CH. VIII. (iii) bable because it had F(v) after it, as shewn in yứŋ, i.e. y a-a, and aia for Faîa or yaîa1. (iii) Nasals are affected in their turn by the following Changes of consonant: we have συγ-καλέω, and ἀγγέλλω (ἀνά); ἔμπειpos and eμßaivo; and numerous others of the same sort. nasals. (iv) Change of τ to σ. (v) Changes of v. (vi) Other exceptional forms. T (iv) In Ionic, Attic, and Lesbian, 7 passes into a before ; as pnoi for pari. This might rather seem a case of simple substitution: but I believe that it first occurred in cases where another vowel followed, as πλovσios for πλουτ-ψο-ς, πλησίος for πλατ-yo-s: when the change would be due to the assibilating influence of the y: then the softer sound was preferred universally; this σ for 7 is also found before v in oú, but in Doric TV is kept: the old form was tva, where again the semivowel could assibilate; and the same is probably true of the suffix -ov'n for -Tuvn, Latin -tuna; for there is a Vedic form -tvana'. The Boeotian, like the Doric, preserves T where the Lesbian and Attic have softened it into σ. I have already mentioned the peculiar Laconian weakening of 0 to σ, which may have begun in the same way as the last change, according to my suggestion. (v) The spirant v is altered by assimilation in certain dialects. Thus Fρόδον becomes βρόδον in Sappho, Εράκος is βράκος; we find βράδινος, βρίζα, βρήτωρ, &c. This change is less surprising, for we have seen that F passed into ẞ in Laconian even without any neighbouring sound to influence it". In the word ope we seem to have a hardening of original v to 4. The old form is sva, which in Greek generally became believe that (through Fe). In this case it is hard to was much more than a spirant. The same change is seen in opo, the dual from tva (whence sva and ou): compare the Latin uo-s, where the t has fallen off. (vi) Lastly come some very peculiar forms which 1 See page 110. 2 Comp. 459. 4 Theok. XXVIII. 11, see Ahrens, 1. 34. 3 Frag. 69, 2. 5 See p. 327. seem to be more probably due to assimilation than any other cause. These are e.g. πτόλις by the side of πόλις, πτόλεμος, &c. : κτείνω by καίνω, &c. It seems impossible to separate Tóns from Sanskrit pura (also a city) and Latin ple-bs, perhaps also po-pul-us (a reduplicated form); and therefore it must be from the root PAR, to fill, which in Greek appears generally as πλa or πλe, in Latin as ple. This evidence excludes any possibility of 7 having originally belonged to the root and then fallen out. It is clearly a Greek insertion. The only explanation of this curious change which I know, does not seem quite satisfactory. It is given by Professor Kuhn1 and adopted by Curtius: that y through indistinct articulation sprang up after π, and was assimilated by the π to T. We have seen above that πу never became oo, as the other hards did; but it is not easy to see why, if the sound πy were difficult, it should not have passed into πe or Tɩ, instead of the very difficult T. Possibly however a new parasitic 8 may have sprung up before the y-the possibility of this will appear in the next chapter-and been afterwards assimilated by the preceding hard letter. This explanation is supported by the form xés. Here again the dental seems to belong to the Greek only: the Sanskrit form is hyas for ghyas, Lat. heri: and here the Sanskrit gives the necessary y: xwv, according to Curtius, is another case in point: the older form is preserved in xaμa-í, with which compare xoaμaλós; and the Latin humus agrees. Here however a different parasitic sound in Sanskrit has produced kshamâ in that language. Perhaps this explanation of these intrusive letters may stand till a better can be suggested. They are certainly not " euphonic" or "strengthened forms:" why did Tóλis require to be strengthened? Still less are they "metrical licenses:" why should a Greek poet have the liberty of arbitrarily inserting an entirely new letter in order to make a word suit his verse any more than an English writer? 1 Zeitsch. XI. 310, see Gr. Et. 453. 2 Gr. Et. 454. CH. VIII. CH. VIII. I. Com 3. General rules of Assimilation in Latin. In Latin also we have complete and incomplete assiplete Assimilation. Complete assimilation may be divided into the same heads as those which we considered in the Greek. Where the first letter assimilated itself to the milation. (i) Assimilation of the first letter. (ii) Assimilation of second letter. second. Thus sup-mus becomes summus, sed-la is sella; dis assimilated very frequently, as in lapillus for lapid(u)lus, esse for ed-se: t passes through s in pet-na, pesna, penna; compare ces-na, cena: g has been assimilated in flamma (flag-ma): very likely, as Schleicher suggests, serra is for sec-ra1. In all these cases the radical vowel was short: therefore the final consonant was not absolutely forced out, but assimilated: and was written after the time when it became customary to write double consonants, not merely to make a little mark above one of them (the 'Sicelicus'). When the vowel was long the consonant was entirely lost, e.g. in suā(d)-uis, and other cases already mentioned among examples of Loss. Sometimes the consonant was lost even when the vowel was short; it was, however, afterwards lengthened by compensation, as in scala (scad-la), squama (skad-ma), perhaps finis (fid-nis), filum (fig-lum), &c.2 (ii) Where the second letter assimilates itself to the first. This change happens to t in superlatives after s or r: as durissumus for dur-i(o)s-tumus, celerrimus perhaps for celer-is-tumus, celerstumus, celersimus; and numerous others. After r, s naturally passes into r, as ferrem for fer-sem, torreo for torseo: and analogously after l 1 Comp. 258, whence these examples are taken. 3 Comp. 262. 4 Mr Roby however supposes (Preface, p. lxi.) that these superlatives are formed with the suffix -umo: and that the double s is erroneous, representing not assimilation but the strong sound of the s. passes into l, in uellem for uel-sem: v also assimilates itself to 7, in mollis for mol(d)uis (Sanskrit, mridus), sollus for sol-uos (seen in soll-ers, soll-ennis, solli-citus, solli-fides, &c.); compare the phonetically different sal-uos; the cause being doubtless the strong sound of l at the end of a syllable. CH. VIII. (iii) Where the two letters pass into another double (iii) Modi sound. This takes place with some past participles in -tus and derivatives in -tor; where the t of the suffix together with the final letter of the root passes. into SS. When the root itself ends in s, no such change commonly takes place, as us-tus, haus-tus, &c. But it occurs regularly when the root ends with a dental: e.g. fissus for fid-tus, cassus for cad-tus, passus for pat-tus, and many others: sometimes the first s vanishes, as in ui-sus, lae-sus, &c.: sometimes the double s is preserved in old Latin, as ussus, diuis-sus. In these cases Corssen gives the following explanation: the s at the end of the root is due to Dissimilation, as equet-ter becomes eques-ter: then fistus assimilated itself to fissus. Mr Roby argues against this view with great force'. He points out that although the first step is quite possible, the second, which assumes the change of st into ss, is contrary to ordinary Latin usage, in which st is a perfectly stable combination. He therefore holds that the t of the suffix first changed into s-a change which is certain in lap-sus, fixus (fic-sus), fal-sus, &c., and highly probable in pressus (prem-sus), passus (pan-sus), iussus (iub-sus), &c., where the root does not end in a dental, and where there is no need for dissimilation of the final consonant. In this way fid-tus becomes fit-sus, and then by regular assimilation (i) fissus. I think that this theory is very much preferable to Corssen's. Passing next to incomplete assimilation we find the first two cases as in Greek. 1 pp. lvii.-lxi. fication of both sounds. II. Incomplete assimilation. CH. VIII. (i) "Euphonic" changes. (ii) Changes caused by nasals. (iii) Change of t to s. (i) The "euphonic changes," by which a hard passes into a soft before a soft; as in segmentum from sec: and a soft into a hard, as actor from ag. (ii) Momentary sounds passed into nasals, as scamnum for scab-num, Samnium for Sab(i)nium, som-nus for sop-nus1, amnis for ap-nis. Dentals passed into n through s; as in penna, already mentioned. It cannot be proved that gutturals were sounded as ng, e.g. in magnus, tignum: but it is probable on the analogy of the other letters. (iii) We saw just above that fid-tus passed into fit-sus, and then fit-sus into fis-sus by assimilation. To assimilation also would seem to be due the change of t in -tus and -tor when in contact with other final letters than t or d. These are chiefly r and . The change however is only sporadic. The t maintains itself in ar-tus, exper-tus, and many others, but suffers change in cur-sor and cur-sus, spar(g)-sus, &c. Similarly in combination with final , t still appears in altus, cultus, ultus, sepultus, &c.: but s appears in falsus, celsus, pulsus, mul(g)sus, uol-sus, and a few more. These cases may be explained as the result of imperfect assimilation. There was probably some vacillation in sounding the dentals, similar to that already noticed in Greek, which rendered corruption of them more easy. Assibilation of the dentals is not found in languages like the Sanskrit, where they are sounded with the tongue in a definite position against the edge of the upper teeth. But where the tongue is only pressed against an uncertain point of the front palate, the position for t and d becomes similar to that for s, and the change would be facilitated by a preceding r or l: by sounding s instead of t after r or l, the necessity of stemming the breath, which rushes forth in a continuous stream, is avoided. In the same way we may account for the change after nasals in pressus, passus, &c.; in which cases, as we saw above, the assimilation afterwards becomes complete: there is 1 Comp. 264. |