Page images
PDF
EPUB

2

5

5

2

5

but in the yévos Atléolov the Brevians is of necessity regularly restricted to an initial syllable, if the verses of this yévos are to be readable and its rhythm to be kept distinct from the rhythm of the yévos coov. The vocal organs experience, of course, precisely the same difficulty in pronouncing the particular sequences named above, whether the vowel of the first syllable is separated from the vowel of the following by a consonant or not, but the method of removing the difficulty is different in the two cases. The language must, of course, in dealing with each case, follow the line of least resistance, and, in the case of the half-vowels i (e) and u in hiatus, all ancient usage, both linguistic and metrical, proclaims that the least and the simplest modification consists in the slurring of these sounds. Hence the original sequences are modified as follows through the law of Brevis Coalescens, as it may be termed : I. u-_ is changed to - 2: Hec. 185 d(i)es ést; Tru. 324 d(i)u quám; Cur. 240 l(i)eni óptumúmst; Tri. 197 (e)am cúro; Poe. 104 eas pérdidít; Mi. 246 e*x (e)a míles ; Am. 1004 m(e)o me aequomst; Ci. 80 s(u)amrem obtinére; Cap. 130 s(u)i-gnáti; Ep. 702 t(u)i-gna-ti. The foregoing changes are brought about properly by the word-accent (which is here, as it happens, reinforced by the verse-accent), but the verse-accent alone may continue the work of prose, e.g. Au. 299 s(u)am-rém (in prose: s(u)ámrem); Mi. 736 quil d(e)orúm. II. Ú-úv is changed to Luui Cu. 671 m(e)am-rem ágere; Cap. 363 t(u)o véteri. III. (a) u 2 is changed in actual speech to -, both quantity and accent uniting to reduce the weak half-vowel, as in (e Jósdem, d(e)órum, by a species of pretonic syncope. (6) Ú-, or rather uś, is changed to %, that is, in cases like deos, involving the weakly uttered half-vowels, quantity prevails over an unstable accent, and produces d(c)ós (cf. Trans. Am. Phil. Assoc. XXXV, 58 n.), while in other and far

1 For the relation of vowel and consonant i and u in Skt., cf. Whitney, Skt. Gramm. 55 ff., 129; Corssen, II”, 768.

2 There is a quite unnecessary outcry on the part of many against ósynizesis in the arsis,' and Skutsch (Sat. Viadr. 144) and Ahlberg (De procel. I, 93 ff.)

2

1

1

3

2

3

more numerous cases like meum-gnátum, the iambic dissyllable loses the accent entirely through proclisis (see below, p. 193): Ba. 124 d(e)órum ; Poe. 1422 (e)a+mus; Cur. 220 l(i)éne; Ep. 338 qu(i)éto; Am. 549 d(i )eos ut; Cap. 881 (me)um qui gnátum; Am. 827 t(u)am-rem cúret; And. 385 ex (e)a-re; Men. 81, Poe. 1338, Ad. 959 m(e )á-quidem senténtia ; Cap. 987 véndidísti m(e)o? patrí; Ep. 582 f. quaé patrém | T(u)o'm vocás; As. 878 t(u)om virúm conspexeris ; Tri. 192 cúres t(u)ám fidém; Cur. 702 dícam m(e)am senténtiám ; Cap. 867 i(u)o a'rbitrátu.

In examples like the following the verse-accent continues the effects of the wordaccents: Mo. 1120 t(u)i-gnatí sodálem (in prose: t(u)ignáti); Ps. 120 t(u)óm tangám patrém; Am. 126 m(e) possém patrí; Mi. 723 d(i)u6 vitám.

The law of Brevis Coalescens may be stated then as follows for the γένος διπλάσιον and the γένος ημιόλιον: An iambic sequence of syllables, if initial, i.e. if forming a single word or a word-beginning, has the value of a single long, in case

3

have concluded too hastily that ‘Synizese in der Hebung ist ein Unding. Une doubtedly, considerable importance should be attached to Latin word-accent, but it should not be converted into a fetish, nor should the dogma be promul. gated that the word-accent in its theoretical form must prevail under all phonetic conditions (see below, p. 193 ff.); compare the treatment of the accented or “arsis' syllable in the contracted forms, di, dis, is, nil, dest, sis (for suis), and in the synizesis forms dein, dehinc, proin, pro.it, quoīd, also in lovem, bis for 0. Lat. Diovim, duis (Paul. Fest. 46, 6 Th.). The argument against synizesis in the arsis is stated by Gleditsch (Metr.: 258) as follows: "In zweisilbigen Wörtern wie eos, tuos und in dreisilbigen wie eorum, wenn sie den Iktus auf der ersten Silbe tragen, wie Trin. 238 éus cupít, eos cónsectátur, erscheint es viel natürlicher an Kürzung der zweiten Silbe in eos zu denken und die Wörter pyrrhichisch zu messen als Synizese anzunehmen. Vgl. ebd. 295 měõ modo et.” U'nfurtunately this argument involves a manifest petitio principii, when it boldly assumes that there is a full and complete “first syllable' in eos, tuos, etc., while, in fact, it is evident that if we have the prose accents — particularly in case of the proclitic pronouns — (e)ós-cupit (or even (e)os-cupit), (e)os-cépit, the verse-accents (e)sscupit, (e)&s-cepit follow naturally; for a fuller statement of my views, see below, p. 185. Especially unfortunate is the example měð modo (Tri. 295), in view of quomodo, cómodo, eiusmodi, quodammodo, ill&modo, etc. Similarly, Ahlberg's chapter on Synizesis is a valuable one, but his theory of synizesis in the thesis, eig. m(e)os sine, but iambic shortening in the arsis, e g. sine měõs, breaks down

[ocr errors]

the former of the two syllables contains the half-vowel u or i, or the similarly pronounced e, in hiatus. This law enables us very largely to see why 'monosyllabic' sc(i)), in(e)o do not occur in verse-closes, except at the end of the anap. oct. and anap. dim. ac., which conclude with the ictus (- -). These contracted forms could not in any case occur in verses which end with an iambus (iamb. trim., troch. sept.), where metrical convention has given the half-vowel the full value of a mora, but they might be expected on first thought to occur in verses which end in a trochee (iamb. sept., troch. oct.), c.g. placét m(e)o like placet mi, Mo. 175. Yet even here we see that the 0. Lat. poets never treated meo as a genuine monosyllable and never placed it at the end of a metrical sentence as such ; in short, the quasi-monosyllabic pronunciation requires an accent to follow the slurred word in an immediate sequence either of the sentence or of the verse (as placét m(e)o ), and not to follow it, for example, at the beginning of another line. Cf. the absence of fac and nil from early verse-closes, and Skutsch's former view, Forsch. 57.4 For further discussion of this question, see below, p. 208.

7

7

3

1

completely, as he himself sees (1.1., 104), when it is applied to 0. Lat. n(ih)il and n(ih)iln[e]. This theory also breaks down completely when applied to the verse of other Latin poets who do not admit shortening, eg. Enn. A. 151 M. (Lucr. iii, 1023) lúmina sís oculis; Auson. Sap. prol. 29 pronúntiáre s(u) as solént sententias; Terentianus Maur, 1609 vim s(u)ám tuetur. Cf. also Seyffert's criticism of this theory, Berl. philol. Wochenschr., 1900, 1611 ff.

1 In anapaestic cantica (the yévos toov), it is perhaps possible, as was formerly assumed by Ritschl and Spengel, that synizesis, like iambic shortening, is applied to other than initial syllables, e.g. to aur(e)o, fil(i)o, but the discussion of this question does not fall ivithin the limits of the present paper, and is of less impor. tance, since the anapaestic rhythm does not accurately represent the spoken language.

2 But in these repeatedly, e.g. Ba. 1153 facito út faciás. || Taceas. tu tuom facito : égo quod dixi haud mútabó, a verse in which Müller (Nachträge, 55) seeks in vain to find a flaw. There is some reason, however, to doubt the admissibility of synizesis in the close of the anap. oct.; see below, p. 207, n. 1.

3 About 1520 verses in all.

4“So erklärt sich ... das Fehlen einsilbiger Formen dieser Worte am Versende. Die lautlichen Bedingungen für die Synizese sind eben nur im Satzzu.

Points of Dissimilarity. Word-Forms and Complexes. — We have seen the basis of similarity which exists between synizesis and iambic shortening; there are also fundamental differences between the two. The most important of these differences are the following: (1) Since synizesis is a much easier and much simpler modification than iambic shortening, slurred forms are proportionally in much more frequent use than shortened forms. (2) Slurred forms ending in the diphthong ae and in s preceded by a naturally long vowel are very freely used, while shortened forms are rare in the case of these endings. (3) Slurring is most freely applied to a large class of words and word-complexes, which, on account of their accentual relations, only rarely and exceptionally admit shortening. The third point of difference is the one which I propose chiefly to discuss, and its extent may best be shown by an enumeration of the different classes of words to which synizesis is freely applied :

(1) Iambic words, e.g. meae, ci, quoii, duos, trium, quia, sammenhange gegeben.” — I cannot say whether the dramatists always retain the full form at the end of a complete sentence, i.e. when there is a change of speaker, but my examples clearly show that this is their regular usage : Au. 154 in rem hóc tùámst. ; Mer. 485; Tri. 197, etc.

1 According to the quite inaccurate statistics of Leppermann (1.6., 78), in forms of declension, the synizesis forms are at least three times as frequent within the verse as the non-synizesis forms (a very decided underestimate ; cf. below, p. 183); on the other hand, the forms without shortening are nine times more frequent in Pl. and nearly six times more frequent in Ter. than the shortened forms. Skutsch (Sat. Viadr. 135) has attempted to explain away the significance of these facts by the special plea that two vowels in immediate contact are particularly conducive to iambic shortening. We have, however, in Latin the principle " vocalis ante vocalem corripitur," not the principle vocalis post vocalem corripitur"; otherwise we might expect to find in Pl. abéă mus, abiúndum, abière, abīžsses, abstinuissem, accipiêmur, accubúisti, acquičscam, advenièntem, dhěnus, aličnus, etc.

2 According to Bömer (1.1., 41) synizesis forms in ne and in s are one and a half times as frequent within the verse as non-synizesis forms, while, on the other hand, forms in ae and s without shortening are seven times more frequent than shortened forms ; cf. also Lindsay, Journal of Philology, XXI, 206.

3 1.c. when the ultima is elided before a long vowel : Mer. 543 qu(i)a úxor rúrist, cf. Mi. 1278 qu(i)a aedis, and Leo's note ad loc.; so also in later poetry: Terentianus Maur. 1090 quia ét variis pedibus loquimur sermone soluto; Venant. Miscell. ii, 15, 8 filius ut dicant quiast creătura dei, cf. L. Müller, R.M:, 323.

6

1

(Ps. 703), deüm, deām, diēs,a liên (always),3 viā, Creo, fuī, scio,4 eo, queo, beo, diū, eo (adv.), etc. (Words of the same

(

a

1

2

1

1 See Abraham, Stud. Plaut. 204 f. The nom. sing. in Pl. is of course always děŭ' (Abraham is clearly mistaken in assuming deus, see below, p. 201); the short vowel is also always preserved in hiatus (dřům, Cap. 865) or in elision before a short vowel, and may be preserved in elision before a long vowel, e.g. the phrase pro děum dtque hominum fidem occurs three times in trochaic and once in iambic verse (Cu. 694; Ep. 580; And. 246; Hau. 61), while pro d(e)um dtque hominum f. occurs once in iambic (Hec. 198). An absolute metrical proof of the contract-forms is afforded by Phor. 764 sed pér-deos átque hominés, where pérdēðs átque is no more possible than filiăs atque would be (cf. A. J.P., XXV, 262, 4173; similarly, Cap. 727, Mi. 541, Tri. šao per-deós atque hómines ; cf.

; Cas. 336 ad-deós minoris. Pl. retains the full forms dčos, deūm, de[um], deorum, deo, exclusive of two cases in the verse-close and of two in cretic and bacchic verse, only six times against seventy-three cases of the contract forms. This extreme weakening of the short vowel, which is even more pronounced than that of the proclitics meo, tuo, is evidently due in part to the trite use of the word in oaths, adjurations, and emotional language in general (cf. the reduction seen in the English Odil's death, 's death, zounds, etc.). Upon the difficult problem of di and dis, the sole O. Lat. forms of the nom. and abl. pl., I shall not venture to express an opinion ; Sturtevant is correct (1.1., 20) in viewing these forms as absolute monosyllables, and rejecting this value for deüm, deo, etc., but quite contrary to his assumption (p. 33), all these dissyllabic forms of deus were verging towards monosyllabic pronunciation in the time of Pl.; for the late and vulgar forms, do, dac, cf. p. 201.

2 According to Gellius (x, 24, 1) the compound diequinti was pronounced in the Republican age 'secunda syllaba correpta,' but Lindsay (Journal of Philology, XXI, 205) justly suspects that synizesis of the first two syllables was the real influence at work here'; cf. d(i)e septimí Men. 1156, but hóc-diè crástini, Mo. 881. The present statement of Gellius is quite similar to that other tradition mentioned by the grammarian Priscian (Keil, III, 511, 20), which our Latin grammars accept so confidently, viz. that 0. Lat. genitives of the second decl., like tăgări (the standard example), are to be accented on the penult. But, as a matter of fact, Priscian probably had no trustworthy information upon this subject, since the form, though sometimes retained in poetry (L. Müller, R.M.2 442), had passed out of actual use centuries before his time, and he had consequently never heard the pronunciation in question ; it is noteworthy that Pl. never uses a proceleusmaticus like sed ingéni, sed aŭxili, and if a vestige of true tradition had reached Priscian, instead of Nigidius Figulus's differentiae causa example (Valéri and Váleri), he would have known that the republican accentuation was both tågūrium and týgări.

3 According to the grammarians (eg. Priscian, Keil, I, 149, 7), Pl. used, after the analogy of liën, a singular riżn instead of *rēn. Whether he used this form also in the pl., and rënés, Cu. 236, represents r(i)ēnés, cannot easily be determined.

* In Pl. apparently only sciò and sc(i)ā; similarly the grammarians noticed that Vergil does not shorten the final o of verbs, and hence, according to Chari

« PreviousContinue »