Advanced search
Publication Cover
WORD Volume 1, 1945 - Issue 2
Submit an article Journal homepage
325
Views
0
CrossRef citations to date
0
Altmetric
Original Articles

On Reconstruction and Linguistic Method

Giluano BonfantePrinceton University
Pages 132-161 | Published online: 04 Dec 2015

  • Strange as it may seem, since it goes directly against his linguistic ideas and methods, this tendency has been clearly seen by L. Bloomfield (Language, New York, 1933), p. 429: “The surface [?] study of semantìcs indicates that refined and abstract [italics mine! meanings largely grow out of more concrete meanings.” He then gives some very good examples, such as Eng. understand, catch on, get, grasp, Ital. capire, Lat. comprehendere, dēfīnīre, ēlīnāre, Russ. pónjat', Germ, verstehen. To these, more might easily be added, e.g., Ital. afferrare, Lat. percipere, coneipere, promittere, permitiere, cögitāre, dēcīdere, putāre, Low Lat. dēuenīre, Eng. become 'fieri,” Gerin. werden, I-E *es-, *bhū-, all having originally a concrete meaning. Bloomfield then goes on, p. 430: “All this, aside from its extra-linguistic interest, gives us some measure of probability by which we can judge of etymologic comparisons” [italics mine]. The better self of the experienced linguist has taken hand of the mechanist Bloomfield against his own will.
  • The opposite phenomenon, although much rarer, also exìsts. In Romanian, for example, Lat. anima has taken over the meaning of cor in the anatomical sense, whereas in the other Romance languages, as already in Latin, cor has added an abstract sense to its original connotation. It is worth noting also that the words for “work” in most European languages originally meant “pain,” “suffering,” “torment,” as Ital. lavoro, Fr. travail, Span, trabajo, Germ, arbeit, MnGk. πὁνoॢ, etc. (cf. on this subject the excellent remarks of H. Sperber, Der Affekt ah Ursache der Sprachveränderung [Halle, 1914], p. 142; Feist, Wörterbuch3, p. 555b; C. Rakovitsǎ. BL 7 [1939]. 99 “sqq.). General bibliography in Schwyzer, Griech. Grammatik, pp. 40 sqq.: Stolz-Schmalz, Lateinische Grammatik5, revised by M. Leumann and J. B. Hofmann, (Munich, 1928), pp. 22 sqq.
  • The existence of semantic “laws” of development, of the same kind as phonetic “laws,” has been asserted by several scholars, e.g., A. Thumb and K. Marbe, Experimentelle Untersuchungen über die psychologischen Grundlagen der sprachlichen Analogiebildung (Leipzig, 1901), p. 86; Zupitza (cf. KZ 43 [19101. 352); Kinkel, LGRPh 23 [1902] 403 (“hier darf also wohl ein gesetzmässiger Zusammenhang angenommen werden”); M. Bréal, Essai de sémantique6 (Paris, 1913), pp. 5 sqq. 9 sqq.; W. Wundt, Völkerpsychologie, 1. 2 (Leipzig, 1900), pp. 487 sqq.; 580 sqq. (chapter on “Allegemeine Gesetze des Bedeutungswandels” and others); G. Esnault, La sémantique, pp. 130 sqq. (in the volume Où en sont les études de français, published by A. Dauzat, Paris, 1935). Against them cf. A. Thomas. Revue des deux mondes, 1902, pp. 582 sqq.; K. Nyrop, Grammaire historique de la langue française, 4 (Copenhagen, 1913), p. 79, § 112; H. Schuchardt, LGRPh 23 (1902). 400; Schwyzer, Grammatik, p. 38, note 3; R. M. Meyer, KZ 43 (1910). 352 sqq.; L. Spitzer, MLQ 4 (1943). 427 (though I do not see how he can write: “No one has ever thought of offering a 'semantic law”); S. de Ullmann, PMLA 60 (1945). 812 with n. 9 (bibl.).—Sperber ZfdA 59 (1922). 49 sqq.; Einführung in die Bedeutungslehre (Bonn and Leipzig, 1923), 56 sqq., holds more or less an intermediate position, although his article in ZfdA has the title Ein Gesetz der Bedeutungsentwicklung and the chapter in question in his Bedeutungslehre is Die Gesetzmässigkeit des Bedeutungswandels; cf. especially ZfdA 59 (1922). 53: “Anderseits aber dürfen uns die bisherigen miszerfolge nicht dazu veranlassen, die suche nach den gesetzen der bedeutungsentwicklung aufzugeben oder gar, wie dies zb. Nyrop [Grammaire, 4. § 112] und de Saussure [Cours de linguistique générale (Paris, 1922), p. 135] zu tun scheinen, die existenz solcher gesetze überhaupt in zweifei zu ziehen. Anzuerkennen, dasz sich im sprachleben irgend etwas regellos und rein zufällig vollziehe, wäre dasselbe, wie unsere Wïssenschaft auszuschalten aus dem kreis jener disciplinen, für die der satz von der gesetzmäsigkeit alles geschehene nicht nur ein axiom, sondern auch eine unerschöpfliche quelle neuer erkenntnisse geworden ist.”
  • But no real “laws” are necessary for semantic reconstruction, just as no phonetic “laws” are necessary for phonologic reconstruction (on the contrary, they are a hindrance). One fact, however, remains established: certain semantic changes are possible and extremely frequent, certain others are impossible or at least extremely rare. Cf. also Sperber, ibid., p. 51: “So hab ich seinerzeit nachzuweisen versucht, dass die deutschen ausdrücke for den geschlechtsact bedeutungsentwicklungen aufweisen, die sich so typisch widerholen, dasz mann geradezu einen bei den verschiedensten worten mehr oder weniger vollständig ausgebildeten Stammbaum der secondärbedeutungen aufstellen kann [Imago 1 (1912). 405 sqq.]. Aber trotz der bekräftigung die dieser Stammbaum nachträglich dadurch erfahren hat, dasz Spitzer [W uS 5 (1913). 206 sqq.] parallelen dazu aus dem romanischen und zum teil auch aus dem slavischen beigebracht hat, bin ich doch heute der überzeugung, dasz derartige versuche niemals zu einen befriedigenden resultat führen werden.” See also Meillet, RCr 53 (1902). 66; A. Dauzat, Essai de méthodologie linguistique (Paris. 1906), pp. 134 sqq.; J. v. Rozwadowski, Wortbildung und Wortbedeutung (Heidelberg, 1904), pp. 72–3, 79 sqq., 87 sqq; Schuchardt, ZRPh 25. (1901). 255; B. A. Terracini, Enciclopedia italiana, s.v. Semantica, p. 336 a; Meyer-Lübke, Wb3, s. v. bagascia.
  • When H. Hirt used, e.g., the ablaut as a means of chronological distinction of the different elements of “Indo European” (as when he declared that the type γένoॢ was recent, because it shows no Ablaut), he employed the method of internal analysis to one single language, viz., his reconstructed I-E (cf. Brandenstein, Wanderung, pp. 15; 173). Cf. also Pisani in his pamphlet Paleontologia linguistica (Càgliari, 1938), pp. 32–33: “La ricostruzione di stadi linguistici più antichi ha luogo attraverso due processi: uno, che io chiamo di ‘ricostruzione interna’, in cui lo studioso, ammaestrato beninteso da ciò che gli insegnano le altre lingue dello stesso gruppo, trae dal materiale monoglottico tramandato (compreso quanto offre la ricerca degli antichi imprestiti, ecc.) tutte le conseguenze possibili per raggiungere fasi anteriori, l'altro di ‘comparazione’, in cui i dati così ottenuti per ogni singola lingua del gruppo vengono confrontati fra loro per vedere se essi risalgono, o non, ad estuali fenomeni. […] La ricostruzione interna ci fa per esempio sapere che in certe parole il greco aveva un σ ove attualmente è attestato uno spirito aspro o uno iato intervocalico, e il latino aveva s dove ci è tramandato un r, che quindi γένεoॢ e generis risalgono a *γένεσoॢ e *genesis; la comparazione ci assicura che s in tali parole era patrimonio indoeuropeo comune. D'altro lato la ricostruzione interna ci fa pensare che il latino arx debba essere una parola assai antica; il fatto che ad essa non si trovino sicure corrispondenze in altre lingue indoeuropee, non ci permette di affermare che arx esistesse già in periodo unitario. Ma i casi del genere di arx sono numerosissimi per ogni lingua indoeuropea, e se non possiamo affermare nulla per ognuno singolarmente preso, possiamo però tranquillamente dire che molti di essi risalgono senza dubbio al periodo unitario; si tratta beninteso di fatti non solo lessicali, ma anche morfologici e sintattici.” Pisani, therefore would use the two methods one after the other; I think they can be used independently. Cf. also Pisani, Mem. Acc. Line. 6. 9. 2 (1940). 264 sqq.; 271–2. More clearly A. Pagliaro, Sommario di linguistica arioeuropea (Rome, 1930), p. 174: “Questa ed altre norme che il Bartoli ha stabilite mettono in luce I'importanza dell'aiuto che si può ricavare dallo studio della distribuzione geografica delle singole innovazioni. Altri elementi si possono dedurre, specialmente per quanto concerne il mutamento fonetico, da indizi intrinseci. Per esempio intorno all'epoca della scomparsa del F in attico si dice che essa è da porre dopo il passaggio attico di—ρη in—ρα a motivo di κὁρη da * κoύρη ion. κὁρFη e prima del passaggio di ιη, εη, υη, in ιᾶ, εᾶ, υᾶ, come si rileva da πoία (πóα)=πoίFη, νέα= νέFη e forse anche da ναυαγòς=ναυ-Fαγoς, ion. ναυηγòς. … I processi contenuti in forme come μέσoς da *medidos, τείχεσι≠ da τείχεσσι≠, βάτις da *βάσις sono certamente posteriori alla scomparsa di ς [sic] intervocalico in greco; analogamente quello contenuto in lat. causa da *caussa [sic] è posteriore alla rotacizzazione di s intervocalico in latino”.
  • Cf. also Introduction7, p. 46: “Une forme d'une langue historiquement attestée ne peut passer pour sûrement ancienne que si elle n'est pas susceptible d'avoir été faite en vertu du système général de la langue où elle est attestée. Ainsi lat. est: sunt et got. ist: sind sont sûrement anciens parce que le procédé par lequel ces formes sont obtenues est étranger au latin et au gotique, et ces formes ont chance de remonter à l'indoeuropéen parce que le type qu'elles représentent est en régression constante depuis l'époque indo-européenne commune […]. C'est surtout avec des anomalies de l'époque historique qu'on restitue la règle de l'époque indo-européenne.” Cf. also ibid., p. 32, and BSL 29 (1929). 37: “Les formations d'ordinaux montrent, une fois de plus, que l'état de choses indoeuropéen ne se laisse restituer ou deviner qu'à travers les formes anomales, et que les dérivés normaux enseignent peu de choses sur la langue commune initiale: règle de méthode qu'on perd trop de vue.” See also BSL 32 (1913). 26 sqq.; Esq.3, p. 19.
  • It is only fair to quote here Jespersen (Language, p. 425, Ch.l, § 7 [“Irregularities”]): “Another point of great importance is this: in early languages we find a greater number of irregularities, exceptions, anomalies, than in modern ones.” However, it is easy to see that Jespersen' idea is connected with a particular conception of linguistic evolution (“primitive languages,” etc.), which I happen to accept fully, but which obviously changes the methodological value of his remark for the problem we are studying here. As an opposite extreme, I shall quote C. Bartholomæ. BB 15 (1889). 30: “Heteroklisie ist gewiss nirgend etwas ursprüngliches” (against which see Bonfante, Emerita 3. [1935] 269).
  • When Feist (Wörterbuch3, p. 19b, against Bonfante), Brandenstein (Wanderung, pp. 16–17), and A. Nehring (in Indogermanen und Germanenfrage [Salzburg and Leipzig, 1936], pp. 67–8; 110; 148 sqq.; 156; 211; Actes du II' congrès international de linguistes [Paris, 1933], pp. 191. sqq.; criticized by Pisani, pp. 18 sqq.) assert that the I-E heteroclitic nouns (*udōr, *udnés, etc.) are older than the*-o-,*-ā-stems, they apply this same method. Cf. also Sköld, Sprachgeographic, p. 16: “Nun gehören eben die heteroklitischen Bildungen einer Zeit an, wo die i[ndo]g[ermanische] Deklination in ihrer jetzt erreichbaren Gestalt, wo vor allem die Genusindikation [?] noch nicht entstanden war.” Cf. also Oertel, Lectures on the Study of Language (1902), pp. 174 sqq.; 320 sqq.
  • I consider such forms as he may, he can, etc., as archaisms, inasmuch as they are survivals of a primitive atemporal era, when ‘preterites’ were used indifferently for the present and for the past (therefore, they were, as a whole, employed much more for the present than they are now, when most of them are strictly limited to the past: he came, he gave, he saw, etc.). This is probably what Sturtevant meant when he wrote Lg 2 (1926). 34 (although he expressed himself in a rather strange manner): “In meaning the Hittite perfect and its preterite are precisely equivalent to the present formations. It follows that the Germanic preterito- prescnts, the prevailing present force of the Greek perfect, such Latin idioms as nōuī ‘I know’ and ödī ‘I hate,’ etc., go back not only to primitive Indo-European, but to Pre-Indo-European.”
  • This principle was already applied—although not enunciated—by Meillet, Introduction7, p. 40: “Mais certains dialectes de Ia Sardaigne ont, d'une part, pira, pilu, et de l'autre, veru; comme la différence entre i et e ne s'explique pas par l'influence des articulations voisines, elle doit être ancienne,” as it is in fact: Lat. pǐram, pǐlum: uērum. To cite another example from the Romance languages, I may point out that the difference between ie- (ye-) and ge-, although preserved only in Logudorian, would he considered old even if we did not know Latin (which has, in inscriptions, iēnuārius, but geminus, geiier), because there is no reason why an original single sound should have split into ge- and ye- in Logudorian (and this quite apart from every geographic consideration, on which see below). Cf. Wagner, LGRPh 39 (1918). 131, n. 1: “Die Zentralmundarten [of Logudoru in Sardinia] halten g- und y- strenge auseinander.” See also Campus, Giorn. stor. della letterat. itcd. 72 (1918). 163 and Sturtevant, Pronunciation2, pp. 166 ff.
  • These examples, with many others, may be found in Gilliéron's well-known works. A good survey of the whole subject is given, e.g., by K. Jaberg, Sprachgeographie (Aarau, 1908); E. Gamillscheg, Die Sprachgeographie (Bielefeld and Leipzig, 1928); A. Terracher, L'Histoire des langues et la géographie linguistique (Oxford, 1929); A. Dauzat, La Géographie linguistique (Paris, 1922); U. Leo, Sprachgeographie und romanische Sprachwissenschaft, ASNS 162 (1932–33). 203 sqq.; I. Iordan, Introduction to Romance Linguistics (London, 1937), pp. 144 sqq.; J. Brøndum-Nielsen, Dialekter og dialektjorskning, Copenhagen, 1927. A good bibliography (up to 1933) is that of J. Schrijnen, Essai de bibliographie de géographie linguistique générale (Nijmegen, 1933); for Romanian examples, cf. the Introduction to vol. 1 of Puscariu's Atlasul linguistic român (Cluj, 1938).
  • Little theoretical work has been done about the “comparative method,” which has generally been taken for granted and used without much concern about its real nature, its applications, and its limitations. Cf. Hirt, Indogermanische Grammatik, 1 (Heidelberg, 1927), p. 98: “Wenn nun auch im Laufe der Zeit die Ansichten darüber, wie das Idg. beschaffen war, in mehr als einer Beziehung gewechselt haben, und wenn die erschlossene idg. Ursprache immer nur der Ausdruck unserer jeweiligen Erkenntnis war, so hat man sich doch eigentlich über die Frage, ob und wie jene Erschliessung des Idg. eigentlich möglich sei, wenig Gedanken gemacht [italics mine!. Erst E. Hermann hat in einem anregenden Aufsatz KZ. 41 [1907]. 1 ff. die ganze Frage entschieden gefördert.” Cf. also A. Carnoy, Restitution de sons en indo-européen et en roman, Le Muséon, n.s., 13 [1912] 187 sqq., which contains many errors of fact and must be used with great caution; C. D. Buck, Lg. 2 (1926). 99 sqq.; Oertel, Lectures on the Study of Language (1902), pp. 30 sqq.; 116 sqq.; 128 sqq.; Sköld, Sprachgeographie, 35; Jespersen, Language, (1923), pp. 80 sqq.; H. Pedersen, Linguistic Science in the Nineteenth Century (tr. J. W. Spargo [Cambridge, Mass., 1931]), pp. 265 sqq.; Pisani, Ricostruzione, and Paleontologia, pp. 30 sqq.; G. Devoto, Gli antichi Italici (Florence, 1931), pp. 43 sqq.
  • The areal theory of linguistics, based on Gilliéron's Linguistic Atlas of France, was elaborated by the Italian school, and has since brought a far-reaching revolution in all linguistic science. Its fundamental principles and methods are given by Bàrtoli in his Introduzione, and he himself (following G. Campus) has extended its applications to other languages than Romance in a series of subsequent articles: for IE see especially the last volumes of the AGIt, the Annuaire de l'institut de philologie et d'histoire orientales et slaves 5 (Brussels, 1937. (=Mélanges Boisacq, 1) pp. 19 sqq.; Mélanges Belic (Belgrade, 1937), pp. 197–202; StBalt 3 (1933). 1–26; RFCI 57 (1929). 333–45; StudAlb 2 (1932). 1 sqq.; Atti del III congresso Internaz. dei linguisti (1935), 164 sqq. (with bibliography); Atti Acc. Tor. 72 (1936–7). 223 sqq. The same method was applied by Campus, Due note sulla questione delle velari arioeuropee (Turin, 1916); Atti Acc. Tor. 54 (1919), 109, n.2; Meillet, MSL 14 (1906–8). 392; 18 (1913). 13; 22 (1922). 218; 23 (1923–35). 148; BSL, 23 (1922). 107; 24 (1924). 193; 25 (1924). 10 sqq.; 29 (1929). 29 sqq.; 32 (1931). 1 sqq.; 194 sqq.; Esquisse d'une histoire de la langue latine3 (Paris, 1933), 16 sqq.; 22 sqq.; 46; 77 sqq.; Ernout-Meillet, Dictionnaire étymologique de la langue latine2 (Paris, 1939), passim; S. Lévi, Fragments de textes koutchéens (Paris, 1938), p. 38; Vendryes, MSL 20 (1918). 265 sqq.; Wagner LGRPh 39 (1918). 126 sqq.; Terracini, AeR n. s. 2 (1921). 99 sqq.; Devoto, Gli antichi Italici, pp. 48; 264; Cultura 10 (1931). 1 sqq.; I problemi del più antico vocabolario giuridico romano, Annali della R. Scuola Superiore di Pisa 2. 2 (1933). 233 sqq.; Storia della lingua di Roma (Bologna, 1940), pp. 6 sqq.; 17 sqq.; Scritti in onore di Alfredo Trombetti (Milan, 1930), pp. 375 sqq. (bibliography, p. 383); Bonfante, BSL 33 (1932). 111–2; ArchOr. 11 (1939). 84 sqq.; IF 52 (1934). 221 sqq.; 55 (1937). 131 sqq.; AnCI 7 (1938). 329 sqq.; Emerita 2 (1934) 269 sqq.; RFCI 63 (1935). 234 sqq.; IAOS. 65 (1944). 177 n. 57 (with bibl.); Nehring, Die Indogermanen- und Germanenfrage, p. 30; Arntz, Germanen und Indogermanen (Heidelberg, 1936), pp. 429–30. The discussion between Campus and Meillet is especially instructive (it eliminates, of course, Sturtevant's doctrine, Lg. 2. [19261 25 sqq.). This method has been applied by Bàrtoli to Uralic and Semitic in Scritti… Trombetti, 175 sqq. (cf. also Atti del III congresso, pp. 165–6), and to ‘pre-Colombian’ languages in Mélanges Van Ginneken (Paris, 1937), pp. 123 sqq.
  • Several of the author mentioned above do not mention Gilliéron or the Italian neo-linguists, and perhaps know nothing about them; but the fact remains that they apply the areal method to language.
  • Bàrtoli application of areal norms to I-E. languages has been criticized by H. Sköld in his Sprachgeographie und Indogermanistik (Beiträge zur allgemeinen und vergleichenden Sprachwissenschaft, Lund, 1931, 1), but the work is quite superficial. The author, although a brilliant linguist, did not make the slightest effort to understand a doctrine obviously new and unfamiliar to him; or perhaps his knowledge of Italian was deficient. I do not think it worth while, therefore, to enter into a discussion of this study. See, however, p. 72: “Brückner [KZ 43 (1910). 315] ist der richtigen Ansicht sehr nahe gekommen, wenn er behauptet, es sei natürlich, wenn (in den Wechselfällen) das Litauische und noch mehr das Slavische Velare statt Palatale Gutturale aufweisen: die Palatalisierung sei ja nur [sie!] im äussersten Osten, bei den Indoiraniern am stärkesten aufgetreten, gegen den Westen hin sei sie schwächer geworden.” This is exactly the areal theory. See also Bonfante, Dialetti, pp. 90 sqq., 22 sqq., 131 sqq., and Sköld, op. cit., pp. 14, 56 sqq. (on πῦρ and ignis, pp. 15–6).
  • The criticism of Brandenstein, Die erste “idg.” Wanderung (1936), pp. 15–16, is likewise mistaken; what he says about French champ and Ladinian tschamp is simply wrong; the two phenomena are surely connected (cf. u>ü, ct>yt, -t->-d-, -k->-g- etc.); and what he says about Lat. est: ēst, Germ, ist: iszt is equally incorrect. The rather superficial remarks of I. Iordan, Introducere in studiul limbilor romanice (lassi, 1932; Eng. tr. by J. Orr [London, 19371), have been answered excellently by Bàrtoli himself, AGIt 25 (1931–3). 37.
  • I cannot discuss here the long paper of Pisani, Geolinguistica e indoeuropeo, Mem. Acc. Lincei, 6. 9.2 (1940). He is, in general, by no means unfavorable to the application of areal linguistics to I-E, and even applies it himself (cf. espec. pp. 115; 297 sqq.; 302 sqq.; also his Paleontologia, pp. 24, 27, 43 sqq.). The discussion with Bàrtoli is found on p. 283 sqq. (on πῦρ: ignis, pp. 302 sqq.). His criticism of Bàrtoli's doctrine on the I-E aspirates (pp. 201 sqq.), on the other hand, is quite coirect, as is that of Sköld, pp. 17–31; cf. Bonfante, Emerita 2 (1934). 79–80; StudBalt 5 (1935–6). 36, note 3. Vendryes (RC 44 [1927]. 472), A. Debrunner (IF 50 [1932]. 212 sq.), Meillet and P. Chantraine also have rejected it.
  • In the Americas, the only studies I know in this field are G. Bonfante and T. A. Sebeok, Linguistics and the Age and Area Hypothesis, AJ Anth 46 (1944). 382 sqq. and Bonfante and Gelb, JAOS. 65 (1944), 169–190.
  • Cf. also U. Leo, ASNS 162 (1932–3). 224: “Sprachvergleichung also ist der methodische Grundzug der Sprachgeographie.” I should, however, invert this assertion and say: “Sprachgeographie ist der methodische Grundzug der Sprachvergleichung.”
  • Cf. also e. g. E. Prokosch, A comparative Germanie Grammar, Philadelphia (1939), p. 35: “If in a given word all, or most [Italics mine], Indo-European languages show the same sound, we assume that this represents the original condition of Indo-European.” Suffice it to remark e. g. that Latin (area isolata) preserves Indo-European *kw (in quis, quia etc.), *gw (in unguō, inguen, anguis), *k (in centum, cor etc.), * g (in genus, genū etc.), *-m, *-s, *-t, the three-phoneme system in centum: genus: haedus, pēs: dē-bilis: ferō etc. etc., all facts which are not preserved in the majority of Indo-European languages.
  • Exactly the same holds true for what is said by Leumann four lines above on the same page, about the flexion of edō: “In der üblichen Weise sind thematisch umgestaltet edō; edimus edunt” (and why not the other persons?). The author does not even feel the contradiction in the following sentence: “Sonst kommen thematische Umgestaltungen erst in der Kaiserzeit auf.” Cf. on this subject my article in Gl 22 (1933–4). 289 sqq.
  • The antiquity of the type fert, bar(∂) tū, ϕέρτε, bhárti is also proved for Greek and Indo-Aryan by the norm of the fase sparita, for bhárti and ϕέρτε are exclusively Vedic and Homeric respectively, and have beeṉ replaced in later stages of these languages by the thematic forms bhárati and ϕέρετε.
  • As the meter shows, bar∂tū (occuring only in Yasna 33.9) must be read baratü, a reading actually found in the MSS Dh 1, S 2, Bh 1, L 2; bar∂tąm, only in Hāąōxt Nask 2. 18, 36, is also unmetrical, and should probably be read bara; abar∂ (Yasht 14), which is quite metrical, is probably < *bhar|ṙ or *bhrṙ, 3d plur. perf. (cf. Bartholomæ's earlier view, in Literaturbialt für orient. Philologie 1 [18841. 18; Arische Forschungen 2 [Halle, 1886]. 51). Av. bar∂tū is therefore, to be compared with Gk. ϕερέτω rather than with Lat. fertō. I owe this information concerning Avestan bar∂tū to the kindness of Prof. L. H. Gray.
  • Leumann, p. 310, admits, however in the note: “Zuzugeben ist freilich, dasz die Erklärung von *som sum als Nachbildung von sont sunt über * somos sumus chronologisch schwierig ist: *som ist uritalisch, aber *sont wohl nicht.” I may remark that the assimilation of only one form of the singular to only one form of the plural (or vice versa) would be quite strange: it does not happen in any language, so far as I now; singular and plural seem to be distinct categories, and analogy operates first of all within them. Span, sois is formed on somos, son (both old), not on soy.
  • This observation on edō, ferō, sum holds true, of course, for uolō, eō, dō, fīō, uomõ, etc.; the *-mi ending, which does not exist in Latin, is an analogical innovation on *-si *-ti.
  • The last defender of this unfortunate idea is Pisani, AGIt 21 (1928). 118 sqq.; Gl 22 (1933–4). 295–6; RIGl 16 (1932). 71–2; 17 (1933). 67 sqq.; Mem. Acc. Linc. 6. 4. 6 (1933). 622–3; 6. 9. 2 (1940). 331.
  • Cf. Pedersen, Le groupement des dialectes indo-européens (Copenhagen, 1925), p. 10: “Or le témoignage de l'indo-iranien et du grec tend à prouver que le génitif de ces mots avait dans la langue-mère [?l indoeuropéenne une forme tout à fait différente, savoir *-o-sjo. C'est de *ekwo-sjo qu'est issu sanskr. áśva-sya, gr. ϊππoιo ‘du cheval’. […] On peut ajouter que la terminaison arménienne (išoy ‘de l'âne,’ de ēš ‘ane’) est identique à la terminaison grecque et indo-iranienne. Or le génitif en -ī de l'italique [sic] et du celtique est tout & fait incommensurable avec un prototype indo-européen en *o-sjo. Les lois phonétiques de l'italique [sic] et du celtique ne nous permettent point de dériver cet *-ī monophthongue d'autre chose que d'un -ī indo-européen.” Cf. also Vendryes, Proc. Brit. Acad. 23 (1937). 366; C. Marstrander, NTS 3 (1929). 245; C. Borgström, ibid., 7 (1934) 122; Meillet, BSL 32 (1931). 196–7; Devoto, Festschrift Hirt, 2 (Heid., 1936), p. 542; AGIt 22–3 (1929). 221; Pedersen, Vergi, kelt. Gramm., 2, p. 84; Schwyzer, Griech, Grammatik, p. 555, n. 4; von Kienle, WuS 17 (1936). 109 sqq.; 122 sqq.; Schrijnen, Collectan, Schrijnen, pp. 57 sqq.; Sommer, Handbuch, 2–3 pp. 340 f.
  • Sanskrit was considered an “archaic” language and was put in a position of preeminence for the reconstruction of older forms. Of recent years, Indo-European linguistics has sought to rid itself of this prejudice, which still haunts many linguists and many books. Hirt writes with good reason Indogermanische Grammatik, 1 (Heidelberg, 1927). 98–9; “Wenn der französische Forscher Pictet für die Wortforschung den Satz aufstellte: partir toujours du mot sanscrit, so galt das nicht minder auf den andern Gebieten. Im Vokalismus und Konsonantismus, in Stammbildungsund Flexionslehre galt das Altindische als die altertümlichste idg. Sprache, und es hat viele Mühe gekostet, diese Ansicht zu beseitigen. Man kann wohl sagen, dass seit Brugmanns Zeiten vom Jahre 1876 an der Weg der Wissenschaft dahin gegangen ist, dieses Vorurteil von der höchsten Altertümlichkeit des Indischen zu beseitigen, […] Wenn also das Indische allein kein unverdächtiger Zeuge war, so schien allerdings jede Schwierigkeit beseitigt, wenn noch irgendeine andere Sprache, insbesondere das Griechische, mit den Indischen übereinstimmte. In der tat, wenn die beiden aus ältester Zeit überlieferten, Sprachen, Griechisch und Indisch, gleiches zeigten, wie konnte man da an dem hohem idg. Alter zweifeln? A ber auch das 1st ein Trugschluss. Ich erinnere nur daran, dasz Griechisc;h und Indisch in der Setzung der Reduplikation im Perfect wöllig übereinstimmen, und dass diese Regelung trotzdem nich indogermanisch ist. Ebenso findet sich in beiden Sprachen das Augment, ja es kommt hier sogar das Phrygisch-Armenische hinzu, und trotzdem scheint es mir so gut wie sicher, dass das Augment keine allgemein verbreitete idg. Eigentümlichkeit war. Man vergisst nämlich bei diesen Erscheinungen […] dass es innerhalb des Indogermanischen dialektische Verschiedenheiten gegeben haben kann, ja gegeben haben muss.” Cf. also Idg. Gramm., 3 (1927). 52; IF 17 (1904–5). 278 sq.
  • Cf. also Meillet, Introduction7, p. 47: “La comparaison de l'indoiranien et du grec ne révèle pas tout de l'indo-européen. Par exemple on a pu croire longtemps que la désinence en *-r &agrave valeur passive [sic] était une propriété de l'italo-celtique [sic]; la découverte du hittite et du tokharien [add at least Phrygian] en a montré le caractère indoeuropéen.”; and see Bàrtoli, AGIt 25 (1931–3). 22: “I fondatori della linguistica ario-europea […] guardavano, e non fa meraviglia per quei tempi, con gli ‘occhiali indiani’ [cf. Hugo Schuchardt-Brevier2 (Halle, 1928), p. 226], cioè credevano, ed era naturale credessero, che la ‘nobilissima’ lingua degl'Indiani antichissimi fosse, se non la lingua madre, almeno la sorella maggiore delle lingue ario-europee.”
  • See also Hirt, Handbuch des U r germanischen, 3 (Heidelberg, 1934), p. 138: “Das Lateinische und Germanische kennen für Konjunktiv und Optatif nur einen Modus, der im Germanischen auf den Optativ zurückgeht, und im Slawischen fehlt selbst dieser. Es ist mir zweifelhaft, ob der Konjunktiv des Griechiischen] Ind[ischen] auf dem ganzen Gebiet des Idg. verbreitet und vor allem, ob er auch einst im Germanischen vorhanden gewesen ist. Sicher überkommen aus dem Idg. sind im Gerni-[anischen] nur Indikativ, Imperativ, Imperativ und Optativ, den man auch Konjunktiv nennt, was aber zu Irrtümern führt.”
  • And Meillet, BSL 23 (1922). 70: “On s'est trop représenté l'indo-européen d'après l'image indo-iranienne pour qu'une réaction ne soit pas utile”: also BLS 31 (1931). 15; Esq. 3, p. 24 and Bàrtoli, AGIt 27 (1935). 11; Studi albanesi 2 (1932). 51 (n. 4); 726 (n. 136); RFCI 56 (1928). 424–5; Juret, Proceeding of the 3d Congr. of Phonetic Sciences, Ghent (1938), 311.
  • We can say too, in a certain sense, that the augment was “I-E”, but dialectal (that is, always limited to one section of the I-E area); whereas, when the older linguists said that a fact was “I-E”, they meant that it existed in the whole “I-E language”—which is as much as to say that all I-E languages once possessed it.
  • All scholars now admit, I believe, that the augment is a late innovation: see Bonfante, Dialetti, p. 130, and JAOS 62 (1924). 104, n. 7; 65 (1944). 185 n. 79; Bàrtoli, AGIt passim; Mélanges Boisacq, 1. 23; RFCl 57 (1929). 336; 339; Pisani, AGIt 21 (1927) sezione Goidànich, 12; Schwyzer, Grammatik, pp. 56, 652, 841; Hirt, Indogermanische Grammatik, 1. 99, § 89 (quoted here in note 23); 4 (1928), p. 171; Schrijnen, Collectanea Schrijnen, p. 84; Meillet, Esq. 3, p. 30.
  • There is one very important point on which I would like to correct, or, rather, somewhat amplify, Bàrtoli's theory of the area più isolata, which he correctly defines as follows (Introduzione, p. 3): “Si dirà più isolato il territorio meno esposto alle comunicazioni.” Now, of course, we know little or nothing about the “communications” in the epoch of the I-E unity, so that I propose to apply this norm in the following way: when we find that a given area (e. g., Latin) demonstrably presents, in, say, 50 cases out of 51 an older form or sound, we may call it an isolated area; and we may consider that in other cases this areal norm definitely speaks in favor of the antiquity of the Latin form as against the antiquity of the Greek, Sanskrit, Slavic, etc., form. It is the usual reasoning by analogy, on which all science is based. Bàrtoli approached this concept of mine in Mélanges Boisacq, 1. 24 (where, however the word “norm” for one specific coincidence between Greek, Sanskrit, and Latin seems to me quite out of place).
  • To Bàrtoli's norms I should also like to add the “aire disloquée et fragmentée” of Dauzat Géographie linguistique, p. 38 (cf. also Pisani, Mem. Acc. Line. 6. 9. 2 [1940]. 297), which is, in certain cases, different from the norm of the lateral areas. Suppose there is a distribution of this kind:
  • B B B B B
  • B A B A B
  • B B B B B
  • or the like. It is clear that if we apply the norm of the lateral areas to a certain section of this area, as we would be allowed to do according to Bàrtoli's method (e. g., the area delimitated by me here with a black line), we would reach the conclusion that B is older than A, which is obviously wrong. But if we consider the fact that the two, three, or more points of A are separated from each other, whereas all the points of B are contiguous, we reach the conclusion that A is older than B; for the most likely hypothesis is that B has expanded in many directions, cutting into two (or three or four, etc.) a formerly united area of A.
  • Another norm to be added to those of Bàrtoli's is, in my opinion, the following. If we glance at the Atlas linguistique de la France (Paris, 1902–10) (e. g., at the maps reproduced by Jaberg, Sprachgeographie) or at the Sprach und Sachadas Italiens und der Südschweiz (Zofingen, 1928 sqq.) or at any other linguistic atlas, we shall frequently notice that, in the center of a territory of A, an innovation B is now spreading, which has not yet cut the area of A to pieces, but is pushing it outwards with its centrifugal movement, something like this:
  • A A A A A
  • A B B B A
  • A B B B A
  • A A A A A
  • Both areas are continuous or unbroken, but A is obviously older than B, because A is peripherical to B. It is easy to conceive how an innovation B can have covered an older layer of A; it is very difficult to conceive, (unless some extraordinarily favorable circumstances should occur, viz., unless B be an isolated area) how A should have expanded over an area of B. For the term peripheric cf. Pedersen, Groupement, p. 20; Meillet, Esquisse 3, 18; 20, and BSL 32 (1931). 5 sqq.; 196; Pisani, Mem. Accad. Line. 6. 9. 2 (1940). 280–1 (with bibliography), although they employ the word in a different way.
  • Pokorny, Berichte des Forschungs-lnst. f. O. u. Or., 3 (1923), 25 ff. (bibi); Terracini, AeR 2 (1921). 112. Even before the discovery of Tocharian, P. Kretschmer (Einleitung in die Geschichte per griechischen Sprache [Göttingen, 18961, 142–3) tried to explain through old prehistoric migrations the obvious affinities between Indo-Aryan and Iranian on the hand, and Celtic and “Italic” on the other.
  • J. Charpentier, Die verbalen r-Endungen der indogermanischen Sprachen, p. 11 (Upsala, 1917 = Skrifter utgifna af K. Humanistiko Vetenskaps-Samfundet i Uppsala 18. 4), and ZDMG 71 (1917). 378 sqq. (similarly H. Junker, LGRPh 36[1915] 69). On the “present” medial AeR n. s., 2 (1921). 109 sqq.; Devoto, Italici, p. 48; AGIt 22–3 (1929). 226 sq.; Meillet, MSL 18 (1913). 13; BSL 24 (1924). 189 sqq.; 27 (1927). c.-r., p. 38; 29 (1928–9). c.-r., pp. 63; 66; 32 (1931). 3 sqq. (“l'hypothèse la plus plausible est donc que ces formes en -r seraient des archaïsmes conservés seulement dans quelques langues indo-européennes”); Esq. 3, pp. 22 sqq.; A. W. M. Odé, De uitgangen met r van het deponens en het passivum in de indoeuropeesche talen (Haarlem, 1924); Leumann, Grammatik, pp. 306–7 (bibliography); Pisani, Mem. Acc. Line. 6. 4. 6 (1933). 566 sqq.; A. Götze, Kulturgeschichte Kleinasiens (Munich, 1933), p. 56; Bonfante, JAOS 65 (1944). 188 sqq.
  • The opposite opinion (the -r- endings are an “Italo-Celtic” innovation) can still be found in Brugmann, BPhW 37 (1917). 1525; 1527.
  • The -r-ending of the 3d plur. perf. is probably different in origin (cf. Meillet, BSL 29 (1928–9), c.-r., 63), but presents, at all events, a similar picture: (1) lateral areas (Lat. fēcēre, OIr. rergatar + Toch. A weñār, B weñāre, Vedic āsúr, čakriré, duduhré, áduhra, Av. ånhar∂, čāxrare; (2) isolated area (Latin, also Celtic). Hittite is once more an island of a lost continent (seker, eter, eper, eser, etc.), as in the case of the preserved velars and labio velars and of the medial “present” endings (see above).
  • Another example is the feminine type with *-sor-, *-sr- in the numerals “3” and “4”, preserved only in the two lateral areas, Celtic and Indo-Iranian: OIr. theoir, cetheoir, Welsh teir, pedeir + Avestan tišrō, čataŋro, Vedic tisr´s, čátasras; cf. Meillet, BSL. 32 (1931). 8; Esquisse 3, pp. 19 sqq.; Vendryes, Proc. Brit. Acad. 23 (1937). 347 sq.; a proof, were any needed, against Sturtevant, that the distinction of gender in I-E is very old, and that its loss is late (but see also Sturtevant himself Lg 2 [19261. 30 sqq.).
  • Oscan deketasiúí (dative) is correctly derived from *dekmtó- > *decentó- by J. Whatmough, Lg 3 (1927). 107 (cf. Umbrian sestentasiaru). The criticism of R. G. Kent, ibid., p. 186, presents a petitio principii: “But there is no warrant in Italic [sic] for the -to-suffix in ‘tenth,’ all the recorded forms being referable to dekemo- [sic]; in fact, no Italic numeral from ‘seventh’ to ‘tenth’ inclusive has -t-suffix.” Kent seems to have overlooked the work of A. Walde Ueber älteste sprachliche Beziehungen zwischen Kelter und Italikern, Innsbruck, 1917 (later corroborated by Devoto, Pisani, Terracini, Bàrtoli and myself in a series of papers and books), who has shown definitively, in my opinion, that Latin and Osco-Umbrian do not go back to a common “Italic” branch.
  • As a matter of facts, a series like Lat. prīmus, secundus, tertius, quartus, quīntus, sextus, septimus, octāuŭs, nōnus, decimus is obviously anomalous, and, therefore, older than e. g. German erste, zweite, dritte, vierte, fünfte, sechste, siebente, achte, neunte, zehnte or Engl, fourth, fifth, sixth etc. The Greek series represents an intermediary stage, the old Latin type being already shattered, but not so completely as in Germanic.
  • Ehrlich, Sprachgeschichte, loc. cit., observed that Latin shows a break between “sixth” and “seventh” (quartus, quīntus, sextus: septimus, octāuŭ;s, nō;nus, decimus); this is doubtless a survival of an ancient I-E duodecimal system, later replaced everywhere by the (more recent) decimal system. Cf. also Eng. eleven, twelve: thirteen, fourteen, etc.; Goth. saihs tigjus: sibuntehund, ahtautehund; Gk. πεντήκoντα, έξήκoντα: έδδoμήκoντα, ὀγδoήκoντα, etc.; OIr. ses-ca: sechtmo-ga, ochtmo-ga. In most Germanic languages, *hund means “120,” not “100”; cf. Brugmann, Gdr. 2 2. 2 pp. 4–5; 26 sqq.; W. Streitberg, Gotisches Elementarbuch 5–6 (Heidelberg, 1920), p. 137; Feist, Wörterbuch 3, s. vv. *ainlif, hunda, sibun-tehund, fidwor-tigjus; KIuge-Götze, Etymologisches Wörterbuch der deutschen Sprache 11 (Berlin and Leipzig, 1934), s. v. Grosshundert; Schrader-Nehring, Reallexikon der indogermanischen Altertumskunde (Berlin and Leipzig, 1917–29), S. V. Zahlen; G. Ipsen, IF 46 (1928). 76; Schwyzer, Griech. Grammatik, pp. 54; 592, n. 4; H. Jacobsohn, KZ 54 (1926). 76 sqq.; Pedersen, Linguistic Science, pp. 3(22 sqq.; Pisani, Re. Acc. Line. 6. 8 (1932). 148 sqq., with bibliog., and Mem. Acc. Line. 6.4.6 (1933). 566 sqq. We can apply to the duodecimal and decimal system in I-E Bàrtoli's norm of the fase sparita (the duodecimal system is older).
  • “Le fait que le type décimas se trouve seulement aux extrémités du domaine indo-européen, en indo-iranien et en italo-celtique, et que l'autre type est celui de la partie centrale suggère l'hypothèse que seul le type decimus aurait été indo-européen commun et qu'il aurait subsisté à deux bouts du domaine, tandis que le type δέκατoς s'y serait substitué dans la partie centrale. […] De nombreux mots de caractère religieux ou politique se rencontrent seulement dans les deux groupes (indo-iranien et italo-celtique): le forme de l'ordinal “dixième” fait partie de ce vocabulaire savant, surtout politique et religieux, qui s'est maintenu seulement in indo-iranien et en italo-celtique et qui a disparu ailleurs; decimus est ainsi à mettre dans le même groupe que lat. iēx, lēx, crē;dō, interdīcō. etc. […] Plusieurs formes du type de lat. decimus sont surement les formes indo-européennes.” Cf. also pp. 31–2: “C'est gr. ὄγδooς qui indique la forme ancienne de 'huitième (à part peut-être le vocalisme -o- intérieur). Mais, isolé au milieu de formes toutes différentes, cet ordinal a disparu partout ailleurs. Seul, le latin, si souvent conservateur d'antiquités singulières [italics mine!, en a gardé une trace fortement altérée.” I cannot follow him, however, when he writes, on p. 29: “En effet, il n'y a pas de particularités phoniques ou morphologiques speciales à l'indo-iranien et à l'italo-celtique”; he seems to have changed his opinion, at all events, in BSL 32 (1931). 3 sqq.; cf. also M. Lejeune, BSL 29 (1928–9). 109 sqq.; 112 sqq.; 116 (p. 113, the norm of the lateral areas finds an easy application to Gk. τρίτoς).
  • Sturtevant's views on the I-E velars (Lg 2 [19261. 25 sqq.), entirely contrary to the results of areal linguistics, have been criticized and rejected even by Ribezzo, AGIt 22–3 (1929). 142 sqq., a scholar who is, in general, a stranger to the geographic conceptions. Sturtevant was apparently unaware of the fundamental works of Campus (see § 18 at the end).
  • Cf. Pisani, Mem. Acc. Line. 6. 4. 6 (1933). 552: “Suono gutturale più o meno intaccato da palatalizzazione”; cf. ibid., p. 562, n. 2. Among recent authors, e. g., Nehring (in Indogermanen- und Germanenfrage, p. 12) and A. Götze (JAOS 64 [1944], 85) write k ǵ; W. Brandenstein (Germanenfrage, p. 232); Biugmann (Gdr. 2 1, p. 542 sqq.), Leumann (Grammatik, pp. 122–3), Sturtevant Lg 20 (1944). 210 and Pisani Mem. Acc. Line. 6. 4. 6 (1933). 552; 6. 9. 2 (1940) 263, 288, 290sqq.) use other symbols. All (except Sturtevant) term them palatals, although sometimes, as in the case of Leumann, with a certain amount of inconsistency.
  • The problem has been stated in its real terms by Bàrtoli (Introduzione, p. 50), who stresses its theoretical implications. The existence of only two velar series (*k, *g, *gh, and *kw, *gw, *guh) was long defended by Meillet (MSL 8 [1893–4]. 277), Hermann (KZ 41 [1907]. 32 sqq.; 46), and myself (Dialetti, pp. 93–4; 122 sqq.; 131 sqq.; 141 sqq.).
  • Here, again, the epigone champion of this dead theory is Pisani, Mem. Acc. Line. 6. 9. 2 (1940). 292–3.
  • On linguistic palaeontology, apart from the classical works of F. G. Eichhoff, A. Kuhn, P. von Bradke, K. Penka, M. Much, S. Feist, V. Hehn, O. Schrader, E. de, MicheIis, H. Hirt, A. Nehring, see now Pisani. Paleon-tologia, with recent bibliography; cf. especially Brandenstein, Wanderung; W. Schulz, KZ 62 (1935). 184–98; F. Specht, Ibid., 66 (1939). 1 sqq.: Nehring, Actes du II⋅ congrès international de linguistes, pp. 191 sqq.; and the volume Die Indogermanen- und Germanenfrage (Sabzburg-Leip-zig, 1936). The areal method was applied with excellent results to linguistic palaeontology by Terracini, AeR n. s. 2. (1921). 105, n. 2; 107; Meillet, MSL 14 (1906–8). 392; 22 (1922). 218; 23 (1935). 148; BSL 25 (1924). 104; 29 (1928–9). 29 sqq.; Esquisse 3, pp. 21 sqq.; 77 sqq.; J. Vendryes, MSL 20 (1918). 265 sqq.; Ernout-Meillet, Dictionnaire 2, s. vv. castus, crēdō, ignis, interdīcō, lēx, pūrus, rēx, etc.; Devoto, Scritti… Trombetti, 375 sqq. (bibliography p. 3,83); Bonfante, Emerita 2 (1934). 269 sqq.; ArchOR 11 (1939). 84. sqq.; Pisani, Mem. Acc. Linc. 6. 9. 2 (1940). 303 sqq.; GSAI 3 (1934–5). 363–4; Paleontologia, pp. 43 sqq.; Nehring, in Germanenfrage, p. 30.
  • Cf. also Nehring, Germanen frage, p. 233: “In meiner schon erwähnten Arbeit [Wanderung, pp. 1; 22 sqq.] habe ich nun auf ein Mittel hingewiesen, chronologische Bestimmungen innerhalb der Masse der Tatsachen vorzunehmen, so dass es möglich war, sie zeitlich zu ordnen. Die Bedeutungsgeschichte gewisser Wörter erlaubt uns nämlich mit zwingender Sicherheit [italics mine] zu entscheiden, was früher und was später ist, und zwar einfach deswegen, weil die Bedeutungsentwicklung in manchen Fällen nicht umkehrbar ist. Ein Beispiel! Latt. pecunia ‘Vermögen’, ‘Geld’ hatte ursprünglich die Bedeutung ‘Viehbestand’; diese Wort- und Sachentwicklung ist ebenso bezeichnend wie begreiflich; hingegen ist es unmöglich, dasz wir bei einem Rückfall der gesellschaftlichen Zustände von ‘pekuniären’ oder ‘geldlichen’ Sorgen reden und dabei unsere Sorgen um die Viehherde meinen sollten! Eine solche Entwicklung ist deswegen undenkbar, weil wohl eine sachliche Verbindung vom Vieh zum Geld besteht, d. h. Vieh als Zahlungsmittel fungieren kann, weil aber unmöglich das Verschwinden des Geldwesens dazu führen könnte, dasz das Wort Geld für Vieh angelegenheiten verwendet wird: einfach deswegen nicht, weil auch in hochentwickelten Zeiten primitive Zustände mitlaufen, für die infolgedessen auch die Bezeichnungen da sind, und die, wenn sie wiederum alleinherrschend werden, keine eigene sprachschöpferische bzw. sprachwandelnde Tätigkeit erfordern.”
  • Other possible methods, based, e. g., on the influence of climate (G. Sarrazin, BB 16 [1890]. 320; G. Schütte, IF 15 (1903–4). 211 sqq., espec. pp. 274sqq.; cf. Jespersen, Language, p. 265; Oertel, Lectures, pp. 191 sqq.; F. N. Finck, Die Aufgabe und Gliederung der Sprachwissenschaft [Halle, 1905], p. 53; Pisani, Mem. Acc. Line. 6. 9. 2 [1940]. 249), of race and heredity (Pisani, loc. cit.; van Ginneken, Raas en taal [Amsterdam, 1935]; Proceedings of the International Congress of Phonetic Science = ANPh 8–9 [1933]. 1. 13 sqq.; 76 sqq.; 2 (London, 1936). 65 sqq.; 77 sqq.; 91; IPs 30 (1933). 266 sqq.; IF 45 [1927]. 1 sqq.; Donum Natalicium Schrijnen [Nijmegen and Utrecht, 1929], pp. 10 sqq.; cf. Sommerfeit, NTS 4 [1930]. 99 sqq.; Grammont, Traité de phonétique [Paris 1933] pp. 175–6), or of cultural psychology (L. Spitzer; K. Voseler; Finck, Der deutsche Sprachbau [Marburg, 1899]; Pisani, op. cit., pp. 233 sqq.; 249 sqq.; Jespersen. op. cit., p. 258; Brøndal, Le Français langue abstraite; E. Lerch) are thus far insufficiently proved to give a solid foundation for linguistic reconstruction.
  • I may add that, according to a French school of thought, “le développement linguistique d'un groupe est dominé par des tendances spéciales” (Sommerfelt, NTS 4 [19301. 92; cf. espec. Vendryes, Mélanges Meillet, pp. 115 sqq.; and the works of Grammont, e. g., Traité, p. 156). It is obvious that the discovery of some stable tendencies of this kind in some language or area might provide us with a valuable instrument for reconstruction of past phases of language. Likewise, we might perhaps observe, with the help of the “Prague school,” that a given system B (morphological, pkonemic, etc.) can develop only from a system A, not from a system C, or D; and can thus reach some conclusions about the “system” of lost languages; cf. espec. Trubetzkoy, La Phonologie actuelle, JPS 30. (1933). 245: “Tout en étant jusqu'à un certain point déterminé par les lois de structure générale—qui excluent certaines combinaisons et en favorisent d'autres—l'évolution du système phonologique est à chaque moment donné dirigée per la tendance vers un but [italics the author's]. Sans admettre cet élément téléologique, il est impossible d'expliquer l'évolution phonologique.” Cf. also van Wijk, Klankhistorie en phonologie, MAWA n. r. 1 3 [1938]. 7–8. One may hope that this fertile thought will be developed further. Cf. now also Hoenigswald, Studies in Linguistics 2 (1944). 78 sqq.
  • The names of scholars who follow the different “methods” are mere indications to help the reader; they do not in the least pretend to be complete enumerations. Moreover, I mention only the authors who gave a more or less clear theoretical definition of every method, not those who merely used it empirically whithout analyzing their own operations
  • This idea, although obvious, is by no means general, since, on the whole, the theoretical ideas of the majority of linguists on this subject are rather hazy. It is doubtless one of the great merits of the Italian neolinguists to have pointed out that several methods not only can, but must be used concurrently, whenever possible—and it is frequently possible to do so.
  • Thus, for example, if we compared Lat. generis with Gr. γένεoς and Skt. jánasas fropi the point of view of areal linguistics alone, we should arrive (on the correct assumption that Latin is an area isolata) at the conclusion that the -r- of Lat. generis is older than the -s- of Skt. jánasas. This obviously wrong conclusion could be easily corrected, even if we ignored the documents of Old Latin and the grammarians (both proving, beyond any possible doubt, that Old Latin had -s-, not -r-, between vowels), by method 7 (usual phonemic change), which indicates that the passage -s- > -r- between vowels is much more frequent than the reverse; then by method 5 (internal reconstruction) the same conclusion would be reached on the basis of a proportion like caput: capitis—agmen: agminis—laudator: laudātõris—genus: generis, where generis is shown to have replaced * genesis; and, similarly, in the cases of es-se: laudā-re; es-sem: laudā-rem; es-se: er-ō. As a matter of fact, this development is of Umbrian origin (cf. Devoto, Antichi Italici, p. 169; Terracini, AeR, n. s. 2 [1921]. 111 sqq.).
  • Likewise, it is easy to show, with the help of method 6 (the anomalous form), that the Greek type ή ιππoς (fem.) is older than the Latin type equa, against Bàrtoli, Mélanges Boisacq, 1, pp. 19 sqq., and Nehring, Actes du- II⋅ congrès international des linguistes, pp. 192 sqq. (with the discussion), and against the indications of areal linguistics. Wacker-nagel, Vorlesungen2 2. p. 24; Meillet, BSL 17 (1911). p. LXIV; 32 (1931). 26–7; Ling, historique et ling, génér., l2, p. 212; Introduction 7, p. 282; Ernout-Meillet, Dictionnaire 2, p. 307; Leumann, Grammatik, pp. 203 sqq.; 364 sqq.; E. Boisacq, Dictionnaire étymologique de la langue grecque 3 (Paris, 1938), p. 380, n. 2, hold equa to be a late (Latin) innovation, while A. Walde, Lateinisches etymologisches Wörterbuch2 (Heidelberg, 1910), S. V. and H. Lommel, Studien uber indogermanische Femininbildungen (Göttingen, 1921), pp. 25, 30–1, are inclined to think it is rather old; Feist, Wörterbuch 3, p. 21; Walde-Pokorny, Etymologisches Wörterbuch der indogermanischen Sprachen (Berlin and Leipzig, 1927–32), 1, p. 112, and Walde-Hofmann, S. V. equus, express no opinion. Cf. also Schwyzer. Griech. Grammatik, p. 457.
  • Another case of the same kind is that pointed out by Pisani Mem. Acc. Line. 6. 9. 2 (1940). 287, n. 1, at the end: Sard, pilu, filu, Ital. (that is, Tuscan) pelo, filo, from Lat. pǐlum, fīlum. The criticism of Pisani is quite wrong; in this case, with the help both of method 3 (areal linguistics) and 1 (impossibility of spontaneous scission), it would be possible, I think, to reconstruct something very near the Latin situation. Namely, Sardinia (an area isolata) would indicate that both sounds (the e of pelo and the i of filo) once had the color i; but Italian would show that there was a difference between them. Now, since this difference could not be in color (cf. Sardinia), it must have been based on something else: on pitch, or stress, or quantity. This, in my opinion, is precisely one of the best examples of the good results which can be reached by the combined operation of two (or more) methods.
  • I had already received the proofs of this article (presented several months before, as can easily be testified by all the editors of Word) when I happened to see the article of Hoenigswald Internal Reconstruction, published in Studies in Linguistics 2, no. 4 (1944). 78 ff. (December). Dr. Hoenigswald reaches to a very limited extent the same conclusions as I do. His assertions are, as far as I can see, nowhere in contradiction with mine; on the contrary they confirm them. I have therefore nothing to change in my paper. I might add that I had already presented an abstract of my main points at the meeting of the Linguistic Society of America, July 1944, in Madison, Wisconsin.
  • In a letter to the Editor of Classical Philology 39 (1944). 187, where he attacked certain assertions of mine. Prof. Sturtevant wrote among other things: “Now, comparative grammar can demonstrate the former existence of a phoneme (formerly called a “speech sound') only in the parent-speech of two or more related languages, as this speech was at the time of their separation; theories about the earlier history of a parent-language are incapable of confirmation by the comparative method. Consequently, there was no possibility of establishing Saussure's hypothesis until we should obtain some linguistic evidence, either documented or reconstructed, that should be capable of comparison with our reconstructed Proto-??.” I hope the present article shows clearly why I do not agree with this statement of Sturtevant's.
  • The other criticisms of Sturtevant's contained in the same letter I have, as I believe, sufficiently answered in the same journal, 40 (1945). 116 sqq.

Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Order Reprints Request Corporate Permissions

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

Request Academic Permissions

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.

Related research

People also read lists articles that other readers of this article have read.

Recommended articles lists articles that we recommend and is powered by our AI driven recommendation engine.

Cited by lists all citing articles based on Crossref citations.
Articles with the Crossref icon will open in a new tab.