¹ Miscellanea in onore di Attilio Hortis 889-913 (1910).

² Modena, 1925.

³ Genève, 1925 (Biblioteca dell' Archivum Romanicum [BAR] 11.12).

⁴ Especially as set forth in Croce's Estetica come scienza dell'espressione e linguistica generale (1901).

⁵ American Anthropologist NS.46.382-86 (1944) ; PMLA 59.877-881 (1944) ; JAOS 64.177 ff. (1944), especially fn. 57; Romanic Review 36.240-3 (1945); JAOS 65.363-4 (1945); Word 1.132-61 (1945), especially 149, fn. 16.

⁶ The first of the items cited in fn. 5 (AA NS.46.382 ff.) is a programmatic exposé of some of Bàrtoli's theories; but it omits the first and sixth of his principles, and fails to mention any of the truly discrepant evidence, or even the fact of its existence, which Bàrtoli mentions throughout. 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.

⁷ In essence, therefore, this article is a review of two books over twenty years old. The justification for so belated a review lies in the situation outlined in this paragraph.

⁸ This principle is rather tautological, since just as soon as the ‘conservative’ area becomes smaller than the ‘innovating’ area, it is automatically re-defined as ‘marginal’ or ‘isolated’.

⁹ The term area seriore refers to areas into which a linguistic feature has been introduced from another area in which the feature in question was previously in use. Under this heading come speech-colonies (e.g. the Judeo-Spanish colonies in the Levant or the North-Italian linguistic islands in Southern Italy) and languages which have borrowed from others (e.g. Celtic, Germanic, Slavic, Albanian, Greek, and Basque, all in relation to Latin). The term later area, though clumsy and unidiomatic in English, is the best equivalent I can find for area seriore; the expression ‘territory later occupied or colonial’ (AA NS.46.385) covers only part of Bàrtoli's meaning.

¹⁰ Italienische Grammatik §§208 ff. (Leipzig, 1890).

¹¹ As Bàrtoli himself notes (58), this theory was later abandoned by practically all scholars, including Meyer-Lübke himself, and replaced by the hypothesis of dialect borrowing. Was this the best example of Neo-Grammarian procedure Bàrtoli could find to discuss?

¹² Language 354 (New York, 1933).

¹³ Programma di filologia romanza come scienza idealistica (Genève, 1922; BAR I.2); Lingua e pensiero (Firenze, 1932) and Lingua e poesia (Firenze, 1937), reviewed together by the present writer in Italica 15.239-242 (1938); Lingua e cultura (Firenze, 1939); and a scries of polemic articles in Archivum Romanicum, vols. 4 through 15.

¹⁴ Cf. Bàrtoli, Introduzione 59: ‘Le leggi fonetiche non sono mai cause, e sono soltanto condizioni’, and Bertoni's books and articles mentioned in fn. 13, especially ‘La legge fonetica’, AR 5.1 ff. (1921), reprinted in Programma di filologia romanza 109-122.

¹⁵ Cf. Bloomfield's statement: ‘Especially during the last decades, linguists have come to see that dialect geography furnishes the key to many problems’ (Language 51).

¹⁶ Cf. DAE and NED s.w.

¹⁷ Thus, for instance, Newton's ‘first law of physics’ is never attested in the universe as we know it, and yet it is the only principle by which the motion of objects in our universe can be analyzed and predicted, and in whose light aberrant phenomena can be interpreted, such as the variation in the orbit of Uranus which led to the discovery of Neptune; cf. Russell, Dugan, and Stewart, Astronomy 399-400 (Boston, 1945).—I am indebted for this reference, and for discussion of this point, to Prof. C. H. Smiley, of the Department of Astronomy, Brown University.

¹⁸ In his comment on Bloomfield's article, A Note on Sound Change, Lang.4.100 (1928).

¹⁹ Der Sprachatlas als Forschungsinstrument 187 (Halle, 1928).

²⁰ Cf. Bloomfield, Language, ch. 19.8.

²¹ Buck, Comparative Grammar of Greek and Latin §84 (Chicago, 1933).

²² Buck §74.

²³ Buck §85.

²⁴ There is, strictly speaking, a third possibility, that of sporadic sound-change : *a > i in some words and not in others, at random and as equally possible outcomes without rhyme or reason; but I do not know of any competent scholar who would uphold this possibility at present.

²⁵ Lang. 4.99-100 (1928).

²⁶ The later addition of other examples by Michelson (IJAL 8.41 [1933]; 10.81 [1939]) and Geary (Lang. 17.304-310 [1941]) is without bearing one way or the other on the validity of Bloomfield's original assumption of PA *çk or of the fundamental principle of the comparative method and of phonemic change.

²⁷ Cf. Jane Daddow [Hawkins], The Speech of the Hudson River Valley 7, 314, 395 (Brown University dissertation, 1941).

²⁸ Linguistic Atlas of New England, point 236, informant 1. The word four is on map 54; father is on map 370.

²⁹ Linguistic Atlas of New England, point 150, informant 1.

³⁰ Even so, how does Indo-Iranian come to be a ‘central area’ and the northern IE languages ‘more isolated’ and ‘less exposed to communications’ than the others (Introduzione 50,51)?

³¹ It is usually assumed that they were, by Neo-Grammarians and their opponents alike; but scholars who are conscious of their basic assumptions realize and state them as such, e.g. Buck, Comparative Grammar of Greek and Latin §1; Gray, Foundations of Language 309 (New York, 1939).

³² This is, after all, the final justification of the ‘starred form’, even admitting that comparativisls may have on occasion made excessive use of ‘starred forms’, e.g. in Romance, instead of seeking documentary and other evidence to begin with (as pointed out most recently by Y. Malkiel, Lang. 21.149 [1945]). Cf. A. Graur, Mots ‘reconstruits’ et mots attestés, Bulletin linguistique 2.11-20 (1934), especially 14-5.

³³ Thus, for instance, even one with no knowledge of Salishan languages may be permitted to guess from the map given by Boas and Haeberlin in their Sound Shifts in Salishan Dialects (IJAL 4.119 [1927])—in which the dialects on the outer edges of the area (coast and inland) at the end of the 18th century are shown as having the innovation tc < k (120-3)— that something may be wrong. The development may of course have taken place independently in the two regions; but one suspects that conditions and developments of later times, when the geographical relationships were different (117-8), may have been ascribed to earlier times, with a resultant improbable situation on the map. (It is all the more remarkable that the diffusionist Boas should have failed to realize this.) Clearly, considerable dialect mixture and borrowing has taken place, as shown by such remarks as ‘Most of the coast dialects belong to the tc group’ (120) ; ‘In the Puget Sound dialects there are a number of exceptions’ (125) ; ‘The study of the sound-shifts in the dialects of the coast is far more difficult than it is in the case of the inland dialects’ (131-2). Cf. C. F. Voegelin, Influence of Area in American Indian Linguistics §2.2 (Word 1.55-6 [1944]).

³⁴ My use of the term ‘American linguists’ here is purely geographical, and intended to refer only to the fact that, as is well known, the tradition of linguistics has, to a certain extent, developed in different directions in America than elsewhere, and that, consequently, some types of linguistic work (e.g. Bàrtoli's theories) are not as well known here.

³⁵ Bloomfield is cited here and elsewhere in this paper, in connection with linguistic geography, not as an ‘authority’ but simply as an illustration. This is done on the assumption (which I believe is well-founded) that Bloomfield's Language is the best-known and most widely studied serious work on linguistics in this country, and that hence a point made by Bloomfield may be assumed to be well known.

³⁶ CP 39.55 (1944), fn. 12.

³⁷ JAOS 65.263 (1945).

³⁸ Cf. Voegelin, Influence of Area in American Indian Linguistics §2.6 (Word 1.58).