¹ Many of the non-Algonquian references given in the footnotes of this paper were found with the assistance of Robert A. Hall Jr. For my understanding of the ‘almost mystical’ version of the drift theory (see §8), I am indebted to Gordon H. Fairbanks (who is not a proponent of this version of the theory). All of my colleagues at Cornell have participated in instructive discussion of this paper, and to all of them I express my thanks.

² Bloomfield's publications on Algonquian are listed here, approximately in order of publication. They will be referred to hereafter as ‘Ref. 1‘ and so on, except that the third and twelfth items will be termed, respectively, the Sound System and the Sketch. The latter includes a bibliography of Algonquian which is complete save for a few items which have appeared since 1941.

1. (1)

   (1) Review of The Owl Sacred Pack of the Fox Indians, by Truman Michelson, AJP 43.276–81 (1922).

2. (2)

   (2) The Menomini Language, Proceedings of the Twenty-first International Congress of Americanists 336–43 (The Hague, 1924).

3. (3)

   (3) On the Sound System of Central Algonquian, Lang. 1.130–56 (1925).

4. (4)

   (4) The Word-Stems of Central Algonquian, Festschrift Meinhof 393–402 (Hamburg, 1927).

5. (5)

   (5) Notes on the Fox Language, IJAL 3.219–32 (1924), 4.181–219 (1927).

6. (6)

   (6) A Note on Sound Change, Lang. 5.99–100 (1928).

7. (7)

   (7) The Plains Cree Language, Proceedings of the Twenty-second International Congress of Americanists 427–31 (Rome, 1928).

8. (8)

   (8) Menomini Texts, Publications of the American Ethnological Society, Vol. 12 (New York, 1928).

9. (9)

   (9) Sacred Stories of the Sweet Grass Cree, Bulletin 60, National Museum of Canada (Ottawa, 1930).

10. (10)

    (10) Plains Cree Texts, Publications of the American Ethnological Society, Vol. 16 (New York, 1934).

11. (11)

    (11) Proto-Algonquian -iit- ‘Fellow’, Lang. 17.292–7 (1941).

12. (12)

    (12) Algonquian, in Linguistic Structures of Native America 85–129 (Viking Fund Publications in Anthropology, Number 6; New York, 1946).

³ ‘A remarkable contribution’, says Sturtevant of Bloomfield's Sketch, Lang. 23.314 (1947).

⁴ Bloomfield has noted the resemblance to older Germanic several times in conversation but not, so far as I know, in print; in Ref. 5, 4.191, the comparison is rather with Indo-European as a whole. Needless to say, the attempts of Reider T. Sherwin, The Viking and the Red Man (4 vols. to date; New York, 1940–46), to establish a genetic connection between Algonquian and Germanic are entirely misguided.

⁵ See fn. 2.

⁶ The principles itemized in this paper are by no means intended to constitute a complete canon of historical or comparative linguistics. The choice is based on the nature of the Algonquian material, not on a general analysis of linguistic method.

⁷ Classical Philology 38.210–11 (1943), 39.218–22 (1944); also the implications of the review in Lang. 16.216–31 (1940).

⁸ Acta Linguistica 2.1–22 (1940–1); IJAL 10.192–3 (1944); and see Robert A. Hall Jr., Bàrtoli's ‘Neolinguistica’, Lang. 22.273–83 (1946).

⁹ The terms ‘synchronic’ and ‘diachronic’ date at least from Ferdinand de Saussure, Cours de Linguistique Générale (Lausanne, 1916).

¹⁰ ‘Contrastive’ is Whorf's proposal (I cannot find the reference), made because the more natural term ‘comparative’ is already pre-empted for a different technical use.

¹¹ Good illustrations of the value of mixed synchronic and diachronic discussion are Ref. 4, 401—which includes also some extremely penetrating remarks on the relation between the synchronic and diachronic points of view—and Bloomfield's paper Literate and Illiterate Speech, American Speech 2.432–9 (1927).

¹² This three-by-three categorization of types of linguistic activity is more general in its application than here indicated: it applies, as a matter of fact, to all the subject-matter of cultural anthropology, of which linguistics is logically a subdivision. This is clearly indicated in Sapir, Time Perspective in Aboriginal American Culture (Ottawa, 1916). There are doubtless some types of fruitful and valid investigation, of language or of culture as a whole, not subsumed in the classification.

¹³ The methodological points or ‘principles’ drawn from the discussion are lettered serially through the paper, from (a) to (p).

¹⁴ That is, we derive our notions of linguistic change in the first instance from the history of those languages, such as English, French, and German, for which there is documentary evidence over a considerable span of time. We then make a comparative analysis of some group of languages whose common ancestor is known to us independently—for example, Romance—assuming the kinds of linguistic change discovered by the first operation, and comparing our reconstruction with the documentary evidence for the parent. Since our inverted predictions in such cases have a considerable degree of accuracy, we feel confident of at least statistically accurate results when the same comparative method is applied to a group of languages for whose parent there is no other evidence. It is true that linguistic science in the 19th century did not follow this course in its chronological development; but this does not invalidate the logical statement. Until the logical interrelationships had clearly emerged, the statements of historical linguistics were somewhat more tentative than afterwards.

¹⁵ That is, Fox (F), Cree (C), Menomini (M), and Ojibwa (O). The abbreviations indicated will be used in the rest of this paper, together with ‘PA’ for Proto-Algonquian, the reconstructed parent (but see §10).

¹⁶ For references see the Sketch. Later contributions by Michelson use approximately the same notation. It is not clear whether this notation was a joint product of Jones and Michelson, or was worked out by one and passed on to the other.

¹⁷ F: at least J M. Cowan, Carl F. Voegelin, and the present writer; O: the latter two, J. A. Geary (other dialects), F. T. Siebert Jr, and—in the summers when O was the language studied in the field-methods course at the Linguistic Institute—a good many others.

¹⁸ It would be tempting to generalize as follows. If a native speaker of a language is trained in phonetics in such a way that he will hear, in any language, at least all those contrasts which are distinctive for his own speech, and will have the symbols easily at hand to record them, he will, in recording his own language, sometimes include subphonemic contrasts, but will never omit contrasts that count. But this statement is far from obviously true. Suggestions of this kind were made by Sapir in Sound Patterns in Language, Lang. 1.37–51 (1925), and more elaborately in La réalité psychologique des phonemès, Journal de Psychologie 30.247–65 (1933), but Twaddell's discussion, On Defining the Phoneme 8–16 (LSA Monograph 16 [1935]), raises serious doubts.

¹⁹ Some such cases are discussed in J. A. Geary, Proto-Algonquian çk: Further Examples, Lang. 17.304–10 (1941).

²⁰ See fn. 19.

²¹ Lang. 22.256 (1946); French Review 17.168–70 (1944).

²² Disagreements about methods of phonemicization are another matter. If ‘phonemic solutions’ are non-unique, as seems likely to me at the present moment, any alternative one will do. There is a gain in clarity and simplicity if each of a set of languages to be compared is phonemicized by the same principles, but if this is not done reconstruction is still perfectly feasible. Against phonemics of any kind: Archivio Glottologico Italiano 31.159 (1940)—one instance of many.

²³ Bloomfield, Language §5.10 (New York, 1933).

²⁴ H. M. Hoenigswald, Internal Reconstruction, SIL 2.78–87 (1944); Sound Change and Linguistic Structure, Lang. 22.138–43 (1946).

²⁵ Obvious from Jones's texts; observed by the present writer in the Kickapoo and Sauk dialects of Fox.

²⁶ Ref. 8, xiv: ‘The texts are here recorded as they struck my ear. Analysis shows that this record (and therefore, if I heard aright, the actual pronunciation) largely obscures the distinction between three short front-vowel phonemes, which here appear as i and e.‘ Since Bloomfield became a fluent speaker of Menomini (a fact which he does not admit, but which is fairly obvious from his discussions), his recording is probably accurate. Nowadays we would cast his statement in the terminology of morphophonemics rather than phonemics, but the fact stated would not be altered.

²⁷ Sketch §33.

²⁸ Max Muller's myth of speakers with vocabularies of only a few hundred words is related. W. von Humboldt, Die Verschiedenheit des menschlichen Sprachbaues §19, gives three stages in the history of language: ‘growth’ before IE, a ‘state of perfection’ in IE, and ‘decay’ since then; for other early references see Lang. 12.101 (1936) and Italica 18.145–54 (1941). Later survivals of these older notions: the subject of review in Lang. 11.154–60 (1935); Emérita 3.257–76 (1935), esp. 272, 274–5; Alf Sommerfelt, La langue et la société (Oslo, 1938).

²⁹ Otto Jespersen, Progress in Language (London, 1894); cf. also his Efficiency in Linguistic Change (Copenhagen, 1941), and the review in Lang. 17.350–3 (1941). Bernhard Karlgren, in reconstructing ancient and archaic Chinese, posits extremely complex phonological structure for those stages of Chinese, and seems (though without explicit reference) to justify this greater complexity for the older stage in terms of Jespersen's theory; see his Philology and Ancient China, esp. 16–18 (Oslo, 1926).

³⁰ See more recent references in fn. 28.

³¹ Ref. 6.

³² Edward Sapir, The Concept of Phonetic Law as Tested in Primitive Languages by Leonard Bloomfield, Methods in Social Science 297–306 (ed. Stuart A. Rice; Chicago, 1931). Also Bloomfield, Language §20.8 (end).

³³ See fn. 17, and Geary, loc.cit.

³⁴ Sketch, fn. 10.

³⁵ Bloomfield, Language §18.14, and his references.

³⁶ Sketch, fn. 13.

³⁷ The English expression fire-water may be a loan-translation from Algonquian; if the former is older, then one or more Algonquian languages may have loan-translated the English.

³⁸ E. Prokosch, A Comparative Germanic Grammar, esp. 34 ff. (Philadelphia, 1939).

³⁹ The discussion in Sapir's Language, chapters 7–8 (New York, 1921), is sufficiently broad to justify (apparently) either interpretation; for the realistic theory, see 184–5 in particular.

⁴⁰ Cogently discussed by C. D. Buck, Some Questions of Practice in the Notation of Reconstructed IE Forms, Lang. 2.99–107 (1926). Buck says (102): the purpose of reconstructions ‘is not to furnish a picture of the parent speech for its own sake, but as a background of the historical relations.' This is in part an answer to the charge of 'irrelevant antiquarianism’ sometimes levelled at the comparative method; even so, we should say that ‘a picture of the parent speech for its own sake’, to the degree attainable, is also a legitimate aim.

When this problem was discussed at Yale in the late thirties, one person cited Sapir as a proponent of the realistic theory, in contrast to Bloomfield as a supporter of formulas. It will be clear from the present discussion that this interpretation of Bloomfield was wrong.

⁴¹ Sketch, §6.

⁴² Loc.cit.

⁴³ In several recent papers (Lang. 20.84–6 [1944], 23.34–42 [1947]), Herbert Penzl has demonstrated, in connection with problems of early English, the difference between phonemic realism and what might be meant by phonetic realism.

⁴⁴ Sketch, §7.

⁴⁵ Since the phonemic systems of the modern languages are different, ‘identical’ must be interpreted relatively: since F and O š and k are elements in differing sets of contrasts, F šk and O šk are not in any absolute sense ‘identical’.

⁴⁶ Sketch, §16.

⁴⁷ Sketch, §17.

⁴⁸ Sketch, §2, referring to T. Michelson, Phonetic Shifts in Algonquian Languages, IJAL 8.131–71 (1935). The implication here is not that Michelson's study is not ‘brilliant’, as Bloomfield terms it, but that the synchronic information on Plains Algonquian available to Michelson was unsatisfactory.

⁴⁹ Sketch, §4.

⁵⁰ Thus Potawatomi is probably historically analyzable on the basis of Bloomfield's P(C)A without modification thereof (C. F. Hockett, The Position of Potawatomi in Central Algonkian, Papers of the Michigan Academy of Sciences, Arts, and Letters 28.537–42 [1942]); but Delaware is probably not (C. F. Voegelin, Proto-Algonquian Consonant Clusters in Delaware, Lang. 17.143–7 [1941]).