The Proto-Germanic Shift *Ā>*Ō and Early Germanic Linguistic Contacts
Total Page:16
File Type:pdf, Size:1020Kb
The Proto-Germanie shift *ã > *õ and early Germanic linguistic contacts* Post-PIE *õ and *ã fromall possible PIE sources are both reflectedin Germanic by a single sound which, because of the sounds thatcontinue it in the individual Germanic languages, is usually reconstructedas *õ forthe latest stage of Proto-Germanic.1This mergeris paralleled by the mergerof the post-PIE shortnon-high back vowels *o and *a in PrGmc. *a. The merger of non-highback vowels is - at least in descriptive - terms a northwesternIE areal feature,2encompassing Germanic, Baltic, Slavic and Celtic, with differencesin the details. In Baltic, only the shortvowels merged,but the reflexof *ã remaineddistinct from that of *o; the opposite is the case in Celtic, where only the long vowels were affectedby the merger,whereas short*o and *a remaineddistinct. The resultantvowel systemof Proto-Germanicis oddly imbalanced, with the shortvowels lacking a rounded mid-highback vowel and the long vowels lacking a long low back vowel. Therefore it has been suspected that the parallelism of developments within Germanic had indeed been perfectand that the erstwhileoutcome of the mergerhad been not a trapezoidor triangularvowel system,but a quadrangularone in which the shortand long low back vowels correlatedwith each other in quality.These vowels could be writtenphonemically *a and *ã, or, if greaterallowance is made to phonetics,[a(:)] or ratherrounded [d(:)]> if not [э(:)]. For the sake of clarity and in contrastto the later vowel system,here the symbols *â and *â will be used with referenceto the low back vowels before the emergence of the classical reconstructable Proto-Germanicstage. * My thanksgo to RobertNedoma and StefanSchumacher for many valuable and importantsuggestions to thisarticle. Part of thispaper was writtenwithin the FWF- fundedproject P20755-G03 'Die altkeltischenSprachreste in Österreich'(The Old- CelticLanguage Remains of Austria; http://www.univie.ac.at/austria-celtica/). Thehypothesis of Schrijver2003 that *ã and*õ remaineddistinct in Germanicin finalsyllables and beforetautosyllabic *n in initialand finalsyllables, is of no consequencefor the lexical items discussed here and will not be examinedfurther. It maybe notedhere that also Proto-Finno-Ugricas well as Proto-Finno-Lappic possesseda quadrangularlong vowel system consisting of *Të õ -й(see Sinor1988: 268,297, 523). Hist.Sprachforsch. 122, 268-283, ISSN 0935-3518 © Vandenhoeck& Ruprecht GmbH & Co. KG,Göttingen 2009 [2010] This content downloaded from 149.157.1.188 on Mon, 9 Dec 2013 05:24:12 AM All use subject to JSTOR Terms and Conditions TheProto-Germanie shift *ã > *o andearly Germanic linguistic contacts 269 Given the ultimatedivergence in quality between the short and the long vowel, it seems rational to surmise that despite their phonemic correlationthere must have existed slightallophonic differencesin their phoneticrealizations, with the long vowel *â tendingstronger towards roundness.This is inferablefrom the veryfact that it ultimatelyresulted in *õ some time in the Proto-Germanicperiod (thus basically Hollifield 1984: 65, but the scenario sketchedby him is rathervague). Why then the phonological correlationbetween the two sounds was ultimately broken,can only be conjectured.3Perhaps the long vowel came under pressure fromthe incipientshift *ë > *œ > *ã, at least in the western and northernvariants of Germanic. Furthermore,*ã may have been introducedas a marginal phoneme by the Proto-Germanicloss of *j between identical vowels, i.e. *aja > *ã (řórhallsdóttir1993: 35-36, citingCowgill 1973: 296). On the otherhand, the monophthongization of *a and a tautosyllabicnasal beforea velar fricativedid not help to fill the gap in the long-vowel system,because it resulted in a nasalized vowel *q (i.e. *anx > *QX),not in an oral long vowel *ã. Thus around the beginningof the historicalperiod, Germanic was on the verge of acquiring a new *ã. In words thatwere borrowed into Germanic or its dialects fromLatin in the imperialperiod, Lat. ã is not reflectedby *o but by the new *ã by default (Kluge 1913: 23-24, 128), e.g. OHG phãl 'stake' <- pãlus, OHG strãzza 'street' <- strãta, OHG kãsi 'cheese' <- cãseus, the OHG agentivesuffix -ãri <- ãrius. Cases where Lat. ã seems to be reflectedby Gmc. *ã are actually borrowingsfrom Vulgar Latin withpretonic shortening, e.g. OHG ratih,retih 'radish' <- VLat. radíc- < Lat. rãdíx 'root'. Two conclusions can be drawn: first, early Proto-Germanic*â had been moved up the phonetic triangle towards*õ by the time of those loans, so thatto the speakers of Germa- nic it was not a phoneticor phonological equivalentto Lat. ã. Secondly, loans fromLatin helped establish *ã as a more or less marginalor even erstwhileloan phonemein Germanic. Nevertheless, a handful of words (partly transmittedin secondary sources,partly foreign loans into Germanic) have been cited as evidence for a relativelylate date for the conjectured shift*ã > *õ during the Proto-Germanicperiod. In particular,it has been claimed that some of the evidence proves thatthe shiftmust have taken place afterthe first contacts of Germanic-speakingpeoples with Rome, that is, in the 1st centuryB.c. or even as late as the 1stcentury A.D (see, for example, Ringe 2006: 145-146). In this article, this evidence will be reviewed 3 See also,for example, Tops 1973, Van Coetsem (1994: 76-81). Hist.Sprachforsch. 122, 268-283, ISSN 0935-3518 © Vandenhoeck& Ruprecht GmbH & Co. KG,Göttingen 2009 [2010] This content downloaded from 149.157.1.188 on Mon, 9 Dec 2013 05:24:12 AM All use subject to JSTOR Terms and Conditions 270 DavidStifter and it will be argued thatits chronologicalvalue is not as unambiguous as has been assumed hitherto. * 1. The Gothic placename Ruma (dat. sg. Rumai, Gal exp A) 'Rome' and the accompanyingi-stem ethnonym Rumoneis* (dat. pl. Rumonim, Rom exp; 2 Tim 1, 17) 'Romans', togetherwith Old High German Ruma 'Rome', rümisk'Roman', etc. (beside innovated Rõma, rõmisk, etc.), Old Norse Rúm, Rúmaborg 'Rome', etc. (beside innovatedRóm, Rómaborg, etc.), and once Old English Rûmwalas* 'the Romans' (beside ordinaryRõm), allow us to set up a Proto-Germanicpair of toponymplus ethnonym*Rümö 'Rome', *Rümönaz 'Roman' (the shift in stem-class of the ethnonymto the i-inflectionin the plural is a specifically Gothic development,see Liihr 1985: 142-143, 147). The two names obviously go back to Latin Rõma, Rõmãnus in some way. However, in themLat. õ is representedby Gmc. *й,4and Lat. ã by Gmc. *õ, very much unlike the regular substitute*ã mentionedabove. This has been used as the prime piece of evidence for an early loan from Latin when the shiftin Germanic had not yet taken place. It has been suggested thatthe names were borrowed at a time when *â was still a closed, rounded sound in Germanic (perhaps approximately[d:]), but neverthelesssuitable to representLat. ã, whereas the best approximation forLatin closed õ was to substituteit withGmc. *й. Thus *RHmõwould allow a rare glimpse at phoneticdevelopments within the reconstructed Proto-Germanicperiod. This is the line of argument,for example, of Noreen (1894: 11-12), Streitberg(1896: 48-49), Jellinek(1926: 182- 185), Schwarz (1951: 21-22), Corazza (1969: 39-40), Hollifield (1984: 65), Ringe (2006: 146). This explanationis quite plausible and cannot be disproved on purely linguisticgrounds. But thereexists a less straightforwardalternative to explain Gmc. *Rumõ 'Rome' and *Rümönaz, an alternative which neverthelessaccounts betterfor the historicaland political environment in which the borrowingtook place in Iron-Age Central Europe. Gmc. *RHmõ, *RHmõnaz could be loans from a Central-European Celtic language like Gaulish (Gaulish will be used here as a shorthandterm for any Central-EuropeanCeltic language; such a source of *Rümö was already suspectedby Luftand Schwarz, accordingto Corazza 1969: 40; 4 Lat.õ is regularlysubstituted by Gmc.*й, e.g. OHG mürberi'mulberry' <- Lat. mõrus,OHG lürra 'pomace' <- lõrea,Middle Ripuarian ur 'hour' <- hõra,and perhaps, withshortening, OHG wlnzuril'winemaker' <- Lat. uïnitôr(Streitberg 1896: 48-49, Kluge1913: 25). Hist.Sprachforsch. 122, 268-283, ISSN 0935-3518 © Vandenhoeck& Ruprecht GmbH & Co. KG,Göttingen 2009 [2010] This content downloaded from 149.157.1.188 on Mon, 9 Dec 2013 05:24:12 AM All use subject to JSTOR Terms and Conditions TheProto-Germanie shift *ã > *õ andearly Germanic linguistic contacts 27 1 cf. also Öhmann 1919). The Gaulish exonyms forthe city of Rome and for its citizens are not directlyattested, but that such words must have existed is beyond doubt,and it is not unlikelythat they were *Rúmã and *RHmãnos,themselves loans fromthe Latin words with adaptation to the Old-Celtic phonological systemby substitutingrounded high *ü for Latin closed <5.In Proto-Celtic,the pentadiclong-vowel system*I-ê-ã- õ-й, inheritedfrom post-PIE, had been firstreduced to a triangularlong- vowel system*ï-â-û (the question of possible remainsof *ë are passed over here), a stage thatstill obtained in earlyGaulish. The gap leftby *õ had not yet been filled again at the time when Gauls and speakers of otherCentral-European Celtic languages were confrontedwith the rising Roman power in the second half of the 1stmillenium B.C. The process of introducinga new õ into the system by the monophthongizationof Proto-Celtic*oy. < Pre-Celtic *eu, *ou and partly*uu happened only as late as the historicalstages of the Celtic languages and is observable in the preservedlinguistic material. A rare furtherinstance of the Gaulish substitutionof й <ou> forLat.