scribed as being of an illiterate stock'. In conformity with this we find that in the best and oldest MSS. the h is often wrongly omitted, as arena, aruspex, &c.: rarely wrongly inserted, as in humor, humerus, &c. These however are trifles to the extraordinary blunders committed by the stonemasons in the inscriptions of the fourth century of the Christian era: such as hac for ac, hornat, hextricata, exhistimantes, &c., quoted by Corssen3. 6. Auxiliary (inorganic) vowels. These vowels are perhaps the farthest extension of the principle which we are considering. They frequently appear to be actual gain, and not loss to a word, causing the addition of a new syllable: and so are difficult to reconcile with the other manifestations of phonetic change. Yet these vowels are distinctly inorganic, as can in almost every case be proved by comparison with other languages. That they should be dynamic is excessively improbable ; what change of meaning is likely to have been intended by sounding layús instead of λaxús (laghu, le(gh)uis, &c.), or αλεγεινός beside ἀλγεινός? It is from vowel-insertions like the last, or from vowel-prefixes like ẻ-x@és for xés, that we gain the conviction of the really phonetic character of these sounds: and accordingly I believe that they arose first from the difficulty of pronouncing a consonantal group, which became much easier when parted by a slight vowel-sound (if in the middle of a word), or if preceded by the same, when initial. Instances of this are well known in modern languages, e.g. in the French état, espèce, espérance, étoile from status, species, sperare, stella, 1 I have already mentioned (at page 20) Roscher's argument, chiefly from Cicero, Orat. 48, 160, in which he maintains against Corssen that this use of the h was a common corruption in the provincial speech, and was just beginning in Cicero's day to steal into the language of educated men. Ib. p. 110. CH. IX. Character sounds: most common in of these Greek. CH. IX. &c., and the corresponding Spanish forms esperanza, estrella, and estava (= stabat): and a vowel similarly prefixed occurs in the late Latin of the fourth century after Christ, as ispirito, istatua. Compare also the Welsh ysprid for spiritus, ysgol for schola, and many others. In these words the y (which seems to me to be the neutral vowel in Welsh) is really irrational; it hardly amounts to a syllable, and so is clearly auxiliary only. But it is not easy to see why this use should be extended to words which do not begin with a compound sound, but only with an easy letter like λ or p, μ or v: and it is especially before these that this inorganic vowel is found; it occurs very rarely before a simple explosive sound; before T, π, o never1. But the reason for this dif ference is not far to seek; a protracted sound has something of the vowel-character about it, and therefore a vowel can easily slip through the lips before it2: in sounding the four letters l, m, n and r especially, the vocal tube is so open that it is easy to let a pure vowel-sound escape at some part of the time during which they can be prolonged. But before a momentary sound the vowel must have been consciously added. Accordingly I believe that this new parasitic sound sprang up before certain liquids and nasals after it had been familiarised to the "Sprachgetühl" by use in cases where it was almost necessary. I do not deny that in some cases a prefixed vowel may be not parasitic, but the remnant of some corrupted prefix, most naturally of a preposition: and this, as might be expected3, is often assumed by Prof. Pott: e.g. he regards ἀμέργω as ἀπομέργω, like απομάσσω, &c. I do not think this likely, because I know no analogy for a similar loss of π; but in some cases such a truncation is doubtless possible. Still in the great majority of cases I have 1 Gr. Et. 678. 2 Compare the irrational vowels in Latin, u before l, and e before ", pp. 276-286. See pp. 105-107. 4 Et. Forsch. II. 386. no hesitation in regarding the vowel as the purely phonetic result of indolent articulation. CH. IX. vowels. I will give examples first of the vowel when initial1. (i) Initial Several have been incidentally mentioned in the previous chapters. Thus a-ornp, as we saw, is from STAR: whence also ἀστράπτω and ἀστραπή: to derive them from a root AS, "to throw" (which occurs in Sanskrit, but not, I think, in the other languages), is not so good. Similarly ἀ-σπαίρω is an easier form of σπαίρω : ἄ-σφι is a Lesbian form of σφί: κίς and χθές have the by-forms ικτίς and ἐχθές: the rare word -τλος seems to be most naturally connected with τλα in ἔτλην: ὄνομα is most likely for -yvoμa. Before a single liquid we have the parasitic e in ἐ-λαχύς, ἐ-ρετμός, ερεύγω (Lat. ructo from rug), έ-ρυθρός, ἐλαύνω for ἐ-λα-ννω" ; α in ἀ-λειφω, ἀ-μειβω, ἀ-μέλγω, &ε.; o in -λίγος, -ρυγμός, and perhaps Ὄλυμπος from Vaμm, by the Aeolic variation of a and v. There are plenty of other examples, more or less uncertain, of the vowel in this connection. The e is probably prefixed before v in elpy for e-Fépy-w (Sk. √vrij), è-Fépo-n (Sk. √vrish), čeíkoow for ẻ-Fikooi (ui-ginti). I have already accounted for the occurrence of this vowel as being one of the many ways in which the Greeks struggled to avoid an initial v. Before a simple explosive sound the best examples are perhaps d-κούω, ο-κέλλω by κέλλω, ό-δάξ, and most likely ỏdous; it is not probable that if the vowel had been original (so that the word should be derived from ED, "to eat"), it would have been lost in all the derived languages; Sanskrit danta, Latin den-ts, Lithuanian dant-i-s, and German Zahn. The cases are very few in all, but they are peculiar extensions of the common usage which I cannot explain. It will be seen that the vowels 1 See Gr. Et. 673, &c. 2 Aesch. S. c. T. 18. 3 Gr. Et. 677. 5 Blomfield (on Aesch. Prom. 191) lays down that where the two forms occur side by side, as kéλλw and ỏkéλw, the first seems to be preferred by poets, the second by prose writers. If this be so, it points to the insertion of the vowel being felt to be a license, on the assumption that Greek and Latin poets always followed the stricter rule. Very rare before an sound. explosive CH. IX. (ii) Medial vowels. regularly found in this use are a, e, o, the strongest vowels. The reason is probably this: at first the prefixed vowel would be the same as the radical vowel', though in actual use there are plenty of exceptions to this rule; and the vowel A in its triple form occurs, as I have already remarked, much oftener in roots than I or U. I pass to the insertion of a vowel between consonants in the middle of a word. These cases are more doubtful, because sometimes the fuller form may be the older, and the vowel have been lost out of it. Thus Schleicher regards the e in opéyw as inserted, and by comparison with Sanskrit varj, arrives at an Indo-European ARG. But the Latin form is reg-o, and we have rak-jan, "to reach," in Gothic. It is therefore at least as probable-I think more so that the older form was RAG; and that 0, not e, was the auxiliary vowel; and therefore opóyvia the older form of opyviá. Other examples, which seem more certain, are given by Schleicher and Curtius: e.g. a-eγεινός mentioned above ; ἤλ-υ-θον from the simple root ελ, whence the secondary root is ελθ; μαλακός by μαλκός (Hesychius)—the 0 in μaλlakós is again the result of dynamic strengthening of the root-doλ-ı-yós from original dargha, Sanskrit dîrgha; dλ-é-¿w beside aλк-ŋ, the Latin arc-eo: TIVUTÓS is from TVU. Apparently we have a suffixed e forming for many verbs a secondary base in common use. Thus μe forms its present stem by the affix yo, as μew: but this stem is phonetically increased by e, whence comes the future μeλλ-n-ow, and the aorist ἐμέλλ-η-σα. μαχ, épéλ-n-oa. Similarly pax," to fight," has a secondary form paye, which is actually used for the present in Ionic, and forms the aorist -μaxe-oáμnv. These new forms, which are rather common, have nothing to do with the formation of stems, for the new vowel commonly runs through them all: and as it is generally rather late, and μαχε, 1 W. Christ, Lautlehre, 19, quoted by Curtius. 2 Comp. p. 76. 4 Curtius gives thirty-eight examples in his excellent School Grammar, which is translated and published in Dr Smith's series, see p. 198. CH. IX. The "con necting produces no change of meaning in the verb, the vowel is probably a mere phonetic insertion, closely akin to the connecting vowel" (the German Binde-vocal), which is so important in the conjugation of verbs. The object, vowel." of that vowel is to preserve the final consonant of a root from all possible injury when it is connected with suffixes beginning with consonants; e.g. in forming the second person of βολ or βουλ, the Greeks insert e before σαι, as βουλ-ε-σαι, βουλ-ε-αι, βούλει, so that in all these changes the λ has not suffered; on the contrary, the Latins, who in a few verbs (as Ves, Ned, vel, fer) do not regularly employ a connecting vowel, lost the final consonant in uol-s, uil-s, uis. I cannot agree with Schleicher, who (following Bopp) regards this vowel as a demonstrative root1, whose original form was a (preserved in Sanskrit and weakened in other languages, in Greek to e and o, in Latin to i and u, according to the following sound.) I can see no proof of such a view: and prefer to regard the vowel as simply phonetic, and belonging in its origin to the class which we are here considering; at a later time of course it became one of the arbitrary forms of grammar. It is worth observing that the oldest verbs (so far as we can trace the historical development of the verb) in Sanskrit, in Greek, and in Latin, do not generally possess these connecting vowels; and it is just as likely that they never had them, as that they had them and lost them. Bopp's objection however is no doubt forcible, that a the strongest of the three primary vowels is least of all adapted for a mere phonetic link; and is not quite met by the reply that a does not occur in Greek and Latin, and that the Sanskrit a is not the full vowel of the primitive language. Still, the evidence seems to me to preponderate for the view that I have given*. 1 Comp. p. 343. 2 Comp. Grammar, vol. 11. p. 694, English translation. * Except perhaps a few verbs, such as dy-α-μαι, πρί-α-μαι, &c. 4 See, on the whole question, Curtius Erlaüt. pp. 107–110 (Engl. trans.). |