meaning of the verb might be overpowered by the preposition, yet its form remained intact. Very different was it with the Latin. Here, as is obvious on the slightest glance, weakening is the rule; when the original form is retained, it is the exception. The primary vowel a, which we have seen elsewhere so rarely affected, is here the greatest sufferer of all, as indeed follows from its occurring in roots more frequently than any other vowel. When followed by a labial or l1, it sometimes does not sink below u: thus we find oc-cup-o (cap), contubernium (√tab), in-sul-to (√sal), and in-sulsus from salsus. But even the labial is commonly unable to stem its downward course: the older forms de-rupio and sur-rupio gave way to deripio and surripio: and hosts of others, such as prohibeo, mancipium, dissilio, &c. will occur to every one. Before final consonants other than labials, the radical vowel sinks as a rule to i: before gutturals, as re-ticeo (√tac), prodigium (ag); before linguals, as profiteor (√fat), Jupiter, or Dies-piter; before nasals, as recino (√can) and inimicus. Further, if the vowel be followed by two consonants as well as preceded by one at least, the vowel regularly sinks to e, as in the examples given above'. In all these cases sufficient strength is supplied to the syllable by the combination of consonants: and therefore the original vowel is sufferea to sink to the dull e, unless retained at some intermediate stage by especial affinities. Thus we have peregrinus from ager, obsecro from sacer; ascendo and aspergo from scando and spargo; but infringo and contingo from frango and tango, because the nasal here is not radical; it is only employed in the formation of the present base from frag and tags. Damno in composition becomes condemno, from annus we have biennium and sollennis; castus passes into incestus, tracto into detrecto; from /cap we get auceps, &c. in the nominative. In some isolated cases, as imbecillus, we find e without 2 See pages 262, 264. 1 Comp. p. 258. CH. VII. CH. VII. the excuse of the two consonants: others, like aequi-perare and per-petior, may be accounted for by the affinities to be hereafter mentioned. E itself is weakened to i in numerous compounds, where it had taken the place of original a in the root: thus lego is colligo in composition; from tenax we get pertinax: but e is often retained, on no very clear grounds, except that these two vowels, as the weakest in the language, exchange place more easily. Corssen can give but one certain instance of o being shortened in composition, illico from in loco: he gives also per-nic-ies, compared with noc-ere: but the root is nec. U is never shortened. There are a few seemingly irregular instances of weakening of long vowels in composition, but always to e or i. Thus halo passes into anhelo; √ag, which is lengthened to ag in ambages, imago, farrago, &c., lets the a pass into i in many compounds, as caligo, vertigo, robigo; long o is weakened to short i in cognitus and agnitus1, and long u to short e in de-iero, pe-iero, but retained in per-iurus. Lastly ē passes into i in two cases -subtilis from tela, and delinire from lenis. It has been already mentioned that these weakenings, although very common, are by no means without exceptions. The prevailing tendency never became universal; and this in most cases is to be accounted for by the sense of the composite nature of the words being retained. Sometimes we can see a reason for this, sometimes not. Thus prohibere acquired the general idea of preventing, losing the primary sense of holding a material obstacle "in front:" hence comes the weakening in form. But, either because of the stronger form of the preposition, or for some other reason, the primary sense of post-habere was felt even when used in cases where no putting behind in space was possible: and hence the retention of the original vowel. Again Diupater passed into Jupiter without preserving a trace in common use of its derivation: 1 Unless we assume (with Corssen, 11. 422) a participle gnõtus (compare nota). In other cases I because of many irregu but the title Janus-pater was felt to be a compound from CH. VII. the use of its first member as a distinct word; and therefore the a never sank to i. And the idea of causation, which is obviously represented by compounds with facere, as tepefacere, &c., prevented the sinking of the vowel, which takes place in conficere, &c. lieve that assimilation has been the larities. This explains why the a maintained its place in per-placet, but not in dis-plicet; in per-facilis, but not in dif-ficilis. Another reason which Corssen has pointed out', by which the change has been prevented, is the necessity for distinguishing between distinct compounds. Thus expando was not allowed to sink into expendo, because of the necessity for keeping it distinct from the combination of ex and pendo: so also it was necessary to distinguish contactum from contectum. But this principle explains a very small number of instances. Lastly, we come to weakening in Reduplicated forms. This process (as Corssen observes) is closely connected with that just described: for Reduplication is really a sort of Composition. In the weakening of the syllable produced by Reduplication, the Greek and the Latin are on the whole in accord. Thus in the formation of present bases the vowel found in the new syllable is regularly i: we have gi-gn-o in Latin, as well as yí-yv-oμaι in the Greek; si-st-o as well as -σT-nμ: sometimes, however, e is found: examples have been given in the section of Chapter VI. especially devoted to the uses of reduplication. So also in the formation of the perfect the vowel regularly used in each language is e: Té-Tʊp-α and πé-πоɩ0-a stand by ce-cid-i and te-tul-i: though it must be allowed that there is much less uniformity in the Latin than in the Greek here; the Latin employs the radical vowel in the new syllable not infrequently; not indeed the heavy vowel a, but o, as in po-posc-i and mo-mord-i, u in pu-pug-i and cu-curri, and the weak i is kept in perhaps the only 1 II. 407. (iii) Weakening in Reduplicated forms: (a) of the reduplicated syl lable, CH. VII. (b) of the radical syllable. two verbs with radical i which have retained the reduplicated syllable, di-dic-i and sci-scid-i. But the strong tendency towards a uniform use of e is shewn by the other forms which were not uncommon in classical Latin—pepugi, peposci, &c.; but the original forins, pupugi, poposci, were again preferred, doubtless through the assimilating influence of the radical vowel. The e was regularly used for radical a, as in dedi (√da), steti, &c. We may infer then that the tendency to regard these new syllables as mere grammatical forms was strong even in Graeco-Italian days: and that while the Greeks after the separation attained to strict uniformity in this matter, the Italians, advancing no further, formed their tenses now on one principle, now on another: from which inconsistency we find in our grammars the anomalies of the "irregular verbs." It is difficult to trace with certainty in Latin the process by which the reduplication was often altogether dropped. Corssen thinks that it began with the compound perfects that in these by the "Old Latin law of accentuation" the accent fell on the first syllable, e.g. ré-tetulit, and thus forced out the e of the reduplication-syllable ; whence ré-t-tulit. (Compare the French "je ne le sais pas," where the e of the ne is lost in pronunciation.) Thus the ear grew accustomed to such possible forms as tulit, and when the "new law of accentuation" came in, and the accent was thrown forward in such words as tetulísti, the e again slipped out and left t-tulísti, tulísti; and in analogy with these accomplished facts the possible tuli also became actual1. The Latin treads its own peculiar path of degradation when it weakens the radical syllable as well; when it allows e.g. pe-pag-i (√pag) to sink into pe-pig-i. The same fate has befallen numerous verbs with radical a— cado, tango, &c. In other cases, chiefly when two consonants follow, e is found instead of i, in fefelli and peperci, 1 I shall have occasion in a subsequent section to explain Corssen's views respecting accentuation in Latin. and in other cases by reason of some affinity, as to the r in peperi (from √par). Sometimes the radical vowel seems to have been lost altogether, as in feci, i. e. fe-faci, then fe-f-c-i; that the verb was really reduplicated in the first instance is shewn by the Oscan fefăcust (i.e. fecerit) and fefacid (i.e. fecit). This weakening of the second syllable would be analogous to that of the second member in a compound (pe-pigi, im-pingo), and produced by the same reason, the resting of the accent at one period upon the first syllable. Corssen however believes that the ā in the Oscan forms was the result of vowel intensification: and that the è in Latin is a parallel lengthening; just as ago became egi. The reduplicated syllable must then have been lost altogether from the combined influence of quantity and accent falling on the radical syllable. This explanation undoubtedly accounts for the long vowel in Oscan, which is otherwise left unexplained. CH. VII. II. ASSIMILATION. change modified by external causes; So far we have seen the results of simple substitution Vowelupon the vowel system of Greek and Latin. A stronger vowel has passed into a weaker one in accordance with a regular scale of vowel-strength, differing indeed for the two languages, but constant in each. Before we look at the cases where this substitution has reached its natural limit-loss-let us see what modifying causes may have sometimes stemmed, sometimes altered, this downward progress what influence other sounds, vowel or consonant, may have exerted in particular cases upon a vowel, which when no such influences were at work simply sank lower in the scale. all in As I have before said, we shall find no such modifi- hardly at cation of the Greek vowels. They were too strong to become the mere reflex of a neighbouring consonant. In Greek; 1 Corssen, II. 427. |