circles of Rome with verses which must have been read in a manner widely different from the pronunciation of common life: in this Latin final consonants were regularly dropped they were often actually omitted upon inscriptions, not merely ceased to be audibly pronounced as in the Romance languages, where they have been fixed by literature even when unheard in conversation. The consonants which most frequently fell away in this manner are the most common final letters s, m, and t. For the loss of these Schleicher gives the following examples. CH. VIII. final s. Final s is dropped upon inscriptions in nominative Loss of cases, like Tetio(s), Furio(s), Cornelio(s). In the older inscriptions, those of the Second Punic war, it is much less often written than dropped': though when the o in these nominatives was weakened to u, the s seems to have been regularly retained. By the beginning of the Empire, s even preceded by u was often lost on inscriptions: and a few centuries later, s had vanished from every case as well as from the nominative. In written Latin of the classical age, as we know, the s was generally kept. Still even here there are plenty of instances where its loss in writing shews how little it was commonly heard. Such are forms like amabare by the side of amabaris, and similar losses in other tenses: forms also like mage and pote for magis and potis. And it was regularly dropped in other forms, only a few traces surviving in Plautus; or in words, which from some old association retained their archaic form. Thus s was regularly dropped in the nominative plural of the o-declension. Yet we find hisce homines in Plautus, magistreis, publiceis, &c. on inscriptions. In the genitive of the a-declension we have familiae, yet sometimes the older familiā(i)s. And lastly, through previous loss of the vowel of the termination, we have pueros, puers, puer. For the omission of final m on inscriptions we need Loss of final m. 1 Corssen, 1. 286. 2 Trin. 877, and Brix's note. CH. VIII. Loss of final t. not go farther than the often-quoted epitaph of Scipio, the consul of A. U. C. 495. This begins, as given by Mommsen in the Corpus, Honc oino ploirume cosentiont R[omani] Duonoro optumo fuise uiro [uiroro-e conj. Ritschl.] Here the m is omitted five times, and written once: Final t, as Schleicher points out, seems to have had the sound of weak d. As such it was sometimes written in the ablative case: as Gnaiuod, sometimes dropped altogether in the same line, as patre(d). Haut is sometimes haud, sometimes hau3. The late Latin shews the t written in personal terminations, as uehit: but the old Latin often omitted it, as in dede for dedit, dedro for ded(e)ront; compare the classical dederunt and dedere. This loss was universal in Umbrian; as it was in the late Latin and the derived modern Italian. Indeed the loss of final consonants is felt much more in Umbrian than in Latin, but not in Oscan. Schleicher suggests reasonably enough that at the time from which our inscriptions date, a common form had established itself among the wide-spread Sabellian tribes, which became the literary dialect, and therefore ceased to vary further. 1 At least in monosyllables; see шI. 1082, and Munro's note on II. 404. 2 Epitaph of Scipio. 3 Cf. loss in Icelandic of the final t in the negative suffix -at, e.g. skalat and skala, Cleasby, p. xxvi. L CH. VIII. III. ASSIMILATION. 1. The Greek Aspirates. I have already, in the account of the Latin aspirates, given the reason why I believe the Greek aspirates to be the result of assimilation. The change of the original breath to the spiritus asper seems to me to explain the changes of these letters in both Greek and Latin, whilst I know no other that does. The original pronunciation of the soft letter, followed by a breath, possible to the original people, possible to the Hindu, and to his descendant', was impossible to the nations of the West, who therefore changed the breath to the more familiar and very slightly different spiritus asper. Even in Sanskrit this occasionally took place; e.g. in hita for dhita, the past participle of √dha, and the root han for √ghan; in these the breath has become the rough breathing, and expelled the d and g. In Latin we have seen that sometimes one member of the new compound was left, sometimes the other. The Greek followed its usual course. Instead of ejecting one of the sounds-a process, as we have seen, rare in Greek-it allowed the second to assimilate the first, and, therefore, instead of gh, dh and bh, the soft aspirates, we have regularly the hard Χ 0, p. That the original aspirates at least passed through this stage is allowed even by those who maintain that X, 0, were sounded in classical Greek not as hard aspirates, but as hard spirants (as they are in modern Greek) corresponding to German ch, English hard th, and the labial f respectively. This view is taken by Prof. Arendt3: the soft aspirates, according to him, became first the hard 1 Thus Prof. Arendt (Kuhn and Schleicher's Beiträge, 11. 289) declares that he has heard a Mohammedan, whose mother speech was Urdû, pronounce these sounds countless times without the slightest insertion of a vowel between the soft explosive sound and the h, and without the soft being changed into the corresponding hard. 2 As in thick, breath, &c. 3 K. and S. Beit. II. 424, &c. Pronunciation of the Greek aspirates. CH. VIII. Probably they were not sound rants. aspirates, and then the hard spirants; the immediate passage would be impossible. Curtius allows the change from the hard aspirates to the spirants, but does not believe that it took place until at least the first century of our era. As it is of some interest to know what was the pronunciation of these important sounds in the mouths of the great men of Greece, I will briefly examine the arguments on both sides. And Arendt argues that the difference of sound between the Greek aspirates (if real aspirates) and the Latin ed as spi- equivalents would be too great for languages so cognate: an argument which certainly does not convince me. when he adds that Onp passes into the by-form pnp (like Latin fera), it is quite true that the difference of sound between the spirants th and ƒ is less than that between the aspirates th and ph; but this does not prove that th could not pass into p'h, or that pnp and fera agree from anything more than accident. Arendt next examines cases where the aspirates occur in combination with other aspirates or consonants; and no doubt in these cases the difficulty of the genuine aspirate is most felt. Words like Zaπþó indeed are as easy on one hypothesis as on the other: σ is easier, as Arendt allows, if ◊ be an aspirate, but he calls in the English pronunciation to shew that can be a spirant in this combination, e.g. in Demosthenes'. But undoubtedly his strongest argument is furnished by the combinations x and p0. It is quite impossible to sound c'hth together fully. This Curtius himself grants'; but he says in reply, I think quite truly, that in no language do we find that in groups of consonants each particular consonant preserves its peculiar value completely under all circumstances. Some one must be partly, if not wholly, sacrificed: this is the very 1 1 He seems to think that "asthma" and "isthmus" are pronounced in England as astma and istmus: surely either the spirant is heard fully, or entirely dropped, as asma, ismus; and he is uncertain whether "sixth' is pronounced as siksth or sikth. 2 Gr. Et. 387. reason of the loss which we saw so frequent in conso- 1 CH. VIII. to prove they were really aspirate sounds. On the other hand, Curtius adduces some positive Arguments arguments to prove that x, 0, were real aspirates, which seem to me on the whole satisfactory. The first of these is the ease with which the h fell off, and left the explosive element alone, in reduplication, &c.; e.g. πÉ-QUка, ἐτέθην, ἐνθαῦτα, the Ionic variant for ἐνταῦθα, &c. Leo Meyer" well points out that the possibility of a reduplicated ƒ in Latin (fefelli, &c.) by the side of πéþvкa, &c. shews the difference of sound between f and ; in the Greek dissimilation was necessary to avoid cacophony. On the other hand, it must be allowed that two consecutive labial spirants would probably have been equally offensive to the Greeks. Curtius' second argument is the pronunciation of these sounds by foreigners, so far as we can judge from Aristophanes; e.g. öрviто таρadídwμ, in the Birds (1679), or the speech of the Scythian in the Thesmophoriazusae : πέρ ̓ ἐγὼ 'ξενίγκι πορμός, ἵνα πυλάξι σοι. These may not be conclusive, but at least they shew that and imperfectly pronounced were more like and than th and f. But the most convincing argument is certainly that drawn from Latin transliteration, at the time when they expressed the borrowed Greek words as well as they could with their own alphabet. If x had been a guttural spirant, surely the Latins would have denoted it by their h, which, as we have seen, had still a guttural 2 Verg. Gram. 1. 43. 1 Gr. Et. 384, &c. |