LINGÜISTICA IX, /2 LJUBLJANA 1969 I LINGÜISTICA IX./2 LJUBLJANA 1969 Uredniški odbor — Comité de rédaction: BOJAN COP, ANTON GRAD, MILAN GROŠELJ, MITJA SKUBIC, STANKO ŠKERLJ — Rokopisi naj se pošiljajo na naslov: prof. Stanko škerlj, Filozofska fakulteta, oddelek za romanistiko, Aškerčeva 12, Ljubljana — Prière d'adresser les manuscrits à M. Stanko Škerlj, Filozofska fakulteta, • Oddelek za romanistiko, Aškerčeva 12, Ljubljana — Natisnila Univerzitetna tiskarna v Ljubljani v 450 izvodih L I N G U I S TIC A LETO IX. 1969 Anton Grad ŠE MALO O IMENIH KOSEZE, KOSEZ Kazno je, da tolikokrat obravnavani1 problem izvora imena in sloja kosesov, »pripadnikov slovenskega družbenega sloja med podaniki in plemstvom«2, ter z njimi povezanega krajevnega imena Koseze tudi pb najnovejših poskusih3 še vedno ni našel zadovoljive rešitve4 ter zato terja še nadaljnjih raziskav. , V svoji obširni in zanimivi razpravi »Praslovensko *kašegfh »Edling«5 se ugledni slavist Pr. Ramovš ni mogel ogreti za Oštirjevo domnevo6, po kateri naj bi ime kosez prihajalo iz starega traškega, še za Karpati izposojenega osebnega imena Koslggas (> *kosegb), ker se mu je to čisto upravičeno zdelo tako časovno kot prostorsko preveč oddaljeno; po Ramovševem mnenju, pa tudi po mnenju drugih raziskovalcev — in kot kažejo tudi zgodovinski podatki' — bi bila beseda *koseg-b, *kasegb, oziroma iz te osnove nastalo in še danes obstoječe krajevno ime Koseze, znana samo najzapadnejšim južnim Slovanom, Slovencem in Hrvatom, medtem ko prav na ozemlju starih Trakov, tj. v vzhodni polovici Balkana in v Dakiji, o njej ni sledov. Ramovš je utemeljeno zavrnil tudi razlago L. Pintarja7, ki je prvi na podlagi toponima Koseze rekonstruiral apelativ kosez, s tem da ga je izvajal iz besede knez, nastale iz germanskega kuning- »ex nobilitate ortus«, kot tudi Lessiakovo preumetno hipotezo8 o nastanku oziroma razvoju kosez < < Icaseg-b < -praslovansko *kozakb, avar. quazaq »svobodnik«, kar naj bi se bilo zgodilo že po naselitvi Slovanov v Alpah (7. stoletje). No, tudi Ramovševa ingeniožna razlaga o nastanku naše besede iz lango-bardskega plemenskega imena *gauting, ki bi po premiku i > s bilo prešlo v "gausing, a to obliko bi bili zapadni južni Slovani sprejeli kot *kosegb, 1 Gl. B. Grafenauer, Deset let proučevanja ustoličevanja koroških vojvod, kosezov in države karantanskih Slovencev, Zgodovinski časopis XVI (1962), str. 176—209; Id., Hrvati u Karantaniji, Historijski zbornik XI—XII (1958—1959), str. 221 s. 2 Gl. P. Bezlaj, Etimološki slovar slovenskega jezika. Poskusni zvezek, Ljubljana 1963, str. 17. 3 Gl. J. Stanonik, še o kosezih, Sodobnost XI (1963), str. 1038—1041. i Gl. B. Grafenauer, Cotsetlas = Kosezit, Sodobnost XI (1963), str. 1147—1150. 5 Razprave Znanstvenega društva za humanistične vede v Ljubljani II (1925), st.r. 303—326. s K. Oštir, Illyro-Thrakisches, Arhiv za arbanasku starinu-I (1924), str. 109. i L. Pintar, O krajnih imenih, Ljubljanski Zvon 1912, str. 600—605. s P. Lessiak, Edlinge — Kazaze, Carinthia I, 103 (1913), str. 81—94. 123 Anton Grad *kasegb, ni mogla zadovoljiti vseh raziskovalcev, ki jih je ta problem zanimal. Ni se strinjal z njo germanist Kelemina9, navajajoč zlasti dejstvo, da se je v langobardščini premik -i- > ss izvršil šele v Italiji10 ter torej langobardski *gausing ni mogel biti sprejet od Slovanov (kot *kosegh) že v severnem Po-donavju pod Karpati, kot je to mislil Ramovš; edinole s tak<5 zgodnjo iz-posoditvijo bi se dala namreč razložiti poznejša razširjenost besede kosez oziroma bolje rečeno njene izvedenke Koseze (Kaseze, Kdzaze) po vsem ozemlju, naseljenem s Slovenci (in morda deloma Hrvati) po njihovem prihodu v novo domovino. Seveda pa se pri vsem tem tudi upravičeno sprašujemo, kako to, da se je to ime obdržalo dejansko le na Slovenskem. Kelemina sam je nato prvi prišel na idejo o romanskem izvoru besede kosez-. romansko-langobardski pravni izraz *casagium, izvedenka iz besede časa (hiša), s pomenom domnicatum, domus principalis11 naj bi bil prišel k Slovencem iz furlanskih, zlasti patriarhovih pisarn, pomenil pa je. grad kot središče uprave; * casagium bi torej po Keleminovem mnenju bil osnova za štiri slovenska krajevna imena Koseze na ozemlju bivše Kranjske, za krajevno ime Kasase na štajerskem pri Celju in za štiri toponime Kdzaze na Koroškem12; toda zdi se nam, da bi za krajevna imena Koseze na bivšem Kranjskem eventualno še lahko računali s predlaganim Keleminovem roman-sko-langobardskim izvorom — seveda'pa je beseda *casagium sama na sebi več kot hipotetična —, pa je po. našem mnenju to mnogo teže s primeri na Štajerskem in Koroškem, pri katerih je tak nastanek zaradi geografske oddaljenosti in zgodovinskih dejstev prav malo verjeten13. ■ Tako se je germanist J. Stanonik, o. c., pred kratkim vrnil zopet na germanska tla pri iskanju izvora naših kosezov: staroangleški izraz cotsetla. »kmet svobodnik«, z mlajšimi variantami coscets, coscez, cozets, cozez, ga po svoji glasovni strukturi močno spominja na našega koseza ter se zato sprašuje, ali ne gre morda pri. imenih cotsetla in kosez — tudi zaradi pomenske sorodnosti — za isti izvor; tudi ko zgodovinar B. Grafenauer — po temeljiti analizi vsebinskih razlik med cotsetli in kosezi — trdi, da gre tudi pri različnih variantah, še bolj podobnih našemu imenu kosez, le za »igro grafičnih substitucij, ki ne morejo prav nič pomagati pri reševanju niti nastanka koseškega družbenega sloja niti podlag njegovega imena«1'1, se Stanonik ne umakne z germanskih tal in se — mislimo, da čisto upravičeno — čudi, da se pri iskanju nastanka imena kosez raziskovalci-lingvisti niso zatekli tudi h Grimmovemu slovarju, kjer bi bili lahko našli germansko sestavljenko cotset = cot (koča) + saeta, set (stanovalec, prebivalec); po Stanonikovem 9 J. Kelemina, Kazaz, kosez, Slavistična revija III (1950), str. 464—465. M Gl. E. Gamillscheg, Romania Germanica, II. Bd, Berlin-Leipzig (1935), str. 271. 11 Meyer-Lübke v svojem Rom. etymol. Wörterbuch besede casagium ne navaja! 12 Koseze pri Ljubljani, pri Vodicah zä Šmarno goro, pri Moravčah in pri Ilirski Bistrici; Kasaze pri Petrovčah zapadno od Celja; Kdzaze v podjunski dolini, pri Št. Petru vzhodno od Celovca, pri Bilčovsu in nad Blačami v ziljski dolini, gl. Ramovš, o. C., str. , 303. u O nekaterih drugih, še teže sprejemljivih etimologijah besede kosez gl. B. Grafenauer, Hrvati u Karantaniji, str, 221 s. m B. Grafenauer, Cotsetlas = Kosezi'!, Sodobnost XI, str. 1150. 124 Še malo o imenih Koseze, Jcosez mnenju bi bili po asimilaciji soglasniške skupine ts v ss v besedi kotset prišli do, oblike Kossat, izpričane v spodnji nemščini, a po konzonantskem premiku t > s do oblike Kossass, Kossesš (ki pa jih Grimm nima oziroma ne navaja)15. Ne trdi pa Stanonik nikjer, da prihaja naše ime kosez neposredno iz geografsko zelo oddaljenega kot.setla oziroma iz njegovih, zgoraj navedenih inačic, ki naj bi se bile iz stare angleščine oziroma z britanskega otočja vrnile na kontinent ter skupno z nemškim Kotsasz povzročile nastanek slovenske besede kosez, kot to nekako domneva J. Mal16. Takšni razlagi bi se — gledano le z jezikovnega vidika — upiralo tudi dejstvo, da so staroangleške oblike cotsets, cascez, eozets, c.ozez in tudi nemška Kotsasz poudarjene na prvem zlogu, kar bi pričakovali tudi za na njih temelječo izposojenko kosez, ki pa ima poudarek na drugem zlogu. 7 Pri iskanju izvora naše besede resnično preseneča predvsem dejstvo, da se ime kosez sploh ne pojavi v nobeni od starejših listin, da ga tudi Pleterš-nik ne pozna, a da ga, kot se misli, srečamo prvikrat in edinokrat šele v turjaških urbarskih zapisih iz prve polovice 16. stoletja (1522—1547) za Žužemberk na Dolenjskem, v katerih je, kot pravi Ramovš, o. c., str. 303, »ista oseba imenovana zdaj s slovenskim, zdaj z odgovarjajočim prevedenim nemškim imenom: Gregor Edlinger — Gregor Kasses, Juri Edlinger — Juri Kasses, Janntsche Edlinger in Michel des Janntsche Kosses sun, Páull Edlinger — Paulle Khases edlinger«. ' Pustimo za sedaj ob strani ta, za presojo naše besede važni zapis, h kateremu se bomo še povrnili, naj pa ugotovimo, da se krajevno ime Koseze, ki naj bi bilo nastalo iz apelativa kosez v njegovi množinski obliki17, prvikrat omenja v listini iz leta 1424 — torej kakih 100 let prej kot domnevani apelativ kosez in sicer v zapisku Edling, Kazezeh za današnje Kasaze pri Petrovčah zapadno od Celja18. Toda čeprav se torej toponim Kazeze najde zabeležen mnogo prej kot (domnevani) apelativ kosez v že zgoraj omenjenih urbarskih zapisih za Žužemberk, so vendar dosedanji raziskovalci precej enotnega mnenja — z izjemo J. Kelemine z etimološko razlago *easagium = grad —, da ñaj bi bili kosezi dali oziroma pustili nekaterim krajem, kjer so prebivali v večjem številu, svoje ime19. Vzrok za takšno razlago tiči v dejstvu, da za naše toponime Koseze, Kasaze, Kdzaze najdemo v stari dobi tudi vzporedno nemško ime Edling (en), ki naj bi bilo pomenilo »plemiško naselje« in ki naj bi bilo prevod slovenskega 15 J. Stanonik, Cotseti in kosezi, Sodobnost XXI (1964), str. 91—93. w J. Mal, Ist das Edlingerproblem wirklieh unlósbarl, Südost-Forschungen XXII (1963), str'. 177. 17 Pričakovali bi Kosezi, toda po akuzativni obliki Koseze je nastala nova nomi-nativna oblika Koseze ter je ime postalo ženskega spola, gl. Ramovš, o. c., str. 307. is Ta kraj se leta 1248 omenja* le z nemškim imenom Edelingen, »a to prvotno (podčrtal A. G.) nemško ime se je kasneje umaknilo slovenskemu (Kassasse)«, piše Ramovš, o. c., str. 303. m Tak nastanek krajevnega imena je seveda možen in celó pogosten, prim. npr. številne kraje, imenovane po Hrvatih na Koroškem v nekdanjem »pagus Crouuati«, v dolini gornje Mure, pa tudi toponim Hrvača pri Ribnici na Dolenjskem, gl. B. Grafenauer, Hrvati u Karantaniji; str. 207 ss. 125 Anton Grad imena Koseze (le-to bi torej bilo starejše kot nemško ime za te kraje, toda gl. opombo 18!); a po enačbi Koseze'= Edling(en) se je napravilo razmerje Koseze: Edling (en) = x: Edling (= plemič) in izluščila neznanka x = kosez; nastal je torej sklep, da je »ob času nemške kolonizacije (tj. od 10. stoletja dalje) slovenščina imela substantiv s pomenom nemškega Edelmann, Edling (= plemič),, kaj ti drugače ta prevod ne bi bil mogel nastati« (Ramovš, o. c., str. 303). Do drugačnega sklepa seveda raziskovalci-lingvisti, ki so skušali najti izvor besede kosez v starih osebnih imenih Kosingas, Kosentzes (kot bomo takoj videli) oziroma v starem plemenskem langobardskem imenu *gausing seveda tudi niso mogli priti. Morda je naš jezik res že v dobi naselitve v novi domovini imel substantiv kosez, ali, kot že rečeno, nikjer ga v stari dobi ne srečamo — kot npr. srečamo slovensko ime župan prvič pri alpskih Slovencih že leta 777 v obliki jopan (F. Kos, Grad. I 290) in od 13. stoletja dalje v oblikah kot supan, suppan ter alpsko nemško Siippan »sodnik, valpet, oskrbnik« (Bezlaj, o. c., sub v. župan). Sklepanje o obstoju naše besede v tako zgodnji dobi temelji namreč predvsem na hipotezi o njenem izvoru iz eponima Kosentzes, ki ga zgodovinar Konstantin Porfirogenet (10. stoletje) omenja kot enega vodij plemena Hrvatov in čigar ime bi po Ramovšu moglo v svoji grški obliki podajati slovansko obliko *kasedz (le-ta pa po Ramovševem mnenju temelji na njegovi domnevani langobardski osnovi *gauting > *gausing), tj. osnovo, ki jo imamo v slovenskih krajevnih imenih Koseze; potemtakem bi naj bili Hrvati prinesli Slovencem ime *kosez20, a čudno in presenetljivo je vsekakor dejstvo, da prav pri Hrvatih oziroma na njihovem ozemlju ne v stari in ne v novejši dobi nikjer ne najdemo apelativa kosez; pač pa srečamo •— šele in edinole v 16. stoletju — samo krajevno ime, in sicer zapisano pluralno genetivno obliko Kaseg ter lokativno Kasezih v Liki, gl. Akad. rječ. IV 879; nastanek tega imena iz starega osebnega imena Kosentzes pa se nam zdi težko verjeten in bi se morda zanj našla boljša razlaga21. Sporno se nam tudi zdi navajati toponim Kossessdorf na Sedmograškem (Ramovš, o. c., str. 315), ki naj bi tudi eventualno- temeljil na domnevanem *kasedz; tu gre po vsej verjetnosti za krajevno ime germanskega izvora, za zloženko iz Kossess + Dorf (vas); Kossess sam pa prihaja — kot je domneval že Stanonik — po asimilaciji ts > ss iz Kotsasz, Kothsasz, nastalo iz Kot (Kothe, Kathe, koča) + saet, set (stanovalec, naseljenec); obliko Kosassen najdemo zabeleženo v Allgemeine deutsche Real-Encyklopadie, Leipzig 1835, VI. zvezek,- sub v. Kothe z razlago: »Kothe, eigentlich Kathe, hiess uršpriing- 20 O težko vzdržljivi enačbi Hrvati = kosezi, oziroma narobe, zgodovinarja Ljud. Hauptmanna gl. M. Kos, GV, -193*2, str. 114, in B. Grafenauer, Hrvati u Karantaniji, str. 219 ss. Morda bi tudi za ta toponim privzeli kot osnovo turško vkaza = upravno okrožje«, kot to domneva za Kazagina in Kazagiče v Bosni B. Grafenauer, Hrvati u Karantaniji, str. 224? Omenjam le kot zanimivost, brez kakega sklepanja za naš problem, da je prav v istem stoletju na beneškem teritoriju bila znana beseda cas(s)asso (= birič) = turško 'asas, gl. Dizionario etimologico italiano, 1955, str. 789. Tudi ni nemogoče za ime »terra Kaseg« v Liki računati z vplivom s severa, kjer je imen z domnevano osnovo kosez mnogo več. 126 Še malo o imenih Koseze, Jcosez lieh im Niedeysächsischen ein Bauernhaus, welches weder Hof noch Laende-reien hat, und dessen Besitzer deshalb bloss zu Hand- und Fussdiensten verbunden ist. Daher heissen Kothsassen, Kosassen, Kossaeten (Koether, auch Hintersassen), zum Unterschiede der eigentlichen Bauern, diejenigen Dorfbewohner, welche eine blosse-Kothe, mithin weder Zugvieh noch Laenderei besitzen«. Grimm v svojem slovarju, V. zvezek, 1898, navaja oblike Kot(h)sasz = homo casatus, bewohner und inhaber einer Kote22. Kossessdorf bi torej nekako ustrezalo naši oznaki Koče,' Kočarji. Ker vemo, da je bila germanizacija tudi na ozemlju Sedmograške močna, nas toponim germanskega izvora ne more presenečati23; res pa je, kot pravi Ramovš, da je glasovni sestav tega imena precej podoben tistemu v naših Kosezah, zaradi česar Ramovš pripušča možnost, ,da je to ime v zvezi s Kosezami; tudi mi smo mnenja, da taka zveza obstaja, samo da bi zanjo imeli drugačno razlago24. Končno se R. Nahtigal25 in Ramovš, o. c., opirata pri izvajanju kosez < < *kosedz < *ka-, *kosegt na krajevno ime v Reziji (Beneška Slovenija), ki naj bi vsebovalo žensko obliko starega *koseg, namreč v posesivnem adjek-tivu, ki ga čitamo v zvezi Mna Kasiginej hörycoe«, citiranem po Baudouinu de Courtenay26. Pri čitanju tekstov, ki jih je B. de Courtenay objavil v 5. poglavju omenjene knjige in ki na približno 20 straneh, v odstavkih 297—368, zajemajo 22 Seveda je Kotsasz iste družine kot flamski Kossaat ter severnonemški Kotsäte, Kossäte < Koiséte, iž česar pO Grimmu srednjelatinski cotsatus, cotsetus; spada pa v to družino tudi že po.Stanoniku navedeni anglosaški eotsaeta, cotsetla »rusticus, villicus«, kar vse izpričuje visoko starost besede v saških oziroma germanskih deželah. J. L. Frisch, Teutseh-lateinisehes Wörterbuch, Berlin 1741, I str. 540, navaja: der Inwohner einer Kote, Kate heisst Köter, Kotsasse, Kotsäte, Kossäte, Kotse »casarius«, lat. barb. cotarius, cotmannus, cotsetus, cotarellus; Kotsassen, in der Magdeb. Verordn., p. 627 »die keine eigene Spann haben«. Tudi starofrancoski cote »cabane« ter izvedenke anglonormanski cotin »cabane« (14. stol.), anglonorm. cottage »tenure d'un jardin«,.staro- in srednjefrancoski cens cotage »cens payé pour un ténement en roture«, cotier »tenant d'un héritage roturier«, pa tudi srednje- in moderno francoski coterie z zanimivim semantičnim razvojem spadajo semkaj, gl. W. v. Wartburg, Französisches Etymologisches Wörterbuch, snopič št. 57 (1957), str. 345 sub v. kot. —žal gesla Kotsass nismo mogli preveriti tudi v najnovejšem nemškem pravnem slovarju Deutsches Rechtswörterbuch, Weimar, 1939 —, ker ustrezni snopič (črke Kot-) še ni izšel. 23 Prim. R. Kötzscke, Allgemeine Wirtschaftsgeschichte des Mittelalters, Jena, 1924, str. 405: »In Siebenbürgen fanden unter König Geise II die Flandrer (aus ,dem Moselland) Aufnahme und erhielten später Zuzug von Norden her durch Sachsen ...«. 24 O še nekaterih drugih krajevnih imenih, ki jih tudi — seveda s še večjo težavo in še manjšo verjetnostjo — spravljajo v zvezo s kosezi, gl; B. Grafenauer, Deset let ..., str. 188, 199; glede furlanskega krajevnega imena Cassacco, ki ga Kranzmayer tudi povezuje s kosezi, je vsekakor verjetnejši romanski izvor, kar navaja tudi Dizionario enciclopédico italiano, Roma, 1955, II. zvezek, str. 855; pripominjam, da kaže še večjo glasovno podobnost s kosezi toponim Caséz v provinci Trento, ki ga pa seveda tudi ne bomo povezovali z besedo kosez. 25 R. Nahtigal, O etimologiji izraza kosez, Slavistična revija VIII (1965), str. 136 s; Ramovš, o. c., str. 312. 26 Baudouin de Courtenay, Materialien zur südslavischen Dialektologie und Ethnographie, I., St. Petersburg, 1895; naš izraz »täna Kasiginej hörycoe« avtor pre- vaja »Auf dem Kasiginen Platze« (= eine örtlichkeit in Stolvizza). 127 Anton Grad področje St61bi.ce (Stolvizza) v Reziji, pa nas preseneča dejstvo, da je beseda Kasiginej edina beseda v vseh teh tekstih z zapisanim g; vse druge imajo namreč namesto prvotnega g pisavo s h, to se pravi dosledno je tudi v pisavi podan znani prehod g > y > h v- rezijanskem dialektu27, npr.: te drühe (druge), móhal (mogel), hlédal (gledat), ha ■ (ga), hré (gre), Buh (bog), hora (gora), H rat (Grad), mlha brátra (mojega brata), brih (breg), ohon (ogenj) itd.; še celó večina romanskih izposojenk kaže pisavo s h namesto z g: butœha (bottega, prodajalna), hràs (grasso, tolst), sohônt (secondo, segondo, drugi), hrdčja (grazia, milost), poleg ligreč, legrêce (allegrezza, slavje, slavnost); prav čudno je torej, da se edino v besedi Kasiglnej pojavi pisava z g, ko bi prav pri njej, ki naj bi bila stara slovenska beseda, pričakovali, če že ne izvršeno palatalizacijo g v z pred vokalom i (prim. zgoraj navedeni hrvatski toponim v Kasezih, poleg genetiva Kaseg\), pa vsaj izvršeni in tudi zabeleženi prehod g v h. Bi torej beseda utegnila biti novejšega romanskega izvora z vrednostjo g = z (it. dž), kot to srečamo npr. v besedi correginat (— korezinat, it. correggere, popraviti, poboljšati) in za kar je izgovor z z dokazan z obliko corrézina (torej pisava z 2!) v Mater., III, str. 51?28 Omenjam še, da za intervokalični g (v tujih besedah)srečamo pisavo gh pred e, npr. v nemighen (it. inimico, sev. it. nemigo, sovražnik), ibid. 350, 412. Tudi poudarjeni i v naši neznanki bi lahko — kot pogosto v rezijanščini — bil prišel iz é prim. še hrih (greh), sic (séci, kositi), tudi mideh (it. medico, zdravnik), a protonični a bi lahko temeljil tudi na e, prim.: nahá prásca (enega prasca), tacij! (teci!), vozam$! (vzemi!), zané (žene), talé (têle), taté (tête), jazik (jezik), raklà (rekla), medtem ko je videti, da protonični o ostane: kozá, kozé, wodó (vôdo), oe'à (oče), poslàt, pomátat (pometati) itd., poleg patbk (potok). Vsekakor sé nam zdi, da je pri besedi Kasiglnej težko in res nesporno misliti na izvor iz domnevane stare slovenske besede *koseg in da ni izključeno, da gre za besedo romanskega izvora; tu bi eventúalno mislili na sev. it. easina, it. caseína, pristava, planšarija, sirarna29 ali pa na furlanski deminutiv ciasine, hišica, koča (od ciase, hiša, it. casa); v zadnjem primeru bi šlo seveda za starejšo izposoditev, tj. pred znano furlansko palatalizacijo ka > > ča, ca; ostane pa še vedno problem intervokaličnega g, ki pa morda predstavlja fonem j ali (d)z v naši besedi? »Tàna Kasiglnej hôricoe« bi torej nekako pomenilo »na kraju, prostoru, trgu pri pristavi ali kjer so hišice, koče, kajšef Lahko pa Kasiginej predstavlja adjektivno izpeljanko iz nekega, žal, neznanega osebnega imena? Vse to bi bilo treba še temeljiteje preučiti (zlasti na kraju samem, če se je ime še obdržalo) z dodatnimi raziskavami in tudi podatki, ki jih žal nismo imeli na voljo. Zgoraj navedeni rezijanski tekst z besedo Kasiginej smatrajo dosedanji raziskovalci, lingvisti in zgodovinarji, za dokaz, da so tudi v tem delu sloven- n F. Ramovš, Kratka zgodovina slovenskega jezika, Ljubljana, 1936, str. 124. 2® Pisava gi podaja izgovor dj, prim. pregián (predjan),, Mat. III 66, je niitar Créai.(djal, dejal, dal), ib. 8. 29 M. Kos, Urbarji Slovenskega Primorja, II. zvezek str. 32, omenja planino Kašina pod Krnom (»cassina in monte Cren«), ' 128 Še malo o imenih Koseze, Jcosez skega ozemlja nekoč bivali kosezi, a drugi dokaz za to najdejo v današnjem krajevnem imenu Koseze pri Ilirski Bistrici na Notranjskem. V zvezi z nastankom tega krajevnega imena pa smo mnenja, da bi še morda lahko zatekli k drugačni razlagi, kot jo nudi domnevana starosloven-ska osnova ,],koseg, Sikaseg, in ki bi mogla veljati tudi za druge Koseze na Slovenskem. , V svoji razpravi «Postanek in razvoj Kranjske30 piše M. Kos, da »so proti koncu 13. stoletja (leta 1292) v notranjskih Kosezah31 imeli alodialno in, od goriških grofov, fevdno pošest gospodje, ki se po Lipi pri Jelšanah imenujejo de Lipia, Lint ali de Tylia; od leta 1323 sledimo v Kosezah naslednike gospodov iz Lipe, rodbini s pridevkom švabi (Swevi, Schwabn) (torej iz nemških krajev došlega rodu! op. A. G.), ki se po notranjskih Kosezah začno imenovati de Edlingen; goriški grofje jim dajejo v fevd »Vest Edlingen« (listini iz leta 1367 in leta 1398); okoli 1381—1383 se zaradi švabov v Kosezah (i. e. Edlingen) prepirajo Goriški z Devinci«. Po Kosovem mnenju v našem primeru ime Edling ne pomeni tujega kolonista — kot je to mislil zgodovinar Lj. Hauptmann32 •—, temveč le kraj, kjer so se ti švabi naselili oziroma si pridobili posest in prevzeli — glede na enak pravni položaj, kot so ga imeli staroslovenski kosezi — tudi njihovo nemško označbo Edling, tj. nemški prevod slovenskega imena Koseze; torej bi slovensko ime takrat že bilo moralo obstajati kot krajevno ime, dobljeno po tamkaj živečih kosezih; žal pa je omenjeno le z nemškim imenom, medtem ko za Lipo čitamo poleg nemškega (Lint,) in italijanskega (Tylia) tudi slovensko ime Lipia. Toda dejstvo je, da se ime Edling(en) javlja v starih zapisih mnogo pred našimi Kosezami — ki se, kot že omenjeno, prvič srečamo z njimi v zapisu iz leta 1424 »Edling, Kazezeh« —, da naletimo na toponim Edling (en) tudi v primerih, kjer slovenskega odgovarjajočega imena Koseze ali podobno sploh ni in ni bilo — obratno pa za slovensko Kazaze pri št. Petru vzhodno od Celovca čitamo, vsaj za novejšo dobo, vzporedno nemško ime Harbach (slov. Horpoh) —, skratka, treba je vendarle računati z možnostjo, da je ime Edling(en) starejše od naših Kosez in kosezov, kar pa seveda pomeni, da nastanek imena Koseze, kosez ne sega tako daleč v preteklost, kot se to splošno misli. Za nemški izraz Edling čitamo v Grimmovem slovarju, III. zvezek (1862) na str. 28: Edeling m vir nobili genere natus, ahd. edilinc, mhd. edelinc, ags. ädeling; Grimm dostavlja: ein schönes, durch edelmann verdrängtes wort. — Po Grimmu je torej Edling neke vrste plemič33. 30 M. Kos, Postanek in razvoj Kranjske, Glasnik Muzejskega društva za Slovenijo, X, 1929, str. 39 ss. 31 Zapis je seveda nemški: Zadlinge, Zedlinge - zu Edling(en)'. 32 v Vierteljahrschrift für Sozial- und Wirtschaftsgeschichte, 1928, str. 273. 33 Iz burgundščine je adhalings, adalings prišel v provansalščino ter ga najdemo v epu Girarz de Rossilho (13. stoletje), toda vedno le v zvezi »vassal i a(d)alenc«: e lui ac bo vassal i adelenc, 4382 (= bil je dober vazal); ib. 2183, 2519; burgundski adalings je po Gamillschegu, o. c., II 184, odgovarjal romanskemu vassallus = (adeliger) Lehensmann. 129 Anton Grad H. Fischer, Schwäbisches Wörterbuch, Tübingen, 1911, II. zvezek, str. 538, pa navaja: Edeling -m; Edler, ter citira med drugim primera iz avgsburških kronik: 50 Pfärd eitel guet auserlesen, stark Volk, darunter etwa vil Edling, 2,91 ter: All Lantzhern und all Ritter und Edlinck, 1, 61, 2, 23. Najnovejši Deutsches. Rechlswörterbuch, Weimar, 1939, II. zvezek, str, 1195: Edfejling, ahd. adaling, ediling; glede pomena pa navaja za langobard-sko, visokonemško, holandsko, spodnjenemško »freier Bauer« (kmet svobod-nik), s primeri kot: herimanniM aut nobiles (?) qui vocantur edelingenses, 1232, Istrien, Böhmer-Ficker, Nr. 320; ez sint dreier hande freien ... die dritten haizent edelinge und sint gepouweren (= Bauern!), 1300, Schawsp. Var./WBS 80 (1875) 313; der ... sol ... sweren, daz er seiner herschefte getriwe sei, ib. 355, n. 5 (1305 Steier. L: Arch. hschr.); berittene freie Grundbesitzer, Werunsky, Oest. RG 499n; die freysüssen und edlinger, die jre aigne güeter und hueben haben, aber doch nit geadlt personen sein, 1535 Krain, LGO 54. Ker nisem zgodovinar, se lahko motim oziroma ne poznam še drugih virov z drugačno razlago za Edlinge, toda vse kaže, zlasti pa najnovejši zapisi v zgoraj omenjenem pravnem slovarju, da je to bil sloj svobodnikov, ki bi ga lahko uvrstili nekako med plemiče na eni ter nesvobodnike (podanike) na drugi strani. Grimmova razlaga — najstarejša — za Edlinge, ki jih šteje k plemičem, se torej ne sklada z razlagami novejšega datuma — v katerih velja Edling le za svobodnika —, kar je gotovo treba pripisovati dejstvu, da Grimm še ni poznal toliko zgodovinskih virov oziroma zapisov kot poznejši sestav-Ijalci raznih slovarjev in je tako povezal besedo Edling z besedo Adel, (plemstvo) ter izenačil besedo Edling z Edelmann. Tudi L. Pintar je istovetil Edlinga s plemičem in enačba Edling(en) = Koseze ga je privedla do tega, da je izraz "kosez skušal izvajati iz plemiško zvenečega imena knez, a tudi poznejši raziskovalci so, kot smo videli, za koseze iskali »visok« izvor (Kosin-gas = glavar, Kosentzes = ime vodje Hrvatov). Da pa Edlingi niso bili plemiči, dokazuje tudi dejstvo, da so opravljali dela, ki bi za plemiča bila nemogoča; tako so opravljali službo stražarjev na gradovih, bili sodni sluge, poštni seli, postavljali so vislice, vodili zločince na morišče ipd.35 34 Za langobardski harimann pravi Gamillscheg, o. c., II 68: harimann bezeichnet zunächst den Kriegsdienstleistenden, dann, den Freien überhaupt, und dringt in das Mittellateinische. ■ 35 O vsem tem in tudi o teritorialni, razširjenosti naselbin Edlingov na slovenskem ozemlju gl. B. Grafenauer, Zgodovina slovenskega naroda, II 106 ss, ter J. Mal, o. c., str. 151, v obširnem poglavju »Das Wesen der Edlinger und deren Dienste«. Mal navaja tudi zanimivost, da še danes prebivalce vasi Orehovica na Dolenjskem sosedje imenujejo »biričevnike« (= pomočnike biričev), ker so nekoč, tj. pred kakimi 500 leti, Edlingi te naselbine morali loviti hudodelce- in postavljati vislice. Seveda pa so v tej dobi in gotovo tudi že prej med Edlinge nemškega roda bili pomešani in so opravljali iste posle tudi svobodniki slovenskega rodu, kot to dokazuje že navedeni žužemberški zapis iz začetka 16. stoletja, a na Koroškem J. Unrest, ki konec 15. stoletja v svoji kroniki (kot prvi) imenuje ustoličujočega slovenskega kmeta z imenom Edlinger (»ein Pawer Geschlacht, genant Edlinger«). 130 Še malo o imenih Koseze, Jcosez Bili bi torej glede notranjskih Edlingov rajši mnenja, da bi morda le kazalo dati prav Lj. Hauptmannu; goriški grofje oziroma plemiči-gospodarji gradu na Premu — ki so bili gotovo tujega, nemškega rodu — so na svojo posest naseljevali tuje nemške koloniste, tudi svobodnike, med njimi tudi Švabe, ki jih (oziroma njihove potomce!) M. Kos omenja kot jabolko spora med goriškimi grofi in Devinci konec 14. stoletja; ti tuji naseljenci pa so tudi po našem mnenju ime Edling prinesli že s seboj, ne pa si ga šele pridobili na kraju svojih posestev na Notranjskem kot prevod oznake pravno enakih že tamkaj naseljenih »kosečov«. Ime Edling oziroma njihova naselišča Edling(en) srečamo že od 12. stoletja dalje na celotnem ozemlju, nekoč in sedaj naseljenem s Slovenci. Naj za štajersko po J. Zahn, Urkundenbuch des Herzógthuw.s Steiermark, Graz, 1872, II. zvezek, str. 259, navedem zapis iz c. 1150: recipiens tale predium quale nos (sc. samostan Admont) apud Edelingin habuimus (gre za Edling pri Trofaiach; isti Edling se omenja v listini iz leta 1164); v neki listini iz c. 1170 nastopa kot priča tudi Engilbertus filius Ratoldi de Edelingen (= Edling pri Gröbmingu); na str. 248 v listini iz c. Í145 čitamo med pričami imena Fridericus de Ettelingin, Liutwin de Etelingen, Heinrich de Ettelingin (ne ve se točno, za kateri Edlingen gre); str. 116 v listini iz leta 1205 najdemo med pričami nekega Rodegerus de Edelingen; str. 495 čitamo v listini iz leta 1240: item in der Dobre quicquid colunt Edlinge. Za Edling, Kazezeh iz leta 1424 gl. zgoraj. Za Koroško navajamo po H. Jaksch, Die Kärntner Geschichtsquellen, Klagenfurt, 1904, III. zvezek, Nr. 1082, za leto 1164: die Güter im Lavanttal in Edling-, Nr. 1163 iz leta 1173—1180 čitamo: ein Gut in ... Edling (pri ,St. Andrä); Nr. 777 v listini iz Jeta 1144 se omenja Edlingen (pri Hartmanns- • dorf pri Priesach); Nr. 887 za leto 1216 čitamo ime Edelinge de Amelricus (gre za Edling jvzh. Spittala); VI. zvezek, str. 192, za leto 1294 čitamo: eine Hube in Edlingen (gre za dar samostanu Vetrinje); str; 76 beremo v listini iz leta 1288 ime Edilhingen (gre za Edling am Zammelsberg, južno od Strass-burga na Koroškem). Na ozemlju bivše Kranjske pravi Krajevni leksikon Dravske banovine (1937), da se kraj Koseze pri Moravčah prvič omenja leta 1238; sicer ne pove s kakšnim imenom, toda gotovo kot Edling (en); za Koseze pri Vodicah nad Ljubljano imamo zapis »aput Edelingen et in Woditz« iz leta 1265, gl. Ramovš, o. c., str. 303 in Pintar, o. c., str. 600; za Koseze pri Ljubljani najdemo zapis Edling bey sannd Margareten iz leta 1453, gl. Kelemina, o. c., str. 464; za Koseze na Notranjskem gl. zgoraj. Tako ne more preveč presenečati, če si upamo domnevati, da je tudi na Notranjskem leta 1292 zabeleženo nemško krajevno ime Edelingen prvotno krajevno ime, imenovano tako po Edlingih, tj. tujih, nemških kolonistih, med katerimi so — kot smo videli — izrecno omenjeni švabi. Slovenska oznaka kraja, tj. Koseze, bi torej bila poznejša, a zelo verjetno se nam zdi, da je nastanek imena Koseze bil prav pogojen s prihodom nemških kolonistov, oziroma drugače povedano, zdi se nam možno, da je ime Koseze germanskega, nemškega izvora, 131 Anton Grad Za katero nemško besedo bi lahko domnevali, da je bila osnova za naše Koseze? V že omenjenih Kosovih urbarjih večkrat srečamo besedo gesäss, gesaezze, gesesse, gesezz, gesezze, kar Kos sloveni s »selišče«; tako na str. 111: (c. 1370) Item 1 gesazze, das Mathyas Glopit inn hat; Item 1 gesezz, das der meins herren von Görz schultazz inne hat; na str. 113: Item Hainczl Fricleins sun von Zelkan hat ze lehen von dem haus ze Reyffenberg ein gesezze, das er selber pawet; na str. 115: Item Friczel von Zelkon 1 gesezz dasselbs, da er selber auf siezet mit eilen den Rechten dy darzue gehören; na str. 117: Item Pluemel ze Görz hat von dem haus Reiffenberg ze lehen ein gesesse in dem dorff ze Görcz. V urbarju za Vipavo iz leta 1499 čitamo, naj Primus Szerottnikh daje od svojega selišča (tj. gesäss) 80 šilingov. Besedo gesaez srečamo tudi na Koroškem, kot kaže npr. zapis iz leta 1317: Graf Heinrich von Görz beurkundet, dass Otte von Reuntal sein hous und gesaz da ze Wynchlern (= Winkleren im Mölltal) ... aufgegeben hat, gl. H. Jaksch, Die Kärntner Geschichtsquellen, Klagenfurt, 1904, VIII. zvezek, str. 112, Nr. 368. Grimmov slovar, IV. zvezek (1897), kol. 3806, navaja za besedo Gesäsz n za starejšo dobo oblike in pomene: stvn. gisäzi, gesäze »sedes, domicilium, tabernaculum, territorium«, srvn. gesaeze, starejše nvn. gesäsze, gešasz; za pomen »wohnsitz, wohnung, gebäude, niederlassung, lagerung« navaja primere kot: stvn. ih zimberön din gesäzze »aedificabo sedem tuam«, St. Galler handsehr. 88,5; die fruoasten arbeite sluog er in Chamis kes&zzen »primitias laboris eorum percussit in tabernaculis, Cham.'77, 51; srvn. ein gesass oder vesten newlich geseezet, Fontes rer. austr., II 20 (iz leta 1459); za pomen »landgut, bauerngut«: All Kössin git (giebt) ain pfund haller uszer irem hof und gesaesz ze Wessingen, monum. zollerana 1, Nr. 426 (za leto 1393); v listini iz leta 1421 hus, hofraiti und gesaesz. Fischerjev švabski slovar daje za našo besedo pomen »wohnsitz«: Uz dem Huse bi unserem Gesaezze, Ulm 1308/Ub. 1, 296; An b-s Hus und Schur, daz man nempt der Nünnelin Gesaehsse. Najnovejši Deutches Rechtswörterbuch, 'IV. zvezek, str. 415, navaja: Gesässe n gesaezze, gesaizze, gesass, gesesse, geseste, geseze, ghesette s pomenom »Besitztum, Wohnsitz«: possessionis kasiezzi Ahd. GL, II 345; vicos ciuitatis kesazze', ib., I 728; unum gesete aduersus E., a. 1190, Erhard II 218; sine gesesse gelegen by dem wasser, a. 1279, Basel ÜB II. Nr. 287; daz gesaizze ze P., a. 1322, Lexer I 895; die da haben ir frey gesez oder hoff, die schollen 1 M von ainem lehen geben, a. 1471, Fr. Austr. II 20, 83, itd. Na podlagi gornjih -primerov se nam ne zdi nemogoče, da bi apelativ Gesäss, ki je imel tudi pomen selišče, bivališče, stanovanjska in verjetno tudi gospodarska zgradba (koča), za notranjske Koseze, za katere vemo, da so v njih živeli (tudi) Švabi (gl. zgoraj švabske primere!) prav lahko prišel v poštev kot osnova za naše Koseze: Slovenci, naseljeni na tamkajšnjem področju, so začeli iz več poslopij sestoječo naselbino tujih kolonistov imenovati prav po imenu Gesäss, ki so jo v pluralni obliki gotovo pogosto slišali od nemških priseljencev. Edninska oblika Gesäss pa verjetno tiči v krajevnem 132 Še malo o imenih Koseze, Jcosez imenu Kases, ki ga je ugotovil Mal, o. c., str. 155, v okolici kraja Rottenmann na štajerskem, ob vznožju mogočnega gradu Strechau. Slovenci so na pol zvenečo nemško medij o g točneje substituirali s svojim k, kot bi to bili mogli storiti s svojim zvenečim g, prim. isti pojav v stvn. krajevnem imenu Sant Gangolf ter slovenski šent Kandolf na Koroškem, pa tudi apelative kot kävtre (vrnila ,lesena vrata) iz srn. gater, kdvklar iz Gaukler, krompir iz bav.-avstr. gruntpirn, itd.; obratno pa za slov. g nemški k v Krka, nemško Gurk,. itd. Prednaglasni e bi bil po disimilaciji prešel v o (ali» ponekod tudi a), izvršila pa se je tudi konzonantska zamenjava z — s v s — z, verjetno pod vplivom slovenskega začetnega nezvenečega k: Kezese > Keseze > Koseze. Beseda je bila verjetno privzeta razmeroma zgodaj, tj. še preden se je v (pogovorni) švabščini oziroma nemščini izvršila redukcija in onemitev začetnega, nenaglašenega e v predponi ge-, pojav, ki ga stavljajo nekako v 14. stoletje. Ali moremo tudi za ostale slovenske Koseze oziroma Kasaze, Kdzaze domnevati isti nastanek? Že Ramovš, o. c., je lepo dokazal, da so vse razlike v našem imenu produkt slovenskega dialektičnega razvoja in ne različnega substituiranja; torej bi mogli tudi pri naši domnevni nemški osnovi suponirati najprej začetno substitucijo v obliki Keseze, Kezese, v kateri bi v nadaljnjem razvoju na Koroškem in Štajerskem začetni nenaglašeni e prešel v a — prim. teta > tata, tele > tale, Ramovš, o. c., — ter še kasneje prenos poudarka na prvi zlog, o čemer tudi govori Ramovš, navajajoč za kranjsko govorico primere motika, navada, poroka proti štajerskemu in koroškemu mötika, navada, rož. pdroka. Predvsem za notranjske Koseze, za katere zgodovinski viri govore o švabskih priseljencih, bi poleg švabske in splošno nemške besede Gesäss kot osnovo za Koseze lahko pritegnili tudi sestavljenko Kottengesässe (Kötte = = koča; torej selišče koč), ki smo jo našli v že omenjenem švabskem slovarju z razlago: »in Lagerbüchern ist mehrmals von »Kottengesässen«, d. h. Wohnungen der Hintersassen36 ... die Rede«;, beseda bi bila seveda tudi doživela močnejšo redukcijo nenaglašenih zlogov in konzonantsko asimilacijo: Kottengesässe > Kotg(e)sässe > Kotsässe > Kosese, Koseze; zlasti s svojim prvim sestavnim delom Kotte(n) bi morda predstavljala še bolj verjetno osnovo za naše Koseze37. Besedo Kottengesässe smo našli le v švabskem slovarju, a je gotovo bila znana tudi drugim nemškim dialektom, npr. bavar-ščini, in bi v tem primeru naši predniki nekatere Koseze bili lahko privzeli tudi od bavarskih kolonistov, dasi je iz starih listin razvidno, da je moralo 36 Hintersasse, po Grimmu, IV/2, kol. 1514, »einer der hinter einem herrn in dessen schütze angesessen ist; po Deutsch. Rechtsw., V 1053, »jeder der in best. Raum im Schutz jemandes sitzt« in »der zwar Schutz geniesst, aber nicht Untertan o. Bürger ist«. 37 Z osnovo (Kotten)gesässe »selišče (koč)« bi bil, se nam zdi, najlaže razložljiv tudi toponim Kajžice za Kazaze, nemško Edling, pri Bilčovsu na Koroškem; morda ne gre le za posmehljivo ime, kot mislita Lessiak in Ramovš, temveč za dejansko zdrobljene posesti in majhna selišča (koče, kajže) obubožanih Edlingov; že Klebel pa je tudi domneval, da je v goriškem urbarju (c. 1300) omenjeni kraj Edling identičen s Chotschaw iz leta 1380, današnji Kötschach v zgornji dolini reke Žile. 133 Anton Grad biti tudi švabskih priseljencev precej na Koroškem in Štajerskem, prim. zapise kot v listini iz let 1074—1087, kjer čitamo med pričami ime Merquart Sueuus (gre za darila salzburške nadškofije samostanu Admont), gl. Zahn, o. c., II. zvezek; c. 1185 nastopa med pričami v neki listini Rudolf Sueuus-, istega leta figurira kot priča Gotfridus Sueuus v neki zadevi samostana Admont; istega leta srečamo med pričami osebo Chonradus Sueuus v zemljiški zadevi za Ribnico pri Beljaku; leta 1205 se kot priča čita ime Fridericus de Suabesdorf v zemljiški zadevi za Leibnitz na štajerskem. Koseze bi se torej po svojem nastanku pridružile skupini iz nemščine sprejetih krajevnih imen tipa Meierhof > Marof (krajev z imenom Marof je v Sloveniji precej: pri Cerknici, Trebelnem, Vidmu, Vojniku, Sevnici, šmarjeti, Vuzenici, Jurkloštru, Pilštajnu), Frlthof (Friedhof) > Britof (npr. pri Kranju, Senožečah, itd.), Reute > Rovte, Rute, itd., da ne govorimo o gradovih nemških gospodarjev, ki so ohranili svoja nemška imena, seveda več ali manj spremenjena, vse do danes v slovenščini: žužemberk, čušperk, Jetrbenk, Bogenšperk, Dornberk, Rihemberk (danes Branik), Pilštajn, itd. Naj še omenimo, da imamo na področju bivšega premskega gospodstva na Notranjskem še danes naselje Britof, torej še en dokaz, da tuji, nemški kolonisti niso bili tako redki v teh krajih in da se je pri Slovencih ohranilo nemško ime kraja oziroma naselbine. V bližini štajerskih Kases, pri Preboldu, pa se srečamo s toponimom žvajga, ki more temeljiti le na srvn. sweige »Alpenhütte, Viehhof, Sennerei«?8. Glede Kosez pri Moravčah, za katere nismo gotovi, ali se' tudi zanje v starih zapisih bere vzporedno nemško ime F.dling(en), pa poroča Mal, o. c., str. 156, o številnih edlingih, bivajočih v neposredni bližini tega kraja pred stoletji. Za oblike Kossas, Kos(s)es in KM,ses iz prve polovice 16. stoletja, ki jih srečamo edinole v že omenjenih žužemberških zapisih, pa bi podobno naši razlagi nastanka toponima Koseze domnevali, da je treba izvor teh besed iskati v že zgoraj navedeni nemški besedi Kossass, Kötsass; v nemščini, ki so jo govorili nemški priseljenci na ozemlju, nekoč in sedaj poseljenem s Slovenci, se je beseda Kossass gotovo uporabljala ter se zrcali- tudi v žužemberških zapisih, v katerih je — tu soglašamo s Kelemino in Bezlajem — rabljena kot priimek (v tej dobi se je raba priimkov že uveljavila), za kar govori tudi zapis »Paulle Khäses edlinger«35, če pa je ta naša razlaga pravilna, imajo žužemberške oblike poudarek na prvem zlogu — kot nemška osnova Kossas — ter bi zato težko predstavljale, kot je mislil Ramovš, prve (in edine) zabeležene primere besede kosez (ki ima poudarek na drugem zlogu)40. 38 Prim. zapis iz leta 1267: unam svega sitam in ..., Jaksch, o. c., IV. zvezek, št. 2927; gre za Schwaige pri kraju Feldkirchen na Koroškem. '39 Tudi prvotni apelativ Edling se je začel uporabljati kot priimek (.Edlinger), kot dokazuje — poleg žužemberških zapisov — ime znane goriške plemiške družine v 14./15. stoletju, potomcev Svabov iz Edlinga (Koseze) na Notranjskem, gl. M. Kos, Postanek in razvoj Kranjske, str. 39 ss, ter meščanski priimek Edlinger v Celovcu. Marof, Frîthof (Friedhof) > > Britof, fréquents, eux aussi, sur le territoire Slovène. Le premier exemple du toponyme Koseze remonte à l'an 1424. Quant à kosez, il n'apparaît jamais dans les anciens documents, car il n'est pas du tout sûr que les noms de personnes Kasses, Kosses, Khâses, constatés dans un document de la première moitié du 16<= siècle, représentent le mot kosez, l'auteur préfère y voir les reflets du terme allemand Kotsasz, Kossass, Kosass »habitant d'Une case, d'une hutte« ou de Gesàsse »propriétaire foncier«. Kosez serait un mot oblike Kossas, Kosfsjes in Khases bile poudarjene na drugem zlogu, a nemška osnova Gesasse bi bila pri Slovencih doživela iste spremembe kot Gesass (selišče) — saj gre v bistvu za isto besedo. Ker pa se beseda Kossas itd. v zapisih pojavi razmeroma pozno, precej kasneje kot toponim Koseze, je po našem mnenju težko misliti na možnost, da bi se pluralna oblika besede Kossas, Kosfsjses, Khases < < Gesasse (= lastnik selišča) začela rabiti tudi za oznako njihovih selišč, kar bi sicer bilo prav mogoče. Verjetno bi tudi za priimek Kaseznlk v Logarski dolini, ki ga omenja Ramovš, Kratka zgodovina .. ¡, str. 10, vzeli kot izhodišče nemško osnovo Kassess (ali Gesasse), če ni izveden iz že sloveniziranega štajerskega toponima Kasaze ali kranjskega Koseze. 135 Anton Grad reconstruit: comme le toponyme Slovène Koseze avait, autrefois, aussi le nom parallèle allemand Edling(en), interprété comme »habitation des gens nobles« (Edling étant compris, erronnément, au sens de Edelmann »home d'origine noble«), Pintar a abouti', grâce à l'équation Edling(en) = Koseze à celle de Edling = kosez, créant ainsi le mot en question. 136 B. V o du§ ek i - THE REPETITION OP PHONEMIC CHARACTERISTICS IN RADICAL MORPHEMES IN SETS OP SYNONYMS PROM INDO-EUROPEAN LANGUAGES* 8.23 SPADE 11 Ir. laige 1 Grk. áme 2 Grk. skapheîon (?) 3 NG patóphtuaro (Grk. patéô) 4 (Grk. ptuárion, ptúon) 5 Lat. pâla W. pal (fr. Lat.) yam- P 502 (s)kep-2 (s)kap- etc. P 930-2 (Bois. 871 f.) pent- B. P 808-9 peu-1 B (8.24), P 827, Bois. 824 pag'- pák'-P 787-8, B, WH 2.236 :kwel- P 545 for W. 6 It. vanga (Late Lat. vanga fr. Gmc., cf. ON vangsni, OHG waganso) 7 Fr. bêche (VLat. *bissus fr. Lat. bis) etc. 8 Sp. laya (Basque: laya) 9 Rum. lopata lO Ir. rame wogwh-nis- wogvvh-nes- REW 9137, WH 2.730, P 1179-80 dwo(u)- dwei-FEW 1.381, 382 f., Gam. 93, REW 1132, P 228-30 laya Cor. 3.57, REW 4957 lcp- B, P 679, Vas. 2.58 er3-l P 338, B . 12 ON reka IS Dan. spade 14 OHG scûvala (?) 15 MHG grabeschït 16 17 Lith. kasiklis 18 ChSl. rylo 19 SCr. ašov (Hung, aso, as) lâgh-P 652, Bois. 561, WH 1.757 f. (: Frisk 2.92) reg'-l P 854-6, B (8.24, 8,27), John. 717-8, FaT 870 sp(h)ë[i]-2 sp(h)6i-P (982)980, FaT 1110, John. 887, KM 721 skëu-5 P (954)955, B, KM 639, 646, FaT 1020 f., 1045 ghrebh-2 P 455-6, B, KM 641 sfek-2 P, (895)919-21 kes-P 585, Fr. 226 f. ereu-2 P (338)868, Vas. 2.555, 558, 449, Brück. 471, 470, Meh. 431, 429 as-B * See First Part, Lingüistica VI, 1964; Second and Third Part, Lingüistica VII, 1 and 2, 1965; Fourth and Fifth Part, Lingüistica VIII, 1 and 2, 1966—1968; Sixth Part, Lingüistica IX, 1, 1969. 137 B. Voduselc 20 Russ. zastup 21 Skt. khanitra- steb(h)-stëb(h)-etc. ste-m-p-P 1011-3, B, Vas. 3.35 cons. kx vowel ex Mh. 1.301, P 634 8.24 SHOVEL 1 Grk. pttion 2 Lat. pâla Br. pal (fr. Lat.) 3 Rum. lopata 4 Ir. sluasat 5 W. rhaw 6 W. siefl (NE shovel) 7 Goth, -skauro 8 ON reka 9 Skt. khanitra- 8.25 HOE 1 Grk. skapânë Lith. kaplys (s)kep-2 (s)kfip- etc. P 930-2, Bois. 870 f. Ls]kSp-2 P 930-2, Fr. 217 f. 2 Grk. skalis 3 Grk. sminfle 4 Grk. mâkella 5 Grk. dikella peu-1 P 827, Bois. 824 pflg'- pfik'-P 787-8, B (8.23), WH 2.236 :kwel- P 545 for Br. lep- B, P 679, Vas. 2. 58 cons, s vowel ux B er3-l P 338, B (8.23) skeu-5 B, P (954)955 sek-2 P (895)954, F 566 f. reg'-l P 854-6, B (8.24, 8.27), Joh. 717-8, FaT 870 cons, k* vowel ex Mh. 1.301, P 634 7 NG tsâpa (It. zappa fr. Illyr. zapp- fr. a calling cry) 8 Lat. ligô 9 Lat. sarculum 10 Lat. bidëns 11 Fr. houe (Frank. *hauwa) 12 Sp. azada (VLat. *asciata, Lat. ascia) 13 NIr. grafan (Gmc., cf. ON OE grafan) 14 Br. pigell (Fr. pic) 15 Dan. hakke (MLG hacke, hacken vb.) (s)kel-l P 923, Bois. 868 f. (s)mëi-2 P 968, Bois. 886 cons. kx vowel ex B, Bois. 602, Frisk 2.163 f. dwô(u)-dwei-P 228-9 lcel-3 P 545 :(s)kel-l Bois. 189 (Frisk 1.392) zapp- B, REW 9599 [s]Ieig-P 961, WH 1.800 ser-5 P 911, WH 2.470 f. ed- P 287-9 kâu- REW 4084, Gam. 519, P 535, KM 293 ag-wesï ak-sï Cor. 1.345, REW 697, F 9, WH 1.71 f. glirebh-2 (B), P 455-6 pikk- B, Gam. 692, EEW 6495 (cf. sub 8.26-3 OE pïc) keg- kek- etc. FaT 372, 390, B, P 537-8, KM 279 138 The repetition- of phonemic characteristics 16 OE mattuc (VLat. *matteüca, *mattea, Lat. mateóla) mat-2 B, REW 5426, 5425, P 700, WH 2.49 ' 8.26 FORK (s. Lingüistica VII, 2, 1965, p. 158) 8.27 RAKE (s: Lingüistica VII, 2, 1965, p. 155) 8.28 HARROW (s. Lingüistica VII, 2, 1965, p. 152) 8.32 MOW, REAP 1 Grk. theridzo 2 Grk. amáo 3 It. segare (Lat. secare) 4 It. falciare (Lat. falx, -cem) 5 Sp. guadañar (Goth. *waitho or Frank., Langob. *wai-danjan) 6 Rum. secera (VLat. *sicilis, Lat. sicllis) 7 Rum. cosi (Slav., cf. SCr. kositi, Russ. kosit') gwher- P 493-4, Bois. 341, Frisk 1.665 f. me-2 m-e-t P 703, Bois. 53, Frisk 1.88 sek-2 REW 7764, P 895 :se[i]k- s3k- sefe WH 2.504 f. dlielg-REW 3175, BA 2.1586, P 247, (WH 1.449 f.) wei-3 Cor. 2.804, REW 9483, P 1123-4, FaT 1361 f., Joh. Ill f., KM 848 cons, s vowel i" P 896, REW 7900, WH 2.533, 504 k'es- B, Bern. 1.581, Vas. 1.639 f., P 586 Ir. bongaim NIr. spealaim 10 W. pladuro (Lat. pala) 11 Goth, sneithan 12 ON sla 13 OE ripan 14 Lith. piauti 15 ChSl. ž§ti 16 Skt. lu- 17 Skt. d§- bheg- bhe-n-g-P 114, LP 347, 340 sons, s, p vowel ex B pag'- pak'-B (8.23), P 787-8, WH 2.236 :kwel-P 545 sneit-P 974, I'1 440, Joh. 913 f., KM 670 (:FaT 1096 f.) slak-P 959, Joh. 920 ff., P 436, KM 652. FaT 1048 f. rei-1 P 857-8, KM 592, FaT 902, 906 peu-P 827, Fr. 584 gwhen-2 P 491-3, Vas. 1.411,428, Brück. 662 f., Meh. 595 lcu-2 P 681, B da- dai- dau-P 175, Mh. 2.31 8.331 SICKLE (s. Lingüistica VII, 2, 1965, p. 158) 8.332 SCYTHE* 1 Grk. drépanon 2 Grk. hárpe der-4 P 206-11 ser-5 P 911-2 * This set of synonyms was erroneously omitted in the list of sets composing ■ the present concept-group (Lingüistica IX/1, 1969, p. 37). Consequently, the number of the included sets amounts to 55. 139 B. Voduselc 3 NG kosa k'es- (Slav., cf. SCr. kosa, B, Bern. 1.581, Russ; kosa) Vas. 1.639 f., P 586 - 4 Lat. faix Sp. (Goth. *waitho or Prank., Langob. *wai-danjan) dhelg-P 247, WH 1.449 f. wei-3 Cor. 2.804, REW 9483, P 1123-4, FaT 1361 f., Joh. 111 f., KM 848 Sp. dalle cons, d* (Late Lat. daculum) vowel ex Cor. 2.105 f., REW 2458, P 247 cons, s, p vowel ex B pag'- pak'-B (8.23), P 787-8, WH 2.236 :kwel-P 545 Ieu-2 P 681, •Joh. 739 ff., FaT 650 siik-2 P 895, FaT 963, KM 704 f., . Joh. 777 f; [s]kep-(s)kap- etc. P 930-2, B da- dai- dau-P 175, Mh. 2.31 7 Ir. spel 8 W. pladur 9 ON le 10 OE side' OHG segansa 11 Lett, izkapts 12 Skt. dâtra- 1.22 CUT (vb.) 1 Grk. temno tem-1 P 1062, Bois. 954, Vas. .3.111, Brück. 60 2 Grk. kópto 3 Lat. secare 4 Lat. caedere 5 It. tagliare (Lat. tâliare, talea) Fr. couper (Vlat. colpus fr. Grk kdlaphos) Fr. ' trancher (Lat. truncâre, truncus) Sp. cortar (Lat. curtâre, curtus) ON skera 9 Ir. snaidim 10 Ir. scothaim 11 NIr. gearraim , 12 W. torri 13 Goth, maitan 14 Goth, sneithan [s]kep-(s)kap- etc. P 930-1, Bois. 492 f., Frisk 1.915 f. sek-2 P 895, :se[i]k- s3k- seli WH 2.504 f. [s]k[h]ai- . P 917, WH 1.129 tal- IiEW 8542,' EA 5.3698 f. P 1055, WH 2.643 kel-3 HEW 2034, WH 1.304, P 545 trenlc-1 REW 8953, Gam. 857, P 1093, x WH 2.710 f. [s]ker-4 Cor. 1.919 f., ■ REW 2418, P 938-9 (s)ker-4 P 938-9, Joh. 835 ff., FaT 1010 f. (s)nadh-P 972-3 cons, s, k* vowel X B (8.57) g'her-6 P 443, B (12.59) . ter-3 P 1071-3, B (9.26) ' :P (1071)1077 ■ [s]mei-P (968)697, F 341 f., FaT 709 . sueit-P 974, F 440, Joh. 913 f., KM 670 (:FaT 1096 f.) 140 B, Vodusek IE ON telgja 16 ON ho*ggva 17 OE ceorfan 18 ME cutte 19 Lith. piauti 20 Lith. re*zti 21 Lett, griezt 22 Pol. r^bac 23 Skt. lu-, 24 Skt. éas- 25 Av. thwar3s- 26 Av. brï- del-3 P 194-6, Joh. 492 f., FaT 1252 käu-P 535, Joh. 177, FaT 426 gerebh-P 392 cons, g* .vowel X FaT 598, Hh. 43, B pëu-P 827, Fr. 584 wreg-P 1181-2, Fr. 725 f., Vas. 2.484, 485, 505, Brück. 476 f., Mch. 434 . cons. gx, r vowel i* B, Fr. 169 remb-P 864-5, Brück. 455, Vas. 2.541 f. leu-2 P 681, B k'es-P 586, WH 1.179 twerk'-P 1102, Mh. 1.539 bher-3 P (133)166 4 Ir. scian 5 ON knifr 9.23 KNIFE 1 Grk. mächaira 2 Lat. culter 3 Rum. cut*it (VLat. *cötitus, Lat. cös, cötis)- mag'h-P 697, Bois. 616, Frisk 2.186, WH 2.3,4 [s]kel-l . P 923-4, WH 1.304 ak'-2 B, P (18)541-2 OHG mezzisahs 7 Lith. peilis" Lett, nazis (Russ. noz) 9 Skt. kr*ti- 10 Skt. sastra- 11 Skt. churikä sëk-2 P (895)919-20, LP 15 gen- P 370, Joh. 332-4, FäT 550, 548 mad-P 694-5, KM 475, FaT 944 cons, p vowel ix Fr. 563 f., ■ Vas. 2.356, B neg'h-B, P 760, Vas. 2.225, Brück. 367, 364, Mch. 329, 326 f. [s]ker-4 P 938-41, Mh. 1.258, 260 257 k'e.s-P 586 ltes-P 585-6, Mh. 1.408 f., 292 9.24 SCISSORS, SHEARS 1 Grk. psalis. Lat. forfices (pl.) 4 It. cesoie (VLat. *caesorium, v Lat. caedere) 5 Sp. tijeras (pl.) (Lat. [forfices] tön-sörias, tondëre) Ir. demess (s)p[h]el-l P 985, Bois. 1073, 890 gwher-P 493-4, B, WH 1.526 f., 532 f. kapP 527, WH 1.159 f. [s]k[h]ai-REW 1475, P 917, ' WH" 1.129 tem-1 Cor. 4.631 f., REW 8784, P 1062-3, WH 2.689 f. dwö(u)- dwei-P 228-9, B, LP 188 141 The repetition- of phonemic characteristics 8 W. gwellaif 9 10 ON skaeri (pi.) Skt. kr*pân*ï- 11 ON so*x (pi.) 12 Lith. zirkle*s 13 SCr. nozice (pi.) 14 SCr. makaze (Tk. makas fr. Arab.) rne-2 m-e-t-P 703 v/eI-4 P 1139 cons bx vowel X B (s)ker-4 P 938-40, Joh. 835-7 [s]ker-4 P 938-44, Mh. 1.261 sek-2 P 895, Joh. 777-8, PaT 944, KM 575 g'her-2 P 441, \VH 1.569 f., A. Leskien, Die Bildung d. Nom. 348 (s. Ling. VII, 2, p. 158) :g'hel-2 P 434, B neg'h-P 760, Vas. 2.225, Brück. 367, 364, Mch. 329, 326 f. Vuk 437 makas Bern. 2.9 6 Fr. hache (Prank. *hapja, cf. OHG happa, heppa) 7 Rum. topor (Slav., cf. Bulg., Russ. topor) 8 Ir. biail 9 Ir. tuag 10 ON barda 11 Lith. kirvis 12 Skt.: kut*hara- 13 Av. taëa- pet-2 P 825-6, WH 2.282 f. [s]kep-2 (s)kftp- etc. REW 4035, Gam. 502, P 930-2 cons, t vowel ex B, Vas. 3.121 f., 3.95, Lok. 1964 bhei(9)-P 117-8, LP 14, 172 [s]teu-l P 1032 bhar-P (108)110 :bher-Joh. 616 f. (PaT 52, 94, 129) [s]ker-4 P 938-40, Pr. 259 cons. K vowel X Mh. 1.223 f. tekth-P 1058, Mh. 1.468, 491 9.27 SPLIT (vb. trans.) (s. Lingüistica VII, 2, 1965, p. 143) 9.45 HEW (s. Lingüistica VIII, 1, 1966—1968, p. 31) 9.25 AX 1 Grk. pélekus (Akkad. pilakku) 2 Grk. aksïnê 3 NG tsekoúri (Byz. sikoúrion, tzikoúrion fr. Lat. secüris) 4 Lat. (bipennis) pilakku Bois. 761, B ag-wesï ak-sï P 9, Bois. 65, Frisk 1.115 f., WH 1.71 f., F 54 sëk-2 B, P 895 sê[i]-k- s3k- sek WH 2.506, 504 f. dwô(u)- dwei-P 228-9 9.46 BORE (s. Lingüistica VII, 2, 1965, p. 154) 9.47 AUGER (s. Lingüistica VII, 2, 1965, p. 154) 1 Grk. 9.48 SAW (sb.) prïôn 2 Lat. serra pris- P 846, Bois. 813 f. cons. s vowel e* WH 2.524 142 The repetition- of phonemic characteristics 3 It. sega (Lat. secare) 4 Rum. ferestrau (Hung, füresz) 5' W. llif (Lat. lima) 6 OE snid 7 Lith. piuklas 8 ChSl. pila (Gmc., cf. OHG fíhala, fila) 9 SCr. testera (Tk. testere) 10 Skt. krakaca- 11 Skt. karapatra- 12 sek-2 REW 7764, P 895 :se[i]k- s3k- WH 2.504 f. füresz B lei-3 slei-B, P 662-3, WH 1.801 f. sneit-P 974, F 440, Joh. 913 f., KM 670 (:FaT 1096 f.) pcu- P 827, Fr. .599 f. peig-1 peik'-B, Vas. 2.356, Brück. 414, P 794 f,, KM 190, Joh. 535, FaT 216 testere B ker-1 k'er-sker- P 567-8, B (:Mh. 1.275) k»er-l P 641, B (:Mh. 1.275) pet-2 P 825-6 9.50 NAIL TACK (s. Lingüistica VII, 2, 1965, p. 155 f.) 9.81 CARVE 1 Grk. glúpho 2 Lat. scalpere gleubh-P 401, Bois. 152, Frisk 1.315, WH 1.610 f. (s)kel-l P 923-6, WH 2.486 f. 3 Fr. tailler (Lat. taliáre, talea) 4 Ir. snaidim 5 W. cerfio (ME cerve) 6 Br. kizella (OFr. cisel fr. VLat. Sciselum, *caesellum, Lat. caedere) 7 Br. bena ON " skera 9 ON grafa 10 Dan. udhugge 11 NHG ausmeisseln 12 Lith. iskalti 13 Lett, téluot (ORuss. télo) 14 ChSl. vajati 15 Boh. vytesati tal- REW 8542, Gam. 829; P 1055, WH 2.643 (s)nadh- P 972-3 gerebh-B (9.22), P 392 [s]k[h]ai-B, REW 1474, FEW 2.40, Gam. 225, P 917, WH 1.129 lihel(3)-P 117, B (9.45) (s)ker-4 P 938-9, Joh. 835 ff., FaT 1010 f. ghrebh-2 P 455-6, 1 Joh. 396 f., FaT 343 kau- P 535, B, ■ FaT 426, Joh. 177 [sJmei-2 P (968)697, KM 472, FaT 709 kel-3 P 545-6, Fr. 211 cons, t vowel X B (9.83), ME 4.171, Vas. 3.91 cons, w vowel X Vas. 1.175 tekth- ,F 1058-9, Mch. 527, Br(ick. 63, Vas. 3.99 143 B. Voduselc •9.84. CHISEL (sb.) 1 Grk. smile 2 Lat. scalprum 3 It. cesello (VLat. *caesellum, Lat. caedere) 4 Rum. dalta (Slav., cf. Bulg. dlato, Russ.-ChSl. dlato, Russ. doloto) 5 W. cyn (Lat. cuneus) . 6 W. gamg 7 Dan. bejtel (MLG bétel, beitel) 8 OE graefsex 10 Lith. kaltas (s)mei-2 P 968, Bois. 885 f. (s)kel-l P 923-6, WH 2.486 f. [s]k[h]ai-REW 1474, BA 2.877, P 917, WH 1.129 delbh-, B, P 246, Vas. 1.360, 359 ak'-2 B, LP 57,' P (18)626-7 cons. gx, n vowel e* B bheid- FaT 69, I' 116-7 ¡»hrebh-2 P 455-6, B sëk-2 P 895 kel-3 P 545-6, Fr. 211 12.352 POINT 1 Grk. akis 2 NG mtitë (Grk. mutis) 3 Lat. cuspis 4 Lat. muero 5 It. punta (Lat. punctum, pungere) Rum. virf (Slav., cf. ChSl. vtrchi, SCr. vrh) ak'-2 P 18-9 itiü-1 B (4.23), P 751 cons. kx vowel X B, WH 1.318 nieuk'-P 745, WH 2.117 f. peuk'- peug'-BA 4.3149, B, REW 6847, P 828, WH 2.389, 383 : wer-2 B, P 1151-2, Vas. 1.190 7 Ir. rind 8 Ir. benn 9 Ir. corr 10 W. blaen 11 Br. beg (Fr. bee fr. VLat. beccus fr. Gall. *bekkos) 12 ON oddr 13 Dan. spids (MHG, NHG spitze) 14 Lith. galas • 15 Lith. smailusis 16 SCr. siljak 17 Boh. hrot 18 Pol. koniec ChSl. kontct 19 Skt. agra- cr-3 P 326-9 :per-2 P 808-16, B bend- P 96, LP 38 [s]ker-4 P 938-44, B mel-1 smel-P 716-9, LP 43 bheg- 3, REW 1013, Gam. 93, P 114, WH 1.99 :bak- P 93, B (12.75) dhe-2 F 235-7, Joh. 166, FaT 787 :wes-4 P'1172, FaT 1524 (s)p(h)el-l FaT 118, P 981, KM 728, 726 f., Joh. 887-9 -gwel-1 P 470-1, Fr. 130 (s)mei-2 P 968 se(i)-3 sei-P (891)915-6, B, Vas. 3.398 gher-3 P 440, Mch. 145, Brück. 159, Vas. 1.311 ken-3 P 563-4, Brück. 252,74, Vas. 1.613 f. ag-ro- eg-ro-? P 8, Mh. 1.18 12.3531 EDGE (of a knife, sword, etc.) 1 Grk. akme ak'-2 P 18-19 144 The repetition- of phonemic characteristics 2 NG athéras (Grk. athër) 3 NG kôpsë (Grk. kópto) 4 Grk. stómá 5 It. filo (Lat. fílum) 6 It. taglio (Lat. täliäre, tälea) 7 Fr. tranchant (Lat. truncare, truncus) Ir. faebar (Brit., cf. W. gwaew) 9 NIr.< bëal 10 W. min 11 Br. dremm 12 Br. lemm 13 Br. neudenn 14 Br. barvenn (Lat. barba) 15 Du. scherp andh- B, P 40-1, Bois. 18 (:Frisk 1.28) [s]kep-(s)kâp- etc. B, P 930-1 stcm-en-P 1035, B, Bois. 914 f. g™heiS-REW 3306, BA 3.1643, P 489, WH 1.497 f. tax- REW 8542, BA 5.3699, P 1055, -WH 2.643 trenk-1 REW 8953, Gam. 857', P 1093, . (WH 2.710 f.) ç'hais-o-B. P 410, LP 10,18, 28, 67 :g'hei-l P 424 f., B (20.26) gwet- P 481, B (4.25) :bheg- P 114, WH 1.99 malt-P 698, B, LP 43 derk'-P 213, B, LP 4 lei-3 slei-P 662-3, B (15.78) (s)në- (s)nêi-P 973, B (6.38), ■ LP 24 bhar-B (4.142), LP 61, P (108)110 (s)lter-4 P 938-43, B (15.78) 16 MHG snide 17 Skt. dharä- sneit: P 974 dho- P 272 (Mh. 2.101) :dheu-3 P 261, B 12.3532 EDGE (of a table, forest, etc.) 1 Grk. kráspedon 3 Grk. cheilos 4 Lat. , ora 5 Lat. margö 6 Pr. bord (Gmc. *bord, cf. ON bord) i 7 Sp. canto (Lat. cantus fr..Gall. *cantos). 8 Ir. brü 9 Ir. cimas 10 Ir. ochar 11 Ir. bil 12 Br. ribl (Lat. rípula) k'er-1 P 574-5, Bois. 509, Frisk. 2.7 f. ped-2 P 790-1 s>helu-[na] etc. P 436, B (4.25), Bois.' 1053, FaT 319 ous-1 P 784, WH 2.218 f., 224 f. mereg'-P 738, WH 2.39 f. bher-3 REW 1215, FEW 1.436 ff., Gam. 122, P (133)138 kam-p-kam-tho-Cor. 1.643 f., REW 1616, FEW 2.232, Gam. 205, P (525)526 bher-P 170, B kem-1 P 555, B ak'-2 P 18-21 cons, x vowel X B rei-1 B (1.27), P 857-8, WH 2.436 145 B. Voduselc IS Goth, skaut 14 ON ro*nd 15 OE snaed 16 OHG trädo 17 Lith. krastas 18 Lett, mala 19 ChSl. viskriltje 20 SCr. ivica 21 Boh. hrana 22 Pol. brzeg 23 Skt. anta- 24 Av. karana- (s)keu-5 P (954)955-6, F 431, Joh. 819, FaT 1011 f., 1044 f., KM 677 rem- P 8644, Joh. 723 f., FaT 876, KM 580 sneit- P 974, B der-4 P 206-7, B [s]ker-4 P 938-42, Fr. 289 mel-8 P 721-2 [s]ker-3 P 935-6, Bern. 1.615 cons, x vowel X Bern. 1.439 gher-3 P 440, Mch. 143, Brück. 155, Vas. 1.304 bhereg'h-P 140-1, Brück. 44, Vas. 1.76, Mch. 47, B (1.27) ant- P 48, B (12.35) cons. kx, lr vowel ex B (12.35) 20.25 ARROW 12.75 HOOK (s. Lingüistica VII, 2, 1965, p. 156) 18.51 WRITE (s. Lingüistica VII, 2, 1965, p. 154) 1 Grk. tókseuma (Iran., cf. NPers. taxs) 2 Grk. bélos 3 Grk. Jos Lat. sagitta (Etruscan?) 5 It. freccia tekw. P 1059, Bois. 975, Mh. 1.467 f. gv.el-2 P 471-2, Bois. 118, Frisk 1.231 f. eis-1 P 299-301, Bois. 378, Frisk 1.730 cons, s, g* vowel ex B. WH 2.464 pel-1 (Fr. fléche fr. Frank. REW 9424a, *fleu[k]a, cf. MDu. vlieke, vleke) 6 Br. bir 7 Goth, arhwazna 8 ON fleinn OE flan 9 Dan. pil (Lat. pilum) 10 OE strael 11 Lith. vilycia 12 Lett, bulta (MLG bolte) Gam. 423, Cor. 2.535 f., BA 3.1712, FaT 238, P (798)835-7 cons, r vowel X B arku- P 67, F 56, Joh. 35, WH 1.64 cons, p, 1 vowel X P 1000, B, Joh. 905, FaT 235 (peis-1) pis-FaT 825, Joh. 1115, KM 542, P 796, V/H 2.304, 302, 307 f. ster-4 P 1028, KM 755 wei-1 ME 4.639, P 1120 ff., Vas. 1.200 bheld- B, P 124 146 The repetition- of phonemic characteristics 13 Lett, sautra 14 Boh. sip 15 Skt. saru- 16 Skt. Mn*a- 17 Av. tig*ri 18 Av. asti sk'eu- P 954-5, B, ME 4.10 (Pr. 969) eons, x vowel X Vas. 3.400, Mch. 499, 368 K'el-3 P 552 cons, x vowel ex Mh. 425, B [s]teig- . P 1016, Mh. 1.507, 525 f. cons, x vowel ex B 20.26 SPEAR 1 Grk. dóru 2 Grk. Lat. lôgchè lancea 3 Grk. aichmë 4 Grk. égchos*4 5 Grk. ákon 6 NG kontári deru-P 214 cons. 1 vowel e* B, Bois. 586, WH 1.757 f„ Frisk 2.133 f., P 832 ftik'-P 15, Bois. 31, Frisk 1.48, WH 1.670 f. cons. K vowel ex Bois. 214, Frisk 1.440 f., WH 1.608, 2.327 ak'-2 P 18-20, Bois. 40, 32' f., Frisk 1.62 k'ent- 7 Lat. hasta Lat. pîlum 9 Fr. épieu (Frank. *speut, c£. OSax. spiot, OHG spioz) 10 Rum. sulit*a (Slav., cf. ChSl., SCr. sulica) 11 Ir. gae 12 Ir. sleg 13 Ir. croisech** 14 Ir. lâigen 15 Ir. manais** 16 ON spjo*r (Late Grk. kontárion, B, P 567, Grk. kontôs) Bois. 491, 434, Frisk 1.913, 820 f. 17 Lith. ietis 18 Lith. ragotine* g'hast-o-1 g'hazd-o-P 412, WH 1.636 (peis-1) pis-P 796, B, WH 2.304, 302, 307 f. (s)p(h)eud- REW 8163, Gam. 372, P 998-9, KM 726, (FaT 1138) . sk'eu-B, P 954-5, Vas. 2.44, Brack. 525, Mch. 484 g'liais-o-P 410, LP 18, 28, 67 :g'hei-l P 424-5, B , WH 1.575 selg'-P 900, B cons. kx, r ■ vowel X B lagh-P 652, LP 29, Bois. 561, WH 1.757 (-.Frisk 2.92) cons. m vowel a B (s)per-l P 990, Joh. 893 f., KM 722 :(s)p(h)er-4 P 992 FaT 1140, 1113, 1125 ei-1 P 293, B cons, r, gx vowel ex Fr. 684, 736 f., Vas. 2.526 f. I* in Av. tig*ri = the voiced guttural constrictive. '* Not in the list, but mentioned in Buck's discussion. 147 B. Voduselc 19 Lett, sk'eps ChSl. kopbje 20 Pol. wlocznia 21 Av. arsti (s)k8p-2 (s)kâp- etc. P 930-2, ME 4.33 [s]kep-2 -P 930-2, Vas. 1.621 welk-1 P 1145, Briick. 627 eres-1 P 335, Mh. 1.125 10 Goth, mëki 20.27 SWORD 1 Grk. ksiphos 2 Grk. mâchaira 3 Grk. âor 4 NG spathi (Grk. spathion, spâthê) 5 Lat. gladius (Celt., cf. W. cleddyf, Br. kleze) 6 Lat. ênsis 7 Rum., sabie (Bulg. sab(i)ja, SCr. sablja fr. Hung, szablya, szab) 8 Ir. colg 9 Goth, hairus cons. kx, s vowel X B, Bois. 678 f. mag'h-P 697, Bois. 616, Frisk 2.186, WH 2.3,4 wer-1 P 1150, B, Bois. 66, 15 f., Frisk 1.117, 23 f. sp(h)e[i]-3 sp'(h)ei-B, P (982)980, Bois. 888 f., 896 kel-3 WH 1.603 f., ,P 545-6 n*si-s P 771, WH 1.406 szab B, KM 617 f. keI-2 P 545 [s]ker-4 P 938-40, F 235, Joh. 835-7 11 ON . sverd 12 ON brandr 13 Lett, zuobens cons, m, K vowel ex F 352 f., Joh. 1096, Vas. 2.128, Bruck. 331, Mch. 290 swer-4 P 1050, B, KM 693 : wer-1 P 1150, FaT 1214, 1560, Joh. 137 f. bher-2 P (132)143-5, Joh. 619-20 g'embh-P 369, ME 4.756 15.45 LOUD 1 Grk. mégas 2 NG dunatôs (Grk. dunatds, diinamai) 3 Lat. clarus 4 It. forte (Lat. fortis) 5 It. alto (Lat. altus) 6 Rum. tare (Lat. talis) 7 Ir. ard 8 W. uchel meg'[h]-P 708, B (deu-2) du-P 218 :deu-3 P 219, Frisk 1.423 f:, (Bois. 204) kel-6 P 548-9, WH 1.228 bhereg'h-REW 3457, P 140, WH 1.535 ff. :dher-2 P 252 ff., B al-3 REW 387, P 26, WH 1,4,31 f. to-1 REW 8543, P 1086-7 er-3 P (326)339 up-(o).- etc. (e)up-s-P 1106-7, LP 19. IE n*si-s = with syllabic n. 148 The repetition- of phonemic characteristics 9 ON liar 10 OE ' hlüd 11 Lith. balsus 12 Lith. gafsus 13 Lith. didis 14 Lett, skan's 15 Lett, skal's 16 ChSl. velijt 17 SCr. glasan 18 Russ! gromkij 19 Skt. ucca- 20 Skt. tara keu-2 P 588-9, Joh. 202 ff„ FaT 451 f. k'leu-1 P 605-6, KM 427, FaT 668 bhel-6 P 123-4, Fr. 32 gwer(3)-4 P 478, Fr. 152, 154 :g'ar- P 352, B (15.44) dei-1 P 183, Fr. 93 cons. s, k vowel ex B, Fr. 794 (s)kel- (s)kwel- P 550, Fr. 544, wel-3 F 1138, B, Vas. 1.180, 181' gal-2 P 350, Vas. 1.287, Bern. 1.323, Brück. 144 :ghel-P,428, Mch. 129 ¡ ghrem-2 P 458-9, Vas. 1.311, 310 Bd- P 1103-4, Mh. 1.99, 101 ter-4 P (1074)1088-9 Mh. 1.497 18.12 SING 1 Grk. aeídó au-6 P 76-7, Bois. 15, Frisk 1.22 f. 2 NG tragoudó (Grk. trago*déó) 3 Lat. canere 4 Ir. gaibim 5 Goth, siggwan 6 ON gala 7 Lith. dainuoti 8 Lith. giedoti 9 ChSl. péti ter-3 B, P 1071-3, Bois. 978 kan-P 525, WH 1.154 f. ghabh-P 407-8, B, LP 365 sengwh-P 906-7, F 419, Joh. 786, KM 709 f. FaT 1227 gliel-P 428, Joh. 382 if., KM 230, 244, 499," FaT 269 cons. dx vowel ix B, Fr. 80 ge(i)- P 355, Fr. 150, Vas. 1.251 :ga(i)-B cons, p vowel ix Vas. 2.422, Bruck. 404, Mch. 305 18.13 SHOUT, CRY OUT (s. Lingüistica VII, 2, 1965, p. 149) 18.41 CALL (vb. = SUMMON) (s. Lingüistica VII, 1, 1965, p. 42) 18.42 CALL (vb. = NAME; BE CALLED, NAMED) 1 Grk. kale'o kel-6 P 548, WH 1.141 f. :kal- Bois. 397 f., Frisk 1.762- f. o* in Grk. trago*déo' = ó with iota subscriptum. 149 B. Voduselc 2 Grk. onomadzo 3 NG lego (Grk. lego) 4 Lat. vocare 5 Fr. appeler (Lat. appelare) 6 Ir. gairim (?) 7 W. galw 8 Goth, haitan enfoj-mti1 P 321, B (18.28), Bois. 704, WH 2.173 leg'- B, P 658, Bois. 563 f., Frisk 2.94 if., WH 1.780 wek™-P 1135-6, WH 2.823 ff. pel-2a REW 542, P 801, WH 1.59 g'ar- P 352, B, (cf. LP 368, § 555, 6) gal-2 P 350, B kei- ke[i]-P 538, F 236, Joh. 198, KM 300 f., FaT 388 etc. 9 Lith. vadinti 10 Lett, saukt 11 ChSl. nareSti 12 SCr. nazvati 13 Skt. abhi-dha- au-6 P 76, Fr. 1177 f., Vas. 1.163 f. kau- keu-k'au- k'eu-P 535-6, Fr. 968 rek-2 P 863, B (18.22), Vas. 2.525, 508 f., Fr. 716 f. :rei-3 rei-P 859-60 :wer-6 P 1162 g'hau-P 413, Vas. 1.447, Brack. 658, Mch. 589, Vuk 402 dhe-2 P 235, B IE en(o)-nin*- = with syllabic n. The chosen 55 synonymic sets are composed of 680 elements. There are 587 established Indo-European roots, 8 imitative radical morphemes belonging to individual Indo-European languages or languages groups, 13 foreign radical morphemes, and 72 unknown roots. The tables Nos' 38 to 42 give the consonantal distributions of the roots contained in the concept-group. As in establishing the etymologies of the single synonymic sets a somewhat unusual distribution of the root vowels came to light, this vowel distribution is separately listed in the table VI. Each of the vowel classes A and E is divided into two classes: the short A and the long A on the one hand, and the short E and the long E on the other. In all instances where by the traditional methods the original length or brevity of the root vowel cannot be determined with certainty, the vowels A and E are taken as short. Due to the interdependence Of the concept groups SHARP and ROUGH the corresponding vowel distribution for the concept group ROUGH is set up in the table VII. The table VIII gives the same vowel distribution of the Representative Sample, and the resulting comparative data are tabulated in the tables Nos- 43 and 44. 150 The repetition- of phonemic characteristics LIST OF ESTABLISHED INDO-EUROPEAN ROOTS OF THE GROUP »SHARP« I. (Type V). I: ei-1 (= root in Pokorny), 20.26 (= Buck's concept-number); ei-4, 8,21 sb.; roots 2, different roots 2. — U: au-6, 18.12, 18.42; u-1, 18.13; roots 3, different roots 2. sum 1: 5 4 II a. (Type CV). L: leu-2, 8.32, 8.331, 8.332, 9.22, 9.27 leu-2, 1.441; roots 6, different roots 2 = LU 6, 2. T: da-, 8.32, 8.33', 8.332; dei-1, 15.45; deu-2, 15.45; dhau-, [s]teu-l, 4.53, 9.25, 9.46; dhe-2, 9.46, 9.47, 12.352, 18.42; dho-, 12.353'; to-1, 15.45; roots 15, different roots 8 = TA 3, 1, TI 1, 1, TU 5, 3, TE 6, 3. K: ge(i)-, 18.12; gwhei9-, 12.3531; ke(i)-, 18.41, 18.42; [s]k[h]ai-, 9.22, 9,24, 9.81, 9.84; gou-, 18.41; g'hau-, 18.41, 18.42; kau-, 8,25, 9.22, 9.45, 9.81; kau-, k'au-, 18.13, 18.41, 18.42; kau-I-, 4.16; lr.eu-2, 12.75, 15.45; k'eu-1, 8.22, 9.46; roots 23; different roots 11 = KI 8, 4, KU 15, 7. P: bha-2, 18.13, 18.41; bhei(S)-, 9.25, 9.45, 9.81; b(h)eu-2, 6.63, 9.46, 18.13; peu-1, 8.23, 8.24; peu-, 8.32, 8.331, 9.22, 9.48; roots 14, different roots 5 = PA 2,1, PI 3, 1, PU 9, 3. S: se(i)-2, 8.21 sb.; se(i)-3, 6.35, 6.37, 9.50, 12.352; seu-1, 9.47; roots 6, different roots 3 = SI 5, 2, SU 1, 1. N: + [s]mei-2, 9.22, 9.45, 9.81 [s. Ill b]; meu-1, 4.58; mtt-1, 15.78, 12.352; me-2, 8.32, 9.24; + [s]ne-, 6.35, 6.63 [s. Ill b]; roots 5+5, different roots 3+ +2 = NI +3, +1, N'U 3, 2, NE 2+2, 1+1. R: rei-1, 8.32, 12.3532, 18.51; rei-3, 18.13; reu-g-, 4.57; roots 5, different roots 3 = RI 4, 2, RU 1, 1. sum 88 a: 74 + 5 35 +2 II b. (Type VC). L: al-2, 15.45, el-a, 6.37; roots 2, different roots 2 = AL 1, 1. EL 1, 1. — T: fld-, 15.45; ed-, 4.27, 4.58, 8.25; edh-, 6.36; roots 5, different roots 3 = UT 1, 1, ET 4, 2. — K: Sik'-, 6.36, 20.26; ag-ro-, 12.352; ag-wesl, 8.25, 9.25; ak'-2, 4.57, 15.78, 1.441, 1.442, 6.36, 6.63, 8.28, 9.23, 9.47, 9.50, 9.84, 12.352, 12.3531, 12.3532, 20.26; ok'"-, 6.37; roots 21 different roots 5 = AK 18, 3, IK 2, 1, EK 1, 1. — P: up-o-, (e)up-s-, 15.45; root 1, different root 1 = UP 1, 1. — S: eis-1, 20.25; ous-1, 12.3532; roots 2, different roots 2 = IS 1, 1, US 1, 1. — N: en(o)-mn*-, 18.42; root 1, different root 1 = EN 1, 1. — R: ar(3)-, 8.21 vb., 8.21 sb.; er-3, 4.661, 12.352, 15.45; erS-l, 8.23, 8.24; ere-s-2, 4.57; ereu-2, 8.22, 8.23; roots 10, different roots 5 = AR 2, 1, ER 8, 4. Sum lib: 42 19 II c. (Type YVC). K: aueg-, 6.35; root 1, different root 1 = UEK 1, 1. Sum lie: 1 1 Sum II: 117 + 5 55 +2 151 B. Voduselc III a. (Type CVC). I,P: [leb-] lal>, 8.21 vb.*; lep-3 = 2, 1.441; jgp., 8.23, 8.24; roots 4, different roots 3. = LAP 1, 1, LEP 3, 2. — LK: lftgh-, 8.22, 8.23, 8.24; [s]leig-, 8.25; leg'-, 18.42; roots 5, different roots 3 = LAK 3, 1, LIK 1, 1, LEK1. 1- Sum L-III a: 9 6 TL: tal-, 15.78, 9.22, 9.45, 9.81, 12.3531; del-3, 9.22, 9.27, 9.45; roots 8, different roots 2 = TAL 5, 1, TEL 3, 1. — TR: der-4, 4.661, 8.28, 8.331, 8.33?, 9.27, 12.3532; deru-, 20.26; dher-5, 4.661; ter-3, 9.22, 9.45, .9.46, 9.47, 9.50, 18.12; ter-4, 9.47, 15.45; roots 16, different roots 5 = TER 16, 5. — TN: dem-, 9.45; tem-1, 9.22, 9.24; roots 3, different roots 2 = TEN 3, 2. — TK: dheig™-, 6.63, 8.22; + [s]teig-, 15.78, 20.25 [s. IV]; dek'-2, 8.26, 9.50; tele™., 20.25; roots 5+2, different roots 3+1 = TIK 2+2, 1+1, TEK 3, 2. ■ Sum T_m fl; 32 + , J2 + ^ KP: ghabh-, 18.12; ghabh-(o)-lo-, 8.26; kap-, 9.24; + [s]kfip-, [s]kab-, 12.75 [s. IV]; + [sjtep-, 15.78, 8.28, 8.332, 9.22, 9.25, 9.45, 12.3531 20.26 [s. IV]; roots 3+9, different roots 3+2 - KAP 3+1, 3+1, KEP +8,4-1. — KT: g'hed-, 4.661; Sc^nd-, 15.78; roots 2, different roots 2 = KET 2, 2. — KL: gal-2, 15.45, 18.13, 18.41, 18.42; gel-l, 1.442; gweM, 6.36, 12.352; gwel-2, 20.25; ghel-, 18.12; gheJu-[na], 12.3532 g'hel-2, 8.21 sb., 8.331; kel-1, 1.442; kel-2, 20.27; kel-3, 8.22, 8.25, 9.22, 9.27, 9.81, 9.84, 20.27; kel-6, 15.45, 18.13, 18.41, 18.42; k'el-2, 8.28; k'el-3, 20.25; k™el-l, 4.661, p,.21 vb.; + [s]kel-1, 9.23 [s. IV]; roots 29+1, different roots 14+1 = KAL 4, 1, KEL 25+1, 13 + 1. — KR: g'ar-, 18.13, 18.41, 18.42; kar-2, 18.13, 18.41; kar-3, 1.441, 1.442; ger-2,'4.57; ger-3, 8.28, 9.46, 12.75; glver (5 )-4, 15.45; g«'eru-, 6.63, 9.46, 9.47; gher-3, 12.352, 12.3532; g'her-2, 8.26, 9.24; g'her-6, 9.22; g'her-s-, g'her-, 1.441, g.28; gwher-, 8.32, 9.24; ker-1 [sker-, k'er-], 8.27, 9.48, 18.13, 18.41; ker-5, 12.75; k'er-1, 4.39, 4.17, 8.26, 12.3532, 12.75; kwer-l, 9.48; [s]ker-3, 6.63, 8.27, 8.28, 8.331, 9.50, 12.3532, 12.75; + [s]ker-4, 15.78, 6.91, 8.27, 8.28, 8.331, 9.22, 9.23, 9.24, 9.25, 12.352, 12.3532, 20.27 [s. IV]; roots 42 + 12, different roots 17 + 1 = KAR 7, 3, KER 35+12, 14+1, — KN: [kam-p-] kan-th-, 12.3532; kan-, 18.12; gem-, 6.63; gen-, 4.16, 6.63, 6.93, 9.23; g'en-1, 6.35; gwhen-2, 8.32; kem-1, 12.3532; kem(5)-4, 8.21 vb.; ken-2, 4.58; ken-3, 12.352; Icona-mo-, 4.16; roots 14, different roots 11 = KAN 2, 2, KEN 12, 9. — KK: kakka-, 4.661; k'a.k-2, 8.21 sb., 8.26, 9.50; geig'-, 15.78; g'egh-, 8.21 sb.; keg-, 8.25, 12.75; k'ek»-, 4.661; koko, kokk-, 4.16; roots 10, different roots 7 = KAK 4, 2, KIK 1, 1, KEK, KOK 5, 4. — KS: kwas-, 4.53; g'hais-o-, 9.47, 12.3531, 20.26; kes~, 6.63, 6.91, 6.93, 8.22, 8.23, 9.23; k'es-, 8.32, 8.331, 9.22, 9.23; roots 14, different roots 4 = KAS 1, 1 KIS 3, 1, KES 10, 2. Sum K-III a: 114 + 22 58 +4 PT: bheid-, 4.58, 9.27, 9.84; bhedh-1, 8.22; p6d-2, 12.3532; pet-2, 9.25, 9.48; roots 7, different roots 4 = PIT 3, 1, PET 4, 3. — PL: bhel-3, 4.57, 18.13; bhel-6, 15.45; pel-1, 20.25; pel-2a, 18.41, 18.42; pel-3a, 6.93; pel-(i)-s-, 1.441, 1.442; * The radical morpheme is lab- according to WH 1.739 f... Therefore, the form lab- given sub laborare 8.11-8, 8.21-5 (Lingüistica VI, 24, VIÍ/2, 131, 152, 158), and sub labi 10.42-3 (Lingüistica VII/2, 139, 146), should be corrected into lab-. 152 The repetition- of phonemic characteristics bol-, 8.28; roots 10, different roots 7 = PEL 10, 7. — PR: bhar-, 6.93, 9.25, 9.50, 12.3531; bher-2, 20.27; bher-3, 6.93, 8.28, 9.22, 9.46, 9.47, 12.3532; bher-, 12.3532; per-2, 6.63; roots 13, different roots 5 = PAR 4, 1, PER 9, 4. — PK: bak-, 12.75; pak', pflg', 4.39, 8.22, 8.23, 8.24, 8.32, 8.331; peig-l, peik'-, 9.48, 18.51; peuk', peug', 12.352; bheg-, bhe-n-g-, 8.32, 12.352; pek'-2, 6.91; pekw-, 1.442; roots 14, different roots 7 = PAK 7, 2, PIK 2, 1, PUK 1, 1, PEK 4, 3. — PS: peis-1, 20.25, 20.26; pes-1, 8.211; roots 3, different roots 2 = PIS 2, 1, PES ■ Sum P-III a: 47 25 ST: se[i]dh-, 6.91; root 1, different root 1 = SET 1, 1. — SR: ser-5, 15.78, 8.25, 8.27, 8.331, 8.332; roots 5, different root 1 = SER 5, 1. — SK: seg'h-, 9.50; sek-2, 4.57, 4.661, 1.441, 1.442, 6.93, 8.23, 8.24, 8.32, 8.331, 8.332, 9.22, 9.23, 9.24, 9.25, 9.27, 9,48, 9.84; roots 18, different roots 2 = SEK 18, 2. Sum S-IIIa: 24 4 ■ NT: mad-, 6.93, 9.23; mat-2, 8.25, 8.28; roots 4, different roots 2 = NAT 4, 2. — NL: mel-1, 12.352; mel-6, 18.51; mel-8, 12.3532; roots 3, different roots 3 = NEL 3, 3. — NR: mer-5, 4.58; root 1, different root 1 = NER 1, 1. — NK: mag'h-, 9.23, 20.27; mak-, 12.3531; meuk'-, 12.352; meg'h-, 1.441, 15.45; neg'h- 6.93, 9.23, 9.24; roots 9, different roots 5 = NAK 3, 2, NUK 1,1, NEK 5, 2. Sum N-III a: 17 11 RT: red-2, 6.93, 8.27; roots 2, different root 1 = RET 2, 1. — RN: rem-, 12.3532; root 1, different root 1 = REN 1, 1. — RK: reg'-l, 6.93, 8.23, 8.24, 8.27; rek-2, 18.42; roots. 5, different roots 2 = REK 5, 2. Sum R-IIIa: 8 4 Sum III a: 251 + 24 120 + 5 III b. (Type CCV). KL: kleu-, klau-, 9.50; k'leu-1, 15.45; roots 2, different roots 2 = JKLU 2, 2. — KS: kseu- [skeu-3], 4.54; root 1, different root 1 = = KSU 1, 1. — PN: pneu-, 4.54; root 1, different root 1 = PNU 1, 1. — SP: (s)p(h)ei-l, 6.63, 9.27, 9.50, 12.352; (s)p(h)yeu-, 4.57; (s)p(h)e[i]-2, 8.23, 9.27, 1 20.27; roots 8, different roots 3 = SPI 5, 2, SPE 3, 1. — ST: sta-, 1.442; stai-, 1.441, 1.442; roots 3, different roots 2 = STA 1, 1, STI 2, 1. — SL: [lei-3] slei-, 15.78, 9.48, 12.3531; roots 3, different root 1 = SLI 3, 1. — SN: (s)mei-2, 8.25, 9.84, 12.352; (s)ne-, 6.36, 12.353' [s. II a]; roots 5, different roots 2 = SNI 3, 1, SNE 2, 1. — SK: (s)keu-5, 8.23, 8.24, 12.353; sk'eu-, 20.25, 20.26; roots 5, different roots 2 = SKU 5, 2. sum Illb: 28 14 III c. (Type VCC, VCVC). ST: ost(h)-, 4.16, 9.27; roots 2, different root 1 = EST 2, 1. — NP: enebh-1, 9.47; root 1, different root 1 = ENP 1, 1. — NT: andh-, 12.3531; ant-, 12.3532; roots 2, different roots 2 = ANT 2, 2. — NK: ank-2, 12.75; onogh-, 4.39, 9.50; roots 3, different roots 2 = ANK. 1, 1, ENK. 2, 1. — NS: n*si-, 20.27; root 1, different root 1 = NS 1, 1. — RP: ereb-, 9.46; root 1, different root 1 = ERP 1, 1. — RK: arku-, 20.25; er(e)k-2, 153 B. Voduselc 8.22, 18.51; roots 3, different roots 2 = ARK 1, 1, ERK 2, 1. — RS: eres-1, 20.26; root 1, different root 1 ERS 1, 1. Sum III c: 14 11 Sum III: 293 + 24 145 + 5 IV. (Type CVCC, CVCVC, CCVC, CCCV). TL-: dhelg-, 6.63, 8.32, 8.331, 8.332; dhelbh-, 8.22, 9.84; roots 6, different roots 2 = TEL- 6, 2. — TR-: derk'-, 12.3531; treb-, 8.21 .vb.; roots 2, different roots 2 = TER-, TRE- 2, 2. — TN-: denk'-, 4.58; root Indifferent root 1 = TEN- 1, 1. — TK-: tekth-, 9.25, 9.45, 9.81; roots 3, different root 1 = TEK- 3, 1. Sum T-IV: 12 6 KL-: gleubh-, 9.27, 9.81; roots 2, different root 1 = KLU- 2, 1. — KR-: ghreib-, .8.26; + [gerebh-] gribh-, 8.28. [s. below]; gwreug'h-, 4.58; grebh-1, 8.27; ghrebh-2, 6.91, 8.22, 8.23, 8.25, 9.50, 9.81, 9.84; gerebh-, 9.22, 9.81, 18.51 [s. above]; ghrem-2, 15.45; k'ers-2, 8.21 sb.; roots 15+1, different roots 7+1 = KRI- 1+1, 1+1, KRIJ- 1, 1, KER-, KRE- 13, 5. — KN-: g'embli-, 4.27, 6.91, 20.26; k'ent-, 20.26; roots 4, different roots 2 = KEN- 4, 2. — KS-: g'hast-o-, 4.17, 20.26; kost-, 4.16; roots 3, different roots 2 = KAS- 2, 1, KES- 1, 1. sum K-IV: 24+1 12+1 PL-: bheld-, 20.25; pelk', pelg'-, 8.28; roots 2, different roots 2 = PEL-2, 2. — PR-: prls-, 9.48; bhreg'-l, 4.57, 8.28; bhereg'h-, 12:3532, 15.45; roots 5, different roots 3 = PRI- 1, 1, PER-, PRE- 4, 2. — PN-: bend-, 4.17, 6.63, 12.352; pent-, 8.23; roots 4, different roots 2 = PEN- 4, 2. Sum P-IV: 11 7 SP-: (s)p(h)eud-, 20.26; (s)p(h)el-l, 6.63, 9.24, 9.27; (s)per-l, 20.26; (s)pen-(d)-, 6.63; roots 6, different roots 4 = SPU- 1, 1, SPE- 5, 3. — ST-: (s)teig-, 6.63 [s. Ill a]; (s)trei-g-3, 18.13; steb(h)-, 8.23; ster-4, 20.25; stom-en-, 12.3531; roots 5, different roots 5 = ST-I- 2, 2, STE- 3, 3. — SL-: slak-, 8.32; (s)leug-, (s)leuk-, 4.57; selg'-, 20.26; selk-, 8.21 vb., 8.21 sb.; roots 5, different roots 4 = SLA- 1, 1, SLU- 1, 1, SEL- 3, 2. — SN-: (s)nadh-, 9.22, 9.45, 9.81; sneit-, 8.32, 9.22, 9.48, 12.3531, 12.3532; sengwh-, 18.12; roots 9, different roots 3 = SNA- 3, 1, SNI- 5, 1, SEN- 1, 1. — SK-: (s)kap-, 8.22, 8.23, 8.25 [s. Ill a and the next]; + (s)kep-, 9.27 [s. Ill a and the former]; (s)keuSd-, 18.13; (s)kel-l, 1.442, 8.25, 9.27, 9.81, 9.84 [s. Ill a]; (s)kel-, (s)k»el-, 15.45; (s)ker-4, 6.93, 9.81, 12.353', 18.51 [s. Ill a]; roots 14+1, different roots 5+1 = = SKA- 3, 1, SKU- 1, 1, SKE- 10+1, 3+1. Sum S-IV: 39+1 21+1 NR-: mereg'-, 12.3532; root 1, different root 1 = NER- 1, 1. Sum N-IV: 1 1 RN-: remb-, 9.22; root 1, different root 1 = REN- 1, 1. Sum R-IV: 1 1 Sum IV:. 88 + 2 48 + 2 154 The repetition- of phonemic characteristics V. (Type CCVCC, CCCVC). TR-: trenk-1, 15.78, 9.22, 12.3531; roots 3, different root 1 = TRE- 3, 1. — PS-: pster-, 4.54; root 1, different root 1 = PS-E-1, 1. — SK-: s'kerd-, 4.57, 4.661; roots 2, different root 1 = SKE- 2, 1. Sum V: 6 3 Sum I-IV: 509 + 31 255 + 9 VI. Y: yeu-2, 8.21 vb.; ye-, 4.57, 6.63; roots 3, different roots 2 = YU 1, 1, YE 2, 1. YN: yam-, 8.23; root 1, different root 1 == YAN 1, 1. Sum Y: 4 3 W: wei-1, 8.26, 20.25; wei-3, 8.32, 8.332; wi-l, 9.46, 9.47; roots 6, different roots 3 = WI 6, 3. . WL: wel-3, 15.45; wel-4, 9.24; wel-7, 1.442, 9.50; roots 4, different roots 3 = WEL 4, 3. — WK: wek™, 18.41, 18.42; wog™h-nes-, 8.22, 8.23; roots 4, different roots 2 = WEK 4, 2. — WP: wab-, 18.13; wap-2, 18.13; roots 2, different roots 2 = WAP 2, 2. — WN: wem-, 4.57; root 1, different root 1 = WEN 1, 1. — WR: wer-1, 20.27; wer-2, 12.352; wer-3, 4.57, 6.35, 6.36, 9.46, 9.47; wer-7, 18.51; roots 8, different roots 4 = WER 8, 4. WL-: welk-1, 8.28, 20.26; roots 2, different roots 1 - WEL- 2, 1. — WR-: werg'-2, 8.21 vb.; wreg'-, 9.22; roots 2, different roots 2 = WER-, WRE- 2, 2. -W-: [dwo(u)-] dwei-, 8.23, 8.25, 8.26, 9.24, 9.25; twerk'-, 9.22; roots 6, different roots 2 = TWI 5, 1, TWE 1, 1. — koi-, kwi-, 18.13, gwes-, 9.50; gwet-, 12.3531; k'wek-, 6.91; roots 4, different roots 4 = KWI- 1, 1, KWE- 3, 3. — swer-4, 20.27; swerbh-, swer-, 9.46, 9.47; roots 3, different roots 2 = SWE 3, 2. — skwerb(h)-, 9.46; root 1, different root 1 = SKWE- 1, 1. Sum W: 43 26 Sum I-VI: 556 + 31 284 + 9 155 TABLE VI SHARP Sh°rt L0.nS I U ShE0rt TOTAL I — — — — .-2 2 3 2 — — — — 5 4 IIa L 6 2 — — — — ' 6 2 T — — 3 1 1 1 5 3 1 1 5 2 15 '8 K —■ — — -- 8 4 15 7 — — — — 23 • 11 P — — 2 1 3 1 9 3 — — — — 14 5 . S. —. — — — 5 2 1 1 — — — — 6 3 N — — — — 3 1 3 2 •' — . — ' 4 2 10 5 R — — — 4 2 1 1 — — 5 3 — — 5 2 24 11 40 19 1 ! 9 4 79 37 II bc L 1 1 . 1 1 . 2 2 T 1 1 4 2 — . — 5 3 K 18 3 — — 2 1 1 1 1 1 — — 22 6 P. 1 1 — ■ — — — 1 1 S —.' — — — 1 1 1 1 — — — — 2 2 N 1 1 — — 1 1 R 2 1 -- 8 -4 — 10 5 21 5 — — 3 2 4 4 14 8 1 1 43 20 III a L 4 2 — — 1 1 — — 2 2 2 1 9 6 T — — 5 1 4 2 — — 25 10 — — 34 13 K 21 12 1 1 4 2 — — 110 47 — — 136 62 P 11 3 — — 7 3 1 1 28 18 — — 47 25 S 23 3 1 1 24 4 N 7 4 — — —r — 4 1 9 6 — — 17 11 R 6 3 2 1 8 4 43 21 6 2 16 ' 8 2 2 203 89 5 3 275 125 III bc L T K 3 3 — — — — 3 3 P 1 1 — -— — —■ 1 1 S — — 1 1 13 5 5 2 2 1 5 2 26 11 N 3 3 3 2 — — 6 5 R 1 1 4 3 — — 5 4 4 4 1 1 13 5 9 6 9 6 5 2 41 24 IV—V L T 15 7 — — 15 7 K 2 1 — — 2 ■ 2 3 2 18 8 — — 25 13 P — — .— — 1 1 — — 11 7 — — 12 8 S 7 3 — — 7 3 3 3 25 14 — — 42 23 N 1 1 — — 1 1 R 1 1 — — 1 1 9 4 — ,— 10 6 6 5 71 38 — — 96 53 I—V 77 34 .12 5 68 34 64 38 298 142 20 10 539 263 The repetition- of phonemic characteristics V. (Type CCVCC, CCCVC). TR-: trenk-1, 15.78, 9.22, 12.3531; roots 3, different root 1 = TRE- 3, 1. — PS-: pster-, 4.54; root 1, different root 1 = PS-E-1, 1. — SK-: s'kerd-, 4.57, 4.661; roots 2, different root 1 = SKE- 2,1. Sum V: 6 3 Sum I-IV: 509 + 31 255 + 9 VI. Y: yeu-2, 8.21 vb.; ye-, 4.57, 6.63; roots 3, different roots 2 - YU 1, 1, YE 2, 1. YN: yam-, 8.23; root 1, different root 1 = YAN 1, 1. Sum Y: 4 3 W: wei-1, 8.26, 20.25; wei-3, 8.32, 8.332; wl-l, 9.46, 9.47; roots 6, different roots 3 = WI 6, 3. . WL: wel-3, 15.45; wel-4, 9.24; wel-7, 1.442, 9.50; roots 4, different roots 3 = WEL 4, 3. — WK: wek™, 18.41, 18.42; wog«h-nes-, 8.22, 8.23; roots 4, different roots 2 = WEK 4, 2. — WP: wab-, 18.13; wap-2, 18.13; roots 2, different roots 2 = WAP 2, 2. — WN: wem-, 4.57; root 1, different root 1 = WEN 1, 1. — WR: wer-1, 20.27; wer-2, 12.352; wer-3, 4.57, 6.35, 6.36, 9.46, 9.47; wer-7, 18.51; rbots 8, different roots 4 = WER 8, 4. WL-: welk-1, 8.28, 20.26; roots 2, different roots 1 - WEL- 2, 1. — WR-: werg'-2, 8.21 vb.; wreg'-, 9.22; roots 2, different roots 2 - WER-, WRE- 2, 2. -W-: [dwo(u)-] dwei-, 8.23, 8.25, 8.26, 9.24, 9.25; twerk'-, 9.22; roots 6, different roots 2 = TWI 5, 1, TWE 1, 1. — koi-, kwi-, 18.13, gwes-, 9.50; gwet-, 12.3531; k'wek-, 6.91; roots 4, different roots 4 = KWI- 1, 1, KWE- 3, 3. — swer-4, 20.27; swerbh-, swer-, 9.46, 9.47; roots 3, different roots 2 = SWE 3, 2. — skwerb(h)-, 9.46; root 1, different root 1 = SKWE- 1, 1. Sum W: 43 26 Sum I-VI: 556 + 31 284 + 9 155 TABLE VI SHARP Sh°rt LoAnS I U ShE0rt TOTAL I _ _ _ _ 22 32 5 4 II a L — — — — . — — 6 2 _ _ _ _ 6 2 T — — 31 11 5 3 11 52 15 8 K — — — — 8-4 15 7 — — — — 23 11 P — — 21 3 1 9 3 — — — — 14 5 S. — — — — 5 2 1 1 — — — — 6 3 N — — — — 3 1 3 2 — — 4 2 10 5 R— — — — 4 2 11 — — — ..........5 3 — — 5 2 24 11 40 19 1 1 9 4 79 37 Ilbc L 1 1— — — — — — — — 1 1 2 2 • T —————— — 11 4 2— — 5 3 K 18 3 — — 21 11 11— — 22 6 P__ ___ __ i i _ _ __ i l ..S — — ............11 1 1 • ■ 2 2 N — — —— —— — — 1 1— — 11 R 2 1 ————— — 8 4 — — 10 5 21 5 — — 3 2 4 4 14 8 1 1 43 20 III a L 4 2 — — 1 1— — 2 2 2 1 9 6 T — — 5 1 4 2 — — 25 10 — — 34 13 K 21 12 1 1 4 2 — — 110 47 — — 136 62 P 11 3 — — 7 3 1 1 28 18 — — 47 25 S — — ______ -23 3 1 1 24 4 N 7 4 — — — — 4 1 9 6 — — 17 11 R — —————— — 6 3 2 1 8 4 43 21 6 2 16 '8 2 2 203 89 5 3 275 125 Illbc L- - - -----____ __ T — — — — — — — — ____ __ K — — — — — — 33 — — — — 3 3 P — — — — — — 11 ____ i i S — — 1 1 13 5 5 2 2 1 5 2 26 11 N 3 3 — ————— 3 2 — — 6 5 Ril— — — — — — 4 3 — — 5 4 4 4 1 1 13 5 9 6 9 6 5 2 41 24 . IV—V L — — —— —— — — — — — — — — T — — —— —— — — 15 7 — — 15 7 K 2 1 — — 2 2 3 2 18 8 — — 25 13 P — — — — 11— — 11 7 — — 12 8 S 7 3 — — 7 3 3 3 25 14 — — 42 23 N — — —— —— — — 1.1— — 1 1 R — — — — — — — — 1 1 — — 11 9 4 — 10 6 6 5 71 38 — — 96 53 I—V 77 34 12 5 68 34 64 38 298 142 20 10 539 263 TABLE VII ■ ROUGH Sh°rt LoAnš I Ü Sh°rt L™g TOTAL I ___ __ i l 33 — —■ — — 4 4 IIa L— — — — — — 11— — — — 11 T — — 2 1.—— 2 2 — — 2 1 .6 4 K — — — — — _ 4 4 — — — — 4 4 P — — 11— — 32 _ — __ 4 3 S — — — — 4 2 11,— ——— 5 3 N — — — — 1 1 11 _ _ — _ 2 2 r __ ___ 3 211 — — 11 5 4 — — 3 3 8 5 13 12 — — ^ 2 27 21 IIb L. — ————.— — — 21 — — 21 T — — — — _._ __ ____ __ K 6 1 — — — — — — 1 1 — — 7 2 p— _ ______ ____ __ S— ——— __•— ______ __ N — ——'——_—'— li__'l i R 3 2 — — — — — — 6 3 — — 9 5 9 3 — — — — — — 10 6 — — 19 9 III a L 3 3 — — 2 1 1 1 3 3 — — . 9 8 T — — — — 11— _ 19 611 21 8 K 13 9 — — 1 1 — — 63 30 — — 77 40 P 8 3 1 1 1 1 — — 19 12 — — 29 17 S — — l l——__ 14 6 — — 15 7 N 2 2 ————— — 3 3 — — ,5 5 R — — —— —— — — 2 1 3 1 5 2 26 17 2 2 5 4 1 1 123 61 4 2 161 87 IIIbcL— —————— — ____ __ T— — — — — — 2 ' 1 — — — — .2 1 K — — — — — — 11 — — — — 1 1 P — — 11— — 11— — 11 3 3 S — — 11 44— — — — 217 6 N 3 1 — — — — — _ 3 2 — — 6 3 R ——————— — 2 2 — — 2 2 3 1 22 44 4 3 54 32 21 '6 IV—V L — — — — — — — — 1 1 — — 1 1 T 11— — — — — — 3 3 — — 4 4 K 3 3 — — 3 21 1 10 8 — — 17 4 P. ________ 7 4__ 7 4 S — — — — 1 1 3 3 15 8 — — 19 12 N — — — — — — _._ -1 1 _ _ 1 1 R ________ 22 — — 2 2 4 4 — — 4 3 4 4 39 27 — — 51 38 I—V 42 25 7 6 22 17 25 23 177 98 10 6 283 175 REPRES. Short Long T TT Short Long TOTAL SAMPLE A A X U E E I — — — — 13 5 18 6 5 2 — — 36 16 IIa L — — 1 1 5 2 12 2 — — 1 1 19 7 T — — 6 2 9 3 27 8 — — 25 4 67 17 K — — 11 2 31 14 33 16 6 2 4 2 85 36 P — — 15 4 / 10 6 24 9 — — — — 49 19 S — — — — 6 1 12 6 1 1 — — 19 8 N — — 6 3 12 7 10 5 ' — — 14 4 42 19 R — — — — 2 1 4 3 — — — . — 6 4 — — . 39 12 . 75 34 122 50 7 3 44 11 287 110 II be L 10 5 ' — — — — — — 6 6 '— — 16 11 T 1 1 A 1 1 — — — — 5 2 1 1 8 5 K 10 ■ 3 — ■ — 4 3 8 4 14 5 — — 36 15 P 3 2 — ;- 3 2 S 3 1 — — ' 1 1 3 3 6 3 — - 13 8 N 10 4 2 1 — — — — 7 3 — - 19 8 R 15 2 — — — — — — 18 8 — — 33 10 49 16 3 2 5 4 11 7 59 29 1 1 128 59 Ilia L. 6 5 1 1 10 4 19 7 15 5 3 3 54 25 T 2 1 — — 11 3 4 2 86 32 2 1 105 39 K 33 18 3 3 4 3 8 4 181 65 1 1 230 95 P 19 8 1 1 8 5 8 5 110 38 1 1 147 58 S 1 1 7 1 2 2 — — 41 15 1 1 52 20 N 9 5 — — 1 1 4 2 53 23 5 2 72 33 R 2 2 — — 9 (V 1 2 2 17 7 — — 23 12 72 40 12 6 38 20 45 22 503 185 13 9 683 282 III be L 2 ¿i — — — — .— — 1 1 — — 3 3 T — — — — 2 1 2 1 K — — ' — — 4 . 4 4 3 — — 3 2 11 9 P — — — — ■ 3 3 1 1 — — — — 4 4 S — — 26 2 13 10 15 7 . 2 1 2 1 58 21 , N 6 5 5 3 — — 11 8 R 1 1 — — — — — — 3 2 — — 4 3 9 8 26 2 22 18 20 11 11 7 5 3 93 49 IV—V L 5 3 —, — 5 3 T 1 1 '— — 1 1 — — 19 14 — — ?1 16 K 5 3 . — — 3 3 1 1 40 24 — — 49 31 P 5 4 — — — — 3 2 15 7 — — 23 13 S 4 4 — — 9 5 8 5 59 28 — — 80 42' N 2 2 8 6 1 1 11 9 R 1 1 — — 1 9 17 14 — 13 9 12 8 147 83 1 1 190 115 I—V 147 78 80 22 166 90 228 107 732 309 64 25 1417 631 The repetition- of phonemic characteristics TABLE 1 REPRESENTATIVE SAMPLE, ROOTS CVC etc., GROUP III—V TABLE 38 SHARP, POINTED, LOUD, ROOTS CVC etc., GROUP III—V Normalized TABLE 39 SHARP, POINTED, LOUD, ROOTS CVC etc., GROUP III—V Actual P T L R N K K S L 9 , 6 — 1 4 19 19 4 62 T 6 — 14 51 20 19 19 1 130 K 9 14 34 56 24 2 1 5 145 K 10 14 34 56 23 1 2 5 145 P 2 23 43 58 16 13 12 7 174 S 23 63 15 7 28 27 27 —. 190 N 1 12 13 12 29 13 12 2 94 R 5 9 — — 2 5 6 1 28 65 141 153 241 146 99 98 25 968 P T L R N K K S L 9 — — — — 5 7 — 21 T — — 33 49 9 12 12 — 115 K 14 3 39 82 21 7 5 21 192 K 14 2 40 82 21 7 5 21 192 P — 16 28 42 12 16 17 9 140 S 33 26 19 12 33 47 46 — 216 N 3 14 7 5 — 14 14 2 59 R 2 5 — — 5 9 9 3 33 75 66 166 272 101 117 115 56 968 P T L R N K K S L 4 — — — ' — 2 3 — 9 T — — 14 21 4 5 5 — 49 K 6 1 17 35 9 3 2 9 82 K 6 1 17 35 9 3 2 9 82 P — 7 12 18 5 7 7. 4 60 S 14 11 8 . 5 14 20 20 — 92 N 1 6 3 2 — 6 6 1 25 R 1 2 — — 2 4 4 1 14 32 28 71 116 43 50 49 24 413 159 B. Vodusek TABLE 4 REPRESENTATIVE SAMPLE, DIFFERENT ROOTS: CVC etc., GROUP III—V P T L R N K K S L 7 5 — 1 2 7 7 2 31 T 3 — 8 24 6 7 8 1 57 K 4 8 GO 26 10 2 1 3 67 K 5 8 13 26 10 1 1 4 68 P 1 13 19 18 7 6 6 5 75 S 12 00 r-l 11 6 14 11 11 — 83 N 1 7 8 6 10 8 8 2 "50 R 3 5 — — 2 2 3 1 16 36 64 72 107 61 44 45 18 447 P T L R N K K S - TABLE 40 SHARP, POINTED, LOUD, L T K 7 7 2 9 19 17 29 7 14 2 4 5 5 7 4 6 14 44 86 DIFFERENT ROOTS K 4 2 20 29 15 5 ' 2 , 9 86 CVC etc., GROUP S 15 20 11 2 11 13 11 — 83 III—V N 2 9 7 4 _ 6 9 2 39 Normalized , R 2 2 — — ^ 4 5 4 3 20 37 44 86 99 58 49 48 26 447 P T L R N K K S L 3 — — — ■ — i 1 2 — 6 TABLE 41 T — — 4 8 3 2 3 — 20 SHARP, POINTED, K 3 1 9 13 6 2 2 3 39 LOUD, K 2 1 9 13 7 2 1 4 39 DIFFERENT ROOTS CVC etc., GROUP III—V P S 7 4 9 9 5 8 1 3 5 4 6 3 5 3 • 34 38 Actual N 1 4 3 2 — 3 4 1 18 R 1 1 — — 2 2 2 1, 9 17 20 160 39 45 26 22 22 12 203 The repetition- of phonemic characteristics RS Roots SHARP Norm. Act. RS Diff. roots SHARP Norm. Act. L 35 27 8 18 12 4 . T 75 68 20 22 33 11 K 60 76 ■ 22 25 25 8 TABLE 42 K 61 77 23 26 26 9 ROOTS CV, VC, P 52 51 15 21 17 6 GROUP II S 32 27 8 16 15 5 N 61 38 11 27 17 6 R 39 51 15 14 24 8 415 415 122 169 169 57 RS SHARP ROUGH SHARP ROUGH Roots Norm. Norm. Act. ' Act. Short A 147 202 210 77 42 Long A 80 31 35 12 7 TABLE 43 I 166 179 110 68 22 ROOTS, U 228 ~ 168 . 125 64 25 GROUP I—V Short E VOWELS 732 784 887 298 177 Long E 64. 53 50 20 10 1417 1417 1417 539 283 RS ' SHARP ROUGH SHARP ROUGH Diff. roots Norm. Norm. Act. Act. Short A 78 82 90 34 25 TABLE 44 Long A 22 12 ' 22 5 6 DIFFERENT I , 90 82 61 34 17 ROOTS, U 107 91 83 38 23 GROUP I—V Short E 309 340 353 142 98 VOWELS Long E 25 24 22 10 6 631 631 631 263 175 161 B. Voduselc CONCLUSION I 1. In our examination of C. D. Buck's dictionary of synonyms we started from the fact that individual synonymic sets show frequent repetitions of identical or similar phonemes in independent radical morphemes. In the work of the Indo-European scholars this fact which clearly goes beyond, the frame of generally recognized onomatopoetic terms has either been overlooked or — because of the theoretical suppositions to the contrary — has recéived no particular attention, even when observed. Despite the tiresome labour required by such an undertaking, it seemed worth while to proceed to a systematic investigation which might establish whether all this can be due to a broader regularity in the parallel naming of the same reality. By means of a comparative analysis of synonyms and by the application of the statistical method we approached, on a limited corpus of material, the old and yet again and again tackled problem of the intrinsic connection between sound and meaning. The thesis which had to be tested on our statistical material was the assertion put as an anti-motto at the outset of our study [1], 21. There is little need to point out that in emphasizing only the difference of the soundform of synonymic morphemes this assertion coincides with the well known principle of the arbitrariness of the linguistic sign [2], 67. Interpreted in statistical terms it amounts to saying that the phonemic distribution of language at all levels and in all component parts of the linguistic forms is in principle independent of the conceptual content. Hence — if the arbitrariness of the linguistic sign were generally valid — in our case the phonemic distribution of radical morphemes in the synonymic sets selected by specific conceptual criteria should not significantly differ from the normal distribution of radical morphemes appearing in synonymic sets taken at random in the dictionary. But it turned out that the seven broad concept groups of synonymic sets chosen and analyzed show great deviations from the normal phonemic distribution. 2. In the course of our investigation we have limited ourselves to a detailed presentation of the data acquired through the analysis and to their illustration in the tables, leaving the facts to speak for themselves. For a definitive, statistically valid conclusion on the overall result of the inquiries as well as on the strength of the individual data this, however, is not sufficient. This is the reason why, in accordance with the requirements of the statistical theory., we have concluded our investigation with four probability calculations, namely: a) the calculation of Chi-square for the tested seven concept groups, in every field — i. e. in the field of discrete consonants, of consonant pairs, and of vowels — in order to find out whether the phonemic distributions of these concept groups, taken as a.whole, significantly differ from the respective normal phonemic distribution. 162 The repetition- of phonemic characteristics b) the calculation of Chi-square for thirty groups of synonymic sets set up at random from our population, in all the fields, for the purpose of establishing whether the normal phonemic distribution, taken as a whole, is stable. c) the calculation of the standard deviations for the individual frequencies of the same at random set up groups of synonymic sets, in order to determine to what degree the individual frequencies in these conceptually unrelated groups are stable. d) the probability calculation using the standard deviations obtained under c), to establish to what degree the individual frequencies of the tested seven concept groups significantly differ from the coresponding frequencies of the respective normal distribution.* ■ 3. The probability inferences are as follows: ad a) The differences between the consonant distributions of the seven concept groups and the normal consonant distribution are in all instances statistically significant. The same is true also of the distribution of consonant pairs, with the only exception of the concept group TOUCH. The differences between the vowel distributions of the concept groups and the normal vowel distribution are statistically significant with the concept groups BLOW, SMELL, SHARP and ROUGH. The probability that these differences might be differences of pure chance nowhere exceeds P = 0.001, this being a appropriately fixed limit for the evaluation of the essential differences of the total distributions [3], 2, 47. ad b) The differences between the phonemic distributions of the thirty at random selected groups and the normal phonemic distribution are in no instance, in no field, statistically significant. The probability of chance is in every case far beyond the fixed limit of 0.001. ad c) The deviations of the frequencies for the individual phonemic elements of the thirty at random formed groups from the frequencies to be expected from the normal distribution all the time move — with a single relevant exception for the consonant pair KR, and four negligible exceptions for LT, PP, NP, RS — within the interval of three standard deviations. The probability of chance is — besides the one case with KR — everywhere far greater than P = 0.0027 which corresponds with phonemic elements with greater relative frequency in the population to three standard deviations, and which is for the evaluation of the individual deviations an appropriately fixed limit [4], 95, 314. ad d) The deviations of the frequencies for the individual phonemic elements of the seven concept groups from the frequencies to be expected from the normal distribution exceed three standard deviations in a number of instances. When taking into account also the relative frequency for the individual phonemic elements in the population, we find with the same * My sincere thanks for the kind' help in the computing operations are due to Mr. E. Zakrajsek, Head, and to Mr. J. Kozak, programmer, The Computing Centre at the Institute of Mathematics, Physics and Mechanics, University of Ljubljana. 163 B. Voduselc phonemic elements as under c) 19 instances of doubtlessly significant deviations in the positive direction and to these correspond 18 instances of doubtlessly significant deviations in the negative direction. Above all in the field of consonants we find positive significant deviations with all the seven consonant classes of high relative frequency: K, R, N, S, L, T, and P. Further ten positive significant deviations are found with the consonant pairs, each of which represents more than 3 per cent of the population [5], 292, specifically with KR, KN, SN (twice), PL (twice), PR, SK, TR, and ST. In the field of vowels we find with the same phonemic elements as under c) two occurrencies of positive significant deviations, both times with the class U. The eighteen negative significant deviations are distributed as follows: in the field of consonants they occur with K (three times), R, N (twice), and T (twice) — altogether eight times; in the field of consonant pairs they occur with NN, LK (twice),'SK (twice), ST, and TR — altogether seven times; in the field of vowels they occur with I, U, and long A altogether three times. Here it should be added that owing to a later combining or splitting up of the individual phonemic elements another 12 significances have been established; six in the field of consonant pairs with the joined pairs SL + NL + TL (twice), TP + PT, PS + SP, KT + TK, and TR + PR, and six in the field of vowels and consonant-vowel combinations, with short A + short E (twice), with long A + long E (negative significance), with K — short A (twice), and with P-U. The probability that all these might be deviations of pure chance nowhere exceeds the fixed limit od 0.0027.* 4. On the basis of the obtained results it is possible to maintain that' the tested thesis about the expression of the same concepts'with consistently various phonemes does not apply to the statistical material of our seven concept groups. Obviously there exists a connection between the significant deviations from the normal phonemic distribution, taken as a whole, and in particulars, and the fact that the synonymic sets of our concept groups have been selected according to specific conceptual criteria. 5. The existence of this connection is clearly demonstrated by our additional tests, in so far the normal phonemic distribution in the thirty * The great difference concerning the distributions under c) and under d) is evident also from the following data: In the 30 groups set up at random we have with the phonemic characteristics representing 3 per cent and more of the population altogether 24 instances of deviations which exceed two standard deviations; once with K, R, L, P, and twice with S and T; once with KN, PL, TR, and twice with KL, SK, ST, LK, TK; once with short A, U, and E. In the seven concept groups, in spite of the- four times smaller number of instances, 31 similar deviations are found; R, S, short A, and -U in the concept group SHARP; -L, PR, TR, -ST, -long A,--I, and E in the concept group ROUGH; -KL, NN, and -long A in the concept group. NARROW; N, -SK, -long A, and -I in the concept group SMELL; -KR, -KN, -NN, -TK, -long A, and E in the concept group SMOOTH; -N, -L, -KL, -NN, and ,-U in the concept group TOUCH; -KL, -NN in the concept group BLOW. 164 The repetition- of phonemic characteristics b) the calculation of Chi-square for thirty groups of synonymic sets set up at random from our population, in all the fields, for the purpose of establishing whether the normal phonemic distribution, taken as a whole, is stable. c) the calculation of the standard deviations for the individual frequencies of the same at random set up groups of synonymic sets, in order to determine to what degree the individual frequencies in these conceptually unrelated groups are stable. d) the probability calculation using the standard deviations obtained under c), to establish to what degree the individual frequencies of the tested seven concept groups significantly differ from the coresponding frequencies of the respective normal distribution.* - 3. The probability inferences are as follows: ad a) The differences between the consonant distributions of the seven concept groups and the normal consonant distribution are in all instances statistically significant. The same is true also of the distribution of consonant pairs, with the only exception of the concept group TOUCH. The differences between the vowel distributions of the concept groups and the normal vowel distribution are statistically significant with the concept groups BLOW, SMELL, SHARP and ROUGH. The probability that these differences might be differences of pure chance nowhere exceeds P = 0.001, this being a appropriately fixed limit for the evaluation of the essential differences of the total distributions [3], 2, 47. ad b) The differences between the phonemic distributions of the thirty at random selected groups and the normal phonemic distribution are in no instance, in no field, statistically significant. The probability of chance is in every case far beyond the fixed limit of 0.001. ad c) The deviations of the frequencies for the individual phonemic elements of the thirty at random formed groups from the frequencies to be expected from the normal distribution all the time move — with a single relevant exception for the consonant pair KR, and four negligible exceptions for LT, PP, NP, RS — within the interval of three standard deviations. The probability of chance is — besides the one case with KR — everywhere far greater than P = 0.0027 which corresponds with phonemic elements with greater relative frequency in the population to three standard deviations, and which is for the evaluation of the individual deviations an appropriately fixed limit [4], 95, 314. ad d) The deviations of the frequencies for the individual phonemic elements of the seven concept groups from the frequencies to be expected from the normal distribution exceed three standard deviations in a number of instances. When taking into account also the relative frequency for the individual phonemic elements in the population, we find with the same * My sincere thanks for the kind' help in the computing operations are due to Mr. E. ZakrajSek, Head, and to Mr. J. Kozak, programmer, The Computing Centre at the Institute of Mathematics, Physics and Mechanics, University of Ljubljana. 163 B. Voduselc phonemic elements as under c) 19 instances of doubtlessly significant deviations in the positive direction and to these correspond 18 instances of doubtlessly significant deviations in the negative direction. Above all in the field of consonants we find positive significant deviations with all the seven consonant classes of high relative frequency: K, R, N, S, L, T, and P. Further ten positive significant deviations are found with the consonant pairs, each of which represents more than 3 per cent of the population [5], 292, specifically with KR, KN, SN (twice), PL (twice), PR, SK, TR, and ST. In the field of vowels we find with the same phonemic elements as under c) two occurrencies of positive significant deviations, both times with the class U. The eighteen negative significant deviations are distributed as follows: in the field of consonants they occur with K (three times), R, N (twice), and T (twice) — altogether eight times; in the field of consonant pairs they occur with NN, LK (twice), SK (twice), ST, and TR — altogether seven times; in the field of vowels they occur with I, U, and long A altogether three times. Here it should be added that owing to a later combining or splitting up of the individual phonemic elements another 12 significances have been established; six in the field of consonant pairs with the joined pairs SL + NL + TL. (twice), TP + PT, PS + SP, KT + TK, and TR + PR, and six in the field of vowels and consonant-vowel combinations, with short A + short E (twice), with long A + long E (negative significance), with K — short A (twice), and with P-U. The probability that all these might be deviations of pure chance nowhere exceeds the fixed limit od 0.0027.* 4. On the basis of the obtained results it is possible to maintain that the tested thesis about the expression of the same concepts'with consistently various phonemes does not apply to the statistical material of our seven concept groups. Obviously there exists a connection between the significant deviations from the normal phonemic distribution, taken as a whole, and in particulars, and the fact that the synonymic sets of our concept groups have been selected according to specific conceptual criteria. .5. The existence of this connection is clearly demonstrated by our additional tests, in so far the normal phonemic distribution in the thirty * The great difference concerning the distributions under c) and under d) is evident also from the following data: In the 30 groups set up at random we have with the phonemic characteristics representing 3 per cent and more of the population altogether 24 instances of deviations which exceed two standard deviations; once with K, R, L, P, and twice with S and T; once with KN, PL, TR, and twice with KL, SK, ST, LK, TK; once with short A, U, and E. In the seven concept groups, in spite of the- four times smaller number, of instances, 31 similar deviations are found; R, S, short A, and -U in the concept group SHARP; -L, PR, TR, -ST, -long A, -i, and E in the concept group ROUGH; -KL, NN, and -long A in the concept group. NARROW; N, -SK, -long A, and -I in the concept group SMELL; -KR, -KN, -NN, -TK, -long A, and E in the concept group SMOOTH; -N, -L, -KL, -NN, and ,-U in the concept group TOUCH; -KL, -NN in the concept group BLOW. 164 The repetition- of phonemic characteristics at random formed groups of synonymic sets nowhere essentially varies taken as a whole and shows almost no essential deviations in particulars. The statistical experiments under b) and c) confirm' that our population of independent roots from Buck's synonymic sets behaves in this respect in the same way as the population of an alphabetically arranged dictionary. In fact, if we take sufficiently large parts from any alphabetically arranged dictionary, chosen at random, we shall see that the phonemic distribution will not essentially vary from one part to another or from the average phonemic distribution of the dictionary [4], 67, On the other hand, essential differences will immediately come up, when we compare the phonemic distributions of consecutively running word groups, chosen exclusively from one initial letter, with those chosen exclusively from another letter: the frequencies of the initial phonemes are in such a case throughout preferred. In our case we are dealing not with an alphabetical but with an onomato-logical dictionary. Admittedly the dictionary is limited to the etymological projections of the root part of the various terms and is also reduced by the exclusion of the recurrent etymologically identical roots with each individual concept, but nevertheless it preserves the typical statistical characteristics of a dictionary with its stable phonemic distribution. A distribution, esentially different from the normal, were to be expected only on condition that the parts taken from, the dictionary are chosen at one time exclusively under one concept group and at another time exclusively under another, and further, that there are specific phonemic characteristics preferred in different concept groups. The results of our investigation prove that such preferences in fact exist. Instances of preference may be seen already in the fact that in small synonymic sets certain phonemic characteristics prevail to the exclusion of others. Frequent instances of preference are found in the recurrence of certain phonemic characteristics where with regard to the normal phonemic distribution these recurrences should not turn up. And finally we get preference through excessive recurrences of certain phonemic characteristics with big synonymic sets, where with regard to the normal phonemic distribution only a smaller number of such recurrences would be justified. The sum of the preferences, of which each for itself is too small for reliable evaluation, gives significant differences, on the whole and with individual phonemic elements, in our seven concept groups. Conversely, in groups of synonymic sets formed at.random the single preferences are mutually eliminating, and the phonemic distributions remain essentially identical with the normal phonemic distribution. 6. The fact that it was possible to detect so numerous phonemic characteristics preferred in definite concept groups, i. e. that it ,was possible to establish systematic morpho-semantic correlations, is based of course on the application of some fundamental methods. First of all the Indo-European phonemic system had to be simplified. The simplification, as it was explained at an earlier stage, was necessitated by larger phonemic classes so that the statistical material should not get too 165 B. Voduselc scattered. Here we must emphasize that the linguistic phenomenon subjected to our inquiry consists not only of the recurrences of the same phonemes and phonemic combinations under definite circumstances but to an even larger degree of the recurrences of similar phonemes and phonemic combinations. If the problem was to be approached in total, a classification into types was indispensable. The inductive classification of phonemes as resulting directly from our statistical material happens in respect of the principal consonantal characteristics to be in an almost consistent agreement with the traditional classification of consonants, i. e. the classification into plosives, fricatives, liquids and nasals according to the manner of articulation, and into labials, dentals, and gutturals according to the zone of articulation. The only difference from the traditional classification is in the separation of the two liquids considered as being two independent phonemic characteristics. This procedure, at variance with the hypothesis that the Indo-European parent tongue had originally only one liquid [6], was fully required by our statistical material. Not only that it was not possible to establish any broader concept group in which both liquids would appear in a joint morpho-semantic correlation but the initial perusal of Buck's dictionary already disclosed that each of' the two phonemes was in an independent correlation with a highly different concept group. The statistical tests carried out confirmed our inductive conclusions and even revealed an unusual coherence of the simplified consonant system. Prom the analysis of the recurrent R in the concept group ROUGH, for instance, we can see that the significance of R is primarily related to the occurrences of the radical morphemes in which R appears with any of the guttural plosives; and from the analysis of the recurrent L in the concept group SMOOTH we can see that the significance of L is at least in part dependent on the supernumerary occurrences of the radical morphemes with an additional labial plosive. If we take the guttural and labial plosives as phonemic types, we get in fact another two significant phonemic characteristics: KR, and PL. Similarly, the recurrent Significant S in the concept group SMELL is related to the supernumerary occurrences of radical morphemes containing either an additional M or an additional N, and the two nasals treated as one single phonemic type give a further significant phonemic characteristic: SN. Three of the four consonant types: K, P, T, and N, which appear as independent phonemic characteristics in the concept groups SHARP, BLOW, TOUCH, and NARROW, appear thus in significant combinations with individual consonants in the first mentioned concept groups ROUGH, SMOOTH, and SMELL. The Whole of the statistical results permits the conclusion that the morpho-semantic correlations of the consonant types are to be attributed to the generic, and hence the more important, auditory-articulatory features of these consonantal categories. Where, on the other hand, the morpho-semantic correlations are related to individual consonants, the individual auditory-articulatory features of these consonants are to be considered equally important as the generic features of the consonant types. 166 The repetition- of phonemic characteristics 7. The establishment of types in the Indo-European vowel system was as well done inductively, but within the frame of the concept groups for which consonantal morpho-semantic correlation had been found. Prom the overall impression of the recurrences of similar vowel elements in these concept groups it seemed that the most appropriate thing to do was to make a statistical verification of the heuristic value of the traditionally assumed five-member vowel system a-e-i-o-u, modified, by joining e and o, into a four-member system. This modification was suggested by the absence of any sign that would point towards a morphosemantic correlation in connection with o, an absence that was probably correctly explained by the fact that owing to the qualitative vowel alternation or some other reason [7], 38, 106, 392, at least the large majority of roots with o sounds, either short or long ones, contained originally an e sound. In the simplified system of the four basic, qualitatively different vowel elements no distinction was made between short and long elements, regarding shortness or length as genetically secondary distinctive features. 8. A special procedure was used in the determination of the vowel classes U and I. These classes came to include not only the independent unreduced u and i root sounds [8] but also the falling u and i diphthongs in radical morphemes, irrespective of the quality of the various initial vowels. Already from the methodological point of view such a joining could hardly be avoided: the root diphthongs had to be taken as a uniform phonemic characteristic because of the statistical comparison with simple root vowels which are counted only once. Also it was not possible to count and discuss separately the u and i vowel sounds on the one hand and the u and i semi-vowel sounds on the other, since in both cases we are concerned with phonologically the same phoneme [7], 393. The decisive factor in the establishment of types was again our statistical material. After a statistical verification in two concept groups, BLOW and SMELL, the vowel class U, established in this way, turned out' to be doubtlessly significant. The parallel class I gave indeed in our statistical material no positive significant result, but the negative significance of that class in the concept group BLOW indirectly proved its real existence. It follows that with the falling diphthongs the initial" vowel subject to quantitative vowel alternation is — at least in the mentioned concept groups — semantically subordinated to the vowel element u and i. The decisive semantic significance of the second vowel elements in these diphthongs agrees with the fact that these elements are not affected by reduction, or rather are even strengthened by it. In their stability they are like the consonantal phonemic characteristics that represent the chief semantic value of the Indo-European radical morphemes [9], 152, [7], 392. Accordingly, we may, irrespective of the hypothesis about the monophthongal origin of the falling diphthongs in imitative words [10], 288, attribute the statistical significance of the U and I classes to the prominent auditory-articulatory features of the u and i sounds which prevail over the respective features of the accompanying initial vowels* . 167 B. Voduselc 9. Important as. are the obtained results concerning the vowel classes U and I, within the frame of the four-vowel system they remained solitary. Only the vowel distribution of the concept group ROUGH seemed to indicate, if taken as a whole, another essential deviation from the normal vowel distribution — but this owing to the extremely low-frequency of the class U and to a slightly less low frequency of the class I and not to any positive significance. One of the reasons for the small number of the established morpho-sematic correlations of vowels was obviously the fact that these correlations were being established in addition to those of the consonants — but the small number could have been caused also by inadequate establishing of the vowel types. To get this clear at the end of our investigation we split up the vowel class A into the class of short A sounds and the class of long A sounds. This, was done with particular respect to the vowel distribution of the concept group SHARP in which short A sounds were markedly recurrent. Besides, we limited ourselves in distinguishing between short and long vowels to the class A-, because particularly here it might be possible to. find an original difference in the quality between the two categories [7.187]. With such a classification of vowels into five basic elements there appeared an essentially different vowel distribution, in contrast to the normal, not only with the concept group ROUGH but also in the concept group SHARP. Nevertheless neither of A classes reached the limit of significance. Only in distinguishing between short and long vowels also with the class E, a close parallelism showed up between the split up classes of A and E. Taking long A sounds and long E sounds together on the one hand, and short A sounds and short E sounds 6n the other, there came to light a clear significance of the joined short vowels in both concept groups SHARP and ROUGH. Conversely, with an opposite methodical approach, separating the various consonantal combinations with a short A sound, we succeeded in discovering the significance of the consonant-vowel combination K — short A. 10. In this way the finally established classification of the Indo-European vowel system consisting of six basic elements, short A — long A — I — U — short E — long E, capable of being joined together or further divided, now according to one of their features now according to another, came to be the most appropriate one for the determining of vowel morpho-semantic correlations- in our statistical material. In the case of the significance of the joined short vowels we are concerned apparently with a morpho-semantically relevant opposition to long vowels, which in the concept group SHARP happen even to be negatively significant: this is to say that we are concerned with the positive significance of the closed syllable in opposition to the negative significance or rather non-significance of the open syllable, the shortness or * Since in the rising root diphthongs with the initial semi-vowels u and i there is no reduction in favour of these initial vowel elements, the rising diphthongs were not included into the classes U and I, and the question of their classification remained open. In view of the comparatively small number of such instances this could be done without affecting the validity of the statistical results. 168 The repetition- of phonemic characteristics length of vowels in the Indo-European monosyllabic roots relying in principle on the central, or rather the final, place in the syllable [11]. The auditory-articulatory features which are the sole decisive factor for the morpho-seman-tic correlations are in this case related only to the syntagmatic structure, i. e. the closedness and openness, of the radical morphemes.* The significance in our statistical material of the consonant-vowel combination K — short A with regard to the specific quality of the short A sound is another somewhat surprising finding since the existence of an original Indo-European A is also for the short A stil controversial. But even if in the statistical calculation we include only the roots containing in Pokorny's dictionary a consistently marked short A, and exclude all cases where A migt possibly be long, the significance of the K — short A combination with. the concept group SHARP is beyond doubt. At this point it must be added that in the normal phonemic distribution almost half of the doubtlessly short A sounds appear in combination with K, while in the same distribution only a quarter of long A sounds appears in such a combination. Finally, in the combination K — short A the guttural is nowhere labialized, this occurs only in the combination K — long A. The sole exception to the rule is found in the radical morpheme (s)lagw.. All this points to the articulatory uniformity od the non-labialized K — short A, similar to the articulatory uniformity of P-U in the concept group BLOW, and to the real existence of the short A sound in the Indo-European mother tongue. The significance of the combination K — short A is in all probability to be attributed to the common guttural auditory-articulatory features of K and short A. But our results give no explanation to the problem whether the class of long A in fact represents an independent vowel quality that would be distinct not only from the vowel quality of short A but also from that of E. It is only the establishment of a particular morpho-sematic correlation appearing in some positive significance of the class of long A that could conduce to the elucidation of this question; until such a'correlation has been established, a distinct treatment of the class of long- A is methodologically required. 11. Along with the establishing of the phonemic types as common denominators in the recurrences of similar phonemic characteristics in the synonymic sets we proceeded with the conceptual categorization of these sets, The categorization was called for by the statistical method, but in order to carry it out successfully suitable theoretical criteria were required. The working hypothesis, which included these criteria, was in the course of this study already briefly explained; but since it represents the main methodological approach in our investigation, we shall try now to explain it in some detail and show how ts empirical and theoretical foundations agree. * If the classes U and I are divided into occurences in closed and in open syllables, and the occurrences in closed syllables are added to short A and E, and occurrences in open syllables to long A and E, we get both with ROUGH and with SHARP positive, significances in the first case and ne;p.tive significances in the second. 169 B. Voduselc ■ The fundamental assumption in the working hypothesis was that the recurrences of phonemic characteristics under consideration have their origin in the imitative representation of reality. One indeed cannot imagine any other acceptable reason for these systematic. recurrences. Theoretically speaking, only two modes of representation of the reality are possible: the imitative and the symbolic [4], 168, or rather, since the second term is not strictly unambiguous, the imitative and the non-imitative, or the iconical and the non-iconical [12], The thesis about the arbitrary nature of the linguistic sign resides in the proclamation of the absolute validity of the second mode in language; the concretely established not-accidental recurrences of .phonemic characteristics in our synonymic sets point, on the contrary, to the first alternative. 12. But what in fact is imitated in the radical morphemes, and what is the mechanism of imitation like? Prom bur statistical material it seems to be evident that the established significances cannot be accounted for by onomatopoeias in the traditional sense of the word, i. e. by reproductions of acoustical phenomena. Unusual recurrences of phonemic characteristics occur frequently with synonymic sets where the denominated phenomena are not acoustically manifesting themselves or their acoustical manifestation is clearly subordinated, e. g: it is only facultative, less intense or less important with regard to other forms of manifestation. This" applies to the majority of the synonymic sets in the concept groups SMOOTH, ROUGH, NARROW, TOUCH, and SHARP, as well as for numerous sets in the groups BLOW and SMELL. It is therefore a necessary conclusion that the significance in all these cases in caused by a specific imitation of non-acoustic phenomena. The idea that such imitations in fact exist, is — as has been seen from Plato's quotation — a very old one; it has been alive through the entire history of the European culture and was again emphasized by the romantic school of the 18th and 19th centuries. At the beginning of this century it was the well known psychologist Wundt [13], 323, who tried to formulate it in scientific terms. But the repeated studies of imitation in language to be seen in the copious bibliography [10], [12] have not led to a uniform view of sholars and to a clear demarcation between the not acoustically conditioned imitation and the onomatopoeia as it is traditionally understood. The uniform and overlapping use of, terms like sound gestures, kinetic onomatopoeia, sound pictures, sound-symbolic expressions, expressive words, clearly reflects this state. And although the recent experimental and statistical studies on artificial and natural linguistic material [14], prior to the present investigation, more than once succeeded in. proving the objective existence of systematic non-acoustic correspondence between sound and meaning, the basic physiological and psychological mechanism, the development, and the historical origin of this correspondence have remained unclear. 13. The first reason for this is to be sought in the insufficient knowledge about the physiological and psychological conditions of man's speech in 170 The repetition- of phonemic characteristics length of vowels in the Indo-European monosyllabic roots relying in principle on the central, or rather the final, place in the syllable [11]. The auditory-articulatory features which are the sole decisive factor for the morpho-seman-tic correlations are in this case related only to the syntagmatic structure, i. e. the closedness and openness, of the radical morphemes.* The significance in our statistical material of the consonant-vowel combination K — short A with regard to the specific quality of the short A sound is another somewhat surprising finding since the existence of an original Indo-European A is also for the short A stil controversial. But even if in the statistical calculation we include only the roots containing in Pokorny's dictionary a consistently marked short A, and exclude all cases where A migt possibly be long, the significance of the K — short A combination with the concept group SHARP is beyond doubt. At this point it must be added that in the normal phonemic distribution almost half of the doubtlessly short A sounds appear in combination with K, while in the same distribution only a quarter of long A sounds appears in such a combination. Finally, in the combination K — short A the guttural is nowhere labialized, this occurs only in the combination K — long A. The sole exception to the rule is found in the radical morpheme (s)lagw-. All this points to the articulatory uniformity od the non-labialized K — short A, similar to the articulatory uniformity of P-U in the concept group BLOW, 'and to the real existence of the short A sound in the Indo-European mother tongue. The significance of the combination K — short A is in all probability to be attributed to the common guttural auditory-articulatory features of K and short A. But our results give no explanation to the problem whether the class of long A in fact represents ah independent vowel quality that would be distinct not only from the vowel quality of short A but also from that of E. It is only the establishment of a particular morpho-sematic correlation appearing in some positive significance of the class of long A that could conduce to the elucidation of this question; until such a" correlation has been established, a distinct treatment of the class of long A is methodologically required. 11. Along with the establishing of the phonemic types as common denominators in the recurrences of similar phonemic characteristics in the synonymic sets we proceeded with the conceptual categorization of these sets, The categorization was called for by the statistical method, but in order to carry it out successfully suitable theoretical criteria were required. The working hypothesis, which included these criteria, was in the course of this study already briefly explained; but since it represents the main methodological approach in our investigation, we shall try now to explain it in some detail and show how ts empirical and theoretical foundations agree. * If the classes. U and I are divided into occurences in closed and in open syllables, and the occurrences in closed syllables are added to short A and E, and occurrences in open syllables to long A and E, we get both with ROUGH and with SHARP positive significances in the first case and ne^p.tive significances in the second. 169 B. Voduselc ■ The fundamental assumption in the working hypothesis was that the recurrences of phonemic characteristics under consideration have their origin in the imitative representation of reality. One indeed cannot imagine any other. acceptable reason for these systematic. recurrences. Theoretically speaking, only two modes of representation of the reality are possible: the imitative and the symbolic [4], 168, or rather, since the second term is not strictly unambiguous, the imitative and the non-imitative, or the iconical and the non-iconical [12], The thesis about the arbitrary nature of the linguistic sign resides in the proclamation of the absolute validity of the second mode in language; the concretely established not-accidental recurrences of .phonemic characteristics in our synonymic sets point, on the contrary, to the first alternative. 12. But what in fact is imitated in the radical morphemes, and what is the mechanism of imitation like? Prom our statistical material it seems to be evident that the established significances cannot be accounted for by onomatopoeias in the traditional sense of the word, i. e. by reproductions of acoustical phenomena. Unusual recurrences of phonemic characteristics occur frequently with synonymic sets where the denominated phenomena are not acoustically manifesting themselves or their acoustical manifestation is \ clearly subordinated, e. g; it is only facultative, less intense or less important with regard to other forms of manifestation. This'applies to the majority of the synonymic sets in the concept groups SMOOTH, ROUGH, NARROW, TOUCH, and SHARP, as well as for numerous sets in the groups BLOW and SMELL, It is therefore a necessary conclusion that the significance in all these cases in caused by a specific imitation of non-acoustic phenomena. The idea that such imitations in fact exist, is — as has been seen from Plato's quotation — a very old one; it has been alive through the entire history of the European culture and was again emphasized by the romantic school of the 18th and 19th centuries. At the beginning of this century it was the well known psychologist Wundt [13], 323, who tried to formulate it in scientific terms. But the repeated studies of imitation in language to be seen in the copious bibliography [10], [12] have not led to a uniform view of sholars and to a clear demarcation between the not acoustically conditioned imitation and the onomatopoeia as it is traditionally understood. The uniform and overlapping use of. terms like sound gestures, kinetic onomatopoeia, sound pictures, sound-symbolic expressions, expressive words, clearly reflects this state. And although the recent experimental and statistical studies on artificial and natural linguistic material [14], prior to the present investigation, more than once succeeded in. proving the objective existence of systematic non-acoustic correspondence between sound and meaning, the basic physiological and psychological mechanism, the development, and the historical origin of this correspondence have remained unclear. 13. The first reason for this is to be sought in the insufficient knowledge about the physiological and psychological conditions of man's speech in 170 The repetition- of phonemic characteristics general; the beginnings of a universally conceived examination of these conditions come only from the second half of this century [15], [16]. In the elucidation of the problem outlined one must needs start from the physiological and psychological data concerning man's speech; until these have been adequately clarified, no definite explanation of any linguistic imitation can be expected. It should be pointed out that the question of the so-called acoustic imitation can no more be regarded as settled; this is demonstrated by the widely known fact that the imitations of this kind vary from one language to another — a fact which the protagonists of the consistently arbitrary nature of the linguistic sign have frequently used to negate any systematic imitation in speech [17]. 14. As now it is certainly true that the examination of the relation between the linguistic sign and the phenomenon to which the sign refers is normally not the linguist's concern [12], this assertion does not apply to the real or presumable imitative terms. A consistent "scientific study of the imitative terms would in addition to a necessary analysis inside the particular linguistic system require a special inter-disciplinary method for the comparison of these terms and the corresponding phenomena, the acoustical-auditory comparison on the auditory, level and the physically-physiological one on the other sensory levels which are involved in the motor-sensory mechanism of man's speech. In this way only — with positive and negative findings — it would be possible to establish definitively the structure of the respective concepts with all the elements of sensory perceptions contained in them and to determine the intermediate link between the phenomena and the corresponding imitative terms. The fact that such investigations — with the exception of some experiments carried out by H. Wissemann [18] concerning the onomatopoeias — have not yet been undertaken is the second and the main reason for the ignorance of the physiological and psychological mechanism of the linguistic imitations. A direct consequence of the lack of such knowledge is the disagreement in view of the role of- the imitative terms in the origin of language. If e. g. a renowned linguist regards the problem of the origin of language as insoluble and for that reason as something outside the scope of the linguistic science, his view is conditioned by a somewhat hasty judgment that the subjective work of a single, be it acute, observer in comparing imitative terms and the acoustic and other elements of the imitated phenomena [19] already suffices for a definitive explanation of the imitative mode in language [20], 19, 270. 15. In these circumstances a working hypothesis was necessary to explain the concept groups suggested by the unusual repetitions of phonemic characteristics' in definite synonymic sets. The basis of this hypothesis are two obvious facts: 1) The linguistic imitation consists in the reproduction of extra-linguistic perceptions by means of the organs of speech on the part of the speaker. 2) Speaking is performed through a motor process in the organs of speech, controlled on 171 B. Voduselc the one hand by an auditory feedback from the ear, and on the other by a tactile and proprioceptive feedback from the organs of speech themselves. It follows that trying to reproduce the extra-linguistic perception the speaker controls this reproduction by the multi-sensory feedback, necessarily acompanying every speech activity. And there exists the possibility that in the process of reproduction and feedback the speaker reproduces and controls anyone of the extra-linguistic perceptions which are of the same nature as those generally taking part in speaking, i. e. not only auditory but also tactile and proprioceptive perceptions. In agreement with this possibility we assume that linguistic imitations of tactile and proprioceptive perceptions occurring wherever in the body really exist. We assume further that the understanding and the acceptation of these imitations on the part of the listener takes place after an active repeating of the particular imitation on his part and experiencing the same process of reproduction and control as experienced by the speaker. Finally, we assume that the identification of tactile and proprioceptive perceptions coming from any part of the body and those arising through speaking in the speech organs in many cases is facilitated by synergism and by sensations involving the entire human organism, taking place in various physiological processes and voluntary motor activities. These identifications are not conceived as conscious: in man's conscious-' ness the phenomenon producing the extra-linguistic perceptions is identified with the sound carrying their linguistic reproductions. 16. Taking into account that the reproduced extra-linguistic sensory perceptions may occur either separately or in combination, we obtain seven possible types of imitated phenomena listed in the table below. Disregarding the difference between the tactile and proprioceptive perceptions and taking them together as non-auditory perceptions, the number of the possible types is reduced to three. In the table, AUD stands for auditory perceptions, , NON-AUD for non-auditory perceptions, TACT for tactile perceptions, and PROP for proprioceptive perceptions. TABLE IX: The basic possible types of imitated phenomena producing: AUD TACT PROP 1 AUD TACT 2 AUD PROP 3 TACT PROP 4 . TACT 5 PROP 6 ATJD 7 AUD — NON-AUD I NON-AUD II AUD III The table shows what type of imitation may explain the unclear significances of the phonemic characteristics in our concept groups. Where the synonymic sets refer to real phenomena that are not acoustically manifesting themselves, it seems plausible to explain the unusual repetition of the phone- 172 The repetition of phonemic characteristics i mic characteristics as an imitation of sensory perceptions listed in the table under II 4, 5, or 6. Wherever the real phenomena expressed in the synonymic sets get at least in part acoustically manifested, the unusual repetition of • the phonemic characteristics may be accounted for by the imitations under I 1, 2, or 3 in our table. In the forming of the synonymic, sets, however, the imitative type 7 or rather III has been excluded for methodological reasons. Our guiding criterion in the forming of the concept groups was to collect in the concept groups only those synonymic sets which express phenonema producing normally tactile or proprioceptive perceptions or both, and to exclude the synonymic sets which refer to other phenomena. In other words: we have limited ourselves to the linguistic expression of that part of reality only which man directly experiences either through his body or on it. 17. Such a decision seemed inductively appropriate from the point of view of our statistical material; otherwise we just would not have known how to form both sufficiently large arid sufficiently homogeneous concept groups covering sets with unusual repetitions of phonemic characteristics; and theoretically we relied here on the great significance of the simultaneous experience of one's own body and of the external reality [21], [22], In this connection it might not be superfluous to note that even the study of modern poetic synaesthetic metaphors reveals that the most frequent source of these metaphors is to be found in tactile, sensory perceptions, irrespective of the single authors [23], 282. And the decisive role of the sense of touch and of the proprioceptive sense in the human organism, which two senses condition all of man's physical operations, has been recently emphasized by the cybernetic science [24], Just as we excluded from our concept groups all the sets with terms for phenomena which are accessible to man only by the auditive way we excluded also the sets with terms for phenomena accessible to man only optically. Since the visual perceptions in the, auto-perception of the speech under normal conditions do not occur, there is a lack of parallelism. This lack does not permit the identification of extra linguistic and linguistic sensory perceptions, and this applies also to all other sensory perceptions which are not considered in our model. All these perceptions may have an indirect share in the linguistic imitation, to the extent to which they are in a local, synergetic or synaesthetic connection with auditory, tactile, or proprioceptive perceptions. 18. Before trying to explain the individual established morpho-semantic correlations, we must tackle the question with what sound material the linguistic imitations under consideration have been performed: with linguistic sounds as they are known or reconstructed, or with pre-linguistic sound elements which were only at a later stage changed into linguistic sounds? Are these imitations primary or secondary, and to what period of the language do they belong? Once more it was pointed out under 5. that the statistical population from' which our results are derived represents an onomatological dictionary, 173 B. Voduselc reduced to genealogically independent Indo-European radical morphemes in each synonymic set. This is of course not a dictionary referring to some pre-lndo-European state where the language morphemes would appear in their pure radical shape. This is a specially worked out onomatological dictionary of the main Indo-European languages; their independent linguistic systems are here taken in consideration with regard to the main periods of their development. Only because all these linguistic systems go back to the same original mother tongue it was possible to reduce the semantically decisive parts of the synonyms, i. e. the roots, to their their supposed shape in the mother tongue, and in this way we obtained the statistical population of the radical morphemes expressed in the same phonemic system and amenable to a statistical investigation. But although the roots of the synonyms go back to the period of the Indo-European community; the synonyms themselves belong to different historical periods. And although the root, abstracted from a particular synonym had most times in the Indo-European period the same semantic content, this is not the rule; often the semantic content of the Indo-European root is new in the synonym. From the fact that nevertheless we have found with the individual concept groups numerous morpho-semantic correlations, two conclusions can be made. First, that behind the established correlations there are certain identical, permanent imitative tendencies that have been working in the historical development with unchanged or not too changed phonemes; second, that among the radical morphemes in the concept groups there must be a considerable amount of old imitative linguistic material, which preserved in the later stages of the Indo-European languages its semantic concept, if not its sound form. 19. On a detailed inspection of the analyzed synonymic sets permanent imitative tendencies are confirmed. This is seen first in the cases of unusual change of meaning where terms whose roots have an outer significant phonemic characteristic acquire in spite of their originally different meaning the meaning corresponding to the particular significant phonemic characteristic. Another confirmation in seen iri the cases where terms with the same phonemic characteristic are substituted, both containing the same or a similar root. Imitative tendencies appear also in a third phenomenon where terms, possessing no significant phonemic characteristic are substituted by terms marked by a such. All these phenomena can be most clearly observed in cases of well recorded historical succession of terms, e. g. in Romance languages if compared with Latin. Finally we know cases of borrowings from foreign languages when terms with preferred root forms for a given concept are borrowed in agreement with synonyms possesing the same preferred phonemic characteristic. In our statistical material borrowings from cognate Indo-European languages only come into consideration for the calculation of significance; analyzing entire synonymic sets, we can observe, moreover, analoguous cases of borrowings from .non-cognate languages. Outside the calculation of the significance remain also the so-called »Urschopfungen« (original creations) 174 The repetition- of phonemic characteristics [25] in the domain of the individual Indo-European languages which contain the very phonemic characteristics found in our statistical material to be significant. Because of the strict restriction of our inquiry to the original radical part of the synonyms root forms significantly changed through mutual contamination or irregular sound transformation, the so-called »Typüsaufpro-pfungen (the graftings of type) are likewise excluded. Further instances of such a reconstructed agreement between sound and meaning appear in the combination of unsuitable radical parts with suitable affixes or by other morphological means, e. g. by reduplication [26], [19], 402. In our synonymic sets certain cases of secondary — not calculated — significance of root extensions belong there. Beyond doubt we have here to do with surprising parallel phenomena of imitative tendencies evident already from the comparatively narrow framework of our statistical material, limited to the common forms of the Indo-European radical morphemes. 20. Considering only the root material taken into account in the statistical calculations of significance, the later imitations, however, provoked by the continued imitative tendéncies, represent an obvious minority in comparisons with the old imitations. Since we have no knowledge of the historical origin of the old imitations we cannot a priori distinguish the possible secondary transformations from a previous non-imitative linguistic material on the one hand, and the Indo-European »Urschopfungen« or the inherited »Urschopfun-gen« from the pre-Indo-European period on the other hand. But as the old imitations come up in great, semantically'closely connected groups, and this in the earliest, still accesible stage of Indo-European languages, their primary origin is, on the whole, much more probable. In favour of this speaks also the circumstance that in single concept groups the positive and negative significant phonemic characteristics are clearly opposed with regard to opposed articulatory areas in the organs of speech, and that a similar opposition is found between the particular concept groups (table XII). And the most decisive in this respect is the fact that it is just the simplification of the Indo-European phonemes into the non-differentiated phonemic classes in connection with the auditory-tactile-proprioceptive categorization which permitted the discovery of numerous and systematic morpho-semantip correlations. If the non-differentiation of phonemic classes and the multi-sensory imitation, as they indeed seem to be, are the two characteristic features of an archaic linguistic stage, then our statistical investigation has brought us to the frontiers of that period when the forming of the original imitative morphemes was accompanied by the forming of linguistic phonemes. Be it as it may, one thing is clear: the imitative tendencies which reach till into our days represent at a time of increasing arbitrariness of the linguistic sign a regression taking place mostly in the peripheric areas of the linguistic and conceptual world. But the old imitations bear — irrespective of the uncertainties of detail — testimony to a quite different, opposite state where central elements of human language and thought are in a well recognizable way build up by the imitative representation of reality. 175 B. Voduselc The five tables in the appendix contain: Table X — a symmetric arrangement of the general and combined consonant characteristics according to their principal articulatory areas in the organs of speech; Table XI — the total calculation of ■Chi-square, in all the fields, for the seven concept groups from SHARP to BLOW; Table XII — the significant deviations concerning the individual phonemic characteristics, in all the fields, for the seven concept groups from SHARP to BLOW; Table XIII — the supplementary data about the six-member vowel distribution set subsequently up, and the respective data abciut the consonant-vowel combinations K — short A, and P-U. Table XIV — the Chi-square results, in all the fields, for the thirty at random set up groups of synonymic sets, and the greatest deviations concerning the individual phonemic characteristics in these groups; Explanation: In the tables the positive significant characteristics are printed with capital letters and the figures for all the significant results are in heavy print. , The statistical data for vowels, consonant pairs, consonants, and consonant-vowel combinations refer everywhere to the morphemic groups I to V; The number of vowels is in principle identical with the number of roots; one root with SHARP, three roots with SMELL, two roots with SMOOTH, and one root with TOUGH, contain, however, no vowel. The number of consonants is reduced by subtracting the repetitions of the same consonant class in the two-consonant roots. The fundamental figures for the REPRESENTATIVE SAMPLE are: vowels 1417, consonant pairs 968, consonants 1936 (morphemic groups III to V) and 415 (morphemic group II), 37 repetitions having been subtracted — altogether 2314. ' The initial figures 533.67 for consonants, 225.63 for consonant pairs, and 325.93 for vowels in the table XIV represent the arithmetic means of the respective numbers occurring in 30 groups of synonymic series set up at random. Each single group was formed by adding at random synonymic series from the REPRESENTATIVE SAMPLE till the limit of 220 consonant pairs was attained, the number of consonants and vowels being thus mechanically determined. If a synonymic series had to appear more than two times in a group, this second repetition was not taken into account. A great deal of the ordered consonant pairs having a small relative frequency, the sums of each two pairs were^ taken for the Chi-square calcula-: tions given in the tables XI and XIV. For the same reason in some cases even the sums of more consonant pairs had to be joined: kk etp. denotes kk, rr, ss, 11, tt, pp, nr, rn, sr, rs, lr, rl; si etc. denotes si, Is, nl, In, tl, It; tn etc. denotes tn, nt, pn, np. At the probability level P = 0.001 the non-significant Chi-square for the distributions of consonants — taking six degrees of freedom — is 22.457. At 176 The repetition- of phonemic characteristics the same level the non-significant Chi-square for the distributions of consonant pairs — taking fortythree degrees of freedom — is 77.419. The corresponding non-significant Chi-square for the five-vowel distributions with four degrees of freedom is 18.467, and for the six-vowel distribution with five degrees of freedom it amounts to 20.515. In the tables XII, XIII, XIV, in each case, the right lower figure is the determined standard deviation, and the left upper figure is the number of standard deviations. For the groups SHARP and TOUCH the figures are adequately reduced. 177 B. Vodusek , MOUTH SHARP ROUGH TABLE X NARROW SMELL SMOOTH TOUCH BLOW LIPS BLOW ,pk PR pn PS PL : pt P BLOW TOUCH tk TR tn ts tl T tp , TOUCH SMOOTH Ik lr In Is L It IP SMOOTH SMELL SK sr SN S Si ST sp SMELL NARROW nk nr N(n) ns nl nt np NARROW ROUGH rk R rn rs . rl rt rp ,ROUGH SHARP K KR KN ks kl kt kp SHARP THROAT SHARP ROUGH NARROW SMELL SMOOTH TOUCH BLOW MOUTH 178 The repetition- of phonemic characteristics the same level the non-significant Chi-square for the distributions of consonant pairs — taking fortythree degrees of freedom — is 77.419. The corresponding non-significant Chi-square for the five-vowel distributions with four degrees of freedom is 18.467, and for the six-vowel distribution with five degrees of freedom it amounts to 20.515. In the tables XII, XIII, XIV, in each case, the right lower figure is the determined standard deviation, and the left upper figure is the number of standard deviations. For the groups SHARP and TOUCH the figures are adequately reduced. 177 B. Vodusek , MOUTH SHARP ROUGH TABLE X NARROW SMELL SMOOTH TOUCH BLOW LIPS BLOW pk PR pn PS PL • pt P BLOW TOUCH tk 1 TR tn ts tl T tp .TOUCH SMOOTH lk lr In Is L It IP SMOOTH SMELL SK sr SN S si ST sp SMELL NARROW nk nr N(n) ns nl nt np NARROW ROUGH rk R rn rs rl rt rp .ROUGH SHARP K KR KN ks kl kt kp SHARP THROAT SHARP ROUGH NARROW SMELL SMOOTH TOUCH BLOW MOUTH 178 The repetition- of phonemic characteristics TABLE XI Chi-square SHARP ROUGH .NARROW SMELL SMOOTH TOUCH BLOW CONSONANTS K + 11.93 + 2,93 - 1.33 -15.15 -11.61 _ 3.31 — 16.99 R + 3.25 + 49.67 - 1.38 - 1.50 - 5.66 — 0.05 0.00 N - 8.86 -14.52 +45.97 + 6.46 - 0.36 — 6.01 - 0.08 S + 5.69 - 0.06 - 0.22 + 12.64 + 0.13 + ■ 1.05 - 0.08 L - 1.75 - 3.36 - 0.67 + 0.01 + 68.14 — 3.98 - 0.26 T -13.33 -11.53 - 1.23 + 0.13 - L43 + 26.88 - 1.56 P - 0.87 - 0.14 - 1.62 +. 0.39 - 1.57 + 0.23 +70.30 45.71 82.84 52.46 36.31 88.93 41.55 89.31 CONSONANT PAIRS kk etc. + 3.46 + 5.72 - 1.72 + 0.82 + 0.18 . _ 1.13 . 0.00 KR + rk + 12.41 + 29.18 - 0.05 - 4.11 - 9.74 — 1.33 - 0.01 KN + nk - 0.01 - 0.31 +42.09 - 4.63. - 3.05 - 1.60 - 2.75 nn -12.37 - 6.98 . + 4.83 - 1.67 - 4.59 - 6.17 - 6.77 SN + ns + 0.37 - 5.35 + 7.88 + 59.31 + 0.81 - 0.12 + 17.26 si etc. + 0.05 - 0.45 + 0.36 + 8.25 . +48.50 — 1.82 - 6.29 kl + lk + 0.85 - 1.66 - 6.54 - 4.19 + 4.64 — 2.63 - 5.57 PL + Ip - 1.72 - 1.62 0.00 + 0.06 + 23.43 — 0.70 + 23.41 pt + tp -.2.33 - 5.12 - 5.27 + 5.34 + 1.01 + 3.97 + 7.71 sp + ps + 2.11 + 1.06 + 0.35 + 0.32 - 0.41 —' 0.12 + 14.26 pk +, kp + 2.78 + 0.18 0.00 - 0.22 - 3.40 + 1.18 + 5.81 PR + rp - 2.30 + 4.04 - 3.62 + 0.01 - 2.56 — 0.83 + 12.01 SK + ks + 34.50 + 2.03 - 7.32 - 0.55 - 1.24 — 0.96 -11.20 ST + ts - 9,73 - 5.74 - 1.42 - 1.02 - 0.72 + 15.78 - 1.63 tk + kt - 9.27 - 4.97 - 1,68 + 1.02 - 5.11 + 9.81 - 1.89 TR + rt - 0.26 + 3.95 + 0.10 - 1.85 + 0.03 + 5.76 - 5.79 tn etc. - 1.15 - 5.15 + 2.94 + 1.14 0.00 + 0.01 + 0.21 95.74 83.59 86.22 94.58 109.51 53.99 122.66 VOWELS Short A ■ + 7.95 + 5.44 + 0.49 - 0.04 - 1.26 + 1.50 - 0.58 Long A -11.16 - 5.04 — 3.02 - 4.51 - 4.91 0.00 - 2.41 I + 0.54; - 3.75 + 0.57 - 6.32 0.00 - 0.25 -10.02 U - 5.96 - 9.26 — 1.76 + 32.23 - 0.81 - 5,34 +31.36 E + 0.67 + 4.94 + 0.37 - 1.28 + 2.73 + 0^87 - 0.53 26.28 28.44 6.21 44.38 9.71 7.97 44.90 179 TABLE XII Significant deviations for the particular phonemic characteristics SHARP ROUGH NARROW SMELL SMOOTH TOUCH BLOW Cons. C. Pairs Vowels 938 413 539 505 233 283 549 238 327 520 208 319 490 739 215 326 280 423 - 565 226 353 KR 4.65 6.25 K 3.44 15.69 -3.83 11.83 -3.25 11.83 -4.22 11.83 R 9.24 6.25 -3.07 6.25 K — Short A 6.08 5.13 3.88 3.72 Short (A + E) 3.61 11.27 5.32 8.16 long (a + e) -3.25 7.02 - long a -3.91 4.71 KN 4.04, 3.08 N -3.22 9.70 -4.01 7.32 7.44 7.32 - nn -3.55 3.48 SN 5.98 3.01 3.81 3.01 SL + NL + TL 3.66 3.00 6.88 3.00 S 3.65 7.25 - L lk PL TP + PT PS + SP P 9.99 6.01 -3.15 -3.11 3.56 2.63 4.56 3.17 4.72 3.17 3.26 2.22 3.19 3.13, 8.63 8.16 PR 4.28 3.38 P-U 6.92 3.29 U -3.18 6.45 6.31 6.45 6.55 6.45 i -3.43 5.94 SK 3.81 4.45 -3.13 3.29 -3.23 3.29 ST -3.62 4.38 4.82 3.90 KT + TK 3.02 4.89 T -3.39 12.77 -3.06 9.63 4.71 11.57 TR 3.84 2.82 -3.79 2.35 TR + PR 3.59 4.12 Chi-square RS ROUGH SHARP Vowels 1417 539 283 Short A + 7.95 + 6.08 Long A -11.16 - 5.05 I + 0.37 — 3.75 U - 5.96 — 9.26 Short E + 1.37 6.49 Long E - 0.78 - 0.60 27.59 31.23 TABLE XIII Short (A Expected Significant deviations SHARP E) 539 375 334.35 Short (A + E) Expected ROUGH 283 219 175.56 40.65 43.44 Calculated stand, dev. 11.27. Calculated stand, dev. 8.16 40.65 = 3.81 11.27 43.44 = 5.32 8.16 Long (A + E) 32 Expected , 54.78 22.78 Calculated stand, dev. 7.02 -22.78 = -3.25 . 7.02 - Statistical data for K — short A RS SHARP ROUGH libe 10 18 6 lila 51 34 17 Illbc 5 2 3 ■IV—V 7 5 3 73 59 29 Statistical data for P-U RS BLOW Vowels 1417 353 IIa 24 18 libe — 1 lila 15 10 Illbc 1 3 IV—V 5 2 45 34 SHARP ROUGH K — short A 59 K — short A 29 ■Expected 27.77 Expected 14.58 31.23 14.42 Calculated Calculated stand, dev. 5.13 stand, dev. 3.72 31.23 = 6.08 14.42 = 3.88 5.13 3.72 BLOW 353 P-U 34 Expected 11.21 22.79 Calculated stand, dev. 3.29 22.79 - 6.92 3.29 TABLE XIV The Chi-square results for 30 groups Cons. 533.67 1. 3.51 2. 5.17 3. 5.96 4. 5.86 5. 3.73 C. Pairs 225.63 19.59 9.84 15.01 22.81, 7.91 Vowels 325.93 3.82 9.63 4.79 3.11 0.59 6. 6.92 7. 7.00 8. 3.52 9. 5.05 10. 3.31 11. 3.58 ,12. 4.98 19.13 18.94 15.05 25.82 9.21 13.19 31.07 5.13 1.40 10.84 2.86 7.48 1.71 0.37 13. 8.48 14. 3.95 15. 7.32 16. 5.92 17. 5.37 18. 8.16 19. 3.52 17.87 12.89 19.41 13.84 21.97. 19.13 11.34 4.66 11.15 7.01 1.91 4.48 10.56 1.39 20. 8.70 21. 4.17 22. 9.48 23. 8.09 24. 3.99 25. 3.57 26. 3.11 14.23 22.18 13.75 17.66 > 7.06 8.12 9.91 9.22 1.17 1.00 10.26 0.87 0.42 10.96 27. 14.96 28. 4.17 29. 5.48 30. 15.58 26.66 16.74 20.92 31.12 5.20 2.72 3.29 1.90 K 2.49 11.83 R 2.06 6.25 T 2.88 9.63 P 2.17 8.16 KR 3.40 2.14 1.52 1.67 2.28 3.29 6.25 3.08 rr 0.00 nr 1.74 . sr 1.44 lr 2.27 0.00 1.58 1.08 rn 1.74 0.73 nn 1.96 2.58 SN 1.86 2.70 rs 3.00 1.37 kt 2.17 2.44 kp 1.95 1.88 The greatest deviations for the particular phonemic characteristics in 30 groups CONSONANTS N S L 1.98 2.46 2.41 '1.32 7.25 6.01 CONSONANT PAIRS KN ks kl 2.08 3;62 rl 0.00 0.00 nl 2.25 1.46 sl 2.31 1.50 11 0.00 0.00 kk 2.48 rk 2.35 nk 2.19 SK 2.18 lk 2.13 0.30 ns 2.62 0.61 3.01 In 1.77 ss 0.00 Is 1.88 0.00 rt 2.36 nt 2.62 ] ST 2.34 It 3.30 1.54 1.22 3.24 rp 2.48 1.17 np 4.38 0.43 sp 2.49 IP 2.28 2.92 2.63 0.37 1.06 1.05 1.21 1.32 tk 2.04 Pk 2.30 3.26 tr 2.39 PR 1.96 2.35 2.63 3.38 tn ts 2.05 2.56 2.03 0.34 pn ps 1.96 2.48 1.58 0.94 tl 2.21 PL 2.06 2.14 3.17 tt 0.00 pt 2.08 tp 1.81 0.00 0.96' 2.00 pp 3.12 0.74 VOWELS Short A Long A I U 2.76 1.80 1.98 2.20 7.14 3.67 5.94 6.45 E 2.01 9.75 The repetition- of phonemic characteristics BIBLIOGRAPHY '[1] A. Meillet, Linguistique historique-et linguistique générale, Paris, 1921. [2] P. de Saussure, Course in General Linguistics, New York, 1959. [3] R. A. Fisher and F. Yates, Statistical Tables for Biological, Agricultural ànd Medical Research, 6'h ed., London, 1963. [4] G. Herdan, Language as Choice and Chance, Groningen, 1956. [5] S. J. Armore, Introduction to Statistical Analysis and Inference for Psychology and Education, New York, 1966. [6] E. Benveniste, Repartition des consonnes et phonologie du mot, Travaux du Cercle linguistique de Prague. 8. Prague, 1939, 27—36. [7] J. Kurylowicz, L'apophonie en indo-européen, Wroclaw, 1956. [8] J. Kurylowicz, Probleme der indogermanischen Lautlehre, Innsbrucker Beiträge zur Kulturwissenschaft, Sonderheft 15, Innsbruck, 1962, 107—115. [9] A. Meillet et J. Vendryes, Traité de grammaire comparée des langues classiques, Paris, 1924. [10] J. M. Korinek, Studie z oblasti onomatopoje (Remarques sur les onomatopées), Prague, 1934. [11] A. Meillet, Introduction a l'étude comparative des langues indo-européennes, Sih ed., Paris 1937. [12] s; Ullmann, Semantics, An Introduction to the Science of Meaning, Oxford; 1962. . ■ [13] W. Wundt, Völkerpsychologie, I. Band, Die Sprache, 1. Teil, Leipzig, 1900. [14] M. Chastaing, Nouvelles recherches sur le symbolisme des voyelles, Journal de psychologie normale et pathologique, 61, (1964), 75—88. [15] R. Husson, Recherches récentes sur les mécanismes de la parole, La Nature, No. 3355, (1963), 415—421. [16] F. Winckel, Kybernetische Funktionen bei der Stimmgebung und beim Sprechen, Phonetica, 9 (1963), 108—126. [17] G. Kahlo, Der Irrtum der. Onomatopöetiker, Phonetica, 5, (1960), 35—42. [18] H. Wissemann, Untersuchungen zur Onomatopoiie, Heidelberg, 1954. [19] M. Grammont, Traité de phonétique, 3"i ed., Paris, 1946. [20] J. Vendryes, Le langage, Paris, 1921. [21] H. Berr, Le langage et l'outillage mental, preface to J. Vendryés, Le langage, Pàris, 1921, VII—XXVIII. [22] F. Mayer, Schöpferische Sprache und Rhytmus, Berlin, 1959. [2.3] S. Ullmann, The Principles of Semantics, 2nd ed., New York, 1957. [24] N. Wiener, The Human Use of Human Beings, New York, 1954. [25] L. Paul, Prinzipien der Sprachgeschichte, 5th ed., Halle a. S., 1920. [26] W. Oehl, Elementare Wortschöpfung, Anthropos 12—13, (1917—1918), 575 f., : 1047 f.; 14—15, (1919—1920), 405 f.; 16—17, (1921—1922), 780 f.; 18—19, (1923—1924), 858 f. (To be continued.) POVZETEK Preiskava Buckovega onomaziološkega besednjaka glede notranje zveze med glasom in-pomenom se sklepa s pojmovno skupino, »oster, špičast, glasen«, obstoječo iz 55 sinonimičnlh serij, in vsebujočo nadštevilno -fonemsko karakteristiko R. Na podlagi končnega statističnega preračuna celotnih rezultatov sledi prvi del razlage, v katerem je obravnavana nenaključnost teh rezultatov in je teoretično utemeljen uporabljeni metodološki postopek. 185 Bojan Cop BEITRÄGE ZUR INDOGERMANISCHEN WORTFORSCHUNG VII* 27. Het. hariuzzi Ntr. »Tisch aus Rohrgeflecht (?)« = Dupl. GISBANsUR.AD.KID (s. Friedrich, Heth. Wb. 57) ist ein Wort mit Suffix -uzzl-, über dessen Funktion s. Friedrich, Heth. El. I2 39 und.Kronasser, Etym. d. heth. Spr.: II. Wortbildung des Heth. 240 f. Es bildet Nomina instrumenta z. Teil aus Nominalstämmen; so sicher in lahhurnuzzi- »Opfertisch (??)« von lahhurä- »Opfertisch« (hier erweitert um ein -n-); sonst deverbal: kuruzzi- »Schneidewerkzeug« von kuer- »schneiden«, warp-uzi- »Bronzegerät b. Baden« von warp- »baden« usw.; doch ist bei kattal-uzzi- »Schwelle (Unter- u. Oberschwelle)« auch an denominativen Ursprung zu denken V'kattala- »unterer«?). Da unser Gegenstand aus Rohrgeflecht gebildet ist, wird man beim Grundwort *hari- oder "harija- wohl an dies »Rohrgeflecht« bzw. »Rohr« selbst denken dürfen. Nun ist die Anknüpfung an anderweitiges idg. Material gegeben: es werden wir idg. *aro-m »Schilfrohr« (Pokorny, Idg. EW. 68) in gr. äron Ntr. ' »Natterwurz; Art Schilfrohr«, ari-saron »eine kleine Art davon«, lat. harundö »Rohr« heranziehen müssen. Der heth. -i- bzw. -;a-Stamm stimmt gut zu gr. ari- in ari-saron, beides aus einem uralten *Hari- »Rohr, Schilfrohr«. Die indogermanische Wurzel hatte im Anlaut einen »Laryngal«, also '"■'Har-, wie das Hethitische erweist; vielleicht ist lat. h- auch ein Rest davon.1 Da heth. h- = Null in anderen idg. Sprachen wohl nur bei Urverwandtschaft vorkommt, da weiter lat. -nd- kaum erst nach etwaiger Entlehnung aus einer Mittelmeersprache abtreten konnte, da nicht produktiv, wird man wohl mit Recht an eine ererbte, zufällig nur im Mittelmeergebiet belegte indogermanische Worteinheit denken müssen; Entlehnung aus einer Mittelmeersprache kommt also kaum in Frage!2 * Unter demselben Titel erschienen: I (Nr. 1—3) in . KZ. 74, 1956, 225—232; II (Nr. 4—10) in Die Sprache III, 1956, 135—149; III (Nr. 11) in Die Sprache VI, 1960, 1—8; IV (Nr. 12—14) in Slav. Rev. XI, Anhang, Lingu., 1958, 49—68; V (Nr. 15—22) in Lingu. VIII, 1966—1968, 165—175; VI (Nr. 23—26) erscheint demnächst in KZ. . 1 Trotzdem dass im Europeisch-Indogermanischen das *H- im Anlaut gewöhnlich spurlos geschwunden ist, - kann hie und da doch ein Rest davon geblieben sein; so vor allem im Lateinischen, wo unorganische Ä-Anlaute oft vorkommen. Mehr anderswo. 2 Gegen Pokorny a. O, u. a. ■ 187 Bojan Čop 28. Idg. *q5-. Das Indogermanische besass eine Wortsippe, die »Zeit« bedeutete und in folgenden Wörtern belegt ist:' alb. hohe »Zeit, Wetter«, aksl. casu »Zeit, Stunde« und apreuss. kisman »Zeit, Weile«.3 Das apreuss. Wort ist Mask., mit -ma- (idg. *-mo-) aus einem "kes- abgeleitet; die übrigen zwei sind vokalische Stämme, das alb. auf *-ä-, das sl. auf *-o- auslautend. Bisher pflegte man allgemein als Wurzel ein idg. *qes- anzusetzen, wobei man fürs Albanische an den Wandel idg. *-s- intervokalisch > -h- dachte. Dies ist aber nicht richtig; denn die Flexion von idg. *es- »sein« zeigt im Alb. 2. Sg. je, das ja auf *esi zuruückgeht. Und die Endung der 1. PI. -mi (jemi »wir sind«, kemi »wir haben«) ist ebenfalls auf eine Urform mit inter-vokalischem *-s- zurückzuführen: *-mesi. Ein weiterer Fall von intervokali-schem *-s- = alb. Null wird später zur Sprache kommen. Alb. kohe müsste also in solchem Fall *fco lauten, ohne -h-.4 Eine Urform, die dem albanischen Widerspruch folge leistet und zugleich auch alle drei Wörter unter einen Hut zu bringen imstande ist, wird wohl *qesk'- sein: apreuss. -s- kann ja auch ein urbaltisches -s- darstellen und dies ist aus einem :-'-sk'- gut herleitbar; alb. -h- aus *-sfc'- ist eine bekannte Tatsache;5 und schliesslich ist slavisches s- aus urbaltoslav. *-&- und dies aus einer Lautgruppe wie *-sk'- ebenfalls gut bezeugt.6 Das indogermanische *qesk'- kann natürlich keine echte Wurzel sein; diese Struktur muss weiter in eine kurze Wurzel *qe- und das bekannte »inkohative« Suffix *-sk'- zerlegt werden. Dass das inkohative Suffix *-sfc'-auch in nominale Gebilde verschleppt werden kann, zeigen mehrere Beispiele.7. ■ - 3 Vgl. u. a. Trautmann, Apreuss. Sprachdenkm. 359 m. Lit.; Trautmann, Bsl. Wb.; Sadnik-Aitzetmüller, Aksl. Hdwb. 221 mit neuen Deutungen, die aber nicht überzeugen. 4 Alb. vjeherr »Schwiegervater«,, vjeherre »Schwiegermutter« wird wohl ein *wesk'uro- < *swek'uro- vertreten, also nicht *swesuro-, das voralb. aus *swesuro-(< "swek'uro-) entstanden wäre; überdies müsste . ein *swe- im Albanischen zu dje-werden, ein überspringen des *s- in den Inlaut ist also auch von dieser Seite her notwendig. 5 Vgl. z. B. die »Inkohativa« auf *::sk'ö, im Albanischen -h, z. B. njoh »ich kenne« < *g'nesk'ö ~ lat. nöscö usw. 6 So aksl. ¿2/sgsto, tysösta »1000« < Hüs-k'mtl usw. 7 Vgl. z. B. *-sk'- in der Gruppe von *ais- »wünschen, begehren, aufsuchen« (Pokorny 16): ai. verbal icchäti (< Hs-sk'ö) »sucht, wünscht«'= av. isaiti > ai. icchä »Wunsch«, arm. aich (< *ais-sk'ä) Untersuchung«, av. verbal isasä »begehre« > nominal aesasa- »suchend, ausgehend auf« = *aisosk'o- in lat. aeruscäre »bitten«, ebenso denominal (von *ais-sk'o- oder *ais-skä) ahd. elscön »forschen, fragen, fordern« usw., vgl. ahd. eisca »Forderung«, auch lit. usw. so: ieskau, iesköti »suchen«, sl. iskati (neben Präs. iskö, auch istö) »suchen«, iska »Wunsch«; zu "prk'sk'ö- »frage, begehre, bitte, verlange« (ai. prcchäti »fragt«, av. pdrdsaiti »fragt, begehrt«, arm. e-harch »er hat gefragt«, lat. poscö »fordere, erbitte«, ir. arco »ich bitte« usw.) ai. prcchä' »Frage, Erkundigung« = arm. harch ds., ahd. forsca »Forschung, Frage« mit forscön »fragen, forschen«; vgl. auch umbr. persk-lu, pesk-lU »supplicatione« u. a. (Pokorny 821 f.); ähnlich in "wen(3)- »streben«, wünschen, lieben, befriedigt sein« ■ usw. (Pokorny 1146 f.): ein sk'- Präs. in ai. väncliali »wünscht«,-Vdavon vänchä »Wunsch«, anord. ösk (< *>wunskö) »Wunsch«, ahd. wunsc »Wunsch« usw., wovon ahd. usw. wunscen »wünschen« usw. usw. 188 Beiträge zur indogermanischen Wortforschung VII Somit bekommt man durch die obige, wohl einzig richtige Analyse eine neue indogermanische Wurzel *qe-. Wenn man sie richtiger bedeutungsgeschichtlicher Analyse unterwirft, kann sie an folgendes iranisches Wort angeschlossen werden, das Benveniste, Etudes sur la langue ossete. (Paris 1959), 25 ff. so evident ausgelegt hat: Bisher wollte man osset. än-cain, Part, än-cad »aufhören, ruhen, Halt machen« mit av. syä-, sä- »sich freuen«, av. Sä'ti-, apers. siyäti- »Freude, Wohlbehagen, Glück« verbinden; mit osset. Wort gehört zusammen sogd. 'nc'y- (an-cäy-), Part, 'nc.'t- (an-cät-) »aufhören, Halt machen«, das den ossetischen Formen genau entspricht. Wie sogd. s't (sät) »reich« und sw- (saw-) »gehen« av. syav-, apers. siyav- ds. < *cyav-, idg. *qjeu-) gegen osset. cäu- »gehen« zeigen, entspricht in Wirklichkeit der uriranischen Lautgruppe öy- im Sogdischen s-; also muss an-cäy- anders interpretiert werden, nicht durch *cyä-\. Man muss also osset. än-cai- und sogd. an-cäy- aus einem Iran. :l'cäy- herleiten (S. 28). Auch morphologisch und semasiologisch müssen beide Wörter von "cyä- getrennt werden; vgl. zur Bedeutung noch osset. cadäg »langsam«, sogd. 'p-c't »Halt, Intervall (Zwischenzeit), Augenblick«, das von einem "ap-öäy- ausgeht. ■ : Man muss also ein iran. *cäy- mit -'cäta- als Partizip rekonstruieren, das »Halt machen, aufhören«, Ende einer Bewegung, bezeichnete; davor standen die Praverbien *ham- und '*apa- (S. 28). Da Benveniste wohl Recht hat, wird man aus diesem *cäy- (Präs. wohl "cäyati) und *cäta- ein idg. *qe-jeü, *qe-to- folgern dürfen. Die Bedeutung wird wohl dieselbe sein wie im Sogdischen und Ossetischen: »Halt machen,, aufhören«. Damit ist aber das preussisch-slavisch-albanische *qe-sk'- gut vereinbar.. Zuerst muss man an das schon erwähnte sogd. 'p-c't (apcät) »Halt, Zwischenzeit, Augenblick« erinnern, das sehr ähnliche Bedeutung aufweist wie unser Wort "qe-sk'- + Suffixe. Dem preussisch-slavisch-albanischen Wort liegt eine Auffassung des Begriffes »Zeit« zugrunde, die darin ein »Aufhören, Ruhen« sieht. Ähnliches in der germanischen Entwicklung der Gruppe *quije- usw. »behaglich ruhen« (Pokorriy 638): got; hweilan »weilen, zögern, aufhören« ~ got. hweila, ags. hwil, ahd. (h)wila »Weile, Zeit« (anord. noch hvila »Ruhebett«), Man wird wahrscheinlich von nun an eine neue Auffassung auch für die slavische Gruppe um aksl. cajati, cajö »erwarten, warten, hoffen« aufbringen können, denn cajö kann von nun ah-eher zu unserem "qe- »aufhören, ruhen, Halt machen« (anstatt zu *g«ei- »worauf achten« (Bedeutung!) bei Pokorny 636 f.) gehören. Darüber an einer anderen Stelle. Eine weitere Gruppe wird auch später zur Sprache kommen. Vielleicht wird man unsere Wurzel *qe- als eine isolierte und verselbständigte Bildung (mit Suffix -e-, wie so oft)8 aus *oq- verstehen dürfen, das in s vgl. zum Suffix e- u. a.: *pel- »giessen, fliessen, aufschütten, füllen, ein- • füllen« neben *pl-e- ds. (Pokorny 798' ff.) z. B. in lit. pilti »giessen, schütten, aufschütten, füllen«, ai. piparti, PI. piprmäs »füllt, sättigt', nährt usw.« — aor. äprät, gr. pleto »füllte sich«, lat. pleö usw. »fülle« u. a. 189 Bojan čop gr. ôknos »Bedenklichkeit, Zaudern«, oknéô »zaudere«, oknërôs »saumselig« stecken kann (die bei Pokorny 774 gegebene Verbindung mit got. aha »Sinn, Verstand« usw. müsste dann entfallen). Zur Bedeutung vgl. oben osset. cadäg »langsam«! 29. Osset. xizyn, dig. xezun »klettern«, auch »weiden, hüten, warten« wird von Benveniste a, O. 55 ff. genauer besprochen. Nach ihm bedeutet es immer »se déplacer en grimpant, traverser un lieu accidenté«; urspr. »grimper à travers«, von der Herde oder dem Hirten »grimper« (trans. oder intrans.) > »paitre« oder »faire paitre«; dazu xizän, dig. xezän »Weideplatz; Stelle, wo man überklettert, übergeht« (urspr. »l'endroit qu'on parcourt en grimpant«); von-der Bedeutung »se déplacer en montagne, traverser un pâturage« kommt xizyn zu »traverser«; trans. »faire paître (en montagne)« ist immer »faire parcourir lés pâtures de montagne«; daraus »protéger«, »garder« (vgl. russ. pasti »paitre (trans.)« und »garder«); schliesslich, wie in d. warten, »garder« > »attendre«., im Digorischen xezun = »surveiller (l'arrivée de quelqu'un); attendre«. ■ - Die ursprüngliche Bedeutung von osset. xizyn ist also »grimper; parcourir un trajet escarpé« (S. 57). Ais Urform ist nach Benveniste ein *xaiz- anzusetzen, dem nur im Sogdischen eine gute Entsprechung gegenübersteht: sogd. z'y-yyz-, z'y-xyz- »rampant sur le soi«, also ein "xaiz- »ramper, se déplacer à quatre pattes«, woraus Benveniste eine'gemeinsame Wurzel xiz- »Se mouvoir à quatre pattes, ramper, grimper« erschliesst. Bei Zupitza, Germ. Gutt. 116 finden wir eine germ. Sippe, die bedeutungs-mässig und phonetisch gut entspricht: mhd. hickelen »springen«, engl, hitch »to move with a jerk usw.«; weiter vergleicht Zupitza noch lit. kelziotis, das ich aber in den mir zur Verfügung stehenden Wörterbüchern vergeblich gesücht habe; weder in Niedermann-Senn-Brender, Wörterbuch der litauischen Schriftsprache noch in Praenkel, Lit. EW. steht ein solches Wort. Die germ. Sippe ergibt zusammen mit der iranischen eine idg. Wurzel "q(h)eig'-, die etwa »aufspringen« bedeutete. Zur osset. Bedeutung vgl. etwa ital. salire 'steigen', urspr. 'springen'. 30. Toch. B sruk- »sterben« bildet das Präsens III: 3. Sg. Med. sruketrd, 3. PI. srukentrd, Partizip srukemane usw.; dazu Konj. V 1. Sg. sraukau usw., Opt. srükoy usw., schliesslich das Prät. I 1. Sg. srukäwa, 3. Sg. sruka usw.9 In A entspricht als Grundverbum bedeutungsmässig wdl- »sterben«, das bekannte Verwandte hat.10 Als Kausativ begegnet aber unser sruk- in der Bedeutung »töten«: Prät. III Med. 3. Sg. sruksüt, Partizip Prät. sasruku.11 9 Vgl. zum Formensystem Krause, Westtoch. Gr. I 304. 10 vgl. van Windekens, Lex. etym. des dial. tokh. 156. Bei Pokorny 1144 f. fehlt das tocharische Wort. u S. Schulze-Sieg-Siegling, Toch. Gr. 481; Thomas-Krause, Toch. El. II 158. 190 Beiträge zur indogermanischen Wortforschung VII Vgl. noch B srukor »Sterben, Tod«.12 Das Präsens III hat gewöhnlich reduzierte Wurzel,1' so auch in unserem Fall; dies wird unser Ausgangspunkt für eine neue Etymologie sein, denn die alten sind nicht stichhaltig.w Toch. -ru- wird einem idg. *-ru- entsprechen; dies *-ru- kann entweder Schwachstufe zu *-reu- sein oder aber aus *-wr- umgestellt sein, wie auch *-lu- aus *-wl-\ vgl. *q"etru-, *qutru- »4« (1. Kompositionsglied) in av. öa&ru-, gr. try-phäleia »Helm«, gall. petru- usw., lat. quadru-, *q"etru-s »viermal« in av. ča&ruš, lat. quater (< *quatrus) neben *quetwr- usw. »4« (Pokorny 642 f.); oder gr.-lat. päon. Hüq»os »Wolf« in gr. Ipkos, lat. lupus, päon. MN Lijppeios, Ltfkkeios neben *wlquos ds. in ai. vr'kas, av. vdhrka-,, got. wulfs, lit. vilkas, aksl. vlbk-h (Pokorny 1178). Der tocharische Stamm sruk- kann also als Entsprechung einen anderen mit -vor- finden: *swrK-, Dann kann unmittelbarer Anschluss an die idg. Wurzel *swergh- »sorgen, sich worum kümmern; krank sein« (Pokorny 1051) eintreten, vgl. ai. sürk-S-ati »kümmert sich um etwas«, got. saürga »Sorge, Betrübnis«, as. ahd. sorga, afränk. sworga »Sorge«, got. saürgan usw., ahd. sorgen, sworgen »sorgen«, alb. dergjem »bin bettlägrig« (< *swörghjö); weitere Verwandte (ohne -w-): air. serg »Krankheit«, lit. sergü, sirgti »krank sein«, aksl. sraga »Krankheit«, srag-b »austetus, torvus«. Toch. sruke- aus "srugho- kann also unmittelbar zur Seite von lit. sergü < *swerghö treten. Es bleibt nur noch die Bedeutung übrig. Das tocharische »sterben« ist eigentlich nur das Ende einer Krankheit; schon dieser Überlegung zufolge wird unsere Etymologie richtig sein. Klare Parallele bildet z. B. die idg. Sippe lsg»el- »stechen« (Pokorny 470 f.), die auch folgende Wörter geliefert hat: lit. gilti »zu schmerzen anfangen«, gälas »Ende, Tod«, gelä »Schmerz«, apreuss. gallan »Tod«, russ.-ksl. želejo, želeti »betrauern«, aksl. želja »Leid«, šalb »Schmerz«, air. at-baill »stirbt«, korn. bal »pestis«, kymr. (a)ballu »sterben«, ahd. quelan »Pein leiden«, ags. ciuelan »sterben«, anord. kvelia »peinigen«, ahd. quellen ds., ags. cwellan »töten« usw. usw. Eine noch bessere Parallele ist im Uralischen nachzuweisen: finn. sure-»trauern, sich grämen«, suretta- »bekümmern, betrüben« und »töten«, surma »Tod« (weitere Verwandte bei Collinder, Fenno-Ugric Voeabulary, Stockholm 1955, 116). Die bedeutungsmässige Ähnlichkeit dieser uralischen Sippe mit dem idg. "swergh-, *srugh- ist so gross, dass ich an Urverwandtschaft denke. Mehr darüber an einer anderen Stelle. 12 Thomas-Krause a. O. 260. • 13 Vgl Krause, Westtoch. Gr. I 66: mit einer Ausnahme nur Verba mit dem Wurzelvokal 5, i, u (3 mit e), vgl. triketSr »in die Irre gehen«, mitentdr »sich' aufmachen«, wiketor »sich entfernen«, lyuketSr »leuchten« usw. usw. 14.Vgl. van Windekens a. O. 116 mit Lit. Beide Etymologien sind wenigstens morphologisch schwierig, da ein Suffix -k- in der Präsexisklasse III nicht vorkommt; hier sind reine Wurzelverba zusammengeflossen; prutk- »eingesperrt sein« hat ein Suffix -tk-, also eine ganz andere Bildung, ebenso s3tk- »sich ausbreiten« und pldtk-»hervortreten« (wenn hierher), mSsfc- »sein, sich befinden« wohl -sk'- der Inkohativa, ebenso musk- »schwinden« usw. 191 Bojan čop 31. Aind. ¿is- »übr'ig bleiben« in Pass. üsyäte »bleibt übrig«, sinästi »lässt übrig«, ¿esa-»übrig«, »Rest« ist bisher ungedeutet, denn keine annehmbare ausserindische, auch nicht arische Anknüpfung ist gefunden worden.15 Die Wurzel ¿is- ist jedoch uralt, denn sie kommt schon im Vedischen vor: Pass. ¿isyäie ist in Rigveda belegt, der a-Aorist ä-äisa-t und der passive Aorist Sesi in Atharvaveda. Auch der Präsenstypus Sinästi muss ererbt sein, da nicht produktiv. Die Laute dieser Wurzel sind so eindeutig, dass sie unmittelbar Rekonstruktion der idg. Urform gestatten: "k'eis- »übrig bleiben«. Dies kann aber in *k'ei- + Wurzeldeterminativ -s- zerlegt werden. Somit bekommen wir eine kürzere Urwurzel *k'ei-, die sofort Anschluss findet. Denn m. E. gehört hierher die baltoslavische Sippe *sei-ra- »verwaist, orbus«, die in aksl. sir-b »verwaist« und lit. seiry's »Witwer«, seire »Witwe« steckt.16 Diese Sippe wurde mit Recht mit av. sae- »verwaist, Waise« zusammengestellt, das aus ''¿aj-i- als erstes Kompositionsglied entstanden ist und-bekanntem Wechsel der Suffixe folgt (1. Kompositionsglied -i-, selbständiges Adjektiv -to-),v entspricht also gänzlich der bsl. Bildung, uridg. also 'k'ei-ro-, *jc'ej-i-.18 . Zur Bedeutung ist die Sippe von idg. *g'he-, *g'hei- »leer sein, fehlen«, »verlassen, fortgehen« (Pokorny 418 f.) zu vergleichen: ai. jähäti »verlässt, gibt auf«, hiyale »wird verlassen, bleibt zurück«, av. zazämi »ich entlasse aus«, ava-zä- »entfernen, verscheuchen«, gr. khäzomai »weiche, fliehe, lasse ab«, gr. khetis »Ermangelung«, khateö »ermangle, bedarf usw.«, kheros »beraubt, leer«, khe'rä »Waise, Witwe« usw. Man wird also ein »verlassen, übriglassen, aufgeben« ansetzen dürfen, das alle Glieder schön zusammenhält; '°k'ei-ro- also ursprünglich »Verlassen«.18« 32. Heth. hä- »glauben, trauen, für wahr (zuverlässig) halten« bei Friedrich, Heth. Wb. 45 ist etymologisch noch nicht erklärt. Es wird konjugiert nach der vokalischen Verbalklasse I 2 b: Präs. Sg. 1 liämi, 2. häsi, Prät. Sg. 1 hänun, 2. häis; Part, hänt-ß Die Flexion ist. also im 15 Die Verbindung mit av. syazd- »zurücktreten vor, aufgeben« usw. ist formell und bedeutungsgeschichtlich falsch. 1« Zur' baltoslavischen Sippe vgl. Trautmann, Bsl. Wb. 301;' Praenkel, Lit. EW. »70-usw. i" Vgl. z. B. gr. kyd-rö-s »berühmt, ausgezeichnet« ~ kyd-i-äneira »mit sich auszeichnenden Männern«, av. ddrdz-ra- »fest« ~ ddrdz-i-raßa- »festen Wagen habend« usw., Brugmann, Grdr. 112 i( 78. 18 Vgl. Bartholomae, Altiran. V,7b. 1547. 18a Zusatz: Ich verweise noch auf die abwegige Analyse des altind. Verbums, die Kuiper, Nasalpräs. 40 und 168 gegeben hat. Auch seine weiteren Verknüpfungen sind wohl abzulehnen, nur über ai. siyate »weicht, schwindet« liesse sich diskutieren. 19 vgl. Friedrich, Heth. Wb. 45 und Heth. El. 12 88 f. 192 Beiträge zur indogermanischen Wortforschung VII Vgl. noch B srukor »Sterben, Tod«.12 Das Präsens III hat gewöhnlich reduzierte Wurzel,1' so auch in unserem Fall; dies wird unser Ausgangspunkt für eine neue Etymologie sein, denn die alten sind nicht stichhaltig.1,1 Toch. -ru- wird einem idg. *-ru- entsprechen; dies *-ru- kann entweder Schwachstufe zu *-reu- sein oder aber aus *-wr- umgestellt sein, wie auch *:lu- aus *-wl-; vgl. *quetru-, *q»trur .»4« (1. Kompositionsglied) in av. öa&ru-, gr. try-phäleia »Helm«, ga.ll. petru- usw., lat. quadru-, *q"etru-s »viermal« in av. ča&ruš, lat. quater (< *quatrus) neben *quetwr- usw. »4« (Pokorny 642 f.); oder gr.-lat. päon. Hüq"os »Wolf« in gr. Itfkos, lat. lupus, päon. MN L$ppeios, Lijkkeios neben *wlq"os ds. in ai. vr'kas, av. vdhrka-,. got. wulfs, lit. vilkas, aksl. vlbk-b (Pokorny 1178). Der tocharische Stamm sruk- kann also als Entsprechung einen anderen mit -wr- finden: *swrK-. Dann kann unmittelbarer Anschluss an die idg. Wurzel *swergh- »sorgen, sich worum kümmern; krank sein« (Pokorny 1051) eintreten, vgl. ai. sürk-S-ati »kümmert sich um etwas«, got. saürga »Sorge, Betrübnis«, as. ahd. sorga, afränk. sworga »Sorge«, got. saürgan usw., ahd. sorgen, sworgen »sorgen«, alb. dergjem »bin bettlägrig« (< *swörghjö); weitere Verwandte (ohne -w-)\ air. serg »Krankheit«, lit. sergü, sirgti »krank sein«, aksl. sraga »Krankheit«, sragi, »austerus, torvus«. Toch. sruke- aus '-'srugho- kann also unmittelbar zur Seite von lit. sergü < *swerghö treten. Es bleibt nur noch die Bedeutung übrig. Das tocharische »sterben« ist eigentlich nur das Ende einer Krankheit; schon dieser Überlegung zufolge wird Unsere Etymologie richtig sein. Klare Parallele bildet z. B. die idg. Sippe !'guel- »stechen« (Pokorny 470 f.), die auch folgende Wörter geliefert hat: lit. gilti »zu schmerzen anfangen«, gälas »Ende, Tod«, gelä »Schmerz«, apreuss. gallan »Tod«, russ.-ksl. šelejo, želeti »betrauern«, aksl. želja »Leid«, salb »Schmerz«, air. at-baill »stirbt«, korn. bal »pestis«, kymr. (a)ballu »sterben«, ahd. quelan »Pein leiden«, ags. cioelan »sterben«, anord. kvelia »peinigen«, ahd. quellen ds., ags. cwellan »töten« usw. usw. Eine noch bessere Parallele ist im üralischen nachzuweisen: finn. sure-»trauern, sich grämen«, suretta- »bekümmern, betrüben« und »töten«, surma »Tod« (weitere Verwandte bei Collinder, Fenno-Ugric Vocabulary, Stockholm 1955, 116). Die bedeutungsmässige Ähnlichkeit dieser uralischen Sippe mit dem idg. ':'swergh-, *srugh- ist so gross, dass ich an Ürverwandtschaft denke. Mehr darüber an einer anderen Stelle. 12 Thomas-Krause a. O. 260. !3 Vgl Krause, Westtoch. Gr. I 66: mit einer Ausnahme nur Verba mit dem Wurzelvokal 3, i, u (3 mit e), vgl. trilcetSr »in die Irre gehen«, mitentdr »sich' aufmachen«, wiketor »sich entfernen«, lyuketSr »leuchten« usw. usw. m Vgl. van Windekens a. O. 116 mit Lit. Beide Etymologien sind wenigstens morphologisch schwierig, da ein Suffix -k- in der Präsensklasse III nicht vorkommt; hier sind reine Wurzelverba zusammengeflossen; prutk- »eingesperrt sein« hat ein Suffix -tk-, also eine ganz andere Bildung, ebenso s3tk- »sich ausbreiten« und pldtk-»hervortreten« (wenn hierher), m3sk- »sein, sich befinden« wohl -sk'- der Inkohativa, ebenso musk- »schwinden« usw. 191 Bojan čop 31. Aind. sis. »übrig bleiben« in Pass. Sisyäte »bleibt übrig«, sinästi »lässt übrig«, ¿esa- »übrig«, »Rest« ist bisher ungedeutet, denn keine annehmbare ausserindische, auch nicht arische Anknüpfung ist gefunden worden.15 Die Wurzel ¿is- ist jedoch uralt, denn sie kommt schon im Vedischen vor: Pass. ¿isyäle ist in Rigveda belegt, der «-Aorist ä-Sisa-t und der passive Aorist iesi in Atharvaveda. Auch der Präsenstypus ¿inästi muss ererbt sein, da nicht produktiv. Die Laute dieser Wurzel sind so eindeutig, dass sie unmittelbar Rekonstruktion der idg. Urform gestatten: *k'eis- »übrig bleiben«. Dies kann aber in *k'ei- + Wurzeldeterminativ -s- zerlegt werden- Somit bekommen wir eine kürzere Urwurzel *k'ei-, die sofort Anschluss findet. Denn m. E. gehört hierher die baltoslavische Sippe *sei-ra- »verwaist, orbus«, die in aksl. sir-b »verwaist« und lit. seiry's »Witwer«, seire »Witwe« steckt.16 Diese Sippe wurde mit Recht mit av. sae- »verwaist, Waise« zusammengestellt, das aus *iaj-i- als erstes Kompositionsglied entstanden ist und-bekanntem Wechsel der Suffixe folgt (1. Kompositionsglied -i-, selbständiges Adjektiv -to-),n entspricht also gänzlich der bsl. Bildung, uridg. also -''k'ei-ro-, Zur Bedeutung ist die Sippe von idg. "g'he-, "g'hei- »leer sein, fehlen«, »verlassen, fortgehen« (Pokorny 418 f.) zu vergleichen: ai. jähäti »verlässt, gibt auf«, hiyale »wird verlassen, bleibt zurück«, av. zazümi »ich entlasse aus«, ava-zä- »entfernen, verscheuchen«, gr. khäzomai »weiche, fliehe, lasse ab«, gr. khetis »Ermangelung«, khateö »ermangle, bedarf usw.«, kheros »beraubt, leer«, khe'rä »Waise, Witwe« usw. Man wird also ein *k'ei- »verlassen, übriglassen, aufgeben« ansetzen dürfen, das alle Glieder schön zusammenhält; *k'ei-ro- also ursprünglich »verlassen«.18*"1 32. Heth. liii- »glauben, trauen, für wahr (zuverlässig) halten« bei Friedrich, Heth. Wb. 45 ist etymologisch noch nicht erklärt. Es wird konjugiert nach der vokalischen Verbalklasse I 2 b: Präs. Sg. 1 Hämi, 2. häsi, Prät. Sg. 1 hänun, 2. häis; Part, hänt-ß Die Flexion ist. also im 15 Die Verbindung mit av. syazd- »zurücktreten vor, aufgeben« usw. ist formell und hedeutungsgeschichtlich falsch. u Zur' baltoslavischen Sippe vgl. Trautmann, Bsl. Wb. 301;' Praenkel, Lit. BW. 970^ usw. Vgl. z. B. gr. kyd-rö-s »berühmt, ausgezeichnet« ~ kyd-i-äneira »mit sich auszeichnenden Männern«, av. ddr3z-ra- »fest« ~ ddrdz-i-raßa- »festen Wagen habend« usw., Brugmann, Grdr. 112 78. 18 Vgl. Bartholomae, Altiran. Wb. 1547. 18a Zusatz: Ich verweise noch auf die abwegige Analyse des altind. Verbums, die Kuiper, Nasalpräs. 40 und 168 gegeben hat. Auch seine weiteren Verknüpfungen sind wohl abzulehnen, nur über ai. siyate »weicht, schwindet« liesse sich diskutieren. 19 vgl, Friedrich, Heth. Wb. 45 und Heth. El. V 88 f. 192 Beiträge zur indogermanischen Wortforschung VII Grunde dieselbe wie in der Klasse I 3 (Stämme auf -äi-), z. B. (dieselben Formen): haträmi »schreiben«, haträsi, halränun, haträes, hatrant-. Da nun die Klasse I 3 auf idg. Denominativa auf *-ä-je- zurückgeht, könnte man auch bei hä(i)- an eine Urform *häje- (idg. !!Häje-) denken; doch diese Urform ergibt leider keinen Anschluss an anderweitiges Material. Es gibt noch eine andere Möglichkeit:- nimmt man an, dass in, unserer »Wurzel« hä- einst ein intervokalisches -w- vorhanden gewesen ist, so kommt man auf ein ''ho(w)o-, idg. *Howo- (genaueres unten) zurück, das wohl eine gute Verknüpfung findet: ein idg. "(H)ow- mit ähnlicher Bedeutung gibt es ja in der folgenden griechisch-lateinischen Sippe: gr. öi'ö,20 öi'omai2(> (Aor. öi'ssato, Si'sthen) »meine, glaube, verfalle worauf«, an-oistos »unerwartet«, später kontrahiert oiomai und oimai, das evident an ein "owis- denken lässt, und lat. ömen, alät. osmen Ntr. »Vorzeichen«, das als deverbales *owis-men zu deuten ist.21 Die griechisch-lateinische Sippe lässt einen Stamm auf -is-*owis- ansetzen, der wahrscheinlich überhaupt nicht verbal ist, sondern nominal, nämlich mit Suffix -is-22 von einem kürzeren Verbalstamm *ow-»glauben, erwarten« abgeleitet; gr. öi'ö usw. also denominal "owis-jö. . , Doch das wichtigste ist, dass man hier mit einem verbalen !fow- bzw. ''How- zu tun hat, das nun auch im Hethitischen steckt: hä- aus "-'Howo-; da die Bedeutungen aller verglichenen Wörter gut übereinstimmen, bleibt nur noch die lauthistorische Frage übrig, wie heth. *Howo- zu hä- geworden ist. Im allgemeinen scheint im Hethitischen das indogermanische intervokali-sche -w- erhalten zu sein, vgl. newas »neu, frisch« (idg. *newos) mit newahh-»erneuern«; misriwant- »hell, glänzend«, misriwatar »Glanz, Pracht, Herrlichkeit«, misriwes- »hell sein, strahlen«, mit denominalem Suffix *-w-, *-went-, und die Flexion der w-Stämme, wie z. B. harnau- »Gebärstuhl«, Gen. harnawas, Dat.-Lok. harnawi, oder assu- »gut«: Gen. assawas, Dat.-Lok. assawi, Abi. assawaza, Instr. assawet, PI. Nom, assawes, assawa usw. Doch neben Abi. idälawaz »von -bösem« kommt aüch idälaz vor und dies bildet den Schlüssel zu unserem Problem: im Abi. war die Urform "-owo-ti,2} das -w- stand zwischen zwei -o- und schwand; erst nachträglich wurde es nach anderen Kasus wiederhergestellt. Wenn wir also annehmen, dass -wim Heth. zwischen zwei -o- schwinden musste, war es analogisch hergestellt im Gen; Sg. -awas statt -as aus *-owos, im Abi. (s. oben), im Dat.-Lok. PI. -awas statt -as aus *-owös (od. ähnl.)?i usw., während es z. B. im Lok. Sg. 20 Das i ist kurz oder lang, ursprünglich ist die Länge (*owisjo). ' 21 Vgl. Boisacq, DEGr. 692; Walde-Hofmann, LEWS II 208 mit Lit., Frisk, Gr. EW. II 566. 22 Dazu vgl. Brugmann, -a. O. 533 f. 23 zur Entstehung des hethitischen Ablativs vgl. meine Bemerkung Lingu IX/1 (Ljubljana 1969), S. 46. 24 Der Dativ — Lokativ (und Genitiv) PI. kann zwar aus einem Dualkasus (mit *-ös = av. -a) erklärt werden, wie Pedersen, Hittitisch 33 nach Benveniste wollte, doch bietet sich noch eine andere, viel bessere Möglichkeit, über die ich an anderer Stelle berichten werde. 193 Bojan öop bleiben rhussté: -awi < *-owi oder *-owei, ebenso im Nom.-Akk. PL Ntr. -awa < *-ow-ä.. Genau so verhielt sich unser *How-: in' der Urform *Howomi (1. Sg. Präs.) schwand -w- regelrecht, ebenso in der 3. PL Präs. *Howonti, im Part. s'Howont- usw., auch in der 1. Sg. Prät. "How-u-m, sogar in der 2. Sg. Präs. *Howosi, sobald das thematische -o- (= heth. -a-) aus der 1. Sg. in die 2. Sg. übertragen wurde (vgl. haträsi < "-äjasi = "-äjosü). Vor dem Themavokal -e- müsste dagegen -w- bleiben; doch trachtet das Hethitische bekanntlich danach, dass das Paradigma möglichst einheitlich ist; so wurde die -Stammform hä- < *Howo- auch in 2. Sg. Prät. häis übertragen, wo sie ursprünglich *Howe- lautete; Hilfe leistete dabei auch' der Typus I 3 (haträ(i)-)•. häis ~ haträes (bekanntlich kann häis auch /häes/ gelesen werden).25 Zusatz: Endgültig wird unsere Vermutung, dass in ha-'einst ein -w- stand, durch die Existenz eines weiteren hethitischen Verbs erwiesen, und zwar durch heth. huslc- »warten, abwarten« (Präs. 2. Sing, huskisi, 3. Plur. hu§kanzi usw., s. Friedrich, Heth. Wb. 77). Das Wort ist augenscheinlich ein Iterativum auf sk'e/o- von einer w-Wurzel, die besten Anschluss gerade an hä- < ''■'Howo- findet; huske- kann direkt aus idg. *Hu-sk'e/o- mit schwundstufiger Wurzel entstanden sein, wie die hethitischen Iterativa noch öfters schwundstufige Wurzelform aufweisen, so etwa kuask- zu kuen- »schlagen; erschlagen, töten« (idg. *gn,Jm-sk'e/o-), zikk- »(wiederholt) legen« zu däi- »setzen, legen, stellen«' (idg. *dh-sk'e/o- oder besser *dhd-sk'e/o-, worüber an einer anderen Stelle), uwansik- zu wen-, went- »futuere« (idg. !iwn(t)rsk'e/o), usk-- »sehen« zu au(sj- ds. (idg. *u-sk'e/o-), azzikk- zu ed- »essen« (idg. Síed-sk'e/o-), akkusk-zu eku- »trinken« (idg. *dqw-sk'e/o-) u. a. Zur Bedeutung vgl. z. B. gr. oknéo »zaudere« — got. ahjan »glauben, meinen« (Pokorny 774); vor allem aber *men- »denken usw.« — *men- »warten«. In allen diesen Fällen kann »glauben, denken, meinen« das Ursprüngliche sein. 25 Die Idee, dass in hä- ein -w- ausgefallen wäre, stammt eigentlich von Prof. Ostir; doch verglich er das idg. *aw- »gern haben« (Pokorny 77 f.), das aber bedeu-tungsmässig stärk abweicht und auch keine guten Voraussetzungen für den Ausfall von -w- bietet. Ostirs Idee vom Ausfall des -w- war leider noch nicht genau umschrieben, Was ich hier nachholen muss. Ostir verglich auch heth. lä- »lösen« mit idg. *lew-»abschneiden, trennen, loslösen« (Pokorny 681 f.) in ai. lunä'ti usw. »schneidet, schneidet ab«, gr. Vyö »löse, befreie«, lat luö »büssen, zahlen«, solvö »lösen«, got. frdliusan »verlieren« usw. usw. Auch hier muss man mit unserer Regel vom Ausfall des -w- zwischen zwei -o- auskommen: Präs. 1. Sg. lämi < *lowo-mi, 2-. Sg. läsi (mit analog. Themavokal -o- wie in häsü) < Howo-si, 3. PI. länzi < *lowo-nti, Prät. 1. Sg. läun < *low-u-m, 1. PL läwen < *lowo-wen, Imper. 2. PL latten < "lowo-ten (vgl. hatratten »schreibet!« aus < *-äja-tten < *-ä-jo-ten mit analogischer Durchführung des Stammauslautes -o-), Part, lant- < Howont-; analogische Entfernung des -■turn Prät. 2. Sg. läis (statt Harnes), 3. Sg. läit (statt Hawet), Imper. 2. Sg. läi k. Howei (statt Hawi); vielleicht so auch 3. Sg. Präs. läi < Howei, doch kann hier auch an ein *lowo-i gedacht werden (mehr anderswo). In anderer Umgebung blieb das -w- der letzten Wurzel erhalten: lawarija- (lawarr-) zerbrechen« hatte urspr. How-, aber das Suffix war wohl *-^rje/o-, also mit -a- im Anfang im Uranatolischen. 194 Beiträge zur indogermanischen Wortforschung VII Grunde dieselbe wie in der Klasse I 3 (Stämme auf -äi-), z. B. (dieselben Formen): haträmi »schreiben«, haträsi, halränun, haträes, hatrant-. Da nun die Klasse I 3 auf idg. Denominativa auf *-ä-je- zurückgeht, könnte man auch bei hä(i)- an eine Urform "häje- (idg. "Haje-) denken; doch diese Urform ergibt leider keinen Anschluss an anderweitiges Material. Es gibt noch eine andere Möglichkeit:- nimmt man an, dass in, unserer »Wurzel« hä- einst ein intervokalisches -w- vorhanden gewesen ist, so kommt man auf ein *'ho(w)o-, idg. *Howo- (genaueres unten) zurück, das wohl eine gute Verknüpfung findet: ein idg. '''fHJow- mit ähnlicher Bedeutung gibt es ja in der folgenden griechisch-lateinischen Sippe: gr. oi'ö,w oi'omai20 (Aor. o'i'ssato, öi'sthen) »meine, glaube, verfalle worauf«, an-öistos »unerwartet«, später kontrahiert oiomai und oimai, das evident an ein *owis- denken lässt, und lat. ömen, alät. osmen Ntr. »Vorzeichen«, das als deverbales *owis-men zu deuten ist.21 Die griechisch-lateinische Sippe lässt einen Stamm auf 4s-"owis- ansetzen, der wahrscheinlich überhaupt nicht verbal ist, sondern nominal, nämlich mit Suffix -2s-22 von einem kürzeren Verbalstamm *ow-»glauben, erwarten« abgeleitet; gr. öi'ö usw. also denominal *owis-jö. , Doch das wichtigste ist, dass man hier mit einem verbalen sow- bzw. "How- zu tun hat, das nun auch im Hethitischen steckt: hä- aus *Howo-; da die Bedeutungen aller verglichenen Wörter gut übereinstimmen, bleibt nur noch die lauthistorische Frage übrig, wie heth. *Howo- zu hä- geworden ist. Im allgemeinen scheint im Hethitischen das indogermanische intervokali-sche -w- erhalten zu sein, vgl, newas »neu, frisch« (idg. *newos) mit newahh-»erneuern«; misriwant- »hell, glänzend«, misriwatar »Glanz, Pracht, Herrlichkeit«, misriwes- »hell sein, strahlen«, mit denominalem Suffix *-w-, "*-went-, und die Flexion der w-Stämme, wie z. B. harnau- »Gebärstuhl«,, Gen. harnawas, Dat.-Lok. harnawi, oder assu- »gut«: Gen. assawas, Dat.-Lok. assawi, Abi. assawaza, Instr. assawet, PI. Nom. assawes, assawa usw. Doch neben Abi. idälawaz »von bösem« kommt auch idälaz vor und dies bildet den Schlüssel zu unserem Problem: im Abi. war die Urform *-owo-ti,23 das -w- stand zwischen zwei -o- und schwand; erst nachträglich wurde es nach anderen Kasus wiederhergestellt. Wenn wir also annehmen, dass -wim Heth. zwischen zwei -o- schwinden musste, war es analogisch hergestellt im Gen: Sg. -awas statt -as aus *-owo.s, im Abi. (s. oben), im Dat.-Lok. PI. -awas statt -as aus "-owös (od. ähnl.)-.4 usw., während es z. B. im Lok. Sg. 20 Das i ist kurz oder lang, ursprünglich ist die Länge (*owisjö). ' 21 Vgl. Boisacq, DEGr. 692; Walde-Hofmann, LEW.3 II 208 mit Lit., Frisk, Gr. EW. II 366. 22 Dazu vgl. Brugmann, -a. O. 533 f. 23 zur Entstehung des hethitischen Ablativs vgl. meine Bemerkung Lingut IX/1 (Ljubljana 1969), S. 46. 24 Der Dativ — Lokativ (und Genitiv) PI. kann zwar aus einem Dualkasus (mit *-ös = av. -a) erklärt werden, wie Pedersen, Hittitisch 33 nach Benveniste wollte, doch bietet sich noch eine andere, viel bessere Möglichkeit, über die ich an anderer Stelle berichten werde. 193 Bojan čop bleiben mussté: -awl < *-owi oder ,s-owei, ebenso im Nom.-Akk. PI. Ntr. -awa < *-oW-ä. Genau so verhielt sich unser "How-\ in der Urform *Howomi (1. Sg. Präs.) schwand -w- regelrecht, ebenso in der 3. PI. Präs. '''Howonti, im Part. *Howont- usw., auch in der 1. Sg. Prät. *How-u-m, sogar in der 2. Sg. Präs. !>Howosi, sobald das thematische -o- (== heth. -a-) aus der 1. Sg. in die 2. Sg. übertragen wurde (vgl. haträsi < *-äjasi = *-äjosü). Vor dem Themavokal -e- müsste dagegen -w- bleiben; doch trachtet das Hethitische bekanntlich danach, dass das Paradigma möglichst einheitlich ist; so wurde die Stammform ha- < "Howo- auch in 2. Sg. Prät. häis übertragen, wo sie ursprünglich *Howe- lautete; Hilfe leistete dabei auch- der Typus I 3 (haträfij-): häis ~ haträes (bekanntlich kann häis auch /häes/ gelesen werden).25 Zusatz: Endgültig wird unsere Vermutung, dass in hä- einst ein -w- stand, durch die Existenz eines weiteren hethitischen Verbs erwiesen, und zwar durch heth. husk- »warten, abwarten« (Präs. 2. Sing, huskisi, 3. Plur. huskanzi usw., s. Friedrich, Heth. Wb. 77). Das Wort ist augenscheinlich ein Iterativum auf