Published online by Cambridge University Press: 22 April 2026
1 In Lang. 22.273 n. 6, Hall writes of me: ‘The first of the items cited in fn. 5 (AA NS. 46.382 ff.) is a programmatic exposé of some of Bàrtoli's theories .... The other items contain, in general, simple assertions of the superiority of Bàrtoli's method, unsupported by other evidence than lists of “authorities” alleged to share the same points of view.‘ It seems that I have filled many pages with simple assertions, in a very Ciceronian way. For the rest, I leave it to my colleagues to decide whether Hall judges my contributions fairly.
2 I do not give a complete bibliography of the publications in which I have treated of neolinguistics, because Hall has done so in Lang. 22.273 n. 5. I shall add, however, the following items, which Hall seems to have overlooked: I dialetti indoeuropei (Naples, 1931); JAOS 66.264 f. (1946); AJP 67.298 ff. (1946); RR 37.247 ff. (1946).
3 Cf. Bàrtoli, Studi albanesi 2.61 n. 56 (1932): ‘Le norme neolinguistiche sono nient'altro che un mezzo per trovare il rapporto cronologico fradue fasi linguistiche (§5) e inoltre un mezzo per trovare la patria dell’ innovazione (§6).'
4 The neolinguists have of course not invented all their principles. Nobody ever invents anything ex nihilo, and all the great discoveries (such as Darwinism or the atomic theory) were several times anticipated. The neolinguists have repeatedly made greatful acknowledgments to the scholars who preceded them—above all Leonardo, Vico, Humboldt, J. Schmidt, H. Schuchardt, G. Paris, Àscoli, Croce, Gentile. Cf. Bàrtoli, Saggi 72 n. 5, 177 n. 14, 275 n. 85; Introduzione alla neolinguistica, Prefazione and 100; Studi albanesi 2.16 ff., 57 n. 51, 62 n. 62, 138, 73. It would however be unfair to deny not only the main new points offered by the neolinguists, but also their great merit in having formed a coherent and logical doctrine out of many scattered, haphazard, and often imprecise observations.
5 It would in fact be easy to show that for many of the so-called phonetic laws affecting well-known fields like Romance, dogmatically presented by the neogrammarians in their manuals, the ‘exceptions’ are more numerous than the ‘normal’ forms. This is recognized by Hall, who however makes the astonishing assertion that ‘scientific principles do not in any way depend on the mere number of attestations’. One wonders on what they do depend.
6 Because of their denial of ‘laws’, the neolinguists are sometimes accused by timid souls of introducing disorder and anarchy into linguistics; but such remarks only show ignorance. In reality, the neolinguists wish to introduce a real and much more rigorous order in place of the false one; cf. the excellent remark of Maver (Slavia 7.147) in his review of Bàrtoli's Introduzione: ‘Il est évident qu'il [scil. Bàrtoli] ne veut pas introduire une anarchie dans la conception des changements linguistiques. Au contraire, il veut nous en donner une notion plus disciplinée et plus riche.‘
6a See also H. Schuchardt, Über die Lautgesetze, Gegen die Junggrammatiker (Berlin, 1885; reprinted in the Hugo Schuchardt-Brevier 143f., 251 f.); G. E. Parodi, Questioni teoriche: le leggi fonetiche, Nuovi studj medievali 1.263 f. (1923/4; but the article was written in 1909); C. Battisti, Ernesto Giacomo Parodi e le leggi fonetiche, La Parola, November 1925; G. Bertoni, La legge fonetica, Archivum romanicum 5.1 ff. (1921); Enciclopedia italiana s.v. Fonetica 623f. (with good bibliography and the main statements of the founders of the neogrammatical school); C. Tagliavini, Enciclopedia italiana 212 s.v. Linguistica. One of the first to oppose the neogrammatical dogma of phonetic law was the Italian linguist G. I. Àscoli, Dei neogrammatici: lettera al professore Pietro Merlo, Archivio glottologico italiano 10.18–73 (1886/8; the letter is dated 1885); Terracini, Perfiles de Lingüistas 16 (Tucumán, 1946).
7 Hall has obviously not read this important work, though he cites it once (Lang. 22.273 n. 1). He never even mentions Bàrtoli's fundamental article in Studi albanesi 2.5 ff. (1932), Accordi antichi fra l'albanese e le lingue sorelle, with the subtitle: Raccolta di fatti e discussioni di metodi. I shall not cite the many later works of Bàrtoli, because they contain nothing very new or important from the theoretical point of view, although they are very rich in the analysis of particular linguistic phenomena, both in the Indo-European and in the Romance field. I will, however, mention at least the following: Alle porte orientali d'Italia (Turin, 1945); Saggi di linguistica spaziale (Turin, 1945); La Venezia Giulia, terra d'Italia (Venice, 1945); Linguistica spaziale, in R. Biasutti, Le razze e i populi della terra 320 ff. (Turin, 1941); La posizione del dialetto nizzardo, Rivista ingauna e intemelia 147–200 (Bordighera, 1941); Lineamenti di linguistica spaziale (with Vidossi; Milan, 1943).
7a 'Physiology is sometimes the metaphysics of our discipline,' said Àscoli (cf. Introduzione 104). Bàrtoli adds: 'it has never taught us either the age, or the home (patria), or the cause of linguistic innovations.'
8 It is true that the neogrammarians distinguish verbally between phonetic phenomena, which they consider purely physiological and mechanical, and morphological phenomena. In the latter they admitted the action of analogy, which they consider psychological; cf. Iordan, An introduction to Romance linguistics 17 ff., 25 ff., and Bàrtoli, Introduzione 96 f. However, this neogrammatical analogy operates quite mechanically, in the same way in every place, country, and language; there is nothing really psychological about it. Accordingly, in spite of what its defenders may say, the neogrammatical conception of language remains, as a whole, strictly materialistic and deterministic. This note may serve to call the reader's attention to the fact that whatever contradictions may appear in the text are the neogrammarians' fault, not mine.
9 Thus it can probably be said (oversimplifying things, of course) that French is Latin + Germanic (Frankish), Spanish is Latin + Arabic, Italian is Latin + Greek and Osco-Umbrian, Romanian is Latin + Slavic, Czech is Slavic + German, Bulgarian is Slavic + Greek, Russian is Slavic + Finno-Ugric, and so on. Gröber was therefore in a sense right in including Albanian among the Romance languages in his Grundriss der romanischen Philologie2, Vol. 1 (1904–6).
10 Cf. also Goidànich, Archivio glottologico italiano 18.364 n. 4 (1914): ‘è un vanto della scuola italiana rendersi ragione delle cause delle alterazioni fonetiche, anche di quelle delle normali.‘
11 But Bàrtoli, altho an idealist, does not despise or disregard any useful ‘positivistic’ method of research, not even statistics. Cf. Studi albanesi 2.43: ‘Ma la storia della lingua è, per questo e per altri rispetti, piú chiara che quella dei costumi, e permette vere e proprie dimostrazioni statistiche.‘ Hall does not like this either; cf. n. 25.
12 The neolinguists (including myself) are frequently called daring by the modern neogrammarians—an accusation which does not make us blush in the least. I could well answer with Bàrtoli (Studi albanesi 2.67 n. 88): ‘anche negli studi linguistici la mia imprudenza mi piace piú della vostra imprudenza.‘
12a The whole movement of linguistic geography (including of course areal linguistics) goes back ultimately to Àscoli; cf. Iordan, Introduction 144 f.: ‘. . . the scientific study of dialect had already been inaugurated, though without any theoretical manifesto, by the great Italian linguist Àscoli, whom Jaberg and Jud, the most authoritative representatives of linguistic geography at the present day, consider as the most distinguished among early workers in the vernacular field.‘ It is from Àscoli, through Paul Meyer, that Gilliéron seems to have received the idea for his great Atlas linguistique de la France (cf. Iordan 145 n. 2).
13 Cf. also Meillet, BSL 22.72 (1921) and 24.80 (1923): ‘Le romanisme est le domaine qui se prête le mieux à illustrer les développements linguistiques et celui où les méthodes qui conviennent à l'histoire des langues se laissent le mieux discuter.‘ And see Bàrtoli, Introduzione 100; Saggi 71 n. 3.
14 Now we should say six, including Phrygian, Lycian, and Albanian; but the theoretical validity of the example remains the same. The position criticized in the text was taken by Meillet, Dialectes indo-européens 87; the neolinguistic point of view was presented by me in Dialetti indoeuropei 145 f., and again (at greater length) in Rivista indo-greco-italica 16.165 f. (1932); Word 1.90 n. 5 (1945); JAOS 64.187 f. n. 78 (1944); and also by Bàrtoli in Studi albanesi 2.11, 53 n. 32; Saggi 296.
15 The same mythical influence of accent was supposed to be responsible for the alleged different treatments of cl in Italian (gli = l'l' before the accent, kky after the accent). Cf. Meyer-Lübke, Italienische grammatik 138; Grammaire des langues romanes 438 ff., 441 f. The neolinguistic solution is in Bàrtoli, Saggi 208 f.; Introduzione 77; Misc. Hortis 918; Bàrtoli–Meyer-Lübke, Grammatica storica 106 f.; Bonfante, RR 36.240 ff. (1945).
16 Meyer-Lübke publicly disclaimed any responsibility for this work, which was entirely revised and recast by Bàrtoli; cf. Italia dialettale 13.227 (1937). The merit or demerit of the work must therefore be attributed solely to the neolinguist Bàrtoli. Cf. also Hall, Lang. 18.117 n. 3 (1942); Bonfante, MLJ 26.157 (1942). Hall's article in Lang. 18.117 ff. brings further and definitive proofs in favor of Bàrtoli's thesis, first presented by Bàrtoli in Misc. Hortis (1910), a work which Hall fails to mention there at all. But Hall does not see the theoretical implications of this phenomenon, or the reasons for the previous error of the neogrammarians.
17 It is of course quite immaterial whether in the particular instances that he mentions, Bàrtoli is right or wrong; he is certainly right on sez, perhaps wrong on guapto. What matters here is the method that he uses, and that is certainly right.
17a It is perhaps fitting to mention that the lexicon forms a system too, just as morphology (morphonology) and phonetics (phonemics) do. The neogrammatical distinction between these categories reveals itself once more as non-existent. Cf. B. Migliorini, Linguistica 37 (Florence, 1946), the sub-chapter entitled Il sistema lessicale, where we read: ‘È lecito ancora parlare di sistema trattando del lessico? Certamente sí: per quanto numerose siano le parole di una lingua, esse non costituiscono una massa disorganizzata, ma un esercito raggruppato in schiere, con un numero di componenti spesso oscillante, ma con quadri ben stabili.—Le parole servono a fissare le nozioni, e permettono cosí al pensiero di stringere come in una rete, semplificandola e classificandola, tutta la realtà.‘ Cf. also Greenough and Kittredge, Words and their ways in English speech 9, 193, 330.
18 It is worth mentioning here three well-known phenomena which the neogrammarians have not dared to deny, but which they have certainly accepted only grudgingly, and have belittled as much as possible, because they obviously did not fit into their mechanistic and strictly ‘historical’ conception of language. These are what the neogrammarians call false (!) or popular etymology, onomatopoeia, and expressiveness. All three, in fact, disturb and break down the dogma of phonetic law; they all prove the active, creative, imaginative power of man in language, which the neogrammarians deny or ignore. Cf. the works of L. Sainéan, Meillet, and J. Gilliéron (especially Les étymologies des étymologistes et celles du peuple [Paris, 1922]); Iordan, An introduction to Romance linguistics 316 ff. Meillet pointed out that in Indo-European (and especially in Latin) the vowel ǎ and the double stops are ‘expressive’.
19 In enunciating his norms, Bàrtoli always adds, in italics, di norma ‘usually’ (e.g. Breviario 70 ff.; on p. 68 he says di solito; cf. Introduzione 13). At one point (Breviario 74) he insists twice, in italics: di norma, non sempre. See also Saggi 214.
20 This procedure has, among other things, the advantage of pointing out to his successors the problems that still need energetic investigation, as can be shown by several instances.
21 E.g. E. Prokosch, A comparative Germanic grammar 35 (Philadelphia, 1939): ‘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.‘
22 Likewise, the sum of the two lateral areas is frequently a smaller area, and the central (innovating) area is the larger one. This is true in the case of the vowel ŏ in fŏcus, mŏrĭt[ŭr] as compared to the diphthongs in Italian fuoco, muore, Spanish fuego, muere, and French feu, meurt (eu from Old French uo, ue, uö):
![]()
The neogrammatical conclusion (see Prokosch, n. 21) would be of course that the type fuoco (fuego, etc.) is older than f⊖cus. And similarly for the Latin futurum perfectum (or exactum or anterior), preserved only in Old Romanian, Dalmatian, Spanish, and Portuguese; cf. Meyer-Lübke, Gr. 2.391; Bourciez, Éléments3 §211c, §257a, §363c, §480.
23 It is true that several of the opinions here attributed to the neogrammarians have never been openly asserted by them, and have even been sometimes strongly denied. The neolinguists' criticism is therefore often leveled rather at neogrammatical practice, which is well known and can be abundantly documented. The philosophical insufficiency of the whole neogrammatical generation and their slight interest in theoretical problems make a discussion of their theories very difficult at times; for their principles, though implicit in all their work, have rarely been formulated by them. I have limited myself to a brief summary of the main points; but of course if more detailed discussion is needed, I can present more material in the future.
24 Cf. Maver, Slavia 7.149 (1928/9). Reviewing Bàrtoli's Introduzione, he speaks of ‘des faits réels, dont on a cherché, pour la première fois, de dégager un système compact’; and of the neolinguistic doctrine he says, ‘tout s'y tient solidement’ (157).
25 Sometimes I fail entirely to understand Hall's criticism, in the sense that it does not sound to me like criticism at all. Thus we read (22.279): ‘For each of his principles, Bàrtoli lists one set of examples that point in a given direction, another set that point in the opposite direction, and then formulates the principle according to the majority of examples cited. Presumably, if a majority of examples were to point to the opposite conclusion, he would have to re-formulate his principle accordingly.‘ In other words, Bàrtoli has formulated his principles according to the facts. What is wrong with that?
25a Cf. Bàrtoli, Saggi 279 (addition to 228 n. 25): ‘L'area meno esposta [= isolata] è stata confusa spesso con l'area minore e con un'area laterale; cfr. Lineamenti [di linguistica spaziale, by Bàrtoli and Vidossi (Milan, 1943)] n. 88 e il periodico slovacco Linguistica slovaca, Bratislava 1939 e 1940, n. 88 (Blaho).‘
26 They do not at all ‘lend themselves easily to manipulation to produce any effect desired’. The examples given in the text, and many others that can be found in Bàrtoli's works, allow only one possible interpretation.
27 A third possibility is of course that Italy is the innovating area, and Gaul a part of the lateral area (together with Iberia and, on the other side, Dacia), as in the following examples (cf. Introduzione 8, 17 f., 40 f.):

That is, sometimes an innovation is created in Italy, and does not expand beyond; then Iberia, Gaul, and Dacia preserve the older form. Sometimes it is created in Gaul, and does not expand beyond; then Iberia, Italy, and Dacia preserve the older form. Sometimes it is created in Gaul, and later expands into Italy; then Iberia and Dacia preserve the older form. Sometimes it is created in Italy, and later expands into Gaul; then likewise Iberia and Dacia preserve the older form.
28 ‘Lateral’ is a word of the same type as ‘left’ and ‘right’, ‘north’ and ‘south’, etc. Would it be inconsistent and contradictory to assert that Princeton is at the same time north of Washington and south of Boston?
29 On the other hand, Hall himself seems to admit the validity of the norm of the isolated area: ‘What linguistic geography can do for us is rather to furnish ... valuable hints as to probable (not certain) interpretation of forms in isolated, graded, and focal areas’ (22.282). As for the word ‘certain’, the neolinguists never use it, well knowing that nothing is certain (cf. Bàrtoli, Saggi 33 n. 9, 72 n. 9). They are proud of their modesty, and have no wish to imitate the arrogance and dogmatism which are typical of the neogrammarians.
30 On the supposed antithesis between linguistic geography, neolinguistics, and areal linguistics on one side, and comparative linguistics, historical method, phonology, philology, and even semantics and lexicology on the other, see also the excellent remarks of Bàrtoli, Studi albanesi 2.32 f., 68; Saggi 73 n. 19, 275 n. 85, 277 f.
30a This article Hall obviously has not read.
31 Cf. also Leo, Archiv für das studium der neueren sprachen 162.224 (1932/3): 'Sprachvergleichung also ist der methodische grundzug der sprachgeographie.' And see Bàrtoli, Introduzione 98 (note to §34): 'Il Terracher ha tutte le ragioni di protestare, BSLPar. 24, p. 302, perché le indagini dei neolinguisti sono fondate anch'esse sui documenti, oltreché sulle aree, e sono, s'intende, comparative, e mirano alla storia ..., sempre alla storia, e molto piú spesso che quelle dei neogrammatici.'
32 The founding date of the school of the neogrammarians is usually considered to be the year 1876, in which Leskien published his work Die deklination im slavisch-litauischen und germanischen. It is now 1947, but Hall wants to continue (with slight amendments) the methods used 71 years ago—before e.g. any linguistic atlas had been published.
33 Moreover, Bàrtoli frequently says that the history or the area of one word or form or sound is ‘similar’ to that of another. See for instance Introduzione 29: 'Simile al precedente [dies ille: dies illa] è il rapporto geografico e cronologico fra venter ille e uenter illa: cfr. spagn. el vientre e rom. vintrea.' Ibid. 47: 'perché il rapporto tra le forme hiems e hibernum è simile a quello tra dies e diurnum.' And Studi albanesi 2.30: 'Alla coppia magis e plus e a quella di basilica ed ecclesia assomigliano—per ciò che spetta alla ricerca della patria dell'innovazione—le coppie che seguono e che si vedono nell'Introduzione: equa e caballa . . ., alius e alter, ecce ed eccum, hoc e istud, super e supra, sumus e simus; e le innovazioni come dormi da dormit e dormis e simili . . . .'
34 In general, Hall's ideas on the subject of linguistic geography and areal linguistics seem to me obscure, vacillating, and contradictory.
35 I may also add that some of these abnormal cases are only apparently abnormal, as I have shown elsewhere (American Journal of Anthropology 46.382 ff. [1944]). Bàrtoli was therefore even more right than he thought. He himself discovered a few apparently abnormal cases, such as esse: stāre, uelle: quaerere, caecus: orbus, stāre: istāre; Introduzione 5, 35 f., 77 f.; cf. also Introduzione 36: ‘Cosicché l'ital. stare è una conservazione apparente, e però anche cotesta anomalia è solo apparente.‘ And see Saggi 90 f.
36 Hall seems not to notice that if he were right, languages would be continually dwindling; for out of all these old pairs of synonyms (plūs and magis, pyr and ignis, etc.), coeval from the start, only one would survive in each contemporary language.
37 The case is exceptional even among new inventions; generally only one term is adopted and establishes itself (penicillin, telephone, typhus). However, even in the case of railroad and railway, one is certainly older than the other; and research into the problem, though it may seem negligible to Hall, is of the utmost importance to the neolinguist, as is all chronological research. Neolinguists aim at being historians in the full sense of the word, and chronology is the backbone of history. Cf. Bàrtoli, Introduzione 94 f.; Saggi 8, 22 n. 7 and n. 9, 117, 248 n. 25, 257 n. 59, 271 n. 25, 275, 297, 105 f.
38 Compare the wise but often forgotten remark of Bàrtoli, Breviario 77 (= Introduzione 21): ‘Ora, fondare una ipotesi sull'analogia di casi isolati vuol dire o fondarla male o niente affatto.‘
39 ‘The general tendency—and a very sound one—in Indo-European reconstruction these last years has been to reduce to a minimum the number of reconstructed phonemes . . . . The progress has consisted in explaining more and more the “irregularities” of each independent language within the language itself, without piling everything onto [the reconstructed] Indo-European.‘ Cf. also CP 40.119 ff.; Bàrtoli, Saggi 274 n. 82; Thurneysen, IF 4.83 (1894).
40 Of course, people have been killed as well as cured by new inventions like inoculation and sulfa-drugs. The neolinguists have made mistakes; who does not?
41 The most significant methodological innovation of the neogrammarians was the postulate that phonetic laws suffer no exceptions: a conception which was definitely detrimental to the progress of linguistics. Cf. especially I. Iordan, An introduction to Romance linguistics (translated and in part recast by J. Orr; London, 1937), last page, which is quoted by me in RR 37.251.
42 On these intuitions see Bàrtoli, Introduzione 68 ff.; Studi albanesi 2.17, 43, 57 f. n. 51, 62 n. 62; Bonfante, Word 1.137.
43 ‘As far as it can be done, of course, every linguistic problem should be examined with the help of as many methods as possible. If the results reached with each of them are the same, they will corroborate each other and strengthen our conclusion; if there is contradiction, this will help us to correct the (always possible) errors of any single method.‘ This is the conclusion of my study in Word.