Classical tradition and Norse tradition in the 'Third grammatical treatise', in “Arkiv för nordisk filologi” 108/1993, pp. 68-79

VALERIA MICILLO

Classical Tradition and Norse Tradition in the"Third Grammatical Treatise" *

0. Some of the most interesting problems related to Olàfr P6rÒarson's TlrlrdGrammatical Treatise are connected with the sources to which the Icelandicgrammarian had access to prepare his work. It is generally held that in the

first part of the treatise, dealing with phonetic and orthographic questions,he avails himself basically of books I and II of Priscian's InstitutionesGrammaticae, while for the last part, handling rhetorical and stylistic matter,he draws essentially from the third book of Aelius Donatus' Ars Maior.However, many passages show a very vague correspondence with thesource(s) presumably used. In fact Priscian's and Donatus' works, in whichconcepts and ideas of the great philosophers and grammarians of ancienttimes meet and melt together, enjoyed an enormous popularity and gave rise

to innumerable commentaries, explanations and collections of quotations,which greatly contributed both to the spreading and to the distortion of theirtheories. Also, other passages in the treatise clearly belong to differenttraditions than the ones most commonly used, and sometimes points ofcontact can be found in traditions of local origin.

The problem therefore is basically that of ascertaining: 1) whether Òl6frused the texts of the classical tradition in a (more or less faithful) "standard"version, or through one or more of the many commentaries of later deriva-tion; 2) which sources other than the traditional ones were available to him.

The present work is an attempt to examine the first part of 6lafr's treatisein such a perspective.

0.1. I wish to point out that I have addressed myself primarily to theinvestigation of direct sources, while indirect or secondary sources have been

considered only upon occasion. Reference is made therefore mostly to theLatin grammatical tradition. However, as is well known, this is ultimatelybased on philosophical and linguistic speculations of ancient Greece. The"classic" tradition of grammar developed on the theoretical foundationsprovided by Plato, Aristotle, the Stoics between others, and it was later

* A preliminary version of this paper was read at the VIII International Saga

Conference: "The Audience of the Sagas", GÒteborg, Al5 Ll-I7, 1991.

Classical Tradition and Norse Tradition in the "Third Grammatical Treatise" 69

codified in Dionysius Thrax's Téchné grammatiké and Apollonius Dyscolus'

syntax. Grammatical studies in Rome essentially followed the patterns pre-

viously established by the Greek scholars but usually failed to achieve

original results. The Latin contribution to grammatical scholarship is none-

theless of great significance inasmuch as it preserved the rich Greek heritage

concerning language. It can be said that later grammarians owed most toGreek linguistic scholarship, although often unknowingly, as it was transmit-ted and formalised in the works of Latin scholars.

This picture underlies also 6l6fr's treatise, but a discussion of the influ-ence of Greek philosophic and linguistic thought on the treatise and on

Icelandic grammatical tradition is beyond the scope of this essay. Greek

sources are referred to occasionally, but no systematic attempt is made to

trace any point back to its ultimate origin.

1. The Third Grammatical Treatise (henceforth TGT) starts, like many

works of its kind, with a chapter on the vox, its definition, characteristics and

different types. I print here the first sentence, together with the relevant

passage in Priscian's Institutiones. (Quotations are from Olsen 1884 for TGTand from Keil 1857-1880, within brackets, for the Latin parallels, unless

otherwise stated.)

1,1 Allt er hlioò, Pat ar um kvikvan-dis ayw ma skilia.

Prisc. lnsl. I,1 (II 5) Philosophi defi-niunt, vocem esse aerem tenuissimumictum vel suum sensibile aurium, idest quod proprie auribus accidit.

We can see at once that the definition of 'voice' in TGT is completely

different from that in Priscian. The only point of contact is the fact that bothregard the 'voice' as something to be perceived through the sense of hearing.

It is also to be noted that while Priscian speaks of vox,6làfr deals with hli6ò,usually, meaning 'sound'. The point is relevant as until at least the laterMiddle Ages the Latin word vox is commonly employed both for the(human) voice and for sound, as an equival ent of sonus, while this would not

appear to be the case with the Norse grammatical tradition. The twoconcepts are generally indicated by different names, viz. hlj6Ò (sound) and

rqdd (voice), for example in TGT and in the Second Grammatical Treatise

(Raschellà 1982:77-78) at least, while the situation is more complicated inthe First Grammatical Treatise (see Albano Leoni 1975:13-14; Haugen1972:6O47 ; Benediktsson 197 2:61-64).

All this already indicates that the formulation of hli6Ò in TGT does notdepend on Priscian. Conclusive evidence to this is a statement by Òl6fr, tobe examined later, set at the end of the first paragraph (1,13), where

'70

Priscian's definition is

viously discussed (see

V:rlcrirr l\licillo

reportecl and cxplicitly contrastcd witlr tlle ()lte l)rc-below).

1;1155ic:al.Iì.aclitionandNtlrscTr.atlititlnirltlre.'.lhircl(irantnlaticlrl.lrclitise..lI

"nratcrial") vs. "spiritual" (: likanilikt vs' andlikr) is implicitly cstablishctl'

It also appears that while,.òorporeal souncls" can be perceivcd "r.ìatuntlly".

"spiritual sounds", it would seem, can not' of t'e

The "corporeal vs. spiritual" opposition is a well-knowl't toprc

ptritosoptrical-theologicai speculation in ancient and mediaeval tirnes' br'rt so

far I have been unabre ,u iina any trace of such a dichotomy in connccrio,

with sound (but sec bclow)' A passage in Audax' howevcr' may bc taken

into c<tnsidcratiiln:

Aud. E.rt'arlttrt: Dt vttcc (VIl 323) Vox quid est'l Aèr ictus auditu stnsibilis'

vocis specics cluot sunt,l i)u"". Qun" sunt'J Articulirta ct confusa. (...) vlx

corpclraliscst,anincolporalis.lsccurrdumstoictlscorporalis,quicamsicdcfini.unt. ut nos in prirlcipio responclinrus Plato autem non esse voceln corpus putat:

"non enitn par.u.,u'", inquit, "aèr' se<l plaga ipsa atque percussio' id cst vox"'

Delnocritus vero ac dcinde E'picurus ex indivisis corporibus vocenì constarc

dicunt, corpus autcm esse efficiens aut patlens'

Attention is ilrawn to this passage also by Olsen (1884:3' fn')' Wc find herc

an explicit link between in" ,o* and a distinction corporalislincorporalis

whichrecallstheoppositionlikamtiktlandliktfoundinTGT.Acloseranaly-sis, however, showì-that the connection is actually rather tenuous' We may

first observe that the Latin adjective incorporalls is not a perfect equivalent

of Icelandic andtikt (: related to the spirit, 'spiritual')' although this may be

aminordifference.ThemaindifficultyisthatAudaxhereisrro/drawingadistinctionbetweentwotypesofsound,onecorporaljs,theotherincorpora./is. In fact he has already àealt with souncl types in the previous c.ontext by

the same <Iichotomy vox articulatalvox confusa usecl by Donatus in his Ar'i

Maior l,t (l{oltz 1981:603, cf' quotation above):

.,Ilownlanykin<lsof.voice,aretobefound?''..Twokincls.''..Whicharethcy.]''

"The articulate 'voicc' and the confused'voice''"

ActuallythequotationfromAuclaxwhichwesawaboveisadiscussionofthenatureofsoundingeneral.Thepointatissueiswhetherthevoxisabodyornot, and Audax reports here the opinions of the foremost philosophers'

Instead the distinction in TGT is based on the origin of sound, which car.r be

eitherfromabodyorfromSomeotherelementhavingthenatureofspirit.Wemayconclude,therefore,thattheaffinitybetweenOl6fr'sdichotomyand the distinction found in Audax is no evidence for a closer connecticrn' It

cannot be exclu6ed, however, that the opposition actually originated within

thecontextofsuchphilosophicalspeculationsasthosereferredtobytheLatin grammarian.

2.1. One important clue to this question might come from the passage of the

Holy Scriptures mentioneA in f'Cf, which concerns sounds deriving from

Apart from Priscian,Aelius Donatus. In his

another well known grammaticus cleals with tltc l'o-r,

Ars Maior he rePorts:

Don. Ars Gramtn. I,1 (lìoltz 1981:603) vox est aer ictus, scnsibilis rtuclitu'

quantum in iPso est.

This definition is shorter than Priscian's, differently formulated ancl partly

problematical because the last phrase, quantum in ipso cr.s/, Itas bccn inlcr-

pretcci in various ways. Again the treatment of hliòò in TGT shows no point

of contact with Donatus, excepting the usual releretlce to the 'voice' as

something perceived through hearing.

2. The next quotation sets forth ol/rfr's classification of souncls.

1,2-3 hlioÒ hrfir margar kynkvislir, ok vcròr hlioÒ pat, sitrì nattvrligil megv

rcyrv greina, af samkvamv tveggia tikama, &nn Qnnvr hlioÒsgrcin &r SV, ler

hnilag ritning s:egir hlioÒa andliga hlvti. likamlikt hlioÒ verÒr annat af lifligvm

hlvtvm, &nn annat af liflasum hlvtvm.

Again, no direct relationship with Latin grammatical tradition can be estab-

lished. The passage above seems to have only one characteristic in common

with conventional sound classifications, that is, the identification of different

types of soun«I. Traditional classificatory patterns can usually bc traced back

either to Donatus or to Priscian. The formcr employs a twofold scheme

originating in the Greek grammatical tradition, thc latter clistinguishes four

basic types with various sub-types.

Don. Ars Gramm. I,1 (Holtz 191t1:603) Onrnis uox aut artictrlitta est aut

confusa.

Prisc. Irr.sr. I,1 (II 5) Vocis trutem differentiae sunt quattuor: articulata, inarticu-

lata, literata, illiterata.

Donatus' simpler pattern apparently enjoyed a greater popularity than the

more complex classification adopted by Priscian. Formulations similar to, ot

identical with that found in Donatus occur in previous and contemporary

grammarians as well as in later texts (see e.g. Diomedes Ars Gramm. lll. De

ioce fl 420]; probus Inst. Art.: De voce [IV a7]; Marius Victorinus -4r.r

Gramm. Z,l-Zlqariotti 1967:661). Further types are obtained by variously

combining and/or modifying the preceding ones'

In Ol6fi's treatise sounds are classified according to different criteria. A

first general distinction is drawn between a kind of sound perceivable.,natuially,', originating from two bodies coming into contact, and another

kind deriving from "spiritual" parts. Here an opposition "corporeal" (or

72

"spiritual" things. A suggestion by J. pedcrsc,, quoted in Finnur.rrinss.rr'sedition of 'l'GT (1927,20, fn.), attempts to explain the sounds uncler consirl-eration as "uforstàelige ly<J", that is, sounds unintelligible to men. Ret'ercnccis nradc to one passage in st. paul's letters, where we read ab.ut a man wrrcr"was caught up into paradise and heard unspeakable words, which it is notlawful for a man to utter,' (II Cor. 12,4).

Pedersen's interpretation might be true. In fact TGT clescrib es thc likum_likt hlj6ò as "perceivabrc naturally", which courd impry that the souncl of"spiritual" origin does not possess the same property. on thc othcr hancl. itmay scem strange that such an irnportant characteristic is not menti..ccl atall in the Icelandic treatise. Also, it is not clear whether wc are conccrncclwith the impossibility for a man to unclcrstand. to hear or to uttcr such a kinclof sound.

Another passage in st. paul (I cor. 2,rz-r3) might in r.y opinion rhrow abetter light on the question. Here "the worcls which man.s wisclom teacheth,,are contrasted with those "which the (Holy) Spirit teachcth,, and this opposi-tion would appear to rccall the "corporear/spirituar" «richotomy in 'r'GT,although in a slightly different perspective. However, even if this refcrencc ismore convinci,g than the prcvious one, it would be unwisc to <Jraw anydefinite conclusions only on the basis of such evi<Ience.

2.2. The description of soun«ls of "spiritual" origin is not carrie«l on i, .r.GT.This may perhaps suggest that oràfr's source À unabre to providc furtherdetails on the subject. In fact a voluntary omission wourd secm rathcrunlikely in view of orafr's well structured anà balanced classification. Unrike"spiritual" sounds, the likamtikt hlj6ò in TGT branches into many sub-rypes:

1,3-11 likamlikt hrioÒ verÒr annat af rifligvm hlvtvm, *nn annat af lir.rivsvrnhlvtvm' Hlioò pat, irr hayriz af liflrysvm hlvtvm, veròr annat af .ij. *ririgvnrskepnvm, &nn annat af v-rgririgvm, annat af samkvamv rcriligra lvtn ok orcrir(-gra. Af roriligvm lvtvnr verÒr rioÒ s."m af hofvÒsk*pnvm ercli, vindvrn okvOtnvm. Af vrgriligvm hlvtvm verÒr lioÒ sem stainvm aeÒa malmi aeòa stren-givm, ok verÒr po piess kyns lioò iafnan af rqring nokcurs likama rifligs ieÒavlifligs. Af samkvamv reririgra lvta ok vrerilig.u u".Ò. lioò sam pa ar vindr *òavgtn eÒa alldr slar sinv afli viÒ iorÒ ,'Òa aÒra vreriliga luti. lioÒpat ar veròr afliflrysvm lvtvm aer svmt ogreinilikt sem vinda gnyr eÒa vatna bytr cÒa reiÒarprvmvr, en svmt hlioÒ er greinirikt aptir natvririgri samlioÒan, pei..i .". phiros.-phi kgllvÒv mvsicanr, ok verÒr pat rioÒ hit efsta ok hit eezta af rrering ring, peirr"'vij., ier sol ok tvngl ok .v. merkistiornvr ganga i, pgr rcr planate lerv [alraÒar,ok hritir pat cerestis armonia rÒa himnresk lioòagrain. pessar stiornvr sagÒiplat6 hafa lif ok skyn ok vera odaÒligar. Greinilikt lioÒ veròr iriflarsvm rvtvm,pat sam ver kgrlvm tistvlikt lioÒ, sem i marmi ok srrengivm ok pipvm ok ailzkyns sgngfrrvm. J lifrigvm lvtvm ok vitlasvm verÒr rioÒ, srm i viÒvm ieÒagrasvm ok !o af hraring nokkvrs reriligs likama. Af lifancri lvtvm peim ..r skyn

Vrlcriir Micillo Classicll 'l'raclition and Norse Tradition in the "Third Grammatical Treatisc" 13

hufa verr)r annat lioÒ. pat lcr rodcl hleitir, itnn annat, pat rr tigi ler rt2ckl. stntfota stapp ir:Òa hancla klapp ok annat slikt.

Olélr builds hcrc a well articulatecl structure of a pyramiclal type, in rvhicheach itcm is usually described, provicled with examples and sonìctinìcsbriefly commentcd upon. This classification can easily be translateci into a

tree sclrcmc, where the first node, likamlikt, dominates two other kinds olsounds: ortc cclnrirrg from aninratc (liJligurn) beings, thc othcr deriving lrontinanimate (liJ'lanunr) things. A furthcr step subdivides the latter soundsaccorcling as such inanirnatc cntities arc 1) nrobilc,2) n<ln ntobile or 3)

mobile things mceting immotrilc oncs.Such a complex classification, which is l'urther developed in the trcutisc,

does not seem to rnatch with any of thclse recorded in the tradition. Atenuous echo rnay bc founcl, pcrhaps, in a passage of the Instituta ArtiumbyProbus. Thc chapte r Dc vot'c starts with a definition of sound followed by a

distinction, also lbund in Donatus, betwcen vox articulata and vox conJ'usu'.

Probus /rrs/. Art.: Dc locr,(lV 47) nunc omnis vox sive sonus aut articulata est

aut confusa. articulata cst, qua homines locuntur et litteris conprehendi potest(...). confusa vero aut anintaliunt aut inanimalium est, quae litteris conprehendinon potest. animalium est ut puta cquorum hinnitus, rabies canum, rugitusfcrarum, serpentum sibilus, aviunr cantus et cetera talia; inanimalium autenì est

ut puta cymbalorum tinnitus. flagellorum strepitus, undarum pulsus. ruinaecasus, fistulae auditus et cetera talia. est et confusa vox sive sonus hominum.quae litteris conprehendi non potest, ut puta oris risus vel sibilatus, pectorismugitus ct cctcru taliu.

In Probus the yr.rx conJusu, a kind of sound which cannot be expressed bymeans of lctters, comes cithcr from animate beings (animalia) or frominanimatc things (irrarrlmaliu). This could make a parallel to the distinctiondrawn by Oldfr in TGT, but not a perfect one. In fact the general structure inthe two texts unclcr consideration is completely different. TGT shows a morecomplex scheme with a greater number of categories, while Probus has a

simpler classification in which a distinction is made between the yox confusaderiving from animalia and that deriving from inanimalia. Instead TGTdistinguishes within the sound of inanimate origin the greinilikt type and rheogreinilikt one. This is a discrimination which is usually regarded as practical-ly equivalent to the articulatus vs. inarticuLaras dichotomy of the Latin'iradition. It is to be noted that also the vox confusa is otien described as

"quae scribi non potest" (e.g. Diom. Ars Gramm. lI De voce [I 420|;Aud.Excerpta: De voce [VII 323], etc.) or "quae litteris. conprehendi non potest"(e.g. Probus, quoted above). In this respect, theretbre, we have in TGT thereverse situation to that found in Probus.

As regards the examples provided in each text, again we find divergencies.

Vrlcrirt Micilltr14

The Latin passagc cxhibits u wicle variety ol itetns, well [talitrrcctl [rtltll

numerically ancl svntacticirlly. rvith a ttcut clistirtction betwcctt uttitttttliu {ittthis contcxt:'aninrals') antl inunittrull«.'l'ho instances founcl in'l'G'l'arc rlot

nrany anrl tcnrl t<l conccntrirto in thc "inanimate" section, to which such

natural clcrncnts as winrl, watcr, earth, thunder, etc. are ascribed.'flrcy can

recur in vitlious colìtexts: wincl an<J water are quoted three timcs, fire twicc.

Man is ntclltiotlcd, olcoursc, anlong animate beings, to which also trees ancl

grass belong, not a common inclusion in the classical gramrnatical tradition.

Anal<rgics, although vague, can be found e.g. between v(ttnu hylr Ltnd

undarwn ltulsus, and musical instruments occur in both tcxts. But thc

Icclandic treatise contains a longer passage regarding music which is tlot itl

Probus. Other affinities may perhaps be found, but as thcy refer to ge neral

or natural phenomena, they are not enough to prove any concrctc rclation-

ship between the texts.

2.3. lt may be interesting to compare Ol6fr's classification of souncls with

that set forth in the Second Grammatical Treatise (: SGTI all quotations and

translations from Raschellà 1982):

sGT (Raschellà 1982:50,2-17) pat er ein grein hlj6Òs, cr frltr vcòr, ctla vatrt còa

sirr cÒa bjòrg eÒa jòr<I eòa grj6t hrynr; petta hlj6Ò hcitir gnlir ok prynrr ok

dunur ok dynr. Sv6 pat hlj6Ò, er màlmarnir gcra, eÒa mannapyssinn; firt heitir

ok gnyr ok glymr ok hlj6mr. Svii pat ok, er viÒir brotna eÒa vhpnin nletast; petta

heita brak eÒa brestir, eÒa enn, sem ziÒr er ritaÒ. Allt eru petta vitlaus hlj(rÒ. En

hér umfram er pat hlj6Ò, er stafi eina skortir til màls; pat gera hòrpurnitr ok e nn

heldr hin meiri sòngfaeri; en pat heitir sòngr. Ònnur hljoÒsgreirt cr sù, scm

fuglarnir gera eÒa dyrin ok sakvikindiir; pat heitir ròdd, en pitr radclir lreita /r

marga lund. (...) Allar pessar raddir eru mjòk skynlausar at viti flcstra marìniì.

En priÒja hljdÒsgrein er sÉ, sem menninir hafa; pat heitir hljoÒ ok rijdcl ok nrhl.

(. . .) En hverju orÒi fylgir minni ok vit ok skilning'

In the treatise. said to have been written between 1270 and 1300 (Iìaschcllà

1982:130), three main categories of sound are distinguished which, as point-

ed out by Raschellà (1982:78), "are presented in a strictly hicrarchical

progression following a growing order of 'significance"'. we start from the

kind of sound produced by inanimate entities, called vitlaus'irrational', then

we find animal cries, defined skynlausar'senseless'. From the point of view

of their significance, the first two categories may be regarded as one, since

no clear-cut distinction can be made between the terms irrationul ttnd

senseless. This sound class is contrasted with that represented by human

speech, about which we are told that "hverju orÒi fylgir minni ok vit ok

skilning", that is, "each word is accompanied by memory, sense, and dis-

cernment". It appears that we are confronted with an implicit opposition

"rational/irrational", which can also be traced in TGT:

Classical Tradition and Norse Traclition in the "Third GrarÌrmatical -freatisc"

75

1,1t)-11 J lifligvm lvtvm ok vitlivsvm vcròr lioÒ, siem iviÒvm ieÒa grasvm ok [traf lrritring nokkvrs reriligs likama. Af lifandi lvtvm peim tr skvn hafa veròr

annat lioò, pat rr rocld hieitir, itnn annat, bat ltr itigi ier rgdC, stnr fota stitl.r1r

rrÒa handa klapP ok annat slikt.

In TGT, too, the opposition is not expressed symmetrically since an adicc-

tive, vitlus 'irrational' is put in contrast with a periphrasis, af liJ'untli lvtvttt

peim cr skyn hul'u ("from livings beings endowed with reason") lt is

ìnteresting to I'totc that, althtlugh thc twtl classifications may not coirlcidc'

the examplcs provided show many points of contact' Practically all of those

quotcd in fCT occur uls<t in SGT, nlaking allclwance for the voices of

animals nevcr rncntioned in ]-GT. Anothcr featurc shared by the twtl

Icelanclic trcatises is the prominent positicln granted to music. The general

imprcssion is that both texts clraw lrom the same sourcc(s), redistributing thc

material in diffcrent waYs.

3. ln his dcscription of voice as a kincl of sound proceedir.rg from "living

beings enclowccl with reason", Ol6fr makes an interesting referencc to the

organs of speech:

1,12 Rodd ar lioÒ fram frcrt af kvikvrndis mvnni, formitrat af .ix. natvrligvnr

tolvm, lvngvnr ok barka, tvngv ok tveim vgnvm ok.iiij. tQnnvm'

The phonation process is described as the rcsult of thc performance of nine,,natural instrumcnts" . Instrurnenturz is a common term among early gram-

marians t0 indicate the parts of the human body concurring to the produc-

tion of souncl, not only in the usual sense 'means, tool" but also with a

metaphoric implication, meaning 'musical instrument'. This is of course a

consequencc of the analogy bctween the production of voice and that of

music. An African mythographer, Fulgentius, explicitly compared the

speech organs to musical instruments already in the fifth century A.D. (cf.

Melazzo 1985422):

Filg. Myth.I, XV (Helm 1898:25) Duo labia uelut cimbala uerborum commoda

moduluntiu. lingua ut plcctrunì quae curuamine quodam uocrtlem format spiri-

tum, palatum cuius concauitas profcrt sonum, gutturis fistula quae tereti mca-

tum spiritalem praebet excursu et pulmo qui uelut aerius follis concepta reddit

ac reuocat.

In the ninth century Sedulius Scottus writes a commentary to Donatus'Ars

Maior containing the same comparison as Fulgentius (4, 30-39, Lòfstedt

1977), while at the end of the century Remigius Autissiodorensis. a commen-

tator from Auxerre, includes the same reference to speech organs in his

commentary to Donatus Minor (17, 15-18, 2, Fox 1902). In the so-called

Commentum Einsidlense in Donati Artem Maiorem, another text belonging

to the tenth century, sometimes also ascribed to Remigius, again we find the

\L

76 Vrlcriir Micilkr

same musical simile (Hagen 1870 220). The above references arc itlso clutltecl

in Raschclliì (1982) and Mclazzo (1985).

Later in time, in the thirteenth century, Michacl of Marbais and Vinccrltof Beauvais also speak of "natural instruments" in relation to the phonatoryorgans (Thurot 1869:41=42,135; Òlsen 1tltl4:35, fn.). The lattcr uscs Pctrus

Helias' commentary to Priscian as a source for book II of his Spcculunt

Doctrinule.

3.1. Icelandic grammatical writing also provides a refcrcncc ttf thc samc

type. In SGT another kind of musical simile is developed in an cxtremelyoriginal way:

SGT (Raschellà 1982:54, 24 26 ancl'72,7(ì*76) Muòrinn ok tungan cr leikviillrorÒanna; ii peim leikvelli eru reistir stafir peir, cr nrf,l allt gera, ok hcndir nrhlit

ymsa svÉ til at jafna sem hòrpustrengir, eòa cru ltestir lyklar i sirnphtinic. (...)Stafasetning sjii, sem hér er rituò, er svir sett til nti'rls, scm lyklar til hlj<iòs i

mÉsika, ok regur fylgja hlj6Òstirfum svit, sern peir lyklIar nriilstòf]unr. Mhlstafireru ritaÒir mcÒ hverri regu bieÒi fyrir ok eptir, ok gera pcir rli'rl af lrcnclingurnpeirn, sem pcir hafa viÒ hljdÒstafina fyrir eòa eptir. K0llurl vér frrt Iykla, scnr

pcir eru i fastir, ok eru peir svii settir hér i spiiziunni, sem lyklar i simphtinic, ok

skal ireim kippa eòa hrinda, ok drcpa svit regustrengina, ok tekr ph pat hljoÒ,

sern pu vilt hatt hafa.

This very peculiar image, which also includes two figures where all thc lctte rs

and their possible combinations are placed, does not compare either with the

traditional descriptions examined above, or with the vcry brief refercnce in

TGT to the "natural instruments" producing human specch. Howovcr it is

interesting to note that here, once again, the two Icelandic treatises appcar

to echo a common knowledge, although elaborated in different ways.

3.2. A clearer correspondence with the passage in TGT under considerationis shown in Roger Bacon's Summa Gramatica.In defining voice, the Englishphilosopher and theologian employs expressions which closely recall those

used in Ol6fr's treatise:

R. Bacon, Summa Gramatica (Steele 1,12 Rodd rr lioÒ fram fart af

1940233) Alio modo [i.e., proprie] di- kvikvendis mvnni, formitrat af .ix.

citur vox sonus prolatus ab ore anima- natvrligvnt tolvm, lvngvm ok barka,

lis, (...) naturalibus instrumentis for- tvngv ok tveim vgnvm ok .iiij.matus, que sunt hec, pulmo, guttur, tQnnvm.

dentes, lingua, palatum, labia.

The two passages are practically identical and even the speech organs are

mentioned in almost the same sequence, inside to outside. Minor differencesare the loss of palatum in the Icelandic text and the displaccmcnt of the teethto the end of the sentence. It is evident. however. that both authors draw

Classical Tradition and Norse '[ì'aclition in thc "Third (ìranttnittical '['rcrtise " 11

from a common source or, possibly, frttm diffcrcnt sources lrelortuine to thc

samc tradition.

4. lmrnediately after the passage quoted above, Ol6fr introduces a tle finititrnof 'vtrico'taken from Priscian's InstitLttiones.l print it bclow togcthcr ri'iththe Latin sourcc:

[.13 ,E'nn priscianus kallar rodd vera Prisc. /n,rt. I,1 (lI .5) Philosophi clefi-

hit gnrnnligstlr lopzirrr lrugg ok uigin- nrunt. voccnt cssc acrerìr tenui:sinturtt

liga ryrvrr-r skilianlikt. ictum vel suum sensibile auriuni. idest quod proprie auribus accidit.

There is n<l worcl-for-word correspondence between the texts but thcrcseems to bc no cloubt as to the origin of the definition. A few differencesmust bc pointecl out. Onc regards the authorship of this clescription of'voice', which is ascribed to "philosophers" in the Latin text, but to Priscian

hirnself by Olifr. Another important divergence concerns the meaningattributcd t<t rr2tld on the one hand, and to vo-r on the other. In fact it is clearthat Priscian is talking about sound in gencral, while Olàfr is referringspecifically to tho human voice, as can be seen from the preceding context(see quotation 1,12). Finally, a basic difference lies in the definition itself.Priscian, likc Donatus and the majority of the ancient grammarians, states

that the'voicc' is aer ictus, that is, 'whippe<1 air'. To Ol6fr the rqtldis lopzirtshatgg,the 'percussion'itself, the 'whipping'. This is not a minor difference toancicnt grammarians, as results e.g. from a passage by Audax, alrcadyquoted abovc:

Autl. lixccrptu: De vtsce (VII 323) Vox corporalis cst, an incorporalis'l Secun-

dunr stoicos corporalis (. ..) Plato autem non esse vocem corpus putat: "non

cnim pcrcussus", inquit, "aèr, scd plaga ipsa atque percussio, id est vox".

The quotation rcports Plato's opinion about the nature of the vox which, he

says, "is not whippcd air, but the whipping itself and the percussion".'lhissame interprctation is found in one ars grammatica in Cod. Bern. 2:

Papias Dc arle gramm. ex Prisciano excerpta: I)e voce f.136'col. II (VIII clxxxi)Vox cst acris tenuissimi ad linguam percussio uel quod proprie auribus accidit.

On thc othcr hand, Sedulius Scottus reports both this definition and the

traditional onc:

Sed. Scot. in Prisciwtum (LÒfstedt 1977:rr5) ltuque quia non omnis tcnuissimus

aer, nisi quolibet pulsu percussus fuerit, uox est, signanter addidit, non solum

uocis aerem esse tenuissimum, sed et ictum, siquidem nulla uox nullusque sonus

fieri potest, si non aliquo pulsu aeris tenuissimi efficiatur. (...) Itlcirco definitursonu\: pcrcu.\sio aeris in indissolttta usque ad auditum. (my italics)

7rì Vulclirr Micilltr

5. The last part of I)riscian's clefinition of vo-r contuins <tnc ctttttrtlvcrsi:tlphrasc, vcl suurrr sen.sihilc ouritutr, id est quotl pr<sprie utrribtt.s ttt'r'idit.

Ancient colnmcnl.dtors already removecl part of the difficulty by acltlltting yt'l

sonunt sansibile uurirtrn instcacl of vcl sLtum etc. TG'I' show's yct lttttlthcrvcrsion: aiginliga cyrvm skilianlikt,literally 'propcrly perccivirtrlc by thc

cars'. Òliifr has totally eliminated the first part and has rctainccl thc litst onc.

explaining it by clearer words. He employs an adjective , skilianlikt, rclatecl

to the verb skilja'to distinguish, to discern, to perceivc', which is rcndered

by intelligibilem in the Latin translation of Snorri's Edda in ESS (1,67)' but

may be closer to percipibilis found e.g. in Marius Victorinus 2,1 (Mariotti1967 66) and in others. The whole modification ailns to attititl a bettcr

comprehension of the passage and appears to fall within a series of sirlplili-cation procedures adopted for didactic purposes.

6. A few concluding remarks. This brief investigation is limitecl to a sllort

passage of Olàfr's treatise, though interesting for its sourcc irlplications. Adefinite assessment of the text would need an accurate stucly of thc whole

work. Mcanwhile I will restrict my conclusions to some observatitltls tlf a

gcneral kind.Wc notice first that traditional sourccs such as Priscian and ilthers cl<l

contribute to the composition of the text. Howevcr, the rtllc thoy play, at

least in the chapter considered, appears to be diffcrcnt frclnl wltitt we

expected. We find quotations scattered through thc tcxt rltthcr thitn a

general underlying structure in rvhich other elements arc insertctl. lt is als<r

evident that some material is drawn from a local tradition, prtltlatlly «lltc

source common to both TGT and SGT. All things consiclcrccl, (ili'rfr itppcitls

to be more than a mere collector of quotations. He sccms to clabtlratc

constantly the material available, so that in the end we get a pcculiar picturc

not really comparable with any of the original sources. Several divergcnccs

which can be detected between these and TGT are in my opinion cluc ttl

Oléfr's conscious intervention, aiming to work out a well-balanccd, carclully

constructed Structure where traditional lcarning and more original clclncnts

of various, especially Norse, provenance combine with peculiar elaborrttiotrs.

References

Albano Leoni, F., ed. (1975), Il printo trattato grammaticale islantlesc. Studi lirì-

guistici e semiologici. Il Mulino, Bologna.

Benecliktsson, Hreinn, ed. (1972), The First Grammatical 7'reatise. Institute of Nor-

dic Linguistics, Reykjavik.ESS: Erlrla Snorra Sturltuonar. Edda Snorronis Sturlaei (184t3-1887), I-lll. Sumpti-

bus legati arnamagnaeani. J. D. Qvist, Hafniae.

Classical Tradition and Norsc Tradition in tho "Third (irammatical 'ltcrttise" -19

Fox, W., ed. (1902), Renigii Autissiodorensis in Artent Donuti llirtor(nt (()ttutt(n

tunt. In acclibus B. G. Teubncri, Lipsiae.IIagen. H.. ed. (1E70), sec Keil. H. (1tì57-1880).

Haugen. L., cd. (19722), First Grantnaticul Trcati.sc.'[hc curliest (itrrnuttiL ltlttttrol-ogy. An editiotr, trun.slutiort and conunerùar,y. Lotrgnran, Lonclott.

I{elrn, R., ed. (189tì), Fubii Pkmciudi.s Fulgcntii v.c. Operu. Mitologiururtt liliri rrL't.

In aeclibus B. G. 'ltubneri, l-ipsiae.IIoltz, L. (1981), Donut et la traditktn de I'enseigrtenrcnt grantntutitule. Iltttlc at

édition critiqu('. Ce ntre National de la Recherche Scientifiquc, Paris.

J6nsson, Finnur, udg. (1927), Ottifr Pòròarson - Mdthtjéòa- og tntilskrLiòsrit. (irurrt-

matisk-retorisk «Jhundling. Det Kgl. Danske Videnskabernes Se lskab. Ilistorisk-filologiske Mcddelclscr XIII, 2. Andr. Fred. IIost & Son, Kllbcnhavn.

Keil, K., cd. (llì57-lfltl0), Grumntatici Lutini. I-VII. In aeclitrus B. ('i.'[tutrrreri.Lipsiac.

- (1870), Ylll. Supplententunt continens Anecdotu Helveticu. Ed. II. Ilagen. lnacdibus B. (i. 'Ieubneri, Lipsiae.

LÒfsteclt, Il., cd. (1977), Sedulius Scotttts. In Donati Artent Muiorenr. Brcpols.Turnholti.

Mariotti, I., ecl. (1967), Murii Vitktrinii Ars (ìrunrnatica. Introduziotrc, t(sto critict,t tcontntcnto. Biblioteca Nazionalc --Seric clci classici -qreci e latini, Vl. Le Monrrier,Fircnzc.

Melazzo, L. (19ti5), "'Ì-he opening of thc so-called Second Grantmatical Trcatise : Ittsearch of the sourccs". P. Janni, D. Poli, C. Santini (a c. di), Cultura clussicu tculturu T4cnnaricu scttentrionalc. Quaderni linguistici e filologici, III. Universitrì di

Macerata, Maccrata, pp. 399-42'1.

Otsen, Bjòrn Magnùsson, udg. (1884), Den tredje og fjerde gr(immutiske ufhand-

ling i Snorres Edrla. Islancls -urammatiske litteratur i middelaldcren, 2. Samfuncl

til udgivelse af gamnrel nordisk litteratur, Kobenhavn.

Raschellà, F. D., ed. (1982),'l'he so-culled Second Granttnatical 'l'reatise. FilologiaGcrmanica. Testi e studi, II. Le Monnier, Firenze.

Steele, R., ed. (19a0), Opera lructenus inedita Rogeri Bacon. XY. Surnnru Grarrtatictt

magistri Rogeri Bacon necnon Sumule Dialer:tices magistri Rogeri Bacon. E typo-grapheo Clarendoniano, Oxonii.

Thurot, Ch. (1869), Notices et ertruits de divers manttscrits latins pour servir ti

I'histoire des doctrines grummaticales au moyen lge. Imprimerie Impériale, Paris.

Repr. Minerva GmbH, Frankfurt a.M. (196,1).