On the Recession of the Latin Accent in Connection with Monosyllabic Words and the Traditional Word-Order 9781463222130

Citation preview

O n the Recession of the Latin Accent in Connection with Monosyllabic Words and the Traditional Word-Order

A n a l e c t a Gorgiana

380 Series Editor George Anton Kiraz

Analecta Gorgiana is a collection of long essays and

short

monographs which are consistently cited by modern scholars but previously difficult to find because of their original appearance in obscure publications. Carefully selected by a team of scholars based on their relevance to modern scholarship, these essays can now be fully utili2ed by scholars and proudly owned by libraries.

O n the Recession of the Latin Accent in Connection with Monosyllabic Words and the Traditional Word-Order

R. S. Radford

1 gorgias press 2009

Gorgias Press LLC, 180 Centennial Ave., Piscataway, NJ, 08854, USA www.gorgiaspress.com Copyright © 2009 by Gorgias Press LLC Originally published in All rights reserved under International and Pan-American Copyright Conventions. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise without the prior written permission of Gorgias Press LLC. 2009

1

ISBN 978-1-60724-629-9

ISSN 1935-6854

Extract from The ^American Journal of Philology 25 (1904)

Printed in the LTnited States of America

I I — ON T H E R E C E S S I O N O F T H E L A T I N A C C E N T IN CONNECTION WITH MONOSYLLABIC WORDS AND THE TRADITIONAL WORD-ORDER.1 PART

INTRODUCTION

I.

AND

PROBLEM.

The present study was originally undertaken with a view to determining the probable cause of the frequent syllable-shortening- which occurs in early Latin verse in connection with short monosyllables, i. e. sed ilium?, sed autem; in its present form, however, it will be found devoted chiefly to the preliminary task of determining the place of the grammatical accent in the wordgroups ^, « — , ^ - and w >-> For it is evident that, after determining the place of the grammatical accent in the flexible tribrach groups sed ea, sed eni(vi), etc., we shall be in a much better position to determine how far the accents sed ilium, s£d autem and the like are due to their analogy. 1 T h i s paper is an extension of a preliminary study on the same subject, an abstract of w h i c h appears in the J o h n s H o p k i n s U n i v e r s i t y Circulars, V o l . X X I I I . A study of the a c c e n t of the trisyllabic word-groups occurring in T e r e n c e has been published by the writer in T r a n s a c t i o n s A m . P h i l o l o g . A s s o c . X X X I V (1903), pp. 60-103. 1 T h e scansion of ille, itnmo, quidquid, nequis, hocin, etc., in some other cases as words of two morae, constitutes a w h o l l y distinct problem. U p o n this question the general principles laid down b y Skutsch must be accepted as final; at the same time Skutsch's conclusions appear to me to call for certain modifications. T h u s metrical theory does not require us to assume that a w e a k final short s y l l a b l e w h i c h is neglected in iambic verse, i. e. illtc, or a w e a k medial short syllable w h i c h is n e g l e c t e d in anapaestic verse, i. e. ferdidi, suffers absolute ' a p o c o p e ' or ' s y n c o p e ' ; it is sufficient to assume that such a syllable was greatly w e a k e n e d in pronunciation. I n g e n e r a l the colloquial iambic poets avail themselves of this license w i t h a definite purpose, viz., in order to preserve unbroken the traditional word-orders, w h i c h exist in connection with the sentence-introducing pronouns and conjunctions, i. e. ille munde til, etc. A study of the question from this point of v i e w w i l l be published elsewhere, and it w i l l be sufficient to point out here that, just as the existence of nempe is disproved by the non-occurrence of nemp(e) with elision, so the existence of ecquis, etc., is disproved by the nonoccurrence of ecqu(a), siqu(em), etc.

148

AMERICAN

JOURNAL

OF

PHILOLOGY.

The question of the recession 1 of the accent upon monosyllabic words which are closely connected both in sense and in pronunciation with some following word, is not a new one. Thus Ritschl in the sixteenth chapter of the Prolegomena, the chapter entitled ' d e Accentu L o g i c o ' , gives the rule (p. C C L V I I I f.) that, so far as regards the accent, an iambic or pyrrhic dissyllable may coalesce with a preceding short monosyllable to form a trisyllabic word, i. e. prò equo, ét erus, quid agam. While Ritschl speaks prevailingly in this discussion of metrical coalescence, due to the fact that two short syllables are closely connected in forming a resolved arsis, there can be no doubt, I think, that he means to imply also a real coalescence in ordinary pronunciation. Valuable, however, as is Ritschl's study of this whole question, his treatment must appear to a critical student of the Latin accent at the present day as wanting at times in definiteness and precision. For the critical student must not only consider the question primarily from the view-point of actual coalescence, but must apply the necessary tests to determine the law of the accent in whole series of word-complexes and groups. Ritschl does not attempt to apply such tests, and it is doubtful whether he recognized the operation of a definite law in these processes. True, he not only holds, as has been already noted, that prò equo, èt erus, quii agam, etc., are accented as trisyllabic groups, but he correctly declares (Proleg., p. C C L X I ) that the accent of dé ìlio, ét Iste, etc., is determined by the same principle, yet upon turning to p. C C L I I I of the same chapter of the Proleg., we are perplexed to find a supposed example set éa (Trin. prol. 10: sei éa quid hùc) quoted in illustration of the thesis that monosyllabic particles of trite use and little weight are rightly placed extra arshn. If this view were correct, it would appear that the dramatists had known two forms of accentuation in the trisyllabic word-groups in question, viz. set eà and set éa, which is far from being the case. For, with a single exception, which is only apparent (Cap. 329: ut éa-quae), the 27 cases 2 occurring in the drama1 For the sake of convenience the term ' recession' is here employed in general of the initial accentuation, w ; J-, w ^ , although it is not in all cases strictly applicable, see below p. 161. * Viz., Ba. 203 ; 472; Cap. 970; 9 4 2 ; C i . 7 4 2 ; E p . 265 ; 532; M e n . 1 8 6 ; Mi. 346; 686; Mo. 160 ; Poe. 1015 ; 1265 ; Ps. 2 7 7 ; 1087; Ru. 1081 ; T r i . 330; 1168 ; A n . 337 ; 837 ; H e . 191 ; 334 ; Eu. 926 ; Ph. 480; 1015 ; cf. 1046 ; T i t i n . com. fr. 98; cf. A t t . tr. fr. 432. Cf. also Seyffert's observation, Stud. Plaut., p. 27 n.,

LATIN

WORD-GROUPS.

149

tists in w h i c h ea, id et, earn In etc., are p r e c e d e d b y a short m o n o syllable, all s h o w the recessive accentuation sit ea, etc., a n d the r e a d i n g set id q u i d hiic q u o t e d b y R i t s c h l from T r i n . prol. 10 is o n l y an unfortunate c o n j e c t u r e of B o t h e ' s for the M S r e a d i n g set ea hiic q u i d introierit, just as F l e c k e i s e n m a k e s a similar inadmissible c o n j e c t u r e sed ia servibat, P h o r m . 83, for the M S reading ea s6rvi6bat. 1 T h e r e m a r k a b l e uniformity which a p p e a r s to exist in the accentuation of sit ea and similar g r o u p s s u g g e s t s an inquiry into the g e n e r a l tendencies of the republican accent. THE

REPUBLICAN

ACCENT.

T h e r e are t w o periods of the L a t i n l a n g u a g e , the accent-laws o f which admit of b e i n g reconstructed e v e n in minute detail. T h e first of these is the republican period, the accents of which are preserved in the d i a l o g u e verse of the dramatists; the s e c o n d is the p e r i o d of v u l g a r L a t i n which g a v e birth to the R o m a n c e l a n g u a g e s and has left its accents e m b e d d e d in the R o m a n c e forms. T h e s e t w o p e r i o d s are separated at their furthest limits b y an interval o f n e a r l y a t h o u s a n d y e a r s , and the accentual c h a n g e s w h i c h t o o k place within this l o n g period of time are n u m e r o u s and in s o m e cases far-reaching, y e t so s l o w l y is each single c h a n g e of accentuation effected in the s p e e c h of a p e o p l e 2 that it s e e m s possible to trace with s o m e precision the history of almost all the important c h a n g e s in the L a t i n accent which o c c u r r e d s u b s e q u e n t to the time of Plautus. A c c o r d i n g to the v i e w s w h i c h are held b y most accentual scholars the L a t i n accent rested u p o n the initial s y l l a b l e of w o r d s and g r o u p s until a time s h o r t l y before the b e g i n n i n g of the literary p e r i o d (see the references g i v e n b y S t o l z in Miiller's H a n d b u c h II 2, p. 101 if., 3 A u f l . ) . T h u s in the time of Plautus the initial accent law was a l r e a d y s u p e r s e d e d , but its effects w e r e still v e r y distinctly f e l t ; that the c o m m o n formula " q i l i s - h i c (hdec)-est?" second syllable.

is n e v e r a c c e n t e d on the

1 N o t a d m i s s i b l e then are the a c c e n t s m a r k e d b y H a u l e r in his edition of the Phormio : prol. 8 et earn; v. 284 i t a e u m , 605 si a b eo. 5 T h u s in our own l a n g u a g e in the case of m a n y words d e r i v e d from the L a t i n l i k e confiscate, contemplate, demonstrate, etc., the contest b e t w e e n the L a t i n a c c e n t and the E n g l i s h recessive t e n d e n c y h a s b e e n g o i n g on since S h a k s p e r e ' s time a n d is not y e t f u l l y at an end, although in other words of

this class, i. e. obdiirate, oppdrtune, contrdry, sepiilchre, etc., (Abbott, spearian G r a m m a r , §490) the L a t i n a c c e n t has l o n g d i s a p p e a r e d .

Shake-

ISO

AMERICAN

JOURNAL

OF

PHILOLOGY.

we shall best describe the republican accent then as strongly recessive in its nature and as seeking every opportunity of recession within the limits of the three-syllable law. T h e extreme recession of the accent is shown in three classes of words : I. In the inseparable composiia which were usually written by the Romans as single words, i. e., expllcat, rent gat, compater, desuper, etc. II. In the separable composita} Under this head I include first the recession in those separable composita which were not infrequently 2 written by the Romans as single words, or whose parts were sometimes joined together by the ifo'v, i. e., cale facit, inter esse (inler erif), circum venit, proptereos, inforo, quamdiu, intered loci, ante volans, ante tulit (the last two written with the ifpiv by Donatus, Keil, IV 372, 2 ff.=Scholl, De acc., p. 92 ; Diomedes, I 434, 36 ff. = Scholl, 1. 1., p. 95 ; Max. Vict., V I 193, 28=Scholl, 1. 1., p. 98) ; and secondly the recession in those combinations which were only occasionally written together by the Romans, and which can only be called composita in the broadest, that is, in the ancient sense of the term j these latter include all the common phrases of the spoken language and also, to a large extent, as we shall see later on, the traditional Latin word-orders, i. e. sedenim (Priscian, Keil, III 93, n f . ) , ctirita, quidego, quidea, netime, non potest, haiiscio, certo scio, etc. III. In the regular accentuation of quadrisyllabic words beginning with three shorts ( y ^ ^ —, proceleusmatic and fourth paeon words) upon the initial syllable, i. e.fdcilius (cf. Stolz in Miiller's Handbuch II 2, p. 101). • F o r a definition of composita, see the locus classicus in Priscian, K e i l , I I 177, 15 ff., and for a discussion of the separable composita see Priscian, I I 183, 12 ; I I I 113, 6 ; ib. 413, 14. T h e ancients included all prepositional phrases among the composita, as is e v i d e n t from Charisius I 17, 3 ; D i o m . I 436, 15 ; Dositheus V I I 389, 4 ; ib. 409, 27, etc. 2 T h i s and subsequent references to the R o m a n system of word-division as extremely fluctuating and uncertain (see, e. g., C I L . I index, p. 609 f.) are intentionally made. T h e practical necessity of a d o p t i n g — o f t e n a r b i t r a r i l y — a fairly uniform word-division in modern texts of L a t i n authors obscures for most readers the w h o l e subject of the L a t i n word-division and of the L a t i n separable composita, and is often misleading even to the critical student. The subject calls for a fuller exposition, but I can only refer here to my brief discussion of the L a t i n word-division in T r a n s . A m . Phil. A s s o c . x x x i v 97-100, and to E y s s e n h a r d t ' s reproduction, to some extent, of the variable worddivision of the M S S in his edition of Martianus C a p e l l a ( L e i p z i g , 1866), viz. et enim and ctcnim,praeter ea and praeterea, etc.

LATIN

WORD-GROUPS.

In the Romance languages we find the Latin composita, both separable and inseparable, preserved in great abundance, and we observe further that many of the composita, which were separable in old Latin appear only as inseparable composita in Romance, i. e. Span. tambien=La.t. tarn bene (tambene), Span, tampoco — Lat. tarn paucum (tampaucum,) Span. ninguno= Lat. nec units (necunus) ; cf. Corssen, Ausspr. II 2 , p. 890. T h e coalescence in pronunciation of the more common monosyllables with the following word is regularly indicated also by the writing in literary Italian and by the doubling of the initial consonant of the second word, i. e. ellui=etlui, ebbene=etbene, checcosa=quidcausa, etc. (Meyer-Liibke, Gramm. d. roman. Sprach. I, p. 508). But at this point all similarity ceases; for, as regards the form and accent of the composita, the Romance languages, as is well known, proceed from a period of thorough-going ' re-composition ' ('de-composition'); cf. G. Paris, Rôle de l'accent Lat. dans la langue française, p. 83; Meyer-Liibke, 1. 1., I, p. 495; Seelmann, Aussprache des Latein, p. 58 ff.; Lindsay, Lat. Lang., p. 199 f.; Stolz, Hist. Gramm. d. lat. Sprache I, p. 188. It is of course not to be denied that single cases of recomposition occur in the oldest literature; thus Stolz, Hist. Gramm. I, p. 187, quotes expars Turpilius, cited by Nonius II 138, 29 Miill.; requaereres Plaut. Merc. 633; conquaesivei C I L . I 551 ; conquaesiverit C I L . I 198, 38 etc., as well as late Latin inscriptional forms like reddedi C i L . V I 3, 20029; condederunt ib. 18850; but it was only at a very late period and only after a long conflict that these processes of disintegration and recomposition finally prevailed over the earlier tendencies towards composition and recession. Since, then, the Romance forms belong to a period of thorough-going recomposition, they commonly show the fall of the accent in the three classes of words just enumerated: I. In all inseparable composita in which the original composition was still felt, 1 i. e., Late Lat. explicat, Fr. esploie ; renégat, Ital. riniega, O. Fr. renie ; compâter, Fr. compère, Span, compâdre ; desûper, Fr. desure ; *ad prope (in old Latin regularly *âd prope, like the compound adverbs and prepositions : dé super, in super, dé /oris, à foris, âd foras, post modo; compare, for the last, Servius ad Eel. I 30), Ital. apruovo, O. Fr. a pruef; *in fo'ris, Ital. infuori; 1 O n l y in cases w h e r e the original composition w a s no longer felt, w a s the recessive a c c e n t upon the prefix retained, i. e. cdllocat, F r . couche, etc. ; cf. G . Paris, 1. 1., p. 83.

152

AMERICAN

JOURNAL

OF

PHILOLOGY.

*assàtis, Ital. assai, Fr. asséz ; *de novo (in old Latin only dé11110), Fr. de nouveau. II. In almost all the separable composita, i. e .

Late

Lat.

adpédem.

(appédem,)

Ital.

appiè

;

iamdiu,

Fr.

jadis; insinu, Ital. insino ; et bène, Ital. ebbene; si bène, Ital. sebbene; tam bène, Span, tambien, etc. III. In the case of words like facilius the change from the pro-antepenultimate accent was completed as early as the middle of the second century A. D.1 In spite of this thorough-going 're-composition,' traces of the old Latin recessive accent have been retained in the Romance languages in the case of a few separable composita. Thus the Romance forms derived from the cardinal numerals viginti, triginta, etc., presuppose a Latin accent upon the antepenult, and are consequently derived, according to d'Ovidio, Ztschr. f. roman. Philol. V I I I 82 ff. and Skutsch, Forsch., p. 160 ff., from Latin composita like viginti-minae, trigintd-dies, etc.2 More direct in its bearing upon the present study is the retention of the recessive accent in the prepositional composita cümìlla,

dé lila,

as s h o w n

by

the

Italian

colla,

della?

cf.

Corssen,

Ausspr. II 2 , p. 889 and Skutsch, Forsch., p. 158; it is noteworthy that this accentuation of cdm-llla agrees also with those statements of the grammarians (Schöll, De acc., p. 192 f.; Seelmann, Ausspr., p. 41), which show that delude,pcrlnde, etc., ' A study of this accentuation which I have made complete for the poets of the empire shows that, after the middle of the second century, the iambic poets completely banished the accent fdcilius from all parts of the verse except the difficult verse-close. - Here the traditional word-order, in accordance with which the cardinals are usually prefixed to their nouns (v. Delbrück, Syntakt. Forsch. I I I 35, and Schmalz, in Müller's Handbuch I I s 2, p. 464) has caused recession of the the accent even upon polysyllabic words, i. e., the accent trlginld arises whenever the qualified noun is an iambic word, that is, equally in all combinations of numeral and iambic noun, and in the rare trigintd-manus as w e l l as in the frequent trigintd-minas. I n consequence of the extreme frequency of some of these combinations, the accent triginta alone is represented in the Romance derivatives. 3 By the side of these forms, as Dr. J. E . Shaw has kindly suggested to me, may be placed alia and dalla, both older combinations than colla and della. T h e question is still open with Romance scholars as regards such a derivation as this, or that proposed by d'Ovidio, A. G. It. I X 71, n. (cf. Meyer-Lübke, Z. R . Ph. X X I 328 f.). O n the other hand, the fact that so large a part of the total use of ego in the colloquial language consists of its occurrence in word-orders like ét ego may possibly have something to do with the numerous Romance atonic forms like Span, yo, Fr. je.

LATIN

WORD-GROUPS.

153

were the common colloquial pronunciations of their time; similarly we find id lili, Büch. C L E . 130, 2 in a poetical inscription as late as the year 50 A. D. (V. Biicheler's note), pointing probably to the persistence of the grammatical accent id- lili, ét- lili, etc. Again the sentence-introducing conjunction et became tonic in old Latin in such sequences (traditional word-orders) as ét-ego, ét-ea, ét-mihi, etc. A t the present day, to be sure, the Romance languages preserve in general only the atonic forms of et, but I owe to the kindness of Prof. C. C. Marden the information that old Spanish has preserved also the tonic forms ye, ie.1 Similarly the interrogative, relative and indefinite monosyllabic pronouns, which are usually atonic became tonic in classical Latin in a vast number of sequences, i. e. quid agis, quód agis, siquid agis,2 etc.; here the Romance languages have preserved both forms, i. e. tonic lSee also R . M e n é n d e z Madrid, 1904, p. 212.

Pidal, M a n u a l elem. de gram, histórica españ.,

2 T h e verse of the dramatists affords the strongest possible evidence of the tonic character of the relative and the indefinite as w e l l as of the interrogative pronouns in these and all similar word-orders. H e n c e I follow Schöll, D e acc., p. 67 (cf. N e u e - W a g e n e r , Formen!. I I 3 , p 430) in rejecting absolutely the teaching by w h i c h the L a t i n grammarians attempt, in imitation of the G r e e k accentual system, to distinguish sharply between the accent of the interrogative and the relative pronouns, attributing the acute invariably to the interrogative and the grave to the relative. A s is w e l l - k n o w n , the grammarians are not thoroughly consistent here ; for while they imitate the Greek distinctions in their statements about the interrogative and relative pronouns, they are by no means agreed in their accounts of the accent of the indefinite pronoun, since Priscian, X I I I 3, 13 f., states that the interrogative and the indefinite (!) qui h a v e the acute accent, the relative qui the grave. W h i l e the interrogative pronoun no doubt naturally receives the accent somewhat more frequently than the relative or the indefinite pronoun, a thorough-going distinction in the accent of the several classes cannot reasonably be maintained ; see the excellent remarks of Schöll on this question (1. 1., p. 67): " P r o n u n t i a t i o autem non solum pronominum, sed omnium fere vocabulorum quodam modo immutatur acriusque intenditur in interrogationibus, ut naturae non sit consentaneum hac re propriam quandam pronominum speciem i n s i g n i r e " . Cf. also Corssen, Ausspr. II 2 , p. 810, on the Latin imitation of the G r e e k distinctions seen in noZof, TTÓÚO^ and Trotón, irooótetc. L i n d s a y , Class. R e v i e w V (1891), p. 402, also speaks somewhat d o u b t f u l l y of the distinction inculcated by the grammarians. T h e very acute observations of W e i l , Order of W o r d s 3 , E n g l , tr., p. 88, have not convinced me that Schöll's argument is inapplicable to the classical languages. T h e grammarians' ordinary rule is also refuted by Donatus ad T e r . H e c . V 4, 25: Nüm quid díxti meó patr{~>\ Num aliquid: acuendum ergo quid.

154

AMERICAN

JOURNAL

OF

PHILOLOGY.

L a t . quem, S p a n , q u i e n ; atonic guem, S p a n , q u e (cf. S e e l m a n n , A u s s p r . d. lat., p. 5 7 ) ; tonic L a t . quid, F r . q u o i ; atonic quid, F r . q u e (cf. M e y e r - L i i b k e , G r a m m . d. roman. S p r a c h . I 506). It m a y be confidently e x p e c t e d that w h e n the p r o b l e m presented b y the n u m e r o u s d o u b l e forms of the pronouns, prepositions and conjunctions in R o m a n c e ( M e y e r - L i i b k e , 1. 1., p. 504 if.) has b e e n m o r e fully s o l v e d than is the case at present, still other cases will be noted of the p r e s e r v a t i o n of the old Latin accent. Y e t no l a r g e number of s u c h e x a m p l e s can b e l o o k e d for in the R o m a n c e l a n g u a g e s , and, a c c o r d i n g to the view w h i c h I h a v e s o u g h t to present here, t h e R o m a n c e forms, which h a v e arisen a t h o u s a n d y e a r s later, cannot constitute, as is c o m m o n l y a s s u m e d , a s o u r c e of the first importance for our k n o w l e d g e of the republican accent or of republican w o r d - f o r m s ; far m o r e important here, in m y j u d g m e n t , is a s t u d y of t h o s e l a w s of the traditional L a t i n w o r d order, w h i c h at o n c e determine the accent of w o r d - g r o u p s a n d justify a free metrical treatment of word-forms. EXTENT

OF RECESSION.

'ENCLISIS'.

T h e present s t u d y of the recession of the Latin accent is limited to the case of m o n o s y l l a b i c w o r d s ; o n l y the dissyllabic prepositions will be included in the discussion. T h e reason for this restriction lies in the fact that m o n o s y l l a b l e s are more c l o s e l y connected in pronunciation with the f o l l o w i n g w o r d than is c o m m o n l y the case with l o n g e r w o r d s ; further, certain wellk n o w n metrical p h e n o m e n a point in a d v a n c e to a c o a l e s c e n c e of the m o n o s y l l a b l e s in pronunciation and the almost total absence of a w o r d - e n d , viz., formations of the iambic a n a p a e s t like quod amas, P h o r m . 504; pol eis, H e c . 788 (cf. R i t s c h l , Proleg., p. C C X X X V I I ; K l o t z , G r u n d z . , p. 307 f.), and the free o c c u r r e n c e of r e s o l v e d arses such as ad eum venit, P h a e d r . I 21, 5 ; cf. H a v e t , ed. P h a e d r . , p. 1 6 0 ; B . S c h m i d t , D e S e n e c a e t r a g g . rationibus metricis, p. 46 f.; L . Miiller, R e s Metr. 2 , p. 169. In g e n e r a l , it is to be noted that a recession of the accent cannot o c c u r a p a r t from s o m e usual w o r d - o r d e r . T h e latter m a y , h o w e v e r , either b e t h e order of s o m e frequent single phrase, s u c h as huius-modi, intered-loci, etc., or it m a y c o n c e i v a b l y be the g r a m m a t i c a l or traditional w o r d - o r d e r which is observed b y all the w o r d s b e l o n g i n g to one part of s p e e c h in relation to the w o r d s b e l o n g i n g to s o m e other part of s p e e c h , i. e., ' d i e tra-

LATIN

WORD-GROUPS.

ditionelle Wortstellung' of Delbrück (Syntakt. Forsche III, 13 ff.; IV, 148 ff.). It is conceivable also that the analogy of phrases and traditional word-orders should in some cases favor a recession of the accent. Wholly exceptional, however, is the peculiar Greek usage which is seen, for example, in the recessive accent of a casual combination like avßpamov nves, and which, no doubt, is an extension, through some long-continued process of development, of the original construction Kara avveaiv, which is seen in the case of avOpcoirov Tiva. I assume for Latin that the accent of iambic words would readily recede whenever these were pronounced in connection with proclitic words, such as the Latin monosyllables can be shown to have been in very large measure, and that under certain conditions iambic words became practically enclitic in Latin. The question may perhaps be asked whether there are not also some formal enclitics in Latin, and whether there are not some cases of recession of the accent due to these. The number, however, of formal enclitics in Latin, i, e., enclitics which, entirely apart from a frequent word-order, throw back an accent upon the preceding word, as may be seen in the Greek ävdpan-öv rives and in a hypothetical Latin *regis enim, is extremely small, and includes only que, ve, ne and a few similar w o r d s ; 1 no student of the Latin accent recognizes the existence of many such particles. The term ' enclitic' is, however, properly and frequently applied by Latin scholars to unaccented and weakly accented words. It has long been recognized that many such ' enclitics' occur in the Latin sentence, but the important researches of Wackernagel have made it possible for the first time to determine these words directly from the Latin sentence. Since the word-order which the enclitics observe has a direct bearing upon the problem discussed in this paper, I shall state Wackernagel's conclusions briefly. Delbrück, Syntakt. Forsch. III, pp. 47, 59, 76, first pointed out 1 Similarly quidem w h e n attached to pronouns b e c o m e s practically a formal enclitic in consequence of the regular word-order, i. e., e'gÜ-quidem, illiquidem, etc.; see L u c h s , C o m m e n t . Prosod. I and I I . In general, however, the existence of formal enclitics in L a t i n is to be d e n i e d ; after monosyllables, to be sure, ego has b e c o m e formally enclitic in locutions l i k e s/d ego, quls ego in consequence of the traditional word-order, but no phenomenon of Plautine verse is k n o w n to me, w h i c h at present justifies the regularity of such enclisis as *reges ego, *püerds ego,*pätres ego,*ämem ego, even when ego in these collocations occupies the second position in the sentence ; hence in no sense is ego a formal enclitic in L a t i n except in association with monosyllables.

156

AMERICAN

JOURNAL

OF

PHILOLOGY.

that enclitic, i. e. unaccented w o r d s a r e d r a w n in S a n s k r i t b y the first w o r d of the sentence, w h i c h is a p p a r e n t l y the most s t r o n g l y accented w o r d , " wie v o n einem M a g n e t , " and this is the case to such an extent that t h e y are r e g u l a r l y found o c c u p y i n g the second position, e v e n w h e n t h e y h a v e n o connection in sense with the introductory word. W a c k e r n a g e l , U e b e r ein G e s e t z der i n d o g e r m . W o r t s t e l l u n g , I n d o g e r m . F o r s c h . , I 333 ff., has since g r e a t l y e x t e n d e d D e l b r i i c k ' s conclusions b y s h o w i n g with g r e a t c o m p l e t e n e s s that the s a m e position-law holds g o o d for m a n y I. E . l a n g u a g e s , so that its e x i s t e n c e in primitive I. E . is p l a c e d b e y o n d all doubt. T h e Latin o r t h o g r a p h y , as is w e l l - k n o w n , e m p l o y s no external m a r k s to indicate enclitic w o r d - f o r m s , and c o n s e q u e n t l y no means of d e t e r m i n i n g the Latin enclitic f o r m s w a s formerly k n o w n e x c e p t t h r o u g h a s t u d y of the atonic R o mance derivatives, s u c h as the reduction of unaccented ille to the R o m a n c e article and of the unaccented me to F r . me, Ital. mi, etc., or t h r o u g h the o b s e r v a n c e of s o m e special p h e n o m e n a , such as the a v o i d a n c e o{ alque 1 and ille in the tonic s i x t h foot of the h e x a m e t e r (cf. L . Miiller, R e s Metr. 2 , p. 277). Wackernag e l ' s researches seem, h o w e v e r , to s u p p l y a direct m e a n s of d e t e r m i n i n g the L a t i n atonies in the precise form in w h i c h t h e y existed in the classical a g e ; in any case, it can scarcely be d o u b t e d that the old association of the s e c o n d place in the sentence w i t h accentual w e a k n e s s is retained to a v e r y l a r g e extent in Latin, as has Ion** been r e c o g n i z e d in the case of f « » , autem, vero, igiiur, quoque, qiddent, etc. (v. the references g i v e n b y W a c k e r n a g e l , 1. 1., p. 406). S i m i l a r l y W a c k e r n a g e l holds, that if the o b l i q u e case of a Latin personal or demonstrative pronoun s h o w s precisely the s a m e peculiarities of position as a G r e e k personal p r o n o u n , whose enclisis is indicated in writing, viz. b y g r a v i t a t i n g r e g u l a r l y t o w a r d s the second position, the w e a k accentual character of the personal or demonstrative pronoun should be cons i d e r e d as established for L a t i n also. T h e s e conclusions, w h i c h w e r e reached t h r o u g h a s t u d y of the w o r d - o r d e r alone, are v e r y notably confirmed in certain cases, as will be s h o w n in detail further on, b y the verse-accents of the dramatists ; for, if w e e x a m i n e at the b e g i n n i n g o f the sentence the e x t r e m e l y sensitive tribrach g r o u p s in which the verse-accentuation must c o r r e s p o n d to the g r a m m a t i c a l accent, i. e. sed ego, sed ea, sed ita, sed 1

O n the w e a k accentual character of the L a t i n conjunctions in general, see

especially the testimony of A u d a x , K e i l , V I I , p. 360, 1 ff.

LATIN

WORD-GROUPS.

IS 7

eni(m), sed eris etc., it appears at once that the initial accent alone was known in these groups to the republican dramatists, and that the second word is here unaccented, i. e. s£d ego, sed eris, etc. T h e case of dactylic groups at the beginning of the sentence, such as sed mihi, atque ita, non ita, is somewhat different, since they are far less sensitive material for accentual investigation and do not admit here the same metrical test; yet it can scarcely be doubted that they were also accented similarly to a large extent in the spoken language. It should be pointed out, however, in conclusion that the only definite relation which the recessive accents sid ego, sed ea, sed mihi bear to Wackernagel's law, consists in the traditional I. E. word-order which the Latin has here preserved; the Latin accent, which results of course from the word-order, conforms in every case to the Latin system of accentuation, i. e. id enim (monosyllable), but more often r£ges enim. (dissyllable) in a non-rhythmical sequence. SEPARABLE

COMPOSITA.

It is not surprising to find that the separable character of many of the Latin composita and the consequent free use of tmesisforms attracted the attention of grammarians even in ancient times and called forth conflicting definitions of ' composition' (Priscian, Keil, III, p. 113, 6; ib. 413, 14; II 183, 12 vs. Orthogr. Bern. II, Keil, Supplement., p. 296, 8; 295, 29). The view here adopted by Priscian, viz., that combinations like respublica, nullomodo, nihilominus, etc., are composita, in spite of the occasional separation of their component parts, is undoubtedly correct upon the whole, yet a real solution of the problem is given only by the use of the historical method and by a study of the various stages through which the word-group passes in the course of its development. In accordance with this method, L e o (Nachr. d. Gottinger Ges., phil.-hist. Kl., 1895, p. 415 ff.) has thoroughly examined a number of Plautine word-groups and clearly set forth the essential principles which regulate their use. The subject still admits of further investigation in matters of detail, but a brief summary, of a somewhat more general character than Leo's discussion, will alone be possible here. In the first place, it may be observed that the use of tmesis-forms was formerly very imperfectly understood, and that it is the especial merit of Wackernagel's investigations to have formulated u

I58

AMERICAN

JOURNAL

OF

PHILOLOGY.

clearly one of the chief conditions under which they freely occur, viz., the weakly accented conjunctions, pronouns and particles, in their straggle to occupy the traditional second position, long retain the power of freely dividing the composita which have otherwise grown into a unity, e. g., Festus 309", 30 M. sub vos placo; Cic. Off. 3, 104 ius igitur iurandum. Secondly, although in our modern terminology we often find it convenient on practical grounds to distinguish more or less sharply between word-groups or phrases and genuine compounds, no clear line of distinction can be drawn between the two classes; see Lindsay, Lat. Lang., p. 361 if.; Stolz, Hist. Gramm. d. lat. Sprach. I, p. 404 if.; Paul, Principles of Language, 2 Engl, transl., p. 371 ff. T h e difficulty of distinguishing sharply between genuine and separable compounds may be said to reach its maximum in the verbs compounded with ante, circum, contra, post, praeter, propter(?), subter and super, which exhibit no clear mark of composition in vowel change and at the same time readily undergo tmesis; they are also especially variable in respect to punctuation in M S S and Inscrr., and it is evident that the grammarians were far from being agreed as to how they should be punctuated; see Marius Victorinus, Keil, V I 23, 7 ff., cf. also Quintil. I 5, 68. With respect to their accentuation, however, it is clear both from the statements of the grammarians and the verse of the iambic and dactylic poets that they were always treated as composita, i. e. inter erit (cf. inter eos), super erit, circiim dedit1, ante volans, ante venit (for the use of the hyphen, see above, p. 150). Again, it is clearly impossible to draw sharp distinctions in this field, when we consider the case of those composita, which were often separated in early L a t i n — most frequently through the influence of toneless words 2 —, but at a later period lost either entirely or almost entirely the power of being thus freely treated. A m o n g composita of this kind belong consue facere, are facere, perferve facere, etc. (examples in Stolz, Hist. Gramm. I, p. 435), fabre facere, lucri facere, manu emittere, animum advertere, super esse (always freely 1 T h u s hexameter closes like V e r g . A e n . V I I I frequent.

474,

cimimsonat

armis, are

2 S e c o n d only to the influence of toneless words in producing tmesis-forms must be placed the metrical necessity of finding a l w a y s an iambus for the closing f o o t ; see, e. g., Seyffert, Berl. Phil. W o c h e n s c h r . 1888, p . 2 3 7 ; N i l s s o n , Q u o m o d o pronomina ap. PI. collocentur, L u n d , 1901, p. 9 ; A s m u s , D e appositionis ap. PI. collocat., pp. 26-34.

LATIN

WORD-GROUPS.

159

separable), male facere, bene facere, etc., palarn facere, magnopere, sacro sanctus, ius iurandum, qua propter, propter ea (for the reversed order ea propter, v. Neue-Wagener, Formenlehre I 676), tarn quam, quo minus, nihilo minus, at tamen, at qui, nescioquis, quid ni, or quinni (examples in Brix-Niemeyer, Mil. 1120), quippe ni (quippini), quomodo, persaepe, quicumque, etc.1 E. g. Amph. 815 qua istaec propter; Cure. 85 super illi fuerit; Cas. prol. 21 opere magno (other examples in Neue-Wagener, Formenl. II 607 f.) Amph. prol. 84 quive alter quo placeret fecisset minus; Rud. 946 at pol qui audies; Aul. 71 nescio pol quae . . . intemperiae; Cas. 370 per pol saepe peccas; Pers. 210 quoi pol quomque, etc. In conclusion I am far from denying that the modern distinction between proper and improper compounds is in some cases a valuable one, yet it is necessary to add that the number of improper compounds and word-groups in Latin is much greater and the number of proper compounds much smaller than is commonly supposed; it may even be doubted whether the latter class includes forms like detorquet and peragit, which even in the Augustan poets sometimes exhibit a species of tmesis, which is produced by the verse-caesura 2 and perceptible to the ear alone (L. Miiller, Res Metr.2, p. 458 ff.). Hence for the purposes of the present study, which freely employs the ancient terminology and has the accent chiefly in view, sedenim (Priscian, Keil, III 93, II f.), inforo, quid ego, quid ea, etc., are to be regarded as separable or improper composita in the same sense as circtim dare, are facere, huiusce modi, etc.; for it is undeniable that in actual use these phrases have often come to denote a single concept, as Priscian's definition requires (Keil, II 177, 15 ff.), i. e. sed enim = sed, quid ego = quid, while with inforo we may compare the English word-complexes indoors, downstairs, and the American downtown; see other examples in Paul, 1. 1., p. 367 ff. So far, however, as concerns purely scientific results, it is far better to discard entirely the conventional terminology, and to base the study of the accent not at all upon 'composition', but wholly upon the traditional word-order. 1

F o r a detailed study of several of these combinations, see L e o , 1. 1., p

417 ff. 2 U p o n this f r e q u e n t ' caesura b y t m e s i s ' , see also Plessis, Metrique § 29, 2°.

IÓO

AMERICAN

JOURNAL

OF

PHILOLOGY.

A C C E N T OF MONOSYLLABLES.

W e learn from the repeated statements of the Latin grammarians that the prepositive monosyllables are as a rule (fere) atonic (Priscian, Keil, III 479, 20, etc.). In a former discussion of this question (Transactions Am. Phil. Assoc., X X X I V , p. 62 f.) I interpreted the rule of the grammarians to mean that the monosyllabic prepositions, conjunctions, pronouns and adverbs are atonic per se, that is, naturally or originally atonic, i. e. quarn bene, quo mi?ius, and that they can acquire an accent only through recession, i. e. qu&m-bene, quo-minus. It is quite unnecessary, however, as I now think, to assume in this way the truth of an abstract proposition, such as the necessary or original atonic character of the monosyllables. It seems simpler and more reasonable, as well as more thoroughly in accord with the usage of other languages, to proceed from the assumption that the monosyllabic particles, quam, iam, et, etc., like all other independent words, have originally an accent, as in fact the grammarians expressly declare; if they very frequently lose this accent, this happens simply because they are subordinated in sense to the other words of the sentence and, at the same time, in the majority of cases, cannot preserve their accent through the operation of the three-syllable law; for it is certain that combinations like quam mdgnus, quam m&xime, et stntit have as a rule only a single accent. Similarly it can scarcely be doubted that examples in which the second word is an iambus or a pyrrhic, i. e. quam bene, et magis, cf. neque potest, neque scio, have commonly no place for two separate accents in the rapid legato pronunciation of common life, which does not especially aim at the painful spelling out of single words or the precise placing of theoretical stresses.1 The question remains whether in the examples just cited the accent falls on the monosyllable or on the principal word ; it will probably be correct to conclude here that both accents are equally correct and equally legitimate, that is, qudm bene, et magis, cf. neque potest, are as normal as quam bine, et m&gis, cf. neque potest, and there seems no reason to suppose that in fugitive collocations such as these, which are not included under any of the traditional word-orders, either 1 Compare the warning, for example, in Gramm. Lat., Suppl., p. 228, 33 H. ( = Scholl, D e acc., p. 128) against pronouncing male sanus with two accents, i. e. male sanus, instead of malesdnus.

LATIN

WORD-GROUPS.

l6l

accentuation e v e r prevailed to the c o m p l e t e exclusion of the other, e x c e p t in the later period w h e n the recessive t e n d e n c y was lost a n d the disinclination to place the accent u p o n a prefix b e c a m e fully established. In all other periods the c h o i c e b e t w e e n the two accentuations is doubtless d e p e n d e n t chiefly u p o n the g e n e r a l r h y t h m or m e l o d y o f the sentence, i. e. u p o n the sentence-accent. 1 T h e existence of the initial accent, h o w e v e r , even in the case of casual combinations l i k e qudm bene, ¿t magis, is v e r y clearly s h o w n b y their free admission in t h o s e feet of L a t i n verse which c o n v e n t i o n a l l y require a tonic s y l l a b l e to be placed in the arsis, viz. the third foot of the trimeter, the fifth foot of the h e x a m e t e r a n d the first foot of the A d o n i c , w h i l e C i c e r o ' s w e l l - k n o w n s t o r y ( D e div. II 84) of cdve n[e] eas2 p r o n o u n c e d nearly as catineas points to the same conclusion. A c c o r d i n g to this view, in e x a m p l e s l i k e t h o s e j u s t cited, w h i c h show the accentuations H magis and et magis e x i s t i n g a p p a r e n t l y side b y side, w e cannot admit, strictly s p e a k i n g , that any ' r e c e s s i o n ' o f the accent has o c c u r r e d in the case of H magis, but must consider the latter in e v e r y w a y an original a c c e n t ; w e can o n l y say that t h e recessive nature of the L a t i n accent renders t w o accents, i. e. it mdgis (cf. neque potest) quite u n n e c e s s a r y in such combinations. T h e term ' r e c e s s i o n ' in its p r o p e r sense is rather to be a p p l i e d to the v e r y n u m e r o u s cases of p h r a s e s a n d w o r d - o r d e r s , in w h i c h the initial accent has entirely s u p e r s e d e d the medial a n d alone remains in use, i. e. hauscio, id scio, idagit, etc.; since, h o w e v e r , ' r e c e s s i o n ' is the most convenient term to e m p l o y , on a c c o u n t o f its b r e v i t y , I shall continue to use it of the f o r m e r class o f cases also, and content m y s e l f with pointing out that this use is in reality inexact. 3 1 English and L a t i n m o n o s y l l a b l e s have many points of similarity in respect to their variable accentuation. F o r w h i l e dissyllables and polysyllables a l w a y s have a fixed accent in E n g l i s h , our monosyllables are treated in each case either as a c c e n t e d or u n a c c e n t e d according to the choice of the p o e t ; cf. D a b n e y , M u s i c a l Basis of V e r s e , p. 32.

' A c c o r d i n g to Skutsch, Forsch., p. 58, = caii »['] 3A

eas.

collection of additional facts, b e a r i n g upon the coalescence of the L a t i n monosyllables in pronunciation and upon the L a t i n system of word-division, is omitted here from considerations of brevity. B e s i d e s the frequent writing of m o n o s y l l a b l e s together with the f o l l o w i n g w o r d in Inscrr. (Corssen, Ausspr. I I 2 868 ff.) and M S S ( W a t t e n b a c h , L a t . Palaeogr. 3 , p. 7 6 ; L i n d s a y , L a t . T e x t . E m e n d a t i o n , p. 14), the approval of this custom by the grammarians (Marius Victor., K e i l , V I , 23, 7 ff.) and its retention in the writing

162

AMERICAN

JOURNAL

OF

PHILOLOGY.

o f m a n y phrases of m o d e r n I t a l i a n , w h i c h involve the prepositions and the sentence-introducing

c o n j u n c t i o n s (see a b o v e , p. 151), it w i l l be sufficient

h e r e to r e f e r to B a h r e n s , praef. Poet. L a t . min., I, p. X I X , and to L . Miiller, R e s Metr. 2 , p. 579, for the t r e a t m e n t of 'in

area'

(v. 5), 'ut

vere'

(11)

'ego

sum' (10: enclisis of the s u b s t a n t i v e verb) as s i n g l e words of six letters e a c h in the i n g e n i o u s l y

constructed verses of the carmina

(Bahrens, I V , p. i 2 o f . ) .

aes est (ibid., v. 5); for iamnunc v. L . Miiller, 1. 1., p. 581.

duodecim

sapientium

T w o m o n o s y l l a b l e s are often similarly treated, as and sivis

similarly c o u n t e d as s i n g l e words,

T h e effect of the traditional word-order, in c a u s i n g

the c o m b i n a t i o n of m o n o s y l l a b i c c o n j u n c t i o n a n d pronoun to be felt as a s i n g l e w o r d , is f u r t h e r seen in the f o l l o w i n g : C o r p . Gloss. L a t . I V , 22, 4 astilla v e r u m ilia femininum

est;

ib. I V 4 8 0 , 1 8 .

S i m i l a r is the s t a t e m e n t of

the g r a m m a r i a n s that a c o n j u n c t i o n , l i k e at (ad), may b e ' prefixed ' to any case of a noun or to a n y v e r b , w h i l e a preposition, l i k e ad (at), can c o m m o n l y b e ' p r e f i x e d ' to one case o f a noun o n l y or to a v e r b through composition (Prise., K . I l l 25, 2 4 f . ; A u d a x V I I 351, 17 f . ; S u p p l . L I ) .

T h u s w e apparently h a v e

a play upon the t w o uses of the prefixed ad or at in P o e . 544: At trepidate saltern: nam vos «Vproperare haud postulo, i. e., at—saltern (like at—tamen) and adproperare. T h e editors c o m m o n l y correct to attrepidate, w h i c h seems u n n e c e s s a r y . — A g a i n the close c o n n e c t i o n b e t w e e n the m o n o s y l l a b l e and the f o l l o w i n g word in pronunciation is i n d i c a t e d b y the c o m p l a i n t of C o n s e n t i u s ( K e i l , V 395, 7) that in the pronunciation of some sic ludit was i n d i s t i n g u i s h a b l e from si cludit. A s is w e l l k n o w n ( L i n d s a y , L . L . , p p . 122, 215), the m o n o s y l l a b l e s *cord, *terr, *esst tried, ted, hisc(e), hose(e), etc., l o n g retained their final consonant in early L a t i n before a n initial v o w e l , w h i l e haud (Caper, K . V I I 96, 4) a n d *hocc (Velius L o n g u s , K . V I I 54, 6) never lost the final consonant in this position, e. g. ' hocc erat, alma p a r e n s ' ; C I L . I X 60, 3 terminus hicc est. C f . also S k u t s c h , Forsch., p. 60 f. ELMIRA, N. Y.

R.

S.

RADFORD.

I I — O N T H E R E C E S S I O N O F T H E L A T I N A C C E N T IN CONNECTION WITH MONOSYLLABIC W O R D S AND THE TRADITIONAL WORD-ORDER. PART

THE

LATIN ACCENT AND THE

II.

TRADITIONAL

WORD-ORDER.

In a previous paper I sought to emphasize the fact that a very great number of Latin word-combinations possess recessive accentuation, and hence, in accordance with current usage, may properly be classed among the Latin 'separable' or 'improper' composita. This form of statement is not, however, sufficiently clear, and it seems worth while to define more precisely the conditions under which a recessive accent arises in the Latin sentence, and to illustrate the actual process by some concrete examples. It is probable, for example, that most Latin scholars, if asked to explain the recessive accent in transfero or trans fero (as it is not infrequently written in the Roman word-division), would say that the accent is trdns fero, and not trans firo, because we have to deal here with a compound verb and cases of genuine composition require a recessive accentuation. Thus the whole explanation is commonly made to turn upon the process of composition, as though in this process we had reached the primary cause of accentual change and did not need to examine also into the effects of simple juxtaposition. S o far, however, as concerns those classical languages which possess a recessive accent-system, such an assumption is wholly unscientific. 1 An historically correct statement must be somewhat as follows: T h e accent is recessive in trans fero, because a traditional word-order, i. e., a preferred and usual word-order, had been established already in I. E., in accordance with which certain adverbs, the so-called verbal prefixes, were commonly placed immediately before the verb (Delbrück, Syntakt. Forsch. V , p. 44 f.).2 In prehistoric Latin this traditional word-order 1 F o r the value and m e a n i n g of composition, cf. the references g i v e n above, A . J. P. X X V 158. 2 T h a t is, in I. E . and also in Sanskrit these adverbs (prepositions) held precisely the same relation to the verb that the n e g a t i v e s non, ne, nec and haud hold to the L a t i n v e r b ; cf. D e l b r ü c k , 1. 1., I V 1 4 7 : " E s war also das V e r -

LATIN

WORD-GROUPS.

257

had no doubt gained still further at the expense of the occasional order, but apparently an invariable order had not been established for any of these combinations; cf. ob vos sacro, sub vos placo Festus 190, 2 M. During the Latin historical period, however, the traditional word-order became virtually an invariable order in the case of some of the prefixes, as in trans fero. No process of composition has, however, taken place originally in any of these verbs, but only juxtaposition, and, from the purely technical point of view, Victor Henry (Comp. Gramm. 2 , Engl, tr., p. 173) is quite right in refusing to admit the existence of any genuine compound verbs in Greek and Latin; in like manner what Stolz (Hist. Gramm. I 404) observes of the combinations denuo, ilico, profecto, etc., applies equally to these so-called verbal compounds: " d a s eigentliche characteristische Moment der Zusammensetzung fehlt bei diesen Verbindungen." Hence the combinations with ante, post, etc., in an adverbial sense, i. e., anteparta (cf. postpartor, PI.), antedicta, contradicta, infrascripta (Inscrr.), postgeniti (v. still other examples in Stolz, 1. 1., 398), though they are regarded with disfavor by many editors of classical texts, are yet, in point of fact, as good Latin 'compounds' as the majority of the verbs in question. Indeed, provided the traditional order can be fully established, it is not even necessary that the two parts of a Latin 'compound' should belong to the same clause, e. g., nimirum = nisi, mirurn est (Ribbeck, Latein. Partik., p. 17). hältniss der Negation zum verbum finitum dasselbe wie das Verhältniss der Präposition, es trat keine Zusammensetzung der Negation mit dem Verbum ein, aber eine enge Verbindung zwischen der Negation und der einzelnen Verbalform. Dieses Verhältniss hat sich in den europäischen Sprachen bei einigen V e r b e n behalten. Im Lateinischen gehört hierher namentlich nescio,

nequeo, nolo [aus *nevolo\, im Slavischen die Verben welche bedeuten sein, haben, wollen, wissencf. also Hirt, D e r Indogermanische A k z e n t , p. 171. It is with good reason then that we not only have in Latin the accents nequeo,

néscio, néclego, nécuter, haúscio, nónnihil, etc., but, as will be shown later, we have, if the negation is prefixed to iambic verbs, regularly nón qaeo, nón facis, néque agis, etc.; thus the old rule has been very largely preserved in L a t i n as well as in Lithuanian and Slavic, that " d i e Negation steht auf einer L i n i e mit den P r ä v e r b i e n " (Hirt, 1. 1., p. 306). I f no L a t i n verbal comfosita are formed with non itself, this is chiefly due to the late development of this particular negation ; cf., however, nonnulli, nonnumqtiam, etc., and, in Inscrr., nonlicebit, nondebuerunt (Corssen I I 881). I may add that we apparently still find in early Latin some traces of the free I. E . position of the negative in relation to the v e r b ; for in T r u . 877 all recent editors read ne facere si velim for the M S re facere (cf. Habich, D e negationum usu Plaut., p. 29), although in Mo. 124 reparcunt ( B C D ) is usually retained (neparcunt L).

258

AMERICAN

JOURNAL

OF

PHILOLOGY.

I have said that an invariable order was finally established in Latin in the case of some of the verbal prefixes ; in perhaps the majority of cases, however, the order never became even approximately invariable, especially in the case of the prefixes ante, circum, inier, praeter, post, super, etc., which form in general only separable composita. Y e t it is easy to show from Latin verse and from the grammarians that the accent was as strongly recessive in those combinations like super erit, antê tulit, circiim ded.it, which possess only a traditional order, as in those which have gained an invariable order like irons fero. The Latin recessive accent is not due then primarily to the process of 'genuine composition', but it is due to the traditional word-order, which includes genuine composition and much more besides. Thus the verbal prefixes are far from being the only words which may be used to exemplify the I. E. traditional wordorder and its effects. Of all the I. E. word-orders the best-known is that in accordance with which the object immediately precedes the verb (Delbrück, 1. 1., III 24), and the question is entirely legitimate whether, in cases where the object-accusative precedes an iambic verb like velim, the accent recedes or not. In fact, we find evidence that both the accent aquâm-velim, and the accent âquam vélim were here known; for Plautus allows such apparent double iambic versecloses as Am. 1058 àquâm-velim; Au. 417 cöctim-decei; Cas. 395 lücrüm-facit; cf. Inc. inc. fab. 92- R . com. misericârdiâm-rogat (Klotz, Grundz., p. 244; Hauler, Einl. Phor., p. 38 n. 1), thus treating these combinations as quadrisyllable words. Similarly, although the word-order is far from invariable (e. g. often dare operam, operant . . . dare, also agis nugas Ci. 581), we find Lachmann's law regularly observed in operâm-datis, -damûs, -daté, -darê;1 nugâs-agis, -agit (10 times; examples in Lodge, Lex. Pl., p. 81); hence Lindsay's view of the admissible character of the accents fidêm-do, codtim-dabo, factdm-volo, missâm-face (Journal of Phil. X X 147; The Captivi. p. 369) seems a probable ' W i t h the single exception, Ba. 98 operdm-d&rë (troch.); anap. is St. 311 operâm-datïs. T h e a c c e n t operdm-do, etc., is w e l l attested b y the critical feet viz., A s . 44Q operam d a s ; Per. 372; Ph. 87 ; — b u t was not the only a c c e n t in use, as M e n . 1009 operam da 3 bo et, and also in i s t ft., as Ba. 103 operam d a ' b o . [I use ' i s t ft.' throughout in the m e a n i n g ' 1 s t ft. of a c o l o n ' ; also numerals written below the line indicate iambic verse ; written in the line, trochaic v e r s e . ] Pesstim dare (9 times) has both an invariable order and accent in the dramatists.

LATIN

WORD-GROUPS.

259

one. Cf. also the pronominal combinations originally containing an oblique case, which are often written as one word, i. e. quodlubet, quoilubet, quolubet, quidvis (Prehn, Pronom. Indef., p. 28 f.), quidvoles, quidvolet, etc., and observe the vowel-weakening seen in simus and libet for sumus and /«¿¿'/respectively (Lindsay, L. L., p. 29). Many indications show, however, that the separate accentuation, as in dquam vHim, was much more usual in such cases, e. g. the non-observance of Lachmann's law in examples like Ep. 691 moräm fäcis, quom 6go; cf. also As. 380 officium fd'cis, etc. It is in connection, however, with certain special classes of words which were pronounced in close connection with the following word, that the effects of the traditional word-order are best seen, viz., the pronouns, the conjunctions, the prepositions and monosyllabic words in general. The case of the monosyllables and the prepositions will be treated here 1 . If the dialogue verse of the dramatists be examined with reference to those I. E. word-orders which place the monosyllabic object pronouns or the monosyllabic sentence-introducing conjunctions and pronouns 2 —also the subject pronouns—before the verb, we shall find that in all these cases the recession of the accent is fairly complete. For we not only always find in a tribrach sequence quod edis, lit als, etc., but in dactylic and cretic sequences all combinations like quod facis, quae cupis, id petis, si sapis, hoc age, non queo, id scio, te volo, te rogo, ne time, etc., are subjected with remarkable strictness to the difficult law of Lachmann respecting the use of dactylic and cretic word-forms 3 , i. e., except in the first foot of a colon, such combinations can enter the verse only in the form quddfacis, 1 F o r a separate treatment of the dissyllabic pronouns and conjunctions, v. my article in the forthcoming volume of the T r a n s . A m . P h i l . Assoc. for 1904. 2 T h e conjunctions, w h e n followed immediately by the verb, often f o r m a complete sentence, as s/sapis, si-facis, and the accent of the more frequent combinations is e x t e n d e d b y analogy ; cf. also the I. E . ' e n c l i s i s ' of the verb in this position (Hirt, 1. 1., 307 ff.), w h i c h has perhaps been preserved in L a t i n . Similarly in the case of clauses w h i c h consist only of subject and verb, the subject-pronoun p r e c e d e s the verb (Delbrück, I I I 13), e. g. tti-facis; in g e n eral, the word-order subject + verb must be recognized as sometimes causing recession in L a t i n , as in the verse-closes Poe. 447 quando ä m ö r - i u b e t ; T r i . 533 ille äger-fuit. 3 Similarly even in P h a e d r u s (I IO, 9 perdidisse quod-peris) quod-petis counts as a trisyllabic word in j u s t i f y i n g an app, double iambic verse-close, and H a v e t ' s correction (ed. Phaed., p. 181) is needless.

A ME RICA N JO UKNA L OF PHIL OLOG

2ÖO

sisapis, t[u]

nepave,

nonpotist,

litdec&t,

iitsolis,

quam

ais, e t c . , a n d n o t at a l l in t h e f o r m quod

V.

r\em\

fäcis,

agis,

quid

ne pave,

nbn

potest-, very rarely as quod fäcis, idvölo, etc. Finally, by extension of usage, the accent recedes upon any monosyllabic pronoun or adverb prefixed to the verb, i. e. tibi ait, hlnc-agis, idm-scio ; the analogy of the v e r y numerous verbal composita like pervidet may perhaps have been an important factor here. It should further be noted that these conclusions which we have reached respecting the accented character of monosyllabic pronouns, conjunctions and adverbs prefixed to the verb, and which are equally true for prefixed pyrrhic pronouns, conjunctions and a d v e r b s , e.

g.,

always

bene-void,

( s e e b e l o w p . 2 6 9 ) , nequevolo,

bene

facts,

ita

a n d n e v e r bene fäcis,

scio,

ego-scio

ego scio,

etc.1,

are in general agreement with the probable history of Latin verbal accentuation. Thus, according to Hirt, Indog. A k z . , p. 1 7 1 , it is extremely likely that Latin originally retained the extensive I. E. ' e n c l i s i s ' of the verb and constantly accented the prefixed adverb 2 , as appears from verbal forms like conficio, earlier *cdn facio, although in historical Latin, to be sure, the accentuation of the prefixed monosyllable or pyrrhic word is limited to those cases which are permitted by the three and four-syllable laws. T h e preceding statement may seem to assign too great a role to an abstract traditional word-order and to take too little account of concrete cases. I have purposely chosen it as the most convenient form of expression, but the principle actually involved in the process is probably this: that in every traditional wordorder so many familiar word-groups like quid agis, quod ames, ' E x c e p t i o n s to this

tendency

are

very

s i b i f o r e ; M i . 1 1 7 u b i s y m u s ; M e r . 778.

f e w , v i z . , in

t h e i s t ft., C a s . 54

W i t h i n the verse I have noted only :

M i . 36 e h e m , s d o ; N o v . c o m . fr. 93 a g e m o v e t e ; t h e f o i l , c o n s t i t u t e n o r e a l e x c e p t i o n : P e r . i g o s e d - i t a v o ' l o te ; A u . 608 t i i - m o d o c a 2 v e ; M o . 20. t r u e t h a t — o w i n g p a r t l y to the r e c e s s i v e t e n d e n c y , p a r t l y to o t h e r the disposition

to p r e f e r

vi w , w w to w w ,

I t is

causes—

is g e n e r a l in d r a m a t i c v e r s e ,

b u t I d o not t h i n k t h a t I a m m i s t a k e n in s a y i n g that t h i s t e n d e n c y r e a c h e s its h e i g h t in c o n n e c t i o n w i t h t h e v e r b a n d in c o m b i n a t i o n s l i k e meus pater

(Lind-

s a y , T h e C a p t i v i , p. 369®), erus metis, e'rus tuos ( e r u s m e u s o n l y in a n a p . a n d in 1st f t . i a m b , a n d b a c c h . ; n o r e a l e x c e p t i o n is P e r . 259 n d m - e r u s m e u s ; i n Cur. 177

either

quo6d-meus

erus or q u o d m e ' u s eriis is p o s s i b l e ) ; cf. a l s o

L u c h s , C o m m . Pros. I I 131. 2

F o r o t h e r t r a c e s o f t h i s I . E . ' e n c l i s i s , ' cf. H i r t , 1. 1., p . 3 0 7 : I m

ischen

german-

A l l i t e r a t i o n s v e r s ist g e w ö h n l i c h e i n s e l b s t ä n d i g e s A d v e r b i u m

b e t o n t als d a s V e r b u m , e. g . A g s . üfp

heran.

höher

LATIN

WORD-GROUPS.

2ÖI

quod facts must arise with a recessive accent that a g e n e r a l t y p e is e v e n t u a l l y established, to w h i c h the accent e v e n of rare c o m binations is m a d e to conform. 1 Q u i t e similarly, in the case of rare v e r b a l composita like exblbo (only M i . 832) or even expüto, the r e c e s s i v e a c c e n t really exists b e c a u s e the m o r e frequent c o m binations like ejfero, expeto, etc., h a v e first established a g e n e r a l accentual t y p e . W i t h this e x p l a n a t i o n of m y m e a n i n g , I shall continue to refer to an abstract traditional order. T h e d e m o n s t r a t i v e , relative a n d interrogative p r o n o u n s also possess an I. E . traditional w o r d - o r d e r , i. e., t h e y are as a rule p r e f i x e d 2 to their substantive ( D e l b r ü c k , 1.1., I I I 35). H e r e also w e find not o n l y a l w a y s quts erüs, hie equos, hie homo, but also with v e r y few e x c e p t i o n s , quis-modüs, hie-loeüs, hie-dies, etc.; for e x a m p l e s of a similar u s a g e in other I. E . l a n g u a g e s , cf. H i r t , 1.1., p. 324. T w o other traditional orders show, at least in t h e sensitive tribrach sequence, a c o m p l e t e r e c e s s i o n : ( 1 ) T h e p e r sonal a n d demonstrative p r o n o u n s and pronominal a d v e r b s attach t h e m s e l v e s to the s e n t e n c e - i n t r o d u c i n g conjunctions a n d pron o u n s ( K ä m p f , P r o n o m . Personal., pp. 36, 31), i. e. sSd ego, quis ea, nisi ita, etc. (2) T h e subject-noun a t t a c h e s itself directly to the S . - I . conjunctions and p r o n o u n s ( D e l b r ü c k , 1.1., V 16, 23), i. e. U erus, sed erus. T h i s habit has a p p a r e n t l y been e x t e n d e d also to the o b l i q u e cases of the noun in this position, and since all s u c h cases i n v o l v e W a c k e r n a g e l ' s toneless ' s e c o n d position', w e h a v e here also p e r h a p s the retention of an I. E . 'enclisis'. T h e cases w h i c h h a v e been s o far mentioned are all of t h e m c l o s e l y c o n n e c t e d with traditional o r d e r s ; on the other hand, o w i n g to the free position w h i c h b e l o n g s t o the adverb3 in t h e sentence, nearly all adverbial combinations s h o w a variable accent 1 I n many cases the retention of an original I. E . ' e n c l i s i s ' is also possible, and in some cases it is h i g h l y probable. 2 J u s t so Priscian, K e i l I I I 31, 5, observes at some length that certain usually ' p r e f i x e d ' words, such as the pronoun hie (e. g. hie homo) and the proh i b i t i v e ne, b e l o n g in almost the same category as the prepositions, w h i c h are the prefixed words (praeposiliones) nar' i^oxr/v, and for the accent of w h i c h see b e l o w (p. 262). 3 T h e adverb, b e i n g loosely c o n n e c t e d with the v e r b w h i c h it modifies, f r e e l y varies in position ; examples in Braune, Observ. gramm., pp. 12, 33, 60; cf. L e o , N a c h r . d. Gött. Ges., 1895, p. 428: " E i n e sehr lose V e r b i n d u n g geht das A d v e r b i u m mit dem V e r b u m , das es bestimmt, eine minder lose mit dem N o m e n oder A d v e r b i u m e i n " ; examples of the latter in B r a u n e , pp. 12, 25, 51.

2Ô2

AMERICAN

JOURNAL

OF

PHILOLOGY.

a n d n o o b s e r v a n c e of L a c h m a n n ' s l a w , e. g . B a . 363 si úsus;

Ad.

708

qui mâgïs m ó r e m ;

Ru.

218 qui mïnûs

mâgïs sérvio;

C u r . 622 t e 2 m a i e p é r d a t ; P o e . 16, e t c . ; a l s o A d . 701 ni m à g i s ; Cap.

430;

we

find

always,

a d j e c t i v e or a d v e r b tám-citó,

however, támdiú,

in

association with

the

tâm-malê.

T h e r e is at l e a s t o n e g r e a t t r a d i t i o n a l L a t i n w o r d - o r d e r , w h i c h is n o t w h o l l y o f I. E . o r i g i n , b u t is in l a r g e p a r t t h e r e s u l t

of

l a t e r d e v e l o p m e n t ; v i z . , t h e o r d e r in a c c o r d a n c e w i t h w h i c h t h e p r e p o s i t i o n is p r e f i x e d t o its c a s e .

H e r e also the accent regularly

r e c e d e d ; for in t r i b r a c h s e q u e n c e s t h e d r a m a t i s t s a c c e n t a l w a y s in opus, a n d in d a c t y l i c a n d c r e t i c s e q u e n c e s t h e y o b e y s t r i c t l y L a c h m a n n ' s l a w , i. e. á patrê, manús

(cf. êmtibî,

vâemihî,

f o o t e i t h e r a pâtre compositum

o r a pâtre,

l i k e apatre,

in marê

sub-sibî, etc.

(immarê),

inforó,

súm-

e t c . ) , a n d o n l y in t h e T h u s if w e t a k e a

w e shall be able to note the

first

separable observance

o f t h e L a c h m a n n - R i t s c h l l a w in t h e m o s t m i n u t e a n d d e l i c a t e p a r ticulars.

F o r t h i s f o r m m u s t e i t h e r b e p l a c e d in t h e v e r s e - c l o s e ,

i. e. á patrê, ( 2 ) as ápatre

o r it m u s t s t a n d w i t h i n t h e v e r s e ( 1 ) a s ( S t . 7 1 , P a n d e d . m i n . ) 1 ; ( 3 ) as ápatr\e\

s h o r t s o f apatre

dpatre;

¿r.

The

c a n n e i t h e r in w h o l e n o r in p a r t b e u s e d t o f o r m

a r e s o l v e d t h e s i s , a n d if e l i s i o n o c c u r , t h e u l t i m a m u s t b e e l i d e d i n t o a n a c c e n t e d s y l l a b l e ; t h u s t h e p e n u l t \a\patr\e\ is d o u b l y i n c a p a c i t a t e d f r o m a c t i n g as a B r e v i s B r e v i a n s . 2 l a r l y it m a y b e s h o w n

from A h l b e r g ' s

\lti\for\o\ Simi-

collection that, alter

a

1 This is a permissible license in the case of all dactylic words (cf. Leo on As. 250; Men. 762, and cf. even Maurenbrecher, Hiatus, p. 203 f.), for the reason that the regular verse-treatment of dactylic words gives a sufficient clew to the reading of the verse independently of quantity; cf. also Per. 398 vél-face ; Eu. 12 qui-petït, unde is; et al. 2 A(çiim, dé, ¿x, prS) pâtre occur 13 times in middle of verse with elision into an acute (Cas. 36; Men. 1112; 1113 ; Mo. 1127 ; Ps. 730; Tri. 771 ; 785 ; Vid. 115; And. 653; Hau. 235; Ad. 951 ; Ph. 607; 879), 12 times in verseclose (Ba. 665; Men. 31; Mer. 64; 68; Poe. 65; Tri. 741; 775; 778; Tru. 649; And. 252 ; Acc. fr. 654; Afran. fr. 310); de pitre, etc., with long penult, of course occurs later (Sen. H. F. 446 ; Thy. 310). Also dd (in, etc.) mare occurs 4 times with elision tn. v. (Ba. 458; Mer. 354; 371 ; Ru. 295), 4 times ult. v. (Poe. 627 ; Ru. 34 ; 898 ; Tru. 564), while we find in mdre (cf. pectóre) once : Tru. 565 hoc in ma're abit. -t- sene, cane, lore, grege, sale, love occur 5 times with elision m. v. (Cas. 320; Ps, 871; Ad. 362; Hau. 1036; Laber. fr. n o ) , 6 times ult. v. (Am. 1125; Tri. 208; Per. 267; Mi. 966; Hau. 759; perioch. 10). No exception is Tri. 152 dttriá-milia (numeral + subst.), nor Tri. 940 a'd caput ámnis (1st ft.), but we find Hec. 842 in bre 2 ve, also once in 1st ft. e nuce, Cur. 55, and in general we find adfdr\um\ (Ps. 1230, usually

LATIN

WORD-GROUPS.

263

preposition, an iambic word is excluded absolutely from the arsis of the proceleusmaticus, i. e. never âpûd ërùm, apud fôrûnt, sine môdd, sine maid, super ânû, quod in manu, but always âpud-erûm, âpûd-forûm, sine-modo, qudd in mâmi. A s shown in the example qubd in mânii (Tri. 914; cf. séd ôptume, épïstulam), the accent, in accordance with its general tendency, may sometimes recede still further, but it may not move forward, as in quod in mânu. It should be added that the absence of vowel-weakening in the prepositional composita is due chiefly to the free occurrence of tmesis-forms (e. g. â bono patre), since, wherever the tmesisforms are excluded, vowel-weakening freely occurs, e. g. per annos (per . . . anno s), but perennis; per agros, but peregre; quotannis (adv.), but quotennis (adj.); in loco (in . . loco), but ilico. Finally, this 'enclitic' attachment of the noun to the preposition was, according to Hirt (1. 1., 43, 299 if.) a familiar construction in I. E. Both Germanic and Greek (cf. eVtVav, v7répfÀopovy Sidrptxa, etc.) preserve some traces of the construction, and Slavic has retained the old rule in actual use to the present day. In primitive Italian, according to v. Planta, Osk.-Umbr. Grarnm. I 597, the preposition received the tone and the noun was apparently in all cases 'enclitic'. In historical Latin the old ' enclisis' is preserved not only in many combinations stereotyped as adverbs, e. g. denuo, sedulo, ilico, comminus, eminus, protinus, antea, inlerea, interdiu, intervias, profectol etc., but as a rule in all combinations of prep, and iambic noun. 2 The recessive accentuation continued here throughout the classical age (cf. Quintil. I 5, 25 ff.; Annianus ap. Gell. V I 7). True, as we have already seen (A. J. P. X X V 151), the processes of ' recomposition ', which give rise to commando for commendo, *de novo (Fr. de nouveau) for dénuo, etc., existed to a limited corrected on other grounds, cf. Kellerhoff, Studem. Stud. I I 83) and cum ma'/ti (Ru. 923, in free troch. (?) Oct., just as in immediately preceding line susci'tët) only as we find fectàre andpectàrï; for cases of latter outside 1st ft., cf. K l o t z , Grundz., 63, 277 vs. A h l b e r g , Corrept. iamb., 46. ' In Pl.'s time often still pràfëcto, as shown especially by the procel. Ps. 201 : id tibi prô 2 fëctô; cf. also St. 614 per hortu 3 m. 2 T h i s is the view now generally held by critical students of the L a t i n accent ; see especially the excellent observations of Vendryes, L'intensité initiale, Paris, 1902, p. 108. I n V.'s view also we have forms like admodum, affatim, instead of *admidum, *affitim, because their ' composition ' or ' recomposition ' is subsequent to the period when the law of vowel-weakening was effective ; ' recomposition ' seems to me here the important factor.

264

AMERICAN

JOURNAL

OF

PHILOLOGY.

e x t e n t even in the earliest period, but t h e y long r e m a i n e d o f s e c o n d a r y importance and did not a c t u a l l y g a i n the m a s t e r y before the late R o m a n c e p e r i o d . In the fifth c e n t u r y the recessive accent a p p e a r s to h a v e b e e n still p r e d o m i n a n t ; for the g r a m m a r i a n s r e p e a t e d l y class o r d i n a r y prepositional phrases l i k e cismare and in locum a m o n g the composita in the s a m e manner as conftcio, omnipotens, etc. (Charisius, K e i l , I 17, 3 ; D i o m . I 436, 1 5 ; D o s i t h e u s , V I I 389, 4 ; ib. 409, 27, etc. 1 ). T h e i r disputes s h o w also that et tarnen, sit tarnen, et qwidem2 a n d a p p a r e n t l y HmihP were still in actual use. T h e verse of the quantitative poets to the latest a g e bears the s a m e t e s t i m o n y to the g e n e r a l recession of the p r i m a r y accent ; thus often Lucil., P o m p o n . , a n d N o v i u s (B. C. 90), as P o m p . 66 a 5 g e a n u s ; N o v . 50 u 5 bi e g o ; also L a b e r . 13 qu6m e g o ; C L E . 1 S o also, notwithstanding G e l l . V I 7, af fdtim was not the only a c c e n t in this period ; for Priscian, I I I 75, 7 K . , w r i t e s : ' affätim ' a G r a e c o an-qne\o\ mjj (7 times in verse-close, 4 times med. v., viz. A s . 9 0 7 ; S t . 2 9 2 ; A d . 7 3 8 ; H e c . 4 4 3 ; cf. also Miiller, N a c h t r . 23), non potêst, etc. ( e x c e p t M e r . 5 5 3 d u m potes, but Ps. 85 read s é d potés ( A ) ; cf. Miiller, P r . 1 3 1 f.), nil morôr (27 times, cf. K e l l e r h o f f , 1. 1., 82 and Miiller, P r . 1 2 6 f.), non tacés ( 1 1 times, but H e c . 527 he :i m, tacës), ût vales, ut valêt, (9 times), lit soles, ût solêt ( e x c e p t B a . 8 0 u t s o 3 l e t in i s t i s ) , jit lubét, lit decét, lit iubês,

quad sciâm, si scids, hie era, etc.'2 F i n a l l y , note especially the v e r s e s in which this accent occurs m o r e than once, a s : Cas. 765: Quin agitis hôdie? quin dalù, siqufddalfsl Cur. 1 4 8 : Vés atnb, vés vôib, vés peto àtque obsecro. Mo. 3 2 g : Si cades, nén cadès, quin cadhm tecum.

med. v. (not infrequent). Although the third class is the least frequent of the three, cases occur quite often enough (Mi. 2 7 3 ; A u . 5 7 4 ; Cap. 440, etc.) to g i v e cogency to the argument offered above. 1 A.led. v. the type et scié occurs 1 3 times ; et sci[o\0^j 8 times. In 1 s t ft. et sciô Ci. 5 8 8 ; — P s . 221 ed. mai. reads sed sci 4 o < e g o > , a common phrase. Ét volé med. v. occurs 1 7 t i m e s ; et vol\o\ 20 times. E x c e p t i o n s occur in 1st ft., as Ps. 9 1 9 hoc vôlô ; Mo. 583 ; Ci. 82 ho'c vôlo âgâtis ; A u . 823. A real exception is perhaps M e n . 207 scin quid vo 2 lo ego té ; [also T e r . H e c . 753 quid vôlô . . . f a c i a s ; H a u . 1 0 2 7 aiit quod vo 3 lo] ; but not T r u . 779 hie nunc vo'lo-scire ; T r u . 2 6 1 . Mi. 6 1 2 se'd volô-scire (for the phrase, v. L i n d s a y , C a p t i v i , p. 366), nor M i . 678 ego-me 6 volô viveré. Although excluded after monosyll. accusative (quod, hoc, te, etc.), volô is frequent after polysyll. acc., as St. 347 mûnditiâs v o l ô ; 674; T r u . 283, etc., and is wholly unrestricted in anap. 2 Compare also the following examples, which, occurring as they do in T e r e n c e , probably involve pyrrhic rather than iambic words (v. Podiaski) :

LATIN

WORD-GROUPS.

413

C f . S t . 285 a g e útplacét, curre iít lubét\ Poe. 1198 ilt sapít. || Ing é n i u m pátris, habet quéd sapít-, A u . 6 5 7 ; M e n . 1 1 6 ; P e r . 482; 5 7 8 ; P h . 799. A l t h o u g h iambic imperatives like cave, vide, vale, abi, etc., a r e used in dramatic verse almost e x c l u s i v e l y in a shortened form ( M ü l l e r , Pr. 1 5 3 ; L e p p e r m a n n , D e corrept. Pl., 82), y e t , in conj u n c t i o n with the n e g a t i v e prefix ne, w e find a l w a y s né negá or né neg[a] ¿ w (7 times, cf. L o c h , I m p v . b. Pl., 21 ; 2 med. v., viz. A s . 3 7 7 ; 9 2 2 ; — 3 times written in B nenega), né pavé (4 t i m e s ; 3 med. v.: A m . 1 1 1 0 ; M e r . 885; Ps. 103), né timé ( 1 2 times; 1 med. v., M i . 1345; i cretic, Cas. 8 3 5 ; — 3 times written netime)\ né duis, né duâs (6 times; 2 med. v.: M e r . 4 0 i ( ? ) ; R u . 1368), and a l w a y s in verse-closes né docé ( e x c e p t anap., A u . 434), né crepâ, né siés, etc., also a l w a y s trié v i d é ( L o r e n z on Mi. 376; L i n d s a y , Class. R e v . V I 404). In conjunction with other prefixes, as inir\o\, rus, nunc, sed, vah, si, etc., we h a v e med. v. intro abi ( A s . 543; C a p . 452; Ci. 770), r ú s abi ( M o . 8; Cas. 1 0 3 — t h e usual punctuation, viz. rus, abi, is incorrect, cf. M o . 74); M o . 585 n ú n c tu abi ( M S S : né tu abi) ; M e r . 749 st, abi ; Ps. 1288 séd v i d é ; H a u . 600 v a ' h , v i d é ; p r o b a b l y A m . 749 ha'nc rogâ.||Mëquidém ; — a n d often in verse-closes, as M o . 580 (núnc abi), C a p . 125 (átque abi), ib. 860 (séd iubé), etc. E x c e p t i o n s to L a c h m a n n ' s law occur o n l y occasionally med. v., viz. M o . 66 tace á t q u e abï r ú s ; R u . 1089 a 2 c tace t ú ; p e r h a p s M o . 810 a 2 h c a v ë ; — b u t frequently in ist ft.,as M o . 929 ( n ú n c abï) ; 187 (quin möne q u a é s o ) ; H a u . 1031 (ét c ä v e ) ; H e c . 223; P o e . 609; cf. C a p . 643 (séd v i d e - s i s ) ; cf. P o e . 358. The a c c e n t ac câve ( B a . 147; H a u . 302) is also frequent med. v.-, further T r i . 11 (et d â t e ) ; A m . 353; M e n . 4 1 6 ; cf. P o e . 292. T h e s e e x a m p l e s show that, e x c e p t in the case of the prohibitive p r e f i x ne, the r e c e s s i v e accent is not so usual with the i m p v . as with other verbal forms, p e r h a p s because s o m e impvs. are often felt as e x c l a m a t o r y a d v e r b s (cf. cedo, as A d . 123 aút c ë d o ) . W i t h v e r b a l forms in g e n e r a l a few additional e x c e p t i o n s to L a c h m a n n ' s law occur, viz. Mi. 1272 út tremit á t q u e ; T r u . 755 no 2 n r e d i s ; cf. C a p . 343 ita-u 6 t velïs (treat otherwise, h o w e v e r , Ps. 479 ; C a s . 242) ; — m o r e often in i s t ft., as St. 58 qui 1 manët út ; A u . 823 ; A m . 703 ; cf. C a p . 572. T h e less usual m e d i a l a c c e n t is frequent in the i s t ft. (cf. perlùbët, C a p . 833), as T r u , 242 si H a u . 1054 ut té dece 3 t. E a ; ib. 666 nón Iice 2 t hominem e s s e ; Phor. 718 noti moro 6 r. U b i ; H a u . 1021 quin sie 6 t itidém.

414

AMERICAN

JOURNAL

négát se; Ci. 453 nil mórór; Ba. 1001; Tru. 235, etc., and bibe si bi 2 bis; Mi. 1314 quín Per. 272 quod da 3 tümst; Poe. PLAUTINE

USAGE

OF

PHILOLOGY.

Per. 50 iamne ábís; 213; Ps. 156; admissible elsewhere, as St. 710 tu iu 3 bes; Mo. 648 has dédít qu.; 374; Ba. 646.

IN P R E P O S I T I O N A L

COMPOSITA.

No critical scholar at the present day is likely to question seriously the recessive accentuation of the prepositional composita (cf. Podiaski, Quomodo Terent. verb, acc., etc., p. 15; Lindsay, Journal Philol. X X 150 ff., and T h e Captivi, p. 370; Skutsch, Forsch. 159; Vendryes, L'intensité init. 108). Thus recession is clearly indicated by the frequency with which the assimilated form of the preposition occurs before nouns in Inscrr. and M S S , i. e. al laevam, im fronte, ec flamma, etc.; v. Corssen P 266 ; I P 871; Neue-Wagener, Formenl. II 783, 868, 905, etc. The chief examples of this sort occurring in the Plautus M S S are brought together by Leo, Nachr. d. Gott. Ges., 1895, p. 419, e. g. appatreni Mer. 962 (C D), impraeda Ep. 108 ( A ) ; summanus Per. 450 (B C D), etc. 1 ; for the occasional assimilation also of conjunctions and other particles, cf. C I L . I l l suppl. 5, 15184, 13 pro se es su[is] ; I V 1895 tan durum; V I 6182 tan cito, etc. T h e lengthened vowel of the preposition is also abundantly attested for combinations like IN FORO (CIL. V I 1566), INSVOS (II 1503), INCOLONIA (II 1676), etc., cf. Christiansen, De i longis, p. 43 ft. Tmesis-forms 2 occur occasionally in poetry, e. g. the caesura after the praep. adposita in Cat. IV 18 per | ímpoténtiá freta is comparable to that after thepraep. composita in Hor. C. II 12, 35 dúm flagrántia dé | torquet; a looser connection may perhaps be sometimes indicated by the very free occurrence in the hexameterclose 3 of tmesis-forms like inter | agendum, ante | domandum (Verg. Eel. 9, 24; Georg. 3, 206), but we may note that Silius Italicus uses quite similarly in the hexameter-close et super\esse (Miiller, R. M.3, 264); cf. Hor. Ep. I 19, 3 ut male | sanus; cf. ' I have not seen J. Dorsch, Assim. in d. Compos, b. PL, 1887. A l t o g e t h e r unwarranted is the view of L. Muller, R . M. s , p. 467, who, on the ground of the occasional occurrence of these tmesis-forms, denies that the dactylic poets treat circumlitora and aboris as composita in the manner described by Quintilian (I 5, 27). 2

3 C f . also the observations of Prof. Humphreys, Trans. A m . Phil. A s s . I X (1878), p. 58.

LATIN

WORD-GROUPS.

415

Verg. Georg. 2, 366 interque legendae; etc. 1 Finally attention may be directed to the characteristic Latin arrangement, by which the preposition is often placed between the attribute and the noun,

e. g. omnibus in locis, lacrimis

cum multis.

Leo, Nachr. d. Gott.

Ges., 1895, p. 418, sees in this construction a survival of the original I. E . postposition of these particles, which is still largely retained in Oscan and Umbrian. Leo's solution may be the correct one, yet I should like to point out that a different view of the construction is at least possible, and that the Latin order omnibus in locis is capable of being explained wholly from the development which has taken place within the Latin domain, and from the tendency which has there prevailed to treat the complex ' preposition-noun' as a unit and to preserve it unbroken in the great majority of cases, e. g. omnibus inlocis, magna cumlaude where we have the adjective agreeing with the compositum.

inlocis, etc.

So far we have spoken only of general considerations; in dramatic verse the recessive accentuation of these composita is clearly indicated by the following special usages : ( 1 ) We find frequent examples of the oxytonesis of trochaic and spondaic prepositions in the critical feet. E. g. Ps. 648 inte4r erum; often intereos, praet6reos 3 (Ritschl, Proleg. C C X X V I I ; 1 Similarly verbal tmesis-forms like ante \ venitis, ante \ parata (which our editors would doubtless in such cases write separately), are probably legitimate in the hexameter-close, although I have had no opportunity to collect such examples. L. Mtiller himself writes in Cat. IV 4 the tmesis-form praeter \ ire on account of the caesura ; similarly, Horace, C. I 2, 34, has the tmesisform circum | volat in Sapphic verse ; on super \ volitantia in Lucan (V 595), see Trampe, De Luc. arte metr., p. 34. As to the main point at issue, the character of prepositional phrases, it is well-known that in the senarius-close forms like sine-mora, in-oppido (Phaedr. V 7. 2 2 > ' H I 10, 19, etc.) count as one word (cf. Miiller, R. M., 532; Havet, ed. Phaedr., p. 178). Hence H . A . J . Munro, Transactions of the Cambridge Philosoph. Society, X (1864), p. 378, states the facts correctly: "An ancient Latin seems to have been able by the sense alone to distinguish in justo from injtisto o r p r a e t e r mtssa from praetermtssa". It is scarcely necessary to mention in conclusion the great number of compounds which have arisen from these familiar combinations, e. g. inforare (in-forum : Plaut.), egregius (e-grege), suburbanus (sub-urbe), etc. 2 One may note in passing the curious oversight of Ritschl (into which Lorenz and the edd. mai. and min. have also fallen) in conjecturing Mo. 1156 pudet | Propter ea quae fecit. B 1 has Propterea qui facit, and if we follow the usual emendation, we can only read Propterea quae (or quia) fecit, cf. And. 414 id propterea.

4l6

AMERICAN

JOURNAL

OF

PHILOLOGY.

Klotz, Grundz. 267), intérse, proptérme, 1 etc. (Klotz, 324);—Enn. tr. fr. 55 inte 6 r deâs; Poe. 1398 ¡nte"r negótiùm ; not conclusive is Enn. tr. fr. 201 ante 2 pedes (cf. ântëpëdés Mer. 780; Ad. 386), nor Cap. 617 inte'2r sacrum. Hence we find inter ' a b s t r a c t e d ' from such cases and used alone, as Mer. 752 q u ó s i n t é r ; possible also is apud-forum Ep. 422, cf. Pompon, fr. 168 R., although Leo, Forsch. 226 f., demands in both these cases the usual âpù(d) forûm, with hiatus or slight change of text ; cf. also with pyrrhic prep. Ps. 140 apu 6 d oveis [cf. Ci. 50 sine 5 mëo ; Poe. 1193 inte:!r alias], though for the sequence ^ ^, ^ ^ we find regularly âpud erûm (Am. 591 ; Cap. 666; Ps. 461 ; 493, etc.), âpud eum ( H a u . 733)i e t c - T h e student of Plautine accent will perhaps wonder why it is that we find in the critical feet such numerous cases of the accents intér-nos, ergâ-me, proptêr-res, etc., where the prep, is followed by a monosyllable, and also numerous cases of the combination ^ - (intêreos), while we find in the critical feet only a few cases like infer deas, where the prep, is followed by a dissyllable; the probable explanation of this difference is that nearly all the combinations involving dissyllables are needed for the formation of verse-closes, e. g. St. 460 praetér pedes ; Mi. 9 proptér virûm. (2) W e find always the accents âpatre, inforó, etc., and not a pâtre, in fóro, cf. A. J. P. X X V 262 f. ; corrupt is As. 463 ipse in manu | habébo, with hiatus, and R u . 206 is needlessly scanned by Skutsch, Forsch. 158, quisquam homo mi obviam, instead of quisquam homô mi obviâm. Pl. also accents in prep, phrases only proptér-viâm, proptêr-maré, praetér-cas\_am~], and never propter viâm, propter mâre, praéter câs\_am\ although we find the latter accent freely used in all other combinations of the same value ( , ^ ¿0, e. g. tânto mâgis (Mo. 8 3 1 ; Ba. 557), multo mâgis (Per. 232 ; Ps. 656), nilo mâgis {Am. 908; Cas. 264). For sûprâ-lacûm admitted as a quadrisyllabic word in iambic verse-close, v. Klotz, Grundz. 244 ; cf. also in the hexameterclose Verg. A. 3, 695 subtér-mare, qui nunc. Similarly we have 1 It is noteworthy that the very frequent cases of ' enclisis ' in intéreos, intérse, pradersfem, etc., are nowhere expressly mentioned by the ancient grammarians, and we must suppose that they regarded such cases as sufficiently provided for by their general rules. T o these rules as cited above (A. J . P. X X V 263 f.) add Priscian, K . I l l 27, 20 (Schôll, De acc., p. 186): annectitur semper praepositio sequenti dictioni et quasi una pars cum ea effertur, quamvis per apfosiiionem proferatur.

LATIN

WORD-GROUPS.

417

a l w a y s äpüdforüm, sine modo, a n d procell. like apüdförüm, sine mödö ( d i s s y l l . n o u n ) are w h o l l y e x c l u d e d (sine b o n o o n l y in a n a p . 1 , e. g . R u . 937 b ), a l t h o u g h , if the g o v e r n e d noun is trisyllabic, the use o f the procel. is quite frequent, e. g . H a u . 30 sine v i t i i s ; E u . 1044 sine d ü b i [ o ] ; PI. fr. inc. 7 a p u d homines ( A h l b e r g , Procel. 135 ff., cites 11 s u c h procell.). (3) If, d u r i n g the republican period, the accent e v e r rested u p o n the noun in c o m b i n a t i o n s consisting of prep, -f- dissyllabic noun, w e o u g h t to find e x a m p l e s of procell. l i k e ita In mdnu quidem In cdput, et in märe, etc. W e find in fact procell. with j u s t s u c h s h o r t e n i n g of the prep., but o n l y 2 w h e n the g o v e r n e d noun is trisyllabic, e. g . R u . 1302 ita in m a n i b u s ; 765 q u i d e m in ca 4 pite; C u r . 6 1 1 v e l in c h l a * m y d e m ; C a p . 1000 fui In la 3 pici | dinas (cf. A h l b e r g , Procel. I 1 3 9 - 1 4 7 , w h o , h o w e v e r , omits R u . 765). T h e a c c e n t m a y not then m o v e forward, but PI. allows it to freely recede further in the case of dissyllabic nouns, esp. in the i s t ft., e. g. C u r . 354 s i ' b i i n m a n ü m ; A m . 1021 ; Ps. 1294; T r i . 914, v. still other e x a m p l e s in Müller, Pr. 3 4 4 ; cf. St. 5 1 7 i'n h ü n c di£m ; C a p . 426 iM üt s c i ä s ; A m . 357 ; C a p . 363. The recession w h i c h all these e x a m p l e s exhibit is that of the quid exprobrds t y p e ( T r i . 3 1 8 ; 250; E u . 233, etc.). (4) W e find L a c h m a n n ' s law carefully observed in the prepositional composita, i. e. w e h a v e either dpatre, inforo, etc. {med. and ult. v.~), or, if the final s y l l a b l e is elided, elision a l w a y s 3 o c c u r s in d i a l o g u e v e r s e 4 into an acute, that is, w e h a v e either or but not ( A . J. P. X X V 262). In other w o r d s , w e m a y h a v e s e n [ e m ] ilium ( B a . 1150), but not ä d s e n [ e m ] Illüm, manu emissus ( A s . 4 1 1 ) , but not e m a n u emissus. I h a v e e x a m i n e d , with the help of R a s s o w ' s i n d e x , a sufficient n u m b e r of prepositional p h r a s e s in PI. to determine ' H e n c e read with synizesis, E p . 426 sine tuo labore ( A h l b e r g , 1.1., 154); Per. 63 sine meo p e r i c l o ; M i . 135 apud suom paternum. ' M ü l l e r , Pr. 181, wrongly emends the corrupt v. A m . 157 to neque in erö (iamb. Oct.), and G o e t z does little better, i. e. in erö sit ; — r e a d with synizesis St. 62 quidem In siio ( A h l b e r g , 1. 1., 154), not quidem In süö. 3 S c a n Cas. prol. 36 cümpatre in Illisce, not in illisce ; so M i . 95 quömodo äd hünc, not quömodo ad h ü n c ; Ps. 1242 (Müller, Nachtr. 15) is corrected from A, or may be scanned obviam ei ültro deferäm (P). V e r y exceptional is C L E . ( B ü c h . ) 2 3 1 , 2 i4n mänü, cf. pectöre. 4 In anap. w e have of course obviam Ignobilis (Ps. 592), inlöcö (Per. 843), hüncdlem (ib. 768) etc., cf. above p. 409.

38

418

AMERICAN

t h i s fact b e y o n d f o l l o w i n g table : Composita

JOURNAL

all q u e s t i o n ;

Elisions I n t o Acute

aforo deforo ïnforo adforum apatre cumpatre clampatrem adpatrem adsenem mare

5 i 4 12 6 2 i 5 2 4

OF

PHILOLOGY.

t h e results a r e s h o w n in

Composita

the

Elisions I n t o Acute

periovem periocum incrucem invia devia propterviam obviam indiem inloco abdomo Total

2 2 i 2 i i H 31 2 2 72

T h e cases of t h e elision of adforum into an a c u t e m a y be c i t e d h e r e : Ba. 902; 1060; Ps. 561; 764; T r u . 313; M e r . 7 9 7 ; T r i . 727; P e r . 487; A s . 245; 367; Mi. 89; 930; A n d . 226; 356; A d . 277; P h . 312; 921. {Total, 17);—similarly aforo\A11. 356; Mi. 578; M o . 998; P e r . 442; P o e . 929; A d . 6 4 5 - , — i n f o r o : A s . 1 1 7 ; C u r . 475; 507; T r i . 651. Finally, t h e p r e p o s i t i o n s a r e s h o w n to be a c c e n t e d b y t h e fact t h a t t h e y a r e o f t e n p l a c e d in t h o s e feet of L a t i n v e r s e w h i c h conventionally r e q u i r e a tonic syllable in t h e arsis, e. g. the fifth arsis of t h e h e x a m e t e r , as a b l o v e s f i m m o ( V e r g . A . 1, 380), a n d t h e t h i r d arsis of t h e t r i m e t e r , as L a b e r . m i m , 1x3 in b o n o a e q u e a t q u e in m a l o ; S y r . s e n t . 135 c u m m a l a f a m a . SUMMARY.

T h e cases of g e n e r a l recession in t h e L a t i n a c c e n t s y s t e m a r e strictly limited in n u m b e r , that is, r e c e s s i o n of this s o r t o c c u r s only u p o n monosyllabic pronouns, prepositions, conjunctions and a d v e r b s , a n d also u p o n dissyllabic p r e p o s i t i o n s a n d p r o n o u n s (for t h e latter, v. T r a n s . A m . Phil. Assoc., 1904). F u r t h e r w h e n c o n j u n c t i o n s , p r o n o u n s a n d t h e like a r e involved, r e c e s s i o n o c c u r s o f t e n in t h e w o r d - g r o u p s ^ ( A . J . P. X X V 160, 260, 269), less often in t h e g r o u p s - - - ( T A P A . X X X I V , 93 if.). T h e c a u s e s of t h i s recession lie in t h e t r a d i t i o n a l I. E . w o r d - o r d e r , 1

No. of cases in both PI. and Ter.

LATIN

WORD-GROUPS.

419

which has been very completely preserved in Latin, and, in a number of instances, they probably lie also in an original I. E . 'enclisis', which has been retained in Latin ( A . J . P. X X V 156, 259 2 , 260, 261 \ 263, 410). Since Latin has preserved no other word-orders in the same degree as those just mentioned, the Latin accent has, in all other cases—except of course in special phrases—acquired a definite place in single words and become a fixed accent (jesie,gebundene Belonung, v. Seelmann, Ausspr. 18; Hirt, Indogerm. A k z . 19). T h e suggestion has been made, to be sure, both by L i n d s a y and by S k u t s c h in recent writings, in which they have developed the theory of Ritschl to its logical conclusions (Journal of P h i l o l . X X 153; T h e Captivi, p. 3 6 2 F o r s c h . 156 med?), that the Latin accent possesses the power of receding over almost any iambic word, e. g. essiirn vocat, and is therefore to a considerable extent a free accent tfreie Betommg). After examining this hypoihesis with interest and without adverse prepossessions, I am led to reject it at every point. F o r it is only in a very limited way that w e find evidences of accentuations like aquàm-velim2 (obj. + verb), and we practically find no evidence at all for accents like rectà-via (adj. + noun) or recié-facis' (dissyll. adv. + verb). Hence we must conclude that a general recession of the accent takes place only within certain definite limits. Further, the conclusions which we have reached are, to a large extent, independent of any very special theory of the Latin accent, but are in accord with certain generally recognized facts of development. F o r both the stress theory of the Latin accent and the musical theory, in the form in which the latter has been thrown by Havet and V e n d r y e s , agree in assuming that in the 1 It is scarcely necessary to say that I accept very largely the v i e w s w h i c h Prof. L i n d s a y has advocated in the works just cited, but I do not accept the v i e w that in the common pronunciation any iambic word can throw its a c c e n t b a c k , e. g. essiim-vocat, dljfractis-velim, rectd-via. In general, there is no t e n d e n c y w h a t e v e r to make accent and ictus agree in the last dipody of iambic lines (v. Schlicher, A . J. P. X X I I I 50, whose statistics unfortunately m a k e no distinct a l l o w a n c e for the numerous cases l i k e in viam, illaéc siet, aliquó modo, etc.), and in accordance with the principle to be stated b e l o w , it is precisely in the last dipody of iambic lines that the developed iambus of Tragedy requires that accent and ictus shall (practically) always disagree, as H. A . J. Munro acutely pointed out, Journal of Philol. V I 75, and T r a n s a c t i o n s of the C a m bridge Philosophical Society, X 386. 2

Cf. A . J. P. X X V 258.

3

cf

. X A P A . 1904, p. 49.

420

AMERICAN

JOURNAL

OF

PHILOLOGY.

pre-literary period an accent of stress or intensity rested upon the initials. In my judgment, however, these conclusions accord better with the stress theory of the accent (cf. T A P A . X X X I V , P- 95)CONCLUSION.

R E L A T I O N OF W O R D

AND VERSE

ACCENT.

The question may be asked whether the conclusions reached in the preceding study throw any light upon the larger problem of the relation which exists between the Latin accent and Latin quantitative verse. 1 Although it is difficult to answer this question within the limits of the present paper, I am unwilling to bring this study to a close without indicating briefly the manner in which I think that its results should be interpreted in relation to the general problem, 3 which has called forth such wide differences of opinion among Latin metrical scholars. In the first place it will be noted that all the tests which have been employed in the preceding investigation are based upon the well-known fact that the verse-accentuation of genere, prodttlo, caldmi\t-a.Kem, dpe\ rire pectdre \ imperl \ osus,4 etc., must agree with the word-tone. This usage means that, in all ordinary forms of pronunciation, the Latin tone greatly weakened a following; short syllable, or two following short syllables, in case the second of the two shorts was not a final syllable. Now it is true, in general, that the various metrical tones or cadences of the ancients afurpos t6vos, Hdt. I 47; Tplfierpot tows, id. I 174) always involve to some extent the song-like utterance known as the peai) niveau or medium genus (Aristid. Quint., p. 7, 23 M.; Mart. Cap. I X 937; cf. Quintil. I 8, 2), and are therefore wholly independent of the tone% or ' I n this w h o l e discussion I use the term ' a c c e n t ' in the sense of the g r a m m a t i c a l accent, the accent of popular speech (serma). H e n c e my position would not be affected in the least, even if one should wish to assume with Z i e l i n s k i , D a s C l a u s e l g e s e t z in C i c e r o s R e d e n , L e i p z i g , 1904, p. 225 ff., the existence of an ' oratorisch-poetischer A c c e n t ' ; for the real subject of m y i n q u i r y is, what relation does this ' o r a t o r i s c h - p o e t i s c h e r A c c e n t ' , i. e., in m y j u d g m e n t , the accent which naturally accompanies the /lian nivr/cu; and the 7rUofia, b e a r to the ' V u l g a r a c c e n t ' . 2 U p o n this relation, see also m y article " S t u d i e s in L a t i n A c c e n t and M e t r i c " , T r a n s . A m . Phil. Assoc., 1904. A still f u l l e r discussion of the subject is proposed e l s e w h e r e . 3

F r o m the time of T e r e n c e , v. K l o t z , Grundz. 276 ff. F r o m the time of Phaedrus, v. T A P A . X X X I V 67. 5 S i n c e the G r e e k and L a t i n accents h a v e both of them a large melodic e l e m e n t and appear to differ only in the degree in which they have developed 4

LATIN

WORD-GROUPS.

42X

cadence (tovos, irpaa-ahla) of the v o i c e in colloquial s p e e c h , indep e n d e n t also, in general, of the w e a k intensive element of the tone. T h i s i n d e p e n d e n c e d o e s not, h o w e v e r , exist in all c a s e s to the same e x t e n t ; for the ancients are a g r e e d that o n e of the metrical cadences, viz. the iambic, a p p r o a c h e s m o r e nearly than the rest to the tone o f colloquial s p e e c h . Hence we should e x p e c t to d i s c o v e r in iambic verse, if a n y w h e r e , traces of the direct influence of the tone, and w e are not surprised to find that the p o e t or poets, w h o a r r a n g e d the L a t i n iambic, i m p o s e d u p o n it from the first the law that it s h o u l d respect the colloquial tone of gene.re, caldml\ tatem and aperire, a n d that a similar law was a c c e p t e d later forp'ectore and e v e n for imperi\ osus. the intensive element, I have no hesitation in following the Roman custom of applying to Roman pronunciation all those terms which the Greeks employ of their own language, i. e. /ie?.oc, dp/iovia, rdvoi;, npooadta, etc. For the sake of brevity and of making my account fit both G r e e k and L a t i n verse, I have also intentionally followed the example of many Roman writers and perhaps of H d t . (cited above) and of D i o n . H a l . (De admir. vi D e m . c. 48; D e comp. verb. c. 11, cf. K l o t z , Grundz. 269 versus Crusius, litt. Centralbl. 1891, 7, 213), among Greek writers, in failing to distinguish sharply between ictus, which is ' stress with a slight musical e l e v a t i o n ' (Humphreys), and the Greek tone, which is ' m u s i c a l elevation with a slight stress'. T h i s lack of sharp distinction seems to me legitimate, since, in the psarj Kivrjctg, the ictus obscures or obliterates the tone, and thus very largely takes its place. In fact, we may safely go considerably beyond the definitions just quoted; see my note on ictus in T A P A . , 1904, p. 51 f. T o that note I now wish to add the following: Besides the examples of roiwf ('musical m o d e ' , ' m e l o d y ' ) cited in Stephanus' Thesaurus, note especially Schol. Marciana in artis Dionys. § 2, p. 307, 38 Hilgard: ¡not; Heyerai ml t6vo