other dialects. The PIE nominative ofthis wordwas *nomn, andas a result theAI form is generally heldto preserve theoriginal vocalism, while all other dialects haveraised /o/ to /u/. But in fact, AI hasinnovated inthis matter andhasassimilated thequality ofthesecond vowelto that of thefirst. Thisis proved by two facts. i) In compounds -wvvpos remains even in AI, thusindicating thato'vvpa developed to o'vo/a
6 Unlessindeed thechange was inaugurated in thePeloponnese and laterspread to AI L. It is possible, butperhaps notlikely, that substratum influence wasatworkhere. See below.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

564

WILLIAM

F. WYATT,

JR.

[I970

we qualityof o'votta, theoriginal in thefreeform:if AI preserved the quality of 2) The -vvuos. of the /u/ explain to hard put be would that onlyon theassumption wordis explicable vowelin this prothetic /a/was roundedto /o/by a following original /u/:sc. *anuma> was*onuma. word ofthis PG form I97I): the *onuma (Wyatt to havea mapon whichthe in somecasesat least, It willbe helpful, of the in orderto show thearea and extent may be plotted changes in changes.7 The map need not be accurate every variouslinguistic it place onlythat provided schematic, can be highly and indeed detail, attested actually of G (moreor less)in their thevariouslocal forms as is (N) dialects, theD and Northwest domain. I slight pre-Dorian them and considering themtogether done,by grouping commonly interesting arenumerous withT, B, A, etc. Thoughthere coordinate out remain must they dialects, in these and developments innovations to areforthemostpartlate,or arerestricted forthey ofconsideration, history. The map, fordialectal DN, and henceareofno importance willshow: groupings, dialectal thetraditional including

DoricNorthwest

Aeolic

T B D N Ar C
Arcado-Cypriote
(3-I40).

L I

Attic-Ionic

7 This has been my procedurein workingon thispaper, using Coleman's list(I07-I3) changes as a basis and adding to it fromBuck's Grammarof the Dialects of fifty-one

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

565

I haveenclosed On this to indicate drawing themapin order that, though doubtless theoutside, influenced from Greek was a single entity.Sucha traditional mapwillserve, butitwould be better and moreaccurate, whilekeeping theoutline of themap,to omitthe intersecting lines, and,instead of dialectal designations, to introduce directional coordinates:

N o rt h

T BL
West

aEast

Ar
C
So ut h
ThusAe becomes Northeast Greek (NEG), AI becomes Southeast Greek (SEG),etc. Earlier changes, atleast, willbe shown byshading those areasaffected by an innovation: later, post-Dorian, changes, D andN dialects, involving individual willhave tobe drawn onactual representations ofGreece in theclassical period. My procedure will be tolist the various isoglosses, todetermine try where the innovation andtodraw lay, what I hopearethe proper dialectological inferences. It willbe most convenient first to take someof themoreimportant of Risch's twenty isoglosses andsee whatconclusions emerge.We canthen ofother treat isoglosses interms oftheframe by then established. In so doingit is important constantly to bearin mind that theestablishment of an isogloss, i.e., a linguistic innovation, does

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

566

WILLIAM

F. WYATT,

JR.

[I970

and commonlinguistic innovation communication not preclude spokethe still all they by that isogloss: only people divided among language. same
*-ti> *-si (Coleman #I0, Buck#I)

i)

tothe restricted shared byArC,Al andL, andthus Thisinnovation, DN, T failed to affect world, oftheGreek parts southern andeastern may be dialects.Thechange speaking thenorthern andB, generally -zi for-ti: language (cf.Hittite of a substrate due to theinfluence for We dialectological purposes. Coleman 70),andifso,isirrelevant areaembracing an innovating ArC,AI shallsee later, though, that If it is a comisoglosses. other by numerous and L is reinforced of Mycenaean (Myc.) I200 (because it antedates mon innovation, had its originin the saw,and probably -si), as Rischcorrectly in havebeenpresent already ofNEG (L) must Speakers southeast. AsiaMinor.8

D
.

~ ~

*.

:.

8 T B have -v&t,,and in general -v-r> -v@- in the verb in those dialects. Hence after-v- similarto thatseen in Skt. panthah thereoccurreda relaxationof articulation not be of the same natureas the assibilafromPIE *pontos. This change may therefore tion of -[ti] which did not require-but allowed-a preceding-[n]-. NonethelessL may have had -vOtin the thirdplural active. See below NEG 6.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOIJ
2) Vpo'S

THE PREHISTORY

OF THE GREEK

DIALECTS

567

I) vs. oTt "near" (Coleman#3

to and does not attempt Risch merely opposesthesetwo forms itis usuand whichtherelic. In fact decidewhichis theinnovation and areinherited, bothforms (Risch66, Colemango) that allyassumed or other the have chosen one independently thatthevariousdialects a consideration of the map, butfrom form. This is indeedpossible, commonto DN T B (and Homer),is theolder thatvoTor4 it appears underthe of Tor4, and thatirpOrt is a development possibly form, correctly observed of -rpo and7rpO'cTOEV "before." Coleman influence here:theinsertion of-p- and theassibilaaretwo isoglosses there that inArC A I L, takes placeas in I) above,thus tionof*-ti.9 Assibilation of-p- occurs whiletheinsertion onlyin A I L, and in Cretan (voprt) and Argolic(7fpot': once, SchwyzerI923: 84, 3, in an inscription two Cretantowns)beside betweenCnossosand Tylissos, mediating ofthe irori and ifoi.It thusseemsthatfpo-rt> rpo'S' is an innovation recitation through spread, perhaps (SEG plusL) which AsiaMinorcoast to to Argos and more firmly of the Homericpoems,sporadically theCretan form unless is a parallel andindependent innovation. Crete, innovation are,this and Cretan and Argolic SinceArC is notaffected,

Locr. Phoc. Ach. Elean Lac. C

T B Cor.
* .*

East T

Meg. Arg. Theran Coan Rhod.

9 FromMyc.po-si we now knowthat we havehereto do withassibilation and not withtheaddition of -s9 to irpoas heldby Buck(58) andColeman (89).

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

568

WILLIAM

F. WYATT,

JR.

[I970

Dorian of ArC, henceafter theisolation musthavc takenplace after of the Peloponnese. Hence, too, we cannotutilizeour settlement Greece during represents a mapwhich map,butrequire first schematic correctly.The this innovation to display periodin order theclassical age bronze indetermining be utilized of-p-cannot or absence presence dialectal relations.IO
3) *totios> ro'uoS (Coleman #I2)

well as zo'oroS< oaos'<o<*totios. Risch's expositionis merelya for-au- > -a-. We cannot be sure, becauseof shorthand convenient thatCypriote sharedin this in the Cypriotesyllabary, deficiencies is generally heldto showthat Myc. to-so innovation. Since,though, did as well. In that did, we may assumethatCypriote Mycenaean all of SouthGreek(ArC AI), is priorto the affected eventthechange to be prior and is likely also to of thePeloponnese, Doriansettlement It is of Greek. area of L Ae into the innovating the movement structure different consonantal thatthe markedly however, possible, the among the continuants-prevented of L-frequent geminates dialects of -ss-there. We will see laterthattheNEG simplification in their consonantal conservative system. It is thus were in general L was not andto state to be moreconservative onlythat bestourselves one of theveryfewshared affected change. Thisone change, by this
I0 From the above discussion I have omittedtwo formscommonlyintroducedin this connection: L TrpE's and PamphylianTTEp7L (Schwyzer I923: 686.7 7fEp- 'p?vt; 686a4 SGDI I260.3 7TEPTKE'SOKE cited in Schwyzer's note to 686a). L TrpE's 7TEpTrESOK, treatise dealing with peculiarthingsthat occurs only once in a sectionof a grammatical a sixthor seventhcentury to Ioannes Grammaticus, happen to /o/in Aeolic attributable Dialekte, Vol. 2 grammarian (Hofmann II.2i6: 0. Hofmann, Die griechischen A.D. Since this form (7Tpos oE' 7rpEs oe) appears in no other sources, [Gottingen] I89). to findphonologicallydistinct earlieror later,and sinceindeed it would be moststrange timesto theAlexandrianperiod identicalwords coexistingfromearliest but functionally in Aeolic, we may safelyremove thisformfromconsideration. It probably restson a aE (Hofmann ibid.244; cf.wp of the Hesychiangloss rEp amisunderstanding and I cannotclaim to be able are more difficult, E,E 7rposRiE). The Pamphylianforms in which theyappear. On thewhole, though, themor the constructions to understand even thoughthis,too, causes trouble,I tend to feelthattheymask 7rE8a.
7TppOS

origin:yEVEut <

for it This changeis broaderthanRisch and Coleman indicate, of their at thetimeregardless affected all casesof -ss-whichexisted
yEVEuut < *genes+ si; 7rouL < Touui<

*pod +si;

as

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

569

exclusively outofa unified byArCandAl, created Greek a division southand conservative between innovating north.IIFrom this (possibly allinnovations pre-Mycenaean) point on,almost in originate DN andAe aretogether thesouth. Thefact that innotundergoing thischange abouttheir to one another: relation provesnothing common notpreserved alone canbe evidence innovations, archaisms, oflinguistic relationship.

B %: ~I::

5.

4)

Athematic infinitive: -vat vs. -tLEv(at)

(Coleman Buck#43) #48,

Again Risch merelystatesthe isoglossand does not attempt to determine where theinnovation lies;andindeed inthis casethere seems little need: -vat characterizes ArC Al L, thusthesame dialectal area we found in i) and 2) above,while-,uEvis shared by DN T B L. L belongs to bothgroups, andprobably therefore reveals thechronology ofthechanges:
I) development of -,uEv 2) addition of-at "Homer hasboth -ro'aaos and &roaoS,and ifwe identify Myc. to-sowith this latter, shorter, then form, we areforced to conclude that Homerpreserves a pre-I200 linguistic form. Thoughelements in thepoemsmaybe thatold,it seems to me unlikely that anylinguistic is. HenceI feel form that Myc.to-sorepresents [tosson] andthat thetwo SG changes ofPG *ts areto be dated as follows:
pre-I200
pOSt-I200

*ts> *ss
*ss> *s

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

570

WILLIAM

F. WYATT,

JR.

[1970

a demanded endings ofthese thedistribution sawthat Risch correctly Myc. though in thePeloponnese, presence to theDorian dateprior lacking. a datebefore I200, is yet alonecouldprove which evidence, prein 3) aboveis clearly seen likethat andsouth north A split into of L in of -puEV,'2 and theinclusion by thedevelopment supposed of -at. In thiscasewe seemto by theaddition Greek innovating on themap:shading twogroupings andhence twoinnovations, find will denote cross-hatching while L, ArC AI will as usualdenote
DNTBL.

is conas far as dialect geography is required is all that Thismuch whatthe also to attempt to determine butit is important cerned, in thesouth in assuming that was. IfI am correct situation original infinitive ending, tothe athematic wasadded infinitive aorist -atofthe notnecessarily, was-(e)n, though possibly, that ending that itfollows It is infinitive. thesameas thethematic thus *-hen< *-sen, from andrendered wasextended this *-en in thenorth that clear therefore the athematic of-,- between insertion clearer bythe morphologically
stemand the endingin itslonger form -Ev. This -st-, like the -a- of
I2

and its had no meaning and the-K- of the perfect, theaorist passive
of -ptbelowtheinsertion too strong. As I shall argue is probably Thisstatement discarded SG but by and hence inherited in its occurrence been PG Homer) (cf. mayhave in favor of -vat. there

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

57I

wasto protect themorphological solefunction oftheatheintegrity matic infinitive. Such that being the itislikely the source ofthe case, will never be but it seems to me most that determined, -slikely it somehow arosein theinfinitive of theverb"to be." Either *e:n of the in DN at least with thethird (< *esen), homophonous plural imperfect *e:n,was felt to be too short, of thesameverb, andwas
again infinitivized by adding -en: *rqvEvthen passed to '1 eV (Et.LEV, Ev) by dissimilation:(dissimilationin the case of 'lqJEv seems

in some andrenders this difficult, explanation dubious);or-i- was felt theverb" to be," andtheinfinitive on the waytocharacterize wasbuilt supposedstem*em-. It is difficult to know when and wherethis innovation had itsorigin-itmust havebeena constant temptation-, butitprobably tookplaceprior to thepassage of-RR- (R = resonant) to -:R- in all dialects save thoseof thenortheast (#9below). If this is thecase,then itwas a very early change indeed. ofthesouth The dialects form either never knewtheinnovative -1tEv because of the Homeric situation), (which is unlikely or, though
knowingit,preferred usingtheold infinitives (*e: n " to be," *titb6n (or
*titb6: n (or *hista'n) n) "to place," *hista: "to stand") extended by another clearmarkof theinfinitive, the-at of theaorist, to inserting morphological mark, which hadthe addeddisadvantage -p-,an unclear a disturbing of creating nearhomonymy withthefirst person plural of thesameverbs. Thischange activeending probably was aidedby the presenceof -at alreadyin all middle infinitives. Thus it is

possible,though perhapsnot likely,that Eretrian Etv Ttl5EWV &tsoiv,

Chian Etv,directly reflect the PG situation, as may also Cyrenaean Ka'r'rtl&Ev&8cov beside `15EEv &0'JLEv (Buck I25). L clearly is either in adding-at to alreadyremodeled doublyinnovative -ptEv;or is inpreserving conservative inadding -uEV,innovative -at.'3 Rhodian and Cretan later innovate by extendincy the onantitv of thethemati

13 L -v in the infinitive of contractverbs (Kipvav o',uvvv KcLAIv) causes some slight troubleif one holds thatsuch verbsare conjugatedathematically (Buck I23): we should expect *KLpvac4evaL *odLuvv1,uevat *KaA47/EvaL. In factformsof thissort do occur in Homer:KaA7)(LevaL IL.IO.I25, StX?7,LEvaL II. 22.265, andprove that this didexist option

at onetime, butfailed to catch on. TheL forms, however, the merely preserve PG andindicate situation, that these verbs were notconsidered to be ofthesame type as other athematic verbs such as -r&9%qjut. Putanother theL rule way, read:add-atto 630.I5) seems to indicate that in L &$'ouu hadentered the -juev.&ucov(Schwyzer
ratevorv of n-contracts.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

572

WILLIAM

F. WYATT,

JR.

[1970

whichdid not get an innovation infinitive -enalso to theathematic, alsoto extending direction, -Vuev in theopposite far. The innovation of mainland to theNE portion was restricted infinitives thethematic fromLyttos(Buck I22). I feel thathere we shouldfollow King Grammar and Generative Linguistics 90-92 (R. D. King, Historical T andB showa generalizathat Cliffs I969]) and assume [Englewood therefore andthat rule, grammatical understood tionofan improperly in NWG or either elsewhere, originated as an athematic ending -p1Ev in T B. (and misunderstood) and was introduced indeedin PG itself, matters, in morphological conservative Since DN was generally with-en coexisted -men that to assume I prefer in theverb, particularly in SG in favorof -en,subsequently already in PG, but was discarded ourfamiliar mapsof I) is thecase,we regain by -ai. If this extended for-en): for-at) and of3) (cross-hatching and 2) above(shading
Greece (and to Homer?) save for the one isolated
TpoFEUTE'1LLEV

5)

EL

vs. ac "if" (Coleman#34,Buck #4I)

Again Risch merelyliststhe isoglossand does not endeavorto as a result and whichthearchaism: whichis theinnovation determine of forprehistoric grouping thisexamplecannotbe used as evidence and be usedunequivocally itcannot stated, dialects. Or, lesspositively is theolder. Etoccurs form as towhich decision an arbitrary without in Ar Al, ai in DN Ae.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI] 6)

THE PREHISTORY

OF THE GREEK DIALECTS

573

(av vs. KE

and Ka, themodalparticle (Coleman 5,Buck#42) #3

The dialectal distribution is muchthesameas for"if," withAr Al forms withK-: KE in showing a'v, and thenorthern dialects showing L T, and in this matter joinedalsoby southern C, Ka inB ND. Both Ae and ArC hereshowdivergences within whathavein thepastbeen form considered united families. We againneedto knowtheoriginal in order is theinnovation, and though Rischseems to to decidewhich he does not undertake favora'v as theinnovation, to show by what of thisproblem, items5) and 6), routeav arose. In pastdiscussions have generally becauseobviously words, havingto do withdifferent as separate matters to be discussed beentreated separately.I propose and hold thatthetwo developed themtogether instead thatwe treat in contact. The question wereso frequently as they did becausethey collocation whichliesbehind becomes:whatwas theoriginal the then -K- because forms? It must of theagreement attested havecontained withAr dK, andindeed because oftheagreement of oftheNG forms we may havecontained KE. Suchbeing Ae andC KE, it must thecase, T andL (together withHomer)havepreserved theearlier presume that was atKEV. with" if" as wellandthat collocation theoriginal situation dialects haveinnovated in one way or another, DN B only All other inlowering need thefinal vowel to -a: thislatter innovation slightly was not,contemporary withtheSG change. notbe,andprobably In SG there a tendency-for reason-to interdeveloped whatever EL of such thequality thetwo vowels that Kav resulted, a form change in Ar (Schwyzer stillpreserved occasionally I923: 656 passim). This did not affect becauseC had losttouch tendency, however, C, either ofSG. or because withthemaintendencies C hadalready innovated by at with?, and theexchange ofvowel quality was no longer replacing thatC failsto possiblethere. There is no reasonto be surprised withAr in thisinnovation, forsurely it is frequently the participate case thata change-particularly a morphological change-does not so as to embrace an entire spread area. Only thosewho imposethe unrealistic area(dialect) be requirement that every linguistically defined will find C KE hard to accept. That this linguistically uniform to interchange low-front withfront-low has not been intendency vented to account forthis by the(atleast) solely one relation is proved

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

574

WILLIAM

F. WYATT,

JR.

[I970

optative person pluralof theaorist SG tendency to replacethethird cases such as well as by certain more isolated -atEv with -Etav,as andHeraclean 'P'4vEta beside ITpoTEpEt`a forA lTpoTEpfxta. 'P'qvatEV'5 all dialects ofthesouth in relevant (saveC, whichis no longer Clearly theresulting redivided thismatter) complexELKav, probablyon the analogyof the frequent oV3Kav,and came up with the new Et av herein their treatAr AI innovate (> Eav in AI). We thusfindthat in itsreanalysis Al, at least, further innovates mentof atKEV, and that of E' and 6'v.I4 Ae and DN B, to the ofthecomplex and itscreation behind, and did in thenextsection, wereleft latter ofwhich we turn about any not experience thischange. Again thisfactsaysnothing Ae and DN sinceagainbothhave merely closer relationship between situation. Our mapwillshow: preserved theoriginal

B D C
7)
OTE'--

ofa--

oKa

"when" (Coleman#36,Buck #7)

T (T o'T'-SchwyzerI923: 565.2 occurs in ArC Al andpossibly in that in L is ambiguous); osra (and possibly T); o'Ka in DN B, thus
OTe (L. R. Palmer in: A Companionto Homer,A. J. B. '4I here follow Palmer 90-92 Eds. [London I962]) and K. Forbes(Glotta27 [I958] I79-82) Wace and F. H. Stubbings, in assuming*eikanas the ancestorof the SG forms. I cannot,however, follow themin (e-grade,o-grade,zero assumingthat Ka(v) is the zero-gradeof KEV. Ablaut relations in IE languages, but are rather grade) are not phonologically conditioned alternants in PIE may have been. Since of what the situation regardless conditioned, functionally betweenKE and Ka, and sinceKEV > Ka(v) is not a phonodifference thereis no functional logical rule of G, I do not feelthatablaut can be invokedto explaintherelationof KE(V) to (K)a(v).

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

575

area(generally NWG) that hasKa forthemodalparticle KE. We may assume that o'Ka alsoreplaces earlier *O'KE withthe final vowel again loweredto /a/, a lowering seenalsoin 7upoo-na "before"(Buckio4-5, Coleman#37).'5 This samelowering all of seemsto have affected NG (againwiththepossibility thatT is an exception), forL haso'Ta (Schwyzer and To'-ra (Alc.PLG 38A II): L, I923: 634.33)and dAAo-ra before thefinal vowel was lowered, had o'TE. Thus we see thatfor dialectal grouping onlythe consonant and not thevowelis relevant. ArC AI L T (and Myc.?) are in agreement in havingorE as their starting pointas against NWG *O'KE. Thus we finda grouping of dialects thatwe havenot seenbefore. The question thenis who has innovated and who has retained the earlier form. Dependingon which areawe chooseas theinnovator we willhaveoneor theother of thetwomaps:

tV
B~~~~~~~

Ar
I5A et'Ira Tczrea, which seem to obey the NWG rule, beside I etre bTetreT has the second vowel dissimilated by the first:44-re> eStra; cf. at'Kev > cec"av.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

576

WILLIAM

F. WYATT,

JR.

[I970

to south innovations tendto movefrom we haveseenthat In general DN B are conto feelthat predisposed and we are thusrather north in T causesone to againin thismatter. Only thesituation servative has been among forT elsewhere a conclusion, twice aboutsuch think awaits our finding decision ofdialects. A final themostconservative a generally orexclude toconfirm either that wouldtend isoglosses other NW. theconservative as against EG grouping also by thefactthatwe are not clear we are hampered At present I believe,only of thesewords. Thereare in fact, about thehistory fourpossibilities: inPG coexisted *O'KE andO'TE i)
2)
*OKE > OTE rE

3)

4)

> * KE
6KE

X>

> OTEelsewhere

inDNB

(>

'Ka)

in werecompeting that twoforms Ifwe adoptI), we then must suppose in "whenever"(velsim.)was generalized PG, and that*oKEmeaning "when" in a correlative and meaning NWG, while5?TE (< *hokwe?), " then"wasgeneralized inallother areas. Thisis not sense with*tokwe though all requirements, but seemsunlikely. 2) satisfies impossible, > /t/-/e/ ifwe canassume a change /k/ weaksemantically, itisperhaps inall areas ofGreece savein NWG (andpossibly inwordfinal position all would satisfy 3) likewise in Homerico' KE in thephrase ess o' KE). of/t/ a change excluded bothphonologically: butseems requirements, and dialectoa palatalizing vowel seems unlikely; highly to /k/ before assumewhat is to be proved:we if we may prematurely logically, area. in a generally conservative an innovation would haveto assume of any of theotherthree. x mayhave a variant 4) maybe merely clearthatit x be neither o'TEnor OKE, it seems that But ifwe require a possibility envisioned (80) and Schwyzer byColeman be *hokwe, must approand one whichseemssemantically (I939: 629) amongothers, will have been simpleand developments priate. The phonological nature nonethelessof a rareand infrequent though straightforward, inall ofGreek as a to *hoke passed suchas to seemexceptions. *hokwe occasionedby the preceding of labialization resultof dissimilation in I OKWS0 OKOta. forwhichcan be found parallels vowel/o/, rounded

bex. bex orJmE may too'IE andtoOKE inPG; O'KEmay both developed

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOIJ

THE PREHISTORY

OF THE GREEK DIALECTS

577

In all dialects, saveforDN B (andpossibly Homeric though, O'KE), the phonologically regular*hokewas replacedby the morphologically clearer in all othercases to *hokwe whichthendeveloped as did -kwe -_TE.I6 Or we can operate a replacement without assuming by simply in all of Greece holding that*hokwe was felt NWG to be comexcept posed of *ho + kwe, two morphemes separatedby a morpheme boundary, whilein NWG themorphological oftheword composition was ignored, andphonological in therest ofGreece, processes, blocked allowed to operate. The preserved (or restored) morphologically intact *hokwe renders clearthefact that this wordcontains therelative pronoun o plus-kwe whichmakesthewhole correlative with *tokwe "then" in a semantically satisfying way. It seemstherefore most likely that a pan-Greekmorphological innovation (or reanalysis) failed to reach NWG, a conservative areaembracing DN plusB. We require thefirst ofthetwo mapsgivenabove:

16 There areobstacles in thewayofassuming *hokwe. i) Myc.haso-te(PY Ta7II.I) for which we should expect *o-qe. 2) C haso"e (Schwyzer 679.I) for which we might rather expect*07Te. 3) L has ora instead of expected orra. The L example, though, iseasily taken care of, for*kwe> -E evenin Ae inthewordfor"and." We mayassume that, though this development didnottake placeinMyc." and,"itdidtake placealready in "when." The samedevelopment will clearly haveobtained in C. The palatalizationof *kw before /e/ seems to havespread gradually, affecting first enclitic and encliticlike words. Hence: i) *hokwe>*hote; 2) -kwe" and" > -TE; 3) *kwe> *te elsewhere (onlyin Ar AI DN).

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

578
'

WILLIAM

F. WYATT,

JR.

[I970

8) /3oA-/,-eA-SeA- "want"

(Coleman #38)

consonant, seems to regard theinitial In this instance Risch,ignoring forifit is not,thefact vocalism of theverbas theinnovation, the/o/ thatArC AI L againagreeis of no importance.I can see no way of thatL again sideswith SG, as in provingthis,but it is interesting thusfar. We will have to assumethatL nearly all casesconsidered in Asia Minorin possession of both *gvel-and *gwol- in this arrived of the the/e/ of replacing of words,and thattheinnovation family reachedit, causingit to discardthe subsequently indicative present by our we are hampered *gvel-. Once again, though, competing word. In any of this and theexacthistory ofthePG form ignorance theless as beingveryimportant, eventI do not regard thisisogloss
(SchwyzerI923: so sinceEast Cretehas fidAcWvraL
I99.I6),

be givenas muchweight as someof theother changes. Nonetheless of I) and2) above: themapis that

and is not to

Av

of Risch'stable,then, we findthe clear Summing up thissection NG and SG postulated by him confirmed.This split splitbetween that communication andisin ceased, abruptly doesnotmean,though, an indexof the factthatGreekspeechnow all probability merely to thePeloponnese. SG as defined from by Risch extended Thessaly if -/s/-), onlyin 3 (-/ss/-> in itspurest, form mostcanonical, appears

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

579

in thischange;and in 4a (-Evas theendingof the indeedC shared athematic infinitive). ElsewhereSG includesalso L: I (*-ti>-si), 2 (p(r)oti>pros), in 8 (foA4b (add-at to theathematic infinitive), 'want'); or does not includeC: 5 (El for"if"), 6 (av as themodal thelowering of final particle). In NG we findbut one innovation, inadverbial a change which inthecaseofKE failed to-/a/ to -/e/ words, reachT and L, though it seemsto have in thecase of oer'. In NG, as Rischcorrectly therefore, we needno dialect boundaries, saw,save in thecaseof Ka vs. KE and Ka vs. T(a),andthese (slight) differences demand a distinction between DN B on theone handand TL on the notbetween DN andAe. All other shared other, isoglosses represent forI prefer to regard -VEv of theathematic as an archaisms, infinitive archaism of theHomeric because situation: was probably present -,/LEv in all areasof Greecebut was discarded in innovating SG. originally notonlyArC AI L butT as well. It doesnot -o "when") includes fitin quite comfortably with the otherchanges of i-8, and I shall postpone discussion ofittilllater. In SG matters aremorecomplex, buteventhere there is no needfor dialectal distinctions of the traditional sort. Some changes affected onlyArC AI, and did not reachL (3, 4a); others affected Ar AI, but to reach L or C (5, 6); butthelargest failed either majority (I, 2, 4b, 8) reached bothC and L. We therefore require three different extents oftheareaofinnovation:

T
B

D~

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

58o

WILLIAM

F. WYATT,

JR.

[I970

T B D
..... ..

D~~

Map 2 showsthe innovating area at its smallest, 3) at its largest. We mayhypothesize areai) first that theoriginal innovating expanded (to 3) andthen contracted (to 2). Thereis no needforthis hypothesis, butifwe wishto tiedialectal innovations to history, we must suppose somesuchdevelopment. blankbecausethere was no change Maps of NG are either (I, 2, 3, > commonto thewholearea(7: -/e/ 4, 5, 8); or showan innovation DN B > or an innovation to restricted (6: -/a/). -/e/ -/a/);

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

58i

.~~~~~.S

A Ar
a) may beolder than b),though such isnota necessary and assumption, b) may beroughly L and B with have at contemporary 3): i.e., the may same oratabout time, the same become time, from separated T, which is merely another way of saying T at aboutthis that timebecame isolated from the rest ofGreek. It is interesting to notethat nearly all ofRisch's isoglosses areSG innovations, and perhaps are moreparticularly SEG characteristics, i.e.,features which inwhatever originated the ofAI pre-Dorian home was. To these we can add several which more embrace theentire innovating area. SEG I)
*tu> su (Coleman #7II)

(Lejeune56), seemsto haveaffected thesameareain which-tipassed to-Si (Buck I know ofnoevidence for C. 58),though

Thischange, whatever theprecise or analogical phonetic details are

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

582 SEG 2) #6)


?EpoS
(Epo'.S,

WILLIAM

F. WYATT,

JR.

[I970

tpos', lpo's., vs. (apo's'

"holy" (Coleman#20, Buck

is ArC AI, and its laterdevelopment tpos and ipo'& appear ofthis word, in AI. The PG form respectively in L and (sporadically) in all other dialects.'7 tapo's, ispreserved SEG 3)
ELKOUL vs. (F)iKart

(Coleman#22, Buck #2)

is againlacking, justified in butwe areprobably evidence Cypriote thatit also possessed ECKOcr. Thereare threeinnovations assuming here:
2)

*/wiy/> */ewiy/> /o/ in numerals /a/ >-/si/ 3) -/ti/ i)

whatwe already andindicates phonological The first changeis purely thatprothesis does not occurbefore knew,namely */wl- in non-SG dialects (WyattI97I). The second changeis partlyphonological, with the/o/ oftheending -Kov-ra butisalsopossibly analogy analogical:
in" 30 etc.,musthave been at work (Coleman 72, Buck 96-97).

Risch#i above.
SEG 4)
-Ko'cnOt

3) is

vs. -Kac-tLot or -KaULOL

"hundred"

2) and 3) of SEG 3), butherewe Clearlythischangealso involves forNG -KacTcot, AI L have-Ko'orot as expected a surprise: though find in having -K&OrfoC. We can ifwe like Ar takesup a middleposition Ar -Ka'-rto > -KOCfLOL (withSG) > -Kccrto undertheinfluence assume D dialects, but to me thisseemsa strange of neighboring linguistic in -KO'O'co is a later to assume that > /o/ event. I therefore prefer /a/ than-tito -si,and thatit tookplace whenArC was no development in the theDorianpresence after G, namely longer partof innovating in C, but I would Peloponnese. We do not know the numerals (FriskI.7I3, ChantraineI968: 458), and in general 17 cEpOS iS cognatewith Skt. isirah whetheraccentedor not, to Skt. /i/:G /e/, G /a/in an unaccentedsyllablecorresponds to Skt. /a/. It is for thisreason thatI consider Iapo's the earlier corresponds generally /a/35-4I, 56-59 (= Haney Foundation = Skt. /i/ cf. my Indo-European form. On G /a/ SeriesMonograph 7 [PhiladelphiaI970]).

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

583 A similar

venture to predict -Kac'cot /o/has spreadalso to T

inAI, andextension occurs with "twentieth": development ElKOaTOS, casethe ElKOLUToS inL, but tcKauTainAr(Schwyzer 6s4gI3). In this
lKOUTO'S'

even *EdKacn). (and possibly (Schwyzer 6I7g=IG

probably AI influence. under

IX 2,506,47)

SEG 5) dative of-o-and-a- stems in-at (Coleman plural #24)


A (sporadic) (regular) -otort.-utforms do occurelsewhere (Pamphylian-Schwyzer I923: 686.i, 3) and Cretan(rarely-Bechtel
-atort); I L 2.729-30, possibly under epic influence), and Coleman (84-85) may

in I -uat L (regular -at occurs -atut in the andA (sporadic noun)

well be right in assuming selection. Otherwise independent we will an assume Ionia-based whichspread, innovation in thenounat least, to L but onlytentatively firmly to A, to be replaced there by -ats, _oLs'.I8 There are a number of otherSG changes whichfailedto spreadbeyondtheoriginal (AI) area of innovation either to L or to ArC. Theyall are to be considered lateand of little importance for dialectal savethat classification dojustify thetraditional they designation of AI as one of thefourdialects of ancient Greek. It is to be noticed that it is onlyby means oflatechanges that we can define AI, againas Rischcorrectly saw. Our ability to define AI withas many isoglosses as we can is due in largepartto thewealth ofliterary andepigraphical material from this areaoftheworld.I9
18 It is generally heldthat-si in thedativecontinues PIE locatival *su, while -ots continues instrumental *-Jis. I feelthatmatters were probably more complicated thanthat, and thatthough A -&at (--gac) mayreflect PIE *-asu, and -oisreflects PIE *-ois, that after -owlais remodeled -a:t, and -asgafter -ols. In mostG dialects -&at wasreplaced there by-ats, andhence appeared -ats, -otS. It was an innovation ofAl L (andindependently (?) Pamphylian andCretan) to create -atat, -oLat. It is generally believed (Docs.84-85andVilborg thetraditional 56-7)that view is supported by Myc. 0-i and a-i which aretaken to represent [oihi]and[a: hi] respectively. Thisviewis,in my opinion, and I believethatMyc. spellings incorrect, represent -otsand -abs (cf. C. J.RuijghMnemosyne Series i ii-i-2; Iv, i i (I958) 97-I I6, particularly Atudes 76-78 = AStudes sur lagrammaire etle vocabulaire dugrec myce'nien (Amsterdam I967) andmyreview of this latter work:Language 44 (I968) 66-2I.

I9 Other SEG changeswhich mightbe mentionedinclude: SEG 6) > a+epos >'7zepos(Buck 24-25); SEG 7) ovvua > ovo,ua (Buck 27 Coleman 74); SEG 8) [u] > [U] (?) (Buck

28);

SEG 9) metathesis of quantity (Buck 37-38,4I); SEG io) v movablein verbal forms (Buck 84); SEG ii) -eS, -as-in thenominative and accusative respectively of

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

584

WILLIAM

F. WYATT,

JR.

[I970

originated in SEG thusfarthatall SG innovations I have supposed to encompass SWG alone,sometimes there sometimes andspread from L. It is now time to investigate merely SWG plus L, sometimes themore in anyway or was merely itinnovated SWG to seewhether originating elsewhere. Of course of changes or lessinert receptacle we arehere Myc.was a SWG dialect, andbecause SWG didinnovate, we date position ofbeingableto datetheinnovations: in theexcellent thosecharacterby Arc butnotMyc.20 Most of themare of course Buck was Since,however, (e.g.) by Buck (I44). istics of ArC listed more features andbecause he treated Myc.evidence, unable to consider to days, I feelit will be profitable these atomically thanis thefashion discuss someofthese characteristics here. SWG
i)

after shared shared byMyc., innovations prior toI200 those I200 those

midvowels > highvowels/l(n)#

either and/o/ become/i/ and/u/ respectively Thisrulestates that /e/ ov final final forviv from position. It accounts orinabsolute before /n/ (Ar vVElvaE-SchwyzerI923 :676) and lV from E'V (Schwyzer I923: 66i.i, etc., Coleman#7), thesefrom Ar; vvEv6a'(uEvos' forby ibid. C. Alsoaccounted (Schwyzer 682.I2) l(v) (Schwyzer 3) from in Ar (a,u'pav of a-stems thisruleare: -av of thegenitive singular inC ('Ovauayo'pav679.I); thethird alone a-stems 654.i6), masculine tenses of the verb: Ar middleendingof secondary personsingular 'o'Aovtv I923: 679.4); the (Buck#I6.4), C EvFpEraaa-rv (Schwyzer Tvavo'vrcv (Schwyzer acv < aro': Ar a'rv 66I.32) C acv' prepositions ofthefirst lowering dissimilatory (Schwyzer 679.8); 07TV < V7T0 with 2 (Schwyzer vowel(Schwyzer 664.I 5, 2I); Karv' < Ka-ro< Ka-ra byrule
656.29);

somewhat cases. The rulealsospread, lastthree forthese sporadically, we getoccasional that ofthewordin Ar withtheresult to theinterior etc.(Schwyzer Mantinea; suchas: ai7TEX0b'VoSv 66i.20) from forms
ofplace"where"in -ov (BuckI02); (Buck98); adverbs 71.,E- andvl5E'is inthe iv for changes oftheverb"to be" (BuckI28). None ofthese oftheimperfect singular third ofthekoineon thebasisofAI. theformation until spread ofMyc.)onlyforSWG. It is (because justification withcomplete 20 We cando this he datesall that to Risch'sscheme notpractical) objection (butprobably a theoretical may thanI200. Some changes to I200 andlater worldto prior in theGreek changes muchlater. areauntil another to I200 in one areabutnotreached havebegunprior

areno C examples &AAv andfinally 656.38); there (Schwyzer

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

585

(Schwyzer665 A 3), and before -/m/-,orvu'EOV 'EpXopuviotss (Schwyzer 664.2I) and61o0o0ts' (Schwyzer Orchomenos. 665A 5) from Of theabove,Myc. has onlyacrv (PY Ta64I.I, a-pu-do-si, frequent), thus inprepositions that thechange tookplacefirst suggesting andthen spread. All theother arelikely forms tobe early butpost-Myceneaean.
rrC1

SWG 2)

/a/>/o//#(C) VC[

(i)]

Condition: no grammatical is conveyed information by/a/ This ruleprovides that/a/ is raisedand roundedto /o/before /r/ in all positions saveinitially andfinally; final before andinabsolute /i/; final position inwords ofmorethan onesyllable, that provided /a/ does not convey grammatical information. The reason for this isrestricting mystself only to Ar-that we want to have Ivat (Buck not *Eaco; FEcEa (ibid.6), not #I6.I), not *Evot; Eaua (ibid.3), *F&Eo; atya (Schwyzer654.2), not *atyo; Ot &lKaaacrati (Schwyzer 66I.I9), not *8&KaUaTro; & caaaco5at (ibid. 17), not *aucraurOot. In all of thesecasesthe/a/ conveys somegrammatical information: neuterpluralnominative-accusative (FETEa); accusative singular of consonant stems (atya); and/or is supported forms of thesame by other with /a/not subjectto raising: category Ecaaa, nominative singular feminine beside &KaCUTaI, nominative plural E'aucadg; ofmasculine athe/a/ of theaorist. It does,however, allow/a/ to passto /o/ before in: ArTETopTav /r/ (Schwyzer I04),2ETop7Tao (652),7avayo'pac 656g etc. (Wyatt (654.26);C KacTEFopyov (679.J), I97ia); before final/i/ inthe third middle singular primary oftheverb:Ar/o'ArTOl (654.9), ending yEV7jTOL (656.2); C KE?TVl (683.6) with further application of rulei) extended to/-i#: finally inSE'KO seen inArSVOEEKO (654.2I),whence also 8EKO'TaV (IG v 2 282.2) and 'KOTO'V (Schwyzer 654.4); and in KaTac> whence toKaTV1bySWG i) above. KaTo, Alloftheaboveruletogether with itsconditions applies alsoinMyc., thusshowingthatthisrule at leastantedates 1200: before /r/: to-no " table" "chair" PY Ta707. I)-cf. va- VToro'8tov (Hsch.), to-pe-za (PY Ta642.I), wo-ze"works" (PY Ep6I7.4); before i, alsoin thethird singularmiddle: di-do-to (= didontoi) PY Ng319.2, e-u-ke-to-qe
Evat and&auacna&a stems;

both infinitives with the /a/ supported by

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

586

WILLIAM

F. WYATT,

JR.

[I970

"nine" e-ne-wo in numerals: PY Eb3s; and finally (= eukbetoiq"e) would "ten" number the that also suggests which (PY Ta642.I), havebeenspelled*de-ko. It mayseem strange-itdoesto me-that rule2) to rule i), forI would have expected rule2) shouldprecede had it For into being. coming forits need rule i) as a precondition with-u- in Myc. spellings I would have expected been otherwise, singular andin thegenitive ofthemiddle tenses (e.g.) in thesecondary must we result and as a masculine. Such is not the case,however, of -o,onewhich types two different masks assumethatMyc. spelling -/o/ laterpassed to -/u/(Ar 'o'Aovrv),and one which remained (Ar 8'Ko).2I SWG 3)
*t'is> aug *kwis>

in greater Therewas a ruleof earlyGreek(whichwe will discuss withwhichthe withRisch #io) in accordance detailin connection like[t'] whenitappeared 22 to something waspalatalized *kw labiovelar withinsave ArC identified which[t'] was in all dialects before /i/, failedto takeplace, and the herited [t]. In ArC thisidentification towards [s] continued. Thus in Ar we finda peculiar development Buck I98, as a following V\ in Mantinea (to be transcribed, character 66I.23, 25), oduEoL(ibid. I4); of 'sr: orLS I923: in forms (Schwyzer from Cleitoror Lousoi, Oc'st inscription withC in an early andspelled 679.23, 29) and at = T (Schwyzer (Buck#i6.4). In C we find:oLrs with of [t8] Ar inscriptions theidentification Hesychius. In later from surof the influence under in elsewhere as was G, perhaps made, t/ labioinvolving roundingDN dialects. Clearlyall developments in most themintact Myc. preserves I200, since tookplaceafter velars the mustpostdate and clearly, too, [t'] > [s] (or whatever) positions; ArC then is,who hasinnovated, in of [t']. The question development to identify or therestof G whichidentified [t'W [t'l with/s/. tendlinpr
21 Myc. forms such as pe-mo(PY Ebi 52) besidepe-ma(PY Er3I2.2) "seed" (= sperma) apply show thatMyc. did not consistently and a-mo(KN So7485) "wheel" (= harmo) information. > /o/only when it conveysno grammatical the conditionthat/a/ 22 There is some question as to what the actual phonetic change was in this case. W. S. Allen (Lingua 7 [I958] II3-I33) speaks of palatalizationof *kwto a labiopalatal, feelsthatlabialihim correctly, while L. Hyman (Language46 [I970] 7I), if I understand zation and palatalization are incompatible. In any event *kw> t in a palatalizing of otherpalatalizations. and did so in a way reminiscent (before/i/), environment

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

587

itsold/t/.Thisisa difficult with butsince in many other question, dialects a fact before (saveNEG) this which [t'] couldalsooccur /e/, almost certainly aidedin identifying [t'] with [t];andsince C, inthis diverging from Ar, the hadno [t']before I presume that /e/; innovating change was to haltthepassage andto identify it of [t'] towards /s/, instead with /t/, with the help ofthe perhaps lessdistinctively palatalized[te] < *kwe much from which may have notdiffered phonetically [te] < *te. Ifso,then this innovation failed to reach andwe Cyprus, maygather that thereason is that time outof forthis C wasat this touch with the rest after Dorian oftheGreek settlement hence world, ofthe the wewillrequire Peloponnese. Hence, too, chronology:
I) 2)

3) Identification of [t'] with[t]in /t/

inthePeloponnese Dorian presence

[kP] > [t ]/

~[i]

(Risch#i o),

TheArC,or SWG,change, was then, though seemingly innovatory, in fact a simple andreprecontinuation oftendencies already present sents a preservation of those and we haveto do with tendencies, conservatism thanwithinnovation. rather We will fillout the relative andfurther chronology discuss the palatalization of *k' below

" seven." It is therefore Lat. septem a conservative tendency, in that it continues a PG rule,one thateither had been deletedor givenup or was constantly entirely elsewhere, beingrepressed because, among other ittended tointerfere with the things, ofcermorphological clarity tain words. Examples ofthe lossofinternal -/s/occur most frequently inLaconian, butappear alsoinArgolic, Elean,Arcadian and Cypriote: Laconian E"TotEhE (SchwyzerI923: I4), vLKahas (SchwyzerI2.3) (ibid. Hohot'8ata (ibid. I2), etc.; Argolic'pahtapt'Sas' i5) evhip8ohats(Schwyzer 97), cio1FEhE (Schwyzer 80.3), 9J-avppo'v (Schwyzer 89.2I), etc.;Elean bvya8Ev'av--L (Schwyzer 424.6), a8EaArChaLE(ibid.I2), iiOl4aoraal (=o irorva0a&-Schwyzer 425.33), etc.;inArthere isbutone

SWG4) /S/ > /h/,+ Thischange, a tendency clearly on the post-Mycenaean, represents '-' q which ofspeakers part ofSWG tocontinue thePG rule*/s/ > /h/ accounts for therelations o,iioat toLat.sequor "follow" and 5ira- to

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

588

WILLIAM

F. WYATT,

JR.

[I970

example-vo'Eor-rL forirouEcrt (Schwyzer 657.I2)-which Buck (56) prefers to regardas a specialcase of dissimilatory loss of thefirst a. Sucha lossseems highly to me,and I feelthat unlikely we haveto do herewiththe surfacing of an inherited linguistic elsewhere tendency avoided, C examples include: 7TOEXO0LEVOV 679.I9), 'OvUjos' (Schwyzer (Schwyzer 683.3), qpovE'O- (Schwyzer combina685.4), andinsentence tion:Ka ac(v)'t Ta- vX`p6v(Schwyzer inWyatt examples 679.5), other I964: VyvE/aS Aafr4 (Hsch.)showsthatthechangeoccasionally affected eveninitial /s/. If we plot on a map of classical Greecethe areasin whichthechange occurs.

Lo c r. Phoc. Ach.
*i:..:.:*...-.,..[...-.

T B Cor.
..,....,.

East T A

Mea.
.....

Cret.

Theran

Coan
Rhod.

isrestricted we find butfor thechange tothePeloponnese, that, Cyprus, in whichMycenaean and especially to thoseareasof thePeloponnese settlement and civilization weremostfirmly established.I therefore regard this change as a peculiarly sub-Mycenaean feature whichwas a or at characteristic of thespeechof thosepeopleswho wereoverrun, ofthe ofa NWG idiom. It isa feature least outnumbered, byspeakers sincethe substrate overwhichNWG speech was laid. Furthermore, in Laconian and towardsloss of /s/is most pronounced tendency is mostclosedialectally, I venture to suggest thatCypriote Cypriote, ofthat of SWG whichwas be a development, and mayindeed variety inLaconia.23 spoken
23 Otherfeatures 56-57,passageof [w] of Lac. Elean Arg. (e.g., rhotacism-Buck to theSWG substrate, alsoattributable to [L]-Buck 47, t > a-Buck 59) areprobably

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

589

SWG 5) Kas'for rKal "and" (ibid. inWyatt collection I964: I73), andinMantinean 5-a complete Ar(Schwyzer 66I.I9), Kd (ibid. inAronly 17); elsewhere Kat occurs.
It is now quiteclearthatKasderives from earlier dart, and theonly questionis what is its relation to Kaal. Thereare three suggestions knownto me: Ka-t < Kat+ Et (WyattI964); *Ka'Trt iS original, and
Kat< *Katr< *Kai-t with metathesis of thefinal twosounds; Kas"< KaTit as byruleI) above(Kiparsky: Glotta 44 [I967] I33): *Ka't iS andKat < *Kai-t bydissimilatory original lossof/t/as inArgolic 'not' besideiro-it(C. J.Ruijgh, Atudes surlagrammaire dugrec etle vocabulaire mycenien 33I-33 [Amsterdam I967]). I nowfeel isright, that Ruijgh
Kas

occursin C (e.g., Schwyzer 679.I)

beside Ka before vowels

and thattherefore the preservation of *kati(>Kac) in ArC is an archaistic feature of thosedialects. Ar replacement of Kasg by Kac neednotbe influenced by D neighbors, butmayrepresent eventhere thechoice ofcompeting Kat. Ifso,C hasagain failed tomoveforward withAr. We cannot know whether Myc. had *kati (or *kasi). SWG 6) Dativewithprepositions instead of thegenitive

It is a feature commonto Ar and C to use thedative instead of the genitive withtheprepositions a&rt 7rEpt ivrep vio' 7rapa 7TE&a Evti (Buck IO8). This changefromthe situation prevalent in all other dialects apparently represents in ArC a merger of dativeand genitive intoone " locative"casewhichis usedafter prepositions, and whichis opposedto boththeaccusative, whichcontinues to be usedwithprepositionsto denote action directed towardan object, and to the genitive,which is no longer a prepositional case. Clearly this tendencyto simplify the grammar-likethe Latin-represents an innovation. Whether ornottheinnovation istobe attributed toMyc. as well is yetuncertain, and depends uponhow one chooses to interpretexpressions containing pa-ro. I prefer to hold withthosewho feelthat theinnovation is ofMycenaean date.24
andweresetin motion by thelossof-/s/. Since, however, they affect individual DN dialects anddo notaffect ArC,they belong to thehistories of these and arenot dialects, to be listed as SWG characteristics. 24In so doingI followVilborg (I22) and Docs. (go) as against Householder (Glotta
38 [I959] i-io), thoughnot withouthesitation. A numberofthefeatures listedby Buck

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

590

WILLIAM

F. WYATT,

JR.

[1970

ismatched or< ar,-toi< -tai,8E'KO < bya case inMyc.:both have

Of the SWG characteristics just mentioned, I, 2, 4 (?) and 6 are innovations.Andofthese butnotwithIv,and2 are onlyi, witha7rv we cannot definitely of Myc. date. 3 and 4 arecertainly later, know about5, and about6 opinions will differ.The casesin 2 areperhaps in later andmostworthy ArC surprising ofnote. Everycase,or type,
&SKa.

Not one ofthese SWG changes of theGreek spread to therest world. of the And yet,givenwhat musthave been the immense prestige dialect ofPylosandMycenae, Greeks we should haveexpected other to in haveemulated did and fact their linguistic usage. That they not, thatthey did adoptinnovations originating in SEG, maywell rankas one ofthegreater surprises ofG dialectal history.25
II

All of the innovations discussed thusfar originated in SG, and particularly SEG. Most of the otherchanges listedby Risch as at least potentially early are eithergeneralGreek tendencies, and/or
(i44)

as characteristic of ArC arein fact preserved archaisms andhence areonlynegative characteristics. Theyinclude:
SWG 7) ovv for o'SE - see below MG 8.

SWG 8) -Kp5ErSg for-KpaT27g-see belowNEG ii. SWG 9) arv-see belowMG II. SWG io) Ov-see belowMG ii. SWG ii) -kt inflection ofcontract verbs-Rischi8. SWG I2) sv==LSE-Risch I2. SWG I3) X forcompensatorily lengthened E. Thisis notan archaism really, butis nota feature ofthemoreprogressive G dialects. See belowMG I. SWG I4) E'S = E' /-C. This was the regularphoneticdevelopment in those dialects whichfailed to create preconsonantal s<. See belowMG 7. SWG i5) subjunctive in -s, --q- seebelowMG io. - seebelowMG 4. SWG i6) article as relative Others havebeentreated in Risch#2 (7Tro) #3(vat) #8(foA-); crts already will come withRisch#To. up in connection 25 Thereare a fewcases of peculiarly Myc. innovations whichfailedto spread:cf. Vilborg22-23, and especially Risch "Les differences dialectales dans le mycenien"
- see below MG 9. SWG i7) 17TO'AtS

Proceedings oftheCambridge Colloquium on Mycenaean L. R. Palmer andJ. ChadStudies, Eds. (Cambridge wick, I966) I 50-57 and G. Nagy "On Dialectal in Pylian Anomalies Texts" Attie Memorie del 1i Congresso Internazionale di Micenologia, Roma 27 Settembre-3 Ottobre 1967 [Incunabula Graeca 25] Rome i969) 663-79 on the possibilityof different

dialects-one innovatory-in Myceneaan.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK

DIALECTS

591

world. The isoglosses are restricted to a smallareaof theGreek include: g) *RR> :R -RR (*esmi> imi -em-mi)
io)
II) I2)

*ke

> Te -re innominative Ot -rot plural ev > 'v& with accusative


a >?1 a

I3)
I4)

ns > :s-is
>ar}
{a

IS)
I6) I7) I8)

-ua
-uev

in aorist of-~c verbs


-lESi

verbs 4Ew -7,& in contract


-W

infirst active plural

I9)
20) 9, 10, II,

*r>pa-po &o infuture ofverbs

Greek I7, I8 involve tendencies whichmerely general failed to reachcertain areasof Greece, henceare in general unlocalizable as to pointoforigin. 13, I5 (?) (Al) 20 (DN) arerestricted to smallareas as Rischpoints of theGreek world.I2-Is arelate, out; I6 isindeterminatebut seemsrather at all but represents late; and I9 is no isogloss pastincorrect interpretations ofthedata. 9) RR>:R all of Greece Thischange affects saveforT andL, andis therefore a pan-Greek phonological a tendency tendency, that failed to affect the ofGreece, most areas northeasterly though itdidreach Boeotia. Risch datesthechangea little laterthanthefirst eight, though still priorto I do notknow. It seemsto me impossible I200; forwhatreasons to datethis change-itcouldbe quitelate-, butsince it failed L, toaffect I suspect-though this is notnecessary fortheargument-that it took place at a timewhenL was not yetin theinnovating areaof Greek, thoughlaterthanthe establishment of NEG speechin Boeotia. C evidence is lacking, and if C also failedto undergothechange, we would be free, not forced, to assumea dateposterior though to the isolationof Cyprus,hence perhapsafterDorian settlement of the Peloponnese. It would thenhave affected only the more or less advanced and central culturally areasof Greece.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

592

WILLIAM

F. WYATT,

JR.

[I970

S~~

ofmaps the conflation either This I) and6) orof2) and maprepresents notbe accidental, andourviewofthedate may b) above. Thisfact we whether choose ofthe will mapI) ormap 2), change depend upon ornot, C experienced this we feel whether that is to say, that change
Io)

*kwe > Te

(Coleman#9,Buck #57)

B ofLinear where naoo theevidence Rischputsthis after change this itgo. A mapdisplaying that andC demands change-assunming innovation--shows: It/ to be the

in all probability the labio-velars post-date involving Developments on be and hence could in the Dorian displayed Peloponnese, presence
in SEG either Greece. They probably a map of classical originated B did that era Peloponnese. It is perhaps or in geometric surprising

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

VVo. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

593

notgo along DN andhence with all therest with ofGreek, andthe in theNE thelabio-velar wasgiven reason is that series probably up earlier than Inthat have ina sense elsewhere. event Aewill innovated, while andnotshared in thecommon C willhavebeenleft behind G topalatalize aswellasbefore note tendency before 28 below). (cf. /e/ /i/ We must be as clear these as possible about thecourse which particular ofwhich changes aretwothings we canbe sure took. There andwhich areshared of G: *kwi by all dialects ti everywhere yields and *kw>pI- a everywhere; (savein ArC in part), theenclitic particle *-kwe as-Te (Lejeune everywhere appears 4I, Buck,62). We included inthe first rule):
I) kw> t/ [i]

therefore need two rules(on the understanding that*-kweis to be

2) kw>pf

la]

rule applies NEGand i), while C apply rule 2), hence:


ia)
2a)

The onlydivergence occurs in which before ofG /e/, most position


kw> t/[i]
kw>p/[aj

or,rephrased:
> t/lirontvowels ia) kw 2a) kw > p/Aow vowels

glosses, we needonlystatethat mostof G appliedia) and 2a) in that order,with 2a) applying before/e/, vacuously while NEG and C appliedthemin thereverse order. Some linguists would be content withthis solution as is,butI am notcompletely.Thoughitisneatand accounts forthedata,it does not explainthedistribution of thedata andhistorical developments involving them and seems insufficient as a result. I prefer, therefore, to assume that thecommon G (for because of Myc. it cannot be PG) tendency, in ruleform, expressed was (and we no longer needto distinguish tworules):

in order to account Clearly, fortheattested distribution of theiso-

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

594 *kw> t'/-[i]

WILLIAM

F. WYATT,

JR.

[I970

> p/-elsewhere26

in NEG and C, and account interruption Theserules appliedwithout an in thoseareas. In the restof Greece, though, for developments innovation tookplace,and ruleone was understood to applynotonly of *kw causing thepalatalization before butto all front vowels, thus /i/, in all The these were: areas, then, to [t'] alsobefore rules /e/. *kw vowels > t'/_front > p/_elsewhere of linguistic We may now expand the chronology changesand events withSWG innovation #3: givenabovein connection historical > kw/-elsewhere Dorian inPeloponnese presence ~e [Peloponnese] 3a) *kw > tp e > p/ elsewhere > t'/_[i] 3b) *kw NEG > p/_elsewhere
2)

I)

*kw> t'/4[i]

4)

inC save [t'] > /t/everywhere

in all other intact that*kw, positions, This chronology though implies whileDN speech of palatalization theprocess had begunto undergo is forso assuming in thePeloponnese. The reason was notyetpresent in treatment of *kwe-betweenAr and C. If this the divergence the then or can be ignored, can in anyway be got around divergence canbe given DoriansinthePeloponnese with tie-in up. In anyandall lies withthoseareasof it is at leastclearthattheinnovation events aloneis of and thatthatfact before *kw /e/, Greecewhichpalatalized fordialectology.27 importance
26 *kw and any cases of *kwwhich happened to /u/, could not appear beforeor after fall in eitherof thesepositionswere perceived already in pre-Mycenaeantimes as /k/, 36 2nd de phonetique grecque Cf. Lejeune, Traite' < *kwekwlos. as in KvcAo& < *kwukwlos I.285-3I7, especially 293-7 Ed., (Paris I955) and Memoiresde philologiemyce'nienne (Paris I958). in NEG, and thatthe 27 It is forinstance possible thatthe labializationrule originated the labio-velarsat a timewhen theywere no more in NEG. restof Greek stillpreserved first, *kwi was broughtto SG, affecting trend We would thenassumethatthepalatalizing

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]
iI)

THE PREHISTORY

OF THE GREEK

DIALECTS

595

ot's i-ortOin nominative plural

(Coleman#28,Buck#5)

Ol occurs inArC Al T L andCretan, whilero0 isconfined toDN B andtheThessalian ofThessaliotis.Once againwe areforced tochoose between two maps

Locr. Phoc. Ach. Elean Lac.

T B Cor.
.:A-:..:
* **.*... @@ss* * . .* . *s-

Meg. Arg. Theron Coan Rhod. .....


..

East T BA M .g A n. The:rar

L I

?.

PIn.og

. . . . .A

Ach:

Elea .Lac. ..: C

Ar Cret.

iSn Ri:Rh6d

and then theother collocations.In themeantime twothings happened: Cyprus was cut off from themainstream oftheGreek world (andindependently developed irein accord. ancewith the NEG rule);andthen theenvironment of *kw > [t'] wasextended toinclude allfront vowels.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

596

WILLIAM

F. WYATT,

JR.

[I970

the PIE to replaceodat with-rot rat,thusrecovering tiveoperated of thehomophonous neuter because situation. o'couldnotpassto -roviewand on thewholeitisbestto adoptthestandard To. Nonetheless DN atleast onlylate, DN B, itreached that, though odreached assume there save in thatarea of DN whichprovided and failedto prevail I conservative, rather to Crete.28 Since Cretanis in general settlers o' rot' overnewlyintroduced that theDN choiceofinherited presume ofo' from expulsion andtotal thefinal was effected onlylate,andthat both postdated the DN realmfailedto reachCrete,and as a result on Crete and in the Peloponnese. For the early Dorian presence of -rot map,butforthetriumph ofot we can useourschematic spread

it Forthough isthe innovation. form we feel upon which depending PG must havehad*toi: that is clear PIE had*toi,itis notclear that andthat in PG by *hoi, PIE *toiwasreplaced that it isjustas likely demonstraofthe many--forms incertain pressure ofthe dialects the

map.29 the classical weneed

A -.

28 Coleman in thismatter. of Cretaninnovation the possibility (74-75) mentions unnecessary. butis probably matters, andwouldsimplify possible, Suchis indeed explanations and psychological in mentalistic to indulge unduly wishing 29 Without aloneof that Cretan thefact to explain itdifficult I find linguistic phenomena, ofpurely in the other of oc'as un-Dorian a conscious one assumes expulsion DN has od,unless feature, hasa given ifoneDN dialect that oftheDorianworld:onewouldexpect areas would haveit as well. Thereare,e.g., the DN dialects of theother one or another in and Doris severior between Doris mitior cases of E'vs (Risch #i2) and thedistinction the of o1willhavetakenplace(after theexpulsion thevowel system. If I am right.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

VoI.

IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

597

East T

. .'

~C ret.

n..

an

an

I2)

Buck#2,3) accusative >' gV with (Coleman#33,

thechange since We needtheclassical development mapforthis in have indeed areas. It may inAegean-facing took begun only place

Locr. Phoc. Ach ...

T B

East T

:A .Elean ,,,,Str
. ~~~~~~~~~~~.C..N..

settlement of Crete)at a timewhenspeakers of DN becameawarethatother dialects existed, and that thepeoplespeaking them always usedoZwherethey wereaccustomed to using bothofandoto. Theytherefore expelled this intrusive and foreign ot in the of Doriansolidarity.Thismaybe truealso of other interests D features, particularly conservative ones, for itistomea striking feature ofD andonethat requires explanation, that it is in itsmainoutlines so monolithic, in morphological particularly matters.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

598

WILLIAM

F. WYATT,

JR.

[I970

inAI(Porzig likely ormore thence, i 5o). Coleman spread and Argolic indepenearly date andfavors the given as unlikely spread (88)regards but is possible, innovation independent Of course dent innovation. ideaof favors Porzig's (seemap)tomerather areaaffected theunified culturally which ofGreece area inthose appears spread.Ingeneral Evs' is date period. Risch's archaic the during andadvanced central were tobecorrect. likely asusual
I3)

a > zq (Coleman#I, Buck #36)

on linguistic I2oo-900 butclearly tobetween change Rischdates this grounds. Sincea and noton dialectological reasoning (phonological) we areawhich areaofGreece (AI), a small to -qonlyin one small passes yetto come was and others casesso farconsidered other knowfrom (to a verylate date),it could have veryhomogeneous linguistically ofRR to :R-#9) and passage theearly (after taken placeat anytime was due to thefact to spread to spread.30Itsfailure havefailed simply of Ae speechin doingso by thepresence from thatit was prevented andAsiaMinor. There in boththePeloponnese LesbosandD speech innovation. ofthis is no needofa map to showtheextent in thatless as the first This sectionhas not been so satisfactory Apparently emerge. seemto of dialects groupings 9 (RR> defulite in SWG or SEG. either :R) movedtowardNEG, havingoriginated 9, at leastin some later than late-probably > te)wasrelatively I0 (*kve C toreach itfailed weakinthat ofGreece-andwasalsorelatively areas NEG evenas muchas 9 did. ii (ot) was a andfailed alsoto penetrate failed to makeheadwayin NWG whichrewhich tendency pan-G relait was in all probability theoldersystem: jected it and retained thanI2 (6VS) which earlier thanio, and was certainly earlier tively and earlier, in thePeloponnese, too, than Dorian settlement requires conor approximately as we shallsee,is laterthan, I3 (a-> 'q) which, with,I4: *pantia> [pansa]> [pasal> miva raTua. It is temporary as it does only AI. affecting geographically, also highlyrestricted
30 Of course of [a:] to [oe:]-a on whatone means.The fronting this depends took early car very place inBoston (= [ka])-probably English likethat seen fronting

inherited place ofthis [E:] couldnothavetaken [at:]with in SEG, buttheidentification > > /a:/ or [pansa] /awa/ onlyafter until a new [a:] = /a:/arose. And thishappened

> [pa:sal= /pa: [pasa] sa/.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

S99

One feature ofthis towhich we willcomebacklater, is that section, phonetic to reach themorphoinnovations NEG,while (9, io) failed logical innovation resistance in NWG. B tooka middle (ii) faced DN in9 andii, butAe inIo. position, joining It is perhaps that theE'v > Evs' area(#I2) includes, interesting butis notcoterminous theareain which inirpo's (#2), and with, p appears inwhich furthermore includes the area to/: passed /RW/ R/(Coleman Buck It is clear that these are late andare re#i6, 49-50). changes to Aegean-facing stricted dialects: all theyprobably originated in Ionia. It is further that ofthese samedialects interesting many have result of compensatory from lengthening but #9. Lessimpressive, still is thefact that HOcoELinteresting, in the occurs in name sea-god's of these many dialects (Coleman 7I), and that in the Z'qv-appears and genitive dative ofZEv'S insome (Coleman Buck 93). H7oaEL#26, occurs alsoin Achaean andAr,inthe latter inherited, Z-rv(orZav-) alsoinElean. Homeric andgenerally cultinfluence seems likely, at inthese least last twocases. Thetable displays the facts.
TABLEI
EVs.

in common/e:/(spelledEl) as thecontraction of E+ E (and product :/, o + o),whether ov,from ornotthey hadEl(andov)as the /1 spelled

I A Argive Theran (E)Cretan Coan Rhodian L Corinthian Megarian Laconian

+ + + + + + + + + + +

wrpo + + + + +

:R + + + + + ?

Et

+ + + +

HocTE-

+ + + ?

(+)
+ + + +

+ + + +

Z'qv+ +

+ + +

Of moreinterest to me at least is i i (rot)which hasthesame geographical extension as #6(Ka) and#7(0Ka), if, that is,we choose toregard theconsonant as thedecisive factor rather than the vowel in the latter case. Inboth #7and#i Iwe find aninnovation common to allG failing tomake its wayinDN B. And, there areat interestingly, least twoother cases inwhich wefind DN B operating asa unit.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

600

WILLIAM

F. WYATT,

JR.

[I970

#37) DN B #i 7Tpo'a5a vs. lrpo'o?2E(v) (Coleman

not in -tkaoccuronlyin DN, though irpoa'a and otheradverbs and Tpo'uJE atMegara atTroizene forEfk,7pocTooE always, oirETto5 appear influence holdAttic we canprobably lastforms (BuckI04-5): forthese in Al L. AboutArC we cannot responsible.7,TpOfE(v) etc. occurs
be absolutelysure, but Ar does have Ov'aUEv (Schwyzer 654.23) and a peocraKo5EV (Schwyzer 664.7); Schwyzer (I939: 628) also mentions
7Tpoo5

I havebeenunableto locate, -rpouor9ayEvE's(Schwyzer which of 7Tpo'o-2E to our assumption obstacle 66i.33, 35), theonlypossible Unto provide counter-evidence. also in ArC, is too uncertain forB, butsinceit wentwithDN in fortunately we haveno evidence it did so that thefinal vowel,we mayassume #6and #7in lowering of NWG to lower final-/e/in here as well. It was a tendency we see also withya besideyE of other adverbial words,a tendency dialects (Buck 24).3I DN B #2 (Coleman#27,Buck #9) 7TpaTrogvs. -7TpCJ0ro withDN B we are embarthecase whendealing As is frequently unable to andareconsequently ofthePG form rassed byourignorance on thehistory differ is theinnovation. Opinions widely decidewhich butit is at least 2.609-I0), words(Schwyzer of these I939: 595, Frisk a from bothwordsderive ofBuck(94),that be in spite or should clear, has and thatone groupof dialects (or areaof Greece) ancestor, single form:to ask thatone theoriginal and theother preserved innovated is to another another chooseone variant (allomorph), groupofdialects tothose lines aresosimilar dialectal sosince the asktoomuch. Themore irpJroS casethat be the itmust l others. Further for ot Troand drawn andthat cannot not*prowatos), 7TpairoS derives from *proatos (probably mostsimple and at leastnot directly.It seemstherefore so derive, an earlier 7rTpdrog. Thissolution economical to assume *1Tpda-roSfor out of the but failsto get us completely in its simplicity, is elegant as everwe wereaboutwhattheearlier woods,forwe are as ignorant motivate was: we can easily *7rpoa-rosg (after 7TporEpos),butnot form that would seem to dictate dimicilior *nrnaaToC. and the etvmolooria
31 Schwyzer (I939: the (cf. 627 n. 4) to identify 628) is inclined,though reluctantly -Oa of DN B with IE *-dhawhich mayoccur in e!ia E'v6a9TE. Clearly I feel that he -a2E(v)in -Trpo'afE. is wrong in this,and thatwe need assume forPG only the suffix

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol.

IOI]

THE

PREHISTORY

OF THE GREEK DIALECTS

6oi

itmaybe. be an analogical of*7TpaaTos, asindeed reshaping A number as thePG form: ofconsiderations urgeus to assume *proatos i) we can do nothing witha *7apaaTog, can connect itwithno other has a lowering word in any IE language;2) [r] notoriously effect in manylanguages, and did so quite definitely (later)in the NW of Greece(Buck 23-24): it mayhave happened in thisword; 3) earlier B we havealready inNWG there seenelsewhere DN #i) that was (#6, a lowering ofmidvowelsin certain positions. It seems bestto assume thesameherealso, and posit*proatos> *praatos> 7rpaTos for DN B. The question is stillopen, though, whether irp6-roSmaynot be an etymological of phonologically reshaping created*praatos. Since a discussion ofthis pointwould takeus too farafield at present, we may leaveitopenforthetime being.3Z Thesetwo seemto be theonlychanges shared by DN B alone(in addition ofcourse to Risch#6,7, I I), andbothinvolve of thelowering a mid-vowel. This is not much,but we shallsee laterinstances of NWG features whichB failedto adoptand NEG innovations which failed to reach DN or B.
*wpoaTog

m There seemslittle pointin discussing in greatdetailRisch'sitems I4-I6 since they areall recent developments: raavaa (I4) isan archaism wherever it occurs;-/ae/-> -a:/-(I5) seems restricted to AI, though it has sometimes been supposed forArC as well(WyattI964: I79-80); thedetails ofaorists to -4w verbs (i6) areso complex that little definite canbe saidaboutthis matter.
I4) 7Tacvora> 7Taora 7TaZua

(Coleman #I3)

Both Risch and Coleman treatthischangeunderthe headingof developments involving *pantia,butclearly is incorrect.Though this
32 The onlyshred of evidence I can concoct fortheassumption that-/oa/> -/aa/in PG andthat regularly therefore in T L AI ArC is analogically *proatos restored from regular *praatos is theword irpadvis' 'withthefacedownwards' whichseems to stem from*pro+ ane:s; cf.Frisk I.I2I, 2.594 and Chantraine I968:97 for 4ne:s. For irpo' doesnotordinarily andwe should experience elision, to developto /f expect :/ -/oa:/(*flEAkdoa > fEAdw-Lejeune Traite 235). -/aa:/, would develop to /a:/ however,

(to /E:/in Ionic).

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

602

WILLIAM

F. WYATT,

JR.

[1970

offaaaa,that prehistory inthe point atsome occur didindeed *pantia in PG, by its phonological already probably was replaced, form as Risch is late, -/ns/ofsecondary *pansa. Simplification successor itbelowlabel asweshall Greek andispan-Greek-Mainstream holds,
such as Crete, backwaters all areas save severalcultural and affects and Arcadia. That Argos failedto go along is perhaps Thessaly,
surprising.

#I3 (*/a:/>/E:/) postdates thechange holdsthat Risch(64) further place(and trivially), in AI, but thismaynotbe thecase. In thefirst Greek worldand henceis area the in a small of #I3 applies only very thechangemayhave taken in mostof Greece. Secondly, irrelevant areas: in different times did takeplace,at different place,andprobably uponthe burst and didnotsuddenly one imagines, gradually, itspread the passageof /a:/to /E:/ and most importantly, scene. Thirdly, of [x:] (< */a:/)with[E:] identification say, the is to to (a -q)-that contrast (= -i)-could not have taken place withouta significant by the triggered = /jE:1was doubtless between[ae:] and [a:]. [ae:] the[a:] of a new /a:/. This new /a:/maybe precisely development in *[pa:sa], of thelossof nasalization whicharosein Al as theresult (in Al) mustbave been: *[pansa]> development for the phonetic *[pa:sa]> [pa:sa]. Thus it is possiblethatPG */a:/was identified is andRisch'schronology > raiua, *pansa withPG */E:/in Al because incorrect.It is, however,also possiblethat/awa/>/a:/ therefore stillholds before [pa:sa]> [pa:sa] in whicheventRisch'schronology [se:]= /E:/,[pa:sa]> [pa:sa], [awa] theparticulars, good. Whatever in Al. havebeenall butcontemporary > [a:] must
I5)
as> a

or 7) (Coleman#5)

a E> a onlyin Al, because is notparticularly important Thischange else it developsto -q.33 Apparently and is late there:everywhere
33 In fact the evidence is restricted to a-contract verbs of which we have secure about ArC T L. L to be sure, examplesonly forAl and DN: henceRisch's uncertainty has -rtpat(insc.) yEAatg(? gramm.-cf. Thumb-SchererI03), but the manner of defromthatin Al DN. In factI suspectthat -&etg < *-ayeis clensionin thisdialectdiffers and -iec < *-ayei had contractedto *-ais and *-ai already in PG. It was the newly created-acls -aet thatcontractedto -is -a in Al but to -ns -n in DN. ClearlyL and theseforms,did not experiencethissecond round of T, since theydid not rethematize contractions.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK

DIALECTS

603

thenormaltendency, thepanG tendency, was foras to contract to [e:] and thenmergewithinherited [E:]. This tendency was, howexisted ever, blockedin Al becausethere there an [ae:]fronted already of [ae] must have been from inherited [a:]. The contraction product somewhat lowerand perhaps backer than[a:]-we maysymbolize it heardin New Englandspeechforwords by [a:], a soundsometimes like ask (J. S. Kenyon,American ioth ed. 27, I76-I84 Pronunciation [AnnArborI9S2])-, and hencemerged in AI with[a:] rather than can be diswith [ae:]. The interaction of panG and AI tendencies playedas follows: [a:] > [:I PanG i) [ae]> [a:] AI 2)- [a:] > [a:] PanG 2) [a:] > [a:]
AI
I)

PanG 3) [X:] > [6:]

i6) Aorist in -;cl (Coleman#40, to verbs Buck#4) Matters here aremorethan confused because we do notknow usually thePG situation-indeed there maynothavebeenone,sincemanyof in question theverbs aroseonlylate(Risch73 note22)-, andbecause seemto have been at work, What is really analogyand borrowing needed,as Kretschmer pointedout in I909 (GlottaI.30 note i), is a examination oftheentire thorough question.34Nonetheless we may be ableto makesufficient for ourpurposes headway byoversimplifying thedatapresented slightly byBuck(II5-i6, Classical 2 [I9o7] Philology Risch (73) and Schwyzer (I939: 737-38). There are three 25I-52), categories of dialects: i) those with-ea in all aorists of verbs in -4w. Here are to be included T C, mostof DN, and partof B. 2) Those dialects whichhave -ea generally, but -aaa whena voiceless velar precedes: Coan, Arg.,Ar. Arg.hasextended this ruleto include also the voiced velars. 3) Those dialects-AI L-which have only, or mostly, in -~c. In these -s(s)- from verbs dialects theunderlying form
34 Suchan examination must withno preconceptions, proceed and must concentrate on whatthestem-form ofthevarious words was felt tobe synchronically in thevarious dialects withno regard paid to etymologies, realor fancied. I suspect that thepicture willturn out to be evenmorecomplex than is currently thought.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

604

WILLIAM

F. WYATT,

JR.

[1970

how itcameto have little a velar:it matters contained form underlying as havingone. In Homer(and Hesiod) one or whyit was regarded GramI948: 340-4I = P. Chantraine, both typesoccur(Chantraine that inassuming Vol. i), and we areprobablyjustified hom6rique, maire case-ea, isthearchaism, in this in Homer,thus element thenon-Ionic usingtheeveryday withI, he is merely whenHomeragrees and that discourse. colloquial in contemporary normal form that case2) thenwe fmd as original, the-6a forms Ifwe maytake -ks-with in thatit replaces situation of theoriginal is a modification of and case 3) is a modification velarprecedes; -ts-whena (voiceless) sounds. ofthepreceding -tsregardless with all-ksitreplaces 2) in that which changes of changes, chronology Hence we needthefollowing times. at various places applyinvarious
i) 2)

andetymoofetymology regardless a dental, contained of theverb and2) the of dialects the in too, i) Clearly, words. related logically

everywhere -{a hence velar, ... dental Ar Al L Arg.Coan > velar ... velar velar
-4c =

hence-*tsa 3) -4- = dental, > * -ssa 4 * -tsa

Al L (B) Ar AI L

6) -/ss/-.> -/s/

> -/ssa/ /V s) -Jsa/

Al L B Ar Al

L B, as well forthefact (Buck II6) that to account Rule S) is inserted suchas KaAE'craaL o,poxaaL:I believe as Homer,have -ss-in aorists in -ss-,and verbs of other thepresence requires thatthisdevelopment TCAE'acruat like verbs from (save in numbers arisen any canhave verbs such of -4Coas a dental. the identification onlyafter froma stem*teles-) *ts,butss< s) is yetproblematic. (with-ir-- from The B situation withpartofB) have C DN T (together that then, reveal, The rules 2) as original-, we wishtoregard situation-unless theoriginal retained areais,as The innovating have innovated. in and B, part AI while L, in and is at itsmostadvanced AI L, a in thesouth, usual,completely enoughbefore. whichwe haveseenfrequently situation
IV

of category of Risch'sfourth We can now moveto a consideration The a date. thatfor whichhe was unableto establish isoglosses,

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. 1I11

THE PREHISTORY

OF THE GREEK

DIALECTS

605

reason forthis is that fortwoofthem ofcontract (i8-lt inflection and i9-[rI) there verbs is agreement dialects forwhich among his reasoning predicts AI DN andAe ArC; andin disagreement, namely dialectal area(DN) stands from theothers: an innovation when apart (orthe retention ofanarchaism) isrestricted toonedialect itcan area, havetaken place(all elsebeingequal) at anytime whatsoever and have simply failed tospread far. very One isogloss, can be removed however, from this at the bundle outset: haddeveloped inPG topa or ap, andthere already isno *[r] agreement between Ae on the onehand andArConthe other,Four rules only areneeded toaccount for the dialectal facts (Wyatt 197Ia):
I)
2)

the other two and 2o-E'w inthe plural but (I7-jiEs- infirst one future)

pre-Greek
PG

4) Lesbian Aeolic

3) Peloponnesian South G

[ra]

ran ror

' /ra/ [fl> */ar/

> /o/ /a/ / -Jabial

> /o/ / rC /a/

[ro]/_(C)Cu

consider developments involving ofthe interrelation[r]indiscussions


shipof theG dialects. With the otherthreeisoglosses matters are not so straightforward.
-REV

Alc. /Spo'p4[oua (possible in 306 fr.I4 ii I7) "run and 1-4WVTES 'having cut' among others. Thus we need no longerto (I29.I5)

Rule i) states that inPG PIE *[r]haddeveloped already an/a/, generallybefore the/r/, though sometimes after it. Rule 2) states that /a/ was rounded to /o/ whennextto /r/ ifthesucceeding syllable contained a /u/: this rule accounts forcrorpvv1tu for*starnumi, as wellas isolated casessuchas thepersonal nameepocroVcrrpolroS (Bechtel Die Griechischen ig2i:25; F. Bechtel, Dialekte, Vol. I). Rule 3) is merely SWG 2) above. Rule 4), the most problematic, accounts for

I7)

in thefirst person plural active (Coleman#43,Buck#II)

It was a strong tendency of PG to replace theinherited *mes of the first plural active ending oftheverbwitha newform *men, possibly in originthe secondary ending. That *mes was originally everywhere present in G seems clearly indicated by Hom. -pEo-Oa beside-,uEOain

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

6o6

WILLIAM

F. WYATT,

JR.

[I970

thereplaceDN resisted pluralmiddle. Andjust as clearly the first in this conservative highly with-REv and is therefore mentof -uesg ofG optfor usis: whendidtherest facing then regard. The question but sincethe we cannotanswerthatquestion, -,LEv? Unfortunately foraboutB all of G (we think, as to reach was so powerful innovation I8), it would seemto have been we do not know-Thumb-Scherer above in conmentioned thantheinnovations earlier possibly early, maywell DN retained that with #7and i i. The reason junction -pEsverbs -HEv in athematic ending be thattheinfinitive (#4)had already an obstacleto the spreadof first thereand offered been introduced for as Coleman reason, be theentire though, -tLEv. Thiscannot, person T at least did affect not of homonymy pointsout (76), considerations both andin theinfinitive, plural person bothin thefirst has-/-LEv which andthematic. athematic
18)

verbs stem ofcontract of durative Inflection (Coleman#39,Buck#65)

T L ArC(BuckI23-24).

in in DN B AI, athematically are inflected thematically Suchverbs

thatboth (7I-72) in feeling and Risch maybe right is mostdifficult, wereknownin PG. thematic (-E'w) and athematic (-Nqp)inflections at ourselves we herefind is complicated by thefactthat The problem conand resulting rules(lossof */y/ of phonological theintersection of inflexion tendency(thematic traction)with a morphological what have restored may denominative verbs). Analogy generally maybe apposite. i) To label destroyed.A fewremarks phonology up if it conjures is misleading in ArC L T athematic the declension L PLF ftAEt in active is I.23) rWi7tu: (Sa. the third singular of pictures rqlcu (Schwyzer634 A 35) UrEbavoL (Schwyzer647.3I). 2) ArC and in L (T is is thematic agree,fortheinfinitive L T do not completely the G to inflect optative tended All ArC. in athematic 3) ambiguous), 'Wv (Schwyzer 4) Ar &VEv8& 665 C 4) T KaTOlKEW`VVIf athematically. 7%jo(PLF 36) and a&lK"El (PLF I.20) (IG IX 2 5I4.3) Sappho's 7TO (other alsoa longvowelthematic conjugation existed show thatthere

verbs these with outdevelopments Sorting

inBuckI24-25). forms

vowel of the stemwas long, and thatthe ending-,ul of the final

the that we may conclude last fact this From

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol.

IOI]

THE PREHISTORY

OF THE GREEK

DIALECTS

607

that Itismybelief first singular may -co. merely replace developments incertain more inArCandL, though differing details, closely represent thephonological ofG, while DN andAI havereanalyzed tendencies and restored theoriginal theresult seenin theother dialects PIE In anyevent I agree ofAI inflexion. Rischthat theagreement with andDN ina pan-Greek atthe trend (extension ofthematic conjugation L T andArCwereout expense ofathematic) cannot be accidental. ofthemainstream ofGreek andhence linguistic progress apparently, failed toshare inthis development.35 theaboveexposition, that--qtu Clearly implying (or -'co) is the olderform, is notan argument to be accepted even and,in order a number cases inwhich the tentatively, requires that weisolate ofother Al andDN, same ora similar dialectal configuration emerges.Since allorover over ofits willagree, wemust date these part extent, changes ofDN speakers relatively late, thus clearly after thesettlement in the For convenience we maylabelthose which Peloponnese. dialects share inthe various innovations Mainstream Greek (MG). MG of course always existed, and manyof thefeatures already discussed (e.g.,/mm/>/: m-#9) which failed to influence certain areasbelonged to it. These,though, and can be kept wereearly from thecases separate about tobe discussed. Others, though lateall all ofGreece affected to greater extents: right, or lesser here we may think ofthe development ofa definite article useof (Buckioo) andthe
adverbs as prepositions to " govern" oneor morecases. These tending latter areimportant, butdo notofcourse serve to isolate linguistically residual wordsand tendencies.I think hererather ofcasessuchas the
35 Clearly ofthis is unsatisfactory mytreatment on linguistic isogloss grounds, butit willunfortunately thepresent, haveto suffice for andis inanyevent sufficient on dialecofwhatI feel cf.n. 33 above. The tological grounds. Fora portion actually happened in theG developments, first it seems to me (to restrict to e-contracts), stage myself was that*-eyo was analogically to *-eyo,probably thea-stems, after changed though also after thefuture andaorist thelossof *-y-contraction possibly *-eso, *-esa. After took place,yielding *-eo *-eis -*ei *-eomen *-ete *-eonti (> *-emen *ete *-entiat leastin in ArC, and L innovated ArC T B). This system was apparently preserved onlyin theconjugation (occasionally) replacing *-eo with*-emi. Thiscaused to appear much likethat ofthethematic optative, whereupon thefirst of a-stems singular developed to -aimi. In MG, however, to *-eo *-eomen, *-eo *-eomen, etc.,developed etc.,andsince in those dialects from *-eis, *-ei, andparticularly *-e:te,differed -qsw --qt--qre,because there thedistinction between theverbs [e:] and [E:] was preserved, wererethematized to -eo -eeis-eei, etc.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

6o8

WILLIAM

F. WYATT,

JR.

[I970

of Eo to as well as vowelsand thepassage consonants use of E's before collected have not of such cases-I probably to. Therearea number my interwhether may well be some question themall-and there cases. The features in individual is correct of developments pretation do tendto show thatT ArC, and to a lesser I shalldiscuss, though, I areasof Greeceduring, B L werenotpartof theinnovating extent period.36 thearchaic suppose, MG
i)

(Coleman#2) lengthening E> Et by compensatory

upon the a nasalor resonant before was lengthened When PG */e/ in quality havebeena [e:] different must theresult lossof *js/or *jy/, it was characteristic that PIE. I consider from the[E:] inherited from and thatit was a markof cultural thisdistinction, of MG to preserve a to preserve thusfailing backwaters to merge[e:] and [E:] in /E:/, Phoc., and Loc. as well as BecauseCor. andMeg., vocalicdistinction. at one timehavebeenpartof all ofDN must thedistinction, preserve MG, and onlylaterbecomeisolated.37This meansthatT B L and that It is interesting out of the mainstream. ArC were the dialects

verbs. or - 'coincontract alsohave-mit they

MG 2)

(Coleman9S, Buck 8I-82) Apocopeof prepositions

I herespeakonly of thosecaseswhichare not due to haplology, and OV< Ka-ra -rov < Kasuch as K' -ro'V for haplologicalforms, rather I think dialects. in occur many < 7Tort rov, Fro rov < iror rov (Schwyzer in 'TTE-ro?S FOLKLarTatS ,TTE&M 1923: 66i.6) and ofAr 7TE from Hom. KaK KEbaA -S (II. I 8.24)andB Kay yav(IG VII 2407.9-Thumboccurin T B L Ar(C)? Hom. Lac. El., and 46). Suchforms Scherer in the restof Greece. I by the longerforms were clearly replaced oftheprepositions, ofsomeatleast forms infact theshorter that believe of thePG shape (or ov-below MG ii), represent suchas trap Kai- aJv forthepresent necessary thesewords,but thisis not an assumption in werefavored forms thelonger argument.We needholdonlythat
36 I have in thissection and have not made an independent reliedon handbooks, inmyaccount. newdata. Hencethere maybe insufficiencies for search ofallinscriptions indeDN [e:] > /E:/ is that 37 I speak whenI sayisolated. All I intend figuratively of that themerger in ArC andAe. It is quite likely ofsimilar developments pendently influence. was due to SWG substrate and [E:] in DorisSeverior fe:1

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol.

101]

THE PREHISTORY

OF THE GREEK

DIALECTS

609

MG and the shorter feltto be infra dig. or (in the case of Homer) archaic. MG 3) Preservation ofwc8a (Coleman#32,Buck#67) inassuming Coleman(ioi) maybe correct that 7TE&a isaninnovation in the faceof inherited eTraE (cf.also Frisk2.38 ), but thehistorical relation of these distribution wordsis unimportant: counts. onlytheir TE8EL occurs as a living form (Buck I07) in L B, and henceafortiori in T, though itis notattested in Ar,and mostlikely in C; itoccurs there; also in Arg. Cret. Ther. It was preserved, no longerunderstood, in compounds in Rhod. Coan Calymnos Meg. Sicily,a factwhich provesthatit was once knownall overtheGreekworld,though we have no proofforthisassertion forAl. It was, though, givenup in thoseareas of Greecewhich were culturally in the archaic central period. MG 4) Demonstrative usedas relative (Article) (Coleman#49, Buck ioi) It is clearenough that os (< *yos-Frisk 2.434)is inherited PIE from (cf.Skt.yah), andthat itistheexpected form oftherelative. Nonethelessin certain outofthewayplaces, andinIonia,notan outoftheway place,henceprobably in mostof Greece, there arosethehabitalso of usingthedemonstrative pronoun as a relative. In mostof Greece this was so resolutely tendency put down thatno traceof it remains, but we do haveexamples from ArC T B L Hom. andI. It is to be noted thatit occursonlyin thosedialects whichhave ot in the nominative plural:Heraclean and Cyrenaean examples represent later independent since developments neither Laconian norTheran, their mother dialects, show any traceof the demonstrative used as a relative;wherever else it occursin DN it is late,and not partof thesamedevelopment withwhich we arehereconcerned.

MG s)

1Fo > Eo (Coleman #6,Buck i-9z2)

The tendency herewas to shorten thelongvowel in hiatus withor withoutcompensatory lengthening of a succeeding shortvowel.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

6io

WILLIAM

F. WYATT,

JR.

[I970

by this T B L Ar(C)? (archaic forms in) HomerEl. werenotaffected remained until so in thatthere -/w/onlytrivially change, C perhaps quitelate.

Buck MG 6) Eo> to> Eo (Coleman 2I-23) #4,


in all of Greekfor [e] to be raisedto [I] There was a tendency [i] in /o/such thatit tendedto mergewithearlier (vel sim.)before all of Greeceseems proved affected that position. Thatthetendency 7co fromEt1ut 'to go' forwhichwe shouldexpect by thesubjunctive was supported by tendency *Eco (< *eyo). Here the phonological factthatthe root of thisverbwas /i/. A simthe morphological was felt pressure, but in the oppositedirection, ilar morphological < *wetesos, and the [i] was also in words like [wetios] <*wetehos of other forms and did identified withtheolder/e/ of theparadigm, rulewas thephonological areas,though, not pass to /i/. In certain without interference: (East) T B L C Cret.Lac. allowed to operate and allowing matters innovatory, I regard morphological considering unchecked conservative. to operate phonology MG 7) E1> EK/_C (Coleman#i8,96; Buck 83-84)

In most areas of Greece E'6 appearedbeforevowels, EK before for ce consonants.This, however,was not the originalsituation, ' farthest' in PG as is proved in all environments by&crxaros appeared occurs eveninAI. In MG. i968: 380) which < *eks-katos (Chantraine EK the to before after was consonants, perhaps simplified however, E'e because tended to pass too, c'/_C analogyof Ev Et., and perhaps, to es and thus become homophonouswith es < &vsl/C. It is in whichEK failedto develop to note thatthosedialects interesting dialects in part Loc. Phoc.) areforthemost partthose (T B ArC Cret., withaccusative didnotdevelopEvsg which (Risch#I2-ArC T B Loc. are conPhoc. Aet.-Coleman 87). Clearlythe two developments nected. or demonstrative MG 8) vv as a particle (Coleman#3o,Buck#44) in T ('rovESchwyzer occurs ofmoregeneral o?8E ov-in themeaning C 682 i6.i), A Ar rovvv ratvvv 657.30) (Schwyzer (Schwyzer 578 9)

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol.

IOI]

THE PREHISTORY

OF THE GREE(K DIALECTS

6II

and vv as a separate occursin B (Schwyzer particle 523 G I65) C in those (Schwyzer 683 6) Hom. Clearlyit is an archaicfeature dialects andwas early replaced else. everywhere MG 9) -7TiOAtso (Buck6I) The general of Greek, of etymology tendency whatever thedetails and pronunciation (SchwyzerI939: 325), was to use 7iAiOs, the shorter formand not vToWAs. Only backwaters like T ArC Cret. and (archaizing) Homer continued to maintain -io'Ats (and in part in use. 7ro'AE,uos) MG io) in -7 Thirdsingular active subjunctive (Buck II9)

Greek PIE a third inherited from of the singular subjunctive ending activevoice in -71(Schwyzer I939: 79I), a form whichnearly everywhere was replacedby -qt on theanalogy of theindicative-Et. Only

in T B Ar(C?)El. was theolderending preserved, at leastin part.


ii)

MG

Preservation of ov (Buck#63)

Onlyin a fewdialects at theexpense (T L ArC) was ov preserved of innovative avad(< *o'va'). It matters littlein thisregard whatthe ofthese relation forms to eachother was.38 It isinteresting, moreover, thatarv appears in thesesame dialects, thussuggesting thatitis the
more archaic form and that possibly Myc. a-pu and ArC adiTv5 may

be archaisms andnottheoutput ofSWG rulei).


I2)

MG

Genitive singular-oto

T and archaizing Only arch-conservative Homerpreserve this form whichexperienced contraction to /l/everywhere else. It would be possible all these to display archaic on a map, features but it would consumespace,and a tablewill do just as well. The
38 There are,it seems, If ava' is to be compared twopossibilities. directly with Avestan ana OP and Goth. ana OE on,then thePIE form must havebeen*on(a), andthis in MG to ana. If,though, havebeenassimilated is to be compared *on(a) must o?v with Skt. anu 'entlang,'thenwe are freeto assume:*anu> *onu (by assimilation of lipfrom*ona withsecondary rounding) > on, a.vax as in other againwill havearisen -/a/ prepositional wordslike7rapa KacaL.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

TABLE 2

Archaic Features
I
2

T + + ? + + + + + + +
+
+ I2

C + ? ? ? + + + + +
+
II

Ar + + + + +

B + + + + + +

L + + + + + +

Hom. Cret. + + + + + + + +

> q Compensatorily lengthened /e/ ApocopeofPrepositions

37rE&a

as Relative 4 Demonstrative Sno remains 6 so> to consonants before 7 Ec6 8 vv


0AtsoVT

+ + + +
+
IO

+ + +
+
+

IO II
I2

- i subjunctive
ov
-OLO

+ +

Total(Maximal)

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

6I3

results are,I think, for all along scarcely we haveknown surprising, T ArCwere that andthat backwater B L showa number dialects, of a good conservative tendencies. TheHomeric would have been figure dealhigher hadit notbeenforthevery influence (probably) strong I exerted which onhislanguage. contemporary
20)

-or'W inthe sigmatic future(Coleman #4I, Buck #I2)

7TvuEVTat Ar.FrogsI22I) andHom. (E&auE-rat II. 2.393), to be sure, havesuchforms, butthey areisolated and occuronlyin themiddle of verbswhichhaveactivepresents (Schwyzer I939: 785). We arefree to imagine that this is whereitbegan, andthat (for somereason) an-eswas inserted between in these rootand ending forms as in liquidand nasalverbs(Schwyzer I939: 786), and that thenwith the loss of -scontraction tookplace: *pleue'somai> *pleue'omai> (inA).39 *pleuouimai When -s- was analogically restored, rAEvorovikat a form(or resulted, type)whichwas thenin general brought intolinewithother futures theregular by utilizing thematic future. In DN, however, this -selo(perhaps old) type, supported by futures likefaAE'wfrom and /a3AAwo lV7KLo, particularly spread so as toembrace allfutures. ?avEo,uat from It is perhaps a toss-up to decidewhether DN has hereinnovated or
39 That -/s/is not original in theseforms is provedby caseslike A KAavaO/,aL Ar. Peaceio8i) whichshould have,bytheregular rules phonological of (KAavaoov'/,ElOa G., appeared as *K '0tLaL (cf. 4E'K-qafrom KaLco 'burn'< *ekausa andKiparsky Language 43 [i967] 623-29). My guess is that the"Doric future" haditsorigin in themiddle of in-/w/-. In order verbs toaccount for AI KAavaovEILca we must assume thefollowing:

Early, andprobably contemporary with creation ofthe #I7, is the "Doric" future which occurs inallDN dialects for which future forms areattested. A (q!veovoiat Eur.Med. 604, 7TAEvorovLtEfa Thuc. I.I43,

present klawyo 2) klawyo 3) klawy6 klaiw6


i)

future aorist klawesomai eklausa klaweomai eklaa klauseomai eklausa

4) KAaLLi

: PG : PG withlossof-/s/: PG with-/s/restored after theroot as seenin thepresent KAavaotLaL EKAavaa: -ao,uat after theactive ofother in verbs -aw

The onlydifference between DN and therest of G is that DN preserved stage3) and extended it-becausemore fully marked as a future-to other verbs originally displaying -so like wpacew SelcoW,

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

6I4

WILLIAM

F. WYATT,

JR.

[I970

inanother innovated have rest ofG may allthe since been conservative extension the consider must we that appear it does way. Nonetheless a NWG innovation. futures ofsigmative active of-eo-to thefuture ofwhatmayhavebeenthe A andHom.showtraces The fact that was thechange that us to assume in no wayforces situation original retention the ona parwith itchronologically toconsider late:I prefer
of -yesg.

here. may be mentioned DN features Two more

Buck#40) 905Ev(Coleman athematic plural DN I) Third #45, verbs ofathematic active plural G dialects person thethird In most andtheimaorists thethird ofthematic after plural wasremodeled inthis whether connection little verb "to be." Itmatters ofthe perfect
theresult was
WcKav

show a all these dialects 679.27),OVE&EIKaEv aorist. of the third pluralactiveformof the athematic remodeling formmaintain (and in Homer) did theinherited Only in D dialects evenin N. itwas replaced itself: (T-Schwyzer 566 2.4):
DN 2)
at' TP
Ka

SO.4), avE1

(Phoc.-Buck #53.5), Eouav (AI L-Alc. PLF


KaTEr&?jav

Eav (B-Schwyzer I923: 440.II),

(C-Schwyzer

(Buck #I5)

to decide L at KE' TL&' (Buck I40). It is impossible C E KE uts., rtsis theoriginal, againit is morelikely, though whichof theseorders thatDN has here amongall non-DN dialects, giventheagreement
innovated.40

rts, Ar EL' ' a'v thatwe mightlabel thisDN B #3) as opposed to Al cEa'v

in B, such in DN (andsometimes wordorder Thiswas thenormal

In this section,then,we have found one new dialectalentity, of and have beenable also to add to thenumber Mainstream Greek, whichwe saw abovein DN B. Therearenow NWG characteristics
40OtherNWG features evenD. NWG 3) -OlS in third and do notaffect arelater low-class (Coleman #25 Buck #26),a later like v7dv-otS andadjectives nouns declension likethatseenin NEG 4 below and thenextfootnote. of thegrammar simplification NWG 5) 32I.2). I923: NWG 4) Ec> /Ip (Buck #25) as in a'piic (Schwyzer held D, and is generally (Coleman#2i Buck #I3) doesinclude T'E1Topesfor-reTacapeg seenin Skt.catvdrah. PIE *kwetwores from inherited to be an archaism,

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

6I5

nine ofthem of which I list below, including theextent ofthespread the feature andwhether itisinnovatory ornot.
minor phonological change -REv in infinitive (?) archaism (Risch#8) DN BTL plural (Risch -rot nominative #II) DN B archaism (Coleman #27)DN B minor phonological 'zparog change Coleman#37)DN B minor phonological -&a, -'ra adverbs change (TL in oera) plural (Risch#i7) DN archaism -PEg first active DN -CuEWfuture (Risch#720) innovation (?) D E5cYv c8ov archaism DN (?) innovation ai r&sKa
(Risch#6,7) DN B

2)

3)
4) 5) 6)

7) 8) 9)

In thislistwe findonlythree possible several innovations, minor andthree phonological changes, inwhich cases archaisms, innovations in therest oftheGreek instituted world failed to penetrate theNW corner.What is more,all these archaisms are morphological, as aremost indeed features on thelist, andwe maytherefore conclude thatNWG was morphologically in the conservative, particularly verbal system.Thisisallthe more important when we remove since, theinnovations of other those dialects dialects, aregoingto appear andmore more likeDN. ThusRisch's statement at older (7I) that ofthe D andAe looked periods language more alike istrue, butisnot forit wouldbe true significantly true, ofanyother dialect areaone cared toselect andcompare with D.
v

Furthermore, though Rischisolated SG and NWG features, he completely ignored innovations and archaisms whichwouldhave madeit possible forhimto identify features pointing to a NEG linguistic area. Had he doneso,he might havefound rather more differences between NEG and NWG evenat earlier stages of the language.We willcomplete the picture ofareal linguistic features by listing here NEG features, those features namely which Buck(147-148) attributes toAe.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

6i6

WILLIAM

F.

WYATT,

JR.

[I970

NEG

i)

Perfect Participle -cv, -ovros

(Buck#S8)

(Coleman68-69) thatthischangedid not It is perhaps surprising takeplace in moreareasof theGreekworldthanit did. As it is, it Homer (Buck II8). It I and (occasionally) took place in Ae, north feet,and like is a simplification akinto Eng.footsfor of thegrammar learners of the by immature have beencreated footsmustcontinually corrected. language. Onlyin NEG was itnotsystematically
NEG2)
7a fori4'a

(Buck #52)

*i form, of the PIE demonstrative in whatever The preservation, I923: I79 by Cret.(Schwyzer shared (seenin Lat. is) is an archaism - r . aiv-r4v . a3rcv . KJ7rptot-Hsch.), Mesavp VII.23, VIII.8,),C (v senian(? IG v I 1390, 206), Homer (II. 6.422, etc.). MG had dis"that"> "thatone" > "one" development cardedit. The semantic there ofuta by ta. purely NEG, as is thereplacement is,however,
NEG 3)
t> E/Cp-C

(Buck #S6)

reads simplyPE= Pt (I47), but in fact the Buck's formulation thep. This preceding he adduces examples (25) all showa consonant to tautosyllabic p, a thistendency thatwe shouldrestrict suggests I923: (Schwyzer borneout,e.g.,by T 'Aptort'6v restriction 597) and and inallAe dialects, occur Examples (Schwyzer 'Aptarovo'ot 590.I). thatwhenL has pt (e.g., in KpLVVW in assuming we maybe justified change I (orMG) influence.Ifthis 620.28), itdoesso under Schwyzer feelthata phonetic I rather is innovatory, it is a minorinnovation: in NEG. wenton unchecked put down elsewhere tendency -Eror (Coleman#25, Buck#6I) NEG 4) dative plural from theIE pointofviewis quiteclear, That-Eootis an innovation also in N foritappears is notso clear, butthat itis a NEG innovation andHomer Corinthian colonies) (Phoc.E. Locr.Elean),D (Cyrenaean, a tendency on to represent rather (Coleman96-97,Buck 89). It seems clearthe to renewand render thepartof G speakers morphologically of seenalso in thespread a tendency dative stems, pluralof consonant
-ots in N (above, n. 40).
-Eaor must have been more common in

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOIl

THE

PREHISTORY

OP THE GREEK

DIALECTS

617

earlier butwas discarded, times, as low class, perhaps nearly everyNE oftheG world where. Onlyin theremote didthis innovation
takeroot.4' NEG5) (Buck#53)
KVppOV-

t>&/.V

Schwyzer 590.20) and L (7rE'ppoXoS Sa. PLF io6, jtErEppa Sa. or Alc. PLF 2 p. 292) that post-consonantal tended to become[i] andto lose /i/ value with syllabic concomitant of thepreceding lengthening consonant (Buck,26, Schwyzer In so doingthese 1939: 274). dialects are an ancient recovering PIE clusters process whereby involving were [j] simplified (bypalatalization) to affricates in thestops and geminates in theresonants.We are likely to be correct ifwe assume (withNagy I07-I2: G. Nagy, Greek Dialects and the Transformation of an IndoProcess European thechange[i]> [i]/.V was [Cambridge I970]) that of wider extent, originally but thatin all dialects (save T and L) depalatalization (i.e., [i]> [i]) took place at a relatively early date. Only remnants (Homericscansion:& ov for 8&4,u0ov I1. 12.2I3, = --- Od. 4.229; A foppas,UTEppos Atyv7rTr'q < fopEcs, urEpEOSj of theearlier situation survive. Thus we are dealing withan archaism: depalatalization (Nagy's term)did not reach T and L, thoughit affected all the rest of G. It is interesting that degemination of resonants (Risch#9)likewise failed to reach T andL. NEG 6)
T>

It is characteristic of T (7roAAtos-Schwyzer558.13,

pluralof verbs #l/v in third

A development [t]> [t']/n. is notunusual (cf Skt.panthah"path" cognate withG 7rTOVoS, Lat.pons),but occurs in G onlyin T andB.
4I There is no needto go through great in order gymnastics to explain theorigin and spread of-caac andto suppose -oLaL as theorigin ofthis form (cf.Schwyzer I939: 564). It doubtless arose where it isfrequently in FE7TEaaL FerEaaL andspread seen, namely from in PG-but was generalized there-already onlyin NEG PG at theexpense of theinherited forms. -caac contains an extra from -/s/thepointofviewofG as against e.g. forthestem 7roAuan, was FEP(E) + ending (-og -l -ov -aaL). In all other obliquecases of s-stem nouns,though, theending beganwitha vowel after and as a result -/e/-,

of courseit spread, in thoseareasof Greecewhereit caughton, becauseof itsgreat in preserving utility themorphemic constitution ofwords (like*possi) whoseroothad disappeared ofthemorphophonemic because rules ofGreek.

another-/e/was " reintroduced"into thedativeplural,thusgivingFE7TE-EaaL. Thence

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

6i8

WILLIAM

F. WYATT,

JR.

[I970

so that inthose dialects, participle even does itoccur inthe Norindeed as wellas phonomorphologically theenvironment we must restrict thisdevelopment to explain to me impossible logically.It seems inonesingle environwithout phonetic assuming that [t] > [t'](=/th/) mark of as the understood morphologically itwasthen ment andthat in whichthatphonetic environment the morphological category plural to [nt'i] in thethird [nti] passed I feelin fact that occurred. as the perceived andthat [t']wasthen ending ofverbs, active primary of thethird third plural mark andhence to all other spread plural this development It maywellbe that endings (data:BuckII2-I4). of/t/, wasthe NEG response tothe SG tendency towards palatalization plural (cfJ. in thethird in fact originated andthat palatalization this something 26 [I962] I33-45, who supposes Jensen, ActaOrientalia inArCandAI L. toall [ti]only similar for Hittite).It spread
NEG 7) [E:]> [e:] (Coleman #32) #2, Buck

oflater G developof-qto Et([E:]> [e:]),anticipatory Theraising inT andB (Buck innovatory, 25). Itisclearly took place only ments, L was minor It probably innovation. but is a rather beganafter inthe SG sphere. already of some however because brief, Allofthe above discussion, required the ofinnovations toNEG. All atleast have peculiar them appearance mentioned archaisms, other features byBuck(147-I48) are important features most ofPG: we neednotinclude andindeed arefor the part I include extent of the inourfinal list ofNEG features.42 them Again nature. feature and its the
42 Other NEG characteristics include: NEG 8) (*kwe>pe)=Risch #IO. of the thanthe genitive rather adjective singular NEG 9) The use of a patronymic rather noun(Buck I34-35). That thisis an archaism nameor a patronymic father's (Vilborg is shown by theMyc. situation thanan innovation I5I). it to NEG in thewordsin which peculiar NEG io) po= pa. Thisis not an isogloss

of eepa- at leastin theretention NEG ii) 9epa- = 9apa- (Buck45). Aralsoshares andArC andL alsohaveKpE-r- forKpar- (L KpE1TOS- Alc. I4I.3, ArC in proper names, I968: 423-4, (Chantraine e-grade of theexpected names). The preservation personal givenup by MG. feature 578-79)is an archaistic DN alone. orwith ofGreek therest either with B hasinnovated In thefollowing cases = Risch#9. of */mm/) NEG 12) (retention

appears (cf. Wyatt I97Ia).

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]
i)
2)

THE PREHISTORY

OF THE GREEK

DIALECTS

6I9

perfect particle -cov (Buck#58) ta for (Buck#52) p la 3) > e/Cr-C (Buck#56) 4) dative -EaUU (Buck plural #6i) (Buck#53) 5) ? > L/_V > /th/ 6) /t/ inthird
plural 7) [e:] > [e:]

TB T B T B T B T

L L L L L

low class innovation (lowclass) innovation archaism low class innovation archaism
minor innovation innovation minor

(Buck #33) T B (Buck #32) T B

Numbers 6) and 7) arerelatively areremoved, late,andifthey there is onlyone realinnovation in thelist, thereplacement of,da withta: the othertwo innovations are not restricted to NEG, but were low classsimplifications of the morphology which sprangup in many areasof theGreekworld,butwhichwereput down elsewhere. For therest NEG was highly in thephonological conservative, particularly sphere. It was as conservative as NWG was conphonologically servative Indeed if one were to combineNEG morphologically. withNWG morphology, phonology one would come verycloseto reconstructing PG. Almostall innovations beganin thesouth.
VI

It is now at lasttimeto sumup all theprevious sections and tryto reacha final conclusion aboutthedevelopment oftheG dialects. We have,in thebody of thetext,discussed a largenumber of linguistic not all of whichare relevant features, to a decision concerning the
NEG = Risch#i8. NEG I4) (-pc inflection ofcontract verbs) NEG i5) (preservation ofov) =MG vii. NEG i6) (preservation of airv)cf.MG #ii. NEG i7) (preservation ofKE) = Risch#6. B andL share thenexttwo features in common. NEG i8) (-/ss/in 4'KcAEaaa) (cf. Risch #i6). This innovation was shared byall SG as well,and is therefore not a NEG feature.It was almost inevitable in L, buttheB development is obscure (to me). NEG i9) (preservation of7Te8c) MG #3. T andB share a fewfeatures whicheither did notreach L or werediscarded there. NEG 20) (-pkev as thematic infinitive) cf.Risch#4. NEG 2i) (yi'vvpuatfor In fact this is theform onewouldexpect inPG from y6yvopuaL) PIE *gigttomai (cf.*onoma > ovvgua), butin all dialects saveT andB the/o/ wasrestored after the-/o/ofother thematic forms.

ofthe NE ofGreece, for occurs dyp4o widely (Coleman ioo).

I3)

yp

= =a'p'

Thisisogloss simply states thata

p4o

did not reachsome

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

620

WILLIAM

F. WYATT,

JR.

[I970

development ofdialectal differentiation inGreece. It is,for instance, clearthat MG (seeabove,pages607-13) together withRisch#i8 (-Esw for and Aegean Greek (see above, page 599) together with -rnpt) Risch (*ens) are, because of no importance for earlier dialectal late, #12 (areal) differences. *pansa>lTa(-)cra likewise is late and can be included in MG. Furthermore all innovations restricted to a single dialectal such as all Risch in and entity #I3(a> -q Al) the SWGchanges catalogued above(seeabove,pages584-90)maybe omitted in our finalformulation sincetheycannot affect discussions of dialectal all the affinities. The following list,though stilllong,contains
TABLE 3

I 2

*ti> si
rpo'S

(Risch I) (Risch 2)

-en 4a inm.
b infin.-ai 6a aiv

3 /Ss/ > /s/

(Risch 3) (Risch4)

S ai>ei

(Risch 5)
(Risch6) (Risch7) (Risch 8) (SEG 2) (SEG 3) (SEG 4) (SEG 5) (Risch9) (Risch io)

'when' oI-' b 'when' -a 8 PoA9 *tu>Su


7a
I0 tEpoS

b Ka

AI + + + + + + + + + + + +

Ar + + + + + + + + + + + +

C + + ? + +

L + + +

DN

+ +

+ + + + +

+ +

+ +

++ ++

(SEGi)

? ?

II
12

ELKOaL
-KoaUoL

I3 I4 I5

-OtaL

*RR>-R
*kWe>te
ol for rot 7Tpo'o6Oa irpiros
-(ev -aEC)

i6
I7 i8
I9 20 2I 22
23 24

(Risch II) (DNB I)

-av for-v
at TlS. Ka

-WJv la
-Ecal

25

(NWG 2) (NEG I) (NEG 2) (NEG 4)

(Risch 20) (NWG I)

(DNB 2) (Risch I7)

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK

DIALECTS

62I

innovations we havediscussed abovewhich canbe ofimportance for early dialectology. From this list we seethat eleven ofdialects which groupings I emerge indescending list ofnumber ofdialects here order affected.
TABLE 4 i)

Dialects Al Ar CL TB

2)

3) AI Ar C L 4)

AI Ar C L T

7a o'T', i6 ot, i9 (?) -(LEV


I -Si, 2 7rpo'S9, 4b -ai,
lepOS., II

I9 (?) -fEV, 2I

Features -av for-v 8 BoA-

S) Al Ar 6) Al Ar C 7) AI Ar 8) AI L B D N 9) Io) L T B II) D N

L T B D N D N

7b-"when"-a
I4-:R,

9 -SU, Io

ElKOat

(?)

3 (?)-/ss/> /s/, 4a-infin,-en 3 (?)-/ss/> /s/, S-aiz> ei, 6a av


I2 -KoclOt, I3 -otau

I5-*kwe>te

6b, 17, I8 -/e/>-/a/, 22 asrtsg ca


23 -WV, 24 la, 25 -Ecac 20 -re'w, 22 al TLS Ka

The question ofcourse then is: inwhat didthese chronological order groupings arise? i.e.,in what order do we list theinnovations which caused the uniform PG tofragment originally into localdialects? AsI haveimplied there is no needto assume earlier, anychronological forisoglosses order, at thesametime developing canhavedifferent andcanspread strengths towider ornarrower areas. Or again, some mayhavedeveloped earlier butbeenweaker, while other later ones, because canhave stronger, spread further such that we tend tothink of them as earlier.It is atleast theoretically possible that thegroupings wehave established areinfact illusory, andthat allisoglosses developed atthe same time. This would beakin toholding that the Greek world fragmented suddenly intomany different unrelated forms of Greek. Such a linguistic event would be highly unusual, though, and, since we find repeatedly bundles ofisoglosses andnotisoglosses inisolation, we may assume thatour groupings do bearsomerelation to actual historical developments. A few observations areinorder. First ofall several innovatory areas aremutually exclusive with other innovatory areas: innovations originating inthe onedonotspread tothe other and vice-versa. Theysplit theGreek world without overlapping. We cantherefore reduce the number ofourgroups toeight byconsidering

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

622

WILLIAM

F. WYATT,

JR.

[I970

characterized as being negatively areas innovatory exclusive mutually byDN, not adopted in areas ArC innovates if visa visoneanother: Al byAI ArC,we mayconsider notadopted inareas andDN innovates characterized DN isnegatively while characterized, Al ArCpositively Table fact has in DN not or whether innovated. features, in these are areas:DN innovations dialectal exclusive themutually 5 shows right. onthe listed
TABLE 5

Dialects Al Ar C .L T B D Al Ar C L T B D D AI Ar C L TB T B D AI Ar C Al Ar C L T B D AI Ar C L T B D Al Ar C L T B D L T B D AI ArC

Features N N N N N N N N
20,22

6b,17,18,2z None 7b None None


I15

None

NEG features -/a/);20 -acrEw22 a' -nS Ka; the low-class (-/e/> DN were6b 7b l7 i8 toreach theonlyinnovations 23 2425. Thirdly, I4 (: R) there; originated 20 -acrE' 22 atl Trs! Ka, allofwhich > -/a/); (-/e/ MG: to fact in and belong are late changes which bothof > te), I5 (*kwe "to the" Dorianinvasion presumably daysatleast-prior intheearly progress.B was part of the by G linguistic DN was unaffected history, ofG linguistic periods in theearliest areaof Greece innovating that conservatism by thesamemorphological was overcome butlater MG when it wentwithDN until DN. Phonologically characterized developments at leastas regards out of the mainstream, it remained B was a NWG thelabio-velars.For muchof itshistory involving from earliertimes. inherited dialect with certainNEG features and whichI havelabeledas low class, theNEG characteristics Finally, do in D N as they to NEG, occurring totally whicharenotrestricted speakers amonglessprestigious literally and Homer,couldhavearisen
I8

6b 7b 17 savefor area innovating inthe AI isnoweachtime Secondly

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK DIALECTS

623

of thelanguageat any time.43 In mostareasmore prestigious and and linguistically conscious one imagines moreconservative speakers influenced theirless fortunately situated fellowsto give up their but in NE, for whatever barbarous solecisms, reason,such usages NEG characteristics prevailed. It is in fact that mybelief 23 24 25 are amongthe earliest to have arisen on Greeksoil,and thattheyarose among the lower classesalreadyin PG times. We may put these in ourchronological changes first list. of isoglosses bundles other will be more Arranging chronologically difficult, butit is at leastclearthat be no AI Ar must changes affecting thanthoseaffecting Ai ArC, and that earlier Al Ar changes involving DN must be later than those AI Aralone. Hencetherelative affecting chronology:
i)
2)

Al ArC

AI Ar
DN

3) AI Ar

AI L (I2 -KOUCOC I3 -otac) can be no earlier Changesaffecting than ArC and those AI couldbe quitelateindeed:I feelthat affecting L, they with or laterthan 2) and 3), and can, perhaps are contemporary be placed as 4) in thislist. It is possiblethatchanges arbitrarily, Al ArC areearlier affecting thanthoseaffecting Al ArC L, forotherwise we might findit difficult to explain a) whyL failed to innovate Al ArC inadopting with theathematic infinitive and -(e)n, b) whyitdid share withT B DN thelowering ofthe final vowelin'when.' Nevertheless thisassumption is not necessary becausenot all innovations so as to embrace spread an entire area,and itis perhaps bestto conflate these two groups and dub them chronologically Al ArC (L). By the
43 In thetext I havenotlabeled 'a for1ldaa low-class feature, though it mayhave been. Whenwe recall that in NEG dialects, > -/io/-/eo/andthat in them (morphoand//i// phonemic) bothappeared as (phonemic) /i/ before backvowels, we realize //e// thatta can have beenunderstood in NEG as //ea// or even//eya//. Thisform may thenhaveappeared to be themoreregular feminine of *hens *hen, i.e., *hens *(h)eya as if displaying a paradigm *hen, like *he:dens *he:deia *he:den. Surely, too,oudens oudemia ouden helpedout in thisdevelopment: theywereanalyzed as //oudem// plus or-//ia// on themodelof *talans *talania (> *talanya) *talan, etc., andoudeia arose. -//s// Givenoudeia andthemorphophonemic equivalence of//eya// and//iya//, ia 'one' wasa likelydevelopment.It was again a simplification of an irregular paradigm, and if we can regard suchsimplifications as low classfeatures, it,too,was a low classfeature.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

624

WILLIAM

F. WYATT,

JR.

[I970

L T B DN (7b) and combine we will in theNG sphere sametokens B D N in thesinglegroup(L T) B DN. This move makesgood aremadecontempor> -/a/ of-/e/ now all cases for sense, phonological we varies. By muchthesamereasoning ary: onlythearea affected (B) themAl ArC L T by labeling two groups our first can combine that to assume no reason seems there and ([L T] B) D N respectively: B, B are any earlierthanthosenot embracing changesembracing thoughsuch of coursecan be done.44 We thusend up with the ofchanges: chronology following
NEG AI ArC L T (B) 2) 3) AI ArC (L)
I) 4) AI Ar

([L T] B) D N

S) AIArD N
Al L

6)

spreading without which, changes alsowith3) arethose Contemporary all of SWG (see above pages 584-go; and beyondSWG, affected only Al with or laterthan6) are changesaffecting contemporary withMG (see above,pages (above,n. I9), 5) can be equatedroughly diversity ofhow dialectal picture We cannow givean approximate stagesof the languagea arose in the Greekworld. In the earliest the geographical PG was spokenover whatever uniform dialectally within however, time. Therewere, oftheG worldwas at that extent levels dialectal area,certain linguistic homogeneous this geographically The classes ofsociety. spoke'pure', in terms oflevels upper definable Greek,while certainof the lower classes-perhaps i.e. inherited, learnedGreekwho had recently non-Greek speakers originally such as those listedas of the grammar introduced simplifications also did not, in Table 3. The lowerlevelsof society numbers 23-25 ofdesignaadopttheupperclassmanner it was notnecessary, because nounor by thegenitive either a man'sparentage by a patronymic ting to usetheolderpossessive adjective butcontinued ofthefather's name, some nameor from from thefather's either (Buck I34-I35), derived Glotta A las and Aitchison characteristic (cf.TeAaELUJvLos outstanding
44As I shall do exempli gratiain the appendix.

607-I3)-

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]
42 [1964]

THE PREHISTORY

OF THE

GREEK DIALECTS

625

oldlabio-velars I do notknow, with labials savebefore itis though /i/ ifso,then enclitic -TEinthose entirely possible: dialects isa borrowing from thespeech ofthemore members ofthe prestigious community. Allthe features inlow-class justmentioned continued but failed speech, to make inroads inthespeech any(save oftheupper sporadic) classes. The G community was at this time still one in a geodialectally graphical, ina social though not sense. Thefirst divisions geographical toarise began when a number ofmorphological innovations in began the southern ofthe part G world, innovations which more or penetrated lessdeeply intothemorenortherly areas. Suchchanges included replacing thefirst person plural morpheme with-pEv (#ig), -,fiES thethird extending person athematic plural E1#Evto E5E(cr)av ending inherited toiwith hoiin thenominative ofthe (#2i), replacing plural demonstrative pronoun *hoke<*hokwe as (#i6), and reanalyzing thus *hokwe, toits too'T' leading passage we (#7). Itispossible, though haveconcluded otherwise above(6I4), that at this sametime *ai kwis ke(n) wasswitched inthe south to*aiken thus kwis, facilitating thelater reanalysis of atKEv as EL(K)av. Thesechanges embraced all of the southern areaofPG,andinthenorth affected those areas which later T L; butB, forwhatever generated reason, though adopting I9 (?) and2I, retained *toi andtheolder (?) *ai kwis at least in part. ke(n), These latter twochanges mayhavebeen later than the first two-and B nowinthe NWG sphere-but itisperhaps still more likely that they wereweaker in force andfailed to appeal to theyeomen ofBoeotia. Those Greeks resident still farther tothe NW ofthe Greek world were conservative highly were linguistically, not affected by thenewer inthesouth, developments andwere proud toretain the older system intact.Their sole effort atmorphological innovation-if such itwas was their creation of thattypeof future tense known later as the "Doricfuture" (#20). Phonologically, though, they were content to allowfinal -/e/ to be lowered to -/a/ in adverbial words, a lapsein which they werealways joinedby theBoeotians, andby theThessalians andLesbians in thewordfor"when." Theseare the only innovations attributable to a NG dialect: from theearliest dayson linguistic innovation wasrestricted tothe south. No newinnovations

I32-38). Whether had already in themorehomely they words s]vs.[mu:s];seeabove (cf.[hy: toconfuse page56I) begun the

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

626

WILLIAM

F. WYATT,

JR.

[I970

that adoptanyoftheinnovations areas aroseinB T D N, nordidthese not,that is to say,tillMG cameinto arosein thesouth, subsequently being. community, At thispoint,then,thoughstilla singlelinguistic in geodefinable differences dialectal some slight Greecedid possess SG innovating terms. There were in fact two dialects, graphical NG (T B L DN). In thenorththere (= AI ArC) and conservative wereforthemost but again theyprobably werefurther differences, sort. The old aristocracy partof a socialand not of a geographical spokea lesspure NWG (= DN), whilethepeasants spokeconservative of SG features an admixture of the same whichcontained variety (= NEG). Therewas no "Doric" or "Aeolic:" just upperclassand lowerclassNG. SG became NG andinnovative conservative The distinction between linguistically to innovate L, continued as SG, withor without greater static and unaffected by newerdevelopments.It whileNG remained ArC AI affected which innovations as I havesaid,that those is possible, SG plusL. Suchanassumption those than affecting butnotL areearlier structure andmorphological thephonological unnecessary: is,though, could that notall SG changes different ofL mayhavebeensufficiently which hadsuch on there. Whileall theSG developments catch easily Greece, on thelinguistic atlasof ancient effect a profound and lasting to all in SG L), (including were spreading SEG, originating probably in this area-again generSWG was innovating also,butinnovations on liketheNEG characteristics-failed to catch nature allyofa careless did not certainly however, to SEG. TheseSWG changes, and spread of SWG and between in communication speakers causeanydifficulty itimpossible for(e.g.) ELKOct to nordid they render SEG at this time, and developover thewhole area even thoughSWG said *kwetortos at this timeweredistinctly The onlydifferences SEG said *kwetratos. and did not prevent forthemostpart, affected pronunciation minor,
common innovation.

AI ArC L, C laterthanthedevelopments affecting At somepoint, and of the SG from the remainder isolated world, became linguistically of to develop-in so faras it did develop-independently continued as *aiken AI Arreanalyzed oftheG world. Duringthis therest period also simplified *eikanand developedEI(K) av (#56), and possibly

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY

OF THE GREEK

DIALECTS

627

* -ss-to * -s-,though it is of course that perhaps the equally likely absence ofreliable C aswell:inthe change affected from early spellings it is myguess that be sure, C preserved we cannot Cyprus though retained * -ss-. Italmost -RR-. certainly ofG dialectal Thelast that which stage prehistory, stage gavemost their distinctive character of thelocal dialects theclassical during arein general that common to Al DN and involves period, changes which involved > /:m/ sometimes B, changes > */mm/ (I4) and *1kw elimination of thelabiovelars. /t/ /./e/ (i5), thatis, thefinal In this meant that terms NWG had dialectological joinedthe innovating hadreplaced andinso doing areaofGreek SWG (byitself becoming NEG. I havelabeled the SWG) andexcluded this changes involving to other MG changes.Theyspread dialectal areas of Greece, entity inthe butnotalways same andindeed didnotalways intensity, affect evenall NWG dialects.We are notto think of all suchchanges or as contemporary. either as sudden It is forinstance since likely, they than embraced Ar,that I4 andI 5 areearlier of the many changes listed with above inconnection the ofMG. Later establishing changes affected limited Lac.alone, areas, e.g.,Al L, AI alone, andfall either on threshold of the classical thevery or within period it,andhence lie outside thescopeof this which is concerned investigation onlywith dialectal prehistory. torecognize We have, the then, ofG dialectal following stages prehistory: itsupper i) PG, with andlowerclass levels; 2) latePG, in which certain SG innovations affected some ofNG; 3) SG (? L + C) in which affected the changes only south, sometimes reaching C but not L, sometimes L but notC; 4) MG; 5) Classical Greek. Atnopoint inthe of G arewe enabled early prehistory better to understand developments byresorting to thefamily tree model of linguistic change towith gether itsconcepts such as Dorian, Aeolic, AI andArC. Indeed wenever canrefer toAeasa whole, andhave instead always tothink of L T B as separate entities, L a low-class PG dialect which later moved intotheinnovating sphere of G; T a low-class PG dialect which remained out of touch withtherest oftheG world from thevery earliest times; B, a low-class PG dialect which affected theconservatism characteristic of NWG. We can speakonlyof those dialects which continued NEG (low-class PG) characteristics throughout their

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

628

WILLIAM

F. WYATT,

JR.

[I970

which as that dialect negatively Doriccanbe characterized history. dialects The SG nearly intact. PG characteristics upper-class preserved oftheir for much butthey were positively, be characterized caneasily saw: SWG changes so,as Rischcorrectly or nearly history identical those changes associated when were for minor.Itisonly the most part dialectal major with MG took that tobeabletodiscern webegin place (likeAl > 71) verylatechanges and it is further linesdeveloping, we times which ustospeak ofDN Al ArC. In earlier finally enable SEG tendenciestendencies-generally canspeak onlyoflinguistic toa greater ora lesser into areas ofthe other which degree penetrated prehistory, endofG dialectal tree, visible atthe G world. Thefamily andhasflourished, history: itgrew, atthe endofG dialectal developed change. It is by thewavesof linguistic for2S00 years, nourished itaway. time these waves sweep that
APPENDIX

my investigation a few lapses, tokeep with I fear In the I have tried, text speculageographical historical and from any and torefrain solely linguistic nor entirely realistic pure isneither toremain sodoctrinally yet tion. And inGreek ofthe for our interest reasons for surely atleast one entirely honest, astothe location isour togain information desire prehistorical dialectology I willno longer oftheGreek appendix people.In this andmovements uses, andwillallow databe putonly to linguistic linguistic require that of the location and movements bundles ofisoglosses todefine and isoglosses isnot such useofthe data that though, peoples.Itistobe remembered, to andthat we may choose conclusions anyhistorical given bythedata, data asdothe far the same authority are and are from having draw suspect themselves. ofthis in early sections paper I have been using Theschematic maps that reasonable assumption relations notscale. It is a highly but mapped the itexpanded from andthat wasoriginally world the Greek small, quite associated time ofchanges size its classical toapproximately bythe PGarea was that thePG world in fact assume Mainstream Greek.I shall with allofThessaly: it may nothave embraced to Thessaly, restricted though form ofGreek ofthe earliest itwasapparently some time before speakers and southern Boeotia atThermopylae andsettle toturn able the were pass distances atvarious allhave been time settled, Greece.Atthis they may atthat time area and inthe aroud Larisa.Itwasinthat from Larisa, plains
thatNEG features (23
-COv24 'a 25 -Ecaat) arose,but caughton, as I

have

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol.

IOI]

THE PREHISTORY

OF THE GREEK

DIALECTS

629

supposed, only among the peasantry. At roughlythe same timecertain otherchanges theentire G world, (I9 -/Ev ? 2I -av for-v) spreadthrough savethat evenamongtheupperclasses, failed to reach DN (though they they did reachB) which may (Herod. i.56) have been locatedat thistimein werenotso powerful Hestiaeotis. Otherchanges I6 -o' for-rot) or (7a o-T' well received, and failedto reachbothDN and B whichmay(Thuc. I.I2) still in thevicinity ofArnein Thessaliotis.I presume havebeenspoken that L was at thistime undifferentiated fromT, and thatundifferentiated SG withT and L savethat it had notadoptedlow-class NEG (ArcAl), identical features, was spokenin Pelasgiotis, perhapson the shoresof the gulfof Pagasae. All the changesI have mentionedtook place withinPG in Thessaly: they arestages i) and 2) ofpage 627 above; they will be displayed on maps3-5 below.45 The next group of changes, thoseassociated with 3) of page 627, may have takensome timeto run theircourse. There are two chronological layersto be considered: thatin whichL innovates withNG and not with SG; and thatin whichL innovates with SG and not withNG. It will, I think, seem most naturalto supposethat SG-perhaps by dint of overThessaly and established colonieselsewhere population-movedaway from in thesouthof Greece,perhaps in Attica, perhaps in Atticaand portions of thePeloponnese, aswellas perhaps in Boeotia: they mayhavetraveled bysea. We cannot knowverymuchofa detailed nature concerning this movement, butsinceit coveredso wide an areaof territory withpresumably so smalla number of people,it musthave takensometime. In anyeventduring the earlier portion of thisperiodchange4a (infinitive -en)tookplacein'SG and > /s/), change3 (/ss/ while change7b (-/e/ > -/a/ in "when") took possibly at thesametime,or slightly later whenL had left placein all ofNG. Either Thessaly,changes6b 17 i8 (-/e/ > -/a/) 22 (*ai kliska) took place and affected DN and B, butnotT or L. Subsequently L, forlonghighly conservative, andin fact identical withT, SG. This switchof allegiancerequires-sincewe are joined innovating trying to account bothforlinguistic developments andhistorical movements of peoples-two populationshifts: L musthave crossedthe waterfrom Thessaly to AsiaMinor, andI must havecrossed to AsiaMinorfrom Attica.46
45 There is nothing us on dialectological preventing grounds from placing I4 (RR > :R) and i 5 (*kwe > pe) at this andassuming time, that merely they failed to reach Larisa andvicinity.We are,though, prevented by theevidence ofMycenaean from positing i S at this andbyinstances early date, ofRR dueto the"Aeolic Substrate" elsewhere in Greece from assigning I4 to this period. We wouldalsoand in anyevent prefer that L T (andwithI 5 B) be moredefinitely separated from therest of MG. 46 At thesametime-or perhaps slightly later(cf.Herod. i.56 and Thuc. i.i2)-B movedintoBoeotiaandDN intoDoris. It might be bestto assume that these move-

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

630

WILLIAM

F. WYATT,

JR.

[1970

withthe traditionally associated the"IonianMigration," words In other placea gooddealearlier.Themovehavetaken must "DorianInvasion," it didnotmean-though doubtless outofThessaly ofthese peoples ments endto communication withThessaly, havemeant-an couldtheoretically to Thessaly. spread innovations linguistic no further butit did meanthat changes ofSG unity I (*ti > si) 2 (rpo's) 4b (-aiin this longperiod During > su)io (LEpOS) II (E'Kool) tookplace. We must 8 (/oA-) 9 (*tu infinitive) over thinly class-spread ofa ruling culture-perhaps a homogeneous assume C, which period same theendofthis ofland. Towards large areas rather from themainland linguistically became cutoff earlier, hadbeencolonized ofthe > /s/) 3? (/ss/ inchanges toshare andfailed 5 (ai> ei) 6a (dv). Because to 1200. changes prior allthese ofMycenaean date we must evidence as the"Dorian that known is ofcourse Thefinal movement ofpeoples the in thePeloponnese, forDorianpresence whichaccounts Invasion" place no canhavetaken movement this andAsiaMinor. Clearly islands, does it neednotandprobably andMycenae: than thefallofPylos earlier for fall. their notaccount isnoneed ona mapchanges stages 3 andfollowing toplace affecting There i and2, maps will that butit maybe wellto havemapsofstages above, a dialectally that caused andthechanges showPG Thessaly schematically unPG. Map i shows differentiated intodialectally uniform PG to split itin dialectal the later PG with upon superimposed differentiated designations

.~~

~~B

over-articulated, still I havesupposed above. Map 2, though thepositions howthe ofthesame PG unity. Map 3 shows a better representation gives a first in theoriginal andspread, split causing first changes (23 24 25) arose
(together completion by 7b, reached 6b I7 I8, setin motion andthat werelater ments ofMalis. thegulf with22) in theareaaround

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Vol. IOI]

THE PREHISTORY OF THE GREEK DIALECTS

63I

NWG
\
~NEG

\,\SG

('~

~~~~~~~~~~~~'

:~~~~~~-

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

632

WILLIAM

F. WYATT,

JR.

[I970

bothby 23 24 25 and by unity.47The fourth map showstheareasaffected the same time I9 and 2I whichleave NWG isolated. At approximately to reach muchof thesamearea,butfailed otherchanges, 7a I6, overspread whileSG was still that tookplacein Thessaly Theseweretheonlychanges on maps included resident there. All otherchangeshave been displayed earlier in thetext.
47Now that ofa mapofGreece we puttheNEG changes on an actual representation innovations: they them as localrather than socialdialectal it becomes possible to regard SG. butdidnotchange PG which reached willhavebeenchanges which affected Central as low class. I havebeenunjust in characterizing them Perhaps therefore

B: thisis map 5.

This content downloaded from 85.69.49.76 on Sun, 2 Mar 2014 12:52:36 PM All use subject to JSTOR Terms and Conditions

Das könnte Ihnen auch gefallen