- G. R. Solta, Zur Stellung der lateinischen Sprache (Vienna: Österr. Akademie der Wissenschaft, 1974), pp. 22 ff.
- Concerning the -r-middle—which is Keltic, Latin, Italic, Phrygian, Hittite, and Tocharian—we may observe that the Greek and Indo-Iranian *-(m)-ai *-sai *-tai form seems to be Baltic also (Prussian asmai, Lith dúomie[-si]) and Goth (bairada, etc.); cf. K. Brugmann, Grundr2 (Strasbourg), III, p. 642, and E. Prokosch, A Comparative Germanic Grammar (Baltimore: Linguistic Society of America, 1938), p. 219. The final Gothic -ă of one mora is the regular Gothic shortening of the *-ai of two morae (Gk ϕερ∈τας). I cannot see how A. Meillet can write (Bulletin de la Société de Linguistique de Paris, XXIX [1928], 29): “il n'y a pas de particularités phoniques ou morphologiques spéciales à l'indo-iranien et à l'italo-celtique, tandis que de nombreux mots de caractére religieux ou politique se rencontrent seulement dans les deux groupes.” Meillet had here not yet reached the Crocean concept of neo-linguistics, which eliminates every difference between lexicon, phonology, and morphology.
- See M. Mayrhofer, Kurzgefasztes etymol. wb. des Altirtd. (Heidelberg: Winter, 1974), s.v.
- I use Italic, of course, in the sense of Devoto, as a very felicitous substitute for Osco-Umbrian. It corresponds to the denomination which the Latins used.
- That the Indo-Aryan palatals were originally only velars with a very slight palatal articulation (as suggested to me by an Indian colleague) is proved by the fact that in the order of letters (which, as is well known, is rigorously phonetic in India) the “palatals” precede the ‘cacuminals”, whereas, if they were real palatals (like E eh and j, as in chin and joint), they should follow. Cf., e.g., W. S. Allen, Phonetics in Ancient India (Oxford: Oxford Univ. Press, 1953; rpt. 1961), p. 52.
- Cf. Soita, p. 15.
- For the Latin words for ‘son’, ‘daughter’, and ‘woman’, see the conclusion of this article.
- On the concept of “negative innovation”, of which he definitely exaggerates the importance, see Calvert Watkins, ‘Italo-Keltic Revisited,” in Ancient Indo-European Dialects, ed. H. Birnbaum and J. Puhvel (Berkeley and Los Angeles: Univ. of California Press, 1966), p. 31.
- I believe that deutsch, that is, OHG diutisc, may have been the “language of the people”, in opposition to Latin (Kluge-Mitzka20 [see n. 11 below] thinks the opposition was to Welsch, that is, Romance): ‘als Adj[ektiv] auf -isch ist d[eutsch] gebildet wie völkisch zu Volk.” This is also the opinion of E. Benveniste, Le vocabulaire des institutions indo-européenes, I (Paris: Éditions de Minuit, 1969), pp. 364 ff.). But Teutoni is anterior to the spread of Latin as the language of learning.
- It seems that *teutá is also Thracian (Tautomedes). I cannot, however, accept in the least what Benveniste says about the etymology of L tōtus (without any mention of my work in Riccerche linguistiche—see n. 13 below), p. 366; I agree entirely, however, about plēbs, πλῆθoς, pleme (ibid.), although the Latin b of plēbs is strange. Carruba wrongly follows him here too (on tōtus).
- I absolutely cannot accept Benveniste's theory (Le vocabulaire des institutions indoeuropéennes, I, p. 366) [followed by Carruba], who denies any connection between Hittite tuzzis ‘army’ and * teutá ‘people’. Suffice it to recall that among the Teutons and the Macedonians the ‘army’ was nothing else but ‘the people in arms’, which decided about public matters (as is proved by many ancient texts); and cf. L populus and populātur ‘devastates, plunders’ (perhaps an Italic penetration).
- Kluge-Mitzka, Etym. Wb. d. dt. Spr.20 (Berlin: Walter de Gruyter, 1967) cites some forms that do not exist: OIr tuoth, Latv tautà.
- I obviously cannot accept what Watkins (see n. 8 above) says about *teutá and rēx (p. 44 ff.), although I am glad he accepts my explanation of tōtus (Riccerche linguistiche IV [1958], 164–176). G. Polomé also accepts my interpretation; but I cannot follow what he (and Szemerényi) say against the “Western” *teutá ‘people’ by pointing to the Iranian evidence. Neo-Persian toda means ‘mass, mob’, and Sogdian (Christian) twdy and (Buddhist) twδ'k mean ‘mass, conglomeration’, meanings that are as far as can be from the precise juridical sense of *teutá in the West (Germanic, and especially Keltic and “Italic”).
- Tōtus is of Italic origin, as I have shown in Riccerche linguistiche, IV (1958), 164 ff. On populus and Italic *teutá see also G. Devoto, Tabula Iguvince2, (Rome: Typis Regiae Officinae Polygraphicae, 1954), II s.vv. totam (p. 440) and poplom (p. 435); now also P. Catalano, Populus romartus Quiritium (Turin: 1974), pp. 113 ff.
- We can see in the Latin a break after ‘6’ (quārtus quintus sextus/ septimus octāuos, etc.) a trace of the Babylonian duodecimal system, as in English eleven twelve (thirteen fourteen) and in Greek έζήκoντα/έβδoμήκoντα, ơγδωκoντα, and so on. OIr cethramad must'be a relatively recent Irish innovation, for MW pedwyryd, fcm.pedwared, MBret pevare ‘fourth’ show the same (old!) ending as Vedic (Thurneysen, Hb., p. 238). I therefore disagree entirely with Sommer and Szemerényi (see n. 20 below).
- See also Warren Cowgill, “Italic and Celtic Superlatives and the Dialects of IndoEuropean,” in Indo-European and Indo-Europeans: Papers Presented at the Third Indo-European Conference at the University of Pennsylvania, ed. George Cardona, Henry M. Hoenigswald, and Alfred Senn (Philadelphia: Uni v. of PennsylvaniaPress, 1970), pp. 116 ff.
- W. Schulze, E. Sieg, and W. Siegling, Toch. Gr. (Göttingen, 1931), p. 199.
- A. Meillet, MSL, XVII (1911-12), 282 ff.
- I cannot follow Sommer (see n. 20 below) when he draws (p. 21, n. 1) δέκατoς from the collective the accent forbids it (a difficulty which he sees himself and for which he finds an unconvincing explanation).
- For Osean we know only Púntiis ‘Quintus’ (personal name) and DEKMANNIÚfS, which points to the type decimus (but deketasiúi perhaps to the type tenth). For Umbrian we have NUVIME ‘nonum’, which points to the Latin type too (and may be due to Latin substratum: note the m).
- Faliscan neuen (G. Giacomelli, La lingua falisca [Florence, 1962], pp. 121 and 262); cf. also L nōnāgintā.
- G. Devoto, I, (Tabulae Iguvinae2 (Rome: Typis Regiae Officinae Polygraphicae, 1954) 353, following Ribezzo, reconstructs, on the basis of Old Umbrian tekvias (i.e., *dek(u)vias) for Umbrian an *ogduwos, which resembles very closely Gk (perhaps from see Frisk, s.v. ). Phrygian has (locative), which is very near to this *ogduwos (with t [from kt] through the influenç of ). Cf. also Goth ahtuda (Sommer [see n. 20], p. 25—who admits that *-tois late; see also p. 30), Lith astuñtas from *aštutas (R. Trautmann, Balt.-slav. Wb. [Göttingen, 1923], p. 15). Cf. Meillet (n. 2 above) p. 32.
- As for the mysterious νν of , we should notice that in Aeolic νF gives νν and therefore ἐν Fα, Horn *εςνἁ- ετες (Sommer, p. 31: ‘ohne Nachwirkung des F wie in ἐνἁτη B 313”) gives in Aeolic ἐννα-; ἐννέα is a dialectal mixture of ἐννα and *enewn(Sommer, p. 27). Cf. also ἐννήκoντα Od. 19. 174, with the commentary of Sommer, pp. 32 ff.
- See on this subject J. Vendryes Bulletin de la Société de Linguistique de Paris, XXV (1925), 37 ff., and Meillet (see n. 2 above), pp. 29 if. The difficulty that Meillet finds (”aucune règle”, pp. 30 ff.) in the voiced stops of Gk ἑβδoμoς, OCS sedmŭ, Gk does not exist, in my opinion: forms like *ἑπτoμoς (Lat. septimus), *ὀκτoFoς (L. octāuos) crossed with *έβδμoς and analogically (the form with F is attested in Aetolian: Carl Darling Buck, Introduction to the Study of the Greek Dialects, 2nd. ed. [Boston: Ginn, 1928], p. 95); it is a known fact that numerals are very greatly subjected to analogical influences because of counting: one two three four five six seven eight, and so on; thus at Heraklea hoκτω hoκτακατιoς hhas its hfrom ℎ∈πτα, because of the series ℎ∈πτά ℎoκτώ *ℎ∈ννέα (cf. ℎ∈νατoς at Thera and Delos, Buck, p. 96). A crossing of with a form (OCS vosmŭ) is, of course, also possible.
- The old type septimus also appears in Hittite, where the only numeral known to us is siptamiya, which can be read septamiya, dative of *septamas according to Sturtevant, pp. 89 and 179; but see also Sommer, (see below), p. 23, n. 1.
- The form ἑβδ∈μήκoντα (see Sommer [see below] p. 24, with sources) is obviously the result of an assimilation (e—o>e—e).
- Concerning the chronological relationship of the suffixes *-moand *-to-, Meillet (”following the Italian school of neo-linguistics”, as Szemerényi writes, p. 70) claims greater antiquity for *-moon areal grounds (see n. 2 above), p. 29 ff. Cf. also M. Lejeune, Bulletin de la Société de Linguistique de Paris, XXIX (1928), 110 ff.
- I cannot discuss here in detail the works of F. Sommer, “Zum Zahlwort,” Sitzungsber. der Bayer. Akad. [Munich, 1950], Heft 7, and of O. Szemerényi, Studies in the IE System of Numerals (Heidelberg: Winter 1960). To Szemerényi I shall merely remark:
- 1. That he completely neglects areal linguistics (nay, shows some hostility toward it).
- 2. That on p. 68 he considers the usage of ‘two decades’, ‘three decades’, etc., instead of ‘20’, ‘30’, etc., as absurd. He does not seem to know that it is very frequent in many Southern Italian dialects (G. Rohlfs, Hist. Gramm, der ital. Spr., 3 vols. [Bern: A. Francke, 1949–54], III, 313 ff.).
- 3. That Keltic, Latin, Umbrian, Greek, Iranian, Vedic, and Tocharian (which, according to him have preserved the older system of numerals—”àrea a ferro di cavallo” [p. 5],—) point very definitely to , not to (p. 68).
- 4. That he entirely neglects Faliscan neuen ‘9’ (G. Giacomelli, La lingua falisca, Florence, 1962, pp. 121 and 262), which seems to me to be certain. Latin has nōnus (nouem is therefore rebuilt on septem and decern). Cf. also Lith deviñtas, beside deši&mtas.
- The same idea (an evident one, in my opinion) on ἑβδoμoς is expressed by Sommer (see above), pp. 17 ff. I follow therefore Brugmann, Grundr.2, I, pp. 631 and 822; his explanation of ἑβδoμoς, sedmŭ, etc. (t becomes d before following nasal) seems obvious to me.
- Cf. J. Vendryes, Bulletin de la Société de Linguistique de Paris, XXV (1925), 37.
- Or nametos? Cf. Meillet (see n. 2 above), p. 32, n. 1.
- Cf. R. Thurneysen, Handb. [Heidelburg, 1909], p. 236.
- Watkins (see n. 8 above) also admits (p. 37) that the Keltic -ameto-s (decametos, etc.) ‘simply combines the two [suffixes], *-amo+ *-eto-” (or *-mo-+*-to-, which is the same thing).
- Greek, a “central” language, has also ἑβδóματoς ὀγδóατς, forms that are astonishingly similar to those of Keltic. They are studied by Lejeune (see n. 20 above), pp. 114 ff., who considers them Greek creations. They may, however, be connected with the Keltic forms through Illyrian (see my work in RIGI, cited in n. 33 below); it would be a very old central innovation which has spread from Greek to Keltic through Illyrian.
- Cf. also Meillet (see n. 2 above), p. 30: ‘Quant au grec, c'est la langue où le type en -τατoς s'oppose à skr. -tamah, lat. -timus (ultimus, etc.), got. -tuma (afturna, etc.).” It is a known fact that the endings of the ordinal numerals are the same as those of the superlative (L postrēmus, imus, suprēmus, etc. =primus, septimus, etc.; E first=Gk ἅρςστoς). Cf. also Gk μέσατoς: Av maδəmō, Vedic madhyamà-h, Goth miduma, OHG mittamo (Meillet, p. 35). Cf. also Lejeune, pp. 114 ff.
- Cf. Loth, Revue Celtique, XLI (1924), 207–208, who quotes three examples of OW ner ‘chef’, ‘maître’ (from *nero-) and says that in MW ner “is not rare”; there are, however, very few cases.
- What A. Ernout and A. Meillet4, Dict. (Paris: Klincksieck, 1967) say s.vv Nero, uir, and p. 822 (which obviously represents the opinion of Ernout and not of Meillet) and what Ernout writes in Philologica, III, (1965), 90 ff. (he tries to find an IE difference between uir and ἀνήρ) does not convince me in the least.
- In Umbrian uir appears only in the ancient dvandva uiropequo, which has its exact correspondence in Avestan pasuvira (Vid. 6, 32, etc.); it is therefore obviously an archaism, probably foreign to the ordinary spoken language. ‘Man (male)’ in Umbrian is homonus (dat. pl.), cf. Devoto, Tab. Iguv., pp. 198. Of course neris found in Osean and Umbrian only in the meaning of ‘princeps, magistratus’; this does not mean that it had not (at least in Osean) the signification of uir: cf. L duumuir, triumuir, decemuir, etc (cf. Pisani, Le lingue dell’ Italia antica, Turin, 1964, p. 61, line 7). In Osean, humuns means ‘homines’.
- Gl., IV, 124, 22; V, 468, 2.
- Suetonius, Tib. 1,2: quo significatur lingua Sabina fortis ac strenuus.
- Even Germanic has not escaped labialization entirely, and precisely before the first consonantal rotation (which made f of p): cf. G, E wolf, G elf, zwölf, etc.).
- See Pedersen, I. 108.
- And sometimes (but rarely, and only before u) *gh>f: e.g., L fundit, G gieszt, Gk χέ∈ι, Vedic Juhṓti. But in these cases too there may have been an Italic influence. Cf., e.g., hircus: Sabine fircus, haedus (edus): Sabine fedus (Varro, De Lingua Latina, V, 97).33On the Western labialization(*kw>p, etc.), see also G. Bonfante, RIGI, XIX (1935), 161 ff. (the table needs some corrections). See now also A. Martinet, “Des labiovélaires aux labiales dans les dialectes indo-européens,” Indo-Celtica, Gedächtnisschrift für Alf Sommerfeit (Munich: Max Hueber Verlag, 1972), pp. 89 ff., reproduced in Évolution des langues et reconstruction, 1975, pp. 169 ff. (who rightly connects the passage of *kw>p with the disappearance of the ancient IE *p (p. 170). This would make Brythonic the epicenter of the innovation *kw>p).
- Cf. A. Walde, Vergleichendes Wörterbuch der indogermanischen Sprachen, herausgegeben und bearbeitet von J. Pokorny (Berlin and Leipzig, Walter de Scuyter, 1927-32), I, 34.
- Slavic has voda, fem., from the root of , Wasser, etc. (*wod-: *ud-), either under the influence of Lithuanian and Latvian (where the word is frequently feminine because there is no neuter) or (more likely) because it has superimposed the word on the feminine type aqua, exactly as in Umbrian the type πûρ, Feuer, Umbr pir has become masculine (purom-e: cf. Devoto, Tab. Iguv., p. 175) because it has been superimposed on the older Latin type ignis (masc.).
- Cf. A. Meillet and S. Lévi, MSL, XVIII (1914), 25, and, after them, A. J. Van Windekens, Lexique étymologique du tokharien, p. 130 (but cf. the absurd etymology on p. 47).
- Cf. G. Bonfante, ‘Diritto romano e diritto indoeuropeo,’ in Studii in onore di E. Betti, II (1961), 88 ff.
- On this problem see now W. Merlingen, Mνήμης χάρςν, II, 1957, 49 ff. (who does not convince me). He discusses the question at length and gives all the recent bibliography, except my Dialetti indoeuropei (Naples, 1931), pp. 181 ff. (where the older bibliography is also given). I maintain in part the thesis I defended there.
- Illyrian too has t (Artos, Mess Aρτας, Ven Artorius, personal names), cf. G. Bonfante, RIGI, XVIII (1934), 223 ff; XIX (1935), 161 ff; Krahe, Die Sprache der Illyrier, II, Wiesbaden (Untermann), 170 and 199; Artorius (Untermann, Die venetischen Personennamen, p. 114) must be connected with Artos, Artas ‘bear’, with the frequent Illyrian and Venetic suffix -orius (cf. Untermann, p. 173). On Messapic Artos see RE, s.v. The name Artos, Artas ‘bear’ should therefore be added under ‘men bearing names of animals’ in Krähe, Die Spr. der Illyrier, I (Wiesbaden, 1955), 70 ff.
- For Hittite, see, e.g., Edgar H. Sturtevant, A Comparative Grammar of the Hittite Language (Philadelphia: Linguistic Society of America, 1933), pp. 74–75.
- Cf. Brugmann, Grundr.2, I, 608 f.
- A. Meillet, Esquisse d'une histoire de la l.l.3 (rpt. Paris: Klincksieck, 1966), p. 19.
- See Watkins (see n. 8 above), pp. 40 f.43Cf. Marstrander, Norsk Tidsskrift for Sprogvidenskap, IV (1930), 426: ‘The Celtic uerbum sciendi varies between three forms: ṷind-, ṷidand ṷeid-.”
- Cf. Pokorny, p. 1126, and Walde-Pokorny, I, 236–237.
- The plural genitive *-āsom (L feminārum) is also probably a very recent innovation coming from the Greco-Italic area, as V. Pisani (Storia della lingua latina, Turin, 1962, pp. 104–105) and C. Campanile (Studi sulla posizione del latino, Pisa, 1968, pp. 102–103) very shrewdly suggest. I agree with Pisani, who thinks that “la spiegazione piú probabile è che l'innovazione abbia avuto primamente luogo nelle fasi preistorica dell'oscoumbro ancora balcanico e del greco e in Italia sia passata dall'oscoumbro al latino.”
- I leave aside, of course, a series of Germanic words which are certainly derived from Keltic, such as (I give the German item, for the most part, but they are found in other Germanic languages too): Amt, Eid, Eisen, Ger, Geisel, Reich, reich, cf. Kluge-Mitzka20 s.vv. Amt and Eid: “Bei der staatsrecht [lichen] Überlegenheit, die die Kelten zur Zeit der Abgabe von Wörtern wie Amt (s. d[ieses]) bewahrt haben (Cäsar, Bell, gall., 6, 24 fuit antea tempus cum Germanos Galli uirtute superarent) ist wahrscheinlich, dasz sie ihren germ[anischen] Nachbarn auch die bedingte Selbstverfluchung vermittelt haben: C. S. Eiston, 1934, The Earliest Relations between Celts and Germans, 65 f.” We can probably add (according to D'Arbois de Jubainville, MSL, VII [1892], 289–90; Les premiers habitants de l'Europe, II, 335, a great authority in Keltic studies): G Erbe, Magd, frei, Goth *wers, dulgs, magus, etc., and all the Germanic cognates, of course. See also S. Feist, Vergi, wb. der got. spr.3 (Leiden, 1939), under the corresponding items. Probably also G Zaun, E town derives from Keltic (cf. Lugu-dūnum, etc., and the corresponding insular forms). See also J. Vendryes, MSL, XX (1918), 284, and M. Hubert, Les Celtes V (Paris, 1950), 176 ff.
- On the Italo-Keltic problem see now Watkins (see n. 8 above), pp. 29 ff. He makes some good remarks, and his conclusion (that there was no “Italo-Keltic group” or ‘language” [p. 58] is correct: strangely enough, however, he willingly ignores (p. 32) the problem of ‘Italic unity”, which is essential for the study of the ‘Italo-Keltic” problem. Watkins definitely goes too far, however, in denying Latino-Keltic isoglosses.
- After I had finished this article I had the opportunity to see the book by Myles Dillon and Nora Chadwick, I regni dei Celti (Italian translation of The Celtic Realms) (Milan, 1968). They deal with the position of Keltic (and Keltic words in Germanic) on pp. 279 ff., with ri and tuath on pp. 137 ff. I have no reason to change anything in my article after consulting this book (otherwise excellent).
WORD
Volume 28, 1977 - Issue 1-2: Celtic Linguistics--1976