however no further change in man-sum. But this explanation will hardly suffice for the few cases where s is found, though the root ends with a momentary sound. Such are lap-sus, lixus, fixus, fluxus, and a few more. For these I think we must adopt Corssen's explanation1: that the change began with those roots which ended in dentals: and that the new suffixes -sus and -sor came by degrees to be introduced through analogy into places where they were not helped by any assimilation. The tendency to soften t to s was an early one, as we see by the antiquity of the change from the older forms, pul-tare, mertare, mantare, &c., to pulsare, mersare, mansum, &c.2 The assibilation of c and t in ci and ti, when followed by another vowel, is commonly assumed to have taken place in old Latin, as it undoubtedly did in the late Latin and the Italian. One part of the evidence for this change is the varying spelling even in good MSS. of words like suspicio and suspitio. The sound of the two must have been very similar before such a change could take place; probably much the same as in our "suspicion." The interchange would therefore be precisely analogous to that between σ and TT, which we saw took place tolerably early in Greek. Corssen, however, who has gone most thoroughly into the question, proves that there is no such wavering of spelling in the inscriptions-our best guide-till a much later period than is commonly supposed. Thus, he says that there is no variety on the most trustworthy inscriptions down to the latest times of the Empire in the following words: contio (i.e. co(n)uentio) nuntius (probably nouentius formed, as Corssen suggests, from nouere, a nominal verb from nouos, on the analogy of Florus, florere, Florentia), setius (of uncertain origin, but with a by-form sectius, which may point, as Corssen thinks, to seg-nis), otium (uncertain), indutiae (for indu-itiae), fetialis (uncertain), dicio (i.e. 1 Krit. Beitr. 426. 2 See Quint. 1. 4. 14. 3 I. 49-67. CH. VIII. Assililation of c and t t. CH. VIII. Assibilation of c, dic-yo), condicio (which has nothing to do with deditio and other nouns formed from √dha, to place), solacium (uncertain), patricius, tribunicius (where the c is part of the suffix) both forms occur in proper names, like Lar tius or Larcius, where a double derivation is quite possible so that the complete confusion of the two spellings did not take place till the seventh century after Christ, though isolated instances doubtless occur much earlier. The best MSS. read suspicio and suspitio, conuicium and late except conuitium; the first in each case is probably right etyin the pro-mologically. There is sufficient evidence of the assibilation of ci and ti among the provincials, which gave rise to the confusion. Ci was sounded as çi and even si in Umbrian, e.g. faciat was façia, at least as early as the third century B.C.; the Umbrian had even a special symbol for this palatal sound: and fasia is Volscian. But for Latin there seems to be no evidence of the change of ci, any more than of the interchange of ci and ti, till the seventh century: while against the change there is the negative evidence of transliteration, e.g. Ovvxía (in the sixth century after Christ), and the Gothic faskja and laiktio for fascia and lectio. vincial dialects. Mr. Roby1 has brought a good deal of additional evidence to bear on the point. Thus, he points out that the same word often varies its form so as to have different letters after c, or to have c final: e.g. hice and hic, dice and dic; was the c first palatal in the older and fuller forms, and then made guttural again in the ordinary form? So also in decem and decumus, cano and cecini, is it likely that the sound changed? or that it was different in lacubus and lacibus, forms which at one period were doubtless used indifferently? Again, Quintilian never hints that c had more than one sound, though he speaks in several places of the superabundance or the deficiency of the symbols of the Latin alphabet. Then 1 Preface, pp. xliii.-1. See also Prof. Munro in the Academy of March 5, 1871. K K with respect to transliteration, Mr Roby thinks if the c The change of ti to si seems to have been earlier and more general: but Corssen regards it as belonging especially to the vulgar Latin (and the other Italian dialects), and not established in the speech of educated Rome till the fourth or fifth century after Christ. It is traceable however in isolated cases much earlier. Such are e.g. uiciens, which has come regularly through uicesiens and uicensiens from uicentiens: similarly amasius and others with the termination -asio are most probably from old -antio-: Acherunsius is certainly from Acherunt-io-s, Hortensius was in old Latin Hortentius2: and numerous names of towns in -esio 1 Roby, p. xliv. 2 Krit. Beitr. 467, &c. CH. VIII. CH. VIII. throughout Italy, as Valesium, Falesii (Latin Falerii), compared with others in -ento, as Laurentum, Valentium; and in -usio, as Canusium, Brundusium compared with Acheruntium, coincide with the other evidence for this change in all the Italian dialects, but seem to indicate that it occurred very slightly in Latin. I infer therefore that in classical Latin ti was sounded hard except in cases where another form in si actually occurs beside the latter. (i) Change of a dental. (ii) Loss of aspiration in Greek: IV. DISSIMILATION. This principle has a more limited application to the consonants than even to the vowels, and for the same reason: there are not many cases in which the occurrence of the same sound twice is unpleasant to the ear. Still, few as they are, they are tolerably certain. (i) One case where Dissimilation acts is common to Greek and Latin: namely, when a dental comes into contact with another dental at the beginning of a suffix. In this case the final dental of the root passes into s. In Greek the following examples may be given: ȧVUT-TOS becomes ἀνυστός: ἀδ-τεον becomes ἀστέον: πιθ-τος becomes TLσTÓS. Similarly in Latin, equit-ter (o) passes into equester: edti becomes est: claud-trum is claustrum. (ii) In Greek, when two aspirates occur too closely, one is softened : ἐ-θυ-θην becomes ἐτύθην; θι-θημι passes into Tienu. So also the suffix -0ɩ of the imperative (IndoEuropean dhi), which is found e.g. in Xô, is changed to when another aspirate precedes, as owent. Similarly if two aspirates occur in the root, one is dropped in conjugation; for example the two forms Túp-w and búπ-σw are referred to a root up. The existence of these doubly aspirated roots has been maintained by Grassmann in his article already often referred to in the twelfth volume of the Zeitschrift1. But where there is no other proof of 1 See also Gr. Et. 51. the existence of the two than the double forms in Greek, it is much better in my opinion to assume only one for the root and then to account for the second (which never occurs in the same word as the first) by the principle of compensation. CH. VIII. To Dissimilation is also due the loss of the consonant and of initial rein the reduplicated syllable of many verbs which begin duplicated with two consonants as ἔκτονα for κε-κτον-α, ἔγνωκα for consonant. ye-уvw-ka. It may be assumed also that the passage of a consonant into the rough breathing in the presents ἵ-στη-μι, ἵ-η-μι is due partly to the desire for a dissimilar sound in following syllables. In eyeipo and some other words it is more likely that the e is prosthetic (as will be pointed out in the next chapter) than that the word was originally 'ye-yeípw. Perhaps too the first consonant may have sometimes fallen away even in simple nouns for the same reason: as in okvos for kok-vos, compared with Latin cunc-tor and Sanskrit çank. But this must rest uncertain1. -aris and -alis. (iii) The only regular application of this principle in (iii) Latin Latin—which is not equally sensitive with the Greek in this respect is the curious change in the termination -aris or alis, accordingly as l is found or r in the preceding syllable. Thus we have uolg-aris, popul-aris, &c.: but mort-alis, later-alis. Similarly the form Pari-lia sprang up beside the more difficult Pali-lia'. There are a few isolated cases of dissimilation in each language, which can be reduced to no rule. Such are φι-τύω (φυ) where the change to seems to be due to the following syllable; v is found in the other derivatives: ảλλŋλ, as Curtius suggests, is another instance of conscious change. So also in Latin ferbui seems to owe its b to the difficulty of sounding the double u: tenebrae has been already mentioned as a possible instance3. 1 See Gr. Et. 660. 2 Corssen, 1. 223; Comp. 267. 3 See p. 100. |