System PIE

The Primary Phoneme Inventory and Sound Law System for Proto-Indo-European

JounaPyysalo

ACADEMIC DISSERTATION

To be publicly discussed, by due permission of the Faculty of Arts

of the University of Helsinki, in Porthania P III, Yliopistonkatu3,onthe22ndofNovember,2013,at10oclock

2

Publications of the Institute for Asian and African Studies 15

ISBN978-952-10-9303-6(paperback)ISBN978-952-10-9304-3(PDF)

http://ethesis.helsinki.fi

ISSN1458-5359

UnigrafiaOyHelsinki2013

3

DerwahreGrund,warum esComte nicht gelang, ein unlsbares Problem zu finden,bestehtmeinerMeinung nach darin, da es ein unlsbares Problem berhaupt nichtgibt. Statt des trichten Ignorabimus heie imGegenteil unsereLsung:Wirmssenwissen,Wirwerdenwissen.

DAVIDHILBERTNaturerkennenundLogik(1930)

4

AABSTRACT

TheIndo-Europeansoundlawsarethebestknownofalllanguagefamilies.Yetmanysound lawsremain incompletelyformulatedduetoafailure inthe interpretationoftheOldAnatolianlaryngeal.Thepostulationofmultiplelaryngeals(atleastthreeinthemainstreamlaryngealtheory)hasledtoasignificantdetourinthereconstructionofProto-Indo-European(PIE).

A single laryngeal PIE * Ri. was already discovered byLadislavZgusta(1951),however,andsubsequentlyitwasconfirmedbyJohannTischler(1977ff.).Thecurrentdissertation studiesunexploredpropertiesofPIE*anddemonstrates thatthis laryngeal had a voiceless (PIE *h) and a voiced (PIE *Y) variant with glottalfricativearticulation.PIE*appearswithPIE*aindiphonemicPIE*aand*a.

This solution to the laryngeal problem allows for a clarification of therelationship between PIE *h/Y and the rest of the phoneme inventory. Segmentalanalysis results in System PIE, the primary phoneme inventory for Proto-Indo-Europeanconsistingof

PIE*a/?*e/*h/Y*i/*k/g*l/*m/*n/*o/*p/b*r/*s/z*t/d*u/.The phoneme inventory of System PIE isminimal: it cannot be reduced and it issufficient to generate attested Indo-European forms. Accordingly, the import ofSystemPIE for Indo-European linguistics iscomparable tomasteryof thebuildingblocksofDNA.

Inaddition,thedissertationmodernizestheessentialIndo-Europeansoundlawsin terms of the laryngeal PIE *h/Y.Due to the advanced stage of Indo-Europeanlinguistics, no entirely new sound laws are presented, because the yet remainingproblems of the traditional sound laws reflect the absence of the comparativeinterpretationoftheOldAnatolianlaryngeal.

The scientific framework used in this study is the comparative method ofreconstruction, recognized as a branch of natural science already by AugustSchleicher.Thedissertationcontributestothedevelopmentofthefieldbyexplicatingthe comparative method by means of predicate calculus, including a preciseformulation of Schleichers intuitive description of the decisionmethod for Indo-European etymology. As such, the reconstruction theory System PIE can bedigitalized (i.e. turned into a programming language that can generate Indo-Europeandatafromreconstructions).

The most reliable etymological and standard dictionaries are used as thematerialof thedissertation.While these sourcespresent thedataandetymologicalsuggestions thatexist todate,no fullcomparativeconclusionshaveyetbeendrawn.Asacontributiontothisvitalareaofthefield,thedissertationpresentshundredsofnewetymologies,which serveaspreliminaryexamplesof theProto-Indo-EuropeanLexicon(PIELexicon),adigitaletymologicaldictionaryofIndo-Europeanlanguagesthatwillbepublishedathttp://pielexicon.hum.helsinki.fi.

5

AACKNOWLEDGEMENTS

Inmanyways,thisdissertationreflectsmyacademiccareer.My studiesat theUniversityofHelsinkibeganwithClassicalGreekunder the

able instructionofProf.MaaritKaimio,Prof. JaakkoFrsen,Prof.PaavoCastrnandDr.ErkkiSironen.ThedemandsanddisciplineofmysubsequentM.A.studiesinculcated inme the value of thorough philological competence in that language,somethingforwhichIowegratitudetotheseprofessorsandmanyothers.LatinwasanaturalcontinuationofGreek,taughttomebyProf.OlliSalomiesandothers.LaterIbecamefamiliarwiththerestoftheancientItalicdialects,anditismygreatpleasuretothanktheLatinistsfortheirassistanceinthisregard.

A solid foundation in the Sanskrit language was laid for me by Prof. AskoParpolaandProf.KlausKarttunen.SoonIbecameparticularlyinterestedintheRig-Vedic language,whichwas thus added tomy repertoire aswell.Avestan andOldPersianwere kindly introduced tome by Petri Pohjanlehto, a PhD student in theCentral Asian Studies department, and I am very thankful for those who wereinvolvedinmytraininginIndo-Iranianaswell.

Being already capable in several ancient Indo-European languages, it wasnaturalthatIwouldtakeupthetaskoflearningthemall.Thisprocessisstillongoing,butitismanageableasdemonstratedbymypredecessor,Prof.PenttiAalto.Tothisend, I have attended lectures by Prof. Anders Ahlquist in Old Irish, Dr. KariLiukkonen inLithuanianandDr.SannaAro-Valjus inHieroglyphicLuwian.Ihavebenefitedovertheyearsfromtheknowledgeofnumerousindividuals,includingProf.JoukoLindstedtinSlavonic,andforthisIamverygrateful.

With time, I have gradually come to depend more and more on my ownresources to learn languages on my own. As I became more familiar with thereconstructionoftheIndo-Europeanproto-language,Iconsequentlygraduatedwitha double M.A. degree in Indo-European linguistics under the kind and ablesupervisionofProf.AskoParpolaandDr.BertilTikkanen.

Sinceembarkingonmyacademicpath, Ihave compileddigitaldictionariesofIndo-European languages for my own personal use. Around the turn of themillennium, I combined these into an Indo-European etymological dictionary.Mylexicographical interests had made me keenly aware of Oswald Szemernyis(1996:31)words:(...) the first taskof the Indo-Europeanist is toworkback to thefullestpossiblereconstructionofIndo-European.Thisprovedtoindeedbethecase,as one can hardly compile aProto-Indo-European dictionarywithout an adequatePIEphonemeinventory.

Having also learned the key Old Anatolian languages by this point, it hadbecome clear to me for some time that the traditional (Neogrammarian)

6

reconstruction was outdated, in particular regarding the laryngeal. Yet myhoneymoonwith the laryngeal theoryproved tobea shortone. IndiscussionswithProf. Jorma Koivulehto, Prof. Raimo Anttila, Dr. Petri Kallio and Dr. SanteriPalviainen concerning the problems of the laryngeal theory, I discovered that itsinaccuracies in the reconstruction of the data could not be overcome, and I amthankfultothesescholarsforhelpingmearriveatthisconclusion.

Whenengagingintheactualwritingofthedissertation,IhadnootherchoicebuttofollowDarwinsexample.Accordingly,IworkedontrueBaconianprinciples,and[] collected factsonawholesale scale []grouping facts so thatgeneral lawsorconclusionsmaybedrawnfromthem.Duringthistime,myacademicadvisorswereDr. Bertil Tikkanen, whose extensive capabilities in the field of phonetics andphonologyhavebeenaconstant,reliableguidance;Dr.MarttiNyman,whosedata-oriented attitude and insights intomethodologywere always held close; and Prof.KlausKarttunen,whose steadfastness has always been a source of encouragementandcalm.

Mystudieshavealwaysalsoincludedaninterestinphilosophy,inparticularthetheory of science, and therefore I followed lectures by Prof. IlkkaNiiniluoto,Dr.HeikkiKannistoandothers.This interest further ledme to study formal logicandmathematicsunderProf.LauriMyrberg,Dr.JuhaPartanenandothers.Lateron,thisinterestwouldresurfaceintheformoflanguagetechnology,andinthatregardIamespeciallythankfultoProf.KimmoKoskenniemiforoursuccessfuldemosincodingtheIndo-EuropeansoundlawsofSystemPIEandtoMr.AleksiSahala,B.A.;bothofthemaremostcapablecomputerlinguists,ifIeversawone.

Iamgrateful toProf.JuhaJanhunen forhisprofoundcomparativeexperienceandacademicleadership.Ithasbeenmyhonourandpleasuretolearnfromhim.

IamalsoindebtedtoDr.AlbionM.Butters,whohascheckedtheEnglishofthedissertation,improvingitandprovidingmewithvaluablelessonsinthatlanguage.

Lastbutcertainlynotleast,IwishtothankLauraandthechildrenAura,Jade,TaitoandTuafortheirgreatloveandpatience.

Intermsofinstitutions,IwouldliketoexpressmygratitudetotheUniversityofHelsinki,theDepartmentofWorldCulturesanditsheadLars-FolkeLandgren,andallotheremployeesof the institution for theirconstant supportandassuranceofasafehavenfortheactualcompilationofthedissertation.

IamextremelygratefultotheFinnishCulturalFoundationforgrantingmethreeannual scholarships and the Emil Aaltonen Foundation for another set of threeannual scholarships. Without this significant financial assistance, the dissertationwouldnothavebeenpossible.

Finally, Iammost indebted to theboardof the InstituteofAsianandAfricanStudiesforacceptingmydissertationforpublicationinitsseries.

7

TTable of Contents

ABSTRACT .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

ACKNOWLEDGEMENTS .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

TABLE OF CONTENTS .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1 COMPARATIVE METHOD OF RECONSTRUCTION IN INDO-EUROPEAN ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.1 SYSTEMPIEANDCOMPARATIVEMETHODASNATURALSCIENCE ................................... 131.1.1 SituationinthereconstructionofProto-Indo-European.......................................... 131.1.2 Formsasfunctionsofphonemesandmeanings......................................................... 20

1.2 PHONETICSANDPHONOLOGYINSYSTEMPIE ................................................................... 211.2.1 Introduction:phoneticsandphonology...................................................................... 211.2.2 Sounds,phonemesandphonetics................................................................................ 221.2.3 ThehistoricalPIEphonemeinventories .................................................................... 23

1.3 SEMANTICS ............................................................................................................................ 301.3.1 Symbolfunctionandsemantics ................................................................................... 301.3.2 SemanticfieldsofPIErootmatrices........................................................................... 35

1.4 MORPHOLOGY ...................................................................................................................... 371.4.1 Morphemesandmorphology....................................................................................... 371.4.2 Onclassificationofmorphemes .................................................................................. 391.4.3 MorphotacticsandPIErootmatrices ......................................................................... 44

1.5 THECOMPARATIVEMETHODOFRECONSTRUCTION ......................................................... 501.5.1 Comparativerelationanditssubcategories................................................................ 501.5.2 Geneticinternalcomparison(Grammarians)............................................................ 511.5.3 Geneticexternalcomparison(Paleogrammarians) ................................................... 531.5.4 Soundchangesandsoundlaws .................................................................................... 551.5.5 Reconstructionandtheprincipleofpostulation........................................................ 611.5.6 Non-geneticexternalcomparison(typology) ............................................................. 641.5.7 Non-geneticinternalcomparison(metalanguage) .................................................... 661.5.8 Thecomparativemethodofreconstruction ............................................................... 671.5.9 Onregularandirregularsoundchanges..................................................................... 71

2 PIE *AND THE INDO-EUROPEAN VOWEL SYSTEM ... . . . . . . . . . . . . . . . . . . . . . 75

2.1 INDO-EUROPEANVOWELSYSTEMANDI. ..................................................................... 752.1.1 TheproblemofOAnat.andtheIEvowelsystem................................................... 752.1.2 Brugmannssystemofeightproto-vowels .................................................................. 762.1.3 OnAnatolianlanguages,corpusandlaryngeal.......................................................... 772.1.4 i.andthereconstructionofPIE* ....................................................................... 792.1.5 i.andvocalismNeogr.*a ................................................................................. 84

8

2.1.6 TheMonolaryngealschool(Zgusta,Szemernyi) ..................................................... 862.1.7 PIE*insyllabicpositionandNeogr.* ................................................................... 892.1.8 i.inenvironmentNeogr.*e* .............................................................................. 902.1.9 DiphonemicPIE*aandPIE*a .............................................................................. 922.1.10 OnpropertiesofthecoversymbolPIE*................................................................ 96

2.2 VOWELSNEOGR.**A*ANDI. ................................................................................. 972.2.1 Introductionanddefinitions ........................................................................................ 972.2.2 ReconstructionofNeogr.*Gr.:OInd.i........................................................... 972.2.3 ProblemsofthereconstructionofNeogr.*.............................................................. 992.2.4 Neogr.*PIE*a..................................................................................................... 1012.2.5 ReconstructionofNeogr.*aGr.:OInd.a........................................................ 1042.2.6 ProblemsofthereconstructionofNeogr.*a............................................................ 1052.2.7 Neogr.*aPIE*aeorPIE*ea ........................................................................... 1072.2.8 ReconstructionofNeogr.*Do.:OInd. ....................................................... 1112.2.9 ProblemsofthereconstructionofNeogr.*............................................................ 1112.2.10 Neogr.*PIE*aorPIE*a ......................................................................... 112

2.3 VOWELSNEOGR.*O**ANDI. ............................................................................... 1142.3.1 Introduction ................................................................................................................ 1142.3.2 ThereconstructionofNeogr.*oGr.:OInd.andBrugmannsLaw............. 1142.3.3 ProblemsofNeogr.*oandBrugmannsLaw........................................................... 1162.3.4 ReconstructionofNeogr.*Gr.:OInd.a........................................................ 1172.3.5 ProblemsofthereconstructionofNeogr.*............................................................ 1182.3.6 Neogr.*PIE*o .................................................................................................... 1202.3.7 Neogr.*oPIE*oa,*oa(BrugmannsLawII) ................................................. 1212.3.8 ReconstructionofNeogr.*Gr.:OInd. ....................................................... 1252.3.9 ProblemsofthereconstructionofNeogr.* ........................................................... 1262.3.10 Neogr.*PIE*,*a,*a,*aor*a ...................................................... 127

2.4 VOWELSNEOGR.*EAND*ANDI.............................................................................. 1282.4.1 Introductionanddefinitions ...................................................................................... 1282.4.2 ThereconstructionofNeogr.*eGr.:OInd.a ................................................. 1282.4.3 ProblemsofthereconstructionofNeogr.*e............................................................ 1302.4.4 Neogr.*ePIE*e*ea*ae............................................................................ 1312.4.5 ReconstructionofNeogr.*Gr.:OInd. ........................................................ 1342.4.6 ProblemsofthereconstructionofNeogr.*............................................................ 1352.4.7 Neogr.*PIE**a*a ............................................................................ 137

2.5 PIEABLAUTANDPIE*INSYSTEMPIE......................................................................... 1392.5.1 PIE*a,*aandtheProto-Indo-Europeanablaut................................................. 1392.5.2 AblautPIE*:*o::*e:*................................................................................... 1422.5.3 ProtheticablautPIE*:*o::*e:* ................................................................... 1462.5.4 AblautPIE*:*o::*e:*withPIE*a,*a.................................................... 1522.5.5 PIE*ainablautPIE*a*oa*a*ea*a ..................................................... 1542.5.6 ProtheticablautNeogr.*a:*oandi. ................................................................. 1562.5.7 SchwebeablautandPIE*......................................................................................... 162

9

2.5.8 OsthoffsLawforAnatolian,TocharianandGreek ................................................ 1662.5.9 EvaluationofhistoricaltheoriesandSystemPIE.................................................... 172

33 PIE *AND RESONANTS PIE *I U L R M N ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181

3.1 ONTHEORIESANDPROBLEMSOFTHERESONANTSYSTEM............................................. 1813.1.1 Introductoryremarksonresonants ........................................................................... 1813.1.2 OnthetheoriesofPIEsyllabicresonants................................................................. 1823.1.3 Thetheoryofsyllabicsonants(Sonantentheorie) ................................................... 1823.1.4 TheproblemsofSonantentheorie............................................................................. 1863.1.5 Theschwasecundumschool ...................................................................................... 1923.1.6 Thecomparativetheoryofsyllabicresonants .......................................................... 196

3.2 SEMIVOWELSPIE*AND*ANDVOWELSPIE*UAND*I............................................... 2003.2.1 Neogr.*+=PIE*+ .................................................................................................... 2003.2.2 Neogr.*u=PIE*u .................................................................................................... 2033.2.3 Neogr.*PIE*u,*u,*u,*u,*uu ......................................................... 2053.2.4 Neogr.*!PIE*! ...................................................................................................... 2113.2.5 Neogr.*iPIE*i ...................................................................................................... 2143.2.6 Neogr.*PIE*i,*i,*i,*i,*ii ................................................................. 2153.2.7 OnSieverssLawandSturtevantsanalysis .............................................................. 2193.2.8 SummaryofPIE*i,*uandPIE*a,*a.................................................................. 223

3.3 LIQUIDSPIE*L*R .............................................................................................................. 2243.3.1 GeneralremarksonPIEliquids ................................................................................ 2243.3.2 FortunatovsLawandPIE*a*a ........................................................................... 2243.3.3 Liquids*rand*lintheNeogrammariansystem...................................................... 2433.3.4 Neogr.*r(consonantaltrill) ...................................................................................... 2483.3.5 Neogr.**(anteconsonantalsyllabictrill).................................................................. 2513.3.6 Neogr.**r(antevocalicsyllabictrill) ......................................................................... 2603.3.7 Neogr.*)(anteconsonantallongsyllabictrill) ......................................................... 2663.3.8 Neogr.*l(consonantallateral) .................................................................................. 2723.3.9 Neogr.*$(anteconsonantalsyllabiclateral) ............................................................. 2733.3.10 Neogr.*$l(antevocalicsyllabiclateral) ................................................................... 2763.3.11 Neogr.*#(anteconsonantallongsyllabiclateral)................................................... 2803.3.12 LiquidsPIE*l/$andPIE*r/*inSystemPIE........................................................... 284

3.4 NASALSNEOGR.*N*M....................................................................................................... 2843.4.1 NasalsintheNeogrammariansystem ....................................................................... 2843.4.2 PIE*n(consonantaldental) ...................................................................................... 2923.4.3 Neogr.*((anteconsonantalsyllabicdental) ............................................................ 2943.4.4 Neogr.*(n(antevocalicsyllabicdental).................................................................... 3073.4.5 Neogr.*'(longsyllabicdental) ................................................................................. 3113.4.6 PIE*m(consonantalbilabial) ................................................................................... 3173.4.7 Neogr.*&(anteconsonantalsyllabicbilabial) ......................................................... 3183.4.8 Neogr.*&m(antevocalicsyllabicbilabial) ................................................................ 3273.4.9 Neogr.*%(longsyllabicbilabial) .............................................................................. 330

10

3.4.10 NasalsPIE*m/&and*n/(inSystemPIE .............................................................. 3323.5 RESONANTSINSYSTEMPIE............................................................................................... 332

3.5.1 Theresonants*iulrmninSystemPIE .................................................................. 3323.5.2 TheevaluationoftheSonantentheorie .................................................................... 334

44 PIE *AND THE PIE OBSTRUENT SYSTEM ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345

4.1 INTRODUCTION ................................................................................................................... 3454.1.1 TheNeogrammarianobstruentinventory ................................................................ 3454.1.2 Neogr.*TThDDh(Decem-Taihunisogloss)......................................................... 3454.1.3 Neogr.*K:K!:K+(Centum-Satemisogloss)........................................................... 347

4.2 THEORIESOFTHEFOURPLOSIVESERIESTTHDDH...................................................... 3484.2.1 NeogrammariansystemTThDDh.......................................................................... 3484.2.2 MeilletsandMagnussonsrootconstrainttheory ................................................... 3484.2.3 ThetypologyTDDhofthelaryngealtheory........................................................... 3494.2.4 Theglottalictheory(GamkrelidzeandIvanov) ....................................................... 3504.2.5 OverviewofthetheoriesofthePIEplosivesystem................................................. 351

4.3 TENUESNEOGR.*K,P,T..................................................................................................... 3534.3.1 MaterialofNeogr.*k,p,t .......................................................................................... 3534.3.2 TheoreticalapproachestoseriesT(tenues) ............................................................ 3554.3.3 SolutionstotheseriesT(PIE*k*p*t) .................................................................... 355

4.4 TENUESASPIRATAENEOGR.*KH,PH,TH ......................................................................... 3564.4.1 Generalremarksontenuesaspiratae ....................................................................... 3564.4.2 MaterialofNeogr.*kh,ph,th ................................................................................... 3574.4.3 TheoreticalapproachestotheseriesTh................................................................... 3604.4.4 ComparativesolutionoftheseriesTh ...................................................................... 361

4.5 MEDIAENEOGR.*G*B*D ................................................................................................. 3664.5.1 MaterialofNeogr.*g,b,d ......................................................................................... 3664.5.2 Theoreticalapproachestotheseriesmediae ........................................................... 3694.5.3 Solutionstotheproblemsoftheseriesmediae........................................................ 370

4.6 MEDIAEASPIRATAENEOGR.*DH*BH*GH ..................................................................... 3954.6.1 MaterialofNeogr.*dh,bh,gh................................................................................... 3954.6.2 Historicalapproachestothemediaeaspiratae ........................................................ 3974.6.3 Criticalcorrectionsandsolutions .............................................................................. 3984.6.4 GrassmannsLawanditsexceptions ......................................................................... 4094.6.5 BartholomaesLawanditsgeneralization................................................................ 413

4.7 SUMMARYOFTHEDECEM-TAIHUNISOGLOSS................................................................. 4144.7.1 SummaryoftheseriesT:Th:D:DinSystemPIE.............................................. 4144.7.2 EvaluationoftheDecem-Taihuntheories ............................................................... 416

4.8 CENTUM-SATEMISOGLOSSORTHETHREEVELARSERIES.............................................. 4174.8.1 GeneralremarksontheCentum-Satemisogloss..................................................... 4174.8.2 TheplainvelarsNeogr.*kkhggh ............................................................................ 4244.8.3 ThelabiovelarsNeogr.*k+*k+h* * h.................................................................. 4274.8.4 ThepalatovelarsNeogr.*""hh........................................................................... 441

11

4.8.5 Proto-Indo-EuropeanvelarsinSystemPIE ............................................................. 4494.9 PROTO-INDO-EUROPEANFRICATIVES.............................................................................. 452

4.9.1 Generalremarksonthehistoricalfricativesystems ................................................ 4524.9.2 ThesibilantsPIE*sand*z ........................................................................................ 4534.9.3 PIE*h/andthepropertiesofthelaryngeal............................................................ 459

55 THE RECONSTRUCTION THEORY SYSTEM PIE .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465

5.1 SYSTEMPIEANDPIELEXICON......................................................................................... 4655.1.1 ThephonemeinventoryofSystemPIE .................................................................... 4655.1.2 TheaxiomatizationofSystemPIE ............................................................................ 4695.1.3 ThesoundlawsofSystemPIE................................................................................... 4725.1.4 ThedecisionmethodofIndo-Europeanetymology ................................................ 4755.1.5 Proto-Indo-European(PIE)Lexicon........................................................................ 477

6 REFERENCES .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484

7 ABBREVIATIONS .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 500

12

13

11 Comparative method of reconstruction in Indo-European

1.1 System PIE and comparative method as natural science

1.1.1 Situation in the reconstruction of Proto-Indo-European

0.ThesituationofthePIEreconstructionchangeddecisivelyafterBed ichHrozns(1917)demonstrationoftheIndo-EuropeanoriginofHittite.Acentury later, ithasbecomeindisputablethatOldAnatolianpreservedalaryngealsegmentHittitethatwas lost in the languages on which the Neogrammarian phoneme inventory andsound law systemwerebased.The laryngeal theory,withMllers advancementofthree laryngeals and the subsequent addition of variants, dates back to the pre-laryngealperiod (1879-1880) and isbasedon aSemitic typology rather than Indo-Europeandata.Accordingly, the theorycannotwin theacceptanceofcomparatists,withtheresultthatthestudyisindeadlock.Withsuchastateofaffairs,Szemernyis(1967:92)assessmentismorerelevantthanever:

What is really needed is a renewed, and unbiased, study of all the available Hittiteevidencewithnoattempttoforceitintothestrait-jacketofpreconceivedtheoriesaboutIEablautorroot-structure.

Indeed, the problemswith the study are caused by a lack of detailed comparativereconstruction based on the current body of greatly enriched data and the newsegmentPIE*,themissinglinkinthePIEphonemeinventory.Itiswellknownthatwhendatachanges, theoriesalsomustchange. It isnotanexaggeration to say thatIndo-European linguistics stands today in the very situation once sketched out byKarlBrugmannandHermannOsthoff:1

Ehemanweiterbaut,bedarfderganzebau,soweiterbis jetztdasteht,einergrndlichenrevision.(1878:xi).

1The laryngeal isconfirmed,owing to the tracesofPIE*outsideofOldAnatolianaswell(e.g. inRig-Vedichiatus,regularlycoincidingwithi.incorrespondences).

14

The quantitative and qualitative improvement of the presentation of the Indo-Europeanmaterial has reached a criticalmass, allowing the solution of allmajorproblems of PIE segmental phonology based on the comparative method ofreconstruction. This window of opportunity will be explored in this study with acompletely upgraded reconstruction theory, called System PIE,which is based onstrict principles of natural science. In essence, SystemPIE consists of the primaryphoneme inventory and the upgraded sound law system forProto-Indo-European,withparticularattentionpaidtothesegmental laryngealPIE* inallenvironments.Assuch,SystemPIEisdesignedtosolvethecriticalproblemsofPIEphonologyandopen the way for a subsequent exploration of the breakthrough, especially in thefields of PIE morphology, etymology and the accent of the proto-language.ConcerningtheseSchwerpunkts,thefollowingpreliminaryremarksarepresented.

1. The reconstruction of the primary phoneme inventory (i.e. the phonetic andphonologicalcomponentofSystemPIE)willnotstartfromscratch.Onthecontrary,owing to thehighlyadvanced stageof the study, the traditionallypostulatedproto-phonemeswillserveasstartingpointsforthecasestudiesandsolutionssuggestedbythe comparative method will be presented for each question. In the order ofappearance,thephoneticandphonologicalproblemsinclude:(a)TheproblemoftheProto-Indo-EuropeanlaryngealPIE*hasbeenpreliminarilysolved by the comparative school with the theory of monolaryngealism (derMonolaryngalismus).According to theproponentsof this theory, there isone (andonlyone) laryngealPIE* inductivelyobtainablefromtheOldAnatoliandata.Thisresult,originallydiscoveredbyLadislavZgusta (1951),hasnowbeen confirmedbyJohannTischlerandhiscolleaguesinHethitischesEtymologischesGlossar(1977ff.),the most noteworthy and reliable etymological dictionary of Old Anatolian inexistence.2 The delay in the breakthrough of the theory has been caused by itsapproximate form, basically consisting only of the realization of the existence of asingle PIE *.With an independent confirmation of the result, the study at handcontinues with a complete study of PIE *, its properties, and the sound lawsgoverning it in all environments. As a result, System PIE implementsmonolaryngealismasafull-scalereconstructiontheoryconsistingonlyofpostulatesofthecomparativemethod.(b) As is well known, the problems of PIE * and PIE vocalism are closely knittogether.At its apogee, theNeogrammarian vowel systemofBrugmann containedeightcover symbols for theproto-vowels.The systemwas inductively reconstructedand it has the necessary minimum of phonemes required for a complete (andthereforevalid) reconstruction theory.Althoughnoadditional correspondence setshave emerged in the newmaterial,Brugmanns system is outdated, particularly in

2InPyysalo2003,aftercomparingalltheexistingPIEreconstructiontheoriesonthesamematerial,Idemonstrated the impossibility of the supported versions ofmultilaryngealism and concluded thatmonolaryngealismisthesolereconstructivepossibilityforProto-Indo-European.

15

termsof therelationof theeight-vowelsystem to the laryngealPIE*consistingofthreesubsets:

1.TheproblemofNeogr.*T:a:(a-vocalism)andPIE*.2.TheproblemofNeogr.*o::(o-vocalism)andPIE*.3.TheproblemofNeogr.*e:(e-vocalism)andPIE*.

ThecomparativesolutiontothesemainPIEablautproblemsandtheirrelationtoPIE*ispresentedinChapter2.(c)Theproblemoftheresonants(orsonorants)PIE*iurlnm,bothindependentlyandintheenvironmentofPIE*,isdividedinto:

1.Theproblemofsemi-vowels/glides*i,u(U)withandwithoutPIE*.2.Theproblemofliquids*rl(L)withandwithoutPIE*.3.Theproblemofnasals*nm(N)withandwithoutPIE*.

ThecomparativesolutionoftheseproblemsispresentedinChapter3.(d)TheproblemofPIEobstruents,independentlyandintheenvironmentofPIE*,isdividedintothreesubsets:

1.Theproblemoffourseriesofplosives(Neogr.*T:Th:D:DY).2.Theproblemofthreeseriesofvelars(Neogr.*k:*:*k,etc.).3.TheproblemofIndo-Europeanfricatives(Neogr.*s/zandPIE*).

ThecomparativesolutionoftheseproblemsispresentedinChapter4.(e)Theproblemsof thePIEphoneme inventoryaredivided intonine subsets.Tothesemay be added a tenth subset: their treatment in a comparatively consistentsystem. Inorder toestablish theprimary characterof thephoneme inventory, it isdemonstrated that no phonemes are absent in System PIE and that the inventorydoesnotcontainanalyzablephonemes(i.e.SystemPIEisminimal).3

2.PIEsoundlaws,comprisingthephonologicalpartofSystemPIE,arethoroughlyupgraded(inparticular,forPIE*),accordingtothecomparativeimplicationsofthenowenricheddata.Whennecessary,thesoundlawsareanalyzedinconnectionwiththeproblems.Thus,BrugmannsLawandOsthoffsLawareupgradedinconnectionwith the vowel system,SieverssLaw andFortunatovsLaw in connectionwith theresonant system and so forth until the segmental PIE sound laws have beencompletelyrevised.

3.ThekeyIndo-European(IE)languagesforthereconstructionofPIEconsistofthehundredmostancient languagesfromthe lastfourmillennia.Split intotwelvemainsubgroups,thelanguagefamilypresentshistoricalsoundchangesinauniquemanner,similarly allowing theprospective reconstructionof their common ancestor,Proto-Indo-European (PIE). To date, thousands of scholars from distinguishedlexicographerstocomparativelinguistshavededicatedmillionsofman-hourstothecodingofthematerial,makingthemostancientIndo-Europeandatafinallyavailable

3Thusallhistoricalproto-phonemeswillbe individually scrutinized for theirexistenceandpossibleanalytical(or polyphonemic)origin,ensuringthatno itemsstandforsimplerproto-phonemes(as isthecasewithGr.J,A,etc.).

16

inapractically complete form.4Thekey featuresofPIELexicon, theetymologicaldatabaseofSystemPIE,formasynthesisoftheseeffortsandcanbecharacterizedasfollows:(a) In terms of the completeness of thematerial, themeasures recommended byBrugmannandOsthoffintheNeogrammarianmanifesto(1878)havebeenadopted:

Jemehr sprachmaterialuns so in lckenloser,durchdie jahrhunderte sichhinziehenderschriftlicherberlieferungzurbeobachtungunterbereitet ist,umsobessersindwirdaran[](1878MU1:vii.)5

Historically speaking, however, theNeogrammarian theory with its emphasis onSanskrit,GreekandLatinwasneverbasedoncompletedata,nordiditclaimtobe.6Thisprovidesawindowofopportunitytofurtherthereconstruction.(b)Inordertoeliminatetheproblemofthe incompletenessoftheNeogrammarianreconstructionand,evenmore, thatof the laryngeal theory thematerialof thedissertationconsistsofthemainbulkofstems(andmorphemes)ofthehundredmostancientIndo-Europeanlanguagesbasedonthemosttrustedmainstreamdictionaries,comparativelysupplementedwithothercriticalsources.

The fullmaterial, in homage to themost capable scholars of in the field ofetymologywillbeseparatelypublishedunderthetitleProto-Indo-EuropeanLexicon(PIELexicon);ithasalreadybeencompiledwithalengthoffivethousandA3pages.Theworkiscurrentlyinanadvancedstage,allowingpreparationoftheinitiallettersofthePIELexiconforpublication.(c)ThePIELexiconisanext-generationetymologicaldictionaryutilizingtherulesofSystemPIE,aspresented in this study.Althoughhardcopyversionscouldbemadeavailable,thePIELexiconisessentiallyadigitalenterprise7withtheultimateaimofaccounting for every recorded Indo-European morpheme. This has been madepossiblebythegeneralprogressoflanguagetechnology,exemplifiedtodaybysimilarproductsinthefield,liketheTITUSproject(ThesaurusindogermanischerText-undSprachmateriel) based in Frankfurt am Main.8 The TITUS project is currentlypublishing archaic Indo-European texts, but links to digital dictionaries are alsooffered on the TITUS website. Due to digital technology, the TITUS project willbecomeavailabletotheusersofthePIELexiconthroughthecommonmaterialdealtwith,allowingforthefurtherimprovementofboth.

4 Bammesberger (1984:9): Seit Beginning unseres Jahrhunderts hat sich hauptschlich durch dieKenntnis des Hethitischen und Tocharischen die Materialbasis fr die Rekonstruktion derindogermanischenGrundsprachewesentlicherweitert.5Zgusta (1951:428): Il estnaturelquune thorienouvelle soit ainsi applique aumatriel lepluslargepossible.6ForBrugmannsnote concerning the incompletenessof all early theories (includinghisown), seeGrundr21:397n1.7ThePIELexicon isdesignedtoallowforanupgradingofdatauntilallIndo-Europeanmorphemesarereconstructed.Thus,thecompletenessofSystemPIEcanbedemonstratedinextenso.8 For the TITUS Program (Das Project eines indogermanischen Thesaurus), see http://titus.uni-frankfurt.de/indexe.htm.

17

4.Throughout the study, specialweight is placed on a strict commitment to thecomparative method and other methodical disciplines. This deserves a briefexplanation:(a)AnthonyFoxcharacterizesearlydiscussionsonthecomparativemethodinwriting(1995:19):

It must be said that nineteenth-century discussions of the method itself, and of theprocedures involved in itsapplication,areratherdisappointing.Although therearemanydemonstrations of the results of the method, no detailed step-to-step explanations orexplicitformalizationsareforthcomingfromthisperiod.

With theexceptionofSchleicher, thisevaluation isgenerally correct.Similar ideaswithanevenmorecriticaltonehavebeenexpressedbyRadoslavKatii(1970:9),aleadingcomparativetheoretician,whowrites:

If this traditional field of linguistic studies is to be incorporated in amodern body oflinguisticdoctrine,thecomparativemethodmustbemadeexplicitanditsproceduresmustbecomemore formal. If amethod is stated explicitly it becomes possible to discern itspropertiesandshowwhyitissuccessfulandwhereitcouldbeexpectedtofail.9

(b) Within this study are found both an explicit presentation of method (seeespecially Chapters 1 and 5) and its formalization in predicate calculus, the bestknown and most uncontroversial scientific meta-language in existence.10 ThisformalizationconsistsofasimplepresentationanddefinitionoftheIndo-Europeanmaterial intermsofpredicatecalculus.11Theusefulnessoftheformalizationwillbedemonstrated in Chapter 5, where the decision method for the Indo-Europeanetymologyisstatedasasimpleformulaofpredicatecalculus.(c)Thepreliminarynatureof thePaleogrammarianphoneme inventoryand soundlaws(basedonSanskrit)andthelaryngealtheory,presentingaSemitichypothesisona Neogrammarian chassis, means that Indo-European linguistics depends on theNeogrammariansmorethantypicallyunderstood.ThismakesthefollowingremarkofDavies(1975:644)relevantforthestudyasawhole:

What historiography [and Indo-European linguistics]most needs now is a series ofattempts to investigate both the neogrammarians concrete achievements (about whichmuchisknown)andtheirtheoreticalpresuppositionsintheirentirety(aboutwhichwearefarlessclear),tocomparethetwo,andsettheminsomesortofhistoricalperspective.

9Asafurthermotivation,Katii(1970:72)referstotheongoing laryngealcontroversy:Theheateddiscussion that arose about the laryngeal theory could become much more fruitful if themethodological problems were made explicit. For a detailed account for the methodologicalinadequaciesofthelaryngealtheory,seeBammesberger1984.10Predicatecalculusisaformalizationoftheuniversalrulesoflogicsharedbyallbranchesofscience.Logicandpredicatecalculusremainthesame,butthebranchesofnaturalsciencedifferintherealobjectsembedded.Forthetranslatabilityofpredicatecalculus intoamodernprogramming languagethatallowsforthetestingofthesoundlawsofSystemPIE,seeChapter5.11 Despite the introduction of notation for predicate calculus, the standard conventions in thepresentationofIndo-Europeandataarefollowedinthisstudy.

18

5.Inonerespect,SchleicherdidbetterthantheNeogrammarians,namelyinviewingthecomparativemethodasanatural science.12Thishighlyconservative tradition isupheld by the author in System PIE and the PIE Lexicon with the principles ofnaturalsciencedulyfollowedthroughout:13(a)The comparativemethod of reconstruction is an empirical science.The Indo-European data is understood like DNA code, carrying genetic information, andthereforenormative.Shouldatheoryconflictwiththedata,correctionsinthetheoryaresoughtinsteadofirregularexplanations,inaccordancewiththethoughtofHansHenrichHock(1991:535):

Givenachoice,analysespostulatingsoundchangesaremorehighlyvaluedthananalyseswhich require analogical or other non-phonetic changes. Similarly, everything else beingequal, analyses operating with regular changes (sound change and/or rule-governedanalogy)arepreferredoverthosewhichrequiresporadicorlessregularchanges.

By seeking improvements in the analysis ofmaterial instead of analogies,the self-correctingprocessofthesciencecanbemeaningfullyupheld.Accordingly,theresultofthemethodis[]testableinprincipleonthebasisofparticulareventsoccurringinspaceandtime(seeEsaItkonen1978:2ff.andMarttiNyman1982:19).Basicallythis amounts to the acceptance of Isidore Dyens requirement (1969:508) that[s]tatements regarding thenatureof theproto-languageareentirely inferentialoranalytical, not assumptive.A theory allowing verification or falsification of everydetailispursued,andaprioristhypothesesarereplacedwithinductiveones.(b)Thereconstructionofproto-languagemeansitsrestorationinascientificmannerthat satisfies high philological, linguistic and comparative standards. Ultimately,reconstruction represents an equivalentof the Indo-Europeandata, compressed inProto-Indo-European formulas. Szemernyis (1996:32) position is compulsorythroughout:

Fromtheoutsetrealism,arealisticapproach,playsadecisivepartinreconstruction,sincethereconstructionofphonetically impossiblesoundsandsoundsequences(=words)canbeconsiderednothingbutanidlegame.

Thereconstructionofproto-language isnothypothetical,butaregulatedproceduredefinedby specificempirical criteria.14Therefore, scientific realism is the standardforthepostulationofreconstructionandconceptformation,whichareonlyallowediftheobjectsareobtainedexclusivelyfromthematerial.15An isomorphicrelationship

12 SeeKoerner (1982:2): Schleichers conception of language []was, at leastwith respect to itsmethodofinvestigation,anaturalscience(Naturwissenschaft).SeealsoFox(1995:24):TheworkofSchleicher and his contemporaries, on the other hand, reflects the growing interest in the naturalsciencesand inscientificmethod: themethodof linguistics istotallydifferentofthatofallhistoricaldisciplines,andisbasicallythatofthenaturalsciences.13Onthestructureofscientifictheories,seeKuhn1973.14AccordingtoSzemernyi(1962),thebasicprinciplesofetymologicalresearcharephonetic,semanticandwordformationcriteria.SeealsoAnttila(1969:35).15Forconceptformationintheempiricalsciences,seeHempel1952.

19

between the objects of the theory and their counterparts in the realworld is thusdemandedonalllevels.16(c)IntheevaluationoftheIndo-Europeanreconstructiontheories,atheory(and/oritssubset)isvalidifandonlyifitiscompleteandsound.17Inthisregard,thecounter-exampleprocedure(i.e.constructingasetofdatafalsifyingahypothesisandleadingto a revision of the theory) is favoured in order to take problems as part of thesolution.(d)Occamsrazor,18ortheprincipleofeconomy(quotedherefromHock1991:538),isadoptedforthepurposesofcomparisonofthetheoriesandtheirsubsets:

Reconstructions should not violate the maxim attributed to the medieval philosopherOccam that eentia non sunt multiplicanda praeter necessitatem entities (in anargument)arenottobemultipliedbeyondnecessity.Putdifferently,thesimplestpossibleanalysisistobepreferred,everythingelsebeingequal.

The converse of the principle,Occams guillotine, is applied in the elimination ofunnecessaryassumptions.19(e)Theexnihilonihilprinciplestatesthatnothingcomesfromnothing.Inpractice,ifameasurablephenomenonexists, itcanbeassumed to reflectapreviouslyexistingstateratherthantoemergefromnowhere.Theprincipleisalsousedintheevaluationofthecompetingtheories.(f)Theruleofunambiguitycanbedefinedthus:fromapropositionpq(porq),itisnotallowedtoinferapropositionporpropositionqunlessporqhasbeenproven.This rule is designed to secure the scientific character of theory by disallowingconclusionsofambiguoushypotheses.(g)Throughoutthestudy,Ficksruleisusedastheprincipleofpostulationtojustifytheentirereconstuction.Accordingtothiskeyprincipleofthecomparativemethod,twoindependentwitnessesarealwaysrequired.20Asaconsequenceofthislimitation,the comparative method of reconstruction in its pure form is the sole form ofinference applied in this study,with the result that the very source codeofProto-Indo-EuropeanisderivedinanobjectivemannerinSystemPIE.

A strict adherence to these principles allows one to demonstrate thatSchleichers view of the comparative method as natural science is accurate. Bystickingtoprinciplesofnaturalscience,nothingbutscienceisproduced.Thecorrect16 For the opposite point of view, see Benveniste (1962:10): On a trop cherch convertir leslaryngales en ralits phontiques. Nous avons toujours pens que le statut qui leur convenaitprsentementtaitceluidtresalgbriques.Loindentregnee,lareconstructionindo-europennesen trouve facilite. Les modles de reconstruction ne doivent pas dpendre dinterprtationsphontiquesencorelargementconjecturalesetquiseraientncessairementhistoriques.17Asystemiscompleteifitgeneratesallthecorrectforms,notifitgeneratesincorrectforms.18ForOccams razor (entitianon suntmultiplicandapraeternecessitatem) in linguistics, seeHock(1986:538-540)andSzemernyi(1977:309).19Ifapostulateisnotnecessary,itismeaningless.20SeealsoBammesberger(1984:11):UmeinlinguistischesPhnomenderGrundsprachezuschreibenzu knnen, mu es in mindenstens zwei verschiedenen Sprachgruppen unverkennbare Spurenhinterlassenhaben.

20

solutionscanbesimultaneously identifiedandcalibratedtomatchtherequirementsofthenowenricheddata.

11.1.2 Forms as functions of phonemes and meanings

0. Katii (1970:146) expresses the key idea of language, forms as functions ofmeaning,asfollows:

[] the languages in genetic research must be defined in the first place as sets ofphonemicstringsthatserveasexpressiontocertaincontents.

Though not sufficient as a general theory of language which is in any case notsought in this study,being strictly limited to the Indo-EuropeandomainKatiisdefinitionprovidesasolidstartingpointforadefinitionofthecomparativemethodintermsofpredicatecalculus.

1.Theattestedformsconstitutingthelexicalitemsoflanguageconsistofthestringof phonemes a1, a2, ..., an and the meaning x (in practice, the translation).Consequently, the Indo-Europeandata canbeunderstood as a setofpropositions(functions)oftheform(a1,a2,...,an)= x.InSystemPIEand inthePIELexicon,thestemsarechosenasthebasiclevelofdescription.21Accordingly,anindependententryisprovidedforeverydocumentedstem,andthedescriptionisunderstoodtobecomplete when all attested stems have been accounted for. An example of thepresentationofmaterialbasedonthestems(arrangedundertherespectiveroots)is,forinstance,theOldAnatolianformation22#-sein(e#-,a#-#a-):

i.e#- (pr.)sein(HEG1:109-10,e-e#-zi[3sg],KBoI53,7) i.a#- (pr.)sein(HEG1:109-10,a-#a-an-du[3pl]) HLu.sa- (vb.)be(CHLu.2.34.1,sa-t[3sg],10.17.6,sa-ta[3pl])

In termsofpredicatecalculus, suchentriesarecombined functions f(g(h(x)))= yexpressing not only the stem and its meaning, but additional information likegrammaticalanalysis(e.g. (pr.), [3sg],etc.),reference(e.g. HEG1:109-110), thelocusoftheattestedform(e.g.KBoI53,7)andsoforth.23

2. In the formalization the following symbols, functions (symbol: )24 anddefinitions(symbol:)areused:

21Hocks(1991:29)definitionisfollowedhere:Ifthemaincarrieroflexicalmeaninginagivenwordismorphologicallycomplex,containingarootplusanaffix,itiscalledastem,suchasword-y,inword-i-er,word-i-ness. Inadditionalso the root,capableof taking inflectionalendings, isunderstoodasaspecialformofstem.22On the topic of organization, compareMatthews (1991:26): For some other languages, such asSanskrit,dictionariesareorganizedbystemsorroots...23Thegrammaticalfunctioncoversthetypesofstemsaccordingtotheirgrammaticalclass, includingverbs(vb.),substantives(sb.),adjectives(a.),numerals(num.),adverbs(adv.),interjections(intj.),etc.24Functions1,2,3,...canrepresentanyproperties(orfeatures)orrelationsoftheargumentsx1,x2,,xn.

21

(a)TheIndo-Europeanstemsarearrangedunderconstant functionsexpressing thesource language (e.g.Aiol.,Alb.,Arm.,Av., etc.)of the item in question, and thescopeofafunctiondefinesthelexiconofthatlanguage.25(b)ThephonemeparadigmsoftheindividualIndo-Europeanlanguages(i.e.thesetsofminimalunitsofthesoundsystem)canbereferredtoastheirconstantinventories.Forthephonemeparadigms,anextensivedefinitionisthereforesetforth.Thus,asanexample,forGreekwecandefine:26

/Gr. 4,5,6,7,8,...,K (theGreekphonemeinventory).

Inpredicatecalculus,therealobjects4,5,6,7,8,...,Kcanbereferredtobytwokindsofobjectvariablesfreeones(a,b,c,...)andboundones(x,y,z,...)bothofwhicharefurthermarkedwithsubscriptsa1,a2,...,an,...andx1,x2,x3,...asneeded.27(c) The phonemes constituting a stem are connected with a sequence function(symbol:+)expressingtheleft-to-rightorderoftheobjectsinvolved(e.g.a1+a2+...+an). Inpractice, it isnotnecessary towrite the sequence function; forexample, theconventionalwriting(e.g.Go.ist)isunderstoodasshorthandforGo.i+s+t.(d)Thecomparative function(thesymbol :)canbesetbetweenany twoarguments/x(a)and/y(b)bysettingtheminjuxtaposition(e.g.i.e#ziis:Go.istis).Ifthecompareditemsareidentical,thenthecomparativefunction/x(a):/y(b)isprovableandidentity(thesymbol=)replacesthefunction;otherwiseitsoppositeisshown(bythesymbol:).(e)Astringofphonemes/(a1,a2,...,an)isamorpheme,ifandonlyifthereexistsanxsuchthat xis itsmeaning(possiblyunknown).Formally,therefore,themorphemesareofgeneralform/(a1,a2,...,an)df x.Astemcancontainmultiplemorphemes,and if so these are separatedby segmentation function (the symbol ) as seen, forexample,with:

OIr.doformag- (pr.)accrore(LEIAM-8,doformaig[3sg]).

3. In thismanner, any Indo-European lexical item can be expressed in terms ofpredicatecalculus(i.e.one-to-onemappingexists).

11.2 Phonetics and phonology in System PIE

1.2.1 Introduction: phonetics and phonology

0.ThebasicsituationisneatlysummarizedbySalmonandSmith(2005:86):

25Thevariablescovering theconstant functions(i.e. languagesanddialects)are/,+,0, ...possiblywithsubscripts(/1,/2,...,/n,etc.).WiththesetheindividualsubgroupslikeBaltic,Celtic,etc.canbedefined.26Thedefinitionsof thephonemeparadigmsof the Indo-European languages,available in standardgrammars,arenotrepeatedhere.27Inaddition,thezerophoneme(representedbythesymbolsor) isusedtomark lostphonemesandthezerograde(bothinIEandPIE).

22

Establishing a phonological inventory is a cornerstone of linguistic description and thesamenaturallyholdsforreconstructingproto-language.

InordertoensurethecorrectreconstructionoftheIndo-EuropeanandProto-Indo-Europeanphonemeinventories,onemustobservethefollowingissues:

1.ThephonemeparadigmsofIndo-Europeanlanguagesbasicallycoincidewiththeinherited alphabets createdby the inventor(s)of the respectivewriting systems. Inthisway, the inherited alphabets contain a received internal reconstruction.Beingempiricallygiven,reinterpretationofalphabetsisseldommotivated,thoughnaturallythepropertiesofthesystemscanbedealtwithbymeansofphonetics,thescientificstudyofsoundsasindividualobjects(Trask,DPhPh:270),andphonology,thestudyoftherelationshipsofsoundsinalanguage(Trask,DPhPh:275-77).

2.InthereconstructionofthephonemeinventoryofProto-Indo-European,onlythestrictestprinciplesofthecomparativemethodareemployed.Inpractice,everyproto-phoneme must be comparatively postulated, based on a correspondence setconsistent with the full data. In particular, the so-called hypothetico-deductivemethod, which is occasionally allowed in historical linguistics and involveshypotheticalproto-soundsandapostulationofpre-proto-language,isunnecessary.

11.2.2 Sounds, phonemes and phonetics

0.Thesoundsofspeechareconcreteobjectswithmeasurableacousticpropertiesorfeatures produced by airflow and the human vocal apparatus, the places ofarticulationandthearticulator.28Strictlyspeaking,asnotwospellingsofasoundareidentical, the concept of phoneme (representing actual instances and/or spellingvariantsa1,a2,...,anofasound/a/)hasbeenintroduced.29

1.Language reaches itswritten phasewhen themeans for its transcription,mostoftenanalphabet,30havebeendeveloped.Thedescriptivenessandgeneralaccuracyof the archaic Indo-European phoneme inventories results from their phoneticcharacter.Unaffectedbyconventions, themain sourceofnon-phonetic spellingsorsimilar factors in the ancient Indo-European alphabets usually reflects the data asdirectlyaspossible,andtheyareusuallyacceptedassuchinacomparativestudy.Interms of minor exceptions, note the following phonological remarks concerningcertainindividualIndo-Europeanlanguages:(a)ContinuingtheSumerian ideogrammatictradition,theOldAnatolian languages(i., Pal. CLu. and HLu.) are syllabic, not phonetic. Consequently, phoneticapproximationsareusedforthepresentationoftheOldAnatoliandata(e.g.i.e#zi

28Forphonemes(sounds),seeLadefoged&Maddieson1996.Forphonetics,seeLaver1994.29ComparethefamousdefinitionofDanielJones1950,accordingtowhomaphonemeisafamilyofsounds.30For the close connectionbetween alphabetand phoneme inventory,compareMeriggi (1966:8):[]diejenige,diedenuraltenBegriffBuchstabeninderneuenMaskierungalsPhonemrettenwill.

23

iswrittenfortheattestedi.e-e#-ziis),apracticealsofollowedinthisstudy.Beingsecondary (builtuponprimarydata), theseapproximationsare susceptible toerror,and comparative evidence is particularly important for the elimination of possiblemistakes.(b)TheIndo-Europeanlanguagesareusuallyattestedintheirowninheritedwritingsystems, but transcribed in the Latin alphabet (except for Greek). The scholarlytranspositionsarenotnecessarily flawless,and scrutinyoccasionally required in thephonologicalconsiderationsinvolvingthelatter.31(c) From a comparative point of view, the allophonicalternation of phonemes iscaused by sound changes in varying environments. Avestan is especially rich inallophonic alternation in its alphabet, possibly reflecting its status as a sacredlanguage. It is not uncommon thatAvestan allophones cannot be explained on asynchronicbasis,but insteadrequireahistoricalexplanationoutsideofthereceivedphonemeparadigm.

2. The comparative method of reconstruction is not primarily interested in thephonemeinventoriesoftheindividualIndo-Europeanlanguages.AlthoughallIndo-European languagespreservesomeproto-phonemesassuch,allhavegone throughmultiple and successive sound changes, leaving the surface level ambiguous to adegree. In particular, the following types of changes are commonplacewithin theIndo-Europeanlanguages:(a)Loss(ordisappearance)ofaproto-soundinalanguage(e.g.PIE*Gr.).(b)Merger(orconvergence)oforiginallydistinctproto-phonemesinalanguage(e.g.PIE*th*dY*kY*YGr.;).(c)Splitofanoriginalproto-phonemeasconditionedbyenvironment(e.g.inPIE*hLat.cgh,etc.).Owingtothesecondarynatureofatleastsomeattestedphonemes,thecomparativemethod of reconstruction eliminates secondary phonemes by postulating therespective sound lawsbeforeentering intoconclusions, thus focusingon theproto-phonemeinventoryasthecommondenominatorofthecognates.

11.2.3 The historical PIE phoneme inventories

0.ThehistoricalPIEphonemeinventorieswillbebrieflypresentedinordertotestthemagainsttheenrichedIndo-Europeandata.Thoughoutdatedincertainaspects,theNeogrammarian phoneme inventory is the common starting point of all Indo-European reconstruction theories (including the one presented in this study), andthusservesasanaturalpointofreferenceforthehistoryanddevelopmentofthePIEphonemeinventory.32

31Foranexampleofafailureintransliterationanditsconsequences,seeChapter4forthediscussiononthevoicedaspirateseries(mediaeaspiratae)ofSanskrit,historicallymiswrittenasOInd.bhdhghjhhinsteadofthepropernotationOInd.bYdYgYjYY.32ForDerLautbestandderidg.Ursprache,seeBrugmann(Grundr21:92-93).

24

Within the phoneme inventory, three functional classes of phonemes, vowels(V),resonants(R)andobstruents(C)aredistinguishedanddealtwithrespectivelyinChapters2,3and4.Beginningwiththe laryngealPIE*, theoverallpictureoftheresearchhistorycanbesketchedasfollows:(a)The laryngealPIE*,which isabsent from theNeogrammarian reconstruction,cannowaddedtotheproto-languagebasedonOldAnatolian,asalreadydiscoveredbymonolaryngealism:

Neogr. (Brugmann,Osthoff,Pokorny,Kronasseretal.) Monolar. PIE* (Zgusta,Laroche,Szemernyi,Tischleretal.)

Thevariationsofthenowoutdatedmultilaryngealismwillbediscussedsubsequentlyintheirrelevantcontexts.(b)At itshighpoint, theNeogrammarian vowel systemNeogr.*V containedeightcorrespondence sets, provided below with the respective vowel system of thelaryngealtheory:

*a-quality: *o-quality: *e-quality:

Neogr. *T *a * * *o * *e *LT. *h2*h2e/ *eh2 *h3e *eh3 *e *eh1

(c)TheNeogrammarian system of sonants33 contained glides (U), liquids (L) andnasals(N),asindicatedinthetablebelow:

Neogr. * i C iV * u %C uV *l C V *r C V *m C V *n C V

Itwasfurtherclaimedthatthelongsonantsstoodfortherespectiveshortones,plusNeogr.*T,nowwrittenas*Hinthelaryngealtheory.(d)TheNeogrammarianobstruentsystemconsistedofthefollowingitems:

Plosives: Fricatives:

Neogr. *p t k k s *ph th kh h kh sh h *b d g z *bh dh gh h h zh h

Thefollowinginitialremarksarerespectivelymadeforeachcategoryofobjects:

1.Themonolaryngealismhas its roots inZgustas (1951)observation that there isone and only one laryngeal PIE * (R i. , CLu. , Pal. , HLu. ), which iscomparatively inferable from theOldAnatolian (and other Indo-European) data.This has now been confirmed by Johann Tischlers Hethitisches etymologischesGlossar (HEG 1977ff.), proving that Zgustas conjecture was both sufficient and

33 Note that in this study, the term resonantis used for PIE *i u r l n m, whereas the termsonantsreferstoBrugmannsandOsthoffssyllabicsonants.

25

necessary. This decisive success provides an inductive starting point for thecomparative reconstruction of the PIE laryngeal, but monolaryngealism has notinferredthepropertiesofthecoversymbolPIE*asanindependentsegmentanditsbehaviourinallenvironments,basedonthecomparativemethod.

2.AsfortheIndo-Europeanvowelsystemand itsrelationtothecoversymbolPIE*,thefollowingtheorieshavebeenproposed:(a) At its high point, the Neogrammarian vowel system consisted of eight coversymbolsforvowels:

Neogr.*T,a,(a-quality) *,o,(o-quality) *e,(e-quality).

Testedagainst theenricheddata, theNeogrammarian vowel system isadequate interms of the number of cover symbols and their derivation. Eight distinctcorrespondencesetscanbeinductivelyobtainedfromthedata,andnoneofthecoversymbols are redundant. In the absence of the laryngeal, the traditional system isoutdated.Inparticular,themutualrelationshipsofvowelsandthelaryngealandtheablautpatternsrequireathoroughrevision.(b)Based on Saussures ideas,Mller (1879, 1880, 1906:vi= ML) presented theclassical three-laryngealism (now competing with Brugmanns comparativereconstructionofproto-vowels)indicatedinthefollowingtable:

Neogr. *T *a * * *o * *e * ML. *A*Ae/ *eA *Oe/ *eO *Ee *eE

This theorywasbasedonSaussures(1878=DS*)single fundamental(inmodernterms pre-proto-vowel) *e34 of two coefficients sonantiques: an a-colouring*A(Neogr.*T=LTh2)andano-colouring*O(=LT*h3),withrulesofcompensatorylengthening and colouring obtained by structural reasoning.35 For the sake ofsimilaritywiththeSemiticsystemoflaryngeals,Mlleraddedyetanotheritem*E(=LT h1) and projected the assumed Proto-Semitic root shapeC1C2C3 onto Proto-Indo-European,36 thus giving birth to the laryngeal theory.37 Unsurprisingly, thislaryngeal theory conflictedwith reality: after the emergence of theOldAnatoliandata, Mllers original proposition of three laryngeals has been graduallydowngraded. By switching to a notation in which E, A, O indicate laryngealspreserved inOldAnatolian andh1,h2,h3 laryngeals thathavebeen lost (ornever

34SeeSaussure(Rec.127):Lephonmea1[=*e]estlavoyelleradicaledetouteslesracines.Ilpeuttreseulformer levocalismede laracineoubientresuividunesecondesonantequenousavonsappelecoefficientsonantique.[...]Dansdecertainesconditions,quinesontpasconnues,a1[*e]estremplacpara2[*o];dansautres,mieuxconnues,ilestexpuls.35TheruleofcompensatorylengtheningreferstothepostulatesLT**eh1O*;LT**eh2O*;LT**eh3O*andthe colouringrulestothepostulatesLT**h1eO*e;LT**h2eO*a;LT**h3eO*o.36Thus,Lindeman(1987:25)writes:Initscommonlyacceptedformthe LaryngealTheoryassumestheexistenceinEarlyIndo-Europeanof(atleast)threelaryngealconsonants[...].37Forthelaryngealtheory,seeHendriksen1941,Puhvel(1960:1-13),Polom1965,Szemernyi1973,Jonsson1978,Lindeman(1982:63-64,1987:78-79),andBammesberger(1984:38).

26

never existed), wemay summarize the subsequent developments of the theory asfollows:

1.Benvenistes(1935=BENV.)assumedthreelaryngeals:twopreserved(*A=i.,*O=i.)andonelostitem(*h1=i.).

2.Kuryowicz(1935:75f.,254f.=KUR.)assumedfour laryngeals:twopreserved(AandO=i.)andtwolostlaryngeals(h1andh2[=LTh4]=i.).

3.Eichners(1973=EICH.)assumedthreelaryngeals:onepreserved(*A=i.)andtwolost(*h1h3=i.).

4.Puhvels (1965=PUH.) theory supposeseand six laryngeals,ofwhich threehavebeen assumedlypreserved inOldAnatolian: *E,A,O and three lost (h1,h2,h3).

38Mllers laryngeal theoryhas split into two subgroups.One favoursweakening theoriginalpropositionofthenumberofpreservedlaryngeals(BenvenisteandEichner)and one adds the number of assumed laryngeals (Kuryowicz and Puhvel) tocompensate:

ML*EAO BENV. *h1*A*O KUR. *A*O *h1*h2 EICH. *h1*A*h3 PUH. *E*A*O *h1*h2*h3

(c) Themonolaryngeal theory of Indo-Europeanvocalism is currently in its earlyphase,inessenceconsistingofthefollowing:

1.Zgusta(1951),thefirsttoreconstructasinglelaryngealPIE*Hcoincidingwithi. ,argues for the favourofa colourless (ornon-colouring) item.Byadding thethreeshortvowels*e,a,oandfollowingtheruleofcompensatorylengthening(*eHO,*aHO,*oHO),Zgustastheoryhasonlyfourproto-phonemes(ZG.*H*e*a*o)andthreerules(ofcompensatorylengthening).

2.Similarly,Szemernyi(1967:96-7=SZ)positsonenon-colouringlaryngealPIE*H(=i.)andsixvowels*e,a,o,,,;thus,hedisagreeswithZgusta,favouringtheoriginalquantityinsteadofcompensatorylengtheningnotrequiredinhissystem.Fromthepointofviewofthedata,itcanbereadilysaidthatthissolutionissuperiorto thatofZgusta,becauseSzemernyis system contains theoriginal vddhi vowelsproventoexistindependentlyoflaryngeals.

(d)Inordertoprovideanoverviewoftheinitialassumptions,thevocalismsandthelaryngealsofthetheoriesaresummarizedinthefollowingtable,whereindicatesacorrespondencesetmissingfromatheory:

Vowels: Laryngeal:

Neogr.*T a o e DS. *A eA (o) eO e eA

38ForPuhvelsmotivationfortheexpansionofthenumberoflaryngealstomorethanthree,seeHED3: v-vi: Those who have insisted on postulating a set (preferably low) number of laryngealsandhewingtothemreligiouslyhavelulledthemselvesintoafalseandprematurecircularity.

27

ML *AAe/ eA Oe/ eO Ee eE ZG. a aH o oH e eH *H SZ. *T? a o e *h

Thetheories lackat leastonecorrespondenceset,withtheresultthatnoneofthemare complete or acceptable as the basis of a comparative reconstruction theory assuch.However,Brugmanns reconstruction is themost accurate description of theIndo-European vocalism,and theabsenceof the laryngeal canbe correctedby theaddition of the critical sound law established by the laryngeal theory andmonolaryngealism:

PIE*R i.,Pal.,CLu.,HLu.:RV/,Gr.,Lat.,etc.

Thus,acompletesetofcoversymbolsemergeswhenthetwotheoriesarecombined:

*T *a * * *o * *e * *.

In Chapter 2, when the cover symbols are replaced with the actual Proto-Indo-Europeanvalues,thissolutionwillbeshownasbothnecessaryandsufficient.39

3.Concerning the resonants, functionallydefinedasphonemeshavingvocalic ()andconsonantal(R)allophones,threetheorieshavebeensuggested:(a)TheNeogrammariansystemofsonantscontainedthepostulates:

Neogr. * i C iV * u %C uV *l C V *r C V *m C V *n C V

Here the long sonants stand for short sonants plus schwa (= + T). In thelaryngealtheory,Neogr.*Tisreplacedwith*Hinacompletelyisomorphicsystem:

LT * i iHC iHV * u uHC uHV *m HC HV *n HC HV *l HC HV *r HC HV

(b)Theschwasecundumschool,initiatedbySchmidt,acceptsBrugmannsandOsthoffs correspondence sets,butexplains theepenthetic svarabhaktivocalismsofthecognatesasreflectingaschwasecundum(writtenas*M)insteadofthezerograde.(c)The third tradition, dating back to the period preceding the theory of syllabicsonants,isthecomparativeone.Accordingtothisview,thoughneverformulatedasafull-scale theory, the identical vocalisms of cognates are directly compared andpostulated to the proto-language when confirmed by at least two witnesses. Thisapproachcanbeillustrated,forinstance,byVernersreconstruction(1877:125):

[G]erm.folliaf. flle(ahd.fullida)=altind.prtdss.,vongerm.folla- voll(goth.fulla-,an.full-r,as.full,ags.ful,ahd.fol)=altind.pr-,dss.

39 For an interpretation of the historical connection between the Neogrammarians andmonolaryngealism, see Eichner (1988:128): Er [= derMonolaryngalismus] bildet imGrunde dieFortsetzung der Brugmannischen Auffassungen vermerhrt um die Ansicht, da man nach derEntdeckungderanatolischenEvidenznichtmehrganzohneLaryngalauskommt.

28

Here,inessence,anoriginalvowelispostulatedbyatleasttwowitnesses:

PIE*pulno- R RV.p%r-,Go.full-,ORus.p&ln&-,etc.

4.Fortheobstruents,functionallydefinedasphonemeswithoutvocalicallophones,theNeogrammarians postulated a system of plosives and fricatives, comprising oftwenty-eightproto-phonemes.

Plosives: Fricatives:

1. 2. 3. 4. 5. 6. 7. *p t k k s (1) *ph th kh h kh sh h (2) *b d g z (3) *bh dh gh h h zh h (4)

As regards these items, discussed in full in Chapter 4, the following preliminaryremarkscanbemade:

1.Columns1-3representthedecem-taihunisogloss,viz.theproblemofthefourmannersofarticulationTThDDh(appearinginrows1-4).

2.Columns 3-5 represent theCentum-Satem isogloss, viz. the problem of thethreePIEvelarseries(Neogr.*k::k,etc.).

3.Columns 6-7 represent theNeogrammarian fricative system, consistingof aseries of sibilants and a series of interdental fricatives (or thorns), but lacking theplaceofarticulationforlaryngeal(s).(a) Three main theories have been presented for the decem-taihun isogloss,consistingoftheseriesTThDDY:

1.Thetraditional(Neogrammarian)theorywithtwentycomparativelyobtainedcoversymbolsforplosives,asalreadyindicatedinthetableabove.

2. The root constrainttheory of Meillet and Magnusson, which claims acomplementarydistributionfortheseriesmediae(D)intherootswithtwosuccessiveplosives,thusimplyingitssecondarycharacter.

3. Based on Saussures suggestion (generalized by Kuryowicz), the series oftenues aspiratae is eliminated by means of segmental analysis in mainstreamlaryngealtheory.

Neogr.*phthkhhkh R LT*p+h2t+h2k+h2+h2k+h2

The remaining systemof three series (*T :D :Dh) is the startingpointof the so-calledglottalictheory,modulationsofwhicharebasedontypologicalconsiderations.(b)ThesecondpartoftheplosiveproblemdealswiththeCentum-Satemisogloss(i.e.the existence of the three velar series (Neogr. *k : : k)).Currently there are anumberofattemptstodealwiththisquestion:

1.TheNeogrammariantheory,consistingoftwelveproto-phonemes(Neogr.*kk;*khhkh;*g;*ghhh), isobtainedthroughthecomparativemethod.Although correct in terms of its contents, the theory is typologically problematic,becausenosatisfactoryparallelsinthelanguagesoftheworldhaveemerged.

29

2. Attempts to eliminate one series by means of environments result in areduction of the system to only two original series. In this regard, all the possiblesubsets of two original phonemes (i.e. *k+*, *k+*k and *+*k) have beensuggested,butwithlittlesuccess.

3.When segmental analyses of the velars (Neogr. *k= *ku) (Reichelt) andpalatals(Neogr.*=*ki)(Szemernyi)arecombined intoasingle theory, the twoapproachesonlyleavetheplainvelarseriesfortheproto-language,thusremovingthetypologicalproblemofhavingthreeseries.(c)TheNeogrammariansystemoffricativesinpartartificial,inpartdeductive(vs.inductive)wasdefectiveintermsofthelaryngealplaceofarticulation.Thesituationisdiscussed separately in thenextparagraph inorder to illustrate theprinciplesofsegmentalanalysis.

5.InordertoguaranteetheminimalcharacterofthephonemeinventoryofSystemPIE, a combinatory analysis of phonemes is carried out for vowels, resonants andobstruentsintherespectivechaptersofthestudy.Thetestingofthepostulatedproto-phoneme systems can be exemplifiedherewith an analysisof theNeogrammariansystemoffricatives,inrelationtowhichonecanobservethefollowing:(a)OfthesibilantsNeogr.*sshzzh,onlyNeogr.*sand*zexistasoutcomesofthecomparativemethod.ThesibilantsNeogr.*shzhwerepostulatedonthebasisofthetypologyof the fourobstruent seriesNeogr.TThDDY (Systemzwang).Since theproper (comparative) reconstruction must be exclusively based on data, theconstructionsleaningtostructuresortypologiesandtheirpostulates(hereNeogr.shandzh)areunacceptable.(b)Thepostulationoftheso-calledthornseries(i.e.thefourinterdentalfricatives)

Neogr.* *h * *h (Grundr21:790)

isbasedonacomparisonofsibilants(inIndo-Iranianandelsewhere)anddentals(inGreek).Thedefinition canbe shown tobeerroneous,because the fulldataof thealleged examples reveal both sibilants and dentals in Greek (and occasionallyelsewhereaswell).No independentsegment is tobereconstructedbecausesibilantand dental extensions (marked I and II) exist simultaneously. The case can beillustrated,forinstance,withthedata:

1.Neogr.*ghho(.)-Erde,Ton(adv.)unter,unten(P.414f.)

I)PIE*ghso-

RV.kpvant- (m.)BeschtzerderErde(WbRV.362) RV.kapvant- (m.)BeschtzerderErde(WbRV.362) Gr.?CBAB- (m.)TonzumBleichen(GEW2:256) Att.bC8IE~- (N.)=bC8-I;~D(SchwyzerGrGr.1:326) RV.km- (f.)dieErde,derErdboden(WbRV.363) Gr.bA8@BD (a.)=Gr.b

30

Gr.I;@ (f.)Erde,Erdboden,Land(GEW2:1098-9) Phryg.674@?4 (f.)MotherEarth(P.414) i.gadan (adv.)unten(HHand.76,HEG1:539ff.)

Both a sibilant and a dental extension exist, due to which the postulation of anunderlyingthornisillegitimate.

2. Neogr. *te- bauen, zimmern, verfertigen, schneiden, usw. (P. 1058-59,KEWA1:612-3)

I)PIE*tes-

RV.tk- (ao.)zimmern,verfertigen(WbRV.511,tkati) TochB.tks- (vb.)chopup,grindup(DTochB.286,tksoym) LAv.t#- (pr.)(inScheite)zerlegen(AIWb.645,t#ti[3sg]) Lat.texo- (vb.)bauen,zimmeren(WH2:678,tex[1sg]) gAv.ta#n- (m.obl.)Bildner,Schpfer(AIWb.645,ta#n[sgG]) Gr.F}I@:- (f.)Handwerk,Kunst(fertigkeit),List(GEW2:889)

II)PIE*tet-

Gr.F}=FK@- (m.)Zimmermann,Handwerker(GEW2:867) LinB.tekton- (m.)Zimmermann(GEW3:183,te-ko-to-ne) Gr.F}=F4

31

especiallymeanings are studied, and as the general problems of the field arewellknownitsufficestorefertothemostrelevantissuesforthereconstructionofProto-Indo-European.41

1.Meaning can be defined inmanyways, parallel or divergent.42 In comparativeIndo-Europeanlinguistics,themainvehicleforthedeliveryofmeaningistranslation.As translation is a concretemeasurable object, it is not intended that it involve aphilosophically loaded discussion about the meaning of meaning.43 It should,however,bekept inmind thatmorphemespresupposemeaningand reconstructionpresupposesmorphemes; accordingly,meaning is by nomeans a trivial concept.44Systems lackingproper reference tomeaning (seeChomsky)areof limited interestforIndo-European linguistics,wheretranslationsplayasignificant(non-trivial)partonseverallevels.45(a)Translationsareofteninterpretationsofmultiplecontextualfactswhereanerrormayoccur.AnexampleofanerroneousmeaningisprovidedbyTischler(HEG1:164-65)explaininghowacertaintranslation

i.apadia- (vb.)schlagen,verletzen,tten(HHand.40)

shouldbepostulated insteadof theearly Diener,Untergebener,whichwasbasedon amisunderstanding of the context. Such corrections, oncemade, can often beverified(orfalsified)bycomparativeanalysis.46(b)Itisnotuncommonforthetranslationofaword(oramorpheme)tobemissing.Thisisparticularlycommonwithhapaxesandinonomastica.Inordertorecoverthisvitalmaterial,Indo-Europeanlinguisticsusesmultiplemethodologiestosupplementthemissingtranslations,butinparticularthecomparativemethod.Asanexampleofsupplementingthemissingmeaning,IquoteanancientCelticpropername:

OGaul.mageno- (PN.m.)-(?)-(ACSS.2:374).

Thoughnotranslationisavailable,themethodallowsforacomparisonwiththelaterCelticitems:

Cymr.maen- (m.)pierre:stone(LEIAM-9) Bret.mean- (m.)Stein(P.709) OBret.cronnmain- (sb.)pierreronde(LEIAM-9)

41Forageneralintroductiontosemantics,seeLyons1977.42Forinstance,typesofdefinitionsincludeostensive,iconic,nominal,extensional,grammaticalandsoforth.43Inthisstudy,hybridtranslationsquotingdictionariesintheiroriginallanguagesareusedinordertominimizethepossibilityoferror.44See, for instance,Nymans sketchof theconnection (1982:32):[...] the so-called sign ruleswhichrelateasignatumtoitssignans,thusmakingupamorpheme(Andersen1980:3)oraphoneme[].45SeealsoMeriggi(1966:3):[]dieasemantischeSprachwissenschaft [],beidermanLauteundFormen,abernichtihreBedeutunguntersuchensoll,istmirsinnlose.46In thiscase,Tischlers translation isnowsupportedby theetymologyHes.\874@B-(LSJ.182)=f874@-schwach,gebrechlich(GEW1:639-40).

32

Walde(andPokorny,P.709)correctlyreconstructedPCelt.*mageno-forthelatter,butastheprototypenowcoincideswiththeactuallyattestedancientform,thelattercanbefurnishedwiththetranslation:

OGaul.mageno- (PN.m.)Stein(?)(ACSS.2:374).

Sincenosound lawsareviolated,andthepostulatedproto-form isreplacedwithanactually attested form of equal shape, the comparisons of the type are allowedregardlessofthesubgroupinvolved.47

2.AsmentionedbyMatthews(1991:223),theproblemoftherelationshipbetweenmorphemesandrealitywasalreadyunderstoodinAncientGreece:

One of the oldest findings about the language is that the forms of lexical elementsgenerally do not bear a natural relation to theirmeanings.AsHermogenes put it in adialogue by Plato, the names of the things are justified by nothingmore than rule andcustom.(Cratylos384d)

However, some modern formulations of the idea, especially the extremeinterpretationofSaussures slogan arbitrarinessofmeaning,doesnot serve Indo-European linguistics inanoptimalmanner. Inparticular, if the rulesmentionedbyHermogenesarenotrecognized,severalactualcriteriagoverningthealternationsofmeaningarelost:(a)ThePIE roots are attested inmultiple vocalizations (including zero), called itsablautbases.Theablautvowelsmodifiedthemeaningoftheroottovaryingdegreesinamannernotyetcompletelyunderstood.(b)ThePIEstemsbelongtovariousgrammaticalfunctions(e.g.verbs,substantives,adjectives, etc.) and their subclasses (e.g. active :medium/deponent : passive andtransitive:intransitive,etc.).Suchalternationsarereflectedinregular(vs.arbitrary)changesofmeaning.

3.The original PIE derivation and the subsequent sound changes have semanticconsequences,especiallyforthefollowingphenomena:(a)Homonymsmorphemeswithanidenticalphonologicalshape,butetymologicallyincompatiblemeaningsarecommonplaceboth inProto-Indo-EuropeanandIndo-European:

/(a1,a2,...,an)x /(a1,a2,...,an)y.

Thecomparativemethodsplitshomonyms,arrangesthemorphemesunderrespectiverootsmnbasedontheirsemanticvalues,andeliminatesmergersintheprocess.(b)Polysemydescribesdifferentbutultimately connectedmeaningsofan identicalsequenceofphonemes,suchas:

/=(a1,a2,...,an) df x1,x2,...,xn.47InthedigitalizedplatformofthePIELexicon,itwillbepossibletolistallthemorphologicalmatchesallowed by sound laws to test the available translations. Even if no match is found, all possibleetymologieshave been attempted and the reasons for their failure systematically codified; this alsoconstitutesascientificresult.

33

Suchvariationcanbetracedbacktoarangeoffactors,suchasthedifferencebetweenthe real objects designated (e.g. ModEng. plain = clear, unadorned, obvious,etc.),thegrammaticalclassesofthestems,andsoforth.Fromacomparativepointofview,polysemyreferstoitemswithacommonsemanticfieldandroot.(c)Synonymsorparaphrasestheforms/=(a1,a2,...,an)and/=(b1,b2,...,bm)withthe same meaning, but distinctive phonetic structure are widespread in Indo-European.48EvenSanskrit,knownforitssynonyms,palesincomparisonwithProto-Indo-European, implying that the one meaning, one form principle cannot befollowed literally in Indo-European linguistics. The principle is helpful indistinguishing formswith incompatiblemeanings, but it should be recognized thatmultipleobjectswithidenticalmeaningaresupportedbythecomparativemethod.(d) It is not uncommon for a stem to have a doublemeaning, thus revealing acompound rather than a simple word. In such cases it is still possible to achievecorrespondencesbysegmentation,as the twomorphemesand twomeaningscanbeattachedtotwodifferentroots.Anexampleofsuchanalysisisfoundin:

Go.aldomin- (m./n.)6C4D:oldage(GoEtD.25).

Herethefirstcomponent(Go.aldo)correspondstothemeaningold,asaresultofwhichGo. min- is leftwith themeaning age,which still currently has no knowncognates, according toLehmann (GoEtD. 25).However, the comparisonwithOldAnatolianresultsinadirectmatchin:

PIE*men-Zeit

i.men- (n.obl.)Zeit(HEG2:171,me-e-ni[sgL]) Go.min- (m./n.)age(GoEtD.25)49

Generallyspeaking,thedataactuallycontainsmoresegmentsthanjustthewords(orstems),andsemantichintsoftenleadtosuccessfulsegmentation.

4. Semantic bridges assumed changes of meaning through a postulated(hypothetical)meaningarerelativetothephonemeinventoryandsoundlawsystemathand. In general, improvements inphonology result in increasedmorphologicaldistinctions, sometimes confirming and sometimes specifying a semantic bridge.Perhaps most often, however, a semantic bridge turns out to be artificial. Anillustration of this can be found with the emergence of PIE * (= i. ). In theNeogrammarian system, a-vocalism (Neogr. *T a ) referred to vowels; notconsidered root radicals, they were therefore allowed to alternate with zero.According to themodern lineof thought,Neogr.*Ta indicatesPIE* (h2),aradical consonant, thus often necessitating distinctionswithin the traditional roots

48Abrief look at the Indo-European synonymdictionaries likeWatkins 19923 andMallory-Adams1997confirmsthatsynonymyiswidespreadwithinthegroup.49 For an alternative extension of the root obtained similarly by Fraenkel, see his outstandingcomparisonofLi. tuoml (adv.) ineinem fort : rightaway (LiEtWb.445)andGo.mel- (n.) Zeit,Stunde(ANEtWb.376).

34

heldtogetherwithsemanticbridges.Therefore,inthePokorny-root*(a)ner-Mann,Mensch:Kraft,Rstigkeit,usw.thefollowingdistinctionsarenowobligatory:(a) PIE *ner *nor- man (P. 765). The undisputed \- inGreek (and Phrygian)impliesthatthisrootoriginallybeganwithPIE*:

Gr.\@~C- (m.)Mann(GEW1:107-8) NeoPhryg.4@4C- (m.)Mann(P.765) RV.nr- (m.obl.)Mann,Mensch(EWA2:19-20) RV.n- (m.obl.)Mann,Mensch(WbRV.748-50,nbhis[I])

(b)PIE*ner-*nor-strength,strong(P.38-39,HEG1:28).HerebothGreekandOldAnatolianindicatethattherootdidnotbeginwithalaryngeal:

Cymr.ner (m.)chef,seigneur(LEIAN-10) Osc.niir- (m.)princeps(LEIAN-10,niir[sgN]) RV.nr- (m.)Held,Krieger(vonGttern)(WbRV.748) RV.n- (m.)Held,Krieger(vonGttern)(WbRV.748) Hes.@C8?@B- (a.)?}64D,B>D(LSJ.1186) Gr.@KC}K (vb.)operate,effect,etc.(Hes.@KC8)b@8C68) CLu.anari- (c.)Rstigkeit,Lebenskraft (DLL26-27) i.anari- (c.)Rstigkeit,Lebenskraft,Vitalitt(HHand.16) Gr.@}CFB- (m.)Hes.l}C4A)@}CFBD(LSJ.1170)50 OIr.nert (n.)force,vigueur,puissance,vertu(LEIAN-10)

Thesemanticbridgefailsnotonlyformorphologicalreasons,butbecauseamanisnotnecessarilychief,heroorevenstrong.Definitely,however,heisabreather,aswassuggestedalreadybyBrugmann(Grundr21:351),connectingGr.\@~Ctotheroot

PIE*en-breath(P.38-39):

RV.sm(...)n- (pf.)leben,atmen(WbRV.50,sm(...)na[3sg]) Go.uzn- (pret.)aus-atmen(GoEtD.385,uzn[3sg]) Osc.anamo- (m.)Seele,Geist,Gesinnung,Gemt,Mt(WH1:49)

5.Finally, it shouldbeobserved that thepostulationofaPIEmorpheme requiresthatboththeformalandsemanticequationsmatch.Therefore,twomorphemes

/x(a1,a2,...,an)dfx = /y(a1,a2,...,an)dfy

are identicalonly ifboththeproposition/x(a1,a2,...,an)=/y(a1,a2,...,an)andthepropositionx=yaretrue.51

50Forthealternationofthemeanings,compareGr.|@4AHerrscher,Herr,Frst(GEW1:102)andLi.vnagasHabicht(LiEtWb.1194).51CompareCampbell(2004:356):Agenerallyacceptedprinciple(advocatedbyMeillet)permitsonlycomparisonswhichinvolvebothsoundandmeaningtogether.Anexceptiontotheruleconsistsoftheformswithunknownmeaning(formula/x(a1,a2,...,an)df-(?)-).Inordertotestwhetherasuitabletranslation can be found, it is naturally allowed to propose equations from amongmorphologicallypossiblematchesinordertoarriveatthemissingtranslation.

35

11.3.2 Semantic fields of PIE root matrices

0.ThePIE roots formed tree-shaped structures called rootmatriceswith awiderange of meanings defining the semantic field of the matrix.52 The existence ofsemanticfieldshasbeenunderstoodeversincetheSanskritgrammariansconstructedroots not restricted to a single but several meanings. The preconditions for acomprehensivescientificstudyofthesemanticfieldshaveonlybeencreatedrecentlyasaconsequenceofthecodificationofthecompleteIndo-Europeanmaterialandtheadvancementsofcomputationallinguistics.HereIwillnotproposeafull-scaletheoryof semantic fields of the Indo-European languages, but simply sketch the generalsituationinapreliminarymannerforthelimitedpurposesofthisstudy.

1.Thealternationofmeaningsofasemanticfield isgoverned,for instance,bythefollowingregularfactors:(a)Thegrammaticalcategoriesofthestemsbelongingtoarootmatrix.Forinstance,itiscommonplacewithinPIErootsthataverbmeaningtogo(withfourlegs),runisassociated with substantives meaning horse, birdand/or foot/leg, an adjectivehasty, anumeralmeaning four, apreposition(s)meaning for(ward), forth, etc.,andanadverbmeaning fast.The subcategoriesof the stems (suchas transitivity,gender,etc.)governregularchangesofmeaning,whichcanbedigitallymanaged.(b)Thefactsoftheexternalrealityarereflectedinthedimensionsofasemanticfield.Thus,inPIE,averbmeaning makeisoftenaccompaniedbyasubstantivemeaninghand(ormoreabstractly,work),anadjectivemeaningcapable,mighty,anumeralmeaning five,andso forth.Thereasons for thealternationarereadilyunderstood(themeaning hand isdefinedby the (five) fingersandactionsperformedby thehand),and thiskindofphenomenoncanalsoberegulated,at least toareasonabledegree.(c)Rootswithparallelextensionswithan identicalmeaning (ornearly so)arenotuncommon in Proto-Indo-European (and Indo-European). This can be illustratedwith the traditional entry Neogr. *mn- moon, month (P. 731), actually a *n-extensionoftherootPIEme-m-luna:53

PIEme-

OInd.ma- (m.)Moon(MonWil.771,Lex.ma[sgN]) TochA.maktt- (m.)dealuna(Poucha212,maktt[sgN])

52Note that the term semantic fieldisusedhere inadifferentsense than in itsoriginalusage.Thestandarddefinitionand its summaryareadvancedbyFox (1995:116)as follows:JostTrier []putforward the theory of the semantic fields (Trier, 1931).According to this theory, it is possible toidentifyareasofthevocabulary(fields)withinwhichmeaningsaremutuallydefininganddelimiting,thus forming systemswhichhave some affinity to those found inphonology andmorphology.Trierillustrated thisprinciplewith an analysisof the vocabularyof knowledgeinMiddleHighGerman,demonstratingthatvariouswordsusedcoveredthefieldinquestionwithoutgapsoroverlaps,andthatthefieldanditsstructurechangedinresponsetoculturaldevelopments.Forthisinternalmeaningofthetermsemanticfield,seefurtherHock(1991:305).53Fortheregularexplanationofvocalismsinvolved,seeChapter2.

36

PIEmn-

RV.m!cat- (a.)denMondverscheuchend(WbRV.1028) Li.mna- (m.)Monat,Mond(LiEtWb.435,mnas[sgN])

PIEms-

RV.candrmas- (m.obl.)Mond-(WbRV.436,candr-masas[G]) RV.ms- (m.)Mond,Monat(WbRV.1036,msam[sgA]) Arm.mahik (sb.)Mondsichel(ArmGr.1:191,mahik)54 Mars.mesen- (sb.)Mond(WbOU.472)

PIEmu-

El.?8- (m.)Monat,Mondsichel(GEW2:227,?8D[sgN]) OIcl.m%lin- (m.)Mond(ANEtWb.395,m%linn[sgN]) OIcl.mundilfari (PNm.)N.frdenVaterdesMondes(ANEtWb.395)

The semantic distinctions originally caused by the extensions remain temporarilyunknown,owing to the incomplete stateof Indo-European studies,but inprinciplethese can also be recoverablewhen a digital study of thematrices as independent(andcomparable)objectsbecomespossible.(d)Semanticfieldsofformallydistinctmatricescanbecomparedwitheachotherintermsofalternationsandparallelsofmeaning.Thusthecommonplacealternationofmeaningsfoot,go,hastyetc.recursin:

pi-gehen;Fu(fortheextendedroot,seeP.795,*pim-)

TochA.pe- (m.)pes:F(Poucha186,pe[sgN]) i.pai- (vb1.)gehen,marschieren,usw.(CHDP:19f.) OInd.paya- (vb.)togo,move(MonWil.585,payate[3sg])

(e) The scope of semantic fields can (and must) be tested using the proceduresketchedoutbySzemernyi(1977:306):

Ifanetymoninvolvestheassumptionofanunusualsemanticdevelopment,theresearchermustre-examinethephonologicalandmorphologicalaspectsofthederivation.

Assemanticalternationscanbeverifiedbymeansofcomparisonorrejectedduetoanabsenceofparallels,themorematricesarereconstructedthemoresolutionsthereare for semanticproblemsand themorepossible it is tobuildahighly regulatedtheory.

2.Generallyspeaking,themostinterestingpossibilitiesinIndo-Europeansemanticslieinnon-arbitraryalternationsofmeaning.

54As evidence againstHbschmanns suggestion of a hypothetic loan (without an Iranian startingpoint),notethea-colouringinArmenianandLithuanianacute,bothwithagreementinPIE*.

37

(a)Usuallytheassumptionofarbitrarinessofmeaningisunnecessaryormisleading:our first and foremost task in (P)IE semantics is todevelop ameansof regulatingnon-arbitrary semantic alternations and providing the study with precise tools toapproachameaningasaninductiveproblemwithasolution.(b)Even if themeanings of the shortest (primary) PIE roots,which serve as thestartingpointsofthematrices,eventuallyturnouttobearbitrary,ourtaskistoprovethisscientificallyinsteadofassumingarbitrarinessapriori.

3.Due to the translatabilityof theIndo-Europeandata into formulasofpredicatecalculus,semanticscanbestudiedasrigorouslyasmorphology.Therefore,insteadofattempting to ignore (or dismiss) it, semantics sh