MA S TER NEGA TIVE NO. 92-80599-13 MICROFILMED 1992 COLUMBIA UNIVERSITY LIBRARIES/NEW YORK as part of the "Foundations of Western Civilization Preservation Project" Funded by the NATIONAL ENDOWMENT FOR THE HUMANITIES Reproductions may not be made without permission from Columbia University Library COPYRIGHT STATEMENT The copyright law of the United States - Title 17, United States Code - concerns the making of photocopies or other reproductions of copyrighted material... Columbia University Library reserves the right to refuse to accept a copy order if, in its judgement, fulfillment of the order would involve violation of the copyright law. A UTHOR : RADFORD, ROBERT S TITLE: NOTES ON LATIN SYNIZESIS PLACE: CHICAGO DA TE : [1 908] COLUMBIA UNIVERSITY LIBRARIES PRESERVATION DEPARTMENT Master Negative # DIDLIOCRAPHIC MICRnFORM TARCFT Original Material as Filmed - Existing Bibliographic Record BKS/PROD Books FUL/BIB NYCG92--B31839 Record 1 of ~ Record added today T lD!WYCG92~B3ia39 CC:966B BLTiam # w J» ?s> DCFfi? IMTs? CSC I? G P Ci f\ REP if MOD ; ' B 1 : -^ FRMn BNR: F I C n '^ FSI:? COL ; Restrictions on Use: 1 00 1 245 10 260 300 LDG QD Acqu i si t ions MYCG-PT Lseng PD :196s/ OR: POL: DM: RR IMNC-- f rNWC Radford, Robert S, Notes on Latin Syn i ;:!es is- f hCmt crof or ml • ChicK^go, =:| bUniversi ty of Chicaqo Press, ~ 1 {-[' l 908] I" 153:1 -IAS p. ORIB 05 -06 -9 2 ri3 : EL. ATCs COM 5? I L C : : E;>!L s •-\ '-> n /•;, /t , ., AD: LJD : 1 1 : GEN s 05-06-92 05-06-92 o BSEi TECHNICAL MICROFORM DATA // X REDUCTION RATIO: FILM SIZE: }AZ2S22- IMAGE PLACEMENT: lA (© IB IID DATE FILMED: iLlJLi.li^ INITIALS__:>>:_j) C FILMED BY: RESEARCH PUBLICATIONS. INC WOODBRIDGE. CT r Association for Information and image Management 1100 Wayne Avenue, Suite 1100 Silver Spring. Maryland 20910 301/587-8202 Centimeter 1 2 3 lyiiiijiiiiiiiiiiniiiiuiiilii TTT Inches 1 iliiiiliiiilim jim IIIIUIIIIJIIIU 8 9 10 iiiliiiiliiiiliiiiliiiili TTT n 1.0 U^ 2.8 1^ ||3-2 lA 11^ li. 1.4 2.5 2.2 2.0 1.8 1.6 I.I 1.25 mi 12 13 14 iiliiiiliiiiliiiiliiiiliii TTT 15 mm m^ MONUFOCTURED TO RUM STONDRRDS BY fiPPLIED IMfiGE, INC. « €^ >^ "ii" i y t aAJo-^- /7x>i OX/i wtmrnttatm /1\n 1/f^ 7T^ 4^-1/«^^^:' ,,s2,xX>.«ea<2_/xj2-_^ a- -r- '^ Notes on Latin Synizesis ROBERT S. RADFORD 'I Reprinted from Classical Philology, Vol. Ill, No. 2, April, 1908 PUBLISHED BY THE UNIVERSITY OF CHICAGO PRESS, CHICAGO ForeieJi Agents: London: Luzac & Co.; Leipzig: Otto Habrassowit2 J If I '} M ?*& I. NOTES ON LATIN SYNIZESIS By Robert S. Radford THE RELATION OF OLD LATIN SYNIZESIS TO THE SENTENCE- ACCENT There is no more familiar phenomenon in the verse of the early Latin dramatists than the quantitative reduction of words which show a vowel in hiatus, e. g., eoSj eorum, deos, deorum^ fui, fuisfi\ yet the precise manner in which this reduction has taken place is still a matter of discussion among philologians. Accord- ing to some critics, iambic shortening is the real influence at work here, and we should pronounce eos, edrum, dedrnm, etc. ; accord- ing to others, a slurring of the first of the two vowels has taken place, and we have to recognize in the treatment of such cases that procedure which is commonly termed by the ancient metri- cians synizesis and by Romance scholars diphthongalization.^ The latter explanation, which finds strong support in the synizesis phenomena of many other Indo-European languages,^ has always commended itself to the majority of Plautine students, but, in becoming its exponents, the latter have usually been content to employ the term 'synizesis' in too vague and indefinite a sense. This word has, in fact, a somewhat variable meaning, and the three great periods of the Roman language, viz., the Old Latin, the Classical, and the Romance, show, upon the whole, three fairly distinct types of the synizesis process. For although all synizesis ' Diphthongalization is not precisely the same process as Old Latin synizesis (Trans. Am. Phil. Ass. XXXVI, pp. 170 flf., 179), but the two processes have many points of similarity, and are often identified (cf. Schuchardt Vokalismus des Vulgdr- lateins II, p. 510). 2 For example, the Romance possessives are derived as a rule from dissyllabic wieifm, tuum, etc., but they often develop independently a diphthongal pronunciation very similar to that of Old Latin. Thus, in O. Span., mio, m& are almost invariably monosyllabic in the proclitic position (Cornu Romania XIII, pp. 307 ff. ; cf. Trans, XXXVI, p. 195, n. 1), and a similar treatment existed in Provencal, as the following lines from Appel's Provenzal. Chresfomathie will serve to illustrate ; No. 108 (p. 159), 144 (La nobla leyczon) : diczent : vene vos en, li beneit del rnio payre ; No. 74 (p. HI), 29 (Raimon Gaucelm ) de la mia mort, per so siatz a mal mes. [Classical Philology III, April, 1908] 153 Robert 8. Radford rests upon the tendency of the short vowels i {e) and u in hiatus to assume a semi-voealic character, and no thoroughgoing dis- tinctkm om comequentty be made between Greek and Roman usage (Zander Ven. limlf ^ cavii),, y«l it is true that the type which is usual ill Greek and Classical Latin is chiefly employed as a convenient and an artistic device for the purpose of intro- iiiciiig''difficiilt wwi^Jbnusiiito the stately and sonorous movement of the verse ( Tram,, p. 167) .' The synizesis of Old Latin dialogue verse, on the other hand, is entirely free from poetic artifice and Wholly •ifoiitMieonB In ito character. Finally, the extensive gynizesis of tie Late Latin period often causes the semi-vowel i to merge itself i& a preceding consonant, to which it gives a platal ciiiiiter, m seen, ia. ft., singe from, *9imya, bras from ^hrmtfum, etc, (Mndsay L(tl Lang., pp. 81, IM, 263). It is not sufficient then to speak of synizesis in general terms, bat il 18' neceBBiirj to nqmiie ipecifically ''into the extent, the cause and tlie »al' ciiractttT" rf tie OM Latin variety. Hence I have sought to show at some length in im article published in the Trans. Am, Phil, Assoc. XXXVI (1905), pp. 158 ft., that pre- cisely lial sequence ^: oyllaUeB' and that position (rf the accent which causes iambic shortening in the case of vowels separated by a consonant, has produced synizesis in the case of the half- vowels I («) and tt in iiatnii. Tiia iw tiid tie difficult quantitative sequence — 1 alike in ddmt frdtrem and in mmm frdtrem, but the method which is employed in escaping from the difficulty is diiefent in the two examples. In caies lili 'flii> iait^/tie Romans naturally tended, as they hastened to pronounce the following accented syllable, to shorten the second syllable of iambic words and word-beginnings, thus giving rise to the phenomena of Brevis Brevians, e. g., domi frdtrem, vdlfiptdtem; in cases like the second, however, the language offers a simpler and an easier method of removing the difficulty in question through the slurring of the initial syllable of the iambic word or word-beginning, and thus exhibits the varied phenomena of Brevis Coalescens, e. g., {e)dritndem, l{i)en6sus, m{e)um frdtrem. With the weakening iThis is also the character of the synizesis which is admitted in Old Latin anapaestic verse, if synizesis be the true explanation of such phenomena, e. g., aur{e)ds ; cf. Zander Vers. Ital., pp. cxxvi ff. Notes on Latin Synizesis 155 of meum in the last example, we may well compare, as Professor Fay kindly suggests to me, the unemphatic and colloquial English possessive which is heard in *ml Lord,' *ml brother,' *ml friend,' and the like. Although many points of similarity, as has just been indicated, exist between iambic shortening and synizesis, there are also impor- tant points of difference, and the numerous cases like dedrum, edmus, qiiieto, tiidm-rem, med-quidem show us plainly that we cannot possibly read all iambic words and word-beginnings with shortening (e. g., dSdrum, edmus, tudm-rem, etc.), and so remove synizesis entirely from the dramatic poets, as C. F. W. Mailer, Skutsch, and Havet have proposed to do. In addition, the vulgar Latin forms of a later period should be closely compared with the early Latin phenomena. These have been most fully collected from late inscriptions and from MSS by Schuchardt, Vokalismus des Vulgdrlaieins II, pp. 441-519 ; III, pp. 295-311, and are referred by him to various subdivisions.' The following citations are especially noteworthy: des {zes), debus {zebus), de for dies, » Schuchardt's treatment of this whole subject is a valuable and suggestive one. He points out (II, p. 44^^) that three phenomena are comprehended as final results under the term synizesis: (1) Consonantization, 'Konsonantirung,' e. g., genua Verg. A. v. 432; (2) Elision, e.g., sem{i)animes Verg. A. x. 396; (3) Contraction, 'Kontraktion,' ' Zusammenziehung,' e. g., reice Verg. E. iii. 96. To the consonantization of the semi- vowels (II, pp. 442, 502) he does not assign an especially important rCle, but classifies his material chiefly under the phenomena of ' elision ' (II, pp. 441 ff. ) and ' contraction ' (II, pp. 505 ff., 510 ff.). While admitting the extreme diflBculty of distinguishing sharply between the two last-named processes, Schuchardt adopts the criterion that 'elision' preserves the quality of the second vowel, as in Thodorus, debus, quescit, while ' improper contraction ' preserves that of the first vowel, as in Thedorus, dibus, quiscit ( II, p. 442). The proposed criterion is, in my judgment, far from being always a con- clusive or a satisfactory one, and leads to a frequent separation of examples which properly belong together. Thus the forms debus and dibu^, quescit and quiscit, which Schuchardt is compelled to treat separately (II, pp. 445 ff. and III, pp. 295 ff. ; II, pp. 513 ff. and III, pp. 310 ff.) may very possibly all alike be the result of contraction, and the variant spelling in these cases probably points only to a pronunciation of the vowel which is intermediate between c and i. It seems safe then to adopt Schuchardt's first form of statement (II, p. 442) and to conclude simply that in all the cases in question the two vowels form a syllabic unity and thus produce 'diphthongalization,' the latter term being here used in a suflaciently broad sense to include combinations like ai as well as those like di. Schuchardt is clearly correct, however, when he maintains further that the word-accent affords no certain criterion between the two processes : " through inversion of the accent a contraction-diphthong may arise out of an elision-diphthong, and vice versa. Beside nofitus = nedfitus = nedfitus stands nefitus = niofitus ; so sos - su6s = s4os = silos (11, p. 443 ; III, p. 333) ; des = dih = dies = dies (II, p. 445) ; capredla = capredla = capriola = capr^la (I, p. 427)." 156 Robert S. Radford Notes on Latin Synizesis 157 diehus, die (II, p. 445; III, p. 295; I, p. 67 ff.; cf. Seelmann, Ausspr, d, Lai,, pp. 239, 323); dis, di{s), dibiis for dies, etc. (II, p. 513 f.; Ill, p. 310); dae, do for deae, deo (II, p. 463; III, p. 298); dende for deinde (II, p. 513); andem for eandem (II, p. 463); sa, so, su (abl.) for sua, etc. (II, pp. 464 fF.); dos for duos (II, p. 467);* dodeci for duodecim (II, p. 467);* quescit, Quetus, quiscit, Quitns for quiescit, Quietus (II, pp. 448 ff., 514 f. ; III, p. 296) ; jmlla, pullae for pwe/Za, puellae (II, p. 518; fustis,^ fut tor fuistis, fuit (II, p. 519).* Schuchardt makes no mistake, I think, in repeatedly comparing these late and vulgar spellings with the Old Latin phenomena ( Vokalismus II, pp. 444, 464, 511, etc.).* The early and the late Latin forms alike give evidence of the weakness of the semi- vowel in hiatus, and it does not seem unreasonable to suppose that the Old Latin dissyllabic pronunciations of quiescit, puella, fuistis, eandem, deinde ( Trans,, p. 182) were always largely pre- *Cf. the Uinbr. contracted form dur 'duo,' from *duur, *duds^ Buck Grammar of Oscan and Umbrian^ §§54, 82; cf. also Span, rfos, Pg. dois, Fr. deux, Wal. doi, *0f. Ital. dodtct, Span, rfoce, Pg. Pr. doze. 'Cf. Ital. /oste, Vg.fostes, Vt.fotz, Fr.fUtes. *0n the early popular form fut (cf. Ital. /u, Pg. foi, Pr. /o, Fr. fUt), see also W. Meyer, K. Z. XXX, p. 341. The precise manner in which the synizesis forms /wsfia and puXla have arisen is not quite clear. The vulgar puila might perhaps be explained as derived from the original form *pAerula (Vanicek Etymol. WOrterbuch^ p. 550) through the intermediate stages *piJ(e)ru/a, piir ''la, but Schuchardt's explanation (II, p. 511) of a contraction-diphthong due to inversion of the accent, i. e., a shift of the accent from the first to the second element of the diphthongal sound, is also an attrac- tive one (see also above, p. 155, n. 1) : '* Der Wortakzent ist hierbei zunftchst indifferent. Aus pu^lla^ fuisse wurden allerdings zunftchst puHla^ fulsse, dann aber (wie sp. vMnte = veinte^ fr. empereur = emper^or = emperedr) piXella^fHisse, wie aus den Schrei- bungen pnlla.fusse hervorgeht. Daher scheint die Oorssen'sche Annahme der Betonung puilla^ fuisse (II, 212 fg.) fflr die Messungen piee//a, fuisse bei Plautuseine unsichere zu sein." Other probable examples of such diphthongs and long vowels as the result of contraction in Latin are coepi from co-epi (Stolz Hist. Oramm. I, p. 155) and (^)ctas (Varro R.R. iii. 16; 28; cf. Caper Gramm. Lai. VII 94. 16) from codcttis (Stolz loc. dt.^ p. 219). Analogy may have exerted an influence upon some of these forms, but Victor Henry's assumption {Conip. Gramm.*, Eng. transl., §73, p. 84) that the con- traction which is seen in coepi has first arisen in forms like cdepisti seems, upon the whole, unnecessary. ^The monosyllabic pronunciation of cuius^ huius does not belong to Old Latin synizesis in the sense in which the term is here employed, but is ' probably due to the loss of j between two like vowels' (Birt Rhein. Mus. LI, p. 247, n.) ; compare also the shortened pronunciation of illius^ istius. Schuchardt ( Vok. II, p. 508) quotes here the plebeian forms cus and hv^, also cuis and huis; see also Corssen 11^, p. 182, and Luchs Studem, Stud. I, pp. 319 ff. J served in vulgar speech and were essentially identical with the late and vulgar quescit, quiscit, pidla, fustis, andem, dende just mentioned.^ This latter supposition, though an extremely prob- able one, is, however, incapable of absolute proof, since a new and independent development might also have produced these forms in the later language. In any case the late synizesis is considerably more extended in its use than that of the early period. For if we except the few and somewhat uncertain examples like evenat, augura, or{i)undi {Trans., p. 169) , we find the Old Latin synizesis strictly limited to the quantitative sequence ^ _, in cases where this is initial;^ the later type, however, is wholly unre- stricted and depends solely upon the weakness of the semivowels in hiatus. Thus the Old Latin type shows in dialogue meters only die, eat, quiescit {Trans., p. 174), but the late language employs also very freely pride (Schuchardt II, p. 445), exatis, exuntes (II, pp. 463 f.), requevit (II, p. 450), facendum, adridat,_ Thodoro, etc. To return to the early Latin occurrence of these phenomena, the dramatists admit synizesis most frequently in proclitic and 'enclitic' words like the possessive or demonstrative pronouns and the substantive verb, which have little appreciable accent of their own (e. g. ni{e)iim frdtrem, {e)dm-rem, f{u)i liber), but they also employ it freely in the case of many substantives and verbs like die, deo, scio, which have the ordinary intensity of tone. It is in the treatment of this last-named class of words that I fear my former discussion was not suflficiently clear, but requires some amplification and enlargement.* Thus, in explaining the occur- rence of synizesis formerly, I properly attached much importance »Cf. Schuchardt Vok. I, p. 59: "Oft ist die Aehnlichkeit zwischen der vulgftren Sprache des 4., 5., 6. Jahrh. n. Chr. und dem alterthflmlichen Latein betont worden. Unndthigerweise ; dies alterthtimliche Latein ist welter Nichts, als vulgftres." 2 The species of syncope by which vowel u was converted into consonant u after h ifi 9» 9 and s, e. g. in larva^ milvos, reliquos, etc. — earlier larua, miluos, relicuos — is still unknown to Plautus and belongs to a somewhat later stage of the language (Lind- say Lat. Lang., p. 46; Capt., p. 20). After other consonants vowel u is simply lost through this process, e. g. in quatt(u)or, quatt{u)ordecim, but these latter forms are scarcely attested for Plautus (Trans., p. 174, n. 3), and are first clearly shown for Ennius, cf . Georges Lex. Wortformen s. v. ; Grdber ALL. V, pp. 127 f . ; Schuchardt II, p. 519; III, p. 311. • 8 Of., however, Trans. XXXVI, p. 193, n. 1 ; p. 195, n. 2; p. 210. 158 KOBERT S. RaDFOBD Notes on Latin Synizesis 159 in several cases to the weakened uses of some of these forms, e. g., to the trite or emotional use of deo and the parenthetical use of scio {he. cit, p. 181, n. 1; pp. 195 f.). No explanation can be really complete, however, which does not recognize the fact that the slurred forms of deo, die and scio occur not only in weak, but also in fairly emphatic uses of these words, which does not meet the very plausible arguments which the opponents of Old Latin synizesis advance at this point. For the latter claim that all iambic words like d^o, die, scid, etc., which have a distinct accent, have their final syllable shortened by the accent, and there- fore cannot well have the first or accented syllable slurred in subordination to the second. Although this argument has been confidently employed against genuine synizesis from the time of Corssen (cf. II, pp. 761 f.) to the present, I believe that it will appear upon closer examination to be wholly fallacious. Thus — to consider first the cases of iambic shortening — although the accent of the single iambic word is undoubtedly one of the factors in this process, yet it is now generally recognized that it is far from being the only factor, or even the chief one. For in actual speech we are not concerned so much with individual words as with the phrase or the sentence. Hence it is not the iambic word as such that we usually find shortened, but the iambic word in certain sentence-phrases, e. g., vold-scire, hene-fdctum, iihi-dico, dedi-d&no, have-frdter (in verse also d^di-dond, hdve-fraUr, vdlo- s0f. also Am. Jour. Phil. XXVII, p. 434. i ilf often unaccented," and cites as examples the frequent shortening of such proclitics and * enclitics' as apiid {m6nsam),^ enim {vSro), tamen {n^queo), quidem {praetor), aptly comparing with these the shortening seen in voMptdtem, senectiltem, and the like. To these cases of weakening I should like to add the almost complete loss of final s which Leo {Fo7*sch., pp. 267 ff.) has pointed out in the subordinate adverbs nimis, satis, magis, and which Hauler {Einl. z. Phor., p. 50) notes also in prius. Such examples show clearly that the principal factor in iambic shortening is not the accent of the individual iambic word, but the accent of the phrase or of the sentence in which the iambic word is placed. Hence, even in the case of those terminations which were finally short- ened entirely, e. g., o, or, at, it and the like, we clearly have a right to assume that the shortening of such words as homo, void, dabo, vetor, vetdt, dedit, etc., began chiefly in sentence-phrases (cf. Lindsay Caj^t., p. 33; Lat. Lang., pp. 210 ff.), although it must be freely granted that the shortening process was here assisted by the accent of the individual word. Important as the individual accent is, it is often profoundly modified in the sentence, and if we wish to obtain practical results in accentual study, our doctrine must not be one of individualism so much as one of collectivism and association. In questions of accent, we cannot, to be sure, neglect the study of the single word, but we must fix our attention still more upon the sentence, since it is the organism of which single words are but the parts and the instruments. Thus the substantive and the verb are uni- versally admitted to be the most strongly accented parts of speech, but even their accent is often greatly weakened in the sentence in consequence of their association with other words, so that, in calling them strongly accented, we scarcely mean more than that they are pronounced with stress in the majority of their uses. We may profitably compare the accent of a simple English sentence such as *I cdll the g6ds to aid'; if three distinct accents are »Leo {Forsch., pp. 226 f.) appears to go too far in maintaining that the usual pyrrhic scansion of apiid in Old Latin verse is due to a definite loss of the final d. The whole particle was greatly weakened in pronunciation, and as a consequence the final consonant was no doubt sometimes obscured; cf. Corp. Gloss. Lat. II, 21, 40: ''ape^ irapd. » 160 KoBEBT S. Radford 1] observed in such a sentence, we recognize that it is spoken with sharp distinctness, but in hurried colloquial speech it is much more likely that only two accents will be clearly heard, e. g., 'I c&U (the gods) to aid,' or '(I call) the g6d8 to aid!', while the words which are inclosed in parentheses will be slurred or treated as subordinate. Similarly, whenever, in a Latin sentence like deos quci^so tit sit sup^rstes {And. 487), the chief accent of the sentence was thrown upon qna^so, the individual word deos was made subordinate to a certain extent, and consequently the accent of the first syllable was not left strong enough to resist slurring under the existing phonetic conditions.* In the case of such a sentence, no doubt there existed originally two forms of pronunci- ation differentiated by the place of the chief accent, viz. d[e)osqiia4so and dSos quaeso, and we may say in general that during one period of Old Latin both d{e)os and deos must have existed, and that one or the other of the two forms must have been used according to the accent scheme of the particular sentence. Owing to the fugitive nature of Latin i [e) in hiatus, however, the former pro- nunciation proved so much easier and more euphonious that, in the time of Plautus, it was almost exclusively in use.* 1 So also, if Plautus has a few cases (chiefly in the first foot) of neglect of common word-accent, as in Amph. 761 de^disse dono hddie (cf. Ahlberg Cotv^ept. iamb.^ pp. 90 ff.), such a passage was probably not pronounced dedisse d6no hddie^ but rather dedisse dono hodie^ the voice hastening over the whole word dedisse and coming to rest, as it were, upon dono. For a somewhat simihir view, which, however, needlessly suggests a word-group dedisse-ddno^ cf. Lindsay Capf.^ p. 36. ^In the case of forms like deo, die, it may perhaps be remembered in addition that, in vulgar Latin, di (de) tended to pass into the sound of y, and at a later period into that of z or of simple d. This latter pronunciation gives rise to the vulgar spellings do^ daen des^ zes<, etc. ; cf. above, pp. 155f. ; Trans. XXXVI. 200; Lindsay ixH'. Lang.^ pp. 49, 84 ; Seelmann Ausspr. d. Lat., p. 323. Thus Lindsay remarks that ' the assibilation showed itself even in the case of accented di, ti.^ The trite use of deus in oaths and prayers, however, is probably the principal factor in producing in Old Latin the much- discussed contract forms di, dis from original *dee (*die), *dees {*dies), but the fuller discussion of these contractions must be reserved for a separate paper (see Am. Jour. Phil. XXIX).— So far as regards the full spelling of the singular forms, examples of dio and dia are cited from the Inscrr. by Seelmann Ausspr., p. 187 (cf. Trans. XXXVI, p. 194, n. 3), and dium { = deorum,) is read by Jordan in Cato 47. 16. It is not quite clear whether we have here weakened forms of deo, deum and dea, or case-forms derived from di(v)us. but the former explanation is more probable. One has sometimes been tempted to assume also (cf. Lindsay Lat. Lang., p. 618) that we find dlu3 as the atonic form of deus in the locution me Dius Fidius, but, in addition to other difficulties, very grave doubt exists as to the quantity of the i in Dius Fidiv^j Notes on Latin Synizesis 161 One other point requires notice. We have seen that the slurred form d{e)os arises originally in such a sentence as deos quaeso; no sooner, however, is this form fully established here than it becomes possible to accent freely in verse d[e)6s quaeso lit vobis dec4t {Ad. 491, 275), as well as to retain the original accent- scheme deos quaeso; cf. also d{i)e quinti and d{i)6 quinti. To sum up: The objections which Skutsch, Ahlberg, and Gleditsch make against synizesis on the score of the accent of the single word d6d have weight only if deo be the sole word or the last word of the sentence,* in short, only if deo be completely isolated and cut off from the society of its fellows, and thus entirely removed from the normal play of the sentence-accent. Old Latin synizesis is produced as the voice is in rapid motion and is hastening to pronounce a following accented syllable; hence it is excluded from the end of the grammatical or metrical sentence, since, in this last position, we cannot say sit deo, but, in order to produce synizesis, we must have the series continued, as in sit deo grdtia. especially if the reading of Asin. 23 is correct {per Dtum Fidium, where Dium is an almost necessary correction for MS deum). Stolz {Indogerm. Forsch. XVIII, pp. 453 f.) suggests that the scansion Dius in this passage is due to confusion with dius, divu^, and argues also for the existence of the form Dlu^, which he derives from Ind.-Eur. dieus. ^On the similarity of sentence-close and verse-close, cf. Birt Rhein. Mus. LI, p. 266; L. Miiller Res Metr."^, pp. 266 flf. 2 The exclusion of sjniizesis from the close of lines which end with an iambus ( w - ) was explained in my former article ( Trans. XXXVI, pp. 165, 179, 195, n. 1, 208) as due solely to the principle of metrical regularity, but since, in the close of a metri- cal sentence like frdtre nied, synizesis would he entirely dependent upon the m,etrical accent and could not occur in actual speech, it seems very possible that it is excluded from iambic verse-closes by the accentual conditions as well as by metrical convention. The metrical accent alone is probably capable of producing some changes in word- forms, but it by no means follows that its power is unrestricted like that of the word-accent; the statement of Trans. XXXVI, p. 176, probably goes beyond our knowledge here, and requires some modification. On the other hand, the non-occur- rence of tetramoric aureds in full anap. verse-closes, which is pointed out by Skutsch (T4pas, p. 131), does not seem to me to require any special explanation. In my judg- ment, no certainty has yet been reached for anap. verse respecting either the limits of shortening or the occurrence of synizesis except in the case of iambic words. I myself am inclined to accept, for every foot except the last, the anap. scansions perdidt, aiireas, since these latter seem to me to rest on plausible grounds of historical develop- ment, which I have briefly stated elsewhere {Am. Jour. Phil. XXVII, pp. 430 flf.). As regards the non-occurrence of aUreas in anap. closes, and its occurrence in the dac- tylic closes of the Augustans {Trans. XXXVI, p. 168), though not in those of Ennius, it should be remembered that the Augustan hexameter has its own ictus, its own T 162 « KoBERT S. Radford II. SOME SPECIAL CASES OP SYNIZESIS Notes on Latin Synizesis 163 It remains to note briefly a few special problems and special developments of Latin synizesis. The first of these problems relates to the rocative of the possessive meus. It is important to remember here that this case of the possessive almost invariably occupies in prose the proclitic position immediately before the substantive, e. g., mi fill, mi pater ^ mi fratres ( Trans. XXXVI, p. 197, n. 1), and it is clear that this position has influenced to some extent the case-form of the vocative plural. Thus, in the case of the nominative plural, mei frdtres is only one of several possible word-orders; consequently nominative plural yw(e)i shows perhaps only approximate syncope and is only quasi-monosyllabic. In the vocative plural, however, 7nei fratres is an almost invariable order, and here we find that m{e)i has been reduced to an absolute monosyllable in Old Latin, and may be fully elided before a fol- lowing short syllable, e. g., Ci. 678 m{i) hdmin^s, mi sp^datdrea (anap. sept. ) ; Mi, 1330 6 mt dculi. In the latter passage, our editions (e. g., edd. min. and mai.) usually accept o mei from very inferior MSS, but the form mi is clearly implied in the read- ing oh mihi of BCD, and should unquestionably be placed in the text;* for the legitimate hiatus, cf. Mi. 1330 6 mi dnime; As. 661 ml 6nime; Cas. 131 ml 01ympi6 (Skutsch Philol. LIX, p. 487; Maurenbrecher Hint, p. 162). Similarly, although many scholars question the contraction of Latin ie into i, the ancient derivation of voc. sing, mi from *mie, voc. of atonic 7nius, remains distinctly the most probable explanation of the form (cf. Lindsay Lat. Lang.y p. 422), and the contraction may possibly have artistic devices, and its own special conventions to facilitate the fitting of difBcult words into the frameworlc of the verse {sy^iizesis Oraecanica). Finally, the total suppression of vowel t and u suggested in Tratis. XXXVI, pp. 169, 204, cannot be con- sidered certain ; it is perhaps admissible to reject as corrupt the half-dozen passages cited in the latter passage, and to retain with Skutsch {Tipas, p. Ill) only St. 39 pol, m«o ammo omnis, since feet like mdlevolente, seoulmini seem to be also legitimate in Old Latin anapaests. MttUer, who scans dttinent in anapaests, is inconsistent in accenting malevdlente (PI. Pr.^ p. 416). »The PI. and Ter. MSS, as Is well known, constantly read mihi for mr, nihil for nil; cf ., for example, Ahlberg Procel. I, pp. 106 ff. Similarly, mihi is not infrequently written for voc. mt, as As. 689 mihi patrone, Mm. 1125 mihi germane, Mer. 947 mihi Bodalis {loc. cit., p. 107). been facilitated by the almost invariable proclitic position which it occupies.* I may mention also the fact that Latin has assimilated the present 'subjunctive' (optative) forms of esse, viz. stem, sies, siet, sient to the two plural forms simus, sltis. Thus the much more fre- quent and more numerous forms have followed the analogy of the less frequent and less numerous ones, and, in view of the fact that the fuller forms remained in use to so late a period, some further explanation of the final outcome here seems desirable (cf. Stolz Indogerm. Forsch. XVIII, p. 470). Zander's explanation {Vers. Ital.y p. cxx) that the l of sit is not due to analogy, but is a Latin con- traction of -ie-,, is scarcely admissible, since the Old Latin form can- not well have been siet, as he assumes, with iambic shortening, but was much more probably siet;^ even the hypothesis which is mentioned by Sommer [Lat, Lautl., p. 577, n. 1) and by Stolz {loc. cit.), viz. that contraction of -ie- to -I- may first have taken place in * enclitic' combinations like potisiet (shortened from potisiet) is not free from difficulties. On the other hand, it does not seem possible, even in the initial iambic sequence, that -ie- should contract directly into -i-, instead of into -e-; for the occasional occurrence on late inscriptions of spellings like dibits {OIL, VI 25540), Quita, inquitare, etc.^ (as well as of QuetuSy quescere, requescere), scarcely points to the production of a genuine i-sound in these cases. Hence I should suggest the fol- 1 Sommer Lat. Lautlehre, p. 446, also wishes to make use of the proclitic position of the vocative to explain the form, but the syncope of *m€^'e to *mct is improbable in the extreme, and is not greatly helped out by comparison with hypothetical ill(e), ind(e)^ etc. ; cf. Am. Jour. Phil. XXVII, pp. 418 flF. Of course the contraction seen in mi^fili^ Valeri and the like is due primarily to the trite and emotional use of these everyday forms ; compare what was said above upon the contractions di, dis (p. 160, n. 2) . *With respect to the orthography, however, the MSS of Cato give only the full form sies in the second person, much more usually siet in the third person singular and equally often sient in the third person plural, and they offer these full forms both in the middle and at the end of the sentence (Weise Quaest. Caton.^ Gottingen, 1886, pp. 46 f. ; Neue Formenlehre IIF, pp. 598 f.). The earlier inscriptions also show only siet and sient in both the positions named. Hence Zander (loc. cit.^ p. cxx) argues with much probability that in the middle of the verse or hemistich, where the Plautus MSS now show only the short forms sim, sis, sit, sint, this strict orthographical uniformity is due to the corrections of the later grammarians, and Plautus himself probably wrote indifferently sim or siem, sis or sies, etc. Our Plautus MSS (P) retain dimoric siet within the verse only in Au. 370 rapftcidarum ubi tftntum si^t in a^dibus. sSchuchardt Vok. II, pp. 444 ff. n 164 RoBEBT S. Radford Notes on Latin Synizesis 165 lowing explanation as possibly accounting for the influence of the plural forms: Weakly accented forms of the substantive verb like siem, siet are necessarily synizesis forms of an extreme type in Old Latin, and therefore very unstable in pronunciation. In other words, they were regularly pronounced within the sentence very nearly as *sem, *set, e.g. s{i)em liber; only at the close of the sentence was the dissyllabic pronunciation siem fully retained, as we may see from the usage of the dramatists (Brock Quaest. gramm., pp. 84 f.; Hauler Einl zu Phor,, p. 63, n. 2). If, then, before the beginning of the literary period, these forms sometimes became genuinely monosyllabic and were pronounced at times simply as *sem, *set, the introduction, through analogy, of -I- from the two plural forms could have easily occurred.' On the other hand, if the weakly accented forms *si&mii8, ^siStis were ever introduced in consequence of the analogy of the singular, they were quickly reduced to *8emus, *setis (cf. the reduction seen in {e)6sdem)y and then assimilated to the short forms. I have stated in the first section of this paper that the effects of the expiratory accent are perceived most clearly in the case of weakly accented words, and I wish to illustrate this principle still further from the later Augustan usage. The poets of the classical age accomplished veritable marvels in checking the use of popular synizesis and in cultivating and developing a more precise quanti- tative pronunciation. Thus they restored deos, sew, dud and even duellum, though this last form had definitely become dvellum or *dellum in Old Latin (Birt Rhein. Mus. LI, p. 73) ; they rescued also very largely meos and eos, although they were compelled by the force of the expiratory accent freely to admit slurring (pre- tonic syncope) in (e)6sdem and {e)(lsdem. It is noteworthy also that they were unable to banish the slurred pronunciation in the case of subordinate particles which were uttered rapidly and with little emphasis like proinde, dein, deinceps, deinde/ cf. quoad »0f. int (Corp, Gloss. II 75. 23) for ewn^, formed under similar conditions on the analogy of tmtis, itis (Stolz Mailer's Handb. 11», 2, p. 161). On the other hand, since audiunt is quite stable in pronunciation, we find no form *audint to show the influence of audimus and auditis, «Cf. dende CIL. VI 30112; cf. also See example in Johnston Metrical Licenses of Vergil, p. 16, n. 2. 2 Latin popular poetry, on the other hand, always retained synizesis in a larger circle of words ; see the examples in Hodgman Hare. Stud. IX, pp. 144, 152, 160, 162 f. ; 166. This later synizesis has much in common with that of O. Lat., but it is no longer restricted to iambic words and word-beginnings, and often resembles externally the Bo-called synizesis Chraecanica {Trans. XXXVI, pp. 167 f.). HM^ 166 Robert S. Radfobd Notes on Latin Synizesis 167 di-fue *bifidum,' etc. {^io\z Hisi, Gramm. I, p. 304; Buck Gram- mar of Oscan and Umbrian, § 102. 3), and compare the late and vulgar double forms des, zes (Schuchardt TI, p. 445; III, p. 295), and ies (e. g., Fabr. viii. 41, cited by Schuchardt I p. 69).' A second case in which Old Latin synizesis is perhaps to be recognized is that of the particle jam, which is most probably an ace. sing. fem. from the pronominal stem i- (Lindsay Lat Lang., p. 570; Walde Etym. Wdrterb,, p. 292; Br^al and Bailly Dictionn, 6tymol.\ p. 140), as quam and tarn are ace. sg. fem. from the pro- nominal stems quo- and to-; in its abverbial use, jam should be compared especially with Old Latin em, which is used both as an ace. sing. m. from the Ind.-Eur. pronominal stem *e/o (Walde loc. cit, s. 17. 'em') and as an adverb with the meaning of Hum' (Paul. Fest. 53. 37 Th.).^ If we assume the usual derivation of jam from the pronominal stem -/-, the question arises whether the consonantization of the i belongs to the Italic or the pre-Italic period. Some arguments may be adduced for the former view, which would evidently involve the operation of Old Latin synizesis. Although the ace. sing. fem. is usually written cam, the spelling tarn, which is at once phonetic and original, is found repeatedly in the MSS of Varro, and the spelling mm is found in a Luceria inscription (Neue IF, p. 381; Lindsay Lat, Lang,, p. 437). On this hypothesis it is natural that the original initial vowel should be consonantized in the simple adverb, but, in agreement with the laws of Old Latin synizesis {Trans. XXXVI, pp. 173 ff., 183), that it should be fully retained wherever it ceases to be initial. Thus, in not admitting synizesis, the compounds etiam, qmspmm, fispkim, and the Old Latin quasi-compound ?«7/ici«m or nunc iam^ jz uT^^l!" '""ItI^^ compare Aiutor for Adiutor, OIL. VI 3, 20752. Schuchardt Vok.U p. 68, and III, p. 24, cites also late plebeian occurrences of aiecit, aiuncfa and the like. j» An adverb im, the ace. of is (cf. inter-im) and glossed expressly by ^J5^, Xo.ir6., te still recognized by Lindsay Lat. Lang., p. 4:^8, and Walde loc. cit., s. v. ^ em,' but the Corp. Gloss. Lat. (II. 75, 36) now reads here i certdmst mihi. Any one who will turn to Langen's own discussion will find that he cites no less than six examples from Plautus of nunc-ixim with the present tense ; after explaining away five of these examples with considerable difficulty, he then bases his distinction upon the one remaining example, Cap. 266, while, according to his own admission, the proposed rule does not hold good for the usage of Terence {Eu. 561) ! Since results obtained in this arbitrary fashion are of little value, it may be worthwhile to state the simple facts of Old Latin usage. There are in all fifty-three cases of trisyllabic numi-iam in Plautus, twenty-eight of these being in verse-closes and twenty-five within the verse On the other hand, the Plautus MSS offer a few cases also of dissyllabic nunc-iam within the verse, where the two parts are not separated by any intervening word. Thus the dissyllable occurs at least twice with the impv. : Au, 451 fte sAne n6nc-iam intro <5mnes (where the ed. min. brackets iam) ; Amph. prol. 38 nunc-iam htic animum (5mnesqua6 loqufir advdrtit^ ; once with future: Poe. 374 n6nc-iam dehlnc erit verax tibi (less natural is the scansion of the ed. min. : ntinciAm dehinc ^rlt verftx) ; once with the present indie. : Cap. 266 ntinc-iam ctiltros fidtin6t. In short, the metrical treatment of nunc-iam like the metrical treatment of a-suo {Trans. XXXVI, p. 175, n.), is wholly independent of the precise meaning; thus we find trisyllabic nuncidm in the sense of 'now at last ' Ep. 135 iUlam amabam 61im: ntlnciam ftlia ctSra imp^ndet p^ctori (Langen : olim: nunc iam, but cf . Skutsch Forsch., p. 107). It is true, of course, that nunc-iam is usually trisyllabic, that it is used chiefly with the impv. and that it 18 commonly equivalent to 'a strengthened nunc, but no other part of Langen's account appears to be established. Lindsay {Bursian's Jahresbericht XXXIV [1906], p. 208 n.) apparently still accepts Langen's distinction and seeks to explain away its difficulties but his discussion shows that he is fully aware of its very dubious character. ' iBirt loc. cit., p. 79, states the rule that j regulariy maintains itself in historical Latin in the interior of words, wherever it occurs between two non-t-vowels. 168 Robert S. Radfobd perhaps possible, however, that the change of m to n in qnoniam may be otherwise explained. Maurenbrecher indeed [Hiatus im alien Latein, Leipzig, 1899, p. 39, n. 4, and p. 84) has boldly assumed that monosyllabic words in m had already developed in Old Latin and in Plautus a hiatus form in w, and he finds examples of this pronunciation in quouiam, conauditus, ciinere {=conhereSj CIL. VI 3282) and the like, as well as in Ital. con amove, Fr. rien, Span, qnien. This view scarcely seems supported by suffi- cient evidence, and still another explanation of the change may be suggested as a possible one. It is well known that the final m of monosyllabic words was regularly assimilated to the initial con- sonant of a following word, and thus freely appeared as n in the proclitic forms con, quon, can, qnen, tan and the like; thus we find con qua, tan dnrnm, cnn dies, cnn hijri, etc. (Schuchardt Vok. I, p. 117; Corssen I^ p. 26r>). An especially notable example of the assimilation of a conjunction quom is that given by Cicero, Fam. ix. 22. 2, i.e., qnom [cum) nos pronounced nearly as cunnos (cf. Birt Ehein. Mns. LI [1896], pp. 94 f!.). Hence in much the same way that an independent form con has been developed in proclitic use from the preposition com [cum), and is sometimes used instead of the latter even in hiatus,* we may perhaps con- jecture that a proclitic form in n has arisen also in the case of the conjunction quom, and that it is this latter which appears in the compound qnon-iamr the chief difficulty which stands in the way of the assumption of an original Old Latin form Ham would thus be removed. Elmira College »E. g. conire, Quint, i. 6. 17 and i. 5. 69: conivola, Paul. Fest. 43. 8 Th.; conin- quere, id. 45. 11 Th. ; although Thewrewk (Paul. Fest. 46. 7) now reads comauditum and comangustatum in place of conauditmn and conangustatum {id. 65. 8 Miill.). ''Further examples of the manner in which the conjunction quom and the preposi- tion com (quom) have influenced each other, are collected by Solmsen Stud. z. lat. LautgeschiehUt^l%