The Collected Works of Edward Sapir VI The Collected Works of Edward Sapir Editorial Board Philip Sapir Editor-in- Chief William Bright Regna Darnell Victor Golla Eric P. Hamp Richard Handler Judith Irvine The Collected Works of Edward Sapir VI American Indian Languages 2 Volume Editor Victor Golla 1991 Mouton deGruyter Berlin • New York Mouton de Gruyter (formerly Moulon. The Hague) is a Division of Waller de Gruyter & Co., Berlin. ? 537 @ Printed on acid-free paper which falls within the guidelines of the ANSI to ensure permanence and durability. Library of Congress Cataloging in Publication Data (Revised for vol. 2) Sapir, Edward, 1884-1939. American Indian languages. (The Collected works of Edward Sapir : 5 — 6) Vol. 1 edited by William Bright. Vol. 2 edited by Victor Golla. Includes bibliographical references and index. 1. Indians of North America — Languages. I. Bright, William. 1928- . II. Golla, Victor. III. Sapir, Edward, 1884-1939. Works. 1990 ; 5-6. IV. Title. V. Series. PM108.S26 1990 497 89-13233 ISBN 0-89925-654-6 (v. 1) ISBN 0-89925-713-5 (v. 2) Deutsche Bibliothek Cataloging in Publication Data Sapir, Edward: (The collected works] The collected works of Edward Sapir / ed. board Philip Sapir ed. -in-chief. ... — Berlin ; New York : Mouton de Gruyter. ISBN 3-11-010104-1 (Berlin) ISBN 0-89925-138-2 (New York) NE: Sapir. Edward: [Sammlung] 6. American Indian languages. — 2. Vol. ed. Victor Golla. — 1991 ISBN 3-11-012572-2 (f) Copyright 1991 by Walter de Gruyter & Co.. Berlin 30. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Printing: Gerike GmbH, Berlin. Binding: Luderitz & Bauer, Berlin. Printed in Germany. Edward Sapir, 1937 University of Michigan, Ann Arbor, Michigan, taken by Kenneth Pike ( Courtesy of Sapir family ) Edward Sapir (1884-1939) has been referred to as "one of the most brilliant scholars in linguistics and anthropology in our country" (Franz Boas) and as "one of the greatest figures in American humanistic scholarship" (Franklin Edgerton). His classic book, Language (1921), is still in use, and many of his papers in general linguistics, such as "Sound Patterns in Language" and "The Psychological Reality of Phonemes," stand also as classics. The development of the American descriptive school of structural linguistics, including the adop- tion of phonemic principles in the study of non-literary languages, was pri- marily due to him. The large body of work he carried out on Native American languages has been called "ground-breaking" and "monumental" and includes descriptive, historical, and comparative studies. They are of continuing importance and relevance to today's scholars. Not to be ignored are his studies in Indo-European, Semitic, and African languages, which have been characterized as "masterpieces of brilliant associa- tion" (Zellig Harris). Further, he is recognized as a forefather of ethnolinguistic and sociolinguistic studies. In anthropology Sapir contributed the classic statement on the theory and methodology of the American school of Franz Boas in his monograph, "Time Perspective in Aboriginal American Culture" (1916). His major contribution, however, was as a pioneer and proponent for studies on the interrelation of culture and personality, of society and the individual, providing the theoretical basis for what is known today as humanistic anthropology. He was, in addition, a poet, and contributed papers on aesthetics, literature, music, and social criticism. Note to the Reader Throughout The Collected Works of Edward Sapir, those publications whose typographic complexity would have made new typesetting and proofreading difficult have been photographically reproduced. All other material has been newly typeset. When possible, the editors have worked from Sapir's personal copies of his published work, incorporating his corrections and additions into the reset text. Such emendations are acknowledged in the endnotes. Where the editors themselves have corrected an obvious typographical error, this is noted by brackets around the corrected form. The page numbers of the original publication are retained in the pho- tographically reproduced material; in reset material, the original publication's pagination appears as bracketed numbers within the text at the point where the original page break occurred. To avoid confusion and to conform to the existing literature, the page numbers cited in introductions and editorial notes are those of the original publications. Footnotes which appeared in the original publications appear here as footnotes. Editorial notes appear as endnotes. The first endnote for each work contains the citation of the original publication and, where appropriate, an acknowledgment of permission to reprint the work here. All citations of Sapir's works in the editorial matter throughout these vol- umes conform to the master bibliography that appears in Volume XVI; since not all works will be cited in any given volume, the letters following the dates are discontinuous within a single volume's references. In volumes where unpublished materials by Sapir have been cited, a list of the items cited and the archives holding them is appended to the References. Contents Frontispiece: Edward Sapir 6 Preface 13 Introduction to Volumes V and VI 15 Section Six: Athabaskan and Na-Dene Languages Introduction 21 Notes on Chasta Costa Phonology and Morphology (1914) 27 Corrigenda to Father Morice's "Chasta Costa and the Dene Languages of the North" (1915) 95 The Na-dene Languages, a Preliminary Report (1915) 105 The Sino-Dene Hypothesis [excerpts from a letter to A. L. Kroeber] (1921) 133 Athabaskan Tone (1922) 141 A Type of Athabaskan Relative (1923) 143 The Phonetics of Haida (1923) 151 Pitch Accent in Sarcee, an Athabaskan Language (1925) 169 The Similarity of Chinese and Indian Languages (1925) 191 Review of Berard Haile, Manual of Navaho Grammar (1926) 193 A Summary Report of Field Work among the Hupa, Summer of 1927 (1928) 195 The Concept of Phonetic Law as Tested in Primitive Languages by Leonard Bloomfield [excerpt] (1931) 199 Two Navaho Puns (1932) 203 Problems in Athapaskan Linguistics 205 1 0 Contents Review of A. G. Morice, The Carrier Language (1935) 207 Internal Linguistic Evidence Suggestive of the Northern Origin of the Navaho (1936) 209 Cornelius Osgood, The Distribution of the Northern Athapaskan Indians [contribution by Sapir]: Linguistic Classification within the Northern Athapaskan Area (1936) 221 Section Seven: Penutian Languages Introduction 225 Preliminary Report on the Language and Mythology of the Upper Chinook (1907) 231 Franz Boas, Chinook [contributions by Sapir] (1911): Diminutive and Augmentative Consonantism in Wishram Post-positions in Wishram Wishram Text and Analysis Modal Elements 243 A Characteristic Penutian Form of Stem (1921) 263 A Chinookan Phonetic Law (1926) 275 L. S. Freeland, The Relationship of Mixe to the Penutian Family [with notes by Sapir] (1930) 283 (with Morris Swadesh) Coos-Takelma-Penutian Comparisons (1953) .... 291 Comparative Penutian Glosses 299 Section Eight: Wakashan and Salishan Languages Introduction 319 The Rival Chiefs, a Kwakiutl Story Recorded by George Hunt (1906) 323 Some Aspects of Nootka Language and Culture [excerpt] (1911) 353 Abnormal Types of Speech in Nootka (1915) 357 Noun Reduplication in Comox (1915) 381 The Rival Whalers, a Nitinat Story (1924) 435 Contents 1 1 Nootka Baby Words (1929) 465 Morris Swadesh (ed.), Salish-Wakashan Comparison (1949) 467 Section Nine: Other American Languages A Tutelo Vocabulary (1913) 471 Review of B. Bibolotti, Moseteno Vocabulary and Treatises (1918) 475 Appendix A. G. Morice, Review of Sapir, Notes on Chasta Costa Phonology and Morphology (1915) 481 A. G. Morice, Chasta Costa and the Dene Languages of the North (1915) 485 A. G. Morice, Misconceptions Concerning Dene Morphology: Remarks on Dr. Sapir's Would-be Corrigenda (1917) 499 E. Sapir, Corrigenda and Addenda to Takelma Texts (1914) 513 Phonetic Key to Publications of Edward Sapir 515 References 525 Index to Volumes V and VI 543 Preface Volumes V and VI of The Collected Works of Edward Sapir are devoted to shorter works on American Indian languages (mainly of North America), including some previously unpublished material. Volume V, edited by William Bright, contains papers of a general nature on typology, classification, and phonetic notation, followed by work on Hokan languages, on the Uto-Aztecan family, and on the relationship of Algonkian, Wiyot, and Yurok. Volume VI, edited by Victor Golla, contains articles on Athabaskan and Na-Dene lan- guages, on Penutian, and on the Wakashan and Salishan families, plus two short papers on languages of other groups. Appendices in both volumes con- tain papers written by other authors which were discussed in papers by Sapir. A combined index to Volumes V and VI appears in the latter. The editors of these two volumes have worked together in planning the entire sequence. Two possible ways of organizing the material were considered. One would be purely chronological, without considering topic; the other, adopted here, separates the articles into topical divisions and then arranges them chron- ologically within each division. This has the advantage, we believe, of making it easier for the reader to consult related papers in close proximity. In addition to the articles contained in these two volumes, a number of arti- cles which discuss one or more specific American Indian languages appear in Volumes I through IV of The Collected Works. These are listed below, orga- nized by language or language group. The volume in which a paper is to be found is indicated by the appropriate roman numeral in brackets. Athabaskan Languages: 1923c, A Note on Sarcee Pottery [IV]; 1924d, Per- sonal Names among the Sarcee Indians [IV]; 1933c, La realite psychologique des phonemes [I]; 1935b, A Navaho Sand Painting Blanket [IV]; 1936e, Hupa Tattooing [IV]; 1936h, Kutchin Relationship Terms [IV]; 1930, A Note on Navaho Pottery (with Albert G. Sandoval) [IV]. Comox: 1939e, SongsforaComox Dance Mask (edited by Leslie Spier) [IV]. Nootka: 1913b, A Girls' Puberty Ceremony among the Nootka Indians [IV]; 1915h, The Social Organization of the West Coast Tribes [IV]; 1919e, A Flood Legend of the Nootka Indians of Vancouver Island [IV]; 1933c, La realite psy- chologique des phonemes [I]. Southern Paiute: 1910d, Song Recitative in Paiute Mythology [IV]; 1933c, La realite psychologique des phonemes [I]. Takelma: 1907b, Notes on the Takelma Indians of Southwestern Oregon [IV]; 1907d, Religious Ideas of the Takelma Indians of Southwestern Oregon [IV]. 14 VI American Indian Languages 2 Tsimshian: 1915g, A Sketch of the Social Organization of the Nass River Indians [IVj; 1920c, Nass River Terms of Relationship [IV]; 1921c, A Haida Kinship Term among the Tsimshian [IV]. Yana: 1908a, Luck-Stones among the Yana [IV]; 1916g, Terms of Rela- tionship and the Levirate [IV]; 1918j, Yana Terms of Relationship [IV]; 1922d, The Fundamental Elements of Northern Yana [IX]; 1923m, Text Analyses of Three Yana Dialects [IX]; 1928j, The Unconscious Patterning of Behavior in Society [III]. Volumes VII-XV, which contain Sapir's work of monographic scope on American Indian languages and cultures, also include some shorter, closely related articles containing lexical inventories and textual analyses. Note that Sapir s Language: An Introduction to the Study of Speech (Volume II) cites some thirty American Indian languages, and his 1916 monograph, Time Per- spective in Aboriginal American Culture (Volume IV), one-third of which is devoted to "evidence from linguistics," cites dozens of American Indian lan- guages or language groups. It should also be noted that all references to specific languages in each article are listed in the indices of each individual volume, as well as in the comprehensive index in Volume XVI. Preparation of this volume was supported in part by grants from the Phillips Fund of the American Philosophical Society, the National Science Foundation (grant no. BNS-8609411), and the Andrew W. Mellon Foundation. The editor also acknowledges the contributions to the preparation of this volume by Jane McGary and the help of Dr. Marie-Louise Liebe-Harkort, editor in chief of Mouton de Gruyter. Introduction to Volumes V and VI It has often been said that Franz Boas is to be considered the father of anthro- pological linguistics in North America, and in particular the initiator of serious research on American Indian languages. But surely Edward Sapir, who began his career as a student of Boas, became the most influential scholar of the twen- tieth century in both these fields . Consider the diversity of the Native American languages on which Sapir did original research — Chinook, Takelma, Yana, Southern Paiute, Nootka, Sarcee, Navajo, and others; or the language families in which he did ground-breaking comparative work — Hokan, Uto-Aztecan, Algonkian, Athabaskan, and Penutian; or the types of studies he carried out — descriptive, historical, comparative, ethnolinguistic, and what would now be called sociolinguistic. Even before his untimely death, Sapir's achievements were monumental; after 1939, his stature as an Americanist only grew, as many of the materials he left in manuscript were edited and published by his students. His stature grows yet more in subsequent volumes of these Collected Works, with the publication of several major collections of texts (Sarcee, Kutchin, and Hupa) and other important longer manuscripts, now edited by students of his students. It is possible to attempt some general comments about the overall course of Sapir's work on North American Indian languages as it is reflected in the pres- ent pair of volumes. Publications from the period 1906-1910 are primarily descriptive, including the first results of field work on Wishram Chinook, Ta- kelma, and Yana. In 1911, typological interest emerges in "The Problem of Noun Incorporation in American Languages" (1911c) and is pursued most nota- bly in the two reviews (1917k, 19171) of works by Uhlenbeck. Comparative lin- guistic research, aimed at establishing relatively remote linguistic relationships on the basis of both lexical and grammatical comparisons, comes to the fore in 1913 with "Southern Paiute and Nahuatl, a Study in Uto-Aztekan" (19131, 19151) and "Wiyot and Yurok, Algonkian Languages of California" (1913h). During the following half dozen years, Sapir's enthusiasm for tracing remoter rela- tionships is manifest in such papers as "The Na-Dene Languages" (1915d), "The Hokan and Coahuiltecan Languages" (1920b, written in 1915), and "A Charac- teristic Penutian Form of S-tem" (1921b, written in 1918). This interest reached its culmination in a drastic proposal to reduce 58 North American "stocks" (as formulated by John Wesley Powell in 1891) to just six "great groups. " This classi- fication, based on grammatical and typological rather than lexical corre- spondences, was presented in a lecture at Chicago in 1920 (the notes for which are published here in "Materials Relating to Sapir's Classification of North American Indian Languages"). With little change, this formed the core for Sapir's influential Encyclopaedia Britannica article on "Central and North 15 VI American Indian Languages 2 American Languages" (not published until 1929, 1929a). After the early 1920s, Sapir s interest in these problems seems to have cooled; however, his last major work in this genre, "The Hokan Affinity of Subtiaba in Nicaragua" (1925b), argues for a Central American extension of the far-flung Hokan (-Coahuiltecan) group, and presents what is perhaps Sapir's most detailed argu- ment for the importance of "submerged" structural features in recognizing remote linguistic relationship. Sapir's sixfold classification and the methodology supporting it constituted, during his lifetime, the most controversial part of his work on North Amer- ican languages (it was never accepted, for instance, by his onetime teacher Boas). It should be remarked, however, that what Campbell and Mithun (1979: 26) have called the "reductionist zeal" of this classification was not unique to Sapir. Large-scale genetic regrouping of North American languages was initiated by Alfred L. Kroeber and Roland B. Dixon, who, in a series of papers beginning in 1913, proposed assigning most of the Powellian language families of California to one or the other of two new "stocks," Penutian and Hokan (Dixon and Kroeber 1913, 1919). Sapir joined in this work only after the groundwork had been laid, and at Kroeber's urging (GoUa 1986: 178). Sapir brought to the task a thorough famiharity with the methods and data of Indo-European comparative philology, and — after a brief period of skepticism — he became convinced that a rigorous application of philological principles to American languages would yield important new insights. He moved from one bold synthesis to another, and his comprehen- sive classification of 1920 must be regarded as little more than a report on work in progress. It is noteworthy, however, that Sapir did relatively little after 1920 either to support or to revise that classification. His 1925 paper on Subtiaba, while introducing some new structural arguments for Hokan, is based on essentially the same group of cognate sets as in his earlier work, and it refers only briefly to the larger Hokan-Siouan grouping introduced in his 1920 lecture. In contrast with the wide-ranging comparative work that had absorbed him during the preceding decade, Sapir s research during much of the 1920s focused narrowly and intensively on one group of languages: the "Na-Dene" stock of his 1920 classification (comprising Tlingit, Haida, and the widespread Athabaskan family). As early as 1906 he had worked briefly, during his Takelma field work, with a speaker of Chasta Costa, an Oregon Athabaskan language; in preparing this material for publication (1914c), he saw Athabaskan as a family having the diversity and the relatively good documentation to make it a match for his skills as a comparativist. He was soon embroiled in controversy with older Athabaskan scholars (e.g. , Father Morice, 1915c, see Volume VI and Appendix to Volume VI); this was exacerbated by his 1915 proposal (1915d) of a genetic relationship among Athabaskan, Tlingit, and Haida. Sapir concluded that only through extensive field work of his own could he hope to accumulate the evi- Introduction 17 dence necessary to convince his critics. His feeling about the necessity of such work became even stronger when, around 1920, he came to suspect that an intercontinental genetic connection between Na-Dene and Sino-Tibetan was a distinct possibility. Sapir's plan for Na-Dene field research was extraordinarily ambitious, and it was never completed. Except for a foray into Haida phonetics (1923d), his field work was entirely devoted to Athabaskan, involving four major inves- tigafions: Sarcee, in 1922; Kutchin and Ingalik, in 1923; Hupa, in 1927; and Navajo, principally in 1929. Only the Sarcee work is significantly represented in Sapir's bibliography; even here the major published study was prepared in collaboration with his student Li Fang-Kuei (Li 1930, see Volume XIII). A good deal of the material collected by Sapir has been published post- humously, but the definitive grammar of Navajo which Sapir planned (and was working on even during his last illness) will never be written. Of his comparative insights into Athabaskan, Na-Dene, and Sino-Dene, we have only fragmentary notes. Sapir's active research career extended from 1905 to 1938, or 33 years. During the first two decades of this period — until his move from Ottawa to a teaching post at the University of Chicago — he was engaged almost exclusively in Amer- ican Indian research, the bulk of it descriptive linguistics. After 1925 his inter- ests began to turn toward other types of study, particularly the psychology of culture; and his linguistic field research virtually came to an end when he moved from Chicago to Yale in 1931. He remained, nonetheless, a central figure in American Indian linguistics, second only to Boas in status and pre-eminent in intellectual influence. Nearly all his important students took up the study of American Indian languages. It was left to them, and to their scholarly progeny in turn, to continue the many facets of his research. We will do no more here than mention the names of Harry Hoijer, Morris Swadesh, George Trager, Stanley Newman, Li Fang-Kuei, Benjamin L. Whorf, Charles F. Voegelin, and our own teacher, Mary Haas. All these scholars have transmitted to their own students not only an enthusiasm for American Indian linguistics, but, even more important, Sapir's commitment to the study of language within the broad- est context of human understanding. William Bright Victor Golla Section Six: Athabaskan and Na-Dene Languages Introduction While Sapir began his involvement with Hokan and Penutian languages as a field linguist, turning to comparative and classificatory studies only after com- pleting several major descriptive works, with Athabaskan the reverse was the case. Except for some unsystematic notes on Chasta Costa collected during his Takelma field work in 1906 (published as 1914c; see below), Sapir had had no first-hand experience with Athabaskan languages when he launched what came to be the most absorbing of his historical hypotheses: the genetic relationship of Athabaskan, Tlingit, and Haida, and the deep connection of this "Na-Dene" family to the Sino-Tibetan stock. Neither the Na-Dene hypothesis nor the possibility of connections with Asi- atic languages were, strictly speaking, original to Sapir (see the discussion of earlier speculation in Krauss 1973), but there is little doubt that Sapir was the first to explore these questions in the light of modern comparative linguistics. The evidence from Sapir's correspondence and manuscripts is that he probably took up this work late in 1912 or early in 1913, shortly after completing a detailed study of Uto-Aztecan (Sapir to Kroeber, 23 December 1912, in GoUa 1984: 71). In the ensuing months he apparently read through all the extant mate- rial both on comparative Athabaskan and on Tlingit and Haida. In the spring of 1913 we find him complaining in a letter to Robert Lowie that Pliny Earle God- dard, a leading student of comparative Athabaskan, was over-cautious: He does not . . . seem to me to get very much beyond descriptive Athabascan sketches cast in parallel grooves. The unifying reconstructive spirit, the elimination of secondary features and the emphasis on essential ones, seem to be lacking, on the whole. It seems to me that with all the experience that Goddard has had with Athabascan he would have felt irresistably drawn by this time to a serious consideration of Haida and Tlingit as possibly genetically related, though remotely, to Athabascan. This may not seem eventually as far-fetched as it does now. But I am afraid that Goddard is rather timid in these matters. (Sapir to Lowie, March 1913.) Three months later he wrote to A. L. Kroeber: A propos of larger linguistic units, which seem to be somewhat in favor just now, I may say that I have been occupying myself of late with Athabaskan, Tlingit, and Haida, and that I have collected enough evidence to convince myself at least of the genetic relationship of these three . (Sapir to Kroeber, 30 May 1913, in Golla 1984: 104) It was during this period that he went back to his Chasta Costa notes and worked them up into a short monograph with a distinctly comparative emphasis. Notes on Chasta Costa Phonology and Morphology (1914c). The publication of this work immediately brought him into the small circle of serious Athabaskanists, of whom the most established and productive was the French-Canadian Oblate priest, A. G. Morice. Father Morice reviewed Sapir s Chasta Costa monograph in glowing terms (Morice 1915a), but followed his review with a more critical 22 V7 American Indian Languages 2 appraisal, "Chasta Costa and the Dene Languages of the North" (Morice 1915b). Sapir replied to Morice's strictures in "Corrigenda to Father Morice's 'Chasta Costa and the Dene Languagesof the North'" (1915c), to which Morice gave rejoinder in "Misconceptions Concerning Dene Morphology: Remarks on Dr. Sapir s Would-be Corrigenda" (Morice 1917). Morice s review and papers are reprinted in the Appendix to this volume. In the assessment of at least one scholar, "in the long run . . . the interaction between Morice and Sapir turned out to be productive," with the two showing "grudging respect for each other" (Krauss 1986: 153-154). The relationship, stormy or not, was essentially between two generations of scholars, and between a man thoroughly familiar with the concrete details of several Canadian Athabaskan dialects and a com- parativist primarily interested in historical reconstruction. As the nature of his Athabaskan work shifted in the 1920s and 1930s from comparative to descrip- tive, Sapir s appreciation of Morice's extensive knowledge of Athabaskan struc- ture grew, as is evidenced by his short but appreciative review (1935c) of Mor- ice's massive grammar and dictionary of Carrier (1932). Between 1913 and 1915 Sapir continued to devote much of his research time to the Na-Dene project, combing the published documentation of Northern, Pacific Coast, and Southwestern Athabaskan languages for material to be com- pared with Swanton's descriptions of Haida and Tlingit. By 1915 he had amassed about 300 lexical comparisons and had begun writing a "systematic presentation" of the material (Sapir to Radin, 17 July 1918, quoted in Krauss 1986: 156; see Dallaire 1984: 169, letter no. 263). At this point, at Goddard's request, he prepared the shorter paper reprinted here, "The Na-Dene Lan- guages, a Preliminary Report" (1915d), iox pub\\cdi\\on'mi\\Q American Anthro- pologist. The manuscript of the full study has unfortunately been lost, although the ledgers in which Sapir entered his Na-Dene lexical comparisons have sur- vived (manuscript 497.3 B63c Na20a.3, vols. 1, 3, and 4, American Philosoph- ical Society Library). Even after the publication of Sapir's evidence Boas and Goddard remained skeptical about Na-Dene, and Boas challenged Sapir's methods in a heated exchange at the annual meeting of the American Anthropological Association the following December. The strength of Boas's opposition (which reached print in 1920 in a scathing attack on the misuse of genetic classification), together with the appearance of Boas's own descriptive study of Tlingit (Boas 1917), seems to have taken the wind out of Sapir's sails, at least temporarily. He devoted hardly any time to Athabaskan or Na-Dene from 1916 through 1920, which was in general a period during which he was more occupied with literary and artistic matters than with linguistic research. Late in 1920, as he was completing his general book Language (1921d), Sapir experienced what he described to Kroeber as a "considerable recrudescence of interest in linguistics" (Golla 1984: 347), particularly in classificatory work. On the one hand, this led to Sapir's working out of a general classificatory scheme for all North American languages, grouping most of them into six "great Six: Athabaskan and Na-Dene Languages 23 Stocks" (1921a). But it also led him to reconsider Na-Dene. As he put it to Kroeber: I am just now interested in another big linguistic possibility. I tremble to speak of it, though I've carried the germinal idea with me for years. I do not feel that Na-dene belongs to the other American languages. I feel it as a great intrusive band that has perhaps ruptured an old Eskimo-Wakashan-Algonkin continuity In short, do not think me an ass if I am seriously entertaining the notion of an old Indo-Chinese offshoot into N.W. America. ... I have already carefully gone over two Tibetan grammars (Jiischke and Foucaux) and find in Tibetan pretty much the kind of base from which a generalized Na-dene could have devel- oped, also some very tempting material points of resemblance. (Sapir to Kroeber, 4 October 1920, in GoUa 1984: 350; reprinted in volume V: 81-83). In the ensuing months Sapir delved deeply into Chinese and Sino-Tibetan lin- guistics, working to some extent under the guidance of the anthropologist and orientalist Berthold Laufer. By the end of the summer of 1921 Sapir had developed the outlines of his "Sino-Dene" hypothesis, and he discussed the matter at some length in a letter to Kroeber (a copy of which was also sent to Laufer), dated October 1, 1921, and printed here almost in its entirety as "The Sino-Dene Hypoth- esis." The complete text of this and a short follow-up letter can be found in Golla (1984: 374-384). More pressing than the need to acquaint himself thoroughly with the Asiatic side of the relationship, however, was the need Sapir felt for a more complete and accurate documentation of Athabaskan, Tlingit, and Haida. Seeing them now through a Sino-Dene lens, it was clear to Sapir that many important aspects of Na- Dene phonology and grammar had been missed by previous investigators. Late in 1921, he drew together extensive comparative evidence on the morphophonology of relativization in Athabaskan. This paper, "A Type of Athabaskan Relative" (1923n), calls attention to the distinctive clause-like nature of Athabaskan poly- synthetic structure and, as Sapir wrote to Kroeber, "insidiously prepares for far bigger things than its ostensible theme" (Sapir to Kroeber, 24 November 1921, in Golla 1984: 386). Especially important in Sapir's eyes was accurate information on the presence of tonal systems, which the Sino-Dene connection made him certain was fundamen- tal throughout Na-Dene. A pitch accent had been reported for Tlingit by Boas (1917), and Sapir himself, working briefly with a speaker of Haida in March 1920, had noted tonal and other phenomena of potential historical importance in Haida, later described in "The Phonetics of Haida" (1923d). Tone had not, however, been definitely reported in Athabaskan, and Sapir was certain that this was an over- sight. To his delight, the first Athabaskan language he had the opportunity to work on, Sarcee, turned out to have a well developed system of pitch accent. In an immediate announcement in the American Anthropologist, "Athabaskan Tone" ( 1922a), he flatly stated that in the light of his Sarcee observations "it is well nigh inconceivable that [tone] should be absent in any other Athabaskan dialect." This view was reiterated in Sapirs full analysis, "Pitch Accent in Sarcee, an Athabaskan Language" (19251), which is less a descriptive study than a general theory of Athabaskan tone illustrated with Sarcee data. The paper, moreover. 24 VI American Indian Languages 2 ends with a list of questions that Sapir felt could be answered satisfactorily only from the standpoint of Na-Dene (1925f: 204-205). His wife's declining health forced a postponement of Sapir s planned visit to the Hupa in 1923, but he was able to spend much of the summer of that year doing productive field work with two young Alaskan Athabaskans, speakers of Ingalik (Anvik) and Kutchin, who were working at a camp in Pennsylvania. In the turmoil following his wife's death early in 1924, and his subsequent move from Ottawa to Chicago in 1925, Sapir carried out little Athabaskan work dur- ing the next two years. That the Sino-Dene hypothesis still strongly attracted him, however, is shown by an interview he gave to Science shortly after arriving at the University of Chicago, printed under the title "The Similarity of Chinese and Indian Languages" (1925o). From the beginning of his teaching at Chicago until his death 14 years later, Sapir's Sino-Dene research had to vie for time with his many other involve- ments. The evidence of his manuscripts is that he did little further with the larger historical questions, although he continued the serious study of Tibetan and Chinese. Certainly his publications after 1925 show little direct concern with the Sino-Dene relationship, or even with Na-Dene, except insofar as the stock was represented in his general classification of Central and North Ameri- can languages (1929a). The major area in which he continued his comparative research was Athabaskan. "A Summary Report of Field Work among the Hupa, Summer of 1927" (1928i) and the Athabaskan portions of "The Concept of Phonetic Law as Tested in Primitive Languages by Leonard Bloomfield" (1931b) give brief glimpses of Sapir's progress in working out the intricate details of comparative Athabaskan phonology in the late 1920s. (In the latter paper, written in 1928-29, Sapir equates his Athabaskan work with Bloomfield's Algonquian in a methodological discussion.) Even here, however, his work slowed considerably in the following decade. He offered a course in Com- parative Athabaskan twice during his teaching at Yale (in 1931-32 and again in 1936), and from the students' notes that survive there is little evidence that Sapir's views had evolved much after 1930. A short survey of "Problems in Athapaskan Linguistics" found among Sapir's papers, apparently dating from about 1932 and published here for the first time, contains little not found in his earlier published work. The statement of "Linguistic Classification within the Northern Athapaskan Area" (1936i), which Sapir provided Cornelius Osgood for inclusion in his ethnographic survey of the Northern Athabaskan Indians, is also unremarkable. From a brief foray into Chasta Costa in 1906 through major field work on Navajo beginning in 1929, Sapir collected a very large corpus of descriptive data on several Athabaskan languages. It was the analysis of this material that came more and more to occupy Sapir's attention after 1925, much of it done in collab- oration with his students, Fang-Kuei Li and Harry Hoijer (also, more briefly, Mary Haas), and with his Navajoist colleague. Father Berard Haile. Only frag- ments of this work saw print during Sapir's lifetime, and some has remained unpublished to the present day. Six: Athabaskan and Na-Dene Languages 25 The materials from Sapir's Sarcee field work in the summer of 1922, his first extensive synchronic study of an Athabaskan language, have been more fully published than most later materials. Sapir himself, in addition to preparing a largely comparative paper on the Sarcee tone system (1925f, see above), had nearly completed a volume of Sarcee texts when he left Ottawa in 1925. This manuscript is being published for the first time in Volume XIII of The Collected Works. Also in that volume are reprinted two studies based on Sapir s mate- rials: Fang-Kuei Li's "A Study of Sarcee Verb-Stems" (1930), originally written as a masters thesis at the University of Chicago under Sapir's direction, and Harry Hoijer and Janet Joel's "Sarsi Nouns" (1963). Sapir also wrote two short papers on ethnographic aspects of his Sarcee work, "A Note on Sarcee Pottery" (1923c) and "Personal Names among the Sarcee" (1924d); both are found in Volume IV. Sapir's Kutchin materials, collected from John Fredson at Camp Red Cloud, Pennsylvania, during the summer of 1923, are much more poorly represented in his published work. Sapir extracted the kinship terms for inclusion in Cornelius Osgood's Ethnography of the Kutchin (Sapir 1936h, printed in Volume IV), but otherwise published nothing of his Kutchin data. As with Sarcee, he had begun preparing a volume of Kutchin texts while still at Ottawa, and during the 1930s Mary Haas, as his research assistant, worked on a Kutchin stem list. In 1961-62 Victor Golla completed a preliminary stem list but did not publish it. The texts and a stem list are being published in Volume XIII of The Collected Works. The Anvik (Ingalik) notes that Sapir obtained from Thomas Reed, also at Camp Red Cloud in 1923, are far less extensive than his Kutchin materials. Essentially a wordlist, of no great descriptive or comparative interest, the material was never utilized by Sapir and is not published in The Collected Works. Sapir collected extensive Hupa data during a northwestern California field trip in the summer of 1927, during which he also worked more briefly on Yurok and Chimariko. He was accompanied by his Chicago student, Fang-Kuei Li, who carried out his own work on Mattole and Wailaki. (For Sapir's lively description of this trip see Volume IV [1927b].) Other than a short "summary report" of his linguistic findings (1928i) — containing, inter alia, the (to Sapir) distressing information that Hupa lacks a tonal system — and the Hupa data incorporated into papers on the comparative method in American Indian lin- guistics (1931b) and on the northern origin of the Navajo (1936f, sec below), Sapir published only one paper based on his Hupa work, a largely ethnographic study of Hupa tattooing (1936e) written for A. L. Kroeber's Festschrift. Sometime during the 1930s he began work on a volume of Hupa texts, with extensive ethnographic notes, but it was far from complete at the time of his death. The texts and notes, edited by Victor Golla, are being published in Vol- ume XIV of The Collected Works, together with a lexical index to Sapir s data. Sapir regarded his Navajo work, begun with a native speaker in Chicago in 1926 but largely carried out in the field in 1929 and in later collaboration with Father Berard Haile, as "by far the most extensive and important linguistic 26 yf American Indian Languages 2 research" he ever accomplished (Sapir to Boas, 12 April 1938, quoted in Krauss 1986: 166). The data he collected, particularly lexical and paradigmatic mate- rial, was extraordinarily rich, and his relationship with his principal consultant, Albert G. (Chic) Sandoval, was especially close. Sapir prepared a large collec- tion of texts for publication , and his correspondence indicates that he was plan- ning to write a full Navajo grammar. In the years immediately preceding his death, he was actively working with Father Haile, a Franciscan missionary and scholar (see Sapir's review of Haile's earlier work [1926f]), in preparing literacy and language teaching materials for Navajo (Krauss 1986: 164-166). Despite all of this activity, at the time of his death in 1939 Sapir had in fact published hardly anything based on his Navajo work. A short paper, "Two Navaho Puns" (1932d), and a comparative-historical tour dc force, "Internal Linguistic Evidence Sug- gestive of the Northern Origin of the Navaho" (1936f), nearly exhaust the list, except for two brief ethnographic notes, "A Navaho Sand Painting Blanket" (1935b) and "A Note on Navaho Pottery" (with A. G. Sandoval, 1930), and a newspaper article describing the circumstances of the 1929 field trip (1929c). A volume of Sapir s Navajo texts was seen through the press in 1942 by Harry Hoijer, Sapir's principal Athabaskanist student, who much later published a Navajo grammar based on Sapir's materials as The Phonology and Morphology of the Navajo Language (Sapir and Hoijer 1967). These two publications are reprinted in Volume XV of The Collected Works, together with extracts from the voluminous correspondence on Navajo linguistics that Sapir and Father Haile carried on between 1929 and 1938 (Berard Haile Collection, University of Arizona Archives, Tucson). Besides work directly attributed to Sapir, the influ- ence of Sapir's materials and interpretation is strong in at least two other pub- lications: Father Haile's Learning Navaho (1941-48), the first volume of which incorporates many of Sapir's insights into Navajo phonology; and Harry Hoi- jer's A Navajo Lexicon (1974), which faithfully reproduces the organization of Sapir's lexical files. NOTES ON CHASTA COSTA PHONOLOGY AND MORPHOLOGY. INTRODUCTION. In a large part of southwestern Oregon and contiguous territory in northwestern CaUfornia were spoken a number of apparently quite distinct Athabascan dialects. The terri- tory covered by tribes or groups of villages speaking these dialects embraced not only a considerable strip of Pacific coast^ but also much of the interior to the east (Upper Umpqua and Upper Coquille rivers, lower Rogue river, Chetco creek and Smith river) ; some of the tribes (such as Tolowa and Chetco) were strictly coast people, others (such as Galice Creek and Umpqua or J\kwa}) were confined to the interior. While some of the Athabascan dialects spoken south of the Klamath in California, particularly Hupa and Kato, have been made well known to students of American linguistics, practically nothing of linguistic interest has as yet been published on any of the dialects of the Oregon- California branch of Pacific Athabascan. It is hoped that the following imperfect and fragmentar}'- notes on one of these dialects may prove of at least some value in a preliminary way.^ ' Outside of a few points in southern and southeastern Alaska (Cook Inlet, mouth of Copper river, Portland Canal) this is the only region in which Athabascan tribes have found their way to the Pacific. * My ^ denotes nasalization. * The material for these notes was secured in a very incidental manner. While the writer was at work on Takelma in the latter part of the summer of 1906, he was living with Mr. Wolverton Orton, a full-blood Chasta Costa Indian. At odd moments Mr. Orton and the writer whiled away the time with Chasta Costa. (273) 28 VI American Indian Languages 2 274 The Chasta Costa (or Cis/ta qlwAs/ta) Indians, now gathered in Siletz Reservation in western Oregon, formerly occupied part of lower Rogue river; between them and the coast were other Athabascan tribes or villages of practically identical speech, above them to the east were the unrelated Takelma.'' Among these tribes of nearly or quite identical speech were the YiV^/gwl or Euchre Creek people, the Tee' /me dA/ne or "Joshuas" of the mouth of Rogue river, the Du/Vil dA/nl, the Ml/klu/nu"* dA/nl, and the GwA/sd. All these formed a linguistic unit as contrasted with the coast people {d/yds/ta "lower tribes") or, as they are now commonly called by the Indians of Siletz, "Sol Chuck" Indians, a Chinook Jargon term meaning "salt water, coast" people; the dialect of these coast tribes was probably identical to all intents and purposes with Chetco. While Chasta Costa and Coast Athabascan are thus more or less distinct, they seem to have been mutually intelligible without very much difficulty, the coast dialect sounding merely somewhat "strange" and "drawn out" to a speaker of Chasta Costa. At least three other Athabascan dialects of this region, however, seem to have differed so much from Chasta Costa as to be but partly understood, if at all^ by speakers of the latter; these are Upper Umpqua, Upper Coquille, and Galice Creek. * It has already been pointed out (American Anthropologist, N. S., 9, p. 253, note 2) that there is reason to believe that J. O. Dorsey was incorrect in assigning the Chasta Costa villages above those of the Takelma (see his map in Journal of American Folk-Lore, III, p. 228). On p. 234 Dorsey gives a list of Chasta Costa villages. Six: Athabaskan and Na-Dene Languages 29 PHONOLOGY. Vowels. The vowels of Chasta Costa are a, a, e (open as in Eng- lish met), e (long and open), o (close as in German Sohn), 0, u (apparently variant of o), u, i (generally open), I, and A (like u of EngHsh hut)\ 6 (short and open as in German voll) sometimes occurs after velars as variant of o {sxo/ld "five," cf. Hupa^ tcwo/la), a (as in English hat) occurs after velars as variant of e {tsxci/xe "child," cf. Carrier^ cezkhehkhe "chil- dren"). Vocalic quantity is of considerable importance in Chasta Costa, not so much etymologically as phonetically. On the whole, long and short vowels interchange on regular mechan- ical principles; open syllables (that is, syllables ending in a vowel) with long vowel regularly shorten this vowel when the suffixing of one or more consonants to the vowel makes the syllable closed. Examples of a thus varying with a are: do/ydc/tla "I won't fly;" do/ydt/tla "we won't fly" (cf. dd/yd/t!a "he won't fly") dad /da "he is sitting down" (cf. da/ 6 Ad /da "I am sitting down") tcIdsL/se "he cries;" tcldch/se/Ve "I shall cry" (cf. tcla/Bil/se "you cry") tc!a/ydsL/se "they cry" (cf. tc!a/yd/6il/se "we cry") nac/tlb "I swim" (cf. nd/tcll/tlo "you bathe") ' Hupa examples are taken from P. E. Goddar^l, "The Morphology of the Hupa Language," Univ. of Cal. Publ. Amer. Arch, and Ethn., 3. * Carrier examples are taken from Rev. A. G. Morice, "The D^n^ Languages," Trans- actions of the Canadian Institute, I, pp. 170-212. (275) 30 ^f American Indian Languages 2 276 An example of c shortened to e is: 7ies/ts!Al/l "I am seen" (cf. lie' /ts!Al/l "he is seen") Original long vowels may lose their quantity even in an open syllable, provided they are immediately followed or preceded by a syllable with relatively strong accent. Such are tela-, na-, and 7ie- in: tc!a/ya/dil/se "we cry;" tc!a/ydsL/se "they cry" ne/7id/ts!Al/l "we are seen" Id na/dit/t!d "don't bathe;" {na/dit/t!d is phonetically enclitic to strongly accented ld\ contrast 7id/dit/t!d/Ve "you will bathe") In general, however, stress accent cannot be said to be particularly well marked in Chasta Costa." Each syllable is a fairly well-defined phonetic unit tending to hold its own against others, so that an approximately level accentual flow with but few peaks results. Such writings as 7id/dit/t!d and tc!d/dil/se, with apparent accent preceded by long vowels, are doubtless but imperfect renderings of forms with level stress on first and second syllables (they might perhaps better be written nd/dU/t!d and tc!a/Bll/se with secondary accent on second syllable). It does not seem that every vowel in an open syllable is organically long; thus e in future -Ve and in -de of Vwl/de "everything" is regularly short. Many such cases are, however, probably only apparent, the short vowel being followed by a glottal stop; thus plural ya- of ya/dAl/ni "they make a sound" should doubtless be ya-. Short a of closed syllables is regularly reduced from long a; original short a becomes /i in a closed syllable. Examples of A thus dulled from original a are: VAc/yAc/Ve "I shall go" (cf. t't/dic/ya "I go;" -yAc = Hupa -yauw) ' Weak stress accent seems characteristic of Athabascan generally. Father Moric6 goes so far as to say, "there is no accent in Den^" {op. cit., p. 173). Six: Athabaskan and Na-Dene Languages 31 277 nd/xAn/do "eight, two less" {;nd/xA- = Kato'^ nqk/ka" "two") do/na/yAct/xwl "I do not vomit" (cf. na/yd/dAdt/xwf "I vomit") VAl/dAc "he runs" {-dAc= Hupa -dauw) jAn/na/'Ac "he will bring" {-'ac = Hupa -aiiw) Ve/A7i/yit/lAl "we are sinking" (cf. Ve/nit/lat' "we drown;" Hupa -lat, -la "to float") Not to be etymologically confused with this a is inorganic A. Whenever a consonant is not followed by a definitely determined vowel and yet, for some reason or other, is not phonetically appended to the preceding syllable, it must begin its own syllable and takes an inorganic, in other words ety- mologically meaningless, ^ -vowel after it. This syllable may either be .completed by a consonant of etymological value (such as first person singular c, verb class signs I, t, I) never followed by a definite vowel or, if it is immediately followed by a syllable beginning with a consonant, this consonant is borrowed to complete the inorganic syllable (-/ closes inorganic syllables preceding d-, tl-, dj-, tc!-, ts!-, tdf-, tc'-, lI-), so that a doubled consonant results of which the first half is of no ety- mologic significance. In some cases, however, as before 7-. and in rapid speech generally, this inorganic consonant is not always distinctly heard; yet in syllabifying words Mr. Orton completed such inorganic syllables with a consonant with mechanical regularity. These syllables with inorganic vowel and consonant are characteristic not only of Chasta Costa but also of Hupa and Kato and doubtless other Athabascan dialects as well. The general phonetic tendency to speak in definite syllables and the further tendency to limit short vowels to closed syllables explain these characteristic Athabascan * Kato examples are taken from P. E. Goddard, "Kato Texts," Univ. Cal. Publ. Amer. Arch, and Ethn., 5, 65-238; and "Elements of the Kato Language," ibid., II, 1-176. ' -a- may be secondarily lengthened from -a-. 32 VI American Indian Languages 2 278 developments. The quality of the inorganic vowel varies for different Athabascan dialects; it is a{u) in Kato as well as in Chasta Costa, i {u before voiced or voiceless w, il or e before post-palatal ^-sounds) in Hupa, apparently e in Galice Creek, (s (probably identical with our a) in Carrier. Chasta Costa xAt/VAl/lal "they sleep" is etymologically equivalent to x/V/lal; X-, third person plural prefix, cannot stand alone and is therefore followed by a and / borrowed from -/'-, while -/'- (verb prefix V- reduced from Ve-, therefore not capable of combining with x- into xaV-) in turn needs a syllabifying a followed by / borrowed from -lal. Other examples of inorganic A, with and without following inorganic consonant, are: VAc/yAc/Ve "I shall go" {t"A- = V- reduced from Ve-) dd/ya/xAt/t!a "they won't fly" {xAt- = x-) nd/xAt/dAl/nic "they work" {xAt/dA- = x/d-, d- reduced from de-) dd/xAn/uAt/VAc "they go to bed" {xAn/nAt- = x/n-) t'e/An/yAl/Ul "he is sinking" (jaI- = y-) Many syllables with final consonant and a- vowel must be considered as radical or at least unanalyzable elements. In not all such cases is ^ a reduced form of a; where a seems a primary vowel, as shown by comparison with other Atha- bascan dialects, it seems best to consider it an organic element in the syllable, though it remains plausible that at last analysis it is but a reduced form of some fuller vowel. Thus, while -yAc has been shown to represent an original -yac (Hupa -yaiiw), -Vac contains a primary a, as shown by comparison with Hupa -tuw "to lie down" (ultimately -Vac is doubtless -V, reduced from -Ve, and suffix -c). Inorganic a sometimes becomes palatalized to i, though there is not enough material available to make it certain just when this change takes place. Examples of this secondary i have been found before c (but not before its developments s and 6) and 5 derived from tc (but not before original 5 or its Six: Athabaskan and Na-Dene tankages 33 279 development) when itself preceded by m, n, or 6 (preceding 7, however, tends to preserve a). Examples are: mis/ki'' "gull" (cf. Kato butc/k'ai') nic/ya "I come" {nic- = cessative 7z- and first person singular c) nic/dac "I dance" t'e/nic/lat "I drown" t'e/dic/ya "I go" {6ic- = durative 6- and prone minal c; cf. t'ed/ya "he goes" without vowel after 6) tclAy/ye/6ic/ya "I eat" ye/ die /I "I saw him" (cf. c/ycd/l "he saw me") Bid/ si "I let him" With -die- contrast -OaO- (both from original *-sac-) in da/ 6 Ad /da "I am sitting;" with -0zc/- contrast -sasI- (from original -*sAd- and -*sAd- respectively) in tcId/sAsi/se "I am crying." -yic- was heard in yd/yic/tia "I fly," but as this is an isolated example (contrast -yAc- in nd/da/yAct/tlo "I bathe" and -yAd- in yAd/Az "I sneezed"), it seems possible that this form was misheard for yd/yAc/tla. Besides -nic- also -nAc- is met with: dd/nAc/t'Ac "I go to bed" and nd/nAc/ An "I stop him;" it is probable that in these forms -nA- is a reduced form of ne- (cf. Hupa tcin/ne/tuw "she goes to bed") and thus not directly comparable with -ni- of -nic-. Unaccented A, itself reduced from a, has in one case {-yAc "to go") been found further palatalized to i: do /Vac/ yic "I'll not go," Id/n/yic "don't go!" (cf. t'Ac/yAc/t'e "I shall go"); this -yic contracts with directly preceding t'A- into -Vac: do/ Vac "he won't go.^^ Original Athabascan ai has in Chasta Costa become monophthongized to i. Examples are: l/gl "white" (cf. Kato i/gai) '" Should probably be miskH'. ''With ihis-t'Ac Kato ta/cac in dd/la/co* ta/cac "not anywhcrt' I went" (P. E. God- dard, "Kato Texts," Univ. Cal. Puhl. Amer. Arch, and Ethn., 5, No. 3, p. 182, 1. 17) is in striking agreement. 34 yf American Indian Languages 2 280 mis/kfi"- "gull" (cf. Kato biUc/k'ai') /zp'-^ demonstrative "that" (cf. Hupa hai) ail as organic diphthong seems to occur but rarely in Atha- bascan. If do "no!" (cf. Hupa dau) may be regarded as distinct from adverbial do "not" (cf. Hupa do), we would have an example of the parallel development of au to o in Chasta Costa. Certain contractions that take place between i of first person plural -it- and second person plural -o- with preceding vowels will be spoken of in discussing the pronominal prefixes. One of the most striking phonological characteristics of Chasta Costa is the disappearance of an original rj^^ or of its representative, nasalization of preceding vowel. Its former presence can always be proved by comparison with other Athabascan dialects that, like Hupa, still preserve it. In the case of all vowels but inorganic A nasalization has left no trace whatever, original q, (from arj), ^ (from et]), and i (from Iry) being reduced to a, e, and t; originally short vowels, on losing their nasalization and thus coming to stand in an open syllable, become lengthened, while originally long vowels in a closed syllable not only lose their nasalization but are shortened. Thus, a syllable si may represent an original sj (or sir]) or si (or sir]), while sil may go back as well to sll as to sjl. Examples of the absolute disappearance of an original 17 are: nd/xe "you paddle" {nd- = *nq-, cf. Hupa nun/ya "you are about") do/yCl/tla "you won't fly" {yd- = ^yq,-, cf. Hupa yum/ mas assimilated from *yun/mas "you are rolling over") tc!dl/se/t'e "you will cry" {tcfal- = *tc!q-l-; cf. tc!dd/se/t'e "I shall cry" with -c- "I" morphologically parallel to -.- "you") '* t is here shortened to i because of following glottal stop. •' i* denotes long I with weakly rearticulated parasitic i. Such "pseudo-diphthongs" sporadically occur in Chasta Costa in lieu of ordinary long vowels. '* i. e., ng of English sing. Six: Athabaskan and Na-Dene Languages 35 281 id/na/yat/xwl "don't vomit!" {yat- = *yqt-, cf. yd- from *yq- in na/yd/dit/xwl "you are vomiting") nel/l "you are looking at him" {nel- = *n^l- ; -I = -'?*, cf. Kato -in' "to see") It "dog" (original Athabascan */?, "^li-q; cf. Hupa Lzw, Montagnais I'in, Hare //'/w, Loucheux /'m, Carrier //, old form loe^'n}^) Nasalized inorganic 4 seems to have acquired a palatal coloring i\ this i then regularly developed to I in open, i in closed syllables. It thus often seems as though Chasta Costa I, i is the morphologic equivalent, for instance in second person singular forms, of Athabascan 77, an equivalence, as has just been shown, due to secondary phonetic developments. Examples oil < i < 4 are : Ve/Bi/ya "you go" {61- = *S4- ; cf. Hupa na/sin/ya "you are going about") ni/dac "you dance" {ni- = *W4-; cf. Hupa nin/yauw "go!") 3'w/u'w f/x/wi "you whistle" {di- = *d4-\ ci. Hupa da /din/ La "run!" assimilated from * da /din/ La na/tcll/ilo "you swim" {tcli-^'^'kylA-'}^ cf. Hupa na/kin/- yun "come eat!") yd/yi/tla "you fly" (7t- = *74-; cf. Hupa ye/win/ya "you are going in") yd/yi/tla "it flies" (7^- = *74-; cf. Hupa na/win/tau "it will settle down" assimilated from ^na/win/tau) Vi/lal "you are sleeping" (/'I- = */'4-; cf. Hupa tin/xaiiw/ne "you take along") verb stem -si "to make" (cf. Hupa -tcwiii) '^ Morice, op. cit., p. 210. Carrier has evidently undergone a development parallel to that of Chasta Costa. All northern Athabascan forms except Carrier (and Chipewyan) are taken from R. P. E. Petitot, " Dictionnaire de la langue Dcnc-Dindji^." '* k^l is "fortis" palatal k, Hupa ^1, Morice's q. 36 VI American Indian Languages 2 282 Examples, in closed syllables, of i, i. e., k plus bilabial/. '' Petitot's tch is our Ic. ** Petitot's p is 7. 40 VI American Indian Languages 2 286 l/do "yellow, green" (of. Montagnais del-thop "yellow;" Hare de-kfwoy "yellow," Hupa u't-tso "green;" Kato L-tso "blue")' In some cases ts seems to have become s: se "stone" (cf. Kato se\ Hupa tse; Montagnais the; Hare kfwe; Loucheux tchi; Carrier tse) As might be expected, Athabascan ts! has regularly become td! in Chasta Costa: dd/de/dil/tB!i "we are sitting" (cf. Hupa na/ya/del/tse, i. e., -tsle, "they lived as before") teiAd/dd "story" Athabascan c is normally preserved as such (e. g., ct "I"). However, it is assimilated to s before 5 and tsl: s/tsli/de "my sickness" {c- "my") nes/tsUi/l "I am seen {-c- "I") As/se/t'e "I shall cry" (from *ac-) s I tsl An I na I 'Ac "he will bring it to me" (c- "me") Assimilation of *sac to sas has taken place in: tc!a/sAsL/se "I cry" (cf. tddci/se/Ve "I shall cry") Original *sAc>*sic, however, regularly developed to die: tclAy/ye/eic/ya "I eat" Original *sac, after being assimilated to *sas, regularly shifted to OaB, unless, as we have seen, it was protected by immediately following I: da/ 6 AS /da "I am sitting" (from *dd/sAc/da) Ve/BAB/lal "I have been sleeping" (from *re/sAc/lal) tcfed/t.'d "I swim across" (probably misheard for tcIe/dAd/tfo) Original s, when immediately following c, also causes it to assimilate; ss, which thus results, is then regularly shifted to 66: yd/yA6/6el "I threw" (from '^yd/yAc/sel) Six: Athabaskan and Na-Dene Languages 41 287 Athabascan tc (sometimes /cw?) is not retained in Chasta Costa, but appears regularly as s: l/sAk "red" (of. Kato L/tcik\ Loucheux ditssigY^ mis/k{!)i(') "gull" (of. Kato butc/k'ai*) sd'/wAs/tsIe "sandhill crane" (cf. Applegate Creek tcd'/- wdc/tc{!)e) -si "to make" (cf. Hupa -tcwin; Kato -tcl; Chipewyan -tsl'') -se "to cry" (cf. Chetco -swe; Hupa -tcwen; Kato -ke'; Carrier -ssd) Chasta Costa sx is found in: sxo/ld "five" (cf. Hupa tcwo/la; Chipewyan sa/so/la/yai^) Athabascan tc! remains, tc! often being shifted, however, to ts! (or 5'22) : tele- verb prefix "across the water" (cf. Hupa tee-, i. e., td.e-, "down to the beach, out of the house;" Kato tce-\ Chipewyan ts'e- "to a body of water") -ts!An "toward, to" (cf. Hupa -tcin, i. e., -tdit); Kato -tc'uiV; Chipewyan -ts'un) ts!i/de "sickness" (cf. Loucheux tssik, i. e., tsUk) -s'at' "to be hurt" (cf. Hupa -teat, i. e., -te!at, "to be sick, to become ill") There is still another set of sibilants in Chasta Costa, which go back to original palatalized (anterior palatal) ^-sounds (gy, k^, k^!). In Kato, Navaho, Apache, Chipewyan, and other Athabascan dialects, as in Chasta Costa, these have become affricative sibilants, without, however, falling together, as a rule, with the original Athabascan te- consonants. In Chasta Costa, k^ has become te\ k^! has become tc! (this te! *> Petitot's tss is our tsl. " Chipewyan forms are taken from P. E. Goddard, "Analysis of Cold Lake Dialect, Chipewyan," Anthr. Papers Amer. Mus. Nat. Hist., vol. X, pt. II. Chipewyan forms taken from Petitot are referred to as Montagnais. ^- It is quite likely that tcl and ts! are here merely auditory variants of ts! (i is mid- way between 5 and c). In Kato tc' , ts' and 5' also interchange. 42 VI American Indian Languages 2 288 does not vary, apparently, with ts!) ; for g^ I have no examples. Chasta Costa and Chipewyan are largely parallel in their development of Athabascan ts, tc, and k^ sounds: Athabascan Hupa Chasta Costa Chipewyan dz d^, d ts ts e te, d ts! tsl td! te!, 6' dj dj dz tc tc(w) s ts tc! tc! ts!, ts! ts! g' g' dj ky k^ tc tc kyf ky! tc! tc! There are thus three distinct series of sibilant affricatives (and of sibilants) in Chasta Costa and Chipewyan, none of which is in direct accord with the original Athabascan sounds; Hupa, it is highly important to note, reflects the original sounds almost exactly .^^ Carrier, it would seem, has also preserved the ky- series. Examples of Chasta Costa tc' from original ky are: dL,tcd/yi "big thing" (cf. Hupa -kya/o "large;" Kato -tcay, -tea' "to be large") ''In his "Analysis of Cold Lake Dialect, Chipewyan," Goddard treats Chipewyan ts and tc as though they were one sound corresponding to Jicarilla and Navaho tc (p. 86). Examination of the various illustrative forms scattered throughout the paper, however, soon convinces one that Chipewyan ts, dz, and ts! correspond respectively to Hupa, Jicarilla, and Navaho fc(w), dj, and tcl; whereas Chipewyan tc, dj, and tc! correspond respectively to Southern Athabascan ts, dz, and ts! and to Hupa k^, g^, and k^!. Thus, the Southern Athabascan ts- sounds represent both original ts- sounds and k^- sounds; perhaps there is a phonetic difference that does not come out clearly in the orthography. As for Kato, Goddard finds no difference between tc- sounds that go back to original tc- sounds and those that correspond to Hupa k^- sounds (" Elements of the Kato Language," pp. 16, 51). However, deictic tc'-, corresponding to Hupa tc!-, varies with ts' and s', thus suggesting ts! as the true sound ; on the other hand, tc'- (to indicate indefinite third personal object) corresponding to Hupa k^!- occurs consistently as tc' (contrast examples of tc'-, ts^-, s'- on p. 50 with those of tc'- on p. 51). It seems plausible, then, that in Chipewyan, Chasta Costa, and Kato original k^- sounds became true tc- sounds, while original tc sounds were shifted to ts- sounds (which are apt to be heard as either ts- or tc- sounds). Six: Athabaskan and Na-Dene Languages 43 289 -tc'u augmentative suffix (e. g., W/tc'ii "horse," literally "big dog") (cf. Hupa -kyo; Kato -ted) Examples of tc! going back to Athabascan k^I are: dAn/tdi "four" (cf. Hupa dink, i. e., di-qkyf) sic! At I de "seven" (cf. Hupa xo/kit, i. e., -k^Ht) tddsL/se "he cries" (cf. Hupa kya/teL/tcwii "it cried, i. e., k^Ia-) tc!- verb prefix indicating indefinite object (cf. Hupa k-, ky-, i. e., k^!-; Kato tc'-) Athabascan possessed sonant sibilants (2, j) and sibilant affricatives {dz, dj). Of these sounds z has been found in Chasta Costa -Az "to sneeze;" dj is illustrated in several forms, but, as we shall see in a moment, does not in these go back to Atha- bascan dj. dz has not been found, though it may exist. 7, as in Kato and Hupa, has become c: ia/cAn "black" (cf. Hupa Lu/hwin<*-cin; Kato i/cmi^; Jicarilla Ll/zl; Nav. Ll/jin; Chipewyan del/zun; Loucheux del-zen) Chasta Costa dj results from / (unaspirated) plus y: qlwAt/tc/At/dja "table" {<*qIwAt/tc!At/ya "whereon one eats;" -ya "to eat") ya/da/yit/dja "we are ashamed" {<*ya/da/yit/ya; cf. yAc in ya/dAcl/yAc "I am ashamed") Of the lateral consonants, only three (/, I, and lI) have been found in Chasta Costa. Original dl may have been pre- served also, but Athabascan did was heard rather as / (unas- pirated) plus Id: yAct/lo "I laugh" (cf. Chipewyan -did, -dldk' "to laugh") -/- is very probably third modal -/- here; while -did really appears as -Id. After c and s, I becomes /: nd/dAcl/nic "I work" (cf. nd/dAl/nic "he works") nd/xwAcl/ye "I play" (cf. nd/xwAl/ye "he plays") qlwAt/dasi/nd "it was lying on it" 44 VI American Indian Languages 2 MORPHOLOGY. Pronouns. Independent personal pronouns: ci "I" ne "we" (probably contracted from *ne/he; cf. Hupa ne/he) nAn "you" nd/ne "you" (plur.) yu "he, that one" yu/ne, yun/ne "they, those" (really demonstrative) (really demonstrative) Examples of possessive pronouns are: cic/la "my hand" {cic is independent cl combined with possessive prefix c-\ literally, "I my-hand") nAn /la "your hand" (that is, nAn n-, "you your-hand") hi la "his hand" {hi is demonstrative) c/na/yd "my eyes" s/tsH/de "my sickness, I am sick" n/ts!l/de "you are sick" nd/ts!l/de "our sickness, we are sick" no/tsll/de/ha "your (pi.) sickness? are you (pi.) sick?" {-ha is interrogative) xd/ts!l/de "their sickness, they are sick" Many nouns, when limited by preceding possessive pronouns, suffix -e, as regularly in Athabascan. Thus, from mAn "house:" cic/mAne "my house" nAn/mAne "your house" A noun followed by another with suffixed -e is to be under- stood as genitively related to it. Examples are: dAne' III tele "person's dog" {ll/tde from ll "dog," with (290) Six: Athabaskan and Na-Dene Languages 45 291 voicing of /- to /-; cf. Hupa Lin "dog," xo/lin/ke, i. e., xo/lirj/kyfe "his dog") tAkAc^'^ hlci^/le "bowstring" (literally, "bow's string;" cf. Chipewyan Vul "rope," possessed form Vu/le) ga/yu ts!i/de "baby's sickness, baby is sick" As reflexive possessive is used xd/dAt- (with -d/dAt- cf. Hupa a/d- ; Carrier cedced-) : xd/dAt/lt/tc!e "his own dog" (used reflexively) Of demonstrative pronouns there have been found: hi* "that, he" (cf. Hupa hai, indefinite demonstrative and article); hi'/tli "that thing" yii "that one" (cf. Hupa yd "that") yu/ne, ytin/ne "those, they" m- "it" (cf. Hupa m-; Kato b-): niAl "with it" de seems to be used as relative in: de ucL/Ve "what I want" This element is perhaps demonstrative in force and related to Hupa de in ded "this," hai/de "this." Totality is expressed by Vwl "all, everything" (cf. Hupa a /tin "all"). Compounded with this element are: Vwl/de "everything" {-de is very likely related to Hupa di- in dl/hwd "something," dl/hwe/e "nothing") do I Vwl/de "not everything" Vwi/dAn "everywhere" (literally, ^'all-at;" cf. Hupa a//iw/- din "every place") Nouns. Primitive non-descriptive nouns, as in all Athabascan dia- lects, are relatively frequent in Chasta Costa. Monosyllabic nouns are: Body Parts. la "hand" (cf. Hupa -/a; Kato -/a*) ** Probably to be understood as lAk/gAc. 46 yi American Indian Languages 2 292 hwd "foot" (cf. Kato -kwe*; Chipewyan -ke) 61 "head" (cf. Kato -si'; Chipewyan -61, -t6i) -ya "hair" (in 6 Ay a "head-hair;" cf. Kato -ga' "hair;" Chipewyan -ca, i. e., -ya) Animals. tdac "bird" ll "dog" (cf. Hupa Lin; Chipewyan Ll) Natural Objects. se "stone" (cf. Hupa tse; Kato se) cd "sun" (cf. Hupa hwa; Kato ca) lAt "smoke" (cf. Hupa hit; Kato uit) Culture Objects. mAn "house" (cf. Hupa diminutive min-tc "hut") Llel "matches" (originally doubtless "fire-drill;" cf. Chipe- wyan L'eL "fire-drill") gd6 "camass" (cf. Hupa kos "bulbs") LloH-e "(its) string" (cf. Hupa Lol "strap;" Chipewyan L'ilL "rope") Primitive, at any rate not easily analyzed, nouns of more than one syllable are: Persons. dAtt/ne, dAne' "person, man" (cf. Chipewyan de/ne, dun/ne; Carrier tcene) ts!d/xe "woman" (cf. Carrier tsekhe; Kato tc'ek) dis/ne' "male" (with -ne cf. probably -ne of dAne') sd/sAs "white man" tsxd/xe "child" (cf. Carrier oe'zkhehkhe; Kato skl-k "boys, children") kel/'e "boy" (perhaps misheard for k!el/'e; cf. Kato k'il/lek "boy") gd/yu "baby" Body Part. na/y& "eye" (cf. Hupa -na; Kato -7ia*\ Chipewyan -na/ca, -na/ce) Six: Athabaskan and Na-Dene Languages 47 293 Animals. BAyAl "grizzly bear" (cf. Carrier sces-e^cel "brown or cross bear") mis/ki!)i{') "gull" (cf. Kato biitc/k'ai') dis/dac "fawn" dA/meV /ke "pelican"^^ nat/qH "duck" (cf. Kato nd'/qH'') ml/tc'd/tsUl/fil "deer" Ve/q!d/lec/re "mink" sd'/wAs/ts!e "sandhill crane" (cf. Applegate Creek tcd'/- wdc/tc{!)e) tclal/tchis/dje "ruffled grouse, 'pheasant' " 6d/gi "kingfisher" dAs/nAl "red-shafted flicker" teiAB/nd/yal/tetde ' 'hummingbird' ' 0c I id e "bluejay" nd/ts!d/le "horned lark" so's/ga/ga "robin" ts!d/ts!uk "wren" kAsis "barn swallow" ga/lal/'e "crow" Many of these animal names, as well as some of those that follow, are probably descriptive verb forms that have become stereotyped. Plants. tcA/pd/yii "flower"'" mt/tlal/tdAd "arrow- wood" do'/de "tar- weed" (probably compounded with Atha- bascan do' "grass;" cf. Hupa Lo/da-itc "an herb") tc!Al/yat/ts!s "sunflower (?)" "This word is humorously used to refer to Democrats, Democrat and dA/mel' /ke exhibiting some similarity in sound. '* This form was obtained independently. " This word is remarkable as containing p, a sound that is normally absent in Atha- bascan. 48 ^f American Indian Languages 2 294 dAl/si "pine" (cf. Kato dul/tcik "yellow pine," from -tclk "red") nd/ife "pine-nut" dA/nAc "manzanita" (cf. Hupa din/nuw; Kato tun/nuc "manzanita berries;" Galice Creek de/rec) mAt/tcfi "cat-tail" cAc/dd' "oak" Culture Objects. xAnAd "canoe" at/ tea "pipe" tclA/BA/gAl "sandstone arrow-shaft scraper" tA/kAc "bow" (probably VAk/gAc\ cf. Kato gqc "yew") *q!d/xA6 "arrow" det/t!e "arrow-point" Abstract. tdlAd/dd "story" yA/wls "whistling" (cf. Carrier ytiyuz "whistling," as noun) tsli/de "sickness" (used with possessive pronouns to indi- cate "to be sick") Several animals are designated by words ending in -tc'u, an augmentative suffix, "big" (cf. Hupa and Kato animal and plant names in -kyo and -ted respectively). Such are: ii*/te'u "horse" (literally, "big dog;" cf. Chipewyan Lln/teo) xd'/te'ii "goose" (cf. Applegate Creek k'q'/te'u. These words are formed from Athabascan xa: Chipewyan xa "goose;" Kato ka') dAe/te'u, des/te'ti "grouse" (cf. Kato due /ted, dus/ted "grouse") Vet/mo/tc'u "pigeon" eu/dc' /tc'u "bald eagle" BAe/dA/li/tc'u "owl" H/tcIe/tc'u "red-headed woodpecker" ga/sd'/te'ii "raven" Six: Athabaskan and Na-Dene Languages 49 295 Nouns ending in -til or -tllni denote "one who has so and so." -ni is, Hkely enough, related to -ne of dAn/ne "person;" -ne or -n is found in many Athabascan dialects as suffix denot- ing "person." Examples of -t!i{ni) are: U/t!l/ni "dog-owner" do I ail till ni, do J at I ill "bachelor" (literally, "not-wife-hav- ing-person." do- "not;" at-, i. e., ail "wife," cf. Hupa ui "wife," Kato aV "sister") Examples of noun compounds consisting of two noun stems are: 6a /yd "head-hair" (shortened from 61 "head" and -ya "hair." Cf. Chipewyan 6l/Ga) ga/lal gwd/yu "red-winged blackbird" (literally, "crow('s) brother-in-law." With this cf. Chipewyan da/tsa/- tcel/le "a small crow," literally, "crow younger- brother"3«). An example of a compound noun consisting of verb and noun is: aI/Az dAn/ne "sneezer" (literally, "he-sneezes person") An example of a compound noun consisting of noun and adjective is: iclac l/66/e "bluebird" (literally, "bird blue") A characteristic type of noun in Athabascan is formed by verbs which, while remaining strictly verbal in form, are used to refer to objects, in other words, are logically nouns. As has been already noted, several nouns of more than one syllable listed above as unanalyzable are doubtless, strictly speaking, verb forms. Quite clearly verbal in form are: nd6/Ll6 "paper" (cf. nal/ilo "he writes") ql'wAt/da6t/gAc "table-cloth" (literally, "it lies or is thrown ^ Goddard, op. cit., p. 1 10. 50 VI American Indian Languages 2 296 down on top;" cf. Hupa -k2as, i. e., -gas, "to throw," and wes/kas "it lay"^^) qlwAt/tclAt/dja "table" (literally, "thereon it is eaten") mAl/Ve/tc!At/ts!Al/lec "smoking materials" (hterally, "there- with it is smoked") Numerals. 1. la, Wjca (cf. Hupa La\ Kato La/ ha') 2. nd/xi (cf. Hupa nax] Kato nqk/ka'); nd/xi la "two hands" 3. t'd/yi (cf. Hupa tak, i.e. Vak!; Kato tak'; Chipewyan ta, ta/ce 4. dAn/tdi (cf. Hupa dink, i. e. dirik^!; Chipewyan dl/Gl) 5. sxo/ld (cf. Hupa tcwo/la; Chipewyan sa/so/la/cai*) 6. k'wAs/Vd/ne (cf. Hupa xos/tan) 7. slclAt/de (cf. Hupa xo/kit, i. e. -K^z7) 8. nd/xAn/do ( = "it lacks two, two less") 9. Ian/ do ( = "it lacks one, one less") 10. hwe'/de Of numeral adverbs there were recorded: Idt/dAu "once" (cf. Hupa na/din "twice," min/ Lun/din "ten times") la/me/q!e/ca "in one time" Adjectives. Of adjectives, or verb stems with adjectival significance, there have been found: wAs/xe, wAs/xd "good;" wAs/xe li "dog is good" txAs/xe/la "rich" {-la is verbal sufhx) du/An/de "bad" (evidently verbal in form, du-, do- is negative; -de probably misheard for -t!e "to be, exist;" cf. Hupa un/te, i. e. An/t!e "there is") "Goddard, op. ciL, p. 281. Six: Athabaskan and Na-Dene Languages 5 1 297 AL/tcd/yt "big thing" (cf. Hupa -kya/o "large;" Kato -tcac) l/gi "white" (cf. Hupa -L/kai; Kato -h/gai) la I c An "black" (cf. Kato -L/cun^\ Chipewyan del/zun) l/sAk "red" (cf. Kato -h/tclk) l/do "yellow, green" (cf. Hupa lit/tsd "green;" Kato -L/tso "blue") "White," "black," "red," and "yellow, green" are characterized by prefixed I {a)-, which is common as adjectival prefix also in other Athabascan dialects. Adverbs. Adverbs of place are: xun "there" (cf. Hupa third personal pronoun xon?): xiin t'e/Bi/ya "there you go" hV xiin Ved/ya "there he goes" txun/la "where?": txun/la VejBi/ya "where are you going?" do/dAt "nowhere" (cf. Hupa -dit- in hai/dai/dit/din "where;" do- is negative) (i^y^/ge "up" (cf. Hare /eg^): dAk/ge Bicl/Vdl "I kicked him up" md'^/dAn "on edge" (-dAn is postposition "at;" md^-K *mq-<*marj-; cf. Hupa ntL/man "each side") Adverbs of time are: xat "then" (cf. Hupa xat "yet, right") xd "quickly" (cf. Hupa xa "yet") xun/de "tomorrow" (cf. Hupa yis/xiln/de "tomorrow"): xun/de do/wa/yAc/l "I'll see him tomorrow" xun/de td/Ad/dd nAl ndcl/si "tomorrow story to- ycu I-shall-tell" xun/de t'Ac/yAc "tomorrow I'll go") 52 VI American Indian Languages 2 298 t'wt/dAn "always" (literally, "all-at"): Vwi/dAti t'Al/dAc "he always runs" t'wt/dAn As/se "I always cry" xAi/tsH/dAn "this evening" (doubtless misheard for xal!-', -dAn is postposition "at." Cf. Hupa xu/Le "in the night"): xAL/tsH/dAn do/wan/yAc/l "I'll see you this evening" Modal adverbs are: do negative (cf . Hupa do) : do/t'Ac "he won't go" do/rAc/yic "I'll not go" do/yd/t/a "he won't fly" dd/As/se "I'm not crying" do/nd/dACL/nic "I'm not working" dd/yAc/l "I didn't see him" do/ned/l "I'm not looking at him" dd/ucL/t'e "I do not want" do/na/yAct/xwl "I do not vomit" . la prohibitive: Id "don't!" Id/n/yic "don't go!" la/yi/i "don't see him!" Id/nd/xwil/ye "don't play!" ld/na/dit/t!d "don't bathe!" Id/na/yat/xwi "don't vomit!" doldd/qle "unable" dol LAn "not much" (cf. Hupa Lan "much," do/ian "little") do I wi I la "of course" (cf. Hupa ddii "it is," he I don "at least") do lid emphatic negative (really verbal inform, "to cease;" cf. Hupa -lan, -lun with negative prefix do- "to quit, leave, desist") : do /Id c/yl/i "you didn't see me" Six: Athabaskan and Na-Dene Languages 53 299 cd'*/djl "all right" (cf. Hupa nil/hwdn/^x "properly"?) cAl/q!we "to be accustomed to": cAl/qlwe na/dAct/t!d "I'm used to bathing" t!V/xun "to keep on:" tlV/xun ne/cAl/l "he keeps looking at me" BAk/gwe "in fragments" ho future prefix (more properly intentive) : hd/ya/yic/t!a "I'll fly" hd/tc!AsL/se "he wants to cry" ho I ill I yit/lo "stop laughing!" do/wa future prefix (probably with dubitative coloring) : do/wa/c/yl/i^/Ve "you'll see me" s/ts!l/de do/wa/ aI/W "I'll get sick" (Hterally, "my-sickness will-become") do/wa/ncL/yan/nAl "he will upset them" dd/wa/it'dt/nTii "they will go to pieces" Postpositions. Athabascan is characterized, among other features, by the use of a considerable number of postpositional elements of chiefly local force. They are appended to nouns or pronom- inal, numeral, or adverbial stems; less often to verb forms, in which case they have subordinating force. Chasta Costa examples are: -dAn "at" (cf. Hupa -din): xAL{!)/tsH/dAn "this evening" Vwi/dAn "everywhere" (hterally, "all-at") Idt/dAn "once" (cf. la- "one") md^/dAti "on edge" al/dAc/ni/dAn "when I tell him" (hterally, "I- tell-him at")*« "Similarly in Hupa -miL "when," as verb suffix, is doubtless simply pronominal ■mi- plus postposition -/. "witli." 54 VI American Indian Languages 2 300 4 "with, to" (cf. Hupa -l; Kato -l): xAUAd/l 7idc/xe "I paddle canoe" (literally, "canoe- with I-paddle") tdlAd/dd riAl ndcl/sl "I tell you story" (literally, "story you-with I-make") td/Ad/dd caI na/yesL/si "he tells me story" (literally, "story me-with he-makes") mAl/Ve/tc!At/ts!Al/lec "wherewith it-is-smoked, ma- terials for smoking" {mA-l- "therewith;" cf. Kato buL "with it;" Hupa mil "with, in") This same -I is probably also found attached to verbal prefix a- (used in verbs of saying) : al/dAc/ni/dAfi "when I tell him" (cf. Hupa ah/- tcit/den/ne "he talked to") -tslAn "toward" (cf. Hupa -tcin "toward;" Kato -tc'un* "to, toward"): s/tslAn/na/'Ac "to-me he-will-bring-it" -me "in" (cf. Hupa -me "in;" Kato -hV "in"): mAn/me "in house" -me/q!e "in, around in" (compounded of -me and -q!e\ cf. Chipewyan -k'e "on"): mAn/me/q!e "around in house" la/me/q!e/ca "all in one time" (cf. la, Wjca "one") Verbs. As in other Athabascan dialects, the typical Chasta Costa verb consists of one or more adverbial prefixes, which may be followed in order by a deictic or third personal ele- ment, a first modal prefix, a second modal element, a first or second person subjective element, and a third modal element or "class" sign; these, not all of which need of course be present, are then followed by the Terb stem itself. The stem often ends the verb form, but may be followed by one or more enclitic elements of modal or syntactic force. The verb form is fre- Six: Athabaskan and Na-Dene Languages 55 301 quently preceded by an adverb or postposition which, while best considered as a non-integral part of the verb, forms a rather close syntactic unit with it. A pronominal object, if present, comes after an adverbial prefix but before a first modal element. Thus, the verb form Vd/yd/Bot/nd/hd "do you (plur.) drink?" consists of seven elements: Vd-, an ad- verbial prefix referring to water; yd-, a second adverbial ele- ment; B-, a second modal element of durative significance; -6-, second person plural subjective pronominal element; -/-, a third modal element, probably intransitive in force; -nd, verb stem "to drink;" and -hd, an enclitic interrogative element. The various elements that go to make up verb forms will be taken up in the order indicated. Adverbial Prefixes, a-, a-, 'a- used with verbs of say- ing, doing, and being (cf . Hupa and Kato a-) : d/dJAn "he says" al/dAc/ni/dAn "when I tell him" (for -1-, see under Postpositions) dd/dAt 'An/ tie "there is not anywhere" This a- is probably equivalent to an indefinite object, "some- thing," indicating what is said or uttered without definitely referring to it. This comes out rather clearly on comparison with a form like yu/wls dAcl/ni "I whistle" (literally, "whistling I-utter"), where no indefinite object a- is required, what is uttered being specifically referred to by yu/wls "whistling." That a- is somewhat in a class by itself as compared with other adverbial prefixes is indicated by its being followed in forms with indirect object by postpositive -/-. yd-, ya- "up (in the air)" (cf. Hupa ya-\ Kato ya^-): yd/yAcl/gAd "I climb" yd/ y Ad /del "I threw" yd/yic/lla "I fly" It is not clear what significance is to be attached to ya- in: ya/dAcl/yAc "I am ashamed" ya/da/yit/dja "we are ashamed" 56 VI American Indian Languages 2 302 ye- "into enclosed space (including mouth)" (cf. Hupa ye-] Kato ye'-, yV-) ye/ydt/ne/la "he bit it" da-, da- "sitting or lying on something above ground" (cf . Hupa and Kato da-) : da/ 6 Ad /da "I am sitting down" dd/de/dil/tB!i "we are sitting down" dd/nAc/Vac "I go to bed" q!wAt/dadt/gAc "it lies thrown down on top," i. e. "table-cloth") qlwAt/dash/nd "it was lying on it" Ve- "in the water" (cf. Hupa ie-; Kato /e'-): Ve/An/yAc/lAl "I am sinking in the water" Ve/nic/lat "I drown" Vd- referring to water (cf . Hupa and Kato ta-) : Vd/yAct/nd "I drink" tde- "across a stream" (cf. Hupa tee- "out of;" Kato tc'e- "out of;" Chipewyan ts'e- "used of approach to a body of water") : tcle/Bii/tlo "I swim across" An- implies disappearance or undoing (cf. Chipewyan *a-, an- "away," implies "desertion or abandonment"): Ve/ An/yAc/lAl "I am sinking in the water" do/wd/ An/nd/yan/nAl "he will upset them" an- "back, hither" (cf. Chipewyan '5- \in-, ai- "back, toward home"): an/yi/al "come here!" tdd-, tela- of unknown significance (cf . Hupa kya-"^^) : tcld/ sAsh/se "I cry" (cf. Hupa kya/teh/tcwe "she heard it cry") tddsi/se "he cries" se'- used with verb of smiling: se' /yAt/lo "he smiles" (cf. yAt/ld "he laughs") ^' Goddard lists forms in kya-, i. e., k^ta-, under ky-; see op. cit., p. 90. It seems better however, to keep them apart. Six: Athabaskan and Na-Dene Languages 57 303 tclo- of unknown significance (cf . Hupa kyo-^^) : tcfo/yit/siL/la "he pointed with his finger" ne/tc!uc/lec "I'll bet you"« u- of unknown significance (cf. Hupa verbs in o-**) dd/ucL/t'e "I do not want" de/ucL/t'e "what I want" nd-, na- indefinite movement on surface of ground or water; horizontality (cf. Hupa and Kato na-): fiAn/ndd/yd/la "he went around it" s/tslAn/na/'Ac "he'll bring (it) to me" yAn/na/Ac/t'e "he will bring here" nd/ni/An "stop him!" nd/ya "is going about, living" i/t'i yAn/na/'d "he brags" (literally, "high, im- portant he-has") nd/xwAcL/ye "I play" nd/dAch/nic "I work" do/wa/nd/yan/nAl "he will upset them" ndcl Lib "I write" xatiaBU ndc/xe "I paddle canoe" nac/t!d "I swim, bathe" na/tc!il/de "you wash" nd/dAcl/de "I washed myself" nd/dAt/t!d/Ve "he'll bathe" nd/xAt/dAl/el "they'll bathe" tdlAd/dd caI nd/Bil/sl "story to-me "you-told, caused" na- "back again" (cf. Hupa and Kato na-), followed by third modal -/-: na/yd/dAdt/xwi "I vomit" yAn- of uncertain significance (cf. Hupa wun- "to pursue *^ Goddard lists forms in kyo, i. e., kyjo-, under ky-; sec op. cit., p. 90. Perhaps k^lo is compounded of k^l- and o-. ♦' This tclii- is probably better explained as deictic tcl- followed by future imperative M-; see note 86. "Goddard, op. cit., p. 115. 58 VI American Indian Languages 2 304 or seek something; to attempt something by per- sistent effort"): yAn/na/Ac/t'e "he will bring here" i/tH yAn/na/'d "he brags" (perhaps "important he-seeks-to-have") ne- of unknown significance: necl/l "I'm looking at him" (cf. ye/ die /I "I saw him") nes/tslAl/l "I am seen" xw{a)- of unknown significance: nd/xwAci/ye "I play" xwAcl/I "I believe (it)" xwAn/ne/diL/ya "you win" (see under first modal 7ie-) yd- ya-, (y) refers apparently to "mouth": ye/ydt/ne/la "he bit it" {yat- may, however, have been misheard for yAt-, with second modal 7-; see note 92) na/yd/6Adt/xwt "I vomit" id/na/yat/xwl "don't vomit!" t'd/yd/6it/nd "we drink" Vdlyit/nd "you drink" (or is y- here second modal prefix? Verbal prefixes of local force which are doubtless primarily postpositions and which are prefixed to adverbial prefixes proper are: nAn- "around" (cf. Hupa -nat\ Kato -nd)\ nAn/ndS/yd/la "he went around it" qlwAt- "on, on top" (cf. Hupa -kut "on;" Kato -k'wut' "on"): qlwAt/tclAt/dja "whereon one eats, table" q!wAt/daBt/gAc "it lies thrown down on top, table-cloth" qlwAt/dasi/nd "it was lying on it" Six: Athabaskan and Na-Dene Languages 59 305 Deictic Prefixes. Under this head are grouped a small number of quasi-pronominal elements of third personal refer- ence which regularly come after adverbial prefixes, if any of these are present. They cannot be grouped with first or second personal subjective elements, as their position is quite distinct from these; first and second modal prefixes may come between. Of deictic elements there have been found: td- denotes lack or indefiniteness of object of transitive verb (cf. Hupa ^-, ky-, i. e. k^!-\ Kato tc'- ): tdAy/ye/dic/ya "I eat" (i. e. without specific object being designated; cf. Hupa yik/kyu/- win/yan "it ate") q!wAt/tc!At/dja "whereon one eats, table" na/tc!il/L!d "you write" (cf. Hupa na/kis/Lon, i. e. na I k^ lis I hi on "she made baskets") nd/tc!l/t!d "you swim, bathe" na/tcHl/de "you wash" (cf. Kato te'/na/tc'ih/deG "he washed it") tdAt/Vit/dAl "we wash ourselves" (cf. Hupa wa/- kin/nin/seL "it was heated through") mAl I V e I tcl At I tsl aI I lee "wherewith it is smoked" (somewhat doubtful, as id- here follows first modal prefix Ve-\ but see note 77) td All ltd "he sucks" (cf. Kato tc'lh/Vot "[make] it suck") It is possible that in this last example td- is third personal subjective (cf. Hupa tc-, i. e. td-; Kato tc'-, ts'-, 5'-), as sug- gested by Ad/tlo "I suck" with its lack of td- prefix. No other plausible case, however, of third personal subjective td- is available, so that its existence in Chasta Costa must be con- sidered doubtful as yet. Generally third person singular subjective forms are dis- tinguished by the lack of any pronominal prefix, but in certain 60 ^^ American Indian Languages 2 306 cases deictic elements are found which are clearly third personal (subjective) in value. These are: dj- (cf. Hupa tc-, i. e. tcf-; Kato tc'-, ts'-, s'-): d/dJAn "he says" (verb-stem -n; cf. Hupa ai/- tcit/den/ne "he talked to them") djAn/la "he says" (cf. Hupa tcin "they say;" Kato tc'in) It is quite likely, however, that dJAn is to be explained as from *dyAii {dy, as we have seen, becomes dj), in which d- is first modal prefix (cf. dt/nl "you make a sound") and *yAn is reduced from *yen {ye- as below; -n to say). ye-, ya- (cf. Hupa y-, yl- referring to other than adult Hupa; Kato yl-) : uaI na/yei/si "to-you he-tells" (contrast ndcl/sl "I tell") caI na/yesi/sl "to-me he-tells" (with second modal prefix s-, 6-; contrast nd/dU/sl "you tell") ya/qled/ya'^^ "he eats" (contrast tclAy/ye/6ic/ya "I eat") tcIe/xAd/tfo "he swims across" (contrast tcfe/- Bit/t!d "you swim across") This X- seems to have no parallel in Hupa or Kato (is it con- nected with third person objective xo- of Hupa, kw- of Kato?). Were it not that -t!d "to swim, bathe" is Used only of singular subjects, one might surmise that nc-'is really plural xA- (see below) . Among deictic elements are further to be reckored certain prefixes that serve to indicate either plurality as such or more specifically third personal plurality. These are: ya- (cf . Hupa ya- ; Kato ya^-) : yu/wis ya/dil/ni "we whistle" (contrast yil/- ims dAcl/ni "I whistle") " qlc- was very likely misheard for te-. Six: Athabaskan and Na-Dene Languages 61 307 yu/wls ya/dAl/ni "they whistle" (contrast yn/wls dAl/nl "he whistles") ya-, ya- tc!a/yd/dil/se "we cry" (contrast tcfd/sAsL/se "I cry") tcIa/ydsL/se "they are crying" (contrast tcfdsL/se "he cries") tc!d/ydl/se/t'e "you (pi.) will cry" (contrast Icldl/- se/t'e "you (sing.) will cry") XA- third person plural (apparently not found in either Hupa or Kato; but cf., without doubt, Chipewyan he- "used for dual or plural of verbs in third person"): yd/xAy/yi/tIa "they fly" {yd/yi/tia "it flies") dd/yd/xAt/t!a "they won't fly" {do/yd/tfa "he won't fly") XAs/se/re/ha "will they cry?" {As/se/t'e/ha "will he cry?") do/xAs/se "they're not crying" {do/As/se "he's not crying") nd/xAt/dAl/nic "they work" {nd/dAl/nic "he works") c/xA/yee/l "they saw me" {c/yeS/l "he saw me") na/xAt/da/yAl/el "they are bathing" t"e/An/xAy/yAl/lAl "they sink in the water" {Ve/An/yAl/lAl "he sinks") xAt/t'Al/lal "they are sleeping" {VAl/lal "he is sleeping") dd/xAn/nAt/VAc "they went to bed" {dd/nAt/t'Ac "he went to bed") x At /VAI/xwaB'' they cough.'' {VAI/xwaB'' he coughs') xa/AI/az "they sneeze" (aI/Az ''he sneezes") First Modal Prefixes. Under this term are comprised a small number of rather frequently occurring elements which regularly come after both adverbial prefixes and deictic ele- ments, but precede another set of modal elements (second 62 VI American Indian Languages 2 308 modal prefixes) which are to be taken up shortly. Their mean- ing is rather colorless. Besides their position they have this peculiarity in common, that they lose their vowel in indefinite tense forms (such as have no second modal prefixes: d-, y-, or W-) and are thus reduced to single consonants. They are: t'e- (definite tenses), t'- (indefinite tenses) seems to indicate durative activity (cf. Hupa te-; Kato te-, t-): re/dic/ya "I go;" indefinite: do/rAc/yic "I'll not go;" t'l/yAc/t'e "you must go" VAci/dAc "I run" (indefinite) tclAt/Vo/dAl "you (pi.) wash yourselves" (indefi- nite) t'/yi/l "he looks around" (indefinite; but see note 69) Ve/BAd/lal/la "I've been sleeping;" indefinite: VAc/lal "I'm sleeping" VAcl/xwAd "I cough" (indefinite) mAl/t'e/tc!At/ts!Al/lec "wherewith it is smoked" (as following td- is deictic, it is more likely that Ve- here is adverbial prefix, not first modal; see note 77). de- (definite tenses; da- before 7-), d- (indefinite tenses) meaning unknown (cf. Hupa d-, du-\ Kato de-, d-)\ al/dAc/ni/dAn "when I tell him" (indefinite) yu/wis dAcl/ni "I whistle" (indefinite) c/na/yd di'/s'aV "my-eyes hurt" (definite; cf. Hupa du I win Ileal "it got sick") nd/da/yAct/l!d "I bathe;" indefinite: nd/dAct/- Ud'/Ve "I'll bathe" na/da/yil/el "we are bathing;" indefinite: nd/- dil/el "we'll bathe" nd/dAci/nic "I work" (indefinite) na/dAcl/de "I washed myself" (indefinite) ya/dAcl/yAc "I am ashamed" (indefinite) ya/da/yil/dja "we are ashamed" (definite) dd/de/dil/lB!i "we are sitting down" (definite) Six: Athabaskan and Na-Dene Languages 63 309 ye- (definite tenses) , 7- (indefinite tenses) meaning unknown : tclAy/ye/dic/ya '1 eat" ya/q!eB/ya "he eats" {q!e- is probably misheard for ye-) ye I Bill "you saw him;" indefinite: do/wa/yl/i/- Ve "you'll see him" yedt/lo "he breaks into laughter;" indefinite: yAt/lo "he laughs" an/yi/al "come on!" This ye-, y- should not be confused with second modal 7-, which will be taken up presently. Two first modal prefixes (/'- and 7-) occur in V/yi/l "he looks around;" that 7- is not second modal here is indicated by parallel definite forms with ye- (see ye/di/l above), further by weak form V- of first prefix (definite tenses require t'eY"^ ne- (definite tenses), n- (indefinite tenses) meaning unknown (cf. Hupa ne-, n-; Kato ne-, n-; Chipewyan we-, nil-): xwAn/ne/dii/ya "you win" (cf. Kato kun/ne/sli/- yan "you win") dd/nAc/VAc "I go to bed" (indefinite; cf. Hupa definite: tcin/nes/ten ''hQlSiy') n/dd" "it is, stays" (indefinite; cf. Kato definite: tc'n/nes/dai "he sat down") nd/nAc/An "I stop him;" nd/ni/An "stop him!" (indefinite) n/do "it is not" (indefinite; cf. Kato n/dd*/ye "there is none") This ne-, n- is not to be confused with second modal n-, which occurs only in definite tenses. Second Modal Prefixes. These comprise three conso- nantal elements {B- or s-\ 7-; and 71-) which are used only in definite tenses and which have reference, as far as any definite *' Moreover, te- in definite tenses seems regularly followed by second modal 6-, not 7-. Yet -I- of yl- causes difficulty; see note 69. 64 yf American Indian Languages 2 310 significance is ascertainable at all, to what may be termed range or span of activity, but not to tense as such. 6- {s- in certain forms) is durative or continuative in force (cf. first modal t'e-, which is regularly followed by 6-) ; w- is cessative, marking the end of an activity or marking an activity which is con- ceived as the end point of a previous activity (e. g., "to come" as contrasted with durative "to go"); y- is the most uncertain, being apparently inceptive or momentaneous in some cases, but clearly not so in others/^ They are, it seems, mutually exclusive elements. In practice their use seems largely deter- mined by the prefixes that precede, n- and y- always begin their syllable, being completed either by -i- ( < *-ir}-) or by subjective pronominal or by third modal elements, which are joined to them by means of -i- or inorganic -A-; 6- (s-) is similarly joined to following subjective pronominal elements, if one is present, otherwise it forms part of the preceding syllable. Exrmples illustrating 6- (s- before I, l) are: t'e/dic/ya "I go;" reO/ya "he goes" (cf. Hupa te/se/yai "I went away") nAfi/ndd/yd/la "he went around it" tdAy/ye/dic/ya "I eat;" ya/q!ed/ya "he eats" (contrast Hupa yik/kyil/win/yan "it ate" with w-) tcfe/dit/tfo "you swim across" (contrast na/da/- yit/tfo "you bathe") dad /da "he is sitting down" (cf. Hupa sit/dai "he lived") dd/de/6il/td!i "we are sitting down" (cf. Hupa de/soh/tse/te "you will stay") tc!d/6U/se "you (sing.) cry" (contrast Hupa winj- tcwu "you have cried") '^ Goddard somewhat doubtfully assigns inceptive force to its Hupa cognate w-; in Kate its cognate g- seems clearly inceptive only in certain verbs; while in Chipcwyan Goddard ascribes continuative value to g-. It would be worth while making a somewhat extcnilcd comparative study of the second modal prefixes of Athabascan, which form one of the most difficult but at the same time important chapters of its grammar. Six: Athabaskan and Na-Dene Languages 65 311 ye /die /I "I saw him" (cf. Hupa te/suw/ifi "I am going to look") na I QUI SI "you told story" (cf. Hupa na/sei/tcwen "I made") Ve/Bi/lal/la "you've been sleeping" (cf. Hupa nit/te/sil/lal/le "you would go to sleep") Bid/ Veil "I kicked him" (contrast Hupa ye/tcu/- wii/taL "they landed" with w-) na/yd/6it/xwi "you vomit" ye/dAdt/lo "I break into laughter" Lad/d/la one was { = Ld "one" plus d/'d/la; cf. Chipewyan ^e/'d "was there") q!wAt/dadt/gAc "it lies thrown down on top, table-cloth" (cf. Hupa wes/kas "it lay") q!wAt/dasL/nd "it was lying on it" xwAn/ne/diL/ya "you win" (cf. Kato kun/ne/- slL/yan "I win") Examples illustrating n- are : nic/ya "I come" (cf. Hupa nei/yai "I came") nic/dac "I dance" (cf. Kato nuc/dac "I will dance") Ve/ni/lat "you drown" (cf. Kato tc'n/ mil flat "it floated there") Examples illustrating y- are: yt/dac "he dances" (cf. Kato tc'/gtm/dac/ kwaii "he had danced") yd/yAd/gAd "I climb" (cf. Hupa ya/wii/kas "he threw up") yd/yi/tfa "it flies" (cf. Hupa na/win/tau "it will settle down") nd/da/yAct/t!d "I bathe" na/da/yil/el "we are bathing" (cf. Chipewyan ni/i/oinL/ilL "take through the water") 66 ^/ American Indian Languages 2 312 fe/An/yAc/Ul "I'm sinking in the water" (of. Hupa da/na/wil/laL "it was floating there") yAci/Az "I've been sneezing" tc!d/yit/siL/la "he pointed with his finger" ya/da/yit/dja "we are ashamed" Subjective Pronominal Prefixes. There are three per- sons and two numbers (singular and plural), making six persons in all. The third persons, as we have seen, are indicated either by the absence of a pronominal element or by deictic prefixes which come between the adverbial prefixes and the first modal elements. There thus remain four persons (first person singular and plural, second person singular and plural) for treatment here. In the definite tenses the pronominal elements are appended to the second modal elements, with which they form a syllable, an inorganic a or i, if necessary, serving to connect them. In the indefinite tenses the pronominal elements are appended to whatever element (adverbial prefix, deictic ele- ment, or first modal prefix in reduced form) happens to precede them. They never begin their syllable except in the com- paratively small number of cases in which the verb form, in- definite in tense, has nothing preceding the pronominal element or, in the case of the third person, nothing preceding the third modal prefix or verb-stem. When this happens, the second person singular and plural and the first person plural stand at the very beginning of the verb; the first and third persons singular, however, begin with an inorganic vowel A-. First Person Singular -c- (cf. Hupa -w-; Kato -c-] Chipe- wyan -s-) : Ve/dic/ya "I go" dicL/Val "I kicked him" nic/ya "I come" dd/nAc/VAc "I go to bed" Ve/An/yAc/lAl "I am sinking in the water" y Act /Id "I laugh" Six: Athabaskan and Na-Dene Languages 67 313 ya/yAcl/gAd "I climb" VAcL/dAc "I run" nac/tlo "I swim, bathe" Acl/t!6 "I suck" acl/Az "I sneeze" In definite tenses with 6- or n- as prefix the inorganic vowel connecting these elements with -c- is regularly i\ this is evi- dently due to the palatal quality of the -c-. In definite tenses with 7- as prefix, however, the normal inorganic vowel, a, is found, due, no doubt, to the velar position of the prefix. In the indefinite tenses the connecting vowel, if required, is always a. Where we have nAc- we are dealing with first modal ne-, reduced to n-, plus -c-, not with second modal n- plus -c-\ contrast definite nic/ya with indefinite da/nAc/VAc. Before s- sibilants -c- is assimilated to -s-: do I AS I se "I'm not crying" {<'^Aclse) Sic- goes back to original *sic- or *sac-. When -c- came to stand before a dental consonant {d, t, /), it was assimilated to -S-, and the inorganic vowel preceding it assumed the form a ; this *sAs- then regularly became 6 Ad-: da /e Ad /da "I am sitting" ye/BAQt/lo "I break into laughter" na/ya/BAdt/xwi "I vomit" Ve/dAd/lal/la "I've been sleeping" Before third modal -1-, die- seems to be regularly retained (cf. Bicl/t'dl above; Sid/ si "I make"). Secondary sas-, not shifted to OaO-, is found, however, before 1{l) when this element is secondarily changed from third modal -/-: tc!d/sAsL/se "I cry" That sasl- here is equivalent to *sasI-<*sacI- is indicated by tc!d/6il/se "you cry;" contrast Bil/t'cd "you kicked him," eicl/Val "I kicked him." 68 VI American Indian Languages 2 314 It is to be carefully noted that -c- (or its reflexes -s-, -$-) is in Chasta Costa found in both definite and indefinite tenses. There is no trace of an element corresponding to the Hupa .^. (-e-), Kato -t-, Chipewyan -1-, which are found in forms of definite tenses. It is quite probable that the -c- of the indefinite forms was extended by analogy. Second Person Singular i- (cf. Hupa -77-, i. e. -r)-\ Kato -W-; Chipewyan n-, ne-, or nasalization of vowel: Vil/dAc "you run" Vd/yit/nd "you drink" na/tcHl/Ud "you write" na/da/yit/t!d "you bathe" nd/dit/t!d/t'e "you'll bathe" yd/yil/gAB "you climb" ya/dil/yac "you are ashamed" nd/dil/sl "you made, told" nd/xwil/ye "you play" xwil/l "you believe it" xwAn/ne/BiL/ya "you win" VU/xwaB "you cough" In all these cases the -i- connects a following third modal ele- ment {-1-, -/-, or -/-) with a preceding prefix. Examples of -/- beginning its own syllable are: tl/Az "you sneeze" hd/il/l "stop!" If there is no third modal element, the -i-, lengthened to close -1-. closes its syllable: dd/ni/t'Ac "go to bed!" t'i/lal "you are sleeping" dd/Bi/dd "you are sitting" yd/yi/t!a "you fly" do/l/se "you do not cry" Six: Athabaskan and Na-Dene Languages 69 315 This i-, 1-, is only secondarily the second person singular subjective element. The original element was doubtless -17- (cf. Hupa). which was reduced to nasalization of preceding vowels; the inorganic vowel, when nasalized, took on i- timbre Finally, when nasalization disappeared, the /- timbre alone remained as the reflex of original -rj-. Where, in many indefi- nite tense forms, the nasalized vowel was other than an inorganic one, there was nothing left of the -77-: do/yd/tfa "you won't fly" nd/xe "you paddle" nei/l "look at him!" (cf. ned/i "I'm looking at him") In such cases the second person singular fell together with the third, as in dd/yd/t!a "he won't fly." First Person Plural (i)t-*^ (cf. Hupa it/d-, -d-\ Kato d-; Chipewyan -/-, -d-^'^) : Vit/lal "we are sleeping" tclAt/Vit/dAl "we wash ourselves" Ve/ nit flat "we drown" da /nit /Vac "we went to bed" ye/dit/l "we saw him" ya/yit/t!a "we fly" Ve/An/yit/lAl "we are sinking in the water" dd/it/se "we are not crying" In Hupa and Kato regularly, and in Chipewyan often, the first person plural subjective pronominal prefix begins its syllable; in Chasta Costa it regularly ends its syllable, unless it has to stand at the beginning of the verb form, when it consti- tutes a syllable by itself (cf. dd/it/se above; do "not" is inde- pendent adverb rather than prefix). " / is here unaspirated, and is thus etymologically identical with d. '" In Father Legoff's Montagnais paradigms -id- or -//- often, in fact regularly, appeals; -i- seems, as in Chasta Costa, to be organic. 70 ^f American Indian Languages 2 316 If the prefix preceding the pronominal element ends in a vowel, the -i- disappears: do/ydt/tla "we won't fly" This does not mean, however, that this -/- is to be considered an inorganic vowel, as is the case in Hupa it/d-. If -it- is followed by third modal -/-, both -/- elements combine into a single -/-, and all that is left of the pronominal prefix is the -i- : t'd/yd/dit/nd "we drink" (contrast t'd/yddt/nd "they drink") If the third modal element is -I- or -/-, -/- disappears and -l- is changed to -/-; thus the first person plural of I- verbs and /- verbs is always formed alike. In Hupa and Kato third modal -f- regularly becomes -/-, but d- is preserved; hence Hupa dil-, Kato dul-. In Chipewyan, however, as in Chasta Costa, -/- not only becomes -/-, but -/- disappears. For Chasta Costa this means that the second person singular and first person plural of /- verbs is identical, provided, of course, that there is no deictic prefix of plurality in the latter and that the verb stem does not change for the plural. Examples of /-verbs are: dd/de/Bil/td!i "we are sitting" (cf. Chipewyan de/dll/d'l "we are sitting") nd/dil/nic "we work" (cf. nd/dil/nic "you work") t'il/xwAd "we cough" (cf. VU/xwaB "you cough") nd/xwil/ye "let us play" (cf. nd/xwil/ye "you play") il/Az "we sneeze" (cf. tl/Az "you sneeze") na/da/yil/el "we are bathing;" nd/drl/el "we'll bathe" tc!a/ya/dil/se "we cry" (cf. tda/Bil/se "you cry") Examples of /- verbs are: yu/wls ya/dil/nl "we whistle" (cf. third person plural: ya/dAl/nl) na/tclil/Ud "we write" (cf. na/tdil/dd "you write") Six: Athabaskan and Na-Dene Languages 71 317 If, in an indefinite tense form, the pronominal element is pre- ceded by a prefix ending in a vowel and is, besides, followed by third modal -I- or -/-, both -i- and -/- have to disappear and there is nothing left of the pronominal element except, in the case of /- verbs, the change of -/- to -/-; nel/l "let us look at him!" (cf. nel/l "look at him!") tcld/yel/se, very likely misheard for tc!d/ydl/se "we'll cry" (cf. definite: tcfa/yd/dil/se "we cry") Second Person Plural 0- {ci-Ylupa. o'-; Katoo'-; Chipewyan rS/lal "ye sleep" t'e/do/lai "ye have been sleeping" t'e/nd/lat "ye drown" yd/yd/ t!a "ye fly" t'e/An/yd/Ul "ye sink in the water" t'd/yd/dot/nd "ye drink" dd/o/se "ye are not crying" No aspiration was heard after o in Chasta Costa. This does not seem due to faulty perception, as /- verbs keep their -/- after o-, whereas, under similar circumstances, Hupa, Kato, and Chipewyan change -/- to -I- {o'-l- becomes ol-). Indeed, in Chasta Costa I- verbs change their -I- to -/- after second person plural 6-. Examples of o- before /- verbs are: nd/dol/nic "ye work" na/da/yol/el "ye bathe;" nd/dol/el "ye will bathe" t'dl/xwAd "ye cough" nd/xol/ye/le "ye play" (for -xwol-) ol/Az "ye sneeze" Examples of -/- becoming -/- after 3- are: ya/ddl/7il "ye utter, make a sound" (cf. third person plural ya/dAl/nl) ne/xd/ol/l "ye look at him" (cf. nci/i "you're looking at him") 72 VI American Indian Languages 2 318 When, in an indefinite tense form, o- is preceded by a prefix ending in a, a and o contract to long a (which, it would seem, remains long even in closed syllables) : do/yd/ t!a "ye won't fly" {<*yad'-; cf. third person singular do/yd/tla with original yd- ; and second person singular dd/yd/t!a < *yq- < *yarj) tc!d/ydl/se/t'e "ye will cry" (cf. definite: tcIa/yd/Odl/se "ye cry;" and contrast tcld/ydl/se "we'll cry" with short -a-) Third Person. As already noted, the third person, apart from possible deictic prefixes, is marked by the absence of any pronominal element. If the element preceding the third modal prefix or the stem consists of a consonant which must begin its syllable, an inorganic -a- is found between the two; if a third modal prefix is absent, the syllable preceding the stem is closed by a consonant borrowed from the first consonant of the stem. Examples of third persons with -a- before a third modal prefix are: ya/dAl/yAc "he is ashamed" VAl/dAc "he runs" nd/xwAl/ye "he plays" yd/yAl/gAd "he climbs" yu/wls dAl/ni "he whistles" Vd/yAt/nd "he drinks" Examples of third persons with -A- followed by an inorganic consonant are: dd/nAt/VAc "he went to bed" (-/- is not third modal; cf. second person singular dd/ni/VAc) Ve/Ati/yAl/lAl "he is sinking in the water" (-/- is not third modal; cf. second person singular Ve/An/yl/lAl) First modal n-, reduced from ne-, however, has in several cases been found without following inorganic vowel and consonant. Six: Athabaskan and Na-Dene Languages 73 319 In such cases it closes the preceding syllable, which may even belong to another word. Examples are: dd/dAt 'An/tie "not-anywhere there-is" (cf. Kato qn/t'e "it is;" Hupa un/te "there is") Idn/do "nine" (really la n/do "one is-lacking") ; nd/xAn/do "eight" (reduced from na/xi n/do "two are-lacking") (cf. Kato 71 /do' /bun "it will not be," but also niit/do' "all gone") cic/mAnen/dd' { = ci c/mAne n/da') "I my-house is" (cf. Chipewyan ne/da "she sat") If the verb form consists, properly speaking, of the stem alone, without prefix of any kind, an inorganic a- completed by a consonant that depends for its form on the first consonant of the stem is prefixed for the third person: As/se/Ve "he must cry" (<*j^//V; cf. dd/As/se "I'm not crying" <*Ac/se) do / wa / aI / W "he will become" (<*/^; do/wa is adverb not influencing form of verb proper. That -/- is here no third modal element is shown by forms like Hupa o/le "let him become") This /I- at the beginning of a third personal form appears also when the verb begins with a third modal element: aI/Az "he sneezes" In this respect Chasta Costa differs from Kato, which need have nothing preceding the stem; with As/se compare Kato tee' "he cried." In the third person of definite tenses with second modal 7- or n- prefix this element is followed by -I, in case there is no third modal prefix present. This goes back, without doubt, to nasalized -i- or -A-, in turn reduced from original -it)- (or -At]-). This nasal element, characteristic of definite third personal forms (except such as have 6-, Athabascan s-, as second 74 VI American Indian Languages 2 320 modal prefix) is found also in Hupa (-in-), Kato {-un-), and Chipewyan (-«-, -in-). Examples are: Ve/ni/lat "he drowns" (cf. Kato tc'n/nul/lat "it floated there," nul- assimilated from nun-; Ve/ni/lat also "you drown") yi/dac "he dances" (cf. Kato tc' /gun/dac/kwan "he had danced") yd/yi/t!a "it flies" (cf. Hupa na/win/tau "it will settle down;" ya/yi/ila also "it flies") di*/s'aV "it pains" {dl*-<*dir]-, contracted^^ from *de/- yir}-; cf. Hupa du/win/tcat "it got sick") Rather hard to understand is: Ve/An/yAl/lAl "he is sinking in the water" One would have expected -7!-, not -yAl- (as seen above, -/- is not third modal, but inorganic). Is yA- reduced from first modal ye-, this form being indefinite in tense? In Hupa this -in- does not seem to be found before third modal prefixes; in Chipewyan -n- {-in-) may, however, occur before -I- and, as inferred from Father Legoff's Montagnais paradigms, also -/-. As for Chasta Costa, what examples are available on this point show that -i- does not occur before -/- (e. g. yd/yAl/gAB "he climbs"). For I- verbs I have no safe example. Before -/- it seems that -i- is present in some cases, not in others: tcld/yit/sih/la "he pointed with his finger" but, without -i-'. na/da/yAt/t!d "he is bathing" On the whole, it seems possible that Athabascan -Ar]- (or -An-) was originally a more freely movable element than it has " Parallel in form to Hupa verbs belonging to Class I, Conjugation 1 D, in which prefixed first modal d- or deictic k^l- contracts with -it}, w- (Athabascan 7-) being lost. See Goddard, op. cit., p. 113. Six: Athabaskan and Na-Dene Languages 75 321 become in e. g. Hupa, being required 'by certain verbs in their definite tenses, but not by others. This is suggested also by Father Legoff's Montagnais paradigms. Third Modal Prefixes. There are three of these: -/-, -/-, and -t-\ they always complete a syllable immediately before the stem. -I- is characteristic of many verbs which are either transitive or, at any rate, imply activity directed outward; in some cases, however, this significance is not obvious. As we have seen, this -I- becomes -/- in the first and second persons plural. Examples of -1-{-l-) are: Bid/Val "I kicked him" tB!AB/da caI nd/6il/sl "story to-me you-told, made" necl/l "I'm looking at him" ndl/de "he washes (something)" nail Lib "he writes" tc!Al/t!d "he sucks" yu/wis dAl/nl "he whistles;" yu/wls dAcl/nl "I whistle"'^ xwAn/ne/diL/ya "you win" If nothing precedes this element, it seems (unlike -/-) to begin its word without preceding inorganic A-: dd/wa/l/t'dt/fiAl "they will be broken" {do/wa is merely proclitic) l/tH "he is important" Verbs in -/- are regularly intransitive; they denote states of mind or bodily activities that may be thought of as self- contained, not directed outwards. A reflexive meaning is sometimes apparent. After first person subjective -c- (s-) it always appears as -/-. Hence the first person singular, the first person plural, and the second person plural of -/- verbs and /- verbs are always alike (but contrast did- < *sAd- with 5Asi<*sAd-). As -/-, when standing after s, becomes -I- also in the third person, the second person singular alone remains '- This verb is irregular, inasmuch as -/- does not occur in the second person singular: yA/wis di/nt "you whistle." 76 VI American Indian Languages 2 322 as an infallible criterion of whether a verb belongs to the l- class or /- class. Examples of -/- are: ya/dAl/yAc "he is ashamed" xwAcl/1 "1 believe;" xwil/i/ha "do you expect?" VAl/dAc "he runs" yd/yAl/gAd "he climbs" nd/dAl/de "he washed himself" aI/Az "he sneezes" VAI/xwaB "he coughs" na/xAt/da/yAl/el "they are bathing" nd/xwAl/ye "he plays" nd/dAl/nic "he works" tddsh/se "he cries;" tela I HI I se "you cry" tc!dcL/se/Ve "I'll cry;" tc!dl/se/Ve "you'll cry" hd/yAci/i "I stop, cease;" hd/il/i "stop!" Verbs in -/- are also intransitive. Examples are: Vd/yit/nd "you drink" ye hat I ne I la "he bit (it)" tdd/yit/sii/la "he pointed with his finger" qlwAt/dadt/gAc "it lies thrown down on top, table-cloth" yAct/ld "I laugh" {-t/lo may, however, represent original Athabascan -did "to laugh") ya/da/yit/dja "we are ashamed" There may be a passive significance in : q/wAt/tclAt/dja "whereon it is eaten, table" With iterative na-: na/yd/6Adt/xwl "I vomit" Verb Stems. The stems that have been determined for Chasta Costa are: -'d,-'a "to have position, to be" (cf. Hupa -ai, -a; Kato -*ai*, -'a'): d/'d/la "(one table) was" Six: Athabaskan and Na-Dene Languages 77 323 -'AC "to bring" (cf. Hupa -an, -uii, -auw "to transport round objects"; Kato -*qn, -^qc): yAn/na/'Ac "he will bring it" -'An "to bring to a halt, stop" (perhaps another form of preceding stem): 7td/ni/An "stop him!" -Az "to sneeze": acl/Az "I sneeze" -ai "to come" (cf. Chipewyan -^as, -*az, -*ais "to travel, used of two persons only"?) an/yi/'al "come on!" -ci, -el "to bathe (plur. subject)" (cf. Chipewyan -el, -eL, -uL "to move on the surface of water"): na/da/yil/el "we are bathing" -'I "to see, look at" (cf. Hupa -en, -in; Kato -*ln' "to look"): ye/dic/l "I saw him" -i "to stop, cease" (cf. Hupa -en, -in "to do, to act, to deport one's self"?): hd/yAch/i "I stop (laughing)" -I "to believe, expect": xwAcl/i "I believe" -ya, -yAc, -yic "to go, come" (cf. Hupa -yai, -ya, -yauw; Kato -yai, -ya, -yac): Ve/6ic/ya "I go;" t'Ac/yAc/t'e "I must go" -ya "to eat" (cf. Hupa -yan, -yun, -yauw; Kato -yan^ -ytl*): tclAy/ye/Oic/ya "I eat;" qIwAt/tcfAt/dja(<-t/ya) "whereon one eats, table" -ya "to win" (cf. Kato -yan, "Kato Texts," p. 146, 1. 13; not listed in "Elements of the Kato Language"): xwAn/ne/dii/ya "you win" -yan "to upset" (cf. Kato -yan "to clear off"?): do/wd/An/- nd/yan/nAl "he will upset them" -ya, -yAc "to be ashamed" (cf. Kato -yafi "to be ashamed"): ya/dAl/yAc "he is ashamed;" ya/da/yit/dja {<-t/ya) "we are ashamed" -ye "to play" (cf. Hupa -ye "to dance"): nd/xwAl/ye "he plays" -lal "to sleep" (cf. Hupa -lal, -lai; Kato -lal, -Iqi): t'cd/- lal/la "he's been sleeping" -IaI "to sink in water;" -lat "to drown" (cf. Hupa -lat, 78 VI American Indian Languages 2 324 -la "to float"): re/Ari/yAc/Ul "I am sinking;" fe/- nic/lat "I drown" -le "to become" (of. Hupa -len, -lin, -hi, -le; Kato -lin% -le): do j wa I aI / le' "it will become" -lee "to wager, bet": ne/tclilc/lec "I'll bet you" -lee "to smoke": mAl/Ve/tc!At/ts!Al/lee "wherewith it is smoked" -Id "to laugh, smile" (cf. Chipewyan -did, -dlok'): yit/lo "laugh!" -lIo "to write" (cf. Hupa -Lon, -Ld, -Low, -Loi "to make baskets, to twine in basket-making;" Kato -Loi, -Ld, -Lon): nail Lib "he writes" -na "to drink" (cf. Hupa -nan, -nun-, Kato -7iqn): Vd/- yAct/nd "I drink" -na "to lie" (cf. Kato ndLjiin/na'' "were left"?): qlwAt/ dasL/nd "it was lying on it" -ne "to bite, seize with one's teeth": ye/ydt/ne/la "he bit it" -nl, -71 "to make a sound, to say" (cf. Hupa -ne, -n "to speak, to make a sound;" Kato -nl, -ne, -n, -nee, -nlL): yu/wls dAl/ni "he whistles;" a/dJAn "he says" -nic "to work": nd/dAl/nic "he works" (cf. Montagnais -7ii "exprime Taction des mains"^^ -dAl "to wash oneself (plur. subject)" (cf. Hupa -sel, -scl "to be or to become warm;" Kato -sll "to steam," -sUl, -suL "to be warm"): tc/At/t'it/dAl "we wash ourselves" -del "to throw": yd/yi/del "you threw" -se "to cry" (cf. Hupa -tcwu, -tcwe "to cry, to weep;" Kato -tcec, -tee'): tcIdsL/se "he cries" -61 "to cause" (cf. Hupa -tcwen, -tcwin, -tewe "to make, to arrange, to cause;" Kato -tcin, -let, -tclL): ndcl/sl "I cause" ■sil "to point with one's finger": tc!d/yit/siL/la "he pointed with his finger" " Father L. Legoff, "Grammaire de la Langue Montagnaise," p. 139. Six: Athabaskan and Na-Dene Languages 79 32.5 -da', -da "to sit, stay" (cf. Hupa -dai, -da; Kato -da, -dai): da I Bi I da "you are sitting" -dAc "to run" (cf. Hupa -dai, -daL, -dauw "to pass along, to go, to come;" Kato -dac "to travel"): VAl/dAc "he runs" -dac "to dance" (cf. Kato -dac "to dance"): nt/dac "you dance" -de "to wash (sing, subject)" (cf. Kato -dec, -de'): nd/- dAl/de "he washed himself" -t'di "to kick" (cf. Hupa -taL, -tul, -tuL, -tal "to step, to kick;" Kato -tal\ -tqL): dicl/t'dl "I kicked him" -t'Ac "to lie down, go to bed" (cf. Hupa -ten, -tin, -tuw "to lie down;" Kato -tin, -tuc): dd/nAc/VAc "I go to bed" -Vat "to break, go to pieces" (cf. Chipewyan -taL, -till "to break"): dd/wa/l/t'dt/uAl "they will be broken" -Ve "to want" (cf. Hupa -te "to look for, to search after"?): dd/ucL/t'e "I do not want;" de/uci/t'e "what I want" -t'J "to be, make valuable" (cf. Carrier til/thi "thou makest him valuable, treatest him as important"): l/Vi yAn/na/'d "he brags" -t.'a "to fly" (cf. Hupa -tau; Kato -t'ac, -t'a'): do/ydc/tla "I won't fly" -tie "to be of (that) sort" (cf. Hupa -te; Kato -t'e): do/dAt 'An /tie "there is not anywhere (one like him)" -t!d "to swim, bathe (sing, subject)": nac/t!d "I swim, bathe" -t!d "to suck" (cf. Kato -t'ot): tclAi/t/d "he sucks" -tdfi "to sit (plur. subject)" (cf. Hupa -tse; Chipewyan -d'i): da/de/dil/tdii "we are sitting" -tshit "to hurt, pain (intr.)" (cf. Hupa -teat, -tea "to be sick, to become ill"): di*/s'at' "(my eyes) hurt" -7« "to go about, live" (cf. Hupa -wai^ -wa "to go, to go about;" Kato -ga, -gai): nd/ya "he goes about, lives" -xe "to paddle" (cf. Hupa -xen, -xuw "to float, used only of plural objects;" Kato -ke' "to bathe (plural only);" 80 VI American Indian Languages 2 326 Chipewyan -kl "to paddle a canoe, to travel by canoe") : ndc/'xe "I paddle" -xwAd "to cough" (cf. Kato kos "cough," as noun; Carrier xwces): VAl/xwAd "he coughs" -xwl "to vomit" (cf. Carrier khti "vomiting," as noun): na/yd/dAdt/xwi "I vomit" -gAd "to climb" (cf. Hupa -kas "to throw"): yd/yAl/gAd "he climbs" -gAC "to throw"? (cf. Hupa -kas "to throw"): qfwAt/dadt/- gAc "it lies thrown down on top, table-cloth" It will be observed that several verb stems are restricted in their use as regards number of subject (or object). This trait is characteristic of Athabascan, as also of other American linguistic stocks. Definite and Indefinite Tenses. My material on Chasta Costa is not full enough to enable me to give a satis- factory idea of its tense-mode system. It is clear, however, that absolute time (present, past, future) is quite subordinate to whether activities are thought of as taking place at some definite time (generally present or past) or are more indefinite as to time occurrence. Indefinite forms are apt to be used for general statements that apply irrespective of any particular time, for future acts, for negative (particularly negative future) acts, and regularly for imperative and prohibitive forms. The contrast between definite and indefinite present forms comes out in: definite: nd/da/yAct/t!d "I bathe" (i. e. am. now engaged in bathing) I indefinite: cAl/qlwe na/dAct/t!d "I'm used to bathing" [ (here bathing is not restricted as to time) f definite: tc!d/sAsL/se "I cry" [indefinite: Vwi/dAn As/se "I always cry" definite: yeSt/lo "he breaks into laughter" (i. e. laughs at one particular point of time) indefinite: ydt/lo "he laughs" Six: Athabaskan and Na-Dene Languages 81 327 [definite: xAt/Ve/lal/la "they have been sleeping" (may \ be said of them at moment of waking up) [indefinite: xAt/VAl/lal "they sleep" Futures, as we shall see, are explicitly rendered by suffixing -Ve to present (generally indefinite) forms; but simple indefinite forms, particularly with adverbs pointing to future time, may often be used as futures in contrast to definite present forms. Examples are: \ definite: na/da/yil/el "we are bathing" indefinite: nd/dil/ei "we'll bathe" definite td!AB/da caI na/dU/si "story to-me you-told" indefinite: xiin/de id! Ad /da uaI ndd/sl "tomorrow story to-you I-tell" definite: n/ye/Bic/l "I saw you" I indefinite: xAL/tsH/dAn do/wan/yAc/l "this-evening I'll- ■[ see-you" I definite: Ve/dic/ya ''I go'' [indefinite: xun/de Vac/yAc "tomorrow I'll-go" Negative presents or futures are regularly expressed by pre- fixing do "not" to indefinite forms; when more explicitly future, -Ce is suffixed to them. Examples of indefinite forms preceded by do are: definite: tdac yd/yi/t!a "bird is-flying" indefinite: dd/yd/t!a "he won't fly" definite: na/yd/SAdt/xwi "I vomit" indefinite: do/na/yAct/xwi "I do not vomit" definite: tc!d/sAsL/se "I cry;" tc!d/dil/se "you cry" indefinite: do/As/se "I'm not crying;" do/l/se "you're not crying" definite: ye/Bic/i "I saw him;" c/ye/Bi/l "you saw me" indefinite: do/yAc/t "I didn't see him;" do/ld/c/yl/l "you didn't see me" [definite: Ve/Bic/ya "I go;" VcBjya "he goes" indefinite: dd/VAc/yic "I'll not go;" do/VAc "he won't I go" {<*VAlyAc) 82 VI American Indian Languages 2 328 Imperatives are simply second person subjective indefinite forms. Examples are: yi/l "see him!" c/yi/l "see me!" nel/l "look at him!" (identical with indefinite present: nei/l "you're looking at him"); ne/cil/l "look at me!" nd/ni/An "stop him!" Prohibitives are simply imperative forms preceded by ia: la/yt/l "don't see him!" First person plural indefinite forms may have hortatory signifi- cance: nd/xwil/ye "let us play!" As regards form, definite tenses are primarily distinguished from indefinite tenses by the presence of second modal prefixes in the former, often also by the appearance of the first modal prefixes in a fuller form than in the latter; the presence of -I- or -i- in certain third person definite forms may also be recalled. It seems, further, that certain adverbial prefixes which have a short vowel (even though in an open syllable) in definite forms lengthen it in corresponding indefinite forms: definite: nd/da/yAct/t!d "I bathe;" na/da/yit/t!d "you bathe;" na/da/yAt/t!d "he's bathing" indefinite: nd/dAct/t!d' /Ve "I'll bathe;" nd/dit/t!6/Ve "you'll bathe;" nd/dAt/t!6/Ve "he'll bathe" definite: na/da/yil/el "we are bathing;" na/da/ydl/ cl "ye are bathing;" na/'xAt/da/yAl/el "they are bathing" indefinite: nd/dil/el "we'll bathe;" nd/dol/el "ye will bathe;" nd/xAt/dAl/el "they'll bathe" [definite: tc!a/yd/Bil/se "we cry;" tc!a/yd/ddl/se "ye cry" undefinite: tc!d/ydl/se "we'll cry;" tdd/ydl/se/t'e "ye will ' cry" These changes of quantity, however, are doubtless only second- arily connected with change of tense, as indicated, e. g., by Six: Athabaskan and Na-Dene Languages 83 329 tela- in definite singular forms: tcId/sAsL/se "I cry;" tcld/- dil/se/ha "do you cry?" It is very likely that we are dealing here primarily with considerations of syllabic and quantitative rhythm or balance.^* In Hupa Goddard has exhaustively shown that verb stems often assume different forms for different tenses and modes. This is very likely also true to a considerable extent of Chasta Costa, but I have but little material bearing on this point. A quantitative change is found in: j definite -el: na/da/yil/el "we are bathing" (indefinite -el: nd/dil/el "we'll bathe" -c characterizes indefinite forms in : definite -ya: Ve/Bic/ya "I go;" Ve/Si/ya "you go" indefinite -yAc: t'Ac/yAc/Ve "I must go;" Vl/yAc/Ve "you must go" negative indefinite -yic: d5/VAc/yic "I'll not go;" Id/t'l/- 1 yic "don't go!" f definite -ya: ya/da/yit/dja{<-t/ya) "we are ashamed" ^ indefinite -yAc: ya/dAcl/yAc "I am ashamed" Pronominal Objects. Pronominal objects are regularly prefixed to the verb. They come before deictic and first modal elements, but after adverbial prefixes. Thus, while not as thoroughly immersed in the verb form as the subjective pro- nominal elements, they cannot well be considered apart from it. The third person singular object is not designated. In form the objective elements are, on the whole, identical with the possessive pronominal prefixes of the noun. They are: Singular 1. c- Plural 1. 7i6- 2. n-, ne- 2. no- 3. — 3. xo- " Hardly stress accent as such. I cannot help feeling that such rhythmic phenomena will turn out to be of fundamental importance for Athabascan generally. 84 VI American Indian Languages 2 330 "He — them" or "they — them" is expressed by means of xl-. c- and n-, when standing at the beginning of a verb form, take no inorganic a- before them (contrast subjective Ac-). The definite forms of yed/l "he saw him" with combined pronominal subject and object are: With first person singular object: Sing. 2. c/ye/di/l "you saw Plural 2. c/ye/do/l me 3. c/yed/t 3. c/xA/yed/i With second person singular object: Sing. 1. n/ ye/Sic/ 1 "I saw Plural 1. n/ye/dit/l you" 3. n/ yed/l 3. u/xa/ yed/l With first person plural object : Sing. 2. no/ye/dd/l "you saw Plural 2. no/ye/do/l us" 3. no /yed/l 3. no /xa/ yed/l For no/ye/dd/l "you saw us" one would have expected *«03 na/nAc/An'o" djAn/la H/tlT/ni bachelor. "Unable I stop him," said dog-owner, t'wi yAn/na/Ac/t'e'"^ xwAn/ne/^iL/ya'"« djAn/la d6/at/t!T/ni. "all he will bring here." "You win," said bachelor ** As in note 94, except that another adverbial prefix, ah-, is present. " do/wa and -tiaI, as in note 94. i-, third modal prefix. -I'at, verb stem. '* ha, interrogative adverb, xw-, adverbial prefix, -i-, second person singular sub- jective pronominal element. -/-, third modal prefix. -I, verb stem, -ha, interrogative suffix. Indefinite present in tense. " A-, of unknown significance. L-, prefix common to several adjectives, -tcd/yi, adjective stem "big." ^'"^yAtt- and na-, adverbial prefixes, -ac, verb stem. Indefinite tense, because point- ing to future time. "" ia, numeral stem "one." -me/q!e, compound postposition, -ca, found also with la alone: Id'^/ca "one." '"' nd-, adverbial prefix, n-, first modal prefix, -l-, second person singular subjective pronominal element. -An, verb stem. Imperative mode. 103 Adverb containing negative do-. Perhaps -qle is postposition (cf. -me/qle). ^^ no.-, n-, and -ah, as in note 102. -c-, first person singular subjective pronominal element. Indefinite in tense, because of preceding negative adverb. "* As in note 100. -Ce, future suffix; here used because idea of futurity is more explicit. ">* xtf /in-, adverbial prefix, we-, first modal prefix. 6-, second modal prefix, -i-, second person singular subjective pronominal element. -L-, third modal prefix, -ya, verb stem. Definite present in tense. Six: Athabaskan and Na-Dene Languages 93 APPENDIX. A few Galice Creek words were obtained from Mrs. Punzie, a few Applegate Creek words from Rogue River Jack. These two Athabascan dialects are probably practically identical. 5 indicates something acoustically midway between 5 and c; r (tongue-tip trilled) and / occur as reflexes of Athabascan n; nasalization (indicated by ') seems to occur, k' and k'w are found as contrasted with Chasta Costa x and xw. Galice Creek. ya'/k'ds "seeds (sp.?);" said to be called bdnax or b4yu in Chinook Jargon tcla/ba/d/k'wa's "brush used for medicinal purposes (sp.?)" L!d'/ddi "tar- weed" (cf. Chasta Costa ild'/de; Hupa Lo/daitc) yel/yat/ts!ai/yt "sunflower" (cf. Chasta Costa, tc I Al/y at /ts/e) giis "camass" (cf. Chasta Costa god; Hupa kos "bulbs") ddl/si "pine" (cf. Chasta Costa dAl/si; Kato dul/tclk) Id I lH "pine-nut" (cf. Chasta Costa nd/de) del res "manzanita" (cf. Chasta Costa dAJnAc; Hupa din/nuw; Kato tun/nuc) md'/ts!i "cat-tail" (cf. Chasta Costa mAt/tdi) sds/da' "oak" (cf. Chasta Costa cAc/dd') i/dd/ge "acorn" (perhaps misunderstood; cf. Kato i/taG "black oaks") Applegate Creek. k'q'/tc'u "goose" (cf. Chasta Costa xd'/tc'u; Kato ka') dAc/tc'u "grouse" (cf. Chasta Costa dAc/tc'u; Kato dAcf- tco) (339) 94 y^ American Indian Languages 2 340 dAclVe' Itc'u "bob-white, quail" k!ai'/dic/tca/we "ruffed grouse, pheasant" k'dn/ta/tc'u "pigeon" (cf. Kato kwl/ylnt) do/s'An/tsfa/ya "screech-owl" si lick lies "kingfisher" tddlkeldi "red-headed woodpecker" tea lwdcltc{!)e "sandhill crane" (cf. Chasta Costa sd'lwAsl- tsle) Editorial Note Originally published in University of Pennsylvania, Anthropological Pub- lications 2(2), 271-340 (1914). Reprinted by permission of the University Museum, University of Pennsylvania. The Athabaskan languages of southwestern Oregon — which are now (1989) virtually extinct — constitute a distinctive subgroup whose relationship to the northwestern California subgroup is surprisingly distant, although the two are usually classified together as "Pacific Coast Athabaskan" (Hoijer 1960). The Oregon group seems to have comprised four distinct, though closely related, languages: Upper Umpqua, Galice-Applegate, Chetco-Tolowa, and Lower Rogue River. The last was spoken in a variety of local dialects, the best known of which are Chasta Costa (on Rogue River about 30 miles upstream from the mouth), Tututni (near the mouth of the river). Euchre Creek (along the coast to the north of the river mouth), and Upper Coquille (still farther north, in the inland area behind Coquille Bay). No full grammatical study exists of these or of any Oregon Athabaskan language. Besides Sapir's description of Chasta Costa, the fullest published documentations of Oregon Athabaskan are of Galice (Hoijer 1966), Tututni (Golla 1976), and Tolowa (Bright 1964; Bom- melyn et al. 1989). Considerable manuscript documentation also exists, partic- ularly of Galice (by Melville Jacobs; see Seaburg 1982) and of Lower Rogue River (by J. P. Harrington; see Mills 1981: 69-76). Some of Harrington s Lower Rogue River material (including a number of aluminum disk recordings) was obtained from Wolverton Orton, Sapir's Chasta Costa consultant. Corrigenda to Father Morice's "Chasta Costa and the Dene Languages of the North."' In view of the fact that Father Morice has reviewed my Notes on Chasta Costa Phonology and Morphology in so evidently a friendly spirit it may seem a bit churlish to point out what seem to me to be either slips or misunderstandings in his recently published paper on Chasta Costa and more northern Athabaskan dialects. If, nevertheless, I venture to do so, it is not because of any desire to minimize the value of Father Morice's paper or to attach an overweening importance to my own very scanty contribution to Athabaskan linguistics, but to help advance our understanding of the problems of Athabaskan phonology and morphology. The chief value of Father Morice's paper seems to me to lie in the further light it throws on the Carrier language, of which previous papers have already shown he has an admirable mastery. I earnestly hope that Father Morice will not be content with the rather sketchy papers he has hitherto given us on the Carrier language, but will eventually publish a complete presentation of the intricacies of its phonetics and grammatical structure. I. " Dr. Sapir's a is my as, almost the sound of u in ' but,' more exactly that of e in the French ^e, te, le" (p. 560, footnote). My a of Chasta Costa words is to be pronounced like a of- German Mann and thus in sound corresponds to Carrier a, not (b. My A is another vowel alto- gether, though often phonetically reduced from original a. It is prac- > See American Anthropologist, N. S., 17, 1915, pp. 559-572. 96 VI American Indian Languages 2 766 tically identical in sound with u of but and doubtless corresponds phone- tically, largely also genetically, to Carrier o^. 2. "This [C. C. tclac 'bird'] is evidently none other than the Carrier tscrz [Morice's s and 2 are described by him in a letter as sibilants midway between 5 and c, and z and j, respectively; they correspond histori- cally to Chipewyan^and t? < Ath, s and z], which in that language means not ' bird ' but 'feather-down' " (p. 560, 1. 15). This is not so evident. In fact it is phonologically impossible. Father Morice's Carrier form is clearly cognate with Chipewyan -d'Hd (Goddard; my -dUd) "soft feathers," Montagnais, tthoepdh (Petitot; tdlA-yd^ in my orthography) "plume servant k broder, cote de plume " (possibly mistranslated for " duvet "), Hare kkwew (my k^\ew), Loucheux tthaw (my td\aw), Navaho ts'os (Franciscan Fathers; my /5!o5) "down feather." These forms imply original Athabaskan *ts\ez (reduced *ts\ez, *ts\es) and *ts\os "down." This would be expected to appear in Chasta Costa as td\Ad (very likely actually found in td\Ad-nd-yal-td\od " humming-bird," perhaps literally "soft-feathers fiy-about-making-a-humming-sound"; for verb stem •tdlod cf. Hupa -tsots, -tsos " to make a kissing-Iike noise, to smack one's lips," Nav. beets' OS " whistle " as noun). C. C. tc\ac can have nothing to do with this tdlAd, but must go back to Ath. *k^\ac or *tc\ac} 3. " This possessive [C. C. -lUc\e '(his) dog '] entails in Chasta Costa the accretion of a sort of suffix which he writes tele, the equivalent of my tse. Now litse means in Carrier, not somebody's dog, but she-dog! " (p. 561, 1. 3). Father Morice's quandary is of his own seeking. -ts\e of his Carrier form is not at all connected with -tele of my Chasta Costa one. Carrier li-ts\e is simply compounded of li "dog" and ts\e " woman." This ts\e (Ath. *tc\e) is found in Carrier tsekhk, Montagnais ttsh-kwi, ttse-k'^ (i. e., tsle-k'e), Hare ttse-Une, ttse-k'u, Loucheux ttse-ndjo, Kato tc'ek, C. C. ts\dxe. Carrier, Montagnais, Kato, and Chasta Costa point to Ath. *tc\e-k'e " woman." C. C. -tele of -li-tcle, however, goes back, not to Ath. *-tcle, but to Ath. *-k^le with glottalized anterior palatal k^l; cf. Hupa -liii-ke " (his) pet " (read- k^le). How this peculiar possess- ive *-k''le (cf. normal -e in, e. g.. Hare se-llin-e " my dog ") arose I can not say. Its isolated character stamps it as probably archaic. Perhaps -jfe"!- of Ath. *-liiiki'\e is connected with -g-, -k of Ath. *lik, *hg-, which is parallel to *ltrj; cf. Anvik hlik (Chapman; my lik), Loucheux V egce-tsellce > Since this was written Dr. Goddard has kindly called my attention to Hupa klyawM (read k^lauW) " birds " which corresponds exactly to Ath. *k''!ac. This elimi- nates Ath. *lc!ac as possible prototype and still more effectively disposes of Father Morice's analysis. Six: Athabaskan and Na-Dene Languages 97 767 " petit chien " (i. e., legA-). (Incidentally, Petitot gives for " chienne " Montagnais I'intse, Loucheux Ventsi. Are these forms errors for -ttsi, -ttsi}) 4. " The- . . . , at least in the north, does not merely mean ' in the water,' as Sapir would have it . . . but it hints at the ' bottom of the water ' " (p. 561, 1. 25). C. C. t^e- was defined as " in the water." That, in Pacific Athabaskan, it has no necessary reference to " the bottom of the water " is showji by several of Goddard's Hupa and Kato examples. Thus, some Hupa verbs with t^e- are translated " in the water it seemed about to tumble," " in the water she floated back," " when he put his hand in it (i. e., the water)," cf. Kato <'e'- verb " she washed them." Evidently Goddard's definition of Hupa /'e- as re- ferring " only to motion into water and under its surface " seems quite adequate. In Carrier thentltlat " thou art precipitately brought to the bottom, thou sinkest," second modal -n- (to use Goddard's terminol- ogy) is terminative in function, so that t^e-n- . . . -tlat necessarily denotes " to sink to the bottom of the water," even if Ce- by itself be assumed to mean " in the water." A wider range of significance for Ath. *t^e- than Father Morice would assign it is implied also by such Navaho forms as tqe-li "water horse"; Navaho tqe- (i. e., my txe-) regularly corresponds to Ath. *Ce-. Navaho tqe- forms are likely to be archaic, as Ath. *Ce- has been almost entirely superseded in Navaho by Ath. */'a- (Navaho tqa-); see below. Even in northern Athabaskan I do not find Father Morice's remark to apply without qualification, even if correct for Carrier. In Anvik Ath. *t^e- appears as te-, tH- (Chapman's orthography); note teyldoutel "floating in the water." In Loucheux Ath. *Ve- appears as tchi- (Petitot's orthography), i. e., tcH- (Loucheux Ve-, i. e., /'e-, does not correspond, according to Loucheux phonetic laws that I have worked out, to Ath. */'e-, but to Ath. *t'a-)\ note tchi-dhitlle "gtre k flot." 5. " Tha- . . . , which he gives as ' referring to the water ' has really that signification [i. e., ' in the water '] " (p. 561, 1. 27). Ath. *Ca- is undoubtedly found employed locally (" in the water ") in prac- tically all Athabaskan dialects; indeed it seems in some dialects (e. g., Navaho, to some extent apparently also Chipewyan) to have extended its sway at the expense of parallel Ath. *t^e-. There is, however, plenty of evidence to show that Ath. *Ca- is frequently substantival, not local, in force, that it is, in fact, an old noun stem (" water, wave, sea," parallel to more wide-spread Ath. */'o " water ") that has become a verb prefix. A local meaning is impossible, for instance, in the numerous verbs of 98 VI American Indian Languages 2 768 drinking found in Pacific Athabaskan (e. g., Hupa ta-win-nan " he drank it," Kato ta-ya'-o-nqn " let them drink," C. C. t'd-yAct-nd " I drink "). A direct substantival meaning, moreover, is obvious in such compound nouns as Navaho tqd-bd " shore " (lit. " water-edge "), while in several Mackenzie Valley dialects Ath. */'a even occurs as un- compounded noun stem (Montagnais t'a, i. e., my t^a, " flot, onde "; Hare t'a; Loucheux t'i:). 6. " As to the verbal stem -al, which he believes to mean ' to come,' I more than suspect that it is but a corresponding form of -ya, which he represents as expressing the idea of ' going, coming,' and should be -yal " (p. 562, 1. 11). That C. C. an-yi-al really means "come on!" and not, like Carrier 'oen inyal, "go on!" (as suggested further on by Father Morice, 1. 26) is conclusively proved by two facts. In my brief Chasta Costa text (p. 337, 1. 3) an-yi-aL occurs as translation of English " come here! " (this text, it should be remembered, was translated from English) under circumstances in which " go on!" would be quite out of place. Further, during my residence among Chasta Costa Indians in Siletz Reserve, Oregon, in 1906, I distinctly remember that an-yi-al was often used by elders in calling children to them. As to whether recorded C. C. an-yi-al, an-yi-aL is to be understood as an-yi-yal, as claimed by Father Morice, or as an-yi-'al, as I had assumed, I now incline to think that -yal is correct. .This is because of such forms as Hupa ivin-yaL " come on " and Kato gUn-yaL " walk," which seem to cor- respond exactly to C. C. -yi-{y)al. Should -'al prove to be correct, it would probably have to be compared with Loucheux -'a " to go," Ath. *-ac " two go " (with dualic -c). 7. " Dr. Sapir furthermore quotes the verb stem -t\o as denotive of the act of swimming, while, according to him, that of paddling is rendered by the radical -xe. Now, in most northern Dene dialects, the former refers to paddling, while the latter indicates the act of navigating, or moving about in a canoe" (p. 562, 1. 28). True, but there is plenty of evidence to show that Ath. *-k^e, *-Ven, *-k'^er}, *-kH frequently refers to or implies paddling. Thus, Goddard translates Chipewyan -ki by " to paddle a canoe, to travel by canoe " (e. g., ta-kt-hwu " when he paddled "). This is confirmed by Father LegofT, who defines 'ke'l (i. e., -k'el): " est le progressif de 'ki, et signifie proprement naviguer, en ramant " (e. g., pes-'ke'l " je rame, j'avance en ramant "). Similarly, for " ramer " in Hare Petitot gives not only e-ttoh, but also e-k'e. And in Anvik I find -kan, -kahl often translated as " to paddle " (e. g., xH-kahl '" he is paddling," ti-qH-kan " he paddled on "). For C. C. -t\o " to swim, Six: Athabaskan and Na-Dene Languages 99 769 bathe " I find no exact parallels. It may, as Father Morice suggests, have primarily meant " to paddle " (denominative verb from Ath. *t\os " paddle ") and changed its force dialectically. 8. " The verbal stem -lal . . . , to which our author attributes the sense of ' to sleep,' has in Carrier the value of ' to dream of ' (with a complement). Might not Dr. Sapir's informant have misunderstood his questioner and thus unwittingly misled him? " (p. 563, 1. i). I think we can manage without this hypothesis of misunderstanding. In Hupa -lal, -laL means not only " to dream," but also " to sleep " (e. g, nit-te-sil-lal-le " you would go to sleep "). Both meanings are given also for Kato -lal, -IqL (e. g., n-to-lqL " let him sleep "). Turning to northern Athabaskan, we find that in Chipewyan (Montagnais) -lal, -lal, -lal, -lal regularly means " to sleep, to fall asleep, to put to sleep " (e. g., Goddard's hi-teL-lal " he is asleep"; Petitot's in-t'es-l'al s. v. " dormir debout," es-l'al s.v. "endormir"; Legoff's in-'tes-la'l " je m'endors "), while for " rever " Petitot gives quite another stem in Montagnais and Hare. C. C. t'e-d-lal-{la) " he's been sleeping," with its prefixed elements t'e- and (in definite forms) -6-, corresponds remarkably to Hupa -te-sil-lal- quoted above, Kato {n-)te-s-laL " he went to sleep," and Chipewyan {ln-)te-d-Lal " he is asleep." In Hupa and Kato -lal, -lal, when meaning " to dream " seems regularly preceded by 7ia- (e. g. Hupa kin-na-is-lal " he dreamed," Kato u-na-s-laL " he dreamed about "). 9. . . . '^ -lal, or rather -tlal. ... As may be seen by Sapir's rendering: t'Ulal. The double consonants tl and ts are of frequent oc- currence in Dene and form as many indivisible groups. The syllables of all Sapir's verbs in the first person plural are wrongly cut up: the t which he attributes to the penultimate syllable should commence the last one: -tlal, -tsoel, -tlat, -thcec, etc. Hence several of his verb stems are incomplete. For instance, -se, 'to cry,' should be -tse (Carrier -tso)\ -si, ' to cause,' should be -tsi (Carrier -tsi); -lo, ' to laugh,' cannot be understood without its /. . . . Were he familiar with the Dengs' syl- labic way of writing their own language, he would have been spared this little inaccuracy " (p. 563, 1. i and footnote i). There are several com- ments I wish to make on this passage. (a) That, in Chasta Costa, not -tlal but -lal is to be considered as true verb stem is quite unmistakably evidenced by such forms as t'l-lal " thou art sleeping," I'o-lal " ye sleep." Were -/- part of the stem, there would be absolutely no reason for its disappearance in these forms (cf., for phonetic comparison, t'd-yit-nd " thou drinkest," I'd-yd-dot-nd 100 VI American Indian Languages 2 770 " ye drink," in which third modal prefix -t- is preserved between personal element -i-, -6-, and stem consonant n). These remarks apply also to C. C. -lat, not -tlat, as Father Morice would have it. Comparison with Hupa and Kato, also Chipewyan, abundantly confirms my own analysis. (6) There is no point whatever in quoting C. C. t'it-lal (not t'ltlal, as Father Morice has it) " we are sleeping " as evidence of a stem -tlal. In Chasta Costa -{i)t- regularly appears as first person plural subjective prefix, except, as in Chipewyan, before third modal -/- and -/- (e. g., ye-dit-'i " we saw him," where Father Morice would hardly claim that -t'i, not -'i, is the true stem). That this -t- is indeed an organic element in the first person plural prefix is shown by comparison with Hupa -d- (e. g., na-dtL-Le " we are painting "), Kato -d- (e. g., na-ddl-yic "let us rest "), Chipewyan -it-, -d- (e. g., e-gU-'t " we saw it "), Loucheux -di- (e. g., i-di-kwoll " nous amenons "). What has apparently misled Father Morice in his analysis of Chasta Costa is that Ath. first person plural (and dual) *-{i)d{e-)- has in several dialects, perhaps by analogy of third modal -I- and -/- verbs where -d- regularly disappeared (I am inclined to think that in Hupa and Kato -d- has in these cases been restored by analogy), been replaced by -i-, -i-, e. g.. Carrier nd-i-ta " we are both sick " (Carrier first personal plural prefix tsoe-, i. e., ts\A-,'^ does not seem to be a widespread Athabaskan element, though Father Jette has recorded for Ten'a ts- forms of like meaning, perhaps to be understood as ts\-); Hare i-ssi "nous deux faisons " {ya-issi "nous faisons"); Navaho ch'i-ne-i-ka " we two carry milk out " {ch'cB-de-i-kd " we carry milk out "). Even in these dialects, however, survivals of the old -d- are found in such cases as allowed of its carrying over into the stem syllable, e. g.. Carrier nt-tas " we two walk," i. e., nl-tlas, morphologically equivalent to n-U-'as (cf. noe-hce-'as "they walk two together"); Hare witta " nous deux allons," i. e., un-tla (Ath. *zi-tlas), morphologically equivalent to w-it-'a (Ath. *z-id-'as); Navaho da^-dt-t'a^ "we put a round object on " (equivalent to da^-d-it-'a'), bd^-ni-'nil " we two put several things on " (equivalent to ba^-n-U-nil), l-gye " we two marry" (equivalent to id-ye, cf. i-ye " he marries "). There can be no talk in such cases of stems -t\as, -t\a\ -'nil, and -gye but only of phonetic resul- 1 1 strongly suspect, as is indeed indicated by Father Morice himself (see "The Dene Languages," 1891, p. 193) that this /foj- is really impersonal in meaning, ta begin with; with Carrierna;-/5a5-/»7 "we walk"; cf. nas-tsi-ya "one walks." For reasons of modesty, perhaps, it may have become customary to say "people (in general) do so and so" instead of " .ve do so and so." This indefinite Carrier tsce- is quite possibly cognate with Hupa U-, i. e., tc!-, of third personal forms applying to adult Hupa (e. g., tce-xauW "he is catching," originally perhaps "one catches"). Six: Athabaskan and Na-Dene Languages 101 771 tants of first person plural -d- plus stems -as, -'a\ -nil, and -ye (Ath. *-7e). (c) As regards syllabic division, I must emphatically disagree with Father Morice. It is quite clear that in many Athabaskan dialects the -/- {-d-) of the first person plural belongs to the following syllable, as we have just seen. This proves nothing, however, for Chasta Costa, nor do I see how familiarity with the Carriers' syllabic way of writing their own language would have materially helped my ear in its perception of Chasta Costa sound combinations. As a matter of fact, my informant Wolverton Orton was particularly careful in syllabifying and I was prac- tically never in doubt, in recording his forms, as to where syllabic di- vision was to be placed. Hence t'ii-lal is not to be " corrected " to tH-tlal, t'e-nit-lat to t'e-ni-tlat, yit-lo to yi-tlo, tclAt-t'it-dAl to tc\At-Ci-tBAl (Father Morice's -tscel). Father Morice's -thcec is simply another ortho- graphy for my -t'^Ac. I may point out that were we dealing in these Chasta Costa forms with such consonantal groups (affricatives) as // and tS, preceding i would appear lengthened to I, because completing its own syllable. That I have consistently misheard both vocalic quantity and syllabic division in these words (e. g., -iHt-dAl for -Cl-tdAl) I cannot admit. Naturally I do not deny that td and td\ also occur in Chasta Costa as true affricatives, though Id seems to have regularly developed to 6. As to C. C.-t-lo (I have no example of C. C. affricative il, dl, nor does it seem to occur in Hupa or Kato), I am uncertain, as I pointed out in my " Notes," whether this is to be understood as directly representative of Ath. *-dlo (as preserved, e. g., in Carrier, Chipewyan, and Navaho) or as analyzable into third modal -t- and stem -Id. For Ath. dl: I, cf. Ath. *-dlo, *-dli " to be cold ": Montagnais e-llu " cold," Kato Id " frost." C. C. -Id " to laugh " might well correspond to Kato -Id " to deceive " (primarily ".to laugh at"?). (d) As regards C. C. -se " to cry " and -si " to cause," Father Morice is going altogether too far when he says these stems are incom- plete for -tse and -ist. As I have already stated in my " Notes " (p. 287), Ath. tc\ of which Carrier ts is reflex (Ath. /c' is preserved as such in Navaho, Jicarilla Apache, Kato, and, as labialized tc'w, in Hupa), has always been simplified in Chasta Costa to 5 (Ath. tc^>ts>s; cf. Ath. ts>td>d). Hence to " correct " C. C. -se to -tse is as justifiable as it would be to " correct " French chef to *kep because, as no one denies, derived from Latin caput. ID. " If Dr. Sapir will allow me, I will also observe that the desinence -tc'ac [probably misprint for -t^Ac], which he gives as a distinct verbal 102 VI American Indian Languages 2 772 element, is nothing else than the plural stem of the same [Carrier] verb nanisthi, whose derivative nthcesthih effectively means ' to lie down, go to bed ' " (p. 563, 1. 7). Father Morice then goes on to compare C. C. -t^Ac directly with Carrier -this, -thez, -thcES (plural stem). These state- ments are misleading. It is evident from my Chasta Costa material (e. g. dd-nAc-t^Ac " I go to bed ") that -t^Ac applies to singular subjects. This is confirmed by comparison with Hupa -tiiW (which corresponds regularly to C. C. -t^Ac) and Kato -tUc (e. g., Hupa tcin-ne-tiiW " she goes to bed," Kato na-nHn-tilc " lie down again "). This Ath. *-t'ec is characteristic in Pacific Athabaskan of " indefinite " ^ forms as contrasted with " definite " Ath. *-t^in, *-t^er}, *-t'i (Hupa -ten, -tin; Kato -tin; Carrier -thi; Montagnais -Vi; Navaho -tqi); in most Athabaskan dialects *-i'e or *-t^el is used in " indefinite " forms (Hupa imperative -te; Mon- tagnais eventual -Ve; Navaho present -tqe, future -tqel). " Indefinite " -c forms are in general apparently characteristic of Pacific Athabaskan. C. C. -t^Ac is phonetically practically identical with Carrier -thces, but not morphologically. Carrier " proximate future " -thces is reduced from present -thes; in other words -5 (Ath. -c) is here found in all forms, " definite " and " indefinite." This plural (and dual) stem Ath. *-t^ec, *-t'ej, *-t^ec (cf. Hupa -tetc; Montagnais dual -/'ez; Navaho -iec; Jicarilla Apache -kec) contains dualic -c (cf. Ath. -'ac " two go "; Navaho -'esh, -'ezh, -tsh " to act upon two animals "). We now see clearly that Father Morice was misled by a phonetic convergence of morphologically distinct, though genetically related, forms. II. " Unless I am very much mistaken, what he adduces as the equi- valent of ' I am seen, you are seen, he is seen,' etc., really means simply: ' people see me ' (French: on me volt), ' people see thee,' etc." (p. 563> 1. 22). Father Morice is, in my opinion, quite right. In fact he merely repeats what I had already pointed out (p. 332): " tslAl-, which appears in these forms, probably contains third modal -I- preceded by deictic ^5!- implying indefiniteness of logical subject: ' mann sieht mich.' Surely Father Morice knows that German man sieht mich is identical in force with French on me voit. I take this opportunity of modifying my analysis of C. C. ts\A-. I do not now think that it is comparable to C. C. deictic tc\-, which denotes lack or indefiniteness of object, not subject, of transitive verbs; this tc\- goes back to Ath. H!(e)-, which is preserved as such in Hupa. C. C. ts\- (Ath. *tc\-) of such forms as nes-ts\Al-'i " one sees me, I am seen," ' I use "definite" and "indefinite" in Goddard's sense. "Definite" tenses are present definite and past definite, all others are "indefinite" (including present in- definite, imperative, eventual, and other forms). Six: Athabaskan and Na-Dene Languages 103 773 however, is clearly subjective and impersonal in force and very probably corresponds to Carrier indefinite ts{oe-), see 9 b above. Goddard's Chipewyan ts'- is probably identical with this Chasta Costa and Carrier impersonal prefix (e. g., ts'e-Lu " he was caught," i. e., " one caught him "); this means that Goddard's comparison of Chipewyan ts'- with Hupa ^(^!)- is incorrect (see p. 133 of his " Analysis of Cold Lake Dialect, Chipewyan "), and indeed we should in that case, as in Kato and Chasta Costa, have expected tc\-. I do not know how to reconcile with these subjective impersonal forms (Ath. */c!-> Kato tc\-, ts\-, s'-; C. C. ts!-; Carrier ^5!-; Chipewyan ts\-) Petitot's Montagnais, Hare, and Loucheux indefinite tse- forms. Could he have throughout misinterpreted is\e- (in his orthography Use-) as tse-? E. Sapir Editorial Note Originally published in American Anthropologist 17, 765-773 (1915). Reprinted by permission of the American Anthropological Association. This reply to Morice's criticism of Sapir's Chasta Costa work (Morice 1915b) prompted a rejoinder (Morice 1917). Both of Morice's papers, together with his original review of Sapir's monograph, are reprinted in the appendix to this volume. THE NA-DENE LANGUAGES, A PRELIMINARY REPORT ^ 1"^HE problem attacked in this paper is that of the genetic relationship of Athabaskan, Haida, and Tlingit. Important morphological, to a less extent also lexical, resemblances between Haida and Tlingit have long been pointed out by Boas and Swanton, resemblances which have led them to assume, though rather hesitatingly, genetic relationship between these languages. Boas has also somewhat vaguely hinted at fundamental resem- blances in structure between Athabaskan and Haida-Tlingit, but no concrete evidence has been given on this point. A full presen- tation of the comparative lexical, phonological, and morphological evidence that serves to show, beyond all reasonable doubt, that Athabaskan, Haida, and Tlingit are indeed but divergent represent- atives of a common prototype is given in an extensive paper on "The Na-dene Languages" now in course of preparation as a memoir of the Anthropological Series of the Geological Survey of Canada. The present sketch, prepared at the request of Dr P. E. Goddard, is merely a rapid abstract of some of the leading points involved. I wish expressly to emphasize the fact that it does not present all the evidence at my disposal. While, however, it does not constitute the complete demonstration of my thesis, I believe that enough is here given to remove this thesis beyond the realm of the merely probable. The term "Na-dene," which has been chosen to designate the hypothetical prototype of Athabaskan, Haida, and Tlingit, will be explained in the latter part of the paper. I. Morphological Features Stem and Word Form. — The most typical and doubtless his- torically primary type of stem form found in the Na-dene languages is the monosyllabic stem consisting of consonant plus vowel; in 1 Read in substance before the Anthropological Association at Philadelphia. 534 106 VI American Indian Languages 2 535 Haida (H.) the consonant may be replaced by a cluster of two consonants which, in cognate words, appears contracted to a single consonant in Athabaskan (Ath.) and Tlingit (Tl.). Examples are: Ath. *-tsi^ "daughter," *t'o "water," *tl!o "grass," *-k'e "foot," *-ne "to speak," *-ya "to stand (plur. subj.)," *-'a "to find," *-ya "for," *-na "around," *na- "again;" H. tc/W "cedar," q/a "harpoon," st/a "foot," Iga "rock," t'a "to eat," q!a "to sleep," xa "to follow," sa "above," gu "at," q/o- "by means of the teeth;" Tl. t^a "stone," nu "fort," xa "enemy," ha "to dig," g'a "to say," ci "to hunt for," t!a "behind," k^a "on," dJL- "quick- ly." Many, perhaps all, elements consisting of a single con- sonant (or cluster of two consonants) are phonetically reduced owing to the loss of a vowel; e.g., Ath. *-n, *-77 "person" <.*-ne; H. stU- "with the fingers" My interpretation of Tl. -/, -ya, and -n as syntactically specialized postpositions diffeis from Swanton's, at least as far as expressed in his grammatical sketch. 122 VI American Indian Languages 2 551 II. Comparative Vocabulary The lexical evidence bearing on the genetic relationship of Athabaskan with Haida and Tlingit comprises, at the moment of writing, over three hundred distinct Athabaskan stems and gram- matical elements which can be, with greater or less probability, assigned to the reconstructed Na-dene language. Only a selection, comprising less than one third, of this lexical material is here presented. The arrangement is alphabetical, from the point of view of Athabaskan. Athabaskan I. a- demonstrative stem 2. Haida a- dit. -ade "elder sister" 3. -ca-rj obligatory future 4. -ca, -cal "to catch with a hook" 5. -d, -de "at, to" 6. da "what?" 7. -do, -da/ "to go, to travel" 8. -da-f) "to drink" 9. -das "to burn" 10. -de/ "several go" 11. del "crane" 12. di "this" 13. -dja hortatory 14. djarj "mud" 15- -ga?7 "to be mouldy" 16. -go " toward" 17. -gid, -yid "to dive" 18. -7a "for," -7an "to" 19. -ya "to go" 20. -ye, xe "grease" 21. -yed "to run" 22. -yel "to be dark, night passes" -sa-rj infallible future djd "bait" -da "to" -dal "to move along" da'dj, {das) "live coals" dal "several go by land" di'la dit. dei "just that way" -dja-T) dit. tc^a'n dit. gu'na "decayed" gtia, gui dit. gi' dit. ga "to," gan "for" -ga "to go in order to" ga't dit. ga'l " night" Tlingit a- dit. a't "father's sister, father's sister's daughter" cat "to seize" -/, -dt "to" da' "what?" dana "to drink" at "several go" du'l dit. de "now" ga "for" -ya "to go to" e'x "grease" get "to get dark' Six: Athabaskan and Na-Dene Languages 123 552 Athabaskan 23. -ye, -yel "to kill, to fight" 24. -ywo "tooth" 25. hai "that" 26. he-, xe- "they" 27. X0-, he- "he, him" 28. -^'a "liquid has position" 29. -k'an, -k'a "to fish with a net" 30. -k^a "on" 31. -^'e personal noun plural 32. k'ene "friend" 33. kla "arrow" 34. kleri "withes" 35. -klan "to burn" 36. -k!e "on" 37. -k'os, k'es "to tie" 38. 4 "with" 39. 1-, la negative 40. la "one" 41. -la "to jump" 42. -lad "end" 43. lo', lok' "fish" 44. -///a "butt; behind" 45. me- "he, it" 46. man "edge" 47. mes "cheek" 48. -n, -77 local postposition 49. -na "to die" 50. -nan "to drink" 51. -ne, -n "person, people" 52. -neg, -leg "to relate" 53. -ne "to play" 54. -ni "to touch, to do with one's hands" 55. no "place of retreat, is- land" 56. -onay{e) "older brother" 57. 5- durative verb prefix Haida gail "to fight" hao " that " xao "liquid" xao "to fish" Tlingit -xa distributive suffix q!a' "harpoon' qian "grass" k^u "to tie" al "with" ila'- "the first' tla' "to dive" tlan "end" -dlga "after" wa- "that" -n, -rj general post- position ntal, nil dit. na "to live; house" nar] dit. ux "tooth" he " this " has "they" hu "he" -k'a "on" -na-x distributive num- eral suffix xo'n "friend" q!a "point" q!a'n "fire" -q! "at" I negative tie'- "one" tUu'k! "cohoes" we "that" wan "edge" wac "cheek" -n "with," also local postposition na "to die" na "people" nik "to tell" ni "to put" mi "fort" hu7ix "man's older brother" s- modal prefix 124 VI American Indian Languages 2 Athabaskan 58. -sen "to hide" 59. sil "steam," -sil "to steam" 60. -/'a "among" 61. /'a "wave" /'a- verb prefix referring to water 62. -t'an "to eat" 63. -t'e "to look for," -t^an "to look" 64. -/'e "to be cold," /'en "ice" 65. -t'e "to lie" 66. i^ez "night" 67. -tla "because of" 68. tla "feather," -t!a "to fly" 69. -tla "tail," tIa-Tj "back- wards" 70. -ties "to step" 71. -tlo "to shoot" 72. -tlod "to rub" 73. tsa "ring-like object" 74. tslai "dish" 75. -tslen "bone" 76. tsli "again" 77. -/c't "grandfather" 78. -/c'ot; " mother-in-law " 79. tela "hat" 80. tcio "fir, spruce" 81. -ifa "for" 82. xa- "up, out of" 83. xfl' "goose" 84. xa-T) "quickly" 85. -xan, -yan "to grow up," xan "old age" 86. xin "song" 87. -ya "to stand ' (plur.) 88. -yan "to eat" Haida sil "to steam" t^a-oan "alongside of" t^a'ria "sea-water" fa dit. /'a/ "cold" /'at, /'t "to lie" -tla dit. tiagun "feather," tlao- " feather-like object ' -tial "back of " 553 Tlingit Sin "to hide" si't "to cook" -Ca'-k^ "in the middle of" /'a'g/ "to chew" tiln "to see" t^a'dj "cold" /'at "to lie," /'a "to sleep" /'c*/ "night" tia-wu "feather" •tla "behind" tla "to step" st^a- "ring-shaped ob- ject" sqlao "to put in a dish" tc\n dit. djo'n dit. tcju "cedar" k'wagi "above" xaha "mallard" xao- "to do a thing quickly" ga "to stand" tluk "to shoot' tlus "to rub" shql "dish" sla'q "bone" tslu "again" tc^a'n dit. sja'x" "hat" -xa-n "to" k^e "upward" can "old person" ci' "song" ya'n " to eat" Six: Athabaskan and Na-Dene Languages 125 554 Athabaskan Haida Tlingit 89. ye- "that," y- "he" gai 'this, that, the" ya "this" 90. -ye "at the foot of, under" -yi' "down in" 91- 92. -ye personal noun plural ye "supernatural being," -yen "to practice sha- manism" -ye-n plural of terms of relationship ye'k "supernatural helper" 93. -ye "in" -gei "into" •ge' "inside of," -yi-k "inside" 94- -ye suffix making relative clause -7i suffix making rela- tive clause 95- yo "that yonder" gu' there" yu "that yonder" 96. -'a "to go" q'a 'to walk" 97- -a "to tell, to sing" g'a "to say" 98. -'in "to see, to look" q\v dit. gen dit. III. Phonology The phonetic systems of Athabaskan, Haida, and Tlingit, despite a good many differences of detail, present important points of similarity. Three types of stops are found in each : intermediate (or sonant), aspirated surd, and glottalized (fortis). Sibilants and sibilant affricatives, k-spirants, and laterals are well developed. A remarkable phonetic feature held in common by the three Na- dene languages is the paucity of labials; b, p\ and p! were clearly not found in Na-dene (b and />' are rare Haida sounds), m existed only doubtfully (Ath. m, whence b in certain dialects, is not equiva- lent to Haida m, but to Haida-Tlingit w), while w was certainly found. Athabaskan has lost the old velar series of stops as such, while Haida and Tlingit have preserved them; on the other hand, the Na-dene anterior palatals, best preserved in Haida, have been lost as such in Tlingit. In the more elaborate paper on the Na-dene languages in course of preparation, the historical relationship of the Athabaskan sounds to their Na-dene prototypes and Haida and Tlingit corre- spondents is systematically worked out on the basis of all the evidence available. Here it will suffice to point out some of the more important correspondences, referring to the numbered entries of the comparative vocabulary for illustrative examples. 126 VI American Indian Languages 2 555 Stopped Consonants 1. Ath. d: H. d, -t: Tl. d, -t (nos. 2, 5, 6, 7, 8, 9, 10, 11, 12, 21) 2. Ath. /': H. /": Tl. /' (nos. 60, 61, 62, 63, 64, 65, 66) 3. Ath. //: H. //: Tl. // (nos. 67, 68, 69, 70, 71, 72) 4. Ath. g: H. g: Tl. g, -k (nos. 15, 16, 52) 5. Ath. /fe": Tl. k' (no. 30) 6. Ath. /fe": H. x: Tl. y (nos. 28, 29, 31, 32) 7. Ath. x: H. jfe'(M^): Tl. k' (no. 82) 8. Ath. k': H. fe' (no. 37) 9. Ath. 3/,(^): H.^ (nos. 17, 87) 10. Ath. ife/pH. g/PTl. g/ (nos. 33, 34. 35, 36) 11. Ath. ': H. g': Tl. g' (nos. 96, 97, 98) 12. Ath. 7: H. g: Tl. g (nos. 18, 21, 22, 23) 13. Ath. y: H. g: Tl. g, 7 (no. 93) Continuants 14. Ath. m: H. w: Tl. w (nos. 45, 46, 47) 15. Ath. n, -7)-. H. n, -77: Tl. n (nos. 3, 8, 14, 15, 32, 34, 35, 46, 48, 49, 50, 51 52, 53- 54. 55- 56, 58, 63, 78, 85, 88, 98) 16. Ath. /, I: H. /, I: Tl. I (nos. 4, 7, 10, 11, 22, 23, 38, 39, 59) 17. Ath. s: H. s, dj: Tl. 5 (nos. 9, 57, 58, 59) 18. Ath. c (j): H. 5, dj: Tl. c (nos. 3, 4, 47) 19. Ath. x{> c in most dialects): Tl. c (nos. 85, 86) 20. Ath. y: H. g: Tl. y (nos. 88, 89, 95) 21. Ath. y (before front vowel): H. g (g): Tl. 7 (nos. 90, 91, 92, 93, 94) 22. Ath. x: H. X, x: Tl. x (nos. 20, 81, 83, 84) 23. Ath. h, x: H. h: Tl. h (nos. 25, 26, 27) 24. Ath. 7: H. g: Tl. 7, -x (nos. 19, 20, 24, 56) Affticatives 25. Ath. I: H. /^: Tl. // (nos. 40, 41, 42, 43) 26. Ath. dj: H. dj, tc' (nos. 13, 14) 27. Ath. /c': H. tc\ dj: Tl. tc' (nos. 77, 78) 28. Ath. tcl: H. /c/: Tl. s! (nos. 79, 80) 29. Ath. ts!: Tl. /i/, 5/ (nos. 75, 76) 30. Ath. tsl: H. til: Tl. tl! (e. g., Ath. *-/5.'e "penis": Tl. tllel dit.; Ath. *-tsIi "to sit [plur. sub].]": H. tl!a-{o-) dit.) Consonant Clusters. — The study of Na-dene sibilants and sibilant affricatives is rather involved and presents several difficulties. The summary given above (17, 18, 19, 26, 27, 28, 29, 30) exhibits some of the main developments. An important group of cases is afforded by Haida stems or elements beginning with consonant clusters whose first element is a lateral (I, tl!, dl) or a sibilant (5). Swanton, in Six: Athabaskan and Na-Dene Languages 127 556 commenting on these clusters, surmised that they were perhaps due to the prefixing of an old morphological element (e. g., 1-, s-). There is, however, no evidence whatever to support this. On the other hand, I have at my disposal upwards of twenty such examples which point clearly to the inference that these Haida clusters were found in Na-dene and correspond to lateral and sibilant affricatives in Athabaskan and Tlingit. The following relations can be estab- lished : A. 31. H.lg-: Ath. ///- (?) 32. H.lq'-: Ath. tU- 33. H. dig-: Ath. tU- (no. 44) 34. H. It'-: Ath. /c'- 35. H. H.'x-: Ath. tcl- (?) 36. H. tUd-: Ath. fsl- (cf. 30.) B. 37. H. sg-: Ath. Is- (?) 38. H. sq!-: Ath. ts!-: Tl. si- (no. 74) 39. H. sg-, (sk'w-): Ath. tc'-: Tl. tc'- 40. H. sqlw-: Ath. tcl- 41. H.St'-: Ath. ts- its!-): Tl. t'- 42. H. St'-: Ath. tc'- 43. H. St!-: Ath. id- Vowels. — The great majority of vowel correspondences is per- fectly intelligible; a certain number of unsolved problems still remain. In comparing Tlingit with Haida and Athabaskan. forms, it is necessary to bear in mind that, under as yet undetermined circumstances, Tlingit a has developed to e (e. g., Tl. xa "to eat": a-xe'-x "he ate"; Tl. de'x "two": daxa-ducu "two plus five, seven"; Tl. /'a "stone": /V-5.' "stones"). A more important prob- lem is presented by Ath. e (doubtless open in quality), which is not to be directly compared with Tl. e. It is clear, both from internal Athabaskan evidence (e. g., Ath. *-t^e and *-i^a "to look for"; Ath. *t'e'- "in the water," *t'a- "water") and, still more, from comparison with Haida and Tlingit, that Ath. e (which must be assumed for the earliest Athabaskan period) has developed from Na-dene a; less frequently Ath. e goes back to Na-dene i. Under what phonetic circumstances, however, Na-dene a has remained as such in Athabaskan or become e is not clear for the present. This I believe to be one of the most important problems of Na-dene phonology. 128 VI American Indian Languages 2 557 Some of the more important vocalic correspondences are: 44. Ath. a: H. a, a: Tl. a, a', a, {e) (nos. i, 2, 3, 4, 6, 7, 8, 9, 13, 14, 18, 19, 25. 28, 29, 30, 33, 35, 40, 41, 42, 44, 46, 49, 60, 61, 62, 67, 68, 69, 73, 74, 79, 81, 82, 83, 84, 85, 87, 88. 96, 97) 45. Ath. e (sometimes reduced to e^)'- H. a, a, «: Tl. a, a, a, (g) (nos. 5, 10, 20, 21, 22, 26, 31, 34, 45, 47, 51, 53. 64, 65, 66, 70, 75, 89, 92) 46. Ath. i: H. i, t, i, ei: Tl. t", , e e (nos. 12, 17, 54, 59, 77, 86, 98) 47. Ath. 0: H. M, 0': Tl. m, «• (nos. 16, 24, 27, 37, 43, 55, 56, 71, 72, 78, 80. 95) Whether or not Na-dene possessed pitch accent must remain un- decided for the present. Its presence in Tlingit and a few remarks by Morice and Legoff as to its possible existence in Athabaskan make this not improbable. Should this prove to be the case, some of the phonological difficulties in Athabaskan and Tlingit vocalism may be solved (e. g., Ath. e < Na-dene d, Ath. a < Na-dene a). All this, however, is quite vague as yet. IV. Conclusion The main conclusion to be derived from the selected morpho- logical, lexical, and phonological evidence that we have passed in review is, I believe, obvious. Athabaskan, Haida, and Tlingit must be considered genetically related. The correspondences are of so intimate a character that mutual borrowing of words and morphological features seems out of the question. It is, however, no less obvious that each of these languages is very distinctive and represents a highly differentiated form of the Na-dene prototype. In no sense can Haida, Tlingit, and Athabaskan be said to form a continuum comparable to that of the Athabaskan dialects when these are compared among themselves. Each Na-dene language has evidently passed through a very long period of development in linguistic isolation from its sister languages. It would be rash, in the present state of our knowledge, to dogmatize on the relative conservatism of the Na-dene languages. I would venture to sug- gest, however, that Haida has remained the most faithful to the original sound system of Na-dene, but that, on the whole, the original morphological features are best preserved in Tlingit. ' This weak vowel is differently colored in different dialects; e. g., Hupa i (less frequently a). Carrier a, Navaho ». Six: Athabaskan and Na-Dene Languages 129 558 Several facts suggest that Tlingit and Athabaskan may have had a common Hnguistic history after Haida had become differentiated, Imt too much should not be made of this. The name that I have chosen for the stock, Na-dene, may be justified by reference to no. 51 of the comparative vocabulary. "Dene," in various dialectic forms, is a wide-spread Athabaskan term for "person, people"; the element *-ne (*-w, *-ri) which forms part of it is an old stem for "person, people" which, as suffix or prefix, is frequently used in Athabaskan in that sense. It is cognate with H. na "to dwell; house" and Tl. na "people." The compound term "Na-dene" thus designates by means of native stems the speakers of the three languages concerned, besides con- tinuing the use of the old term Dene for the Athabaskan branch of the stock. An important ethnological consequence of our linguistic results is that a demonstration is at last given of the northern provenience of the Athabaskan-speaking peoples. So long as Athabaskan was counted a separate linguistic stock, there was no conclusive k priori reason for considering its Pacific and Southern branches as having spread out from the northern group. Under the present circumstances a southern drift of Athabaskan-speaking tribes can- not seriously be doubted. The center of gravity of the Na-dene languages is clearly in the northwest, in southern Alaska and adjacent parts of northern British Columbia and southern Yukon Territory. Owing to the great linguistic gulf separating Haida and Tlingit, I should be inclined to consider the coast of southern Alaska, the present home of the Tlingit Indians, as the most likely region in which the Na-dene languages developed. The Athabaskan branch of the stock undoubtedly formed a relatively undifferentiated unit long after Haida and Tlingit had become differentiated from each other. The Athabaskan dialects have so many distinctive traits in common that it is perfectly evident they have had a long history in common. They may be considered a specialized interior offshoot, just as Haida is a specialized island offshoot. 130 Vf American Indian Languages 2 Editorial Note Originally published in American Anthropologist 17, 534-558 (1915). Reprinted by permission of the American Anthropological Association. Nineteenth-century Russian linguists speculated about a possible rela- tionship among Athabaskan, Tlingit, and Eyak, an isolate of the Copper River area (Krauss 1964; Pinnow 1976). Sapir was almost certainly unaware of this work, and saw data on Eyak only quite late in his career, after Birket-Smith and de Laguna's ethnographic field work in 1933. The germ of Sapir's Na-Dene hypothesis, as he himself notes (p. 534), lay in the speculations of Boas and Swanton (especially Swanton 1908: 472-485) about a historical connection between Haida and Tlingit. It was the position of Haida, rather than that of Athabaskan or Tlingit, that was central to Sapir's view of the relationship. Haida, with its relatively uncomplex morphology, seems an improbable con- gener for languages so thoroughly "polysynthetic" in spirit as Athabaskan and Tlingit, but in Sapir's view the convoluted morphosyntax of Athabaskan- Tlingit was a late development, and the more open structure of Haida words represented either an archaic situation or a quite different development from the "isolating" Na-Dene proto-language. Most of Sapir's lexical comparisons with Haida depended on the assumption of morphosyntactic changes of this magnitude. Sapir's Na-Dene proposal was not well received. Both Boas (1920) and God- dard (1920) disparaged Sapir's use of a genetic hypothesis to explain resemblances — particularly morphological similarities — that they felt could better be explained as borrowings. The issue of Athabaskan-Tlingit-Haida "morphological borrowing" remained alive until at least the 1950s, when Dell Hymes in an important series of papers proposed a technique of "positional analysis" to assess the historical value of structural resemblances of this sort (Hymes 1955, 1956). Research in recent decades has clearly established the genetic relationship of Athabaskan and Eyak (Krauss 1964, 1965). A genetic relationship between Athabaskan-Eyak and Tlingit is accepted by most scholars, but is seen as dis- tant and problematic (Krauss 1968, 1969; Krauss and Leer 1981; Pinnow 1966). In the recent literature, it is this grouping that is usually referred to as "Na- Dene." Krauss has called Sapir's belief that Haida belongs in Na-Dene "an illusion" based on "mistranscription, misanalysis, mistranslation, and/or mis- interpretation" (1979: 841). Levine, in an influential paper (1979), has given a detailed critique of the "spurious" evidence on which Sapir based his hypoth- esis, tracing many of the supposed errors to Swanton's deeply flawed descrip- tion of Haida (1911). In his recent general treatment of linguistic relationships in the Americas, Greenberg (1987: 321-330) rebuts Levine and wholeheartedly accepts Sapir's more inclusive version of Na-Dcne , which he sees as an intrusive group unrelated to any other American Indian linguistic stock. Pinnow also regards the relationship of Haida to Athabaskan-Eyak-Tlingit as likely to be Six: Athabaskan and Na-Dene Languages 131 genetic, and has published considerable lexical evidence for the relationship (Pinnow 1985a, b,c, 1986a, b, 1988, 1990), as well as a comprehensive history (in German) of Na-Dene research from 1798 through 1976 (Pinnow 1976). Sapir's ms. corrigenda on his own offprint are as follows: Original For: Read: p. 545, 1. 32 -ga -ga p. 546, 1. 6 ga-laer) §a-laer) p. 553, entry 63 t'i!n t'in p. 553, entry 79 sja-x" s!a-x" p. 553, entry 80 tcju tc!u ga ga p. 553, entry 87 On p. 535, lines 1-3, the text from "the consonant ..." to "Tlingit (Tl.)." is bracketed with a question mark by Sapir. On p. 541, footnote 3, Sapir indicated the last sentence of the note with the word "revise!" On p. 547, last paragraph, Sapir questions the statement "Chipewyan-/7fl. . . *man)/' On p. 549, end of first paragraph, Sapir questions the sentence "Such an example . . . 'ring')." In the Comparative Vocabulary, Sapir questions the Tlingit forms in entries 22 and 33 and the Haida forms in 60, 74, 89, and 95; adds Tlingit na 'to drink' to entry 50; and recalculates the total entries from 98 to 95. On pp. 555-556, Sapir questions nos. 30, 33, and 38 in the list and states that "This paragraph [headed "Consonant Clusters"] should be eliminated. Swanton is right." On p. 557, first text paragraph, Sapir questions the sentence "Should this prove . . . Na-dene ^/)." On p. 558, Sapir notes, beside the first three lines, "More certain now." The Sino-Dene Hypothesis [excerpt from a letter to A. L. Kroeber] I have long wanted to write you about Nadene and Indo-Chinese, but my evidence accumulates so fast that it is hard to sit down and give an idea. Let me say this for the present. If the morphological and lexical accord which I find on every hand between Nadene and Indo-Chinese is "accidental," then every anal- ogy on God's earth is an accident. It is all so powerfully cumulative and inte- grated that when you tumble to one point a lot of others fall into line. 1 am now so thoroughly accustomed to the idea that it no longer startles me. For a while I resisted the notion. Now I can no longer do so. The chief stumbling-blocks in the way of a general acceptance of the syn- thesis would be: 1. Failure to realize the very exceptional type of language to which Nadene belongs. It is really quite alone in America, so far as I can see. The contrast between it and Eskimo, Wakashan, and Algonkin is tremendous. 2. Failure to realize that the Nadene languages are not one-third as synthetic as they look. Go at analysis half-way decently and get into a critical perusal of connected text and you soon realize that the complex verb breaks down into a cluster of very live elements, each of which has a syntactic or positional value, not merely as "affix" but as radical element. Haida in particular, I find, is extremely analytic. It knows no "prefixes" and "suffixes" at all except for certain important survivals that Swanton does not even mention (e.g. causative and denominative s- in s-kit to 'handle a club, to club' < kit 'club', cf. Tlingit sl-, transitivizing prefix; also Haida /- = Tlingit it- = Athabaskan "3d modal" /-, also primarily transitivizing). What Swanton calls affixes are all independent stems entering into composition, or even little verbs. There is no "tense suffix" in Haida, merely a series of enclitic phrases consisting of demonstrative + particle verb of being. His "continuative" -gAn, e.g., is simply 'that-is (duratively)', his "imminent future" -asan is really -'a-sa-i} 'this-will-be (durative)', and so on. It all crumbles to pieces at the least touch. I think the same will prove true of Athabaskan-Tlingit, though here the integration is more thorough. But I no longer seriously believe we have the right to consider anything preceding the "second modal" elements (Ath. 7-, n-, and v-) as part of the verb and am beginning to doubt if even these elements and the subjective pronominal "prefixes" are part of the true verb. I think it more than likely that such an Ath. form as *yasectk'os (purely theoretic form ad hoc\ Not to be mis- taken for genuine Ath.) 'I picked up a flexible object' is to be analyzed as *ya se c l-k'os 'up it-is (that) I handlc-a-flexible-object'. *l-k'os is the verb; the rest is a series of somewhat reduced independent elements that follow in a definite order. 3. The third prejudice to overcome is the nature of Indo-Chinese itself. 134 VI American Indian Languages 2 Modern Chinese is a very secondary development. The most typical represent- ative of the earlier stage is Tibetan — which is startlingly Nadene-like. It has those fundamentally important "3d modal" elements of Ath., Tlingit, and Haida (e.g. du- 'to be together': s-du- 'to cause to be together, to assemble'; in fact, 4 of its more important verb prefixes, which are "voice" elements, seem to me to correspond in form and meaning to Nadene elements — s- to Tlingit- Haida s-; r- to Tlingit-Haida-Ath. /-; d- (medio-passive) to Tlingit-Ath. d-, survivals also in Haida; nasal prefix to Ath. ''3d modal" -n-, -r)-, of mysterious value but probable active intransitive). Moreover, Tibetan has vocalic ablaut in its verbs (e.g. X]-gex]-s 'to fill', perfect b-kax], fut. d-gax}, imperative /c'oij). Again, the transitive verb is really passive, as in Tlingit. In both, for instance, you would say 'Man-by horse kill' = 'The man killed the horse', Tlingit agen- tive -tc corresponding exactly to Tibetan instrumental -s. In both Indo-Chinese and Nadene, postpositions are of extreme importance and serve to subordinate preceding verbs and clauses. Indeed, reading Tibetan text gives you precisely the same feeling as reading Haida text. I wish I had time to illustrate. In both groups the fundamental element is really a noun, the verb a kind of denominative structure. In brief, I should say that the similarity in feeling between Tibetan and Nadene is at least as close as between Latin and English, probably closer. Thus the theoretical road to a synthesis is clear. And the lexical evidence is startling. You would be amazed at some of my material. Things like: 1. Tlingit k'a 'surface'; Navaho k'd 'surface': Tibetan k'a 'surface' 2. Chinese fan 'charcoal': Haida s-fan 'charcoal' 3. Old Chinese ti 'this': Ath. di 'this' (Ath. di really means ti) 4. Old Chinese ti 'pheasant': Ath. di 'partridge' 5. Nadene k'u 'hole' (TI. k'u-q'" 'hole', t'a- tu-k'^ 'cave' = 'rock-interior- hole', yoL-k'o 'to fall into a hole'; Nav. k'o, e.g. ts'e-k'o 'rock-hole' = 'canyon'): Indo-Chinese k'u 'hole' (dozens of forms, e.g. Tib. /:'w-r) 'hole', Karen k'u, with falling tone, 'to dig a hole'). These are only a drop in the bucket. Naturally it is a big problem and there are going to be hundreds of knotty points to unravel. But I do not despair. My present plan is to proceed as follows. First, to prepare part 1 of a Nadene com- parative study, to consist of my present lexical material (about 300 comparable radical elements, to which I add constantly). In this I would give reconstructed Athabaskan but also actual Ath. dialectic forms. Before publishing parts 2 and 3, on morphology and phonology, which need much preliminary work, I intend to publish special papers on selected portions of Nadene grammar, e.g., certain archaic post-positions; or demonstrative stems; or general points of syntax. In this way I shall be keeping the problem live and accumulating experience for the definitive Nadene study. Of course I shall have to do Ath., Haida, and Tlingit in the field. I want particularly to gather a large amount of purely lexical material. People do not realize how scanty is our material, and for my purpose, which is comparative, I need stacks of it. What Goddard gives us is a miserable pittance — and wretchedly analyzed or not analyzed. Did you read his attack on my Nadene? You can have no idea of the laughable errors he commits. It is the Six: Athabaskan and Na-Dene Languages 135 work of an utter groundling that does not know his own material. What do you think of a man who expects you to unravel the complex phonology of Nadene by drawing up an alphabetical list of Tlingit ''stems" and "matching" them with random "stems" from Kato or Ten'a ad libitum'? Very much as though you "matched" Sanskrit words now with French, now with Portuguese. Great method, what? And what do you think of a man who rules out comparisons because he does not "know" the Ath. form in question? Particularly when the form occurs in a book (Chapman's Ten'a) which he has "edited"! I may reply to Goddard, but it is really no use. He is a man of no more than average linguistic ability, completely at the mercy of his local sentimental memories, and abso- lutely without vision as to the older drift of Ath. He probably imagines his lists of stems are the last word on the whole subject. The degree to which he has failed to analyze his material is shocking in the extreme. I shall not broach the Indo-Chinese part of the problem till I have moored myself more completely in Nadene. The final plan is: 1. a Nadene comparative grammar to be published in 3 parts (possibly an Ath. etymological dictionary as a side-show); 2. a Nadene-Indo-Chinese demonstration; 3. a more general treatment of the evolution of the whole group, showing how old types have been replaced by new ones. As a starter, I am at work now on a paper on Haida phonetics, which may interest you when you see it. P.S. I cannot resist the temptation to give a somewhat livelier idea of the remarkable way in which lexical elements are interwoven in Nadene and Indo- Chinese. I have some cards along, so don't need to trust to memory. I shall give an idea of the richness of some of my entries by dealing with a group of related words. In Ath. we have a stem *fu, post-vocalic *-Iu, which may be rendered as 'coiP or loop'; e.g. Nav. to loop', as vb.: Nav. -lo 'to catch with a rope', Jic. Apache -lo' 'to lasso', Chipewyan -lu, -tu 'to be caught in a net or noose'. So far, so good. Here our friend Dr. P. E. Goddard would end. But it is difficult to believe that Ath. */76>/'rope, strap' (found in all dialects) is unconnected. How? With- out going into details (it would take too long), I may say that I feel justified in analyzing *t'lo-l into *t-'lo--i. How *-'lo-, -'lu is related to *{u, -lu I cannot yet tell, but I strongly suspect Nadene had both / and '/, and in related stems. As you will see from my Haida paper, Haida has both / and '/; in Tlingit '/ probably became /'. I should guess that 7 is causatively related to /,- Ath . *///, -lu is intr. : 'loop; to lie coiled', *- '/w would be 'to cause to be coiled, to make a loop, to tie around' (possibly -lo' is a secondary form of *-'lu). Now -/ we know to be con- tinuative or usitative; and /- is medio-passive. Hence *t-'lo-l is what is always caused to loop around, what loops about something', in other words, rope, strap'. We learn important things from such an analysis: that "3d modal" ele- ments were welded with verb stems and appear in nominal derivatives; that there was an old alternation /.• '/ whose significance remains to be discovered. That we are on the right track is confirmed by another common Ath. stem 136 ^^ American Indian Languages 2 whose formation is precisely parallel to that of *t'to-i. This is *t'iei 'fire-drill'. Fortunately we are here not dependent on Goddard's material alone. From Petitot we learn that in Hare and Loucheux there is a verb *-/e (-d-le, -t-le, -l-le) meaning 'to revolve' (words involving it are: 'virer au cabestan', 'cylindrique', 'tourbillon', 'tourner', 'se tourner'). Hence *r-7e-/ is 'what keeps turning itself, what revolves drill-like'. This parallelism of *r7o-/ 'rope' to */7e/ 'fire-drill' is, of course, highly suggestive. It shows that many of Dr. Goddard's "stems" may not be pure father-Adam radicals. And we see that Ath. fl fails to correspond to Haida and Tlingit fl for a reason. It is a secondary development in probably all 3 groups. Such a sound as /' appears in cognate words throughout; not so fl — which fact alone casts a reflex light on our analysis. Let us proceed. To Ath. *hi, -lu is clearly related Ath. *-lui 'to wrap around': Hupa -loi 'to tie, to wrap around', Kato -//' 'to tie up' (old Ath. form possibly causative *-'lui). And further, having once allowed Ath. t'{ to analyze itself into medio-passive t- (d-) + '/, we do not feel we are doing anything ungodly to analyze Ath. *fVw' 'grass' into *t-'lu' 'what is wound (in basketry)'. This analysis of 'grass' is helped by denominative Ath. verbs, e.g. Hupa t'h-n, -t'lo, -t'lo-W, -t'lo-i 'to make baskets, to twine in basket-making'; Nav. -t'lo, -t'lg-, -t'lo-l 'to tie (e.g. the hair)'. We may summarize all this as follows: Ath. *lu, -lu 'coil, loop' — *lu 'to be caught in a noose' — *'lu 'to catch in a noose' — *'lui 'to wrap around' r-7w' 'what is twined' > 'grass' (denominative vb.: 'to twine in basketry; to braid hair') *t-'lo--l 'what is always looped' > 'rope, strap' Now comes the fun. Indo-Chinese lu as follows: Tibetan lu-i] 'a strap, slung over the shoulder or round the waist, for carrying things' Miao /o-r) 'bridle' (close o) Tib. metaphorically: caus. s-lu 'to cause to be snared', i.e. 'to ensnare, beguile, seduce' Angami Naga te-rhu 'sly' {rhu < h-ru; Tibeto-Burman hi-, hr- > Ih-, rh- is exceedingly common; h- is common as causative prefix, e.g. Tib. s-l- often par- allel to Ih-) T'ai group: Siamese roi^^^ 'enfiler' (numbers indicate tones in H. Maspero's orthography); White Tai roi, loi 'enfiler' And now Chinese: ///- group: lui]^-^ (numbers for tones according to usual Modern Pekinese system) 'a cage, to snare' < Old Chinese (i.e. 7th Cent, forms, which I have worked out carefully from Karlgren's tables) lu[o]x]- (- = level; / = rising; \ = falling tone). luv]<-^ (in hi^^^> t'ou(-> 'halter', i.e. 'snare- head') < Old Chinese /w/o/r)-;/w^-^^ 'girdle gem' 'hempen thread' < luo-; lo^^^ 'conch, Six: Athabaskan and Na-Dene Languages 137 spiral, screw' < lud- {a is a dark-timbred a- vowel); lo^-^ 'lines in the palm' < lud ?lei'^^ 'to creep, cling to', same character also read lei^^^ 'series, connected' (words evidently refer primarily to creeping vines) < ludi-, ludil. These Chi- nese lu, lud forms are paralleled by another set in lui- (cf. Ath. *-lui above): lei^^^ 'to bind' < lywie- (< *lwie-);lei''''> 'acreeper (asof melonor pea)' (not sure of reconstruction, but phonetic element in character suggests initial lu or lywi); lii^^^^ 'silken thread, a hank' < lyhi/. And metaphorically we have luj}^'^^ 'foolish, to impose on' < lu[o]r\\; lii^'^^ 'deceitful' < />'h'/o\ (i.e. 'winding about, ensnar- ing with blandishments'; cf. Tibetan s-lu above). Other probably connected Tibetan words are: Ihub (i.e. causative h-lub) 'to bind, tie, fasten (e.g. ornaments to the ear)'; k-lub 'to cover (e.g. the body with ornaments)'; Iwa-ba 'a woolen blanket'. Observe how well the Ath. forms integrate with the numerous Tib. and Chi- nese forms. But we are not done. Very likely connected with Ath. *-lu is Ath. *-lu-s 'to drag an animal by a rope': Nav. -Ids, -lo-z, -Ids classifier verb "denot- ing a single animal as an object: the inference is that the animal is led by a rope" (Franciscan Fathers); Hupa -los 'to drag, to pull along'. With these forms I feel inclined to compare (though here I feel far more hesitant) Chinese lo^^> tsz<^^ 'mule' {tsz<-^^ is merely 'son', often used to make nouns) < lud-; also lu 'donkey' < lywio-. The parallelism between Ath. and Chinese would be a convergence from related radicals rather than a specific etymological parallel. But we are far from finished. Perhaps related to Ath. *-lu is an important classifier verb *-/e, *-la\ Nav. -/e, -la, -let 'to handle a long, flexible object, as a rope, quirt, leather, hide, etc' (Franciscan Fathers). Parallel to this is a set of Indo-Chinese forms in *la, *le (Ath. e is often parallel to a; just how related I do not yet understand, possibly reflex of old alternation a: [a): Tib. causative s-le, Ihe (< h-le) 'to twist, plait, braid the hair, to make a basket, to knit'; s-le 'a coarse basket'; s-le-po, s-le-ba, s-le-bo 'a flat basket' {-po, -ba, -bo are "articles"); Iha-s, Ihe-s < h-la-s, h-le-s 'braid, wicker-work, texture; twisted cake or bun'; Ihe-s-ma < h-le-s- 'the act of twisting, plaiting' (-ma is "article"); /a«-/?w 'braid, plait, tress of hair' (-/?« is diminutive); lan-ts'ar 'ornaments worn in the hair'; le-brgan 'diapered design of woven fabrics'; le-na 'the soft downy wool of goats below the long hair; fine woolen-cloth'; Ida-ldi 'a kind of ornament of silk or cotton, a fringe or tassel' < d-la- {dl- regularly > Id- in Tib.; very easy to illustrate); Idan-mgo 'the yarn-beam of a loom' < d-lah- {mgo 'head'); Idem-Idem < d-lem 'flexible, supple, elastic, pliant'; Ideb- < d- leb- 'to bend round or back, to turn round, to double down'. 1 am not so certain of this last, which brings us into a large set of forms in la- and lo- referring to 'turning, turning back', which may well be related to our present set but which I prefer, for brevity's sake, not to go into just now. Now Chinese. We have two series: *la and *//. Based on *la are: lan'-> 'basket with handle' < Idm-; lan^^^^ 'rope, hawser' < Idml; lao'-^> (tezi) 'netted case' < Idk, same character also read h^-^^ 'joined, to tic up'; lei^'K also read h<'---'^> 'to rein in, to strangle' < hk (perhapsbcttcr to /fv- scries above); /o'^'^ 'net, sieve' < ld-\ lo<^) 'shallow open basket' < Id-. Based on *// are: li<~> 'ornamented girdle' 138 ^I American Indian Languages 2 < bie- (?); W^^ 'basket' < />'/-; //^- "^^ 'rope to tie a boat', character also readi/'^-^'' 'a well woven gauze' < xyil (?), which probably means older *h-li (loss of post- consonantal / is now well established for Chinese). Is it not impressive that Ath. and Indo-Chinese *lu seem to have reference chiefly to "looping," Ath. */e, la and Indo-Chinese *la, li to "handling a long flexible object, twining, basketry"? Now let us return to Nadene. In Tlingit we have //r' 'fine basket'. Remember that Tlingit has no voiced /, only voiceless /. Further, experience shows abundantly that many Tlingit forms in final conso- nant (particularly if glottalized) are compounds. Hence we may suspect that lit' really means old *li-f(a). Reference to Ath. gives us *-t'a 'receptacle': Hupa -fa 'sack'; Kato -fa 'pocket, blanket fold'; Nav. -dzis-fa 'pocket' {-dzis 'hollow, semi-tubular' ?). This somewhat unsatisfactory parallel is buttressed by Haida fao-fa 'box' (almost certainly misheard for tao-fa 'food-receptacle'). Tlingit ///' is therefore probably li-fa 'receptacle (for small objects) of twined basketry'. We have further /'/ 'woolen blanket' < *// or *//' (final glottal stops seem regu- larly to affect the first consonant; I have good evidence for this). Based on *la or *lia we have also in Tlingit fleq' 'tentacles of squid' *t-'le-q' (with medio-pas- sive r-; Tlingit e is always related to a- forms) < 'what is long and twists itself? (cf. fundamental meaning of Ath. *-/e, *-la above). And in Haida we have dli-n 'arm of devil-fish', which I analyze as d-li--n {d- medio-passive; -n perhaps causative, to which there are good Ath. and Tibetan analogies; -//•- < Im- ?) < 'long things that twist themselves about'? I am quite likely to have misinterpreted here and there, but the Nadene and Indo-Chinese parallels seem highly suggestive to me. Of course, this sort of group-parallelism is not isolated. I shall refer to another interesting group, without going into details. I spoke of Ath. *-/e 'to revolve' and its derivative *flel 'fire-drill'. To these forms belong Haida hi 'to surround, move around'. In Indo-Chinese we have a well-cham- bered Tibetan set: re and ril (e.g. res 'change, turn, time, times'; i}-g-re 'to roll one's self ; causative s-g-re 'to roll' ; ril 'round' ; causative s-g-ril 'to wind or wrap round; to roll, wrap, or wind up; to wag (the tail); to roll (a stone)'. Here may also belong Haka kut-hrer 'ring' {kut- 'hand'; h- common Tibeto-Burman causative prefix), though Conradi sets this to Tib. forms in la-, le- referring to "twining." Editorial Note Originally published as part of letter 332 (October 1, 1921) in Victor Golla (ed.). The Sapir-Kroeber Correspondence. (Survey of California and Other Indian Languages, Report 6.) Berkeley: Department of Linguistics, University of Cahfornia (1984). Six: Athabaskan and Na-Dene Languages 139 Sapir's manuscript Sino-Dene "dictionary," including more than a hundred lexical comparisons, is in the Library of the American Philosophical Society (manuscript 497.3 B63c Na20a.3, vol. 2; cf. Kendall 1982: 28). Nearly all the entries date from the early 1920s, and it is clear that Sapir went little further with the idea than the speculations contained in this letter. Of these specula- tions (see also 1925o, later in this volume) Krauss has written: "Sapir was . . . carried far beyond any objectively justifiable conclusions by his enthusiasm for the idea" (Krauss 1973: 963). Few would challenge the accuracy of this assess- ment. The connection is, however, a plausible one, both on linguistic and anthropological grounds, and it continues to attract attention. Robert Shafer, a Sino-Tibetanist, published two papers in support of the Sino-Dene hypothesis (Shafer 1952, 1957), but, being based neither on Sapir s Sino-Dene files nor on extensive comparative Athabaskan or Na-Dene data, his work cannot be con- sidered an authoritative statement. Greenberg, Turner and Zegura (1986) have proposed that Na-Dene (including Haida) represents the language of an Asi- atic population that entered the New World about 7,000 years ago, and that it is distinct, both physically and linguistically, from all other American Indian pop- ulations. While this is consistent with Sapir's Sino-Dene hypothesis, Green- berg, in his recent world-wide linguistic classification, does not link Na-Dene with Sino-Tibetan or with any other linguistic group, considering it to be one of 15 fundamental linguistic families in the world (1987: 332-337). Pinnow (1990), relying mainly on Greenberg 's own evidence, believes that a good case can be made for viewing Na-Dene as intermediate between Old World languages (especially Sino-Tibetan) and New World languages (especially Greenbergs "Almosan-Keresiouan" phylum). Athabaskan Tone A part of the summer of 1922 was spent by the writer at Sarcee Reserve, Alberta, in studying the language of the Sarcee Indians. A series of texts was obtained as well as supplementary grammatical material. The most important single point that appeared was the fact that Sarcee has a well-developed system of pitch accent. Fundamentally this system has a striking resemblance to the Tlingit tonal system described by Dr. Boas, though secondary developments have complicated the Sarcee system considerably. The tonal resemblances between Tlingit and Athabaskan constitute an important further argument in the Nadene theory recently put forward. Hints on Athabaskan tone are to be found also in Father L. Legoffs study of Chipewyan (Grammaire de la Langue Montagnaise). Father Morice makes a few isolated references to tone in Car- rier, where it is almost certainly a feature of importance judging from brief MS linguistic notes taken by C. M. Barbeau among Carrier Indians at Hagwelgate. Some years ago P. E. Goddard noted tonal differences between otherwise iden- tical second person singular and third person subjective forms in the Hupa verb; these observations, based on the study of [391] Rousselotgraphs, agree with the Sarcee results obtained. So fundamental is tone to Sarcee morphology that it is well nigh inconceivable that it should be entirely absent in any other Athabaskan dialect. Editorial Note Originally published in American Anthropologist 24, 390-391 (1922). Reprinted by permission of the American Anthropological Association. Sapir's references to earlier work are to Boas (1917) for Tlingit, to Legoff (1889) for Chipewyan, and to Goddard (1907) for Hupa "Rousselotgraphs." Morice later commented on Carrier tone more extensively in his full treatment of the language (1932). At the time of Sapir s Sarcee work, tone systems had been described for many African and Asian languages but only rarely noted among American Indian languages (Sapir's own description of a pitch accent in Takelma being one of these instances; see 1912h, Volume VIII). We now know tone to be a wide- spread feature in the Americas, particularly in eastern North America and in lowland South America. Ironically, comparative evidence in Athabaskan (and Athabaskan-Eyak) now indicates that tone was not present in the proto-lan- guage but developed in several Athabaskan subgroups (but by no means in all) as syllable-final consonants were simplified or lost (Krauss and Golla 1981: 142 Vf American Indian Languages 2 69-70). Sapir, in later fieldwork, encountered at least two Athabaskan lan- guages without tonal systems, Anvik and Hupa. In the latter case, where Sapir had gone to the field with every expectation of confirming Goddard s earlier reports, he was clearly nonplused (see Sapir 1928i, reprinted in this volume, and Sapir s letter to Kroeber, 28 June 1927, quoted in Krauss 1986: 163). A TYPE OF ATHABASKAN RELATIVE As is well known by students of Athabaskan linguistics, the Athabaskan adjective is in form a verb. Even the simplest, non-pronominal or third-personal, form regularly contains either a " first modal " prefix - (generally de- ^, d; ne-, n-; t'e-, t'- ; or ie-, i-), a " second modal " prefix {se-, s-), a " third modal " prefix (gene- rally -/- or -i-), or a first (or second) modal prefix followed by a third modal prefix (forms in de-l-, ne-l-, nl-i- are typical). It is unneces- sary to give examples here, as they may be readily supplied from the special dialectic monographs (see, e.g., Goddard, Legoff, Mo- rice, Petitot), Father Legoff, however, in his Grammaire de la Langue Montagnaise \ calls attention to an interesting class of adjectival derivatives (" nouns formed from'adjectives ") in which the prefixed elements are lost and the bare stem appears in the relative ^ form. Some of his examples are : 1. I follow Dr. Goddard's convenient terminology, without thereby committing myself in the least as to the term ""modal ". 2. I use e as a formula for the reduced or " pepet " vowel which has different forms according to dialect or according to varying phonetic circumstances in one dialect (e.g. i, less frequently a, e, a. or u, in Hupa ; a in Kato and Chasta Costa ; e, a in Chipewyan). It may often be shown to be a reduced form of an older Atha- baskan e or /. 5. Montreal, 1889. See p. jo. Petitot and Legoff use *' Montagnais " in the sense of Chipewyan. 4. By " relative '" I mean the form assumed by nouns when they are qualified by preceding elements (nominal or pronominal) and by verbs when they are used as rela- tive clauses. The fundamental Athabaskan relative suffix is probably -e, -e (-ye, -y^), but the actual dialectic forms are often involved by the operation of various di-l-ba (to be) gray : bay-z the gray one dz-l-gai (to be) white : gay-z the white one m(d)-du-i (to be) short : du-z the short one dz-yzl (to be) squat : yzl -s the squat one dz-bal (to be) round : bao the round one dz-l-zzn (to be) black : :^zn the black one dz-yo' (to be) shaggy : yo' the shaggy one dz-l-Bo' (to be) yellow : Oo' the yellow one. Such forms as bg^, ;(en, yo' , and 60' look for all the world like unmodified stems, but there is every reason to believe that they are relative forms, like gay-z and yel-z, that have either fallen together with or that differ in some res- pect from the parallel stem forms found in the adjective-verbs. Possibly the phonetic record is defective ^. Legoff says of these forms : " This kind of phonetic laws. Frequently the e, e is dropped but is then apt to leave a trace in the voicing of the prece- ding consonant, now final (e.g. the relative form of Hupa -fa AMONG < Ath. *-t'ax is -t' au < -t' av.\ the regular Hupa development of Ath. *-t'a-{, reduced from *-t'a-f-e). 5. I write ^a^-e rather thad ba-y^, though Legoff speaks of a suffixed -ye, because comparative evidence demons- trates the existence of the stem-form -bay- a.s well as -ba' . -ba'-i and Petitot's Montagnais -ba-a seem to presuppose Ath. *-mah-e. Navaho -bat, e.g. gim i-bai gray hawk, is Ath. *-may-i; d. further Hupa -nmi, Kato -bai. I am modifying Legoff's and Petitot's orthography and that of the Franciscan Fathers so as to bring them into har- mony with current Americanist usage. 6. Goddard writes di-l-{7.ti to be black, but tia^iiz- ^an' BLACK K0.\ (Texts and Analysis of Cold Lake Dialed, Chipewyan, apamnh, X, 1 10). Does this mean that -iin is the relative form of -inn ? The relative form of the corresponding Navaho -/t« black is -/««, with lengthened vowel. The formation of the relative in Athabaskan has never been properly studied. It is one of the fundamental problems of the language. 144 VI American Indian Languages 2 137 substantive is hardly used except as sobriquets which men give one another or as names which people give to animals, in order to distinguish them. And then they are always followed by the word ya:(^t, which means little". As a matter of fact, the type is illustrated in other connections, e g. : d--k\oTO BE BALD I t^i-k'zit THE BALD-HEADED ONE, BALD-HEADED (Pet.) (h)o-rt-iur '< to be slippery (Pet.) : t' an'-xurz SLIPPERY ice (Leg.) These two examples differ in an important respect. In t'xn'-:(urs and numerous other com- pounds of its type the first member (ice) desig- nates the properly denominating concept of the group, which is then qualified by a relative form (jur-i) of Legofi's " sobriquet " type. In t^i-k\lt the first member of the compound (head) is not the properly denominating con- cept of the whole, though it is itself qualified by a relative form (^k\l-t), again of the sobri- quet type. The group tHi-k\l-., taken as a unit, is to be understood as qualifying a third, un- derstood, noun. We can express tliis by saying that while slippery is relative to ice and ice not relative to another concept, bald is relative to head and bald-head to person. As far as such a form as tOi-k'eoi is concerned, it makes no difference whether the qualified noun is expressed or not. Obviously the difference bet- ween t'oin'-iure. and thi-k'zle is analogous to the English difference between red breast and (robin) redbreast. Whether there is a pro- sodic difference (one of stressor pitch) between the two Athabaskan types does not appear from the evidence, but it is possible that they are not formally identical. As there is no genuine line of demarcation in Athabaskan between " adjective " and " verb ", one may expect that forms both of 7. Petitot writes 0- for /;o- (see Goddard and Legon). -r- is a postvocalic form ol d- in Chipewyaii. the type bays, the gray one and t'ixn'-:(tm slippery ice may be based on " verb stems ". This is exactly what we find. A few examples are : Mont. -Y? TO MELT ; ds.-l--(j to be melted (Pet.) : dlts '(in-z grease easy to melt, meltable grease (Leg.) Mont, -ti'a^ to cry : ts'ay-e. the whimpering one, grumbler (Pet.) Hare -k'^i to lie habitually : k'^t-t liar (Pet.) It scarcely needs to be pointed out that LegofT's " sobriquets " are simply qualifying terms in the relative form, the noun referred' to being unexpressed. It is not a far cry from compounds like t^i-k'tlt the bald-headed one and dUs -^in-e. meltable grease to such clipped forms, say, as k'toe. the bald one and -(ins. WHAT melts (easily). Such forms are in type identical with LegofFs hayt the gray one. Ot ts'a-^t grumbler, Petitot remarks that it is deri- ved from ts'ay tears; we would then have to interpret ts'ayz as the one with tears rather than as the whimpering one. Petitot's analysis is perfectly credible, for forms of the type of bayz may be directly formed from noun stems, as I shall show from Navaho evidence. But I hope also to show that the difference between -ts'a-( TO CRY and ts'a-; tears is purely a matter of translation, not of intrinsic Athabaskan form. In Navaho there are a great many prefixless or radical qualifying terms in relative form. They are freely used with or without preceding nouns and are based on " adjectival ", " nomi- nal ", or " verbal " stems. They all denote permanent or characteristic attributes and easily take on the character of sobriquets or regular clan or personal names. Examples of " adjec- tival " origin are ^ : 8. My Navaho examples are taken from tiie Vocabuhry of the Navaho Language of the Franciscan Fathers (2 vols., St. Michaels, Arizona, 1912). Six: Athabaskan and Na-Dene Languages 145 138 n-jun-i NICE (relative form) ; «i-/p" to be gocd : jun-i NICE n-nt'i TO BE LONG : nts, nf^-i long n-t't'lro BE WIDE : t'd WIDE h-ts'o to BE YELLOW : h ts'o-i YELLOW EARTH (place name) ii-jin TO BE BLACK : hwo' ji'n-i tooth-black, DECAYED TOOTH J ya'' ji''n black LOUSE a-i-i'sg's-i to be slim (also -t'sgs, -t'sos-i, -t'so's) : ts'i' t'so's-i head-slim, the slim- haired ONE (man's name) hwo' c-gi'j-i TOOTH wHicH-is-MissiNG (c- assimi- lated from S-, " second modal " prefix) : hwo' gi'j-i MISSING TOOTH ; bo-hwo'' gi'j his-tooth MISSING di-tc'oc-i stubby (relative form) : ts'i' tc'o'c-i head-stubby, the stubby-haired ONE (man's name) a-gud-i SHORT ; Montagnard i-gor-z (Pet.) : ga gu'd-i ARM-SHORT, ARMLESS ; k\ gud-i SHORT- FOOTED ia-gai to be white : ga' ts'o gai' rabbit-large- WHITE, WHITE JACK-RABBIT Examples of *' verbal" origin are : -t'lo, -t'ig', -t'ioi to weave : Hog-i 9 grass- weavers, SIA INDIANS -y^'d, hwud, -hwul to run : t'o' hwul rapid WATER (place name) This type is doubtless actually well repre- sented in Navaho, but the material is scanty or not easily accessible. To it belongs probably Tucson (Arizona place-name), said to mean BAD-SMELLING WATER ; the Navaho (or Apache) form is probably something like to tc'a (cf. Nav. -tc'iUy -tc'a, tc'ji to smell). Examples in which the related word is a noun are quite numerous. The reference is not to the concrete content of the noun as such but to a person or object, expressed or implied, that IS conceived as the possessor of or as in 9. Or directly to t'io' grass? some way related to the thing defined by the noun. Examples are : k'ai WILLOW : k'a''-i those who have (or are CONNECTED WITH) WILLOWS, WILLOW CLAN yo' BEAD : yo-0 those who have beads, bead CLAN fca' HAT : t'cah-i he with the hat (man's name) gic CANE : gic-i HE WITH THE CANE (man's name) ca'c KNOT : ca'j-i knotty xvi burden, bi-yf'l his burden : ye'l-i the one WITH A burden, he WHO IS SLIGHTLY HUNCH- BACKED (man's nick-name) fcoc WORM : hwo' fco'c-i tooth -wormy, a hol- low TOOTH yii SUPERNATURAL BEING : t'o yf WATER THAT HAS SUPERNATURAL BEINGS, DANGEROUS WATER (place name) t'o WATER : na' t'o'-ho enemy that is connected WITH WATER (?), ENEMIES AT THE WATER, ISLETA INDIANS cac BEAR : na' cdc-i ennemy that is bear, bear ENEMIES, HANO INDIANS Relative forms like t'cah.-i bear an obvious formal similarity to such English derivatives in -ed and -y as bearded and knotty. But the resemblance is more apparent than real. The relative form of the noun is not a true " adjec- tival " derivative of the noun, as is shown by the fact that morphologically parallel forms are built on stems conventionally set down as " adjectival " or " verbal ". The genetic relation between these Athabaskan relative forms and the possessed form of the noun (e. g. Montagnais t'sah-i his cap [Leg.], si-t'sa-a MY CAP [Pet.]; Hupa hWm song : xo-hW\.n-[n^'. HER song ; Chasta Costa c-man-e my house) on the one hand and the subordinated form of the verb (e.g. Mont, i^-yt one grows up : {s--ye- h= GROWTH, m-^^-i-l-dyd we are afraid : ns—,"- t-l-dyed-i we who are afraid [Leg.] ; Hupa -lal TO FLOAT continuously : na-na-t-lnl-B it floating; Nav. di-c-Jnuuc i yell : d\.-la-hwuc-i 146 VI American Indian Languages 2 139 ONE WHO YELLS MUCH, howler) OH the Other is obvious. In Navaho the relative form with final vowel (generally -/ ; old -e assimilated to -cJ after radical o)is probably no longer felt as identical with the possessed nominal form with final consonant (cf. bi-ye-'l his burden with yz'l-i above), but the general consensus of Athabaskan evidence makes it higly probable that such alternations as -yi'l and yd-i go back to Athabaskan alternations of type *^z'l-e: *'(el-e. In other words, even in Navaho the possessed forms of the nouns are simply redu- ced relatives (perfectly analogous to xe'i : -yfl is d:(d MOUNTAIN : yo-l-gai' d^i'l shkll-white mountain). We have already seen that there are analogous doublets in Navaho for the rela- tive forms of " adjectival " stems (cf. gi'j-i and gi'j .missing, jm-i and ji'n black above). It is not the purpose of this paper to discuss the functions and the fundamental significance of the Athabaskan relative. I hope to show in a future paper that it is a feature that goes back, both in general form and as an actual phonetic element, to the Nadene period and that it con- sists in essence of an old particle, probably a demonstrative stem, that could be freely added to any word or group of words to relate it to an expressed or understood person or thing. The primary function of the Athabaskan (and Nadene) relative is thus an exceedingly wide one, of which the particular usages listed in our grammars are but specific applications or rather English (or French) translations. The lengthen- ing of the stem vowel and the voicing of a final voiceless spirant '° are merely secondary phonetic phenomena due to the presence of 10. In part no doubt retention of originally final voiced spirant. 1 believe it to be probable that in such Athabas- kan alternations as '/Vo/ strap : *tlol-e, *t'iol-e STRAP OF (one) it is the -/- which represents the old conso- nant, preserved because of the following relative element, and that the -i of the absolute form is du^ to a secondary unvoicing of'the old -/. the relative element. As this element became reduced to zero, these secondary phenomena tended to take over the properly relative func- tion. Just as we have the alternation of final -i : -/-, of -X (-') : -Y-, of -X (-') : -y-, of -c : -/-, and of -s : -;(-, so also these alternations occur initially " ; e. g. Mont, iue fish : se-llut my FISH (Pet.), Hare xi burden : se-yzl-t my bur- den, Hupa hi smoke : m'.[l]-h[t]d-t his smoke, Nav. Sin song : bi-yi'n his song (in Athabaskan terms *xen.: *-yen-i ; cf. Mont, cen : -yzn-Zy Hare ci : -yin-e). Here too the alternation could only have been due to phonetic circum- stances to begin with. If a word was closely connected, in thought and in position, with a preceding word or element, the voiced spirant (say /) was retained or the voiceless spirant was voiced. Athabaskan *xei burden : *cB-yel-e my burden thus originally meant no more than that two radical elements {*ce me and *xel bur- den) were united into a phrase with the help of a following denominating element *-e, *-e : me-burden the, i.e. my burden. The *-e preserved the -/ of *xel, ordinarily *xei, while the voiceless x passed to -y- in intervo- calic position '^ It is very doubtful if there was any specific function connected with the X- : -y- interchange. In time, however, there 11. In final position note also -r; (or nazalization) : 12. Possibly an old y- was here preserved, but passed to unvoiced x- when not protected by an immediately preceding vowel. The alternations listed in the text na- turally apply to primary Athabaskan. In certain dialects some of these alternations ceased to operate freely be- cause phonetic laws divorced the consonants that had originally belonged together. Thus, in Hupa the old Athabaskan interchange of x- : -y- (preserved in Chipe- wyan, Hare, and Navaho) had ceased to be a live pro- cess because of the falling together of Athabaskan .v and k' into Hupa x and the change of Athabaskan y to Hupa w ; the corresponding final alternation of -x : -y lingered on as -' : -w, -u. Note also that in Pacific Athabaskan voiced sibilant spirants have been leveled with voiceless sibilant spirants. Six: Athabaskan and Na-Dene Languages 147 140 can be no doubt that the voicing of initial spi- rants came to be felt as intrinsically, not me- rely mechanically, connected with the relative function. Hence such detached forms as "^^eUe THE ONE WITH A BURDEN (Nav. yel-i '^), "jUH-e THE GOOD ONE (Nav. jun-i), *jen-e the black ONE (Mont. :(enn). It is probably because of the intrinsically " relative " significance of " adjective " stems that these regularly begin with a voiced spirant if the initial consonant is a spirant. Such forms as Hare ^'^'e-e liar and Navaho t'^'I WIDE bring home to us the highly impor- tant fact that the actual " radical elements " of Athabaskan verb and adjective forms are more freely isolated than one might at first be- lieve to be possible. A careful study of all the available material would tend to show that these radical elements have a considerable mo- bility, that they are not far removed from the status of independent monosyllabic " words ", and that the complex " word " of our Atha- baskan texts and paradigms feels a great deal more like a closely knit phrase or sentence than has yet been suspected or, at any rate, explicitly demonstrated. I hope to show in due time what is the true nature of the various " prefixes " and " suffixes " that render the morphology of the Athabaskan verb so complex in appearance. It will appear that each and eve- ry one of these elements is a relatively self- contained unit in the sentence, either a deter- minative or an actually predicating element. Much of the " vagueness " of meaning or function that we feel to attach to many of these elements is simply an index of our inability to carry over the Athabaskan manner of expres- 13. Athabaskan y is preserved in Navaho only before a. It appears as y before i and e, labialized to w before 0. The alternation x : y, however, leads one to suspect that this " y " is not phonetically identical with the y that corresponds to common Athabaskan y. sion into precisely equivalent English (or French) form'"*. For the present I shall content myself with a few random examples suggesting the mobility and essential concreteness "> of the verb stem. The Hupa verb stem for '' to flow " is -Un, -hr„ which has numerous cognates in other dialects. Forms like t'cewishnt't it will flow OUT and nzdhn it always flows do not seem to suggest the possibility of combining the bare stem freely with other concrete elements. I have found no examples in Hupa of -/sn or -hr, so used, though they may 'of course exist. But this stem (Athabaskan *-lin, *-leri) is clearly related to Athabaskan *-// (e.g. 'Mont, -^a-i-d-li, Hare yi-d-li couler a terre [Pet.]), probably also to Athabaskan *-lej (tears) flow (e. g. Mont, dc-l-hz, Hare dt-l-W [Pet.]). Athabas- kan *-li appears in Hupa as -k (reduced to -/) ; it is found, without formal prefixes of any kind, in certain compounds : no-U dam, wa- terfall (lit. down-flow or halt-flow), iz-l- d'.r, together-flow-place (village name). Si- milarly, -xa{u)W, an " indefinite "form of-.vd liquid has position, is directly compounded 14. Very much as one who tries to see in a French lo- cution like Qu'est-ce qu'il a ? an exact equivalent of our English What has he ? would find it difficult to get the proper form-feeling of the elements est, ce and qu\ 15. In a large number of Athabaskan verb forms it is impossible to assign to the radical element (that is, the " verb stem " proper) the kind of concreteness that would in our eyes yield the actual concrete significance of the form. It does not follow that the stem has not a defini- tely concrete significance of its own, clearly apprehen- ded by the native form-feeling.. All that we have a right to say is that the summing of significances inherent in the theoretically independent elements of the complex does not seem naturally to lead to the precise idea that we express in such and such a way. The true difficulty from our naive standpoint is in such cases not that the stem is a « vague » element but that it seems inapplic- able, just as to one unacquainted with English idiom the notion of " fall " in such a sentence as The friends had a falling out is inapplicable. A a matter of fact, I be- lieve that there are few languages in America that feel their " stems ",.and elements generally, as definitely and as concretely as Athabaskan, Tlingit, and Haida. 148 VI American Indian Languages 2 141 with sa MOUTH : sa-xa(u) H^ liqjjid which has BEEN PUT IN THE MOUTH, ACORN SOUP. In Na- vaho, again, the verb stem -ua to live (e. g. X'.-n-c-na 1 live) may be used as an unmodi- fied element in a compound ; k'a'-na''-ni ARROW-LIVE-PEOPLE, LIVING-ARROW PEOPLE (clan name). Such examples could be multiplied con- siderably. If the monosyllabic *' verb stem " may thus be isolated in practice as a more or less freely movable element, capable of conveying a definite notion in its own right, we cannot but con- clude that the purely formal difference between verb (and adjective) stem and noun stem be- comes a tenuous one. What is to prevent us from interpreting the -h of Hupa no-lz as a noun meaning flowing or current, no-h and it-l-d\.rj meaning properly down-flowing (not a secondarily aominalized form of an inherently verbal to down-flow) and reciprocal-cur- rent-locality ? Might not the Navaho k'a'- na'-ni be just as well interpreted arrow-life- people ? As a matter of hex., I cannot see that anything seriously stands in the way of such an explanation, and its adoption would at once make clearer a number of morphological pecu- liarities. Among such peculiarities are : 1. the ease with which a great many evident nouns are transformed into " verb stems " (e. g. Hu- pa ^'fl DRESS : -k'a TO wear a dress ; Kato djit] DAY : -djvr^ TO BE day) ; 2. the frequency with which " verb stems " with aclearly defined verbal lorce, if we may trust all appearances, take on, when isolated, an abstract or concrete nominal significance (e. g. Athabaskan *-yan to pass THROUGH life : *xan old age, maturity ; *-dlo' TO LAUGH : *dlo' laughter ; *-lo to snare, to BE CAUGHT IN A NOOSE : N'av. lo'^ SNARE, LOOP); 3. the fact that a number of verb stems refer not to specific activity but to a class of objects 1 6. Voiceless spirants initially, voiced spirants in post- vocalic, or originally postvocalic, position, according to the typical Athabaskan rule. (e. g. *-'a« TO HANDLE A ROUND OBJECT ; *-^'oj TO HANDLE A CLOTH-LIKE object). I hope later to take up this fundamental question and to show that in a verb form it is not the " verb stem " that is the distinctively verbal element but, where found, the" third modal " element; that all " verb stems " are in fact nouns not only in theoretical origin but in actual usage ; and that verbs translated according to the forms TO DIE, TO BE SEEN, and TO KILL fall iuto patterns more accurately rendered by death is, sight TAKES PLACE, and TO MAKE SLAUGHTER. If this interpretation is correct, an element like -an is not properly a " verb stem " indicating some kind of activity or state with reference to a single round object but is actually a noun which means, or originally meant, a round object. The three classes of verbal usage listed above would fall into a single category applicable to all other verb forms as well. Indeed, it will appear that this theory of the, essentially nominal character of all " adjective " and " verb " stems simpli- fies enormously the whole aspect of Athabas- kan (and Tlingit) morphology. Meanwhile, whether or not we are willing to go so far in the present stage of our knowledge as to accept the nominal theory of verb radicals, it is clear enough that the Athabaskan relative forms discussed in this paper belong together. If a Navaho form like Ccah-i is to be interpreted as THE one HAVING (ot CONNECTED WITH) A HAT, we may venture to interpret a verbal deriva- tive like Hare k''^t-z as the one having false- hood, an adjectival derivative like Navaho jun-i as the one having goodness. There is certainly no serious point of morphology that would make such an interpretation impossible. As it is, it is sometimes an arbitrary matter whether we assign a given relative form to an adjectival or to a nominal source. In Navaho ga k\s-i one-armed, armless, k\s-i short, crippled, DEPRIVED OF may be looked upon as an adjectival (or verbal) formation (cf. gnd-i Six: Athabaskan and Na-Dene Languages 149 142 above). Identical, however, with Navaho k\s, which I have not found in its bare form, is Anvik Ten'a k'xf) piece (oF)(Athabaslian *k'es). Clearly it makes little or no difference, from the Athabaskan standpoint, whether we ana- lyze ga k'ls-i verbally as the one whose arm is CUT OFF(cf. Navaho verb -k'z, -ki'\ -k\ to cut WITH A knife), adjectivally as the one who is SHORT OF AN ARM, or nominally as the one WHO HAS AN ARM-PIECE, AN ARM-SEVERANCE. In the same way, it seems an indifferent matter whether we interpret Montagnais ts'ay-t ver- bally as THE ONE WHO IS (aLWAYS) CRYING Or nominally as the one who (always) has tears, WEEPING ; or Navaho yil-i (a sobriquet for a hunchback) verbally as the one who is wont TO carry (a burden) (c(. Navaho -xe or -xd, -yr, -xi-i, TO carry on one's back, to handle A burden), or nominally as the one who (al- ways) HAS A BURDEN. Editorial Note Originally published in International Journal of American Linguistics 2, 136-142(1923). THE PHONETICS OF HAIDA TABLE OF CONTENTS Introductory. I. Consonants. The Consonant System. The Intermediates. The Unaspirated Hard Stops. The Aspirated Surds. The Glottalized Stops and AfFricatives. The Voiceless Spirants. The Nasals and Voiced Spirants. The Glottalized Nasals and Voiced Spirants. The Laryngeal Consonants. Secondary Consonantal Processes. Initial Consonant Clusters. II. Tht' Syllable. Syllables with 1-Vowel. Syllabic Nasals. III. Vowels. Qualitative Changes. Vocalic Quantity. Diphtongs. W. Stress and Pitch. INTRODUCTORY The following notes on the sounds of tlie Skidegate dialect of Haida are based on mate- ri.il which I was fortunate enough to secure iVom Peter R. Kelly, a well educated Haida Indian who is at present engaged in missionary work among the Indians at Nanaimo, Vancou- ver Island. Mr. Kelly visited Ottawa in March, 1920, as member of an Indian deputation to the Canadian Government and was too much occupied to give me more than a few hours. In spite of the brevity of my notes I believe the insight gained into Haida phonetics is suf- ficient to warrant this paper. I cannot, of course, give an adequate account of the Haida sound- system, but purpose merely to present data supplementing Dr. Swanton's brief statement '. The phonetic system employed in this paper is explained in « Phonectic Transcription of In- dian Languages » (Smithsonian Miscellaneous Collections, vol. 66, n° 6). A remark or two on the general impression produced by Haida maybe of interest. I took several opportunities to have Mr. Kelly speak Haida connectedly and was thus enabled to hear it long enough to form a definite image ot its acoustic quality. It is one of the most remarkable languages that I have ever heard. Indeed, I cannot recall having at any time heard connected speech that appeared more definitely possessed of individuality. The great frequency of nasal consonants {n, r,), the con- stant occurrence of sonorous ^-sounds, the pro- fusion of /-syllables (see below, p. 152), and the musical cadences are probably the chief determinants of this individuality. Haida is very far from being a harsh language. On the contrary, it was voted a beautiful language by all who heard Mr. Kelly's recital of a Raven myth. Several of us in Ottawa heard connect- ed Mohawk, Tsimshian, Nass River, Thomp- son River, Shuswap, and Danish at the same time. If we were asked to rate these seven languages on the score of acoustic appeal, I believe the consensus of opinion would be a division into loin' groups : Haida as an easy I. See pp. 210-215 of I. R. Swanton, HuiJu, tin Illiis- lialive Sketch (Bureau o( American luhnology, Bulk-tin ^O, pt. I, pp. 20)-2S2 [lc>10]). 152 VI American Indian Languages 2 I 144 first; Mohawk as a fairly pleasant, but none too close, second ; Tsimshian, Nass River, and Danish as a moderately uneuphonious third ; and Thompson River and Shuswap as an exe- crable last. These remarks are of no great scien- tific value, but they may be of some interest none the less as serving to bring home the fact that the « harshness » of certain West Coast languages results from the printed page rather than from their actual articulation. I. — Consonants. Thk Consonant system. — Swanton recog- nized 28 organically distinct consonants in Haida. I believe his table errs in two respects : in not including a number of sounds which he recognizes as existing but does not consider as elements of the fundamental sound pattern of the language ; and in neglecting to take ac- count of certain sounds that he did not hear. To the former class belong the anterior palatals and the labialized gutturals and velars. To the latter class belong the glottal stop ' and a set of glottalized nasals and semivowels. As to the anterior palatals, Swanton remarks, " An anterior palatal series might be added to these, but the sounds to be so characterized seem only palatals followed by a close [read " front "J vowel. " I do not know if this is originally true or not, but I think there can be no reasonable doubt that the anterior pala- tals are felt as a primary series. They occur before /-, »-, and rt-vowels, which last they co- lor to ci. It is true that in such a form as xdi ARM one sometimes hears a slight z-glide (x'di), but I do not think we have the right to conclude that xdi is felt as a secondarily modi- 1. lam not referring to the glottal stop, if it is a glottal stop, of Masset Haida, which corresponds to Skidegate velar ^, but to a primary glottal stop in Skide- "ate itself. • fied form of xiai, though it is of course pos- sible that it may go back to such a form. At any rate, I prefer to consider the anterior pala- tal series as a phonetically well defined prima- ry group of consonants and the /-glide, when it occurs, as the secondary fact. I do not hear gia''da.i the blanket, for instance, but gd''da.i. As a matter of fact, Mr. Kelly's ear proved ex- tremely sensitive on the diflfrence between the k- and ^-series. The word for eyebrows was at first recorded sk'd-'dji, but this pronunciation, in spite of the a-vowel, did not satisfy him. The correct form is sk'd''dji. There is even reason to believe, if my record is to be trusted, that there is a difference between the ^- and the ^-series before /-vowels. Thus, I heard go'xxg'H- gx TO BE starting TO BURN, not -^i/- (misheard for -^J- ?) ; t'a'gr'gx TO be through eating, not -W"- as in p'sdi'r, two blanket-like OBJECTS. However, I have not enough evidence on this point to be justified in speaking with confidence. As regards the labialized gutturals and velars, Swanton sometimes writes according to the form Java (i.e. k'zva) to strike, at other times, according to the form sgod'nsir, (i. e. sgwa''- nsvr^ one. Here again I think there is no rea- sonable doubt that we must look upon the la- bial element (whether written lu, u, or 0 in Swanton's material) as a constituent element of a primary labialized ^-sound. Counting the anterior palatals, the two sets of labialized k- sounds, and the new glottal and glottalized consonants, we have 47 primary consonants in Haida. They may be arranged as follows (see p. 145). Of these consonants, I did not myself obtain an example of aspirated p but set it in the table because of Swanton's form djA'pAt to SINK SUDDENLY, which I interpret as dju'p'at. It may, however, be really djxp'at (i.e. djxp -\- \t; cf. X3.T,\' eye), in which case p' would have to be removed from the table. Aside Six: Athabaskan and Na-Dene Languages 153 145 Aspira- Glottal- Intermediate ted ized Stop Surd Stop Surd Stop Voice- Glottalized Glottalized less Voiced Voiced Nasal Nasal Spirant Spirant Spirant Labial b p' m 'jn Dental d t' /' n 'n Palatal Sibilant (AfFricative) dj tc' t'c Anterior Palatal i Js' ¥ Guttural g )i k' n '■n Labialized Gut- tural gw k'w k'w Velar S f > Labialized Velar gw q'w qw Lateral ' dl ti fi Laryngeal ' X X xw y yw i h w w from w and 'w, which are common, and m, which is not rare as syllabic final, labials are very uncommon in Haida. Most, if not all, examples of initial m and 'w may be suspected of occurring in Tsimshian loan-words. Besides the 47 primary consonants that 1 have tabulat- ed, three secondary ones must be noted : syl- labically final p and /, which are true surds but unaspirated, and spirantal velar y ; p and / are secondary forms of b and d, while 7 is some- times heard as a variant of g between vowels. It is barely possible that we should distinguish between guttural q and velar r,, but I did not hear the latter sound. The INTERMEDIATES. — By " intermediates" are meant unaspirated " voiceless lenes, " con- sonants pronounced with the voicelessness of typical French surds (/?, /, k) but with the les- ser energy of our sonants {b. d, g). To the ears of English-speaking persons they are more apt to sound like sonants than surds, while French- men would be almost certain to set them down as typical unaspirated surds. The Haida inter- mediates are identical with the corresponding series heard in Iroquois, Athabaskan, Takelma, I. First three laterals are affricative. Yana, Achomawi, and Miwok. A recent oppor- tunity to study Mandarin Chinese phonetics has made it clear that these typical American " intermediates" are absolutely identical with the unaspirated surds of Chinese, which sound distinctly " softer "than the surds of French or Italian. It is possible that the Haida interme- diates are sonant at the moment of release, but their general effect, if carefully heard, is cer- tainly not that of sonants. They are here written as sonants because it is convenient to adhere to establisjied usage and because it is advisable to. keep p, t, ^ for the " harder" secondary forms that may result from them under appro- priate circumstances. The iv of giu and giv and the / o( dl are fully voiced. I supect that the d o{ dl'is at least partly voiced also. A word as to g. This is generally a very firmly pro- nounced stop, hut it seemed to me that a faint uvular trill could sometimes be detected as a glide between the g and a following vowel. Examples of intermediates are : a''bx TO CHEW FOOD FOR A CHILD dxr, THOU sdx'lISvr, FOUR k'u'dx LIP djr'djW TO BE SPLIT UP 154 VI American Indian Languages 2 1^6 dja-'dy. WOMAN t'/r EAR digud/x'rga MY DAUGHTER t'a-'gx TO EAT giva'i ISLAND ga''.'na bucket ga'xa'' CHILD gtva- sea-bird (sp.) dla'i peace, quietness After an accented short vowel Swanton heard a / before dj and d. Hence he writes klA'tdju and xA'tdjii small ' for what I heard as k'xdjir ; similarly, klA'tdAla small ones ^ The firm, voiceless attack of the dj after a markedly short vowel created the illusion of a syllable-closing /. As his classifiers k !At- and xAt- are illustrated only before dj, it is almost certain they should be read k'a- and xol- (Swan- ton's x and .V, it should be remembred, are my X and X respectively). He himself writes xA- in xA'dAla, the plural of xA'tdju '. That our analysis is correct is demonstrated by k'oLSgwan- S\:r^ ONE little OBJECT, k\sdir, TWO SMALL objects. The point is of some interest for N.idene, as it leaves Haida ^'a-, classifier for small objects, identical with its Tlingit cognate -k'x to be s.mall. The unaspirated hard stops. — When b and d appear at the end of a syllable, they sound much more like our/) and / than do the ordi- nary intermediate ^ and d. They impreess the ear as the normal English p and/; in which the breath release has been suppressed. Examples are : t\'pdjlf TO BE TALL AND STRAIGHT dx'pdju TO BE VERY SHORT "» 1. Op. cit , p. 231. 2. Op. cit., p. 241. 3. Op. cit., p. 276. 4. Swanton gives t.'Ap- as a classifier for short and protruding objecis (op. cit., p. 234), which does not correspood to my data. He does not give dAp- as a clas- di''gltga MY CHILD q'e't SPRUCE (cf. q'e-'d-x to be spruce) Sget RED ^i''x\t to. PICK UP (a canoe) That the final stops are not aspirated is a noteworthy fact in view of the common Ame- rican Indian rule that final stops, whether in- termediate or aspirated surd in origin, are re- leased by a markedly audible breath (e. g Oji- bwa, Tlingit, Nootka, Comox, Takelma, Yana, Paiute). In this respect Haida difl"ers from its remote relative Tlingit and is in accord, it would seem, with certain Athabaskan dialects. Goddard writes, e.g., Hupa Lit smoke, while the Franciscan Fathers write lid in Navaho. It is likely that 'an unaspirated surd is tneant in each case. If I may trust my memory of Chasta Costa, there too final stops are unaspirated, e.g. l-t'at TO GO to pieces. The aspirated surds. — These are "hard" and markedly aspirated surds, much like our English p, t, k when initial before vowels ex- cept that they are even more strongly aspirated. They are identical in every respect with the aspirated surds of Athabaskan, Tlingit, Takel- ma, Yana, and Chinese. Examples are : t'a- TO eat st\ from tc'i'nga grandfather tc'i''djlC TO BE a BIG, BLOWN-UP THING k'd'xwx OUTSIDE Vmi POINT, NOSE sk'w'djW TO BE SMALL AND ROUNDISH k'wa'i HIP q'a''dji' head (J ' I Tig XT, TO BE SEEING sifier. Apparently there was some confusion between these rhyming classifiers. 5. Swanton defines the classifier tci- (op. cit., p. 227) in terms of « such objects as full sacks and bags, pillow, s, etc. ». Mr. Kelly stated Ic'i- referred primarily to the blown-up stomach of a seal used asm float in fishing. Six: Athabaskan and Na-Dene Languages 155 M7 q wa'i ROPE tia'S BRANCH tc'a'tidju'gx TO BE FAT Swanton remarks, " It is doubtful whether d and / [i. e. /'] and dj and tc [i. e. Ic'] really exist as recognizedlys eparate sounds " ' . I do not see how there can be any reasonable doubt on this point, da' thou and t'a' to eat ; dji--, classifier for cleft objects (like hands), and tc'i'-, classifier for blown-up objects ', are as per- fectly distinct as g and k' or g and q\ (Swanton does not list dji'- as classifier. He either did not isolate it as a classifying element or he confused it with tc'i'-. Examples of its use, besides dji''- djw already quoted, are sila'i dji'-sgwa''nsir, ONE hand [literally, hand cleft-one], stki'i dj'.-sdi'r, TWO hands, k'u'dx dji'-sd\r, TWO lips. There is no doubt that there are plenty of examples of dji'- in his material. Note, for instance, dji'wAl [Ma.sset] roots of fallen trees). The glottalized stops and affricatives. — These are the well known stops and affri- catives pronounced with simultaneous glottal and oral closure and with glottal release follow- ing upon the oral release. Swanton says, " Some speakers bring these out very forcibly, while others p.iss over them with considerable smoothness ^ ". Mr. Kelly pronounced them quite as smoothly as any other consonants. Their essential nature is certainly not to be explained as due to " urging more breath against the articulating organs than can at once pass through 5 ". If there is a true " fortis " series in Haida and Athabaskan, it is the aspirated surds, which are indeed pronounced with an excess of breath. Examples of glottalized surds are : 1. Op. cit., p. 210. 2. Op. cit., p. 210. 3. Op. cit., p. 210. t'a' gun feather t'a''r,OLl TONGUE st'a'i FOOT /Vw red cedar qoLuCcdoL cheek q'a't'car, to be going in ^'d'll leg * k'i''dji seal-stomach k'a't DEER k'w'dx LIP k'wd'i TO WAIT q'a'ri HEMLOCK q'e'sdir, two spherical-like objects qWCl'Tt TO BURST fia''doLn gorge C'tll WE ' I may note that/V tends to move front in its position — either to that of an nnterior palatal c- sound (/V) or even to that of an .f- sound (e. g. i''knCsui3. dla'' msdir, two gigantic people ; but also q'y.'nt'c:dx cheeks. Does t^adx corres- pond to Swanton 's -djit (op. cit., p. 260) ? The VOICELESS spirants. — These require no special comment, x is pronounced like ch of German ich. As Swanton remarks, s often interchanges with dj^, which does not normally occur as a syllabic final. Swanton states that " s becomes dj before most vowels ". Inasmuch as both dj and s occur between vowels (e. g. i'dji' to BE : isdxr, TO BE CAUSING TO BE ; but also " participial " -asi, -as'.\ would it not be better to say that -dj- becomes -s just as -b- and -d- become -p and -/ .'' We would then have one final (-j) representing two distinct intervocalic consonants (s- and -dj-). Examples" of voiceless spirants are : sdw'nsOLT^Xy. EIGHT a'k'oSU THIS THING xdi ARM 4. See p. 152 regarding syllabic /- sounds. 5. Op. cit., pp. 214, 215. 156 VI American Indian Languages 2 h8 t'a-'xidlgJ. TO BE ABOUT TO EAT xa-'ya sunlight xwi' to be cold Xa' DOG X7Lr,\' eye xd neck xiua-'igcr a thing that is loose la\ ia, il i igiinnl THREE The nasals and voiced spirants. — Of the three nasals, n and c may occur as either sylla- bic initial or syllabic final. In such a word as 'l(i''(ar,a THEIRS the guttural nasal q must be considered as belonging to the final syllable, in such a word as xoc'r^'c eye to the first, m, as we have already seen, is rare as an initial but not uncommon as a final. Examples of final m are : t^'mdjtr TO BE something THIN AND ROUNDED fy-m lice (Mr. Kelly considered this word as connected with the classifier t\m-. This mat be only a folk etymology, however.) dld''nidjn' to be a gigantic person ga''mdju- TO be a wide thing ^d'mdjir TO be a large (canoe, blanket) Several Haida syllables ending in m seem to belong to a set of classifiers : t'xm- THIN and rounded dlam- gigantic, corpulent (?) ga'm- wide and rounded ^iim- large (in reference to canoes and blankets) igum- (Swanton) large and roundish (e. g. rattles) Fa;//- (Swanton) small and roundish js,'rt'm- (Swanton) large around (?) dla'm-, ga'm-, and gam- are not listed by Swanton. My ga'm-, however, may be the same as his ga-w- (op. cit., p. 235) ; I am cer- tain of the velar g, for the element was also recorded zs -(am-, e. g. ^w -(am-sgiva'nsu, one big ear. Although the evidence is far from complete, I would risk suggesting that this set, of which there are doubtless other, members, contains a common element -;//- indicating something like rounded, all around. This view is strongly supported by the fact that several of the m- clas- sifiers are clearly related to other classifiers without -m or with final -p. Thus, to t'am- may correspond t'a- (Swanton) coiled and flexible; to ga'm- evidently corresponds ga'- FLAT (e.g. pr -(a'sgiva-'nsir, one flat ear); ^dm- is clearly related to p'- canoe, blanket- like (e.g. tin' p'sgiua'fiS'.T, one canoe and ^d''at grsdi'r, two blankets : tiw ^d'mdjus A large canoe, tiw gd-'mdxh sdvr^ two large, SPREADY CANOES, m'' ul gamdoiJa gi'sdir, two large blankets) ; igxm- probably belongs to iga- (Swanton) branching objects, k'xm-, as Swanton himself points out, is derived from k'x- SMALL ; and slla'm- belongs with sthp- (Swanton) slim. Parallel to the -m- series is a -p- series (fo^p-, doip-, stixp- [Swanton], t'ioip [Swanton], audita/)- [Swanton] = probably sk'oip-^. Of these, tap- may go with t\m- ; stiap-, as we have seen, with slia'iii-; t^ixp-, judging from Swanton's one example, which refers to the surface of the moon, goes well with his L!- thin and FLAT (as I shall show later, this must be interpreted as t'il-, a reduced form of /'fa-, t'ia-); and skap-, which in his one example refers to the curled tail of a dog, may go with sk'a'- roundish (used of eyes, water-drops, berries). The meaning of the -p- series is more obscure than that of the -m- forms ; possibly stubly protruding covers the case. Ond" sur- mises that the complete set of Haida classifiers is a more complex subject than Swanton's data represent. Further, that the i- and s- which begin so many of the classifiers (and other noun and vtrb stems) are vestiges of an older classi- ficatory system that was related to the /i- and s: classifiers discovered by Boas in Tlingit. The Six: Athabaskan and Na-Dene Languages 157 NOS. 3-4 whole subject of Haida classifiers needs a rene- wed and intensive study. Examples of n and r, are so numerous in the forms scattered in this paper that I do not need to give further examples here. Unglottalized w does not seem to be as com- mon a sound as glottalized w. Examples of w, y, and /.are : ha- Wit HURRY ! t'sCWx'ga TO BE A CEDAR-BOARD yd-'goLlari ancestors ga''yir sea q'a'fcaya''gvanton's hao), often contracted to ir , u, 0. The vocalic hiatus without glottal stop is thus the etymo- logical equivalent of an old h. Examples of the verb i''dj\., '.5 are : la 'o' na'.t -;a c'dj-.-ri i am in a house tllgU".'' U tx isdxr, I CAUSE IT TO BE AWAY, 1 PUT IT AWAY See also the first example under demonstra- tive u, 0. The consistent absence of the glottal stop in this verb may possibly be explained as a slurring, but I think it more likely that it is to be interpreted as due to an old /; that has disappeared. I would suggest that i'dj\, is is a I. a- slurred from 'a-, the demonstrative stem this corresponding to ';tw- that. Six: Athabaskan and Na-Dene Languages 159 151 later form of hi'dj'., his, and that this verb is to be understood as composed of a demonstra- tives stem hi'-, hi- (cf. hi''-dll-goLl come here ! and Swanton's hi-t!A-gA'n then, hi nan only). Justasthis /;/- parallels /;fl-(cf. Swanton's hx-n LIKE, AS FOLLOWS ; general demonstrative hao^if , u,o; " article "*hai, parallel 10 gai, in e. g. *na hat house-the > na'i), so *hi-dji, *hi-s TO BE parallels an old *ha-s, preserved as Swanton's " participial " -as, -is, -es. Such a phrase as Swanton's nxn gaxd'gas one who was A child is to be interpreted, it seems to me, as «ar( gaxa-'-g a-s one child-be it-is, contracted from an older gaxa'-ga has. Such endings as -^-5? (S wanton) are probably to be interpreted as -ils\ rd" ing to accentual or other phonetic circumstances with full syllables in a', a, a (e. g. la', 'la in 'In' .va'r/'." HIS EYES, na''i •;'. 'la ga't'car, HE is GOING INTO THE HOUSE : '/ 'i''ii.r,7. tc'r'djw HE IS A BLOWN-UP (obese) MAN ; /fl I : // ; -dxloL adjective plural : -dj.il). That Swanton too heard, though he did not explicitly record, /- Six: Athabaskan and Na-Dene Languages 161 1)3 syllables is shown by such accentuations as Z,'-, /'-, I'-andL!'-. Examples of /-syllables are : k^all LEG da''ll RAIN (recorded as dal by Swanton, which failed to satisfy Mr. Kelly's ear. This word is not only clearly disyllabic, but the accented vowel is higher in pitch than the /- syllable. Cf. Swanton's ddla-ge'ii-si rain fell, BBAE 29 : 12, 1. 8.) 'a'll PADDLE qui'' FOREHEAD (should have been record- ed q'u'Il or qui') 'IJgx'nga COUSIN 'h;a'i'^';a keeper dx'n U YOU I. . . (not to be confused with dxnl swelling) Sge-'il TO CRY titil TO RUB one's hand against tlc'll five q'w'dll to go aboard a canoe hi'dllgal come here! tilga' earth tilgWV' AWAY til^u'nui SIX t'il'^d' to soak (a dry salmom) 'i''t'ilgd-fa ours When an /-syllable beginning with /, 7, or dl is pronounced with very weak stress, the length of the / is of course reduced, though it never loses its syllabic character. We may then write simply /, '/, dl. The syl'able -ilr, (reduced from -/ay;) was heard as -ir,, with syl- labic q, in lia'^ait, (i. e. tia'"al avj) sgu'a''ris:T, gou ten it-is one missing, nine. In interpreting the phonetics of Swanton's texts, there is no genuine syllabic ambiguity in the case of initial L-, L-, L!-, and /- before consonants and of final -L and -L/, which necessarily represent /-syllables. It is different, unfortunately, with initial / before consonants and with final -/, -/, and -L, which may repre- sent non-syllabic or syllabic laterals. It is im- possible to tell offhand whethersuch orthograph- ies as igeda-i and ga'lxua represent igc'daJ or ilge'da.i, ga''lxu'a or go'lhwa. It is hardly conceivable that the morphology of Haida can be adequately understood without an exact knowledge of its syllabification, for the Haida " word " is essentially a group of significant syllables. The recognition of /-syllables is likely to put many points of grammar in a new light. Thus, we may surmise that the two forms of the first person plural (subjective fa-la-r, and objective ''/'-t'il) are closely related, ''/'-til being a reduced form of *\''-t\xya. The true basic forms would then be fa and 'c-fa, -la and -Ix-r, being pluralizing elements. Syllabic nasals. — It is quite possible that we should recognize also syllables of type n and syllabic q, alternating on phonetic grounds with syllables of type w. and r,a. The alternation of-r; (-n) and -/;« in cases like i"7.-/; (or i''i:.n) fsdx sdvr, two men and i''hr,x sgiua''nsiri one man should perhaps be interpreted as an alternation of i"'/iy;y; {'.''hnn by assimilation) and i'7tr,a. This consideration may explain the constant interchange in Swanton's material of forms like -gAn and -gAni, -agAri and -agAni, -gin nnd -gtni, in which -/ (i. e., -i) can hardly be a " perfective " element, as Swanton assumes. For the present, I cannot say whether Haida distinguishes final -n and -r, from -n and sylla- bic -Y). All I can say is that I feel strongly that while Haida reduces the quantities of syllables freely (e. g., in such a gamut as 7^/', 'la, 7a, '//, '/). it resists the extinction of syllables, and that if syllables ar.^ actually lost, they are lost phoneticallybut not psychologically. HI. — Vowels. I am able to give only a preliminary idea 162 VI American Indian Languages 2 1)4 of the Haida system of vowels. The vocalic nuances seem to be due primarily to secondary phonetic causes rather than to basic etymolo- gical differences. It is quite probable that there are only three organically distinct vowels : a, z, //. Each of these runs through a gamut of quantities and qualities that give the lan- guage a far greater vocalic variety than the simplicity of the fundamental vowel scheme would argue. Qualitative changes. — The a-vowel seems to be the most liable to phonetic change. Its fundamental quality is that of German a in Mann, e. g. da''ll rain, but when short, whether accented or not, it very frequently assumes the duller timbre of a. Thus, -ga TO BE and fa i vary with -gx and // (i. e. /x) in : ia ga l'a''s[a i am eating something a'k'os u il l'a-''(xsgx i shall go and eat THIS thing The a-timbre is particularly common before nasals, e. g. -gx-r, (continuative of -ga^ and -^a-n('.) (past definite oi -ga), /'aw lice. (It seems fairly evident to me that Swanton's temporal -gxn [op. cit., p. 247], future -sga [p. 249], and continuative -gah are merely com- binations of " declarative " -ga to be or to be so and certain particles that he has not properly isolated : -r,, -r,a to be continuously ; -s, -sa- ABOUT to bk ; and -n -««. to be at a given MOMKNT OF TIME, TO HAPPEN. It is clear that they occur also without -ga. Boas has already pointed out the analysis oi -s-ga- [see p. 249]. I hope to show at a future opportunity that the whole tense-modal system of Haida is nothing but a loose compounding of demonstrative elements and particle verbs and that the synthe- tic nature of this scheme is more apparent than real. Thus, Swanton's " infallible future -asah [p. 249] is merely a verb phase : 'a-sa-r,- [a] this-will-be [durativkly]). After anterior palatals and y, a (a) appears as a (as in English mat), perhaps even as z (as in English met), e. g. : k'd'xwx outside ^'d'll LEG (also heard as k'e'l) Xdi ARM t'il'^d- to soak (a dry salmon) sk'd-'dji eyebrow ^d'gxn MINE ^d''da.i THE blanket ^dm- wide and sprhady (classifier) yd''gxlxq ancestors After sibilant afFricatives and also before r, there is a tendency for x, particularly if unac- cented, to pass into i (English i of bit). Before yj this seems to take place particularly after laterals (/, /) and sibilants. Examples are : sdvr, TWO (d. sdxnsi-f] four, dissimilated from sdx'n-sdvr, gi-- ga-d- >■ gd'-d- ^a'-m- >> gdui- (closed syllable, hence with shortened vowel) Again, g'rw-ai the ear : ^w ear can be best explained as palatalized from an old ''gaw-, final -an o{*gaii contracting to gw . This inter- change of -/'tf- : -w seems to take place also after s, e. g. Skidegate sfi (Swanton) to say : siw-As (contrast Masset sdw-^. A basic saw- best explains the alternations. Another alter- nation of similar form is that of tin' canoe : tilw-a'i the canoe (Swanton always writes Lu-a'i). This again is doubtless the reflex of an old tlaii : tlaiu-. For the /-vowel, /, i, and e were heard as variant timbres. The t-timbre is quite frequent, perhaps normal, for the long /- vowel, e. g. : 'c'til^dya OURS A-a'r/f EYE tilglUf' AWAY djr'gO'-(OL SEVEN '.■ ilTfT. MAN Examples of the less common ^-timbre are : c'dji-q IS xei REGION OF THE NECK (cf. xd neck). The characteristic timbre of the u-vowel is u (as in English put), here written u. The cor- responding long vowel was heard partly as o' (e. g. go'dx TO START A eire), partly as w (as in English fool). The latter seems to occur chiefly after sibilants and anterior palatals (e. g. -dju- TO BE so AND SO, gU' EAR, 'y/r"Qt« big). A glide a was noted in 'lyai^-^a keeper. In ^d''at BLANKET, the long a vowel, modified to d after g, reasserts its proper quality before the final / and thus appears as a broken vowel. Vocalic quantity. — Quantity is a diffi- cult matter in Haida. It is likely that there are etymological quantitative distinctions, but it is impossible to be certain, as the actual quan- titative variations are clearly largely due to secondary lengthenings and shortenings of the fundamental vowel. Thus, as already pointed out, 'la he may be shortened to 7a, 'II, '/or lengthned to 'la' . The determinants of this quantitative variation are probably pho- netic rather than morphological, but I doubt if stress accent is the only or even the decisive factor, as both the long and the short vowel may occur in an accented or unaccented syl- lable. I suspect that the distribution of quanti- ties is the resultant partly of inherent quanti- tative distinctions (e. g. long a' in yd''gxhr, ANCESTORS : short a, a in k\'dju' to be short), partly of a tendency to establish a rhythmic equilibrium. This equilibrium seems to depend on several factors, chief among which are the nature of the syllable (a closed syllable tends to be short, an open one long), the place of he accent, and the grouping of the syllables in phrases. It is impossible to give rules at present ; the subject is evidently complex. At the same time I do not feel that the quantities are distributed ad libitum, rather that they remain to be discovered. The Haida type of quantitative variation is somewhat similar to that of Athabaskan, if I am not mistaken. Presumably the Tlingit feeling for quantity is analogous to that of Haida. Boas merely remarks that " the quan- titative value of Tlingit vowels varies conside- rably " '. In the body of his grammar he unfor- tunately pays no attention to the subject of quantity. The quantitative peculiarities of the Nadene languages are in striking contrast to those of the neighboring languages. In Es- kimo, Wakashan, and, I believe, in Tsimshian as well, the inherent quantitative value of a vowel is clearly felt and, on the whole, rigidly preserved. In these languages vocalic quantity I. F. Boas, GiamiiiatiCid Notes on the Laiu:;iiaQy of llie Tliu^it Indians, Univ. of Pa. Anthr. Publ., vol. VIII, no. 1, 191 7, p. ; I. 164 VI American Indian Languages 2 156 is as much a matter of etymology as is the consonantal framework of the word. Phonetic variations in the quantity of vowels such as are found in Haida would be well nigh un- thinkable in Xootka. Examples of quantitative alternations are : a — g^i'-y^'' CHILD : ga'Xx i''i':r,x CHILD MAi.K, boy; gaxx .ija'dx child female, girl s^'d''dji' EYEBROWS : di' sk'ddj w v'dji' IT IS MY EYEBROWS sda-'nsar,xj. eight : sdy.'ns'.r, four na-' house; na'ga'i yfl 'a in the house : na I'e' 'sgwa'nsvr, one house sgwa''nsi.T, OYiE ; q'a-'dji q'e'sgwa-'nsir, one head; k'un dji'sgwa-'iisir, one nose; k'nn k'xsgwa-'uy.r, ONE LITTLE NOSE, POINT ; ou' ''^d-sgwa-'ns\r, ONE (flat) ear : gir -'amsgwa'n- S'.r, ONE BIG EAR ;^M' k' x'sgwanS:r, ONE LITTLE E\R;m-'dai ^rsgwa'nsir, one blanket, shawl i — ; q'a''dji head, hair of head : q'a'dji 7/ — ; la-'o nw'gai vt- q'a't'car, he is going INTO A house : na''i vi 'la q a't'car, he is GOING into the HOUSE dji-'djir TO BE SPLIT up, CUT UP : stia'i djisgwa-'ns'.r, one band ; stia'i djisdir, two HANDS /•V/V" TO BE, e''djirt TO BE (duratively) ; ia' '.sdxr, I CAUSE TO BE, PUT // — sk'a''dju' TO BE SMALL AND ROUNDISH ; ^i'dju-'gx (it) is LONG AND MASSIVE, SPREADY; tilwa'i u gdmdju'gxn the canoe is large : xd k'lf'dJH neck IS short; dx'pdjti very short; litr gaiiid/us a large canoe di'sk'ddjw MY eyebrows; ia''o' la'gd^'a na'i godxgx I PUT HIS HOUSE ON FIRE ; a'k'os U {( l'a''gx I AM EATING THIS ; '■lVx!s\' 0 gO'Xx- g'.'lgxr, THAT IS STARTING TO BURN Diphthongs. — There are two diphtongs in Haida, (7/ and an. Each of these exists in two quantitatively distinct forms — short (a/, often contracted to e' ; an, ou^ and long (a'/, a.i, sometimes heard simply as ^Z; a'o\ a'o, ao'). The latter type seems to result from contraction. The uncontracted short ai is well illustrated by the enclitic articles -ai (probably <; *hai), when appended to a stem ending in a con- sonant, and gai, e.g. : St'i''iuai THE SEA-EGG na''gai y<.' o into a house According to Swanton ', the Skidegate ai appears in Masset as e', but Mr. Kelly, a Skide- gate Haida, pronounced a number of fl/-forms with ^"-vowe!, e.g. : qe'- classifier for large roundish things (cf. Swanton s q!ai-, p. 232) yd''ge' PARENT <^ yd'' gai (cf. T^\\iv.yd''gxlxr\) t'e'- classifier for bulky lying objects (cf. Swanton's t'ai-, p. 227) q'e't SPRUCE (cf. Swanton's^flf/TREE, p. 271) tic'il FIVE (probably contains iia- ; cf. tlx, til- TO touch, s-tla hand) Possibly the Masset-like g'-forms are gaining currency among the younger people at the expense of the older az-forms. The contracted e' is, of course, not to be confused with the e' which is merely a variant of r. The long a'i (a.i) probably always results from the contraction of fl -f- ai, e.g. : na'i THE house < xd'i < old *xa-(h)ai ; *xa itself passed to xi (Swanton's xi^) as *^a passed to^^r'. The short ^/<-diphthong is written ao by 1. Ot). cii., p. 213. 2. Op. cit., p. 226. 3. See page 155. Six: Athabaskan and Na-Dene Languages 165 f$7 Swanton. I heard it as an ordinary a -\- u and did not feel that the labial element was low- ered to o-position. Examples are : 'a'uga MOTHER (Swanton's aoga^ sq'au GROUSE (Swanton's sqao) It seemed to me that it was rather the a that tended to become modified in the direction of u ; thus, I heard gm to be wanting (in tia-"air, sgwa'' ns'.Tt gJu nine) and 'o''tiga, almost 'o''ga, as variant of'a'uga. This tendency o( au to con- tract to a monophthong is carried through, e.g., in ga''yu' sea (cf. Swanton's gayao, p. 272), ^tf ear (*^aw, *gau, and in w, 0', u, 0 as reduced form of demonstrative hau. The long a'o, also heard as wo' and ao', is markedly distinct from- short au. It seems to result always from the contraction of a -|- (/?) au, e.g. : 'la'O', 'la'o HE (emphatic form ; «< 'la-[h\aii) ia'o', lao' I (emphatic form; a Language (University of Cali- fornia Publications in American Archaeology and Klhnology , vol. 3, 1905), pp. 117, ir.o. 3. The Franciscan Fathers, A Vocabulary of the Navaho Language, 2 vols., Saint Michaels, Arizona, 1912. 4. Op. cit., vol. I, sub" give " and " lake ". I have adapted the orthography of the Franciscan Fatiiers to my own. In my orthography accents following vowels are indicative of stress; wlien placed on the vowel, of tone. Six: Athabaskan and Na-Dene Languages 173 189 stem, e. g. i'yi "you marry": iyt "he marries;" txadi'gis "you wash yourself" : txa'digis "he washes himself" *. I shall now proceed to a brief discussion of the Sarcee tone system. It is impossible in a brief compass to do justice to the subject, which is diflicult and involved. Sarcee tone cannot, properly speaking, be discuss- ed as an isolated chapter of the phonology. It interpenetrates each and every aspect of Sarcee morphology. None of the elements of the language, whether radical, prefixed, or suflixed, can be accurately understood with- out a consideration of its tone and of the tone shifts which it undergoes in combination with other elements. All that I can attempt here is a preliminary treatment, designed to give some insight into the tonal peculiarities of an Athabaskan language. 1 hope to give a full and system- atic study of Sarcee tone in a later work on the phonologv and mor- phology of this language. My method at the present time will be, in the main, to list and discuss a series of apparently homonymous sets of forms. This informal and inductive method will probably convey more to a public necessarily but little acquainted with Athabaskan morpho- logy than a more compact and generalized style of treatment. It will be convenient at times to refer to the general or reconstructed Athabas- kan form of a word. Aly reasons for the reconstructions - can hardly be given here but must be reserved for later comparative studies. In any event, the Athabaskan dialects do not dilTer greatly on most points; the degree of ditTerentiation is comparable to that of Romance, Slavic, Semi- tic, or Bantu rather than of Germanic, Indo-Iranian, or Hamitic. Phonetics. A few preliminary remarks on Athabaskan phonetics are required. The Sarcee sound system does not reflect the original Athabaskan system as well as do those of Chipewyan, Hupa, and Navaho. There have not only been important shifts of vowels and consonants, but levelings of certain consonants (e. g. the old ^-sounds and r-sounds ^ have combined into a single i-series, the Sarcee c-series being, in the main, of anterior 1. Op.cit., vol. I, pp. 13, 14. 2. In my comparative Athabaskan researclics I have used chiefly the following dialects and authorities : Chipewyan (Petitot, Logoff, Goddard); Hare (Petitot) Loucheux (Petitot); Beaver (Goddard); Sarcee ((ioddard, Sapir); Kulchin (Sapir) Ten'a of central Alaska (Jetle^ ; Ten'a of Anvik, hjwer Yukon ((-liapnian, Sapir) Carrier (Morice) ; Hupa (Goddard) ; Kato (Goddard) ; Chasta Costa (Sapir) ; Navaho (Franciscan Fathers) ; and Jicarilla Apache and Apache proper (Goddard). 3. My c is Lepsius' s. 174 yf American Indian Languages 2 190 palatal origin ; Athabaskan y and y are distinguished only before origi- nal a) and vowels (e. g. old i and i, the pepet vowel, have become leveled to i) ; weakenings or- disappearances of certain syllabically final consonants (e. g. -n drops if preceded by a short vowel, which was pro- bably nasalized at one time ; -g and -d are weakened to -', the glottal stop), with resulting interchange of lost, or weakened, and retained con- sonant (e g. Ath. *-en : *en-C> Sarcee-i : -in-a ; Alh.* -e^: *-td-e^ Sarcee -a': -ad-i) ; and contraction of vowels (e. g. Ath. -ai-, -e'e-, -e'a-, -fe-, -ee- may all now appear in Sarcee as -a'"-, see below). Once the phonetic changes peculiar to Sarcee have been worked out, however, it is seen that this dialect is a perfectly regular, if somewhat disturbed, represen tative of its prototype. The Sarcee sounds are : I. Vowels : a. Short: a (a), fl, i, u {o) b. Long (or half-long) : a' , a' , i", o' {w) c. Over-long (with glide-like rearticulations) : a-", a-', '.••■, o-" («•«) d. Diphthongs ( " to cease ", which, however, must not be slurred over). Variations of stress do not seem to be coordinated with differences of tone. Sarcee syllables are of five types : a. Open syllables ending in a short vowel (e. g. wi, t'a). h. Closed syllables with short vowel followed by a voiceless consonant or, far more rarely, by two voiceless consonants (e. g. nd, Va\ the only finals allowed are -t\ -is', -tc' , -', -s, -c, -x,-\ -i, -st\ -ctc\ -lie' \ more rarely -k\ -t's, and -ft). c. Heavy syllables w^ith long, or half-long, vowel followed by a voiced consonant (e. g. ni'l, t''a'^\ the only finals allowed are -:(, -/, -/, -« ; historically speaking, syllables in -r [', *-^y]> ^" -0' (£'Js^a'' jjiven above is to be read k' xg' \.f s'' a'' {-g' zzz -kg-, i. e. unaspirated yoiceless ^releasingin intermediate g). We shall not indicate these purely mechanical gemina- tions. A final glottal stop, whether an old Athabaskan element or of Sar- cee origin, is absorbed in the following geminated consonant, e. g. a"'sdiy'.nC " he has said to him" -j- inferential particle la becomes a'^sd'-j'-U'd'a. There are other important rules of sandhi which we must neglect here. Tone The tone system followed in this paper is that used by Father Schmidt in " Anlhropos " and by Westermann and other Africanists. In this sys- tem a indicates a high tone, a a low tone, and a a tone of intermediate pitch. The intlections are indicated by combinations of these symbols : a represents a tone falling from high to low, a falling from high to middle, a falling from middle to low, a rising from low to high, a rising from low to middle, a rising from middle to high. In writing Sarcee I have found it convenient to leave the middle tone unmarked [a = Sar. m-), which is used in certain forms, is itself variable in tone (e. g. amid " you do ", nits\' " you are crying"). In sit^i the si- is just low enough to bring oiit the inherent high tone of the stem -t'i. Just what tone a variable syllable takes cannot be foretold on the basis of simple rules ; the various syllables of this type do not all fol- low identical patterns. The tones so far illustrated belong, then, to three categ^ories : inherent or fixed tones, tones controlled by definite grammatical requirements, and tones controlled by contrastive requirements. It is naturally impossible to draw a hard and fast line between the last two categories. All, or nearly all, primary syllables have a level tone. This applies to syllables with a long vowel as well as to short-voweled syllables; e. g., in forms like n'lts^i' " you are crying " or gdisyc:{ " they have run olT " OT wiictc^o' " leg-big " or yi'an " it has lain " the long vowel of -ts^c, -)'i';(, ~tc'd\ -a'n must be pronounced on a fixed register — middle, high, or low — without the slightest upward or dow^nward inflection. The numerous inflected tones ^ of the language arise from the contraction of two, rarely three, vowels of different registers. The theory of these inflected tones is simple enough. The vowels melt into a single long (over-long) vowel without losing their tones, which tlius combine into inflections whose tendency and range is determined by their component elements. Hence, a -\- a becomes a' (half-falling from high level), a -\- {. Yet far less nnmerous Ihan in Chinese or in certain other American lang-nages, e. g. Takelma, in which we have primary rising and falling inflections. The quick change from level lo level, with comparatively few slurs, is highly characteristic of the general How of Sarcee speech. Six: Athabaskan and Na-Dene Languages 179 195 a becomes d' (fuU-fallingrfrom high level), a -\- a becomes a' ( half- falling- from middle level), a -\- a becomes a' (half-rising from middle level), a -\- d becomes a' (full-rising from lo^v level, a -\- a becomes a' (half- rising from low level) '. If the contracting vowels have the same tone there results a long (over-long) vowel of level tone ((i", a' ", or a"); long level-toned vowels of the same phonetic type result also from vowel -|- /. Only a few examples are needed to illustrate these tonal rules : /Wi isfscx, blanket I-wear >> t'sidi'sfsix^ k^d adi moccasin without, barefoot>» k^a'di nd-\-c-atc again-it (indef.)-I-keep-handling-one-object, I handle a certain object time and again > nd'cqlc Si-l^una my sickness>> sd':(i'ma da'" Ictc'o here I-seize-it>> da'ctc^o gu-d-s-tid thus-I-do> giva'stid \-nl-i-n'.-i it (indef.) -he-admires-the, the one v^ho admires^ Inl'nl: k^u dstid fire(wood) I-do, I make a rire>> k'wa'stid (note disappear- ance of -' in crasis) n'.sgdywlsl' dnd downward he-becomes, he stoops^ nisgd^(iu\sa"nd mit'uwa iss'd his-water I shall-heat > mit'-iava-ss'd t'd:(d-d-yi-s-\ grief-thus-him-treats, he is in mourning>- f'^i^a'^yu't t's4-i-s-sl noise-he-is-sounding, crow>> fsd'ssi t^u-^{w\-\-a water-in-it (indef.) -he-handles-one-object, he gets water The only certain example of an inflected tone in Sarcee that I know of which cannot be analyzed into two simple tones is the low falling tone, -a'-, of the first person plural subjective, e. g. gumca'Cd'j " we suckle them", nxsa'io' " we get moist ". Full historical knowledge would proba- bly indicate that this element is contracted from two simpler elements of middle tone and low tone respectively. It goes without saying that the high, middle, and low levels have no absolute fixation. Nor are they definitely fixed relatively to each other. The interval between high and middle or between middle and low or between high and low varies widely according to the care or emphasis of articulation. What under some circumstances constitutes a movement from high to low may under others be equivalent to no more than a half fall. 1. These formulae do nol refer to vocalic quality. The quality of the contracted vowel is dependent on a number of factors, which do not concern us here. 2. As several of these examples show, the rules of contraction apply not only within the " word " (a somewhat elusive concept in Athabaskan) but in sandhi between words (" crasis "). 180 VI American Indian Languages 2 196 Each level and each inflection must be judged or intuitively felt in the context of the preceding and following tones. Now one of the most cha- racteristic and also one of the most puzzling things about the Sarcee tone systtMU in practice is the tendency, within a phonetic phrase or breath group, to gradually lower the absolute register. This means that a high tone following on a low or middle tone that is itself preceded by a high tone tends to be a little lower than the first high tone and to be perceived by the ear as a middle tone. Many of the ''middle " tones that I have recorded in running text or in specimen sentences or even within the single word are really pseudo-middle tones — " dropped high tones ", as we might call them. I feel that the}' are not normally identical with true middle tones but keep above the normal middle register. The form dn\ld " you do " already quoted was often recorded as dnda, with a secondary middle tone, instead of the theoretically required high tone, on the last syllable, yet, unless my ear deceives me, the tone cadence of this word is not identical with that of words like dit'ibdja' " it is flabby ", diszuuga' '' it is velvety ", nUc^it'ca' " it is tiny ", in which the diminut- ive -a' falls regularly on the middle tone (cf. , with another tone cadence, nidot'ia' " it is light in weight ", S'.sowa' " it is a little sour "). In other words, the " dropped high " really falls between the high and the true middle level. Where there is an immediately adjacent middle tone to serve as a standard of comparison, this difference can frequently be direct- ly perceived. In the phonetic group ayd t^unihi ^' the one who wrings out water " the cadence of the second word is not adequately suggested by the orthography. In making the record I noted that the group t^umhi was really t'linihi on a high register (i. e., not equivalent in cadence to forms like dnlla with true low -n\). In fiimhi, in other w^ords, the -m is a true middle and the -hi a " dropped high ". The analysis is not difficult here, because we have an immediately preceding -|*a with a clearly mark- ed low tone to serve as gauge ; moreover, we have the analogy of other related forms in middle-tone -m and of hundreds of analogous relative (or participial) forms in -L A good example of the progressive fall in register is the sentence k'iyldji. ytca '* coat in-I-go, I put on a coat ", in which i indicates a lower tone than the normal low tone fixed by the preceding wnrd. The theore- tically correct form of the sentence is k\yidji y\cd, but actually no two tones are quite alike in it. Neither of the " middle " tones as the sen- tence was recorded {-dji and -c'(i) is a true middle, but merely a " dropp- ed high ", the second a trifle lower than the first. Thus, the three high tones of the sentence are graduated downward, and the two low tones are also graduated downward in a parallel curve. Functional register Six: Athabaskan and Na-Dene Languages 181 197 chang-es are compounded, in other words, with a rhetorical register change. Just as there is a " dropped high " tone, there ia also not rarely a " dropp- ed middle " (or secondary low) tone, further a " raised middle " or secondary high tone. These niceties can hardly be properly studied with the unaided ear. They require the assistance of mechanical devices^. In any event, how^ever, it Avould be a great mistake to lay too much stress on variability of tone in Sarcee. It is probably no more disturbing a factor in the understanding of the basic tone theory of the language than in Chinese or any other tone language, each of which has its special laws or tendencies of tone sandhi. Tone differences in otherwise identical words. We shall devote the rest of this paper to a brief analysis of a number of cases of forms which dilfer only in tonal respects. In this way we shall gain a cumulative idea of the importance of tone in the study of Sarcee structure. In many cases the forms in question originally dilTered also in other respects and fell together completely in the consonantal and vocalic framework because of the operation of various phonetic laws. Thus tone was left as the last reflex of outward distinction between such forms. One must beware of makinsr the error, which is so commonlv made with Chinese and other tone languages that possess numerous near- homonyms, of supposing that the tone differences arose for the purpose oi keeping apart words which would otherwise be indistinguishable. As a matter of fact, perfectly analogous tone distinctions prevail in groups of non-homonymous forms, w^here tone is not " necessary " as a gran^mati- cal device. F'urthermore, there are many examples in Sarcee of true homonyms, of two or more w^ords which belong to distinct form cate- gories but are identical in all phonetic respects, tone included. 1 . a. -^a'^dd'l he has eaten the berries b. -^a'dd'l we have eaten the berries Form a. goes back to *^alddl, in Athabaskan terms *ys.-l-di'l, '' in 1. I hope to secu e phonograpliic samples of ottier Athabaskan dialects, so as to be able to work out some of the more elusive tone problems -with the help of strictly objective material. [Since this article was written such material has been secured from a Kulchin informant]. 2. These Athabaskan reconstructions, it is hardly necessary to say, do not imply that we are certain that precisely such forms were in early use, but merely that, if they existed, tiiey probably had such a phonetic form. The reconstructions are practically Athabaskan formulae for the dialectic forms. 182 y^f American Indian Languages 2 198 which *)'£- is contracted from *yc-ye'. Form b. g-oes back to *'(a'ldd'l, in Ath. terms *-(-€'-l-di'l (perhaps contracted from *'^e-\-l-di'l). -l-dd'l is perfective of imperfective -l-dal " to eat berries". In form a. *^a- /, d and / become t!, g and k become k.'\ while all surd and exploded consonants become sonant to express the augmentative (i. e. p and pf become b, t and /.'' become d, k and k! become g, q and q! become ^); in the case of the velar consonants a possible change to the "fortis" to denote the diminutive is attended also by a more forward, i. e. palatal, articulation (i. e. g and q become, not q!, but k!). More- over, the sibilant consonants c, tc, and tcf on the one hand, and s, ts, and tsf on the other, are related to each other as augmentative and diminutive consonants, while dj may sometimes, though rarely, be employed as the augmentative grade of tc and ts (e. g., idjik-) djik ' big wagon,' from itsiktsik ' wagon.' The guttural spirant X becomes x in the diminutive form. Subjoined are a few illus- trations for the purpose of making the process more easily under- stood. The normal word in Wishram for ' hip-joints ' is ck.'dlkal. The diminutive of this word is skfdlkal, in which, it will be noticed, the c of the first word has been changed to s in consonance with our rule. The word skldlkal would be appropriately used to designate a baby's hip-joints, for instance. On the other hand the augmentative would require a change of the fortis k! to a sonant g — hence cgdlkal is used to denote ' big hip-joints,' as of a giant. Similarly, while a^.^^a*/ with velar fortis (^/) is the normal word for 'knee,' akliixt with palatal fortis (yk!) and guttural spirant pronounced farther front i^x) is the diminutive, and aodxt^\\\\ sonant velar (^) the augmentative. Not infrequently there is a slight change of meaning accompanying the phonetic change. Thus, while itdVndn (masc.) denotes ' eagle,' itts.'V non (neut.) with diminutive consonantism means ' bird '; ik.'dlaniat denotes ' stone,' but igdlainat with augmentative consonantism means ' rock.' It 236 Vf American Indian Languages 2 538 must not be supposed that this characteristic consonant-gradation is confined to the noun ; it is found just as well in every other part of speech. An example of its use in the verb will serve to give an idea of its rhetorical possibilities. InigEltcim is the normal word for ' I struck him with it.' If the verb-stem -tcim appears, with diminutive consonantism, as -ishn, it implies that the person struck is small ; if the verbal prefix -gEl-, which implies in this case intent to hit, is pronounced -k/sl-, the implication is that the missile used is a small one. Hence we have four forms : inigAlicim ' I hit him with it,' inigHltsivi ' I hit him (a child perhaps) with it,' inik.'Eltcim ' I hit him with it (something small),' and inik.'kltsim ' I hit him (a child) with it (something small).' It would seem then necessary, so far as Chinookan grammar is concerned, to allow as a regular grammatical process, alongside of reduplication, vowel change or "ablaut," and pre-, in-, and suffixation, a fourth process — consonant-gradation or " ablaut." Turning again to morphology, there was one feature which was well calculated to arouse a certain degree of surprise. The work which had been done on Lower Chinook disclosed a paucity of tenses that is, on the whole, quite in accordance with the general morphologic character of many American linguistic stocks. In VVishram, however, I found that it was necessary to distinguish carefully six tenses : ist, a tense characterized by the prefix ga- (before consonants) or ga/- (before vowels) in certain cases option- ally by the prefixed consonant w-), which refers to time long past, say more than one year ago, and which is used regularly in the re- cital of myths ; 2d, a tense characterized by the prefix ;/z- (before consonants) or fiig- (before vowels), used to refer somewhat indefi- nitely to time past and which is used in speaking of events that happened say less than a year ago, yet more than a couple of days ; 3d, a tense characterized by prefixed nn- (before consonants) or tia/- (before vowels) and sufifixed -a, which seems to refer to recent time exclusive of to-day, more specifically to yesterday ; 4th, a tense characterized by prefixed t- (before consonants) or ig- (before vowels), which refers to an action already performed to-day ; 5th, a tense characterized normally by suffixed -/", referring to an action now going on but, as it seems, with the implication of its soon being Seven: Penutian Languages 237 539 completed ; and 6th, a future tense, normally characterized by pre- fixed a- (before consonants) or al- (before vowels) and suffixed -a} Besides this series of six positively characterized tenses, I should not omit to mention that some verbs, when referring to present time, are morphologically tenseless, and seem to form their immedi- ate past tense by a verbal prefix -/- which ordinarily denotes action toward the speaker.^ In this connection I may also mention a group of verb-forms which are characterized by the consonant / (assimilated in nasal surroundings to «) suffixed or infixed to the verb-stem, sometimes by -lal (or -7ian) suffixed to the verb-stem. These forms denote frequentative or continuative action and, as a rule, do not allow the verb to be further characterized by a tense element. They may then, from a certain point of view, be considered as forming a seventh tense — the present tense with no implication of comple- tion.^ The most interesting point about these /-frequentatives is ^Examples — Tbnse: ^a-FoRM ni-FoRM «a-FoRM f gay^y^ J 'he went ' ' «/yuya f wayuya nalnya. 1 galaya. \_ 'she went ' nigiiya. _^rttcigElkEl Wi'tcigitkEl W(7tcigElgEla ' he saw him' galiwi-x. m'gixaix «ff/ixuxwa ' he became' ^^'flcgnux 'they two did to me' wicgnatx wrtcgniixwa /-Form Pres. /-Form FuT. rt-FoRM i iyiiya. fyiii/ \ui/ f ayuya \ a/uya ?tcigEikEl (tssiklElutk/) atcigfelgElrt r^ixux /cgnux 'he looks at him fl/ixiixwa flcgnuxwrt cgnux/ ^Thus uxt (=a -\-tt -{- xt) means 'she is seated,' but 'she was sitting' is rendered by dtxt, in which the prefix -«- has been changed to -t-. Cf. , for this interchange, ^wgwat 'they fly (away from me) ' and k k! (g) ^, q k!, (g, k) (^) qx kx (^) q! k!, (kx) (qx) p! (p!) h t! (t!) d k! (k!) g c s, ts (c) tc ts (tc), (?)dj tc! ts! (tc!), dj s (s) c 244 VI American Indian Languages 2 Normal ts tsf X X Diminutive (ts), ts! (ts!) X (X) Augmentative tc, dj td, (?)dj (X) (?)x On the whole, there is a distinct tendency to have all the consonants of a word bear a consistent diminutive or augmentative coloring, though absolute con- cord in this regard is by no means always observed. In general it may be said that c and s sounds are most easily varied in accordance with our rule. Final non-affricative stops seem incapable of change. It often happens that the nor- mal form of a word is itself partly diminutive in form owing to its meaning; in such cases the form may be still further "diminutivized" if it is desired to give the word a more than ordinarily diminutive force. Thus -k!ac- in il-k!a'c-kac 'child' is evidently a semi-diminutive form of the stem-syllable -kac; 'little child, baby' appears in more pronouncedly diminutive form as itk!a'skas (Wishram Texts 176.3). The following table of body-part nouns will serve as a set of examples of diminutive and augmentative forms. The diminutives would naturally refer to the body-parts of a tiny child, the augmentatives to those of an abnormally large being, as a giant. Normal Diminutive Augmentative i-p!a'qxa 'flat-headedness' i-ba'qxa (dim.) i-gE'tc 'nose' (aug.) i-k!E'tc i'i-pc 'foot' i'l-ps [640] i-qxwi't 'leg' i-khwi't a-q!o'xl 'knee' a-k!u'xl a-go'xl a-ms'luqtan 'cheek' a-mE'luk!tan i-mElExtk!u'lamat 'tongue' i-mElExtgu'lamat i-mi'ct 'lips' i-mi'st i-k^cxa't 'mouth' i-k!"sxa't wa'-kcEn 'finger' wa'-ksEn is-qxu's 'eyes (dim.)' ic-qxu'c id-mE'qco 'face-hair' id-mE'kso i-k!wa'yat 'crown of head' i-gwa'yat a-tckE'n 'shoulder' a-tsk!E'n wa-qxa'tc 'breast' wa-kxa'ts^ i-kxa'tc 'tooth' i-k!a'ts i-q!a'qctaq 'head' i-ga'qctaq ic-k!a'lkal 'hip-joints' is-k!a'lkal ic-ga'lkal is-q!wd'gwdst 'jaws (dim.)' is-k!wa'gwast a-mu'q!w>al 'paunch, stomach' a-mu'gwal 1. Cf. wa-q!a'lc 'thorn', dim. Wa-kxa'ts (Wishram Texts, p. 261). Seven: Penutian Languages 245 Examples of other than body- Normal it-q!u'tcu 'bones' i-tcfi'au 'snake' i-tsi'ktsik 'wagon (dim.)' i-cgi'lukc 'wolf (aug.)' da-ga'c 'yellow' i-cga'n 'cedar board' i-kfa'lamat 'stone' a-k!a'munaq 'fir' il-k!a'ckac'c\\M (dim.)' a-t!u'-gagilak 'good, strong woman' [641] part nouns are: Diminutive it-q!utsie'lxlEm 'dog' (literally, eater of small bones) i-ts!i'au is-ts!i'ktsik 'buggy' il-skli'luks 'new-born wolf cub (Wishram Texts 56.30) a-qx-k!i'c 'gold' wa-ska'n 'box' wa'-tsk!un 'cup' Augmentative i-dji'au i-dji'kdjik 'heavy truck' il-k!a'skas i-ga'lamat a-ga'munaq a-du'-gagilak 'strapping big woman' In these lists, (dim.) and (aug.) mean that the words so designated are wholly or partly diminutive or augmentative in consonantism owing to their primary significance. In i-pfa'qxa, for instance, the diminutive notion implied by p! is easily understood if we remember that head-flattening is associated with infancy. In some cases a consonant change involves or is accompanied by a vocalic change; it seems that the change of a to m or £ has in itself more or less diminutive force (cf. wa'-tsk!un from wa-ska'n with ila-k!d'its 'very little' [Wish- ram Texts 176.3] ordinarily -kfaits 'small'). The case of i-cga'n as compared with wa-ska'n and wa'-tsk!un illustrates the fact that the diminutive form of a noun often has a specialized meaning of its own. A few more examples are: Normal i-tc!i'ndn 'eagle' i-tc!i'laq 'cricket' i-q!apca'lwac 'turtle' a-tca'la 'grindstone' Diminutive il-tsfi'non 'bird' i-ts!i'laq 'grasshopper' is-k!a'psalwas 'lock (of door)' a-tsa'la 'file' It will be observed that several nouns on becoming diminutive in form at the same time change to a more suitable gender, masculines often becoming femi- nines (e.g., wa-ska'n), neuters (e.g., ii-sk.'i'luks), or diminutive duals in is- (e.g., is-k!a'psalwas). Most examples of diminutives and augmentatives hith- erto given have been formed from nouns that in themselves have no necessary diminutive or augmentative force. Other examples than those already given of 246 ^f American Indian Languages 2 words with inherent diminutive force, and hence with at least partial dimin- utive consonantism, follow: a-k!u'ksk!iiks 'ankle' a-p!u'xp!ux 'elbow-joint' i-pfu'xc 'cotton-tailed rabbit' a-t!antsa 'crow' (contrast i-cka'lax 'raven') i-skfu'lya 'coyote' (? cf. i-cgilukc 'wolf') a-gu'sgus 'chipmunk' a-p!una'tsEktsEk 'mosquito' (? cf. - bEna 'jump') is-ga'k!aps 'hat' i-k!a'its 'smallness' (contrast -^ail 'bigness') i-k!a'stila 'crab' H-xan '(somebody's) child' i-sklwo'latsintsin 'swallow' wa-tsk!E'nLX 'nit' Particularly instructive as indicating a live feeling for diminutive consonan- tism are such words as a-lik!u'k 'chicken' and a-lap!u's 'cat' borrowed from Chinook jargon (p in -pus would not be consistent [642] with diminutive s). It is perhaps not too far-fetched to recognize augmentative consonantism in the fol- lowing nouns: i-ga'nuk 'beaver' i-gii'nat 'Chinook salmon' (contrast wa-tsu'iha 'blue-black salmon') i-CE'lqcElq 'porcupine' ic-ga'kwal 'eel' i-du'iha 'buffalo' ic-kcku'ct 'testicles (contrast is-qxu's 'eyes') i-gu'cax 'sky' ic-gwo'lala 'gun' wa'-itc 'tail of mammal' Ic-li'ct 'fish-tail' (contrast is-p!i'ost 'tail of bulb, dried fish') It sometimes happens that a change to diminutive consonantism implies not so much the diminutiveness of the object referred to as a sense of endearment. This seems particularly true in the case of certain terms of relationship: N on- diminutive Diminutive -qcE-n 1 'man's son's -/:.'fl'c-w-c 'paternal grandfather' ga'c-u (vocative) j child' -gak-an 1 'man's daughter's -ga'/c.'-w-c 'maternal grandfather' ga'g-u (vocative) /child' -gi-an 'woman's son's child' -k!i-c 'paternal grandmother' Seven: Penutian Languages 247 Interesting as examples of augmentative consonantism are the names of Coyote's four sons, all of which are derived from words denoting body-parts of the salmon. The augmentative consonantism implies the lubberliness of Coyotes sons. Body-parts of salmon i-k!la'tcin 'salmon-head gristle' i-ksa'lk!uts 'backbone of fish' i-qfwi'nan 'fin' a-k!a'tk'^tgwax 'adipose fin' (? better -qfa'tk'^tgwax) Names of Coyote's sons Sipa'-glatsin 'Big Gristle' (Wishram Texts 66.5) Sipa'-ksalguts 'Big Backbone' (Wishram Texts 66.6) Sapa'-gwinan 'Big Fin' (Wishram Texts 66.7) Sapag-a'thHgwax 'Big Adipose Fin' (Wishram Texts 66.8) As has already been remarked, the noun is not the only part of speech that illustrates the consonantal play here discussed. Adverbs and particle verbs of appropriate meaning sometimes show diminutive consonantism: tsfii'nus 'a lit- tle'; sdk! 'to whistle'; sa'u sau 'to whisper' (contrast Lower Chinook cdu)\ Lower Chinook k!a 'and' may be diminutive to ka. The diminutive form of a particle verb denotes a less intense state of being or activity than its correlative form. Sometimes its meaning is considerably specialized: [643] N on- diminutive tcic 'cold' ma'ca 'to spoil' gut 'to break up (earth) by digging' Possibly also: wax 'to pour out' Iq.'up 'to cut' Diminutive (ts/u'nus) a- its d's 'just (a little) cool' (Wishram Texts 190.15) ma'sa 'to be ashamed' kfu'tkfut 'to pluck' wax 'to set on fire; to bloom' fkfup 'to shoot' The dual in is- is not the only example of a diminutive form of a purely gram- matical element. The diminutive stem -qlwa'lasup 'fast running' occurs with possessive prefixes showing diminutive consonantism. Thus the normal ele- ments -tea- 'her' and -cda- 'of them two' appear as -tsa- and -st!a- in i-tsa-qfwa'l- asup 'she runs fast' (Wishram Texts 66.9) and i-st!a-q!wa'lasup 'they two run fast' (Wishram Texts 66.13). Similarly, in a song (Wishram Texts 94.23), where the reference is to is-p!i'ast 'tail of bulb', a noun of diminutive form, the pro- nominal element cd- and the post-positive local element -ha appear as -st (? better st!-) and -p!a. Thus: staimapid' giskip'.i'ast 'it-alone-at the-my-tail' Finally the verb may show diminutive consonantism, partly in the stem itself, partly in its local and adverbial prefixes and suffixes, partly and most frequently 248 VI American Indian Languages 2 in its pronominal prefixes. Examples of verb stems in distinctly diminutive form are not exactly common, but certain cases seem clear enough. Thus gaqiulatla'- ulx 'he was tossed up' (Wishram Texts 84.26) and gatciutat!a'mElq 'he swal- lowed him by sucking him in' evidently contain a diminutive form of the verb stem -lada- 'to throw away'; ^//w'^^wajc 'it trembles' (Wishram Texts 116.10) and gasi'ximk!na-uk"atsk 'he looked around' (Wishram Texts 30.6) show dimin- utive consonantism both in their stems {-skw- and -k!na-u-) and in their first incorporated pronominal objects (dual s-), the latter verb also in its adverbial suffix -tsk, doubtless the diminutive form of -tck 'up from position of rest'; gats(s)altsgi'ma 'he laid her belly up' (Wishram Texts 56.27) shows diminutive consonantism in both stem (-tsgi) and incorporated pronominal subject (-ts-) and first object (dual -s-). We have already given -tsk as an example of a derivative suffix with dimin- utive consonantism. Other such suffixes are -p!a 'slightly out (of position)' (from -ba 'out') in ayulapfa'tcguxwida 'it will tilt up', literally 'it will spon- taneously move out up from its sitting [644] position' (Wishram Texts 184.10) and tsu (from -tcu 'down') in ililu'stsu '(water) moved down into the (hollow place)'. As examples of diminutive forms of local prefixes may be given -klEl- (from -gEl- 'directed toward') in ga-tssi'k!Elutk 'he looked at him' and its reflex- ive correlative -xeI- (from -xeI) in gasi'xElutk 'he looked'; -sklEm- 'under' in iniasklEmla'datcu 'I threw it down under her' is doubtless diminutive to -gEtn- 'next to' (cf. -tcu and -s-tsu above). The only examples of diminutive consonantism in the pronominal prefixes of verb forms occur in the case of ts (for tc, third person masculine subject tran- sitive) and s (for c, third person dual subject intransitive and transitive and object transitive). Whenever the object of the transitive verb (or the apparent subject, really first object, of the "half-transitive" verb) is diminutive in form, the pronominal prefixes tc and c appear as ts and s; the ts by no means implies the diminutive character of the transitive subject. Examples are: I'wi gatssu'x isie'nqxoq 'he looked at his fish-line' (Wishram Texts 140.28), where the incor- porated pronominal dual element -s- of gatssu'x refers to the diminutive dual object is-ie'-nqxoq 'his fish-line', while the pronominal subject -ts- 'he' agrees with the object in diminutive consonantism; gatksu'klam {-Iks- always appears for -skt-) 'the two (women) came home with the (baby)' (Wishram Texts 2.12), the diminutive dual -s- referring to the grown-up women, not to the baby ; ga^f- ngatklagwa'x gas ktenaklwd 'st 'it-waves-freely-over-me-my-feathered-cloak' (Wishram Texts 142.5), where the first object -s- of the half-transitive verb refers to the diminutive dual noun s-tenakfwd'st '(small) feathered cloak'. Par- ticularly noteworthy in this connection is the idiomatic use of a diminutive dual object -.s- referring to an implied, unexpressed noun of diminutive significance; there need not even exist such a diminutive dual noun to which reference, if desired, could be explicitly made. A good example is: gaksi'lutk 'she cradled him', literally, 'she put the-two-small (objects) down to him', where 'the two small (objects)' refer to an implied word for 'cradle', though the word for 'cra- dle' in actual use is a masculine (i'-lkau). Similarly, verbs of jumping and somer- saulting have an incorporated diminutive dual object -s- referring to 'the two Seven: Penutian Languages 249 small (feet)', though the actual word for 'feet' is plural (i't-pc). Examples are: gaksu'bEna 'she jumped'; gasixmi' Lgwa 'he turned a somersault' (Wishram Texts 82.18); and gats(s)altsgi'ma 'he laid her, belly up'. The [645] most trans- parent example of the use of an incorporated diminutive dual object to refer to an unexpressed but existing noun is afforded by certain verbs of looking, in which the -s- has reference to is-qxu's 'the two eyes'. A frequently occurring example of such a verb is gatssi'klElutk 'he looked at him', literally, 'he put the two small (eyes) down toward him', the -tc- and -gEl- appearing in their dimin- utive forms -ts- and -k!El- to agree with the object -s-; gasiximk!na'-uk"atsk 'he looked around' is another such verb. As a rule, it will have been observed, a verb form tends to be consistently diminutive or non-diminutive in its consonantism. It is at least possible, how- ever, to limit the application of the diminutive idea to some specific element of the action by "diminutivizing" only some corresponding element of the verb form. An example already published elsewhere will again do service here. The normal word for 'I struck him with it' is inigE'ltcim. If the verb stem -tcim appears, with diminutive consonantism, as -tsim, the implication is that the missile used is a small one. Hence we have four forms: inigE'ltcim 'I hit him with it'; inigE'tsim 'I hit him (a child perhaps) with it'; iniklE'ltcim 'I hit him with it (something small)', and iniklE'ltsim 'I hit him (a child) with it (something small)'. To be sure, such examples are very uncommon and the one just given is perhaps little more than a linguistic tour de force. Nevertheless, it shows very clearly how thoroughly alive is the feeling for the significance of consonantal play. Post-positions in Wishram [650] Wishram, differing markedly in this respect from Lower Chinook, makes rather considerable use of a series of post-positive particles [651] defining mate- rial case relations (chiefly local and instrumental). As most such relations can be expressed by means of local and adverbial prefixes and suffixes in the verb, the denominating parts of speech being in apposition to incorporated pro- nominal elements, this use of postpositions must be considered as un- Chinookan in origin; the fact that some of the postpositive particles are phonetically identical with corresponding Sahaptin case suffixes proves the whole process to be borrowed from the neighboring Sahaptin linguistic stock. As a rule, such postpositive particles are used with denominating parts of speech (nouns, pronouns, adjectives), but some of them may also be suffixed to predicating words (verbs, particle verbs); in the latter case the predicate is to be considered as substantivized syntactically, though not morphologically, and is used subordinately to another predicate. Wishram thus utilizes its postposi- tions to some extent in the building up of subordinate clauses. Where a noun or 250 ^' American Indian Languages 2 other denominating part of speech has been already represented in the verb by an incorporated pronominal element, its relation to the verb and to other nouns in the sentence is necessarily already defined, so that no postposition is neces- sary; even here, however, it not infrequently happens that a postposition is pleonastically used (compare such English possibilities as "He entered into the house"). If a noun is modified by a preceding attributive word (demonstrative pronoun, numeral, noun, or adjective), the postposition is used with the modi- fying word. The postpositions, with examples illustrating their uses, are listed in the following paragraphs: 1. -ba (-pa) 'in, at\ With this element should be compared Yakima -pa 'in'. Examples illustrating its use with nouns and pronouns occur with very great frequency, so that only a few need here be given. cikxa'-imdt ci'tfix yak"cxa'tpa 'half of it lies in his mouth' 4.3^ gaklakxa'-ima ilkfa'ckac akni'mba 'she put the child in the canoe' 2.11 atgadi'mama da'uyaba wi'lx 'they will come in this land' 6.17 gayu'yam ixtpd' wilx 'he arrived at one land' 6.28 itcqxE'mEm axqxatcpa 'I am sick in my breast' 12.27 gatci'upmt it.'o'xwatckpa 'he hid it in the bushes' 18.25 gatu'ya yaxka'ba 'he went up to him' 20.10 (one can also say gaiiglu'ya 'he went to him' with local prefix -gEl-) [652] gadiq'.Eltxi'uba icia'gitcba ya'k"cxatpa wamtfu'xiba 'they went out through him at his nostrils, at his mouth, and at his ears' 28.24 gahi'xuni yaga'iipa wi'mal 'it floated in the great river' 48.7 alxu'ya wa'tcktib' itga'qpuks 'let us go on the tops of the grass' 70.26 (literally, 'the-grass-at its-tops') Observe that the first two examples illustrate its pleonastic use; the nouns yak'^cxa't and akni'm have been respectively anticipated in the verb by the pronominal elements -/- and -a-, while their local relation to the verb is defined by the prefix -k- 'on' following these elements, -ba is also used with demonstrative stems to form adverbs of place where: da'ba 'here'; kwo'ba 'there'; ia'xiba 'yonder'. As subordinating element, -ba denotes 'where'; less frequently it indicates cause. It is suffixed either to the verb itself, or, similarly to the case of the modified noun, to an adverb or particle preceding the verb. Examples are: cta'xya I'nadix qla'tsEuba gatccgE'lgElx 'across yonder (were) the two where he had first seen them' 8.10 (literally, 'first-at he-saw-them') galiktcYptck gatcc^ElkE'lxpa 'he came to land where he had seen them' 8.5 e'wi gali'xdx gayaxa'limaixpa 'he looked back to where he had thrown himself into the water' 8.6 ma'sa gali'xox qfu'mba gagi'iix 'he was ashamed because she had disturbed him in his sleep' 58.26 (literally, 'disturb-in-sleep at she-made-him') 2. References are to Wishram Texts. Seven: Penutian Languages 251 2. -/amr (often with palatalized a as -ffl>?i/, -iemt) 'to, from'. This suffix is proba- bly Chinookan in origin; it may be plausibly analyzed as verb stem -/- 'go' + verb suffix -am 'arriving' + tense suffix -t. This analysis would explain its two apparently contradictory meanings. It tends to draw the accent to itself. Examples are: ickte'lgwiptck wimalia'mt 'they collected (driftwood) from the river' 2.2 nigElga'ba iciagitcia'mt 'it flew out of his nostrils' 80.29 (literally, 'out of him from his nostrils') gacx^klwa'x txtdqtia'mt 'the two returned to their house' 2.12 gayukfwi'xa ilaxni'miemt 'he swam to the person's canoe' 18.23 mxa'tcktcam wimatia'mt 'go to the river and wash yourself 22.18 (literally, 'go-and-wash-yourself to-the-river') gatciu'k"{ itq^Udmt ittcqoa 'he took the water to the house' 28.8 As subordinating element it may be translated as 'to where'. An example of its use after verbs is: [653] asEmxElu'tka a'tpxiamd aga'tax 'you shall look towards the east' 188.21 (liter- ally, 'she-comes-out to-where the-sun') 3. ba'ma 'for, belonging to'. This is evidently the Yakima suffix -pama 'for'. Examples of its use with denominating words are: na'ikabam' amtklni' dama itqagi'lak 'for my sake you two will go and get me the woman' 62.25 ya'xtau laxka'bama IgiubttcEma 'that (fish) he obtains for himself 186.4 gaqxd'gwigax itsH'ndnks wi'lxpama 'animals were taken belonging to the country' 16.13 ctmo'kct gactu'ix ntca'ikabama 'two of our men (literally, us-for) went on' 216.16 da'nbama qxe'dau mxu'lal 'what for do you speak thus?' 132.24 tga'tqwdm luwa'n qa'xbabama 'he has come I know not where from' 128.17 (literally, 'what-in belonging-to') k.'a'ya kwo'babama idE'lxam tcduxt 'he had not made people belonging to there' 44.23 gi'gwalbam' itk'.i'tit 'underclothes' (literally, 'below-for clothes') Less commonly bama may precede. An example is: bam ' iiJ'wulx aktugwi" ilk'.a'lamat 'he carries rocks for (i.e. , in order to gain) strength' 186.17 (cf iLxe'wulx bama 188.2) When used at the beginning of a predication, bama gives it the meaning of a clause of purpose. Examples are: ba'ma ta'-itcka a'lEm' atcludi'na 'in order that he might kill them' 54.2 (liter- ally, 'for them will he-will-kill-them') bama capca'p qiuxu'nnit ika'ba 'for chopping up the ice' 188. 19 (literally, 'for chop-up it-is-always-made the-ice') When accented {bama'), it is used after predicates to mean 'ever since'. An example is: 252 ^f American Indian Languages 2 nk'.a'ckacbama' k/d'ya qxantcix itctcgE'ruEin 'ever since I was a child I have never been sick' 190.9 4. (E)nEgi 'with, by means of, less frequently 'made out of. It seems to be the Yakima genitive case ending -ngi. Examples are: axk' E'riEgi amcgiu'xa Iqfo'p 'with it you will cut it off 12.4 Lqfo'p gaigi'ux aqE'tiEkc E'riEgi 'they cut it off with the stone knife' 18.5 gatkld'qi' aiakcE'n EtiEgi 'he counted them with his finger' 18.19 itia'ma ngi gayu'ya 'he went by means of a round-pointed canoe' 38.21 iga'hEnac E'riEgi gatdu'x 'he made them out of young oak' 4.13 [654] Less frequently ngi may precede. Examples are — xa'u xau galxu'x ng' ilkcE'n 'they combed themselves with the hand' 78.10 ayakla'lamat ngi wa'nux 'his pipe (was) made out of a stomach' 94.9 5. a'mEni 'made out of, less frequently 'with'. It is perhaps the Yakima -nmi. Examples are — sd'q" itk.'a'lamat a'mEni akitxax 'it is entirely out of stones' 82.13 isklu'ly' amEni isga'klaps aqsu'xwa 'a hat is made out of coyote' 182.7 alklwa'dit amEni aqiu'xwa 'it is made of tule' 182.9 itq/u'tc' a'mEni tsE'tsEx gaqtu'x itkfa'munaq 'they split trees by means of antlers' 182.14 6. -pt 'up to' is used to form adverbs out of demonstrative stems: dapt 'up to here'; kwopt 'up to there, then, enough'; ya'xpt 'up to yonder'. Probably etymologically identical with this element is -bEt, frequently added to verbs or other words in the predicate to form temporal clauses. Examples are — gatclE'mquit Iqa'wulqt gagiula' dabit 'he spit blood when she threw him down' 14.11 galikta'tckpEt pla'la igi'xox 'when he had come up out of the water, he stopped' 22.18 lE'p(b)Et atxu'xwa anigElgd 'ya 'when he dives, I shall take hold of it' 18.20 nk'.a'ckacbEt 'when I was a boy' 188.8 aga'lax alaxu'xwa yaxtadi'wi gali'xux gaixo'qbEt 'the weather will be as it was when they came together' 130.27 When rhetorically lengthened to -bd't, this post-position has a general cumulative significance; with verbs it is best translated 'as many as'. Examples are — ^wE'nEmabd'd Hgwo'mEx antk.'wa'lalaqwida 'I shall be absent as much as five days' 122.12 kwo'pt natcdup^Enayabd't 'that many (ropes) as he had apportioned' 188.6 qxa'ntcipt alkixa'tgway' atciulxamabd't 'he piles up as many as he tells him to' 186.19 I Seven: Penutian Languages 253 diwi (emphatic dd'wi) 'like'. This element is very likely of demonstrative origin, and so does not perhaps belong here. It is freely used, however, as a post-position, and so may be included. Examples are — ickla'li diwi datdi'p 'striped like a basket' 166.2 iya'tqx ilgwa'Ulx diwi 'his body (was) like a person's' 166.17 naika dd'wi itcE'lgulit 'exactly like my appearance' 104.10 Demonstrative Pronouns and Adverbs in Wishram [625] Near 1st person Near 2nd person Near 3rd person Near 3rd person (formed from ya'xdau) Masculine da'uya(x) da'ya(x) [ya'xdau \ya'xda(x) ya'xia(x) yakd'xdau Feminine da'ua(x) da'wa(x) a'xdau a'xda(x) a'xia(x) akd'xdau Neuter da'uia(x) da'{a(x) la'xdau ia'xda(x) ia'xia(x) {akd'xdau Near 1st person Near 2nd person Near 3d person Dual da'ucda(x) da'cda(x) f cda'xdau \cda'xda(x) cda'xia(x) Near 3d person (formed from ya'xdau) cdakd'xdau Plural da'uda(x) da'da(x) da'xdau da'xda(x) da'xia(x) dakd'xdau Plural, persons da'uda-itc da'da-itc da'(u)la-itc , da'(u)a-itc da'xdauaitc la'xdauaitc a'xdauaitc da'xiaitc la'xiaitc a'xiaitc dakd'xdauaitc hikd'xdauaitc akd'xdauaitc Note. — It is somewhat doubtful whether ya'xdau should be so read or as ya'xdau. (x) in personal and demonstrative pronouns is deictic in value. -ka may be added to demonstratives in -/7c. Elements -tfa and -tHkc are perhaps "diminutive" forms of demonstrative pronominal stem da 'this' and personal plural dike. 254 VI American Indian Languages 2 Following is a list of the demonstrative adverbs of the Wishram dialect: Locative up to towards, on . . . side Stem da(u) da'ha 'here' dapt dabd't 'little ways further on' kwobd't Stem kwo kwo'ba 'there' kwopt iyax da'ba ^^A^y Stem iaxi ia'xiba'yondef ia'xi 'away, off ya'xpt iaxd't 'further on' Stem di di'ka 'here' (pt also in di'gat (18.17) (dika data' 92. \\) qa'n tcipt 'how long?') [626] Stem gi g/>/ (18.17) {i'wat 'to you (place)' i'wa 'thus, there' (106.22) [iwa'tka (158.24) Note. — Compounded with gi are also da'ngi 'something'; qa'tgi 'somehow'; qxa'matgi 'somewhere' (96.11). Related to di'ka and di'gad is perhaps digu'tcix 'perhaps' (96.17); also di'wi 'like'. In -xi we have, besides ya'xi, also (aga) du'xi 'oh, well'! (60.4). Note. — Ya'xa 'indeed ' (also in quct i'axa 'as it turned out'); au (perhaps = aw', a'wa, and related to Chinook _vflwfl) in da'n au ayamlu'da 'what, pray, shall I give you?' (154.6); yaxa'wa 'however'. Note also kwo'bixix 'right there, not very far'. -a'dix forms: a'ngadix 'long ago'; ixtka'dix (192.2); ina'tkadix (192.5). With stem dau: kwo'dau 'and'; da'ukwa 'just as before'; qxi'dau 'thus'. Independent Personal Pronoun These correspond to Wishram forms recorded by Sapir: na-ima 'I alone' Uiinuidikc, dci-iniadikc , dimadikc 'they alone' mci-ima "thou alone' Ixa-imadikc 'we (incl.) alone' [627] Besides these, Doctor Sapir has recorded in Wishram the following: Shortest form: nafx) T ya(x) "he da'-ifc 'they' la'-itc 'they' (Wishram Texts 48.4) a'-itc 'they' inclusive: nai't'.a 'I too' ya'xt'.a 'he too' la'-it'.ikc 'they too' Ixai't'.ikc 'we too' da'-it'.ikc 'they too' a'-it!ikc 'they too' 3. References are to Wishram Texts. Seven: Penutian Languages 255 He remarks that the demonstratives of the third person (ya'xia) seem mor- phologically parallel to first and second personal emphatic pronouns (na'ya); that the demonstrative element -/- is characteristic of the first and second per- sons, -X- of the third; as in na-i-ka T ya-x-ka 'he' na'-i-tfa 'I too' ya-x-t!a 'he too' na'-(i)-ya T ya'-x-ia 'he' These elements -/- and -x- are probably identical with Chinook -/"- and -x-, -x in x-i 'lu and x-ix-, xax. Modal Elements [578-579] 2. ni-. This prefix is confined to the dialects east of the Kathlamet. It takes the form nig- before vowels, like the preceding a-, prefix indicating a change from one state into another, translated 'to become' in intransitives and as an aorist in transitives. It occurs in transitive and intransitive verbs. It expresses a somewhat indefinite time past, and is used in speaking of events that hap- pened less than a year or so ago, yet more than a couple of days ago. ni-y-u'ya 'he went' {ni- 'past'; -y- 'he'; -uya 'to go') nig-u'ya 'she went' (the same before vocalic element; -a- 'she', being con- tracted with -u- into -u) ni-tc-i-gil-kErhe saw him' {ni- 'past'; -tc- 'he'; -/- 'him'; -g// verbal prefix; -knl 'to see') [579] 3. a-. [Prefix indicating the future in intransitive verbs.] In the dialects east of the Kathlamet it is used also with transitive verbs. a-tc-i-gE'l-kEl-a 'he will see her' {a- 'future'; -re- 'he'; -/- 'him'; -^eI- verbal prefix; -kEl 'to see'; -a 'future') 4. ga-, before vowels gal-. This prefix is confined to the dialects cast of the Kathlamet. It expresses time long past, and is always used in the recital of myths ga-y-u'ya 'he went' (see analysis under 2) gal-u'ya 'she went' (see analysis under 2) ga-tc-igE'l-kEl 'he saw him' (see analysis under 2) n- may be used in place of this prefix. 5. na-, before vowels nal-. This prefix is confined to the dialects east of the Kathlamet. It refers to recent time exclusive of today, more specifically to yesterday. Its use is analogous to that of the preceding. 256 VI American Indian Languages 2 WISHRAM TEXT' By Edward Sapir Coyote and Itc!e'xyan 673 Aga' kwo'pt^ gayu'ya* isklu'lya'^ wi'tlax." Na'2wit' gayu'yam;* Now tlu'ii he went Coyote again. Straightway he arrived going; galixE'ltcrna(i® isklu'lya gwa'nisini'*' qtiilatla'iiiElqt" idE'lxam" he heard Coyote always they findef.) are always the people Mwallowing them down 674 itclE'xyan.'^ Qxa'damt" gayu'y"^ ikni'm** na'wit gatcigE'lga" Merman. Whither it went the canoe straightway he got hold of it itclE'xyan; gatciutatla'mElq^'* ka'nawi'® dan.^° "NaitI'" a'g'" Merman; he always swallowed it down every thing. "Me too now atcnuiatla'niElEqEma,"" isklu'lya galixtu'xwa-it." Aga kwo'pt he will swallow me down," Coyote he thought. Now then gayu'y' isklu'lja; gatcigE'lga yag'ail" ikla'munaq.^^ Aga kwo'pt he went Coyote; he got hold of it its bigne.ss the tree. Now then 675 La'x" gali'xox.'* GatcigE'lga itclE'xyan, gaqiulat!a'raElEq." insight he made him- He got hold of merman, they (indef.) swallowed self. him him down. Na'wit ittcqo'ba^" gi'gwaP' isklu'lya galixi'maxitam^'' wi'lxba.*' straightway in the water below Coyote he arrived falling on the ground Aga kwo'pt gatcugi'kfil^* Igabla'd'* idE'lxam; Jgabla'd'^ akni'ni" Now then he saw them their multitude the people; their multitude the canoes axu'xf* kwo'ba^® gi'gwal iltcqo'ba. Aga kwo'pt gatcigE'lkEl*" they are there below in the water. Now then he saw it piled together isklu'lya itclE'xyan yago'mEnii^' qxwoL^'' iki'ax." Aga kwo'pt Coyote Merman his heart hanging it is. Now then 676 gaqiu'lxam"* isklu'lya: "Ya'xdau"'' itclE'x3'an yago'mEnii.'' Aga they (indef.) told Coyote: "That Merman his heart." Now him kwo'pt Lq!6'p"'= gatci'ux;"^ Lqlo'p"*^ gali'xox*^^ itclE'xyan yago'niEnii. then cut he made it; cut it made itself Merman his heart. Aga kwo'pt ka'nawi gatkxEni'yutck" sJi'q"''^ akni'm kwo'dau" Now then all they each floated up out entirely the canoes and of water idE'lxam kwo'dau isklu'lya. the people and Coyote. Aga kwo'pt gali'kim" isklu'lya: "Lga^^ pu" qii'ma^" ma'ima" Now then he said Coyote: " Perchance would how you alone itclE'xyan qxi'dau" amdu'xwa" idE'lxam? Da'uya^^ wi'gwa" aga Merman thus you will do to them the people? This . day now 677 kwo'pt^® qxi'dau amdu'xwa idE'lxam. Na'ika" isklu'lya yamu'lxam.^* that far thus you will do to the people. I Coyote I have told you. them Kwa'ic*" da'uyaba®" wi'lx atgadi'mama" idE'lxam. Kwo'pt Soon in this land they will arrive coming the people. Then alugwagi'ma,®^ 'Qxi'dau ^ex«^ gatci'ux^* isklu'lya itclE'xyan.' they will say, 'Thus exercising he did to him Coyote Merman.' supernatural power Kwo'pt a'ga itclE'xyan pla'l'^^ amxu'xwa."" Then now Merman being quiet you will make yourself." Seven: Penutian Languages 257 1 A connected English translation of this text will he found in Sapir's \\ ishram Texts, Puhlicatlons of 673 the American Ethnological Society, ii, 4 1 , 43. The Indian text as here given has been very slightly normal- ized from its form as there published (pp. 40, 42). « Used partly with weak temporal force, partly as mere connective in narrative. It is frequently prac- tically untranslatable into English. 3 kw6pt, THEN, AT THAT TIME, is regularly used with preceding aga to mark new step In narrative. It can be analyzed into demonstrative stem kwo- (or ktoa-) that (= Chinook go there) and local suffix -pt UP TO (so AND so) FAR. Neither of these elements occurs freely, kwo- is not used to form demon- strative pronouns, only occurring stereotyped in several adverbs; l)esides kwopt we have kwo'ba there (note 39), and kwo'dau and (note 46). -pt also hardly seems to occur except stereotyped in adverbs; cf. dapt AS FAR AS THIS (related to da'ba, this-in=here, as kwopt is to kwo'bc), and yaxpt, as far A3 THAT YONDER, from ya'zi off yonder). See also note 56. * ga- (gal- before vowels) = tense prefix denoting remote past, regularly used In myth narrative. - y-^ 3d per. masc. subj. intr., referring to isk.'u'lya. before consonants it would appear as -i-, while gal- would then appear as tense prefix (ga-y- = gal-i-: see notes 9, 28, 32, 47). -u- = directive prefix away from SPEAKER, -ya = verb stem to go. * i- = masc. noun prefix with which -y- in gayu'ya is in agreement, -sk.'u'lya = noun stem coyote, apparently not capable of analysis; perhaps loan-word from Klickitat spi'lya. Chinook has another stem, -t.'d'lapas. « Composed of wi'tfa again and deictic particle -i: cf. da'uya (note 54) and da'uyaz this. wi'Ha is most plausibly explained as stereotyped adverb from wi-, masc. noun prefix (originally independent masc. pronoun? See notes 19 and 33), and -t!a, emphatic particle added to pronouns, too, also (see note 21). According to this analysis wi't!a(x) was originally formed from *wi as ya'xtfa(i) he too from ya-X' HE. Originally it must have meant that (masc.) too, but was later generalized in meaning. ' Rhetorically lengthened form of nu'it immediately, right away. When thus lengthened to na'wit, it seems to imply direct, unswerving motion without interference of other action; it may then be rendered as STRAIGHT ON or ON AND ON. 8 As in note 4, except that instead of verb stem -ya we have its shorter form -y. -i- (as in yu'it he goes; cf. also note 61). To this is suffixed verb suffix -am arrive while — ing, go (or come) to do . Several verb stems have two forms,— one in -a, and one without this -a (e. g., -pa and -p to go out; cf. gali'pa SHE WENT OUT With atpx SHE COMES OUT). 9 gal- = tense prefix ga- before vowels, -i- = 3d per. masc. obj. before reflexive element (reflexive verbs have, morphologically speaking, no subject). -xEl- = indirect reflexive composed of reflexive element -r- and local verb prefix -I- to, into, -tcmaq = verb stem to hear. galixE'ltcmaq means literally to him- self HEARD. TO HEAR SOME ONE is expressed by -x-tcmaq with prefixed transitive subject and object pronominal elements. 10 Adverb not capable of analysis. " q- = indefinite transitive subj. -t- = 3d per. pi. obj. tr., referring to idE'lxam. -u- =directlve prefix (very many verbs have this "directive" -u- even when no definite idea of direction away from speaker seems to be implied). -tatfaniElq- is exampleof rarely occurring compound verbs. -Za^'a- is "diminutive" form of verb stem -lada- to throw down, away (in this case its meaning seems to correspond somewhat more closely to that of its Chinook cognate -Lata to pull back); -mElq- is best explained as verb stem -mEq. (or -mq-)TO vomit with infixed -I- of frequentative or continuative significance (that -Ms not really part of stem is shown by form itciulatfa'maq he swallowed him down); pull back -i- vomit may be construed as meaning vomit backward, draw to one's self and swallow, -t = tpnse suffix of present time. Observe peculiar sequence of tense, he heard . . . they sn- allow them down. Verbs that are dependent on other verbs, chiefly of saying or perception, are always present in tense, no matter what tense is logically implied; cf. below gatcigE'lkEl . . . iki'az (note 43) he Syi if it ... it /5. '2 id- = 3d per. pi. noun prefix, in concord with -t- in preceding verb, -ham (-E- is inorganic) = noun stem village (wi'liam village is formally masc. sing, of idE'lxam people); -liam is evidently related to -li (see note 33). " j- as in note 5. -tc.'Eiyan = noun stem merman, protector of fishermen (see Wishram Texts, p. 40, note 2; p. 42, note 2; p. 256, note 2); no etymology suggests itself. Syntactically itc.'E'iyan is subject Implied, but not grammatically referred to, by q- of preceding verb. This clause can hardly be considered as quite correct; properly speaking, itc.'E'xyan should go with tclulat.'a'mElqt. 258 Vf American Indian Languages 2 675 •< From interrogative stem qia- (or qa-), seen also in qa'xba what-in? = where? qa'xia of what KIND? and qa'ngi what-with? = how? -damt = local suffix toward found suffixed to several adverbs (cf. ca'ialadamt toward above, gigwa'ladamt toward below). This -damt is evidently related to local noun sufTix -iamt to, from. Qxa'damt here introduces indirect question, and may best be translated as no matter where. li = gayu'ya. Final vowels are regiilarly elided when following word begins with vowel. For analysis of form, see note 4. •« i- as in note 5. -knim = noun stem canoe. This stem can be only secondarily monosyllabic, for otherwise we should have * wiknim (see note 3.3); its Chinook cognate -kanlm shows original dissyllabic form. See also note 37. 1' ga- = tense prefix as in note 4. -tc-.= 3d per. masc. tr. subj., referring to following itc.'E'xyan as sub- ject, -i- = 3d per. masc. tr. obj., referring to ikni'm of preceding clause as object. -gEl- = verb prefix of adverbial force, toward (with purpose, intent to reach); it here replaces directive -u- of most transitive verbs, -ga = verlj stem TO get hold of, seize; it is possibly to be identified with verb stem -ga stick to, its particular active significance being gained by use of transitive pronominal prefixes and verb prefix -gE 1-. 18 ga-tc-i- as in note 17, -i- here referring to following dan. -u-lat.'a'-mElq as in note 11. " ka'nawi all, every is most probably compounded ofkana- all together (found in such niuneral fonns as ka'nactmokct all-the-two = both and, with unexplained -m-, in kanEmlu'nikc all three PEOPLE) and old 3d per. masc. demonstrative pronoun *wi (cf. note 6) now no longer preserved as such (except in such petrified words as wi't.'a and ka'nawi), but specialized, like its corresponding fem. wa-, as 3d per. noun prefix (see note 33). These old pronouns *wi and *wa are best explained as substantivized from pronominal elements-!- (masc.) and -a- (fem.) by means of demonstrative element w- (or «-); this latter element is probably identical witli -u- in demonstrative stem da'u- this (found also as da-; see note 54), and with ( hinook -o- in demonstratives near 3d per. {I'OLa, locta, iota), ka'nawi must origi- nally have meant something like all (of) that (masc), but, like wi't.'a, was later generahzed in signifi- cance, ka'nawi is here, as often, rhetorically lengthened to ka'nawi to emphasize its meaning of totality. ""> Interrogative and indefinite pronoun referring to things, what, anything, something. Though not provided with any sign of gender, it is always construed as masculine, hence -i- in gatciulat.'a'mElq. Its correlative can (Kathlamet ban) referring to persons, who, anybody, somebody, is always neuter in gender; he swallowed everybody down would he gatciulat.'a'mElq ka'nawi can. 21 Elided from na'it.'a (see note 15). Composed of 1st per. sing, pronominal stem nai- (seen also in na'- ika I) and emphatic suffix -t.'a too, also (see note 6). All independent pronouns in -fco can be changed to emphatic pronouns by merely replacing -ka by -t.'a (e. g., ya'ika he becomes ya'it.'a he too). Syntactically na'it.'a here anticipates -n- in following verb (see note 23) as 1st per. sing. obj. M = a'ga (see note 15). This particle is very frequently used before future verb forms in conversation. 23 a- = tense prefix of future time, -tc- = 3d per. masc. tr. subj. -n- = 1st per. sing. tr. obj. -u-lat.'a'- mElEq- as in note 1 1 {-E- before -q- is inorganic). -Em- = connective before future suffix -a: verbs that are continuative or frequentative in form regularly use this connective -Em- before certain suffixes (such as future -a, cessative -tck, usitative -nil), -a = tense suflix of future time; in \V ishram verbs regularly form their future by prefixing a- or al- (before vowels) and suffixing -a. It is somewhat difficult to see why this form should be frequentative; one would rather except atcnulat.'a'mEgwa. 2< gal-i- as in note 9. -x- = reflexive element; literal translation of verb would be (to) himself thought. •lux^w)- = verb stem to think, -a-it = verb suffix of rather uncertain significance here; it is found in all tenses of verb but present, where it is replaced by -an {iilu'iwan he thlnks). 26 ya- = i-ya-. i- = masc. noun prefix, determining gender of noun stem -gail. -ya- = 3d per. masc. possessive pronominal prefix, referring to masculine noun ik.'a'munaq. -gail = abstract noun stem big- ness, yagailik.'a'munaq the tree's bigness may, like all other possessive constructions, be construed either attributively (the big tree) or predicatively (the tree is big). Its attributive character is here determined by presence of true verb (gatcigE'lga) as predicate. •6 J- as in note 5. -k.'a'munaq = noun stem tree, stick, wood. This word is difficult of etymologic analysis, yet can be no simple stem; -k.'a- is undoubtedly to be regarded as noun prefix (cf. ik.'a'lamat ROCK, perhaps from verb stem -la to move), -k.'a- is most plausibly considered as "diminutive" form of verb stem -ga- to fly '- r in air (as first elenient in comi)ound verbs); cf. itciuk!wa'la he whetted it with itci'ula HE filed it, and miugwala' da-ulx i threw it up on top (of something) with iniula'da-uLf I threw it up. 2' Particle verb. Though verbal in force, it is purely adverbial morphologically, having no grammatical form of its own. In regard to tense and person it is defined by following verb, which serves as its form- giving auxiliary. » gal-i- as in note 9. -i- = reflexive clement. -6- (modified from -«- because of preceding and following velar consonants) = directive prefix; ordinarily reflexive -i- replaces directive -u-, but there are several Seven: Penutian Languages 259 676 verbs that retain it even when reflexive in form, -i = verb stem to do, make, -x-u-x to do to one's SELF, MAKE ONE'S SELF, is regularly used to mean become. For other forms of verb stem -i see notes 43, 53, 64, and 66. ^ ga- as in note 4. -q- = indefinite tr. subj. -i- = 3d per. masc. tr. obj. -u-lat/a'-mElEq as in note U. FOTms with indefinite -q- subject are very commonly used in Wishram in lieu of passives. '• il- = 3d per. neut. noun prefi.x. -t- = inorganic consonant, serving as glide between I and c. , except that -ga- = .3d per. fem. possessive pron. prefix (merely homonymous with -ga- of note 35), referring to akni'm. iT a- = 3d per. fem. noun prefix; though many fem. dissyllabic stems have wa- (e. g., wala'la pond), it is here replaced by analogy of iknt'm (see note 16), as in related nouns i- and a-, wi- and wa- generally pair off respectively, -knim as in note 16. logically akni'm canoes is plural, morphologically it is fem., being so referred to in aiu'xt, (note 38); another example of fem. as plural is wa'mwa maggots, masc. wi'mwa maggot. 38 a- = 3d per. fem. intr. subj., referring to akni'm. -i- = verb prefix on ground, on bottom (?) -u- = directive prefix, -xt = verb stem to lie, sit, be placed, corresponding in use to Chinook -c. This verb stem allows of no formal modification by means of tense affixes. 3" Composed of denionstrativestem kwo- (see note 3) and local suffix-6a Csee note 33): that-in = there. ♦0 As in note 34, except that incorporated obj. is -i- = 3d per. masc, referring to yago'mEnil, and that -gEl- is unmodified. i in Kalapuya (p. i8o) is incorrect, as However, further evidence serving to hnk Chinook both with Kalapuya and with Coos- Siuslaw-Alsea will no doubt have been discov- ered by Frachtenberg by the time this paper is published. The greatest surprise was still awaiting me. Tsimshian occupies a peculiarly isolated posi- tion. In its morphological aspects it offers hardly any specific points of resemblance to the neighboring Nadene languages (Haida, Tlingit, Athabaskan). With the Wakashan- Salish-Chimakuan group to the south it shares, e. g., initial reduplication to express distribu- tive and plural ideas and a series of classifying suffixes appended to numeral stems. In its deeper morphological features, however, Tsimshian stands quite apart from these lan- guages. Lexically too, aside from a number of more or less patent loan-words, there seem to be no stems that Tsimshian shares with any of the neighboring languages. A tentative com- parison with the Penutian (or, if one prefers, hypothetically Penutian) languages of Oregon revealed a considerable number of correspon- dences both in the lexical material and in some of the more intimate and fundamental features of the morphology. Should it be possible to demonstrate (and I am fairly sanguine that it can be demonstrated) thai Tsimshian is a de- tached northern offshoot of Penutian, we would be compelled to face a most interesting fact in linguistic differentiation and in the distribu- tion of American tribes. The data for the various assertions I have made in this paper I expect to present /;/ extenso in the future. The fundamental type of Penutian language seems to be a predomi- nantly inflective one, as exemplified say in Miwok, Yokuts and, in more specialized form, in Takelma. The tremendous morphological Takelma possesses no ts ; al-ts-M red (pp. i8o, i8i) should be corrected to al-ts-\il. Seven: Penutian Languages 265 60 divergences exhibited by the various languages here grouped together are due, it v»'ould seem, to two main types of change : the gradual breaking down of inflective forms and the consequent growth of more analytic forms (e. g., in Costanoan and, in a less extreme way, in Coos) ; and the evolution of special synthetic developments, as in Chinook, often, no doubt, under the influence of alien languages, as in Maidu and Tsimshian. It is, of course, entirely premature to group the Penutian languages genetically, but the following provisional scheme may be of some s'ight value as an aid to convenient visualization. A. Californian group [ Miwok \ Costanoan 2. Yokuts 3. Wintun 4. Maidu B. Oregon group 1 . Takelma 2. Oregon Coast group Coos Siuslaw Yakonan 3. Kalapuya C. Chi nook an D. Tsimshian It may well be that Frachtenberg's as yet unpublished data on Kalapuya and Yakonan will force a readjustment of the languages under B ; as for group A, possibly Wintun and Maidu form a closer unit, analogous to that undoubtedly comprised by Miwok and Costa- noan '. I. See A. L. Kroeber, The Chumash and Costanoan Languages, UCP 9: 237-271, particularly 259-263(1910); also The Languages of the Coast of California north of San Francisco (/oc. cit., 191 1, pp. 273-455), pp. 306-508. One of the most striking of the more inti- mate structural peculiarities of Takelma is the presence of a large number of stems of type cv, cv(c2)-, i.e. of stems with repeated vowel. In the verb, this method of stem formation has, to a large extent, taken over a grammatical significance, in some cases characterizing the aorist stem, in others, less numerous, the verb- stem proper. It is, however, frequently illus- trated in the noun^ adjective, and adverb as well ^. Examples are : yana ' acorn viahai large waga- TO BRING (non-aorist) baxam- to come (aorist) herein- KIDNEY o yeuue^ perhaps sehe- TO roast (non-aorist) lehei- to drift dead to land (aorist) yiwin speech -xiivi- TO ROT (non-aorist) ginig- TO GO SOMEWHERE (aotist) nwt'op' SEED-BEATER Olom FORMERLY odo- TO HUNT FOR (nou-aorist) Jop'od- TO RAIN (aorist) Silli'V'k' CRICKET ilyftfs- TO LAUGH (aorist) Turning to Coos, we find that this same type of stem formation, though apparently not pro- vided with formal significance, is very common. It is so very characteristic of the noun, indeed, that Frachtenberg's failure to mention it in his Coos grammar must be set down as a curious oversight. By actual count I find in his Coos 2. For details see The Takelma Language of South- western Oregon (BBAE 40, part 2, 191 2, pp. 1-296), §§40, 86, 109. 3. lam omitting all stress and pitch accents in this paper, also normalizing all orthographies to correspond to the' recommendations of the Phonetic Committee of the American Anthropological Association. 266 VI American Indian Languages 2 6i vocabulary ' no less than 140 examples of such nouns (and adjectives), 35 of verb stems, and 9 of adverbs. These figures show conclusively that we are dealing with a real formative feature of the language, not with an arbitra- rily selected congeries of words. A few of the examples are : yabas maggots yalaq gray (of hair) fiianat crowd Cola't'iS, CRACK (jaiva CHEEK maha- to scare ak'auak- to stick out kayal- to decay qala"- to hurt //rt/a'^'- to bark hakwai as if ehe GONE luebel stomach tsehes alive ^el'eths corner tlpene wings en'ek- to stick out eqe to die (plur.) xeyetl to twist to one side lek shouts luekx- TO stoop kweek to live ehenlc far away Cccd MAT kzvims feather W'.t'.n blood f'./"u: razor-clam hkiuKt feathers i'tud- TO RECOGNIZE h'.tl- TO CARRY hmiS- TO HOLD BACK INFORMATION ^l//-^- TO OVERTAKE IV'.xkm- TO STEAL I. See pp. 191-215 of Coos Texts, CU i, 1913. ydki' LONG AGO tfyW RAINBOW nwytis anus tso'no' BOTH ways It does not seem plausible that these and numerous analogous examples are merely secondary phonetic developments due to pro- gressive or regressive vocalic harmony ^. To at least some extent Coos possesses forms with repeated vowel that are of grammatical sig- nificance, notably certain irregular plurals of nouns and adjectives, e.g. : tSi'y-UX^' SMALL knes HUNCHBACK plur. tsi'yt'-ne kene'yes-e The very irregularity of such forms points to their archaic character. They are obviously analogous to such Takelma adjectival plurals as : -t'gem BLACK -t'gcy-ap-x ROUND ba'Hs i,ONG plur. -fgeme-t'it' -t'geye-p'-it' ba-Has-iV Analogous forms seem to be found in Lower Umpqua (Siuslaw), though apparently not so characteristically. I do not feel as clear about them as yet as about the corresponding Coos formations. Examples are : yafqa-a"' hole hayci'tit WIDOW halca-t LONG pa'iiqa medicine-man taha-mk quiver ca'ya penis t'saxan stomach k\iyak EAGLE qa'xal fish-net XlVil'ka HEAD iazvat- to gamble 2. Sec Frachtenberg's Coos, BBAE 40, 2 : § 7. Seven: Penutian Languages 267 62 amha- to be willing atas ONLY ivi'imtCKKT prs<.p FISH-HAWK misi'-a" ELDER SISTER t\hmS CHILD CriUVSt SUNSET tSimd'z' MUSKRAT tsi'kwi LEGGINGS hmilta MATCHES pmd-t SHARP aiorhin- to lose zui'tsm- TO CAMP mU'l'U BREECH -CLOTH BELT tlxmu'hv BOW-STRING Many more forms of this type, though not as many as in Coos, could be cited, but, on the whole, one gets the feeling that it had ceased to be a live and productive method of stem formation. An apparent tendency to loss of vowels and consequent formation of consonan- tal clusters would seem to have counteracted the persistence of the type. Comparison with Coos seems to confirm this, e.g. : L. Umpqua (Siuslaw) Coos nia''q"li crow tipa'^'n, ipz't'v wings ma'qati t'ipenc In both Coos and Siuslaw, it should be noted, forms with repeated vowels are sometimes related, as in Takelma, to monovocalic forms, e.cr. : Coos manat crowd t'ipenc wings, feathers xalaw-is heat L.U. valqa-a" ■ hole piurcja medicine-man nianta-^- to keep compan\- tlpe wings xahu-is hot yafq-' i-lq- to dig p:iiq- to dance a mtdicinc dance Whether the type of stem with repeated vowel is found also in Alsea (Yakonan) must wait for an answer until Frachtenberg's mate- rial is available. This applies also to Kalapuya. The Kalapuya forms that he gives in his com- parative study number several of the type, e.g. : thanaq fly qalavi silverside salmon qauwan Chinook salmon palam drunk (Chinook Jargon) qoloq swan (Chinook) diigulhu' owl niu'Iukwa cow but, suggestive as such examples are, not much can be done with them when torn apart from their grammatical and statistical context. In Chinook and Tsimshian the type cannot be demonstrated to exist. It is evident from inter- nal evidence and from its numerous and often complex clusters of consonants that vowels have dropped out very frequently in Chinook, so that an originally present type of stem for- mation (c)vc,v(c2)- would be expected, under the destructive influence of phonetic processes, to have disappeared as such. Possibly, however, compar,ative evidence and closer study of Chinookan material may eventually reveal the former existence of the type. Such forms as -kala MAN (cf. also Wishram -gi-kal husband), -ka-holf EEL < *-ka'kwale' (cf. Wishram -gakwal\ -saJa fresh-water clam, /av.'//; six (cf. Wishram tx.^ni), and ca'y.m grizzly bear (cf. plural caya'm-ukc) are, for the present, sugges- tive but too isolated to constitute convincing evidence. As soon as we turn to the Penutian languages of California, we are at once struck by the large number of nouns, adjectives, and verb stems thatareformed on the pattern(c)vc,v(c2)-. Examples from Wintun ' are : I. Taken from S. A. Barrett's vocabulary, pp. 81-87 of The Ethno-oeography of the Povio and jicighhoring Indians (University of California Publications in Ameri- 268 VI American Indian Languages 2 63 apti- mother's brother tahal tongue daka belly anak knee kama bed taral mush-paddle kada deer-snare aka bitter k'afta to sleep hara to go (D) ' elct infant k'me rib keicel house wene medicine ii'nik nose (Southerly dialect) iniil milk tkvi-l rattlesnake hi'ii' shell-beads t'iki sand luini to see (D) ko'so'I lungs poto intestines mrko' arrow no'Io' smoke do'ko' knife tco'nr fish-net cotok pestle SH'iio' nose (Xortherly dialect) pi fill' heart irdii'i rabbit-skin robe tii-nfl magnesite beads tii'nii'k cradle swrii't burden net yu'rw rain ku'pirm fingers ti'tcw- younger sister lu'iiiw dead iinrhii' to sin" can Archaeology and Ethnology, voL 6, 1908, pp. i- 332). I. Forms marked (D) are quoted from R. B. Dixon's Outlines of Wintun Grammar (Putnam Anniversary Volume 1909, pp. 461-476). It is evident at a glance that the type is very much in evidence in Wimun. By actual count I find that considerably more than a third of the nouns and adjectives listed by Barrett for the « Southerly » dialect belong to it. Equally evident is the type in Maidu. From Dixon's grammar ^ I may quote : luaka' meat tsaka- pitch tsa'zva teeth paka sinew ma' tun hip yaha' good ivasa' bad -tala- to crush heme gopher sede blood benek tomorrow /;/;// eye hini net ki'ivi' back ;///;//■ nipples yiini arm pili' dung si mi month tsibi' nails -piuili- to roll -tibil- to wind around U'dhol heavy koJo- to rotate orw- head ko'zuo' armpit yoso' field-mouse kono' baby nolo' pack-basket bono' ear poko' sun koyo' valley WO' no'- to die 2. Maidu, an illustrative Sketch (Handbook of Ameri- can Indian Languages, BBAE : 40, part i, 19 10, pp. 679- 734). Seven: Penutian Languages 269 64 buhl tail butu fur suhu' smoke ku'lw evening viu'su' face It is clear that the type is as much alive in Maidu as in Wintun, and, as in Wintun, it seems, further, to be rather more characteristic of the noun than of the verb. This may be only apparent, however, as our knowledge of the verb morphology of these languages is far from complete. Turning to Yokuts, we find the type of stem formation still very much in evidence. I select from Kroeber's lists '. Examples of nouns are : V'apnd shoulder ma ,ad tears kabad wings tabak deer-rick capan coal wadak head-net k'twz't back tthzty head louse tczhvr, fog bidik faeces dimik prairie falcon tyipin sky idik water cikid arrow ip'i7i ground acorns P'ir,ik nose mi'kJc throat lir,it earth wdho'cit panther iirta youth oty^ hair p:>tJ penis I. See The Yokuts Lau^uas^e of South Central Califoinin, UCP 2 : 165-577(1907). For this paper I use only Yaudanchi forms (see pp. 240-254 for vocabularv). comot lungs ■rpho grizzly bear coyod antelope topo■r^ buckeye tcoxJtc soaproot so'hn tobacco cohd hole odJt ball luocok belt putuc acorn The list might be greatly extended. A large number of verbs also belong to this type, e.g. : awaty to dislike tawac to be thirsty paxat to mourn for tcadax to turn dixid to make a basket pinity to ask pitid to tell pitciiu to catch icitid to pull hair luinis to be ready xityiw n be angry tdxitc ■ be sick i'dik to sing iv:)d:> to dodge od:^y to be on ciitnx to skin dukud to bury dumuk to sweat k'uyuk to scratch nuhuk to kneel hupnc to select Before certain suffixes these stems, like other disyllabic stems, lose their second vowel, e.g. : diikttil (imperative) ) dxkud-ji (past) ) ( duhd-un (present-future) ( duhd-ut (passive) According to Kroeber this stem alternation is purely phonetic, not morphologic, in charac- ter, so that direct comparison with the stri- 270 VI American Indian Languages 2 65 kingly similar stem alternations of Takelma (^e-g. zonst p'eleg- TO go to war : non-aorist p'elg-) is not feasible. However, this matter certainly needs further investigation, particu- larly as not all of Kroeber's forms seem to conform to his phonetic rule (loss of second vowel before a suffix beginning with a vowel). It may well be, on the other hand, that what was in remote origin a purely phonetic alter- nation in Takelma gradually assumed a mor- phological function. There is nothing to in- dicate this at present, however, for Takelma. In Miwok (Moquelumnan) the presence of the type (c)vc,v(c),- is as easy to demonstrate as in Wintun, Maidu, and Yokuts. For examples of nouns and adjectives I quote from Barrett's vocabularies of Coast Miwok ' : nawa old man ata elder brother saka cradle panak red-headed woodpecker patca poison luala Indian potatoe heleke neck ewe milk wene medicine killi antler ki'ti'lak butterfly pi'tci' nails ko'no' bow loko tule olok ocean posol lungs mo'M condor koto grasshopper tso'to'i short pu'tw infant pu'lu'k belly ku'lu'm bone pu'tcw wild onion I . See The Etbno-geography of the Porno and neighboring Indians, UCP 6 : 68-80. u'kw hand tirnu'k cradle su'ku'i sand mu'lu'-ta black n'lu'-ta red hu'hf stinking From Kroeber's material ^ the following verb stems are cited : yar^a to sleep kata to shut ete to see etepo to lie on one's stomach kelpe to swallow nete to count nipito to sit with folded leg hili to pinch oiuo to eat koyok to see toloye to hear lutsu to ascend uhii to drink iinii to come ynnu to kill kusH to sit with stretched leg nuxii to undress uku to enter iitcii to stay yiiku to swing Costanoan, finally, is perfectly analogous in this respect, as in so many others, to Miwok. Examples of nouns and adjectives are again quoted from Kroeber > : ama person wara body patcan blood 2. See his Miwok sketch (pp. 278-3 I9)in The Languages of the Coast of California north of San Francisco, UCP 9 : 273-435 (191 1). The Northern Sierra dialect is quoted. 3. See vocabularies in The Chiimash and Costanoan Languages (UCP 9: 237-271 [1910]). Seven: Penutian Languages 271 6G lawan bow wasar wind lanai neck hatac foot tcara sky wahan snow hesexem old woman £)'Ci beard /c^jyw jackrabbit t^d;'(?u rabbit wetel large j/;n' child kinir fish ^/7m belly wznzf heart liti bow tt/fx/ fish worox hair /co/^o boy o/co ear koro foot, leg rono^ bow moto cloud yokovi ice /o/oi knee co/o^ moccasin to/o fire locko-, laskci- white ^o/^o- black xntii belly kunuc pipe «;/n *i-tcE-m- >> tcE-m- (W. tc-m-') " he (subj.) -thee (obj.) " really indicated " ihey-that (it is) -me ", " he -that (it is) -thee ". It this is correct, the original difference between the intransitive and transitive phrase must have been one of sentence idiom. " He goes " was expressed as " he goes ", but *' he kills her " as " he that (is) (who) kills her "'. This deictic or demonstrative *o-p'- can only be a reduced form of post-accentual -ka, which occurs freely in Chinookan numerals, pro- nouns, and adverbs as deictic element (" only, just "). Examples are W. i'xt-ka " just one ", na'i-ka " I ", a'x-ka " she ", Lower C. nd'm- hi " I alone ", e'-ka " thus ", kaiua't-ka " soon ", nau'it-ka " indeed ", W. iiua't-ka " to yon (place) ". This deictic '-ka, in turn, is obvious- ly merely an enclitic use of an old demonstra- tive stem ka " that " which is no longer in free Chinookan use but which survives in Lower C. ka, c-ka " and " and as petrified temporal ka- " that (time) " in Lower C. ka-iua't-ka " soon " (< " to just that [time] ", parallel in form to W. i-wa'l-ka "to just yon [place] ") and ka-wi'x' " early " {c{. luiix't '' tomorrow ", W. wax "dawning") ; cf. also W. tense pre- fix ga-, ga-l- of remote past time. All these Chinookan elements (ka, ka-, -ka ; ga-; -gE-), finally, are reflexes of a wide-spread demon- strative stem *ka " that ", often used as a gener- al term of reference, found in other Penutian languages (e.g. Coos -kd in xd-ka " he " ; Takelma j^^ " that"; Yokuts/fert " that "[vis.] ; Miwok i-ka " that " ; Tsimshian -_^e absent connective, -ga absent demonstrative added to final noun in sentence). 3. It is fairly clear that the two fundamental factors in the development of the somewhat irregular morphology of Chinookan were a 280 VI American Indian Languages 2 I 10 strong and movable stress accent and, as a result of this, the tendency for vowels to drop out and for originally independent elements to melt together into complex assemblages. Thus the old sentence, which seems to have been constructed on rather simple, analytical, lines, tended more and more to petrify into a highly synthetic sentence-word. We have already hint- ed at the probability that the phonetic change of g to tc antedated the inclusion of certain elements in the verb. Internal evidence makes it practically certain that at least the tense pre- fixes were late in coming into the verb com- plex. In the first place, the tense prefixes of Lower Chinook differ considerably from those of the upper dialects ', so that it looks at though an old set of temporal particles or adverbs (Lower C. a ; «-; W. a, a-l; i, i-g-^ na, na-l ; ni, ni-g ; ga, ga-l \ ;/-) had coalesced with the following pronominal prefixes of the verb in the independent life of the various dia- lects. Moreover, these elements do not behave as though they had ever coalesced into a phonetic group with the early Chinookan forms of the transitive forms for " he" (^iicc-) and " she " (^ags-'). Thus, in Wishram we have forms like i-g-i'-ux " she made him ", in which the tense prefix /- does not palatalize the following -g- to -Ic-, no doubt because it did not enter into the verb complex until long after the palatalizing effect of an /- had spent its force. If the /- had been prefixed at the time that the pronominal element^^- " she "still exis- ted in the fuller form *agj-, it would have re- quired an intervocalic -g~ and the form *ig-ag3'- would have arisen (cf. modern forms like ig-a- -tpa "■ she came out of the house "). Similarly, a form like Lower Chinook atcE tax " he made them "evidently arose before the tense prefix a- was part of the verb complex, for a- could not have palatalized an original *-_^?- to*- to- while the I. Op. cit., pp. 577-79. older pronominal form */to- would have requi- red as tense prefix the prc-vocalic «-, hence *n-iicy-. 4. It is a well known linguistic phenomenon that similar or identical sounds, groupings of sounds into phonetic patterns, or phonetic pro- cesses may characterize a number of indepen- dent languages or even linguistic stocks within a continuous area. Such examples are sugges- tive of phonetic interfluences between distinct languages presumably through the medium of bilinguals. The change oi g or k to dj or tc because of the palataUzing influence of a preceding or following front vowel (/ or e) is perhaps too general a process to warrant our attaching much importance to its occurrence in a number of contiguous languages. Neverthe- less it is of some interest, and, it may be, of historical significance, to point out that the change of ^- sounds or of palatalized k- sounds to Ic- sounds is found in a continuous or nearly continuous area from a northern point on the west coast of Vancouver Island south to the mouth of the Columbia. All the Nootka dia- lects, both NootKa proper and Nitinat-Makah, have altered the original Wakashan anterior palatal Ic- sounds, preserved in Kwakiutl, to corresponding palatal sibilant affricatives; Kwa- kiutlj g' (^'•)and t (^)appear as Nootka tc; t \ (^') as tc' ; and x (.v) as c. A large number of Salish dialects, furthermore, have altered the original unlabialized /.'-sounds to /^-sounds. I am quoting Dr. Boas' personal statement on this point and am unable to give the geograph- ical distribution of the Salish /^-dialects. I should perhaps add that the Lower Chinook k'- Q-) sounds, which correspond to ordinary k- sounds in Wishram, are a comparatively recent dialectic development before /-vowels and that they have nothing whatever to do with the old, general Chinookan, change of 0^ to tc after /-vowels which is the subject of this paper. Seven: Penutian Languages 281 Editorial Note Originally published in International Journal of American Linguistics 4, 105-110(1926). Silverstein (1977), working with richer comparative data, has reaffirmed Sapir s analysis of the Chinookan ergative prefixes, although he posits a basic zero marker, rather than a-, for the third person masculine subject/agent pre- fix. Like Sapir, Silverstein sees "heavy categorial influence from the languages surrounding [Chinookan] on the coast," resulting in a rich elaboration of ergative inflectional morphology out of an earlier "nominative-accusative phra- sal-enclitic syntax" (1977: 154). Sapir's ms. corrigenda on his copy are as follows: Original For: p. 106, col. 1, first table p. 107, col. 2, last para. ta- p. 108, col. 1, 1. 11 -kc- p. 108, col. 1, 1. 22 in animate p. 108, col. 1, fn. 1 See, p. 108, col. 2, 1. 1 age- p. 108, col. 2, 1. 3 charasteristic p. 108, col. 2, 1. 8 whith p. 108, col. 2, 1. 26 q:'8' p. 109, col. 1. 4 gE-L- p. 109, col. 1. 5 -gE-M p. 109, col. 1. 18 to her," p. 109, col. last 1. tachedas p. 110, col. 1. 25 itce- p. 110, col. 2, 1. 29 Kwakiutl, g- Read: Reverse headings "intran- sitive" and "transitive" ta -kc inanimate See ag^- characteristic with -GE-L- -GE-M- to her" tached as /7C9- Kwakiutl g- THE RELATIONSHIP OF MIXE TO THE PENUTIAN FAMILY ^ By L. S. Freeland. Morphological THE structure of Mixe is rather bare and scanty, at least when compared with the rich and fairly intricate grammatical patterns found in some of the Penutian tongues of California and Oregon. One gets the im- pression that in Mixe the morphology has worn thin. 2 There is a strong tendency to fall back on word-order to express relational ideas. New morphemas appear which are but one step removed from concrete seman- temas. In some cases these "empty words" still retain their full concrete meaning in other connections, as for instance the word mid which is the verb "to have". We find it in Mixe used to express the concept "with, of" very much in the same manner as the "belong" of Beche-de-mer. In Miwok or in Maidu this would be expressed by a comita- tive suffix. ^ The Mixe material for this study was obtained by J . de Angulo in Oaxaca during the course of a linguistic survey of that region for the Department of Anthropology of Mexico under the Direction of Dr. Manuel Gamio. After writing this article I sent it to Dr. Edward Sapir for criticism and suggestions. Dr. Sapir was kind enough to send me a Ust of further cognates from Takelma, Coos, and other languages from the northern Penutian group. Dr. Sapir was also kind enough to write some marginal notes which I have taken the liberty to reproduce as footnotes with the initials K. S. In his communication to me Dr. Sapir says that Prof. Roland B. Dixon had already been struck by the Penu- tian character of Zoque, and had sent him a list of Penutian cognates in this language. Zoque is very close to Mixe. The credit for the discovery of a Penu- tian language in southern Mexico belongs therefore to Dr. Dixon, by priority. ^ Cf. analytic wearing down in Mukne and Coos as contrasted with Miwok and Takelma. — E. S. One somehow gets the impression that the Penutian morphology having reached too great a degree of intricacy in its usual methods for expressing the relational, has thrown the whole baggage overboard and started on a new tack. This may be due in some measure to the influence of other Central American tongues which seems to be in the direction of sparseness of linguistic expression. Nevertheless it is true that a very similar drift is at work in some places in California, for instance in the western dialects of Miwok as compared with the eastern dialects.^ Whatever may be the interpretation, the fact of the morphological sparseness of Mixe remains, and naturally precludes extensive morphological comparisons. The evidence for classing Mixe in the Penutian family must therefore necessarily^ be largely lexical. But in spite of its limited grammatical apparatus Mixe possesses morphological traits that have a strong Penutian flavour. We will consider these briefly: (l) INTERNAL MODIFICATIONS OF THE RADICAL. As in Maidu, in Miwok and in Yokuts, so in Mixe the radical often appears in several forms so closely allied that it is sometimes difficult to say which is the primary one. ^ This is just what happened in Chinookan, where older Penutian features are only survivals and a new autonomous and rather complex morphology was built on top of the older system by means of coalescences of sequences. — E. S. 284 VI American Indian Languages 2 2q In Maidu these symbolic changes express derivational concepts, in Miwok and Yokuts temporal aspects, in Mixe differences of person.^ Examples (a hyphen indicates the omission of irrelevant affixes) : Maidu : a) tot to drag ; tttt to touch ; tat to stroke ; tet to reach for. b) tea to eat grass; tco to eat meat. c) yopom-ia to throw; yapam to slap. d) witcap to tear; witcep to tear a little; witcohit to wring; witcup to tear off. e) ivitcot to skin a hare or small animal; iviicut to scrape hides; witcat to split with the hand; ivitcet to split a small thuig. Miwok: 2 to play to die to smoke to run Yokuts : cadik, cadak to awake; tcadxin tcadax to turn; ipe, epi to get water; hiwet, hcut to walk, go, move; tawidc, taudj to overcome ; waid, waadi to break up ; diikdu, dukud to bury; tui, toy to shoot. iESENT TiME Removed Freque FROM Present. TATIVE awi-n- awin'- dwni tcdmci- tcam'ic- tcdnric- hdksi- hakis- hdk'is- huwd't- huwdt-- huwta Mixe: to get up to fall to like to sleep I pedelk-as ri'gunac-els n-zoik-bes may thou pedetk m-gunac-el n-z6ik-he may he pedtg kiivac-el coik-be mey we pedelk-matc n-gunac-edom dsok-medj mau-medj ye pedtk-tetc ni-gunac-tel m-dsok-taba mau-detc they pedik-nedj kunac-tel tsok-teba mey-dj (2) THE INCORPORATED PRONOUN. Yokuts and Mukne (Costanoan) are appar- ently lacking in incorporated pronouns. Southern Maidu uses independent forms in the ordinary statement, but possesses v^hat seems to be a depleted series of suffixed pronouns (possibly old objective forms), used Examples : in subordinated expressions of time and in hortatory and optative modes. Miwok is rich in pronominal series, specialized in their use for present, past, future, subjunctive and the Hke. Although the}'' are much less numerous the Mixe forms come closest to those of Miwok. 3 Maidu Present Miwok Past and Future Mixe I -5, -se -m -/ -is n- thou -ne, -no, -men -s -n — m- he -i — — — y-, IV we -es, -nes, -hese -mas -m -ma, -mats n- ye -maw, -mem -tos -ton -te m- they -i -P -k -te y-, w Both Miwok and Mixe show a considerable amount of fusion and obliteration of elements ^ Cf. internal vocalic changes of verb themes in Takelma, Coos, Lower Umpqua. — E. S. - These alternations are amazingly like the Takelma in putting together their verb forms. This gives often quite a similar impression: ones, e. g. ydan- "to go": fut. yana-. — E. S. ^ Takelma like Miwok and Mixe. — E. S. Seven: Penutian Languages 285 30 Miwok Mixe I have eaten i'wicak I eat hi' kcebes thou ■i'wicas hi'kceb he i'wica heykc we i' w'icakmas hi'kcemets you i'wicak tos hi'kcteh they i'wicap hi'kcteb (3) THE VERBAL PREFIX. Rather a large number of the Mixe verbs in the list are bi-syllabic, and many can be analyzed into two elements. They suggest very strongly the Maidu verbs with prefixes or first-position-stems such as he- action with the foot, ka- action with the hand, and the like. Mixe probably has nothing like the number of these prefixes that there are in Maidu, but three or four are quite clearly discernible and very suggestive: Maidu: heye creep; bedokoi run; bedoi kick; helom step over; hewet hurry, kadokok push away; kapadum roll up; katut touch; katcaduk grab; pai foot, trail; -pat after. Mixe: petc-h to climb; pegig to run; pedig to rise; pasl-h to go out; pa- along, by; pa-otsm to go along; pa-witsm to creep; pa-nas to pass by {nas "the ground") ; pa-boy to chase; pa-son to follow; pa-wep to kick [wep to strike) . kuhath to pay ; kudigx to put ; kutuk to command; kuyat to play; kunas to fall {nas ground); ni-pan to cover; ni-kes-b to be on top of something high [kos, kes above, on, upon); ni-wits to close; neksm to go, ni-neksm to go for the benefit of some one else; kutuk to order, ni-kutuk to order through the agency of another; kes-wits to place; kes-ets to place to boil {ets-b to boil); niwis-dud to open {niwits to close) ; nipan-dud to uncover [nipan to cover) ; rswits to take off [kes-wits to place, ni-wits to close) ; rspit-n a skirt [pit to make thread) [rspit-m also means "to tangle"); rsmats to let go [mats-b to take, pats-b to find) ; rsmats trotisers ; II. Semasioi,ogicax^ (l) NATURAI, EIvEMENTS: 1. sun: si Cf. Maid, sa fire; Miw. hi, hit sun, day; Muk. icnien, hismen sun; Wint. sun sun. 2. moon: po Cf. Maid, pombok moon (also poko sun, poho night, poketcuk star); Yok. opodo, upic moon; Wint. po fire; Tak. be svm; Kal. pyd-'n sun, day. 3. shine : tuks Cf . Miw. lulep- shine ; Muk. tuxts day. 4. night: su, us Cf. Miw. sien- be night; Wint. sinol night; Tak. xii'n night (Tak. x regularly comes from 5). 5. cloud: yots Cf. Maid, ya; Tak. hat; L. Um. hiai ; Pen. * h-yai. 6. sky: tsap Cf. Muk. tcarak. 7. star: musa Cf. Muk. mur night. 8. fire: hen Cf. Miw. wik-, wik-, wul-, wel-; Yok. licit, oxit; Muk. xii, xihi, xute go for a fire, light a fire ; Tak. -'/cm/- to set fire to, to catch fire; Coos cutl to set fire, Hcil- to burn, 'tcwel- fire; Pen. * swilihe, *'lswil- to bum (cf. "to burn" next semantema).^ 9. bum isai Cf. Maid, tco; Miw. tsup. 10. salt kan Cf. Muk. akes, aks. 11. dry tes Cf. Miw. tsutul. 12. stone: tsa Cf. Miw. cawa; Muk. isin. 13. earthquake: pumimb Cf. Wint. pomoko. 14. hole: Aw/ Cf. Maid, tuke; Muk. kutui. 1 In making the following list of semantemas I found it necessary to alter somewhat the orthography of the different sources in order to obtain a common basis of comparison. The system of transcription we have followed is that of the American Anthropological Association. It may be well to point out that in Mixe, s and c are interchangeable ; i' and f'are the "unrounded" forms of u and o; in Mixe the dynamic stress accent is invariably on the semantema, and this in turn is so seldom of more than one syllable that a di-syllabic stem may be looked upon confidently as resolvable under further scrutiny. For Miwok, Wintun and Maidu, I have made use of my own unpublished material and of that of Mr. J. do Angulo, also of the grammar of Maidu by R. B. Dixon in the first volume of the Handbook of American Indian Languages. For Pomo I have made use of my own and J. de Angulo's unpublished material, also of The Ethno- geography of the Pomo Indians by S. A. Barrett. For Mukne I have used the Mutsun Dialect of Costanoan by J. A. Mason. For Yokuts I have used The Yokuts Language of Southern California by A. L. Kroeber. Many of the kinship terms are to be found in E. W. Gif ford's exhaustive work "California Kinship Termi- nologies". * Pen. *swil- "fire" ? : Pen. *'tswil- "to burn" ? — E. S. 286 VI American Indian Languages 2 15. water, river; ni Cf. Maid, dim wet; Miw. nuni- to be submerged; Wint. hm boat. 16. wet: 50^ Cf. Miw. cumek to soak up. 17. rain: tub, tsuy Cf. Miw. upa; Yok. ciwek drizzle. iS. wash: iL'ibux Cf. Maid, dp to dive; Miw. upux bathe, swim; Yok. ep-, ip- swim; Muk. upxi to sip; Tak. 'pag- to bathe. 19. wind: poh Cf. Maid, biye; Miw. pwc to blow; Yok. piuate whistle; Muk. put blow, pusninyis whirlwind; Wint. pul whistle; Tak. -p'ouw- to blow; Chinook po to blow; Kal. ptU to blow; Coos pjix- to spout; Tsim. bmi to blow. 20. red: (saps Cf. Muk. lasas. 21. green: stisA Cf. Maid, kotcis; Muk. icutsu. 22. then: we/ Cf. Miw. ne this; Muk. «e this. 31 23. Temporal affixes: Past te- Cf. Miw. -/it; Muk. -te, -kie. "essential" : -b, -n Cf. Muk. -n infinitive; Tak. -n noun suffix, sometimes used to make verbal nouns, exactly as in Mixe; Maid, -n infinitive ending, "contmuative" -its, -ats (forms of the verb to be) ; Cf . Miw. -ntcu to live, also a continuative ; Mukne tcira always, continually. 24. big: mi, mik, mix (V. also old) Cf. Yok. nioxodo old; Muk. muk adult; Tak. mahai. 25. little: mutsk. Cf. Muk. pusnt. 26. round: pik Cf. Miw. pol; Yok. huk-\ Wint. bakak. 27. Numbers:' Mixe Maidu Miwok Yokuts Mukne Wint. I tig wite, suii kerje yet emet- ete 2 matsk pen otto punoi utxin pampe 3 tugig sapm toloko soopin kapxan ponol 4 niaktask tcii oyisa hatepanai usit, utit emus 5 mogosk mawik [ma hand) macoka yitecinil parues ete-sem 6 tuduk timbo temoka tcolipi nakitci serpol 7 ustug penimbo kenekaki 7iomtcin takitci serpote 8 tuktug pentcii kawinta monoc taitimin pan-emus 9 tastug tciinbo wo^e nonip watsu, paki pan-emute 10 mak matcam na^aica waditc matstt pampa-sem (two hands) 20 ipc inaiduk na'a kay 28. more: maak Cf. Miw. manik. 29. no, not: ka-, kedi. Cf. Miw. ke, ken, ket; Yok. k'a>nu; Muk. ekwe, kwe. 30. yes, positive, good: oy Cf. Maid, he; Miw. hi; Yok. houu; Muk, he; Wint. 0. 31. thus: sa Cf. Miw. saka like, as; Muk. sata. 32. if: pen. Cf. Muk. pini. (2) animals: 33. dog: uk. Cf. Maid, suku; Miw. tcuku; Yok. tce-jej; Muk. hutce; Wint. cukut; Tak. 'tsixi. 34. bird: hon Cf. Maid, hu to fly, horn nest; Miw. hotju egg; Yok. hoy egg. 35. eagle: wits Cf. Miw. wipayak; Yok. witisnl, eagle, witc, condor. 36. snake: tsan Cf. Muk. lisana. 37. frog: ink Cf. Miw. wataksay, kotola; Wint, watak. 38. fish; ak Cf. Maid. mako. 39. flea; picg Cf. Muk. wipsur. ^ Note to "two" Chin, mokct., Mixe matsk cannot be directly compared to Mixe us- in us-tug "7" (= 2 -|- 5) and to Miwok otto and Mukne utxi-n but may be related if we assume m- as archaic durative intran.sitive prefix, such as we have far more transparently in Hokan and Muskogian. — E. S. 40. louse: ag Cf. Muk. rax. 41. bedbug: /// Cf. Maid, dt^ louse. (3) plants: 42. wood, tree: kep, kip Cf. Miw. kapiim, bark; Yok. yapkin ; Muk. xipur, tree. 43. brush, bush, forest: yuk Cf. Miw. yomyum, brush; Yok. yawud; Maid, yo, flower. 44.. tobacco: xuyge Cf. Miw. hutia. 45. moss: sarjk Cf. Muk. sasuk. 46. fruit: uts Cf. Muk. owos. 47. flower; pux Cf. Maid, pu, to blossom. (4) man: 48. person, people: hay Cf. Maid, maidik; Miw. miw; Muk. muwe; Coos md; Yok. mai. 49. man: yayek, yadyek Cf. Miw. naya-yak, Yok. muk-yamk. (also Coos -iyag plural suffix in terms of relationship). 50. woman ; tosdyek Cf . Miw. occayak, Yok. latc-yamk. 51. old man: mix (also "big") Cf. Maid, muk, big; Yok. moxodo; Muk. muk, adult. 52. foe, enemy: was Cf. Mukne wayas. Seven: Penutian Languages 287 3^ 53. youth, boy: nakn Cf. Yokuts notco. 54. father: tyedj, tat Cf. Maid. te\ Yok. -atet; Muk. ete, grandfather; Win. tata. 35. mother: tagh, nana Cf. Maid, na, ne; Miw. ita; Yok. -ajaj; Mukne ti/a, parents: Tak. ni-, -hin; Kal. 7ii; Low. Chinook -wafl; Coos e'n-dtc {-ate kinship ending); Coos nik/a "mother!" (voc); Tsim. (Nass R. dialect) nc-^i "my mother" (-H "my"), nd'd (voc.). 56. grandfather: ap Cf. Maid, opa, pa; Miw. apa, father; Mukne apa, father; Win. apa; Yok. bapa, bap', father's mother; Tsim. (Nass R. dial.) no-beb-i', my mother's brother, bip' (voc). 57. elder brother: ats Cf. Miw. ata, tatci; Mukne taka. 58. yoimger brother: its Cf. Maid, tu, t'i; Miw. tcale; Mukne tare; Miw. iDi'' (Sapir MS.). 59. younger sister: vis Cf. Wint. utcii; Miw. Wa'* (Sapir MS.). 60. uncle: haym Cf. Maid, yam, kam; Miw. kaw\ Yok. kawa, komoy. 61. elder sister: isyo Cf. Maid, cti; Miw. De'De* (Sapir). 1 62. aunt: tsugu Cf. Yok. guiha, nusus. 63. child: tt»;g Cf. Miw. aijst, son; Mukne rwis, son. 64. son: wza»; Cf. Mukne mos. 65. daughter: ^e/, wyi's. 06. diminutive or endearing suffix (suggested by Sapir) -s, -c, -j (e. g. wi'-y, daughter; Cf. ur)-g, chUd. Pen. stem perhaps *ariu, child; dimin. *a'yiu-si > Mixe tji-s, Miw. ajj-si, son, Mukne ini-s, son). With stems ending in -t, -d this old diminutive combines in Mixe to form -ts, -ic, -dj, hence: tyed-j, father (probably dim. form of tat) ; at-s, elder brother : Miw. ata ; it-s, younger brother: Maid, ti, Miw. iDi'^; ut-s, yoimger sister: Miw. Wa'\ This dimin. *-s! survives also in other Pen. dialects: Miw. ay-si, son; S. Coast Miwok (Barrett) taiyi-s, man; kuleyi-s, woman; hena-s, boy; oyi-s, old man; potci-s, old woman( ?) ; Mukne ini-s, mo-s, son; Wintun utcu, younger sister (diminuti- vized < *utu < *utaP); Tak. -xi, -x < *-si, *-s (in hap-xi, child, haap-x, one's children, cf . haap'-, child in other combinations) ; Chinookan (Wishram dialect) -c in -k!acu-c, paternal grandfather, -gak ! u-c, maternal grand- father, -k/i-c, paternal grandmother (these stems are etymologically related to respective reciprocals: -qcE-n, -gaka-n, -gia-n), further in wi-n-am-c, my father (stem -am) and wa-n-aq-c, my mother (stem -aq-) ; Coos -ca, endearing suffix (e. g. huumik-ca "dear old woman"; umd-ca- ^ c- of Maidu cti and -sy- of Mixe tsyo may be diminutivizing. tc, uma-c "grandmother": vocative uma. This Penutian *-si diminutive is characteristic, it would seem, as contrasted with its vmdoubtedly cognate Hokan *-tsi {*-'tst). — (E. 8.)^ 67. I: es, -s, -es Cf. Maid. -5, -mus; Tak. -xi "me", from < *-si. 68. I: n- Cf. Miw. kan; Yok. na; Muk. kan; Wint. nanu "my"; Coos n; I/Ow. Ump. na "1", -n "I, me"; Tak. -'n, -n (trans, subj.); Chin, -n- "I, me" (Wishram dial, also indep, «a.) 69. thou: migs, m- Cf. Maid, mi, -mam, -mem; Miw. mi; Yok. ma; Mukne me; Wint. mi; Tak. ma; Chin, -m-, -mi-; Tsim. — n (thee); m-. 70. he: yt Cf. Maid, -i; Miw. /. 71. we: -m,, -em, -ma Cf. Miw. -m, -me; Yok. mai; Mixkne makse, fnak; Tak. -aw; Coos -ami; Tsim. -Em. 72. ye: -t, -te, -ta Cf. Miw. -tos, -ton, -tok, -tc. 73. which: wud Cf. Mukne watt "someone"; Coos wit; Low. Ump. watc. (5) BODY PARTS AND ACTIVITIES: 74. eye: win, is "to see" Cf. Maid, hin; Miw. c'in-ti, eye, c'iy, to see ; Yok. oil, to see ; Mukne xin ; Wint. wini, to see, ca, eye; Tak. al-xi-g-, -xi-k-, (Tak. -xi- < *-si-), also -xanw- ("to look") < *-san-w-.- 75. mouth: aux Cf. Miw. awo; Yok. ata, ca; Mukne xai. 76. tongue: _va»e Cf. Maid, eni; Miw. He/??/, leiip; Yok. tadxat, palat "project tongue"; ^lukne lase; Wint. tahal; Tak. ela; Coos Ae//a ; Low. Ump. i^al. 77. to drink: ug Cf. Miw. wcczt; Yok. uk; Mukne ukis; Tak. iigw-. 78. to speak: mo^ Cf. Maid, mo-, ma-, action with mouth; Miw. mo-, answer; Mukne }>!ons. 79. to answer: azoy Cf. Miw. has-, ask; Mukne osehe. 80. word: kops Cf. Maid, ka, say; Miw. /fa-, say; Mukne ko, say. 81. to call: mo Cf. Maid, wo; Miw. «'o; Low. Ump. waa, say, speak, tell; Chin, wawa, talk; Tak. yaway-, talk (dissimilated from *waway- ?). 82. nose : hikb Cf. Maid, hikii; Miw. A«/j ; Mukne hus. ^ Nominalizmg suffix -c = Mukne -s verbalizing suffix (e. g. yitc-s "language" <; rite "to speak"; = Tak. -{a)x infinitive suffix of intransitive verbs < *-(a)s; = Low. Ump. -s verbalizing suffix (waas "language" << waa "to speak"). ^ In other words we have two old Penutian stems: *(i)sa or *(i)say (> Miw. ciy-, Wint. ca, Mixe is, Tak. -xi-) and *{i)san, *(i)sat > Miw. cln-ti, Yok. cil, Mukne xin. Maid, hin, Tak. -xan-w-, Tsim, saal "to notice", perhaps also Coos xil-, "to look around". — E. S. 288 VI American Indian Languages 2 33 83. ear: tatsk Cf. Miw. tokoc; Yok. luk; Mukne liiksiis. 64. hand: ki Cf. Maid. ka-. with the hand; Miw. eku. uka « *esu. Miw. k = Mukne s. regularly. — E. S.); Yok. ko, to hit with hand; Muk. isu; Win. kupum. 85. left hand: ava Cf. Mukne an-si. 86. to take: mats Cf. Mukne nm;tr, get; Low. Urap. mate-, to lay; Tak. mats lag-, put, place; (Pen. base *mats- > Ta^. mas-g-); Chin. (Wishr. dial.) -A--n»a, to lay, -a-l-ima-lx, to put into water; Tsim. mag-, to put dowTi. ; Maidu me, to catch. 87. to give: mo, to have: mid Cf. Maid, me; Miw. a>«-; Muk. ;trMm, cumi. 88. to do, make, work: dun Cf. Miw. taiiha; Yok. /nc; Muk. /w« finish. 89. to put, place: -wits Cf. Maid. wi-. pull, press; Miw. wtk-; Tak. -ttyhA, to spread out. go. foot, leg: t'ik Cf. Maid. /^A, jump; Miw. tek, kick, Aa/f, foot, tirji, hip; Mukne /^A^, walk on toes. 01. knee: koc, Cf. Tak. gouk; Low. Ump. qoq; Chin- -q!oxl. 92. to go: ois Cf. Maid, u-; Miw. «-. 93. to walk: wag Cf. Miw. iwi-n, weeta; Yok. wadix, pass by; Mukne twa//, come, ze/a/^, go; Tak. wi-, to go about. 94. to go: y- Cf. Maid, i-ye; Miw. yoc-; Chin, -ya, -i-\ Tsim. yaa; Tak. yaw-; (Cf. Hokan *iya, to go). 95. action with the foot, walk, trail, etc.: pa- (pason, to follow, paboy, to chase, pesamb, to go out) Cf. Maid, be-, with the foot, puiyi, hunt. track; Miw. puy, move, depart; Mukne paya, run, payta, himt. 96. to stay : wi (also in : tti^ic, a bed)^ Cf. Tak. teiay-, to sleep ; Kal. ivai, to sleep ; Tsim. woq, to sleep ; Coos, haya-ti, to lie down (Coos A often comes from iv) ; Yok. woi-, to sleep. 97. to sit: uy Cf. Maid, in-; Yok. uy, to lean; Mukne one. 98. bone: pak Cf. Maid, -/'o^, -puk; Wint. /jaA. 99. skin: po Cf. Maid, ^tt, posa. 100. neck; yok-yi Cf. Yok. ogun; Tak. gwen; Coos kwin-ts; Low. Ump. kwininicu. loi. rib: /la/c Cf. Maid, /cz ; Miw. Atf/c, bone; Mukne /ca/c, bone. 102. phallus: tsii-k Cf. Maid, utcu, piss; Yok. icwyo ; Mukne tcoxo, pudenda. 103. testicles: istca Cf. Miw. caii-tal; Mukne tcatia. 104. anus: pxut Cf. Miw. poii, navel; Yok. poiodo, intestines; Mukne puttus, belly; Wint. pot. 105. to urinate: tatsp Cf. Miw. ota, otso; Mukne tcala (Cf. Wint. tcunu-s; Tak. xan. — E. S.). 106. heart, belly, thought: hot Maid. ho7i; Yok. hut, hon, to know, horj, heart; Mukne xtitu, belly; Coos haw-, to imagine, think; Low. Ump. hai, opinion; Tak. heivhaw-, to think. ^ 107. fear : tsok Cf . Miw. ceki ; Yok. dote : Mukne susu. 108. love, like: tsok Cf. Maid, -sak-; Yok. icik; Mukne otciko. ^ < "Sleeping instrument". See -i nominalizing suffix. — E. S. ' Maidu ho-n and Yokuts ho-r] probably nouns in -n, cf. "infinitive" and noun-forming -n, from base *ho- < Pen. *haw-, "hau-; Mixe ho-t probably factitives in -t from same base. — E. vS Editorial Note Originally published in International Journal of American Linguistics 6, 28-33 (1930). Ancestral Mixe-Zoque is now supposed by most authorities to have been the language of the Olmecs, the oldest of the archeologically known high cultures of Mesoamerica (Campbell and Kaufman 1976). Sapir's "Mexican Pcnutian" (Mixe-Zoque and Huave) was expanded by Whorf ( 1935) to include Mayan, Totonac, and Uto-Aztecan. Greenberg (1956) accepted the addition of Mayan and Totonac but excluded Uto-Aztecan. Seven: Penutian Languages 289 Greenberg has recently (1987) reiterated his version of Mexican Penutian as part of a hemisphere-wide classificatory scheme bolstered by extensive lexical comparisons. Neither Whorfs nor Greenberg's claims have met with favorable reaction from specialists in Mesoamerican languages (see Campbell 1979: 964). The connection of Huave to Mixe-Zoque, proposed by Radin (1916) and accepted uncritically by Sapir, is now considered improbable. The most likely affiliation of Huave is with Otomanguean (Campbell 1979: 964). COOS-TAKELMA-PENUTIAN COMPARISONS Edward SAPiRf and Morris Swadesh 1. Introduction 1.1. Date of manuscript 1.2. Description 1.3. Phonology 1.4. Why withheld 1.5. Symbols and arrangement 2. Sapir's comparisons 1. The Coos-Takelma-Penutian compari- sons presented here were made by Edward Sapir. Morris Swadesh has prepared them for pubhcation and has provided the intro- ductory information and discussion. 1.1. Sapir's manuscript, which now forms part of the Franz Boas Collection of the American Philosophical Society, evidently dates back to about 1914, the publication year of Leo J. Frachtenberg's Coos, An Illustrative Sketch, Extract from Handbook of American Indian Languages (BAE-B 40.2.297-429). It surely represents an im- portant part, though far from all, of the lexical evidence Sapir refers to in A Charac- teristic Penutian Form of Stem (UAL 2.58-67), pubHshed in 192L In this article he concurs with Frachtenberg's suggestion of Penutian affinities of certain Oregon languages, stating: All this is very interesting to me, as it chimes with conclusions or hypotheses I had arrived at independently. On the appearance of Frachten- berg's Coos grammar it soon became clear to me that the morphological and lexical resemblances between Takelma and Coos were too numerous and fundamental to be explained away by accident or plausibly accounted for by borrowing. The appearance of Frachtenberg's Siuslaw material has only tended to confirm this impression, further, to make it perfectly obvious that Coos and Siuslaw, as Frachtenberg announces, are divergent representatives of a single linguistic stock. Meanwhile comparisons of Takelma, Coos, and Siuslaw with Dixon and Kroeber's Penutian group of California (Costanoan, Miwok, Yokuts, Wintun, and Maidu disclosed an astonishing num- ber of both lexical and morphological correspond- ences. ... In spite of our slight knowledge of most of the Californian languages involved, I succeeded in getting together what I believe to be a quite respectable mass of evidence tending to unite the southern languages with those of Oregon into a large and highly differentiated "stock". In the same article Sapir refers to evidence for still further Penutian relationships. It is possible that other manuscripts will yet be discovered containing this material. Much of it may have been in the form of marginal annotations made by Sapir in books and re- prints, like those compiled in Comparative Penutian Glosses by Edward Sapir, in the Boas Collection. 1.2. The present manuscript consists of 13 pages 8x11, handwritten in Sapir's usual small writing. There are 152 numbered groups of cognates, with columns for Coos, Takelma, Penutian. There is no other title. At least two shades of ink are distinguishable, a dark blue for all but set 152, and a lighter blue for the latter and for additions to the first 151 sets. Since the additions include items in the Coos column marked "Mil." (that is Miluk dialect of Coos) and four annotations of 'see L.U.' (that is, Lower Umpqua or Siuslaw), one may infer that the main list was made before and the added notes after the appearance of Frachtenberg's Coos and Siuslawan sketches. The original Coos items must then have been based on Frachtenberg's Coos Texts, which appeared in 1913. This assumption is corroborated by the fact that Sapir's copy of Coos Texts has marginal glosses of Takelma comparisons in what seems to be the same ink as the manu- script list, and there are 'L.U.' annotations in the lighter ink. All the foregoing observ^a- tions on the ink used in the manuscript were made by inspection only. The identifications seem for the most part quite obvious, except for those parts of the manuscript which have been most affected by light. 132 292 VI American Indian Languages 2 133 Besides the entry numbers the manuscript has a set of check marks varying with a diagonal criss-cross at a few points. Sapir used such marks to register a later critical reexamination of his work. That is, at some time after he had made the list, he must have gone through it again, with constant reference to the sources, checking the entries which seemed plausible and putting the criss-cross by the ones that on reflection seem improbable. During this process he would often insert additions to the original material. In the present manuscript, we find that the criss-cross sometimes confirms and sometimes overrides an earlier hesitancy re- flected by parenthesized question mark or annotations (e.g. in 57 after the Yokuts form: "perhaps better to Tak. -ha-n, no. 55"). Three asterisks between entries 55 and 56 may be a pause mark, showing where Sapir had temporarily stopped in his check-up. A short horizontal line between 86 and 87 is evidently another pause mark. 1.3. The manuscript does not include any full analysis of phonological relationships. However, one finds phonological notes at various points. Some of these merit general comment. The occurrence of Yokuts d for earlier 1 is indicated or implied in a number of places (8, 9, 19, 50, 54, 71, 72, 74, 78, 79, 83, 84, 87, 94, 95); Newman's work' has since shown that this is a regular dialectal development in Wikchamni, while the other dialects preserve the lateral. In two cases (10, 23) we have Wintun X, Takelma s recon- structed to t; item 7, disregarding the first Coos form (Hanis dialect), suggests that I or X might be a better reconstruction. The development of Takelma x from earlier s is posited in 11, 35, 41, 61, 75, 99, 121, 123. The origin of Takelma s from earlier c is suggested in 129. This fits in with the fact that Takelma has no c even though it has s and the com- bination s?, the latter being pronounced as a glottalized affricate. Sapir's idea was evi- dently that pre-Takelma s changed to x, ' See Stanley S. Newman, Yokuts, Viking Fund Publications in Anthropology 2 (1944), p. 16. coinciding with original x, and that new instances of s developed out of older c and ch. The evidence in general seems to bear out Sapir's theory, but the proof is complicated by the presence of more than one s-sound in Yokuts, Coos and perhaps other Penutian languages, and by the existence of two x-sounds in Coos and Siuslaw. Robert Shafer has attempted a clarification of Penutian k and x-sounds,^ but he does not fully suc- ceed; his material is insufficient in quantity and does not properly distinguish between probable cognates and similarities which might have resulted from borrowing. How- ever, the problem may be expected to yield to patient research in the future. Where Sapir gives two or more spellings for a word, they may be variant phonetic recordings of one and the same form, dialect variants (as in Yokuts), or functional vari- ants (e.g. verb stem and aorist stem in Takelma) . The first kind of variation is here eliminated to the extent possible by using phonemic spellings. The other two types have been retained because of their value in tracing phonological relationships. For Yokuts it was desirable and easily possible, by consulting Newman's mono- graph and with additional data kindly pro- vided by Newman, to specify the dialect of most of the cited forms. Where Sapir gives a single form and it coincides with the Yawel- mani dialect, it has been labeled Yawelmani (Yy). Where Sapir's form is from another dialect or includes more than one dialect, the Yawelmani has been added after Sapir's Yokuts entry. 1.4. Why did Sapir not publish his Coos- Takelma-Penutian evidence?' The answer is perhaps suggested by the following remarks (UAL 2.59, 1921): Unfamiliarity with Alsea (Yakonan) and Kalapuya made it impossible for me to follow * Penutian, UAL, 13.205-19 (1947). ' Some of Sapir's comparisons and structural observations are included in L. S. Freelander, The Relationship of Mixe to the Penutian Family, UAL 6.28-33 (1930). Seven: Penutian Languages 293 134 the chain of evidence geographically. Neverthe- less, Chinookan points of contact soon manifested themselves too persistently to be brushed aside. After hesitating for a long while to take up seri- ously the possibility of affiliating Chinook, one of the most isolated and morphologically specialized languages in America, with the Penutian languages of Oregon, I now find myself forced by the evidence to admit such an affiliation as not only possible but decidedly probable. . . The greatest surprise was still awaiting me. Tsimshian . . . Should it be possible to demon- strate (and I am fairly sanguine that it can be demonstrated) that Tsimshian is a detached northern offshoot of Penutian, we would be com- pelled to face a most interesting fact in linguistic diflferentiation and in the distribution of American tribes. These comments show that Sapir con- sidered the Coos-Takelma-Penutian rela- tionship to be part of a larger grouping which he felt could be demonstrated. Evidently he was awaiting the appearance of Kalapuya data and was hoping to work out in detail the relationship of Chinookan and Tsim- shian. He felt sufficiently sure of himself to publicly state his theory, but perhaps he thought it well to wait with the specific lexical data until he could present the evi- dence for the fuller picture. The case would then be comparable to that of Sapir's theory relating Na-dene and Sino-Tibetan.'* When students asked Sapir why he did not publish his evidence for this connection, he said it might be better to hold off until he had worked out and published the full details of Athapaskan phonology. However, in retro- spect, one can only regret that this material was not published long ago. 1.5. Sapir's 1914 orthography is here modernized and spellings are given in phonemicized form to the extent possible. In the case of Takelma, we take the aspirates to be clusters of stop with h and the glot- talized to be clusters with glottal stop; Takelma accent is here reduced, with reason- able justification inferred from Sapir's de- * See discussion (by Morris Swadesh) of Athapaskan and Sino-Tibetan by Robert Shafer, UAL 18.178-81 (1952). scription,^ to two types, a normal accent (') with tonal variations according to position in the word, and a rising or sustained long accent (') occurring on long vowels or on vowel-sonant groups in contrast to falling tone for the normal accent in this position. Points of phonetic usage: x is for back velar spirant, c is for sibilant affricate (ts), H for shibilant affricate, double letters are used for long vowels, r is for voiced velar spirant (in Coos). The languages are abbrevi- ated by using the first letter for Takelma, Coos, Siuslawan, Yokuts, Wintun, the first two letters for Chinook, Miwok, Maidu, Mukne (Costanoan — Freelander's term). Yy is used for the Yawelmani dialect of Yokuts. For convenience in printing, Sapir's columnar arrangement of the manuscript has been dropped. Sapir's diagonal cross for improbable comparisons is retained, his check marks are omitted. His parenthetic question marks, also indicating doubt, are retained but are placed after rather than before the entry. Other than in such matters of arrangement, punctuation and phonetic symbols, the materials are given as fisted by Sapir. 2. 1. T kuu-x-m/e (?) (see S); Mi kule, kulei wife, woman. Ma kiila woman. 2. C k'^aaxaX how; T gaP bow; Mi kono, W kul-sak how. 3. C walwal knife; T wiih-A;m/e; Mi hulaya knife. 4. tuu good (aor. tuu\\^iu-k- to be good); Mi towi-s good. ^5. C asoo again, asoow-is second; Mi os(s)a, otta two. 6. C e'^n-ee^ mother, niika mother!; T hin-, ni- mother; W nen-^m, nenin. Ma ne, Mu ana mother. 7. C huuI nose, Miluk fin-nuuq; T sin-, sinii-x- nose; W Xinik, suno, Y tiiDiik nose, Yy tinik. 8. C helta tongue; T ela tongue; Ma eni, W tahal. Mi le-mtep, le-tip, Mu las, lase, lasa, Y talxat, '''aladis, m-ada-t tongue, Yy talxat tongue, "^aalat- to lick. 9. C kxla leg; T k«el- leg; W koli leg (<*k*ili ?), Mu kolo, koro foot. Mi kollo * Takelma, Handbook of American Indian Lan- guages, 40.2.17-18 (1922). 294 VI American Indian Languages 2 135 foot, Y kada-Sa lower leg, Yy kalassa'i*. 10. C Miluk Hil-li leg; T sal- foot (<*tal-?); W Xel-ma (<*tel-?). 11. T xuu''*n night (for x:s, cf. xi water below); W sinol night. 12. T b§e sun, day. poo now, today; Ma poko sun, W po day, Y 'i'opo-do sun, "i^op-di day, Yy "^op moon. 13. T n6ox rain-, W luha rain. 14. T p'^ii^re; W po fire. 15. T s'i'aaw lake; W nahi lake. 16. T som mountain (<*tom?); W toX, ho\ mountain. 17. T tan rock; W Son rock. 18. C Sximl black bear; T x§,mk grizzly bear; W Silai bear, grizzly bear. 19. T -s'i'il red; W tulu-ka, tede-kit red. 20. C bni ^ood; W laiyok good. 21. T ma thou; W mi thou, [see 47]. 22. T pam- up, pam-ls sky; W panti-be up, panti on, upon. 23. tqanX- (?) [to hit, to strike with instrument] ; T saansan-, sana- to fight with, spear; W Mia to shoot (<*tana?). 24. T -t present participle, adjective suffix; W -u-t future participle, -t adjective suffix with sub- jective noun, Ma -do- present participle. 25. T -khi?, -ki? if; W -kila conditional. 26. C -uu interrogative suffix; W -wi inter- rogative. 27. T -de?, the? / (intr.), future -tee, -thee; W -da /, Mi -t, -te /, me, my. 28. C iin not; T ?anii? not; W eleu (<*ene-u?), Y am, Yy ?ohom, Mi (E.) ela. 29. T yaan-, yana- to go;W hara to go. 30. T p^nx hunger; W bira to be hungry. 31. C tk"*- to kick; T tkuuntkan-, tkuun- to kick; W qow(a) to kick. 32. -eenii reciprocal verb suffix, -inii (relatives) to one another; T -an- reciprocal verb suffix; W -pu-ra re- ciprocal verb suffix (?). 33. T ?al- to (?), ka-nan in; W el in, into. 34. T pay- out of house (vb. prefix); W pat out. 35. C yipsn three (?) (see S [Siinx]); T xipini three; Ma sapwi three, Yy soopin, Mu kapan, kaphan, kapxan three. 36. C SuuX- to set fire to, m\- to burn, Hweel^re; T -s^iiluii-, -s^iil- to set fire to, to catch fire; Ma sa, Y ?oSit, Yy ?osit, Mu Soto fire. 37. C mee human being; Mu ama person, Ma mai-dii Indian, mai-ki boy, Y may person. 38. T yap?a person; W yapaitu person. Ma yepi man, husband. 39. T tak-ax- head; Mu taxa-S head. 40. C x'^iHux'^ ^ head (?); T ?ulu-k?-i- hair of head; Mu ut, ^| uri, uli hair, Yy ?otow head. Ma ono head. 41. C '))[ to drink; T xl water; Mu si water. Ma sewi river. Mi kik water. 42. C Xipi^ arrow; Mu tep-s, temo-x, nemo arrow. 43. C Iqawe to die; T lohoy-, loho- to die; Mu laku- dead. 44. C cimsimt to sleep (plur., redupl. <*tim-t or reformed from *ir)tim > *t-ir)tim); Mu eten, enen, etin-i to sleep. Mi eM, Y ?er)tim, Yy ?entam a sleep-inducing substance. (Is Esselen acin- after all merely borrowed from Costanoan?). 45. T ti inter- rogative enchtic; Mi mau-ti whof hi-ti whatf Ma -de interrogative, Y ti enclitic, Mu -sa, -s (< *-ti). 46. C n- /; T -?n I—him, -n (fut.) I— him; Ma ni /, W ni, Yy na?. 47. T ma thou; Mu me, men, mene. Mi mi, Yy ma?, W mi, Ma mi thou. 48. C Xtaa earth (?) (see S [Xa?ay]); T tkaa earth; Ma kau ground. 49. Ch -k-, -x- my; T ki /, -te-k, -t-k my, koom we; Mu ka, kan, kana /, Mi kanni 7, -ka, -ka-n my. 50. C -en numeral suffix so and so many times; T -n, -(t)an j adverbial numeral suffix; Mu -na adverbial ' numeral suffix, Ma -nini, Y -d, -1, Yy -? . . . il'. 51. C puux"^- to spout; T phoophaw-, phoow- to blow; Ma bo to blow. 52. T mena bear; Y moloy bear. Ma mode brown bear. 53. T maan- to count; Ma -mak- to know, count, measure. 54. C -enii to do, to make something; T -(a)n(a)- causative verb suffix; Mi -ne causative, Y -la, -da causative (for 1 < n, cf. no. 50), Yy -aala. 55. C -n, -ni-, -eene plural element found in few nouns (e.g. mee-n human beings, ceey-eene stnall pi., qe-ni-xet short pi.) — its irregularity and rarity show it to be survival from remote past; T -ha-n, -an noun plural (not very frequent); Y -n, -in plural of pronouns, -in numeral collective, W -1 dual of pronouns, -li-t dual obj., -le-t plur. obj., -li-n dual pos., -le-n plur. pos., Ma -nono many (?). 56. C -e, -ii petrified noun plural (e.g. tummeeX-e old men, timil-ii men, keneeyes-e | hunchbacks); -a petrified noun plural (e.g. f aXimaq-a big pi.); Y -i, -a noun plural (for Seven: Penutian Languages 295 136 Y ablaut, e.g. 'i'onmid: ?onemad-i, cf. C toomiX: tummeeX-e, knes: keneeyes-e), Yy -i, -a subject case with plural stems. ^57, C hinna so and so many each; Y -hin collective (e.g. yapkan-hin many trees) (perhaps better to T -ha-n, no. 55). 58. T ^uuk'^- to drink; Yy "i^ugun- to drink, Mu ukis, Mi uhu. 59. T ka that (indef.); Y ka that (vis.), Mi i-ka that. 60. C t that there, ta so, such, te- article with pos. pronoun; T ii-ta (ka) that yonder; Y ta that (inv.), Yy ta that. 61. C -is noun suffix (often suffixed to verb stems: Xeey-is language Xee-c he spoke), -s, -t-s, -en-is nouns of quaUty; T -(a)x infinitive (often used as verbal noun: S ?ipn-ax speech); Mu -s noun-forming from verb stems (e.g. ni^i-s language < nm to speak), W -s verbal noun suf., Mi -s noun ending (?), Y -0$ noun suf. 62. C -ne it is (e.g. n-ne it is I); T -(a)n-, -in- petrified intransi- tive suffix; Y -in- intrans. (?), Mi -ne, -fie in- transitive verbifying. 63. C ii-ta emphatic particle; ii-ta(ka) that; Mi i-, i-ni, i-sa, i-mo, i-ka, i-ti that. 64. T -n noun ending (e.g. yiwi-n speech < yiw to speak); Ma -n in- finitive ending. 65. T -a'i'n person, people of; Yy -? . . . inin people of. 66. C axaax maternal uncle.; T has- maternal uncle; Y 'i*aga$ mother's brother, Yy "i^aagas. 67. C ^inl^in eyebrow; Y nimejid eyebrows and lashes. 68. C cneex beard; Y jamo^ beard, Yy daamut-. 69. C k'*'in-c throat, neck, k'^'in- to swallow (Coos kw: T k^, cf. k^'ees-is wind, T k^alt); T k*en- neck; Yy ?oogun, Ma kuyi neck. ^70. Xpene wings, feathers; Y ^a'i'ada fur feathers (?). 71. T telk-an buttocks; Y teda anus, Yy iooto*^. 72. T nihwlk'' black bear; Y du?uxun bear (<*lu?uxu-). 73. teec lice (see S); T fi^eld louse (?); Y tehet head louse, Yy ti'i^it. 74. T laap leaf; Y dapdap leaf (<*lap-). 75. T bix-dl moon (?) (x < 5); Y 'i'upiS moon (cf . sun), Yy "^op. ^76. C t(^een soot; Y neheo fog. 77. T tk^'d thunder; Y taka'i'a thunder. 78. C lahX-is earth; Y diioiit earth (< *luout). 79. T hok--ai, -h6k-h-al hole (<*-Sok-al); Y Sogod hole (<*Sogol), Yy sogol. 80. T tkism- green; Y nii'i'imat green. 81. T -tkSm black; Y qimgutan black. 82. T mii now, then (particle) ; Yy 'i'ama'i* and then, then. 83. C liixUi to pass out, to pass by, landing place; Y day (< *lay), lay to step, to kick. 84. C al-$ toy; T loo-l-, loo- to play, loo-sf toy; Y do to play (<*lo). 85. C tooh-, tous- to hit, strike; T t^omom-, toom- to hit, kill; Y do battle, (?), do5 to beat, overcome (?), Ma -tul- to break flat thing (?), tup-, tus- to break (cf. 117). 86. C Xou- to eat; Y duy to eat (<*\uy). 87. T -(a)l-, -Iha- continuative, frequentative; Y -ad (<*-al) continuative, -li frequentative. 88. C maux- to chew (up) (cf. Ch); Y mok to swallow, Yy meeki-. ^89. C nix(t)- to touch; Yy niti- to squeeze with hand. 90. C tou- to fall; Y ?ot- to fall. ^91. C t^- to come in; Y tax- to come. 92. C $anx- to shake (see S); T s'i^elel-, s?eel- to rattle (?) ; Y ta^it to shake. 93. C ^au- to come apart, to pull apart; Yy tatay- to break, Y taw to overcome. 94. T helel-, heel- to sing; Y 'i'uduk-, Yy ^ilik- to sing. 95. C welex- to stoop, to lie down; Y wodo to dodge ( < *wolo), Yy wolooyee-. 96. C w^kin- to touch with stick; Yy woto- to hit with a stick. 97. C yak"-, yak'^-t- to pick, to gather (see S); Y yam- to gather seeds, yitw- to gather. 98. T -koyo-k*^-, -koy-k*!*- to touch, nudge; Yy koyo- to butt. 99. C xil- to look around (?); T -xanaw- to look (out) (?); Y Sil'i- to see, Yy sil'i-. ^100. C xint- to go fast, to run (?); Y Silit- to jump, Yy siUt-. 101. T wayaan-, way- to sleep; Y woy- to sleep, Yy woo'^uy-. 102. T mahay large; Y met, mat, may- large, Yy mayaahay. 103. T khay- woman, girl; Yy gaa'i'ina woman, girl. ^104. T tola hollow tree; Y ton, toga-S digger pine. 105. C k6s shell used for ornament; Y gu^i^ beads. 106. T s^fxi dog; Y neSe^, ^exa, Ma sii, Mu huhu dog. 107. T hSiiy cloud; Ma yaa cloud (?). 108. T nf, nii- teats; Ma mini nipples (dissimilated < *nini?). 109. T -k?otk'i*at-, -k'i*oot- to break in two; Ma -kot-, -kut- to divide. 1 10. T kele-k, kel-k- to drill for fire; Ma -kel- to perforate. 111. C qal- to dig (?); T k'^olol-, kool- to dig; Ma -kol- to bore (?). 112. T kala-p-, kal-p- to twist (thread) by rolling ; Ma -kol- to 296 VI American Indian Languages 2 137 roll. 113. T k'^ataay-, k'^'aat- to pick, pluck; Ma -ket- to graze. 114. T lopap-, loop- to pound (acorns, seeds); Ma -lop- to move with friction (?). ^115. C six- to shake off; Ma -sil- to shake (?) (see no. 92 [in different ink, showing that comparison with 92 was prob- ably made after the original entry and after the entry had been marked with criss- cross]). ^116. T tama-k?-, tam-k*^- to choke; Ma -tap- to squeeze (?). 117. T -tk^eltk^al-, tk'^eel- to break in two; Ma -tala- to crush (?). 118. T t?alal-, taal- to crack; Ma -tala- to crush (?). 119. C -aami /, we 2, or we-thee, you 2, or you; T -(i)k-am we (fut. subj. intr.), -(a)nak-am we (fut. subj. trans.), -am us, -t-am our; Mi -ma(si) our, -m, -me us, me (subj. 1 [sic]). 120. T -nk he (fut. trans, subj.); Mi -k, -ko him, he (subj. 1, including future), W -k he (past). 121 . T -xi me, I (subj . of passive) ( < *-si) ; Ma -s(i) /, Mi mu-su, mo-s, mu-$u, $i-ma I — thee. 122. C -aa'^is thou, ye 2, or ye — me, u^ 2, or us; W -s-ka, -s-ke-n thou, ye 2, ye (-ka, -ke- as in Ma -nka-no thou, ye 2, ye, -nka-s we 2, -nke-s we?). Mi -s thou, -tok-su ye (subj. 3: present and perfect). 123. T -x non-agentive (quasi-passive or reflexive), -al-x intransitive (-x < *-s) ; Ma -us reflexive, Mu -s-tap, -s-tapse impersonal, passive (?), Y -wi-5, -wi-s reflexive. Mi -si passive. 124. C C xeen-is sick; T xil-am sick (< *xin-an) ; Mi hali sick. 125. T -molo'i'mal-, mol^mal- to turn things over, .stir food in basket-bucket; Mi mole to spill. 126. T phili-phal-, philphal- to squash (insects), to whip (children); Mi pilapa to pinch (?). 127. T -hit, -thit plural of ad- jectives; Mi -ti plural of adjectives and verbs, W -te subjective pronoun plur. 128. T -k inferential past; Mi -ke, -ka past. 129. T -sii, -saa agentive (-s- < *-c-); Y -i^, -ic agent, Yy -ic. 130. C -a-'^eiwat causative frequentative (-eiwat frequentative), -iya-t causative, -e-'^et (-a-'i*at) causative passive (-et, -at passive); Y -i, -u, -a causative. 131. T -(a)pa'^ let us'.; Ma -p6, -pe, -pe'^'e let us I ^132. C -qm, -xm to be in a condition, to be in the act of; Mi -imi continuative (?). 133. C -H instrumental noun suffix; Mu -cu comitative (?). 134. C -et (-at) past passive; Y -t passive, -n-it future passive. 135. -iiyas plur. of nouns of relationship, -iiye, -eeye adjective plur.; Mi -ya noun plur. 136. C han about to, hanX shall, will; Y hi future particle. 137. C k^'a it seems, as if, hak'*'a-i as if, kind of; Y akam perhaps, it seems. 138. C hi'i particle indicating surprise; T his almost, trying to but in vain; Y hiqa perhaps. 139. C hei emphasizing particle, hi enclitic em- phasizer indeed; Ma -hehe only, just. 140. C qa-1 down, below, under (on him Well, it is said his friend Warrior-all-round- took his in canoe. the-World k'le'LEnxe qa^s k-!e'ltse^stales mak'inxE'ndalaxox awE'nxa^yaxsEns sharp-edged and cut around with it close to this edge of our (knife) he SE^ya'x. Wa, g i'FEm^la'wise Ifi'lgEwe k!e'La^yasexs hair. Well, as soon as, it is said, met where it was cut when la'e ne'xodEx Lle'tsEma^yas xo'msdasa q!a'k-owe. Wa, 5 then pulled off the scalp of the head of the slave. Well, he la'^lae LaxSvuito'dxa qla'k'o. Wa, g'i'FEm^la'wise la^sta'xs it is said they pushed out of slave. Well, as soon as, it is said, he went into canoe the he the water, when ga'xaeda ha'Emats! LE%a' ne^na'ne LE^wa' no'EniEmala ho'qawEls came the cannibals and the grisly bears and the fool-dancers went out la'xa lo'bEkwe qa^s la qa'qayaxa qia'kaxs la'e dzExwae'sEla from the winter-dance and went pursued the slave then ran along house they when he lax LlEma'esas Qa'logwise. Wa, he'^mes la qaqaya'atsa at the beach of Crooked- Beach. Well, and that is then place of pur- suing of the ne^na'naq LE^vva' no'EnlEmala qa^s sa'k'apleq. Wa, la'^'lae 10 grisly bears and the fool-dancers that they tried to Well, it is said, him spear him. na'papleda wao'kwaq. Wa, la'^laeda ne^na'ne Ui'qaplEses tried to throw several at him. Well, it is said the grisly bears tried to strike stones with their xexElya'yowe laq. Wa, we'gaa^latla lax ^na'lanegwesas claws at him. Well, not he reached, to the beach on south side however, of Qa'IogwIsaxs la'e ya'x%alesa. Wii, laE'm 1e4 la'xeq. Wa, Crooked-Beach when he fell down dead Well, then dead there. Well, on beach. he is he'x'^ldaEm^'la'wisa no'EnlEmala sEsox"sE'ndxa q!a'kox"de, immediately, it is said, the fool-dancers cut up in pieces the former slave. Wa, g-i'l^Em^'la'wise gwal sa'kwasoxs la'eda ha'Emats!a jr Well, as soon as, it is said, he finished was cut up when then the cannibals 344 VI American Indian Languages 2 129 GEORGE HUNT ho'qunts!es la'xa LlEma'ese qa^s la ha^mx-^l'dEq. Wa, went down to to the beach that they go eat him. Well, beach la'^lae o'gwaqa'^ma ne^na'ne LE%a' no'EnlEmala ha^mek'la'la it is said, also the grisly bears and the fool-dancers asked to eat laq. of it. Wa, wel?.x"dze'^lae ge'x-^idExs la'e '■"wi'^laq qaxs Well not it took, it is said, long when they (ate) him all up for ma^itso'gugiya^eda bEgwa'nEme ha'^ma'pxa q!a'k'oxMe. Wa, 120 were the men eating the former slave. Well, la'^laeda ne^na'ne, yi'xa it is said the grisly bears, that is the ma^le' two da'nEma, novices (taken ones). he'x-^idaEm immediately le'nEmax SE^ya'x-das took away his past hair ha^mg'i'^layo laq. Wa, was given to Well, food them. qa^s le'nEmapleq. and took it from they each other. Wa, Well, laE'm then he'^mesexs and that is when lE^ma'e he k*!e'lakas6^s was killed by TsEx%l'de qae's hayo'te TslEx^i'de. Wa, la^me' ya'k-aweda Fast- Runner on account rival Throw-away. Well, then was beaten of his the 10 g-i'gama^e Ts!ex^ chief Throw- "I'de la'xes ^nEmo'kwe away by his friend TsEx^'wl'de. Fast- Runner. Wa, g-i'l^Em^la'wise gwal a'mleda ma'^le' Well, as soon as, it is said, finished playing the two ne%a'nesa grisly bears with the sa'bEkwe SE^ya'sa skinned hair of the q!a'k-oxMaxs la'e TsEx%l'de former slave, then Fast- Runner da'x-^'ldxa took the sa'bEkwe SE^^ya' skinned hair qa^s la ts!as lax and he went gave it to TslEx^i'de. Throw-away. Wa, la'^lae Well, it is said he ^ne'k-a : " Wa, qa'sta, said, " Well, friend. la^mo'x now this qo'sL," ^ne'x-^lae. will be he said, yours," it is said. Wa, g-i'P'mese gwa'lExs la'e Well, as soon as he finished then ho'x^wuita went out of canoe 15 ^wl"leda k i'm'yaxdaxa x i'sa'lax-de. Wa, g-i'FEm^a'wise all those who had surrounded former disap- Well, as soon as, it the peared ones. is said. Eight: Wakashan and Salishan Languages 345 130 dza'qwaxs la'e a^'yi'lkwas TsExHvi'de qa's^Ida. Wa, k-!e's'lat!a evening came, then attendants of Fast-Runner walked Well, not, however, the (to call). it is said, g-a'laxs g-a'xae %i'^laeLeda ga'la Kwa'kugula. Wa, long when came all into house the first Kwakiutl clans. Well, he'x'^idaEm^la'wise ya'lasE^weda ma^e' immediately, it is said, were tamed the two ne'^na'na. Wa, Well, g-i'FEm^a'wise gwa'iExs, as soon as, it is said, finished they with it, la'aEl then it is said ya'laqweda sang their sacred songs the grisly bears. ma'lo'kwe teto'x^wlda. Wa, g-i'pEm^la'wise qlwe'FedExs la'e dE'nx^ededa 5 war-dancers. Well, as soon as, it is stopped speaking, then began to sing said, they the ne^na'gade. Wa, g-a'xeda ma^lo'kwe teto'x^wid ^yEx%uit!a'- song- masters. qa^s lllEl came out and of house, they Well, came the la^stali'lEle went around in house two la'xa at the war-dancers dancing as they lEqawa'lllasa fire in middle of the •o'kwe. house. Wa, Well, dEma. laE'm^la'wise then it is said they gwal finished dE'nxElasa singing with 'nE'msgEme one (round) q!a'm- song. Wa, Well, he'x'^ldaEm^la'wisa immediately, it is said, the ya'yaq!antemiie la lax speaker of the house went to the qlwae'lasasa standing-place of the ma^lo'kwe two teto'x^wlda. war-dancers. Wa, Well, laE'm^lawise then it is said 10 wuLa'sE%eda ^nEmo'kwe to'x^wid la'xes axe'xsdESE%a. Wa, was asked one war-dancer for her desired (thing). Well, he'x-^ldaEm^lawise to'x^wide %e'ka: " Wa'laqeleg'anu^x" immediately, it is said, the war-dancer said, qEnu^x" lEqwi'lasE^e la'xwa lEgwe'lex," that we be made fire [put into fire] into that house-fire," la'^Iaeda ya'yaq!antemile e'talas la'xcs it is said speaker of the house repeated it to his the " We desire ^ne'x-^lae. she said, it is said. ^nE^me'mote clan Wa, Well, yi.s the wa'ldEmasa teto'x^wide. Wa, he'x-^idaEm^^lawisa mo'kwe a^^yi'l.x^'.s 15 word of the war-dancers. Well, immediately, it is said, the four attendants of 346 VI American Indian Languages 2 131 TsExSvi'de ya'q!eg-a^ia. Wa, laE'm4awise ^ne'k-a: "^ya, pepExala', Fast-Runner spoke. Well, then it is said they said, "O shamans (dancers), wc'gadzaxins ^nE'mplEna na'nageg-exg-axg-a wa'ldEmg-asg ins let us for one time obey this desire of our ^ne^nEmo'kuk-. friends here. Wa, Well, we'g'a go on ax^e'dEx ma^Exsa' ts!a'ts!ax"- take two flat short roof- SEtna qEns we'gl yiLEdzoda'lasEk- laq," ^ne'x-^lae. Wa, boards that we go on tie them on to them," they said. Well, it is said. 5 laE'm^lawise Le'^IalasE%a ma^lo'kwe ba'bEbaklwa, yix then it is said were invited two warriors, that is QEnx%ida'yowe LE^e's ^nEmo'kwe Frowned-upon and his friend ^na'xulalese qa Warrior-all-round- that the-World, ga'xes yiLEdzo'ts la'xa ts!a'ts!a6x"sEme. Wa, they come and tie them on to the daEm^la'wise ately, it is said, 10 qa^s and they Wa, Well, qa^s and they Wa, Well, g"a xe came ax^e'tsE%a were taken pa'xalelEm put them down short roof-boards. ma^lExsa' two flat Well, he'x-^I- immedi- ts!a'ts!Ets!a6x°sEma short roof-boards la'xa ogwewali'iasa g-o'kwe. in the rear of the house. laE'm^awise then it is said da'x'^ItsE%a were taken the ma^lo'kwe two teto'x'wid war-dancers naEriLEdzoda'yowe la'xa ^na'FnEmxsa tsIa'tslaox^sEma. were laid on their backs on the each flat laE'm^lawisa then it is said a g-i'Itla dEnE'm long cedar-bark rope yaLa'4ayosa ts'.e'daqe to'xw^ld instrument of women war-dancers tying of the Wa, laE'm^laxaa'wise he' Em Well, then, it is said, also, that LE'^wa and the short roof-board. ax^etsE^wa qa was taken for ts!a'ts!aox"sEme. short roof-boards. gwe'x-^ItsE%a thus was done to the 1 5 Wa, laE'm^lawise gwa'lExs la'e Well, then it is said they finished when then ax^etsE^wa were taken ''nEmo'kwe. other one. LESLEkwe' thick Eight: Wakashan and Salishan Languages 347 132 tE'mg-ik° lEqwa' qa^s qElxasusta'lasE'^we lax awi'^s- blocks fire-wood and they were piled up at the around tasa lEgwi'le. Wa, a'^mese la nExts!E%e'da the fire. Well, and only then was in centre the lEgwl'le laq. Wa, laE'm^lawise ha'lsElaEm^la heltsla'pElcda fire in it. Well, then, it is said, hardly could look over a g'i'ltlExsde bEgwa'nEm lax xixtsla'laq. Wa, laE'm gwa'lala tall man in the putting head Well, then it is ready out into it. qae'da ma^lo'kwe teto'x^ld qo lal latslo'^yoLo laq. 5 on account two war-dancers if will will be put in into it. of the then middle Wa, he'^maaxs la'e ^ne'k-a teto'x^wide qa^s laxLa'nowe Well, that when then they said the war-dancers that should be put they on top la'xa lEgwi'le. Wa, la ax^e'tsE%a ma^lExsa' ts!a'ts!aox"sEme on the fire. Well, then were taken two flat short roof-boards la ^na'^nEmax iyaak" LE%a' la nELEdza'yaatsa teto'x^wlde, Wa, made just like also the now place of lying on war-dancers. Well, back of the la'^lae ax'^e'tsE^a q!wa'xe, Wa, la'^laxae ^na'^nEmax-iyaak" it is said were taken hemlock- Well, then also, it were made just like branches. is said, they LO^ qeqExIma^yasa a'laklala to'x^wlda. Wa, laE'm- 10 also hemlock head-rings true war-dancer. Well, then the of the ^lawise qEximda'yo la'xa q!a'q!Ek*o ma^lo'k" tsle'daqa. it is said were tied around to the slaves two women. they Wa, laE'm^lawise la ne'nELEdzodayo la'xa ts!a'ts!Ets!aox"sEme. Well, then it is said now were laid down on on the short roof-boards, they their backs Wa, laE'm^lawise yii^eda'yoweda gi'ltle dEnE'm laq lax Well, then it is said were tied the long cedar-bark to in ropes them the gwa'laasasa ma^lo'kwe teto'x'^wida. Wa, laE'm^lawiseda same manner as two war-dancers. Well, then it is said the the ma^o'kwe ba'bEbaklwa ^ne'x'xa ma^lo'kwe ts!e'daq q!a'q!Eka : 15 two warriors said to the two women slaves. 348 Vf American Indian Languages 2 133 " ^ya, sa'sEm, gu'no gwaLlEXLa'lalaxo qa'so lal lax'La'not "O children! do not scream if you will you will be I put on top axga'da lEgVVi'iek'. Wa, he'^maaxs k-!e'seLaqos on this ■fire. Well, that when you not gwaLlEXLa'laLOL, wa, la'LEs mo'plEnxwa^s^Eml k"!es g'ax you will scream, well, then you four (times) day not come will, q!ula'x'^idELOL. Wa, g-i'pEml^its gwaLlEXLa'laLoL, wa, you will come to life. Well, as soon as you will you will scream, well, 5 la^me'sEnu^x" kwe'xaplELOL qa^s lelE'^'laos. Wa, laE'ms XEk'la'l then we shall strike the so that you die. Well, then you will stay nape of your necks away IcIe'^U la'xaq," ^ne'x-^lae. Wa, laE'm^lawise ^na'xwa ya'qleg'a^e- will be by this," they said, it Well, then it is said all spoke dead is said. da ^we'^woselaga ts!e'daq q!a'q!Ek-a. Wa, la'^lae ^ne'k-a: poor women women slaves. Well, it is said said, they "We'g-a a'Em ha'^lilalax laE'mx- iak!we'masg-anu^x" ne^na'qek* " Go on only do it quickly this we are strong in our hearts qEnu'^x" k*!e'se gwa'LlEXLa'la qEnu'^x" k-!e'sel ga'ial qEnu'^x" for we not scream so that we shall not shall be that we long 10 ga'xei e'tledEl q!ula'x-^ida," ^ne'x'da^x^^Iae. Wa, laE'm^lae shall come shall again come to life," they said, it is said. Well, then it is said they gwa'llla. Wa, la'^laeda no'Enlamala Le'^lalaso^s TsEx^wi'de were ready. Well, it is said the fool-dancers were invited by Fast- Runner qa^s that they la go la go and La'g ililaxa la ya'gudzayaatsa teto'xSvlde, qa*s war-dancers that they carry up the now place of lying tied on board of the Lax"LE'ndEs la'xa lEgwi'le. put it on top on the fire. ga'xeda no'EnlEmalaxs came the fool-dancers when 1 5 gwe'gudza. Wa, winter-dancers. Well, la'e then laE'm^lae Wa, Well, ^na'xwa all g-i'FEm^la'wise as soon as, it is said, qlwa'ga^lila stood up in house the xo'lexulila. Wa, then it is said they were all confused Well, (running about). Eight: Wakashan and Salishan Languages 349 134 laE'm^lawisa it is said the x'i'lplede turned around no'EnlEmala fool -dancers La'g-a^lllaxa took up in teto'x^ide war-dancers qa^s the la'xa in the ogwewali'iaxs rear of house when la'e then they la'xa go'kwe. in the house. Wa, Well. la'^lae it is said they X'i'lpled turned around la'xa at the and they la^stali'lElas went around in house with them o^stall'Jasa door of the g'o'kwe. house. Wa, Well, la'^lae it is said they e't!ed again lag'iyoll'lElaxs went to rear when la'^lae it is said they la'e then they axa'Hlaxa put down the heyak ili'lEla la'xa helk*!ote'waliie. Wa, went back to at the right-hand side. Well, inner room a'lak-!ala teto'x^id qa^s q!ula'Fedeq. Wa, la'^lae La'g-a^lllaxa true war-dancers and hid them. Well, it is said took up the they they ma^lo'kwe q!a'q!Ek*o qa^s le laltla'lii la'xa gEmxote'walile. two Wa, Well, laxs when la'^lae it is said they slaves and went they le^stall'iElas went around with them went out at the left-hand side. qa^s and they le went la'e then la'^laeda it is said the g-a^ll'lEm stopped pled around %Elg'a^li'lEma was stopped ^nEmo'kwe he^stali'lElaya ^nEmo'kwe one one [other] la'xa at the LE'*wa and the was taken around qa's and he la'xa to the q!a'k"o. slave. le'El went Ogwewali'iaxs rear of house o^stali'lElaxs door when la'e when they la'e then they ^nE^ma'x-^it at same time o^sta'li- door Wi, Well, ^weI- was xi'l- turned waxsano'dxa were on each side of the qElxase'lakwe piled-up lEgwi'la. fire. Wa, Well, la"^lae La^na'kulamatsE^wa it is said were placed upright one after another the ma%'x"de qIa'qiEkoxs la'e la'xxana. Wa, laE'm'^lae havve'xa two former slaves when then were put on Well, then it is never they top (of pile). said they gwa'LlEXLalaxs la'e we'qumaxa. Wa, laE'm^lae lelE'^la. Wa, screamed when were shoved Well, then it is were dead. Weil, they down. said they 10 15 350 Vf American Indian Languages 2 135 gi'FEm^la'wise q!u'lx-^idExs la'e ax^e'tsE%a ma^tsE'me as soon as, it is said, burned to ashes then were taken two they xa'xExatsEma qa^s g-a'xe ha'nEmg-a^lllEm la'xa ogwewali'le. small boxes and came were put down on floor at the rear of house, they Wa, he'^latla le'da ^nE'msgEmeda o^sta'lile. Wa, la'^lae Well, that, however, was one at the door. Well, it is said it is said, ax'-e'tsE^a gi'lt!a klipLa'lae qa^s k!ip!e'tsE^we xa'lxEqasa were taken long tongs and they were picked up the bones of 5 ^naFnEmo'kwe q!a'k-o qa^s le k-!ipts!a'layo la'xa ^nal^nE'msgEme each (person) slave and then were put in into the one to each (round) they with tongs xa'xatsEma. Wa, g-i'FEm^la'wise ^wl'^lts!axs la'e yikuyE'ntsE^wa box. Well, as soon as, it is were all in then they were covered said, they yi'ses ye'yik"ya^e. Wa, la'^ae ha'nga^lllEm la'xa ogwewali'lasa with their covers. Well, it is said were placed on at the rear of the they floor ts!a'gats!e. Wa, la'^lae mo'plEnxwa^se ^na'las he gwae'lExs winter-dance Well, it is said, four times day day of that being thus in house. house la'e ya'laqweda teto'x^wide. Wa, laE'm qlula'x-^ldbola. Wa, then sang their war-dancers. Well, then they pretended to Well, sacred songs the become alive. 10 laE'm^Iae a'lax-^ld la ^ya'k'owe TslEx^i'de la'xeq. Wa, la'^lae then it is truly then was beaten Throw-away after that. Well, it is said said TslEx^i'de ^nex- qa'^s le wl'naxa Mo'tsladxwe g*a'sa la'xa Throw-away said that he would make war Nootka going at the go upon the through dzE^la'lasa ^nE'mgese. Wa, la'^lae te'nox%ed la'xa wa. Wa, lake of the ^nE''mges. Well, it is said poled canoes on the river. Well, they g'i'FEm^la'wise la'gaa la'xa t!ex i'las TiE'se la'e yu'dux"sEnd as soon as, it is went to the road of Nootka then in three pieces said, they Inlet they LE'mx'^idxes ^ya'^yats!e qa^s la ^we'k'ilk-ilaqexs la'e e'k'!e^sta split their canoe and went carried it on their they went up they shoulders when 15 la'xa nEg-a', Wa, g i'FEm^la'wise la'g'aa lax was TIe'sc to the mountain. Well, as soon as, it is came to the river of Nootka said, they Inlet Eight: Wakashan and Salishan Languages 351 la'e then they Wa, Well, tlE'mx-^ldxes ^a'^yatsle qa^s yo'Ix-^ide la'xa sewed together their canoe and they drifted down at the g'i'PEm^la'wise la'x'SE^yod la'xa vvaxs as soon as, it is arrived at at the river said, they mouth sex"sale'sEla la'xa tlo'kwaxs la'e a'Em paddled through at a narrow passage then only when they la'EnaLlEme. Wa laE'm^lae arrows. Well, 136 wa. river. la'e then they q!a'yaxaxs were startled when la'e then they ha'nialExsElaso^sa were shot at by lE'^le then it is was said killed TslEx^i'de Throw-away qlu'laxa was alive that LE'we's and his g'a'xe le'Elote. crew. ae'daaqa. back. Wa, Well, Wa, Well, la'^lae it is said he'^me that is the ^nEmo'kwa 5 one q!a'lag-iltsa reason of know- ing of the ga'la first Kwa'g'ul Kwakiutl qexs that lE'^mae they lelE'^la. were killed. Wa, Well. la^me^ then ^ya'k-owe Ts!Ex*l'de la'xeq. Wa, was beaten Throw-away after that. Well, laE'm la'ba. then end. Editorial Note Originally published in Boas Anniversary Volume. New York: G. E. Stechert, 108-136(1906). This paper, a contribution to the Festschrift honoring Boas on the 25th anni- versary of his doctorate, was Sapir's first published work. Student records at Columbia University (Murray and Dynes 1986) show that Sapir enrolled in Boas's introductory course on "American Languages" in 1903-04 while he was an undergraduate in Columbia College, and that he continued with a second course in 1904-05 during the time he was enrolled as a Master s candidate in Germanic Philology. According to the Columbia catalogue for those years, the "translation and grammatical interpretation of Indian myths" formed part of the work of both courses, and it is entirely possible that Sapir edited "The Rival Chiefs" for this purpose. The text was one of those transcribed by George Hunt. Boas's native Kwakiutl collaborator, and the original manuscript is preserved among Hunt's Kwakiutl materials in the Columbia University Library. 352 VI American Indian Languages 2 George Hunt, a native speaker of Kwakiutl, gathered and transcribed large quantities of ethnographic and Hnguistic material for Boas during a 40-year association. Boas first met Hunt during his initial field trip to the Kwakiutl in 1886. In 1893 Boas arranged for Hunt to visit the World Columbian Exposition in Chicago as part of a delegation of Kwakiutls, and their close collaboration dated from that time. To facilitate their work Boas taught Hunt to write Kwakiutl phonetically, and Hunt began collecting texts for transmittal to Boas for editing and publication, with two of the resulting monographs explicitly co- authored (Boas and Hunt 1902-05, 1906). Boas described Hunt s work in some detail, including his orthographic practices, in his introduction to The Religion of the Kwakiutl Indians (1930: ix-xviii). See also Helen Codere, "George Hunt and Boas" (in Boas 1966: xxviii-xxxi). Sapir, impressed by the productiveness of the Boas-Hunt collaboration, established similar working relationships with several native speakers, includ- ing Pete McGuff (Wishram), Tony Tillohash (Southern Paiute), Albert San- doval (Navajo), and Alex Thomas (Nootka). The orthographic practices of these men later formed an important part of Sapir s famous discussion of the psychological reality of phonemes (1933c). Some Aspects of Nootka Language and Culture [excerpt] The Linguistic Relationship of Kwakiuti and Nootka The Wakashan linguistic stock is divided into two main branches, the Kwakiuti and the Nootka or Aht; the former embraces Kwakiuti proper, Xaisla, and He'ttsa^q", the latter Northern Nootka (from about Cape Beale north to Cape Cook on the west coast of Vancouver Island) and Southern Nootka or Nitinat (south of Cape Beale to Cape Flattery). By careful comparison of the two Wakashan branches one can in part reconstruct a Wakashan "Ursprache," but the actual differences between Kwakiuti and Nootka are in fact very great; they differ perhaps as much as Slavic and Latin. [16] As regards phonetics, Kwakiuti and Nootka, while both showing charac- teristic Northwest Coast features, differ rather considerably. The sonant or inter- mediate stop series of Kwakiuti is absent in Nootka, Kwakiuti p and b for instance being replaced by Nootka p. Besides the ^-series, which Kwakiuti and Nootka possess in common, Nootka has a c-series, which is doubtless derived from the Kwakiuti and Wakashan /:--series, which in turn Nootka lacks; thus Kwakiuti g- and k- are cognate with Nootka tc, k-f with tcf, and x- with c. There is no / in Nootka, n corresponding to both Kwakiuti / and n. The velars q! and x, while somewhat infrequently found in Nootka, are not the regular Nootka represent- atives of Kwakiuti q! and x; q! has developed into a peculiarly harsh and choky glottal stop, which I write ^, x into a strangulated-sounding h which I write //, these two consonants respectively resembling Arabic 'ain and ha; ordinary ^ and h are also frequently found in Nootka. As regards phonetic processes, Kwakiuti and Nootka agree in allowing no initial consonant clusters in words; initial Kwakiuti and Nootka ^m, ^'n, ^ w, ^'y, and Kwakiuti ^/ are undoubtedly related to ordinary Kwakiuti and Nootka m, n, w, y, and Kwakiuti / as are Kwakiuti and Nootkap.^ t!, kf, iJ, tsf, qf, Kwakiuti k-!, and Nootka tcHo non- fortis Kwakiuti and Nootka/?, t, k, l, ts, q, Kwakiuti k-, and Nootka tc. In both Kwakiuti and Nootka certain derivative suffixes "harden" the final consonant of the stem; thusp, q, and i, become Kwakiuti /7.^ qf, and ^7, Nootka /;.^ ^\ and ^'y. The "softening" of Kwakiuti seems to be represented in Nootka by but a few stray phonetic processes. Syllabically final glottal stops and glottally affected consonants — such as -^/ and -p! — which are common in Kwakiuti, are entirely absent in Nootka. Medial and final consonant clusters are not as freely allowed in Nootka as in Kwakiuti, / often serving in Nootka to lighten them (cf. Nootka -qEmit, 'round thing', with Kwakiuti -^Einl 'mask'). All final vowels and 354 VI American Indian Languages 2 Stopped consonants in Nootka are aspirated. Peculiar to Kwakiutl is the change of A:-stops to spirants (x, x", x-) before consonants, whereas in Nootka they remain; in this point Nootka seems more archaic than Kwakiutl. In general morphology Kwakiutl and Nootka are quite similar, [17] despite numerous differences of detail. In both the stem is, as far as its meaning allows, indifferently verbal or nominal and one or more suffixes are required to give rise to definitely verbal or nominal complexes; in Nootka a suffixed-'/' is often used to substantivize a verb form. Both Kwakiutl and Nootka are absolutely devoid of prefixes, most of the elaborate grammatical mechanism being carried on by means of suffixes, to a lesser extent by means of initial reduplication, and, in Nootka, consonantal changes. The suffixes of Nootka and Kwakiutl express similar ideas and are used in more or less parallel fashion, though the number of suffixes that are etymologically related form but a small percentage of those found in either; so far about ninety Nootka suffixes have been discovered that are entirely or in part cognate to Kwakiutl suffixes. Examples of local suffixes shared by Kwakiutl and Nootka are: Kwakiutl -o-^yo 'in the middle', Nootka -^'win'; Kwakiutl-A7e<7"'in the corner', Nootka -nikw-; Kwakiutl -atCls 'down river', Nootka -atis; Kwakiutl -tsfo 'in', Nootka -ts!o'; Kwakiutl -k-E 'top of a box', Nootka -tci' 'full'; Kwakiutl -!a^ 'on the rocks', Nootka -!a^a'\ Kwakiutl -es'on the beach', Nootka -is; Kwakiutl -/7'in the house', Nootka-//; Kwakiutl -xs 'in a canoe', Nootka -qs, -lans. A few examples of body-part suffixes are: Kwakiutl -Ids 'cheek', Nootka -as; Kwakiutl -xo 'neck', Nootka -as-Haul 'chest'; Kwakiutl -dp! 'neck', Nootka -dpfaf 'back'. Important temporal ele- ments held in common are: Kwakiutl -l 'future', Nootka -^dq-L, -^il; Kwakiutl - x-^ld 'inceptive', Nootka -ci-L. There are some striking agreements in verbify- ing derivative suffixes, as: Kwakiutl -lexst 'to desire', Nootka -Hh" 'to try to get', -st!iH^ 'to have as goal'; Kwakiutl -!a 'to go in order to', Nootka -fas; Kwakiutl -k-!dla 'to make a noise', Nootka -^en' (= Wakashan *-q!Ela); Kwakiutl -g-a^i 'beginning of a noise', Nootka -^aL (= Wakashan *-q!a^l)\ Kwakiutl -qlES 'to eat', Nootka -lis; Kwakiutl -nuh' 'to have', Nootka -nak\ Examples of nominal suffixes are: Kwakiutl -aano 'rope', Nootka -d^nul 'long'; Kwakiutl -gas 'woman', Nootka -^aqs; Kwakiutl -asde 'meat', Nootka -act' 'dried meat'; Kwakiutl -mis 'useless', Nootka -mis 'mass'; Kwakiutl -p!e-q [18] 'stick, tree', Nootka -p.'U 'long board-like object', -q- 'tree'; Kwakiutl -(x)^'Enx 'year, season', Nootka -q'^itcH" 'year', -fitcH" 'season'. On the whole it seems that Nootka has a rather larger number of derivative suffixes than Kwakiutl, many quite special ideas being expressed by means of suffixes where there seem to be no Kwakiutl equivalents. A few examples are -a/ 'blanket'; -'m//' 'son';-^^ 'daughter'; -Htut 'to dream of; -!d^il 'to ask for as a gift in a girl's puberty ceremony'; - tfota'' 'to give a potlatch for'; -yaqn'^ 'to sing a song'; -Hl 'to begin to sing a song'; -lini 'to give a feast of; -Hd' 'to buy'. Both Kwakiutl and Nootka make use of two kinds of reduplication, one in which the first consonant, first vowel, and second consonant of the stem are repeated, and one in which only the first consonant and vowel are repeated; the 1. ! denotes a "strengthening" of the preceding consonant. Eight: Wakashan and Sallshan Languages 355 former type is employed in forming iteratives, the second in forming plurals or distributives and with certain suffixes (such as Kwakiutl -fa, Nootka -fas 'to endeavor, to go in order to'; Kwakiutl -^ydla 'to go to look for'; Nootka -////" 'to try to get'; Nootka -kfok'" 'to look like'). In Nootka the repeated vowel is in all cases the same as that of the stem, in Kwakiutl the second type of reduplication has a definite vocalism {e in some cases, a in others) in the reduplicating sylla- ble. In Kwakiutl verb stems ending in vowels inserts- after the first, k- after the second syllable of the iterative, while Nootka iteratives of like form insert l and y; Nootka sd- 'to crawl' forms iterative sd' Lsdtc, -tc being probably identical with Kwakiutl -k-. One other striking resemblance of detail between Kwakiutl and Nootka may be noted: both Kwakiutl diminutives in -Etn and Nootka nouns in -kwin' 'toy' require reduplication of the stem. In regard to pronominal development there is considerable difference between Kwakiutl and Nootka. While there is, practically speaking, but one series of personal pronominal suffixes in Kwakiutl, there are three in Nootka (represented, for second person singular, by -e^its, -k\ and -sok'), of which the second and third are etymologically related: the first Nootka series is used in indicative forms of verbs, the second in subordinate clauses, interrogatives, and possessive forms of nouns, while the third seems to be confined to [19] certain modal forms. Kwakiutl has distinct forms for first person plural inclusive and exclusive, while Nootka has only one form for both. Pronominal objects are, to at least a considerable extent, incorporated in Kwakiutl; in Nootka, however, only in the case of the first person (second series) of the imperative. A great degree of complexity in pronominal forms is brought about in Kwakiutl by the combination of the pronominal affixes with syntactic (sub- jective, objective, and instrumental) and demonstrative elements. Nootka has none of this syntactic and demonstrative complexity of the pronoun, but a series of forms is found built up of the second pronominal series and an element -tc implying that the statement is not made on the authority of the speaker. Almost all Nootka and Kwakiutl words are noun or verb forms, there being almost no particles properly speaking. Such apparent Nootka conjunctive and case particles as ^onoi. 'because', ^oyi' 'when, if, and ^okwil 'to' are mor- phologically verb forms built up of a stem^o- 'a certain one, thing' and deriva- tive verbifying suffixes. There is, however, in Nootka a syntactically important conjunctive element ^ani' 'that' to which may be appended pronominal affixes of the second series and which may perhaps be considered a particle in the proper sense of the word. The "empty stem," Nootka ^6-, is cognate with Kwakiutl 6- 'something,' which, however, is used primarily in noun forms. Other Wakashan "empty stems" are: Nootka ^ap-, ^am-, Kwakiutl dps-, used chiefly in forming nouns of body-parts that occur in pairs, and Nootka hit-, hi- 'to be at', Kwakiutl he- 'that'; peculiar to Nootka is hin, hit- (before "harden- ing" suffixes) 'to be or do (as indicated by derivative suffix)'. In regard to vocabulary Kwakiutl and Nootka differ greatly. Considering the very striking morphological agreements between them it is somewhat disap- pointing to find comparatively few stems held in common. It is highly impor- tant, however, to note that many of these are rather colorless in content and 356 Vf American Indian Languages 2 thus hardly to be suspected of having been borrowed in post-Wakashan times. Such are Kwakiutl ^nd-, Nootka ^nds 'daylight'; Kwakiutl g-dl-, Nootka tcdn- 'to be first'; Kwakiutl ax- 'to do, be', Nootka 'o//- 'to be'; [20] Kwakiutl we-, Nootka wi-, wik" 'not'; Kwakiutl g a 'S C b ■J rt« .9 s •■ £ 1 «,7 "3 0 1 "S GJ 0] •*j 1 "O a fl •g t£0 5 0 b (U § "S'm d 3 d ■0 1 a CO 0-S So 0 " 0 a 3 ■3 fl •c 3 ■3 ft 0 ~ 3 1 V.S *^ e •« i .c c rt u S Si s 5 Q ':; i Js^a ^ '^. ^ s 5 c 3 "a ■:3 ■^ "a ":4 ^ ^ e S 1 > 1 > •2 §"* ■'1 •1 § fl fl 0 0 ■5 S fl OS J) 0 73 •»' • ?"^ g 0 :" ^2 >• fl _fl §^' TJf; •2^ "1 > i 1 ■-T3 •Beg 3 3 .9 0 " 3 "„, *• :§l.8 .tJT3 3 fl ■5t) 3 d _, > "1 > "0 a ■ft *ft 08 2 •^-0 '3 * 3s. '•2 ■■V a B ■ — aj 3 a " •oSg 3 v> 3 fl a 5 1 > g" '.a ■♦^ _fl 2 "3 > J3 T3 ■5 ^ fl 3 i > S 3'S 9> .. " 2 S'S' •0 ■0 ■0 T3 ?-g5 ?-g2 •a 4) t-H 1 ■3 I •3 •3-0 C9 gg§ S 0 fl { < -»! J<_ < -«! < ««! _■< M ^ H_ W ,H_ H H Pui E 0 i° .2 fl ■ g a T3 0 ^ 5 1 jS £a SS 0 '■3 '•z K. 1 a: 5 01 .= 1-1 _ "J Jo a Xi 0 fc OS 0. 1 '•5 i g^-3 3- t5 1 .2£ fc> c U) 0 c ^ 'c i 0 V r c — V e* w 0 OS a 3 0 ^ u sg> $ *" u 0 a-^ a 0 E V D. "0 a a >. H ft*) al 3 — 0 S.-H it c « 0-5 a ill 0 a 0. a "5 i 15 1= < g-3 2 OS 0 13 3 K fl 0 1h 0. V E d g 2 «) a ■g T) fl ot XI I 0. I a 3 u fl 0 £ 4) a ■a V 0 OS 0 0 1 Eight: Wakashan and Salishan Languages 379 Editorial Note Originally published as Memoir 62, Anthropological Series 5, Geological Survey, Department of Mines, Canada. Ottawa (1915). Although the formal patterns of consonantal symbolism, particularly dimin- utive/augmentative devices, have been much discussed in Americanist work (for an excellent survey see Nichols 1971), the rhetorical and literary uses of these have been noted only sporadically. Some attention has been given recently to the phonetic stereotyping of the Coyote and similar figures in myth (Aoki 1975: 190; Toelken 1969). Hymes's study of the expressive resources of Takelma traditional literature (1979; revised in Hymes 1981: 65-76) is closest to the spirit of Sapir's work. Noun Reduplication in Comox, a Salish Language of Vancouver Island INTRODUCTION. One of the most characteristic grammatical processes of a group of Northwest Pacific Coast languages, embracing the Tsimshian, Kwakiutl-Nootka, Salish, and Chemakum linguistic stocks, is initial reduplication, employed in both noun and verb forms to indicate a variety of grammatical concepts, chiefly those of plurality, distribution, and iteration. The Salish languages in particular are known to make exuberant use of reduplication for grammatical purposes, but the subject, which seems to bristle with irregularities and intricacies of detail, has never been adequately treated for any of the numerous dialects of the stock. Indeed, a thorough grammatical study, at the same time phonetically adequate, of a Salish language, is still one of the desiderata of American linguistics. During the autumn of 1910, while prosecuting ethnologic and linguistic research for the Geological Survey of Canada among the Nootka Indians now living in two reserves near Alberni, B.C., opportunity was incidentally found to gather some linguistic data on Comox, a Salish language spoken on the east coast of Vancouver island near the present town of Comox. The dialect represented in these notes seems to be 382 ^f American Indian Languages 2 Comox proper (Q!6mgx"s), with which Uohos, spoken on the mainland of British Columbia, was stated to be identical. SdlolV'' was stated to be a northern dialect of the same language. This term is evidently identical with Boas' Qatloltq, which he uses to apply to the most northern group of Coast Salish tribes, excluding Bella Coola, inhabiting "Discovery Passage, Valdes Island, Bute and Mala-^pina Inlets."^ Boas adds, "The Catl61tq are called K'omoks by the Lekwiltok'" (southernmost Kwakiutl tribe.) The informant was Tommy Bill, an Indian of mixed blood, whose father belongs to the Ts'.ictiatH" tribe of Nootka Indians, while his mother was a Comox, he himself living with and being to all intents and purposes a member of the Hopdtdas'atH" tribe of Nootkas. His knowledge of Comox was obtained in his earher years, when living among his mother's people, whom he visits from time to time; it is only fair to add that he speaks mainly Nootka and English nowadays and does not claim to have a perfect command of Comox. However, the rather elementary character of the data obtained, together with convincing internal evidence derived from their study, leaves no room for doubt as to the essential accuracy of the material here presented. Most of the time spent on Comox was taken up with securing material pertinent to the problem of reduplica- tion in nouns. For most of the nouns obtained, plural, diminu- tive, and diminutive plural forms were secured, all of which involve various types of reduplication. Our linguistic material thus naturally divides itself into three heads, not to speak of a small number of nouns that are always used in reduplicated form. A few introductory remarks on Comox phonetics and some supplementary data are also added. I. PHONETICS. Vowels. The short vowels found in the Comox material secured are: a (as in German Mann); a (as in English bat); e (short and open as in English 7net); e (short and close as in French ete); i (short and open as in English hit); i (short and close as in French fini); o (short and open as in German dort); ' See F. Boas, First General Report on the Indians of British Columbia, Report B. A. A. S., 1889, 5th Report on North-Weatern Tribes of Canada, p. 10. Eight: Wakashan and Salishan Languages 383 0 (short and close as in French beau); and u (short and open as in English 'put). Of these vowels, e, i, and i are etymologically one sound, which is modified by phonetic surroundings; simi- larly, o and u. Velar consonants tend to lower preceding or following i to e (possibly sometimes e), while certain consonants (particularly s and I) tend to palatalize i to i. e and o, which latter does not occur often, are doubtless etymologically related to e and o respectively, but seem in every case to be clearly kept distinct from these, a is not common. Corresponding to each of the short vowels is a long vowel (long d, however, has not been found). These are indicated as: d (as in German Bahn); e (long and open as in French mere, or as in English hear, but without "r-vanish"); e (long and close as in German See); i (long and open as in English beer, but without "r-vanish"); i (long and close as in English see); 6 (long and close as in English roll, or as in German Sohn) ; 6 (long and open as in English born, but without "r-vanish"); u (long and close as in English rule) ; and u (long and open as in English poor, but without "r-vanish"). Similarly to the corresponding short vowels, and under parallel phonetic circum- stances, e, %, and I are variants of one sound, etymologically speaking, though i is often to be interpreted as lengthened form of inorganic vowels, in which case it does not seem to vary with e and l; 6, u, and u are likewise representatives of what is etymologically a single sound. 6 does not often occur; it is probably etymologically related to 6. e occurs often and cannot be considered a mere variant of e. As not infrequently happens in American Indian languages, the long vowels are not always held out with even stress, but end with short rearticulations which give the whole vowel in each case a quasi-diphthongal effect. Such vowels have been noted by the writer in Takelma, Southern Paiute, and, at least to a moderate extent, in Nootka; Boas has noted them in Tsimshian. While they occur to a considerable extent in Comox, they cannot as in Takelma be considered the normal forms of the long vowels; sometimes the short rearticulations seem to serve as glides to following consonants, particularly velars. The quasi-diphthongal long vowels are here indicated by long vowels followed by superior short vowels, the vocalic 384 VI American Indian Languages 2 quality of the latter being indicated as in normal short vowels. There are found: a"; e%' e^; e' (occurs before anterior palatal consonants) ; 1'; i'; i^ (occurs before velar consonants) ; 0° and o"; and w". A number of cases also occur of short vowels followed by weak rearticulating vowels; such are e", q^, and i' (here the * is a glide to the following velar consonant). Some of these may well represent secondarily shortened long vowels. Differing from such long or short vowels with quasi- diphthongal character are vowels that are secondarily diph- thongized by a vocalic glide whose timbre depends wholly on the following consonant; such is i" in /cwpw"mi"a;" "hill," in which the second " is a glide due to the u- timbre of the final consonant. Short vowels of somewhat obscure quality are also found, either representing dulled forms of normal short vowels or being of inorganic origin and meant to lighten consonant clusters or serve as glides. Such vowels are: a (as in English but, yet sometimes less clearly marked in quality), which is sometimes inorganic, sometimes dulled from a; e (obscure vowel with e- quality) ; and / (very short rather unclear i). At times short vowels are so weakly articulated as to be barely audible; these are rather "murmured" short vowels of etymological significance than merely glides, timbre-echos of preceding consonants, or voiceless vowels. Examples are: ? in Idl^bo'm' "small clam" (-^?6- reduced from ld'"b- in I6"°bnm' "clam"; yet in this case ? can just as well be morphologically dispensed with and phonetically explained as a timbre-echo of -ol-); ^ in qt'w^x "steel-head salmon" (that ^ is organic, despite its dull quality and extreme brevity, and reduced from a, is indicated by Nootka qe'wan "steel-head salmon," with which Comox qt^w^x is evidently identical; borrowing has doubtless taken place); "^ and " in hew^qen" "swan" and its diminutive hew'^qAdol. Another class of "murmured" vowels (German 'Murmel- vokale") is formed by weakly articulated, yet not voiceless, vowels occurring in syllabically final position after glottal stops ('). Such vowels are only in part "murmured echoes," i.e., reduced repetitions of immediately preceding fully voiced vowels (such are a'", e'*, V\ ai'\ o'?, d'"; vowel breakings of this type occur often in American languages) ; in some cases we have Eight: Wakashan and Salishan Languages 385 also murmured vowels after glottal stops that are of different quality and etymologically distinct from immediately preceding vowels (such are a'' and a'O- Some consonants, notably glottalized ("fortis") consonants, are apt to be followed by timbre-echoes dependent in quality on the preceding vowel. This simply means that the oral resonance chamber characteristic of a vowel may, failing to be materially disturbed by the following consonant position, linger on and thus become acoustically noticeable as a voiceless (sometimes aspirated) vocalic echo; if the consonant is a spirant, the vocalic timbre may be audible during its production. Examples of such unaspirated timbre-echoes after glottalized consonants are: " in pld'alats!'^ "skunk" and ° in k!6°ddt!° "porpoise." In Id^g^et!" "herring" the t! was heard with definite a-timbre despite preceding e. After u (o)-vowels syllabically final A;-sounds are regularly followed by echoes (aspirations when consonant is not glottalized) with w-timbre. Hence A;'", A;.'", X", q'", q!", and x" (see below for orthography of fc-sounds). These sounds, however, are also very frequent after unrounded vowels, as in Id'^dak'"' "skin;" in such cases they represent original labialized A;-sounds (see below). Aspiration with definite w-timbre is also found after t, as in sdW" "woman." Excluding such inorganic diphthongs as are formed by vowels and following glides (e.g., i"), there have been found as true short diphthongs ai, au (also au), di, ei, and long diphthongs di, du. Vowels normally forming diphthongs that do not so unite, each preserving its full value, are separated by . (thus, a.i as distinct from true diphthong ai). Stress accent is indi- cated by ' over vowels. Consonants. The consonant system of Comox is fairly full, including, as it does, eleven distinct series that differ according to place of articulation. As regards manner of articu- lation, six distinct series are to be recognized (voiceless stops, glottalized or "fortis" stops, voiced stops, voiced nasals, voiceless spirants, and voiced spirants), though by no means all of these are represented for all places of articulation. The voiceless stop and glottalized stop series are complete, the voiceless spirants nearly so, while the others are quite defective. All these consonants may be represented in the form of a table : — 386 VI American Indian Languages 2 6 Laryngeal glottal) Velar Imbialiekd velar Guttural Labialized guttural Pbe-quttural (anterior pala tal) Dorsal lateral Palatal sibilant Alteolar aiBILANr Alveolar Labial Voice- less 8TOPS Q qw k kw kv Glot- TALIZED STOPS q.'w k! k.'w tr! is! a Voiced STOPS Nasals I (voiced contin- uant) dj id) (b) Voice- less spirants Voiced spirants c is pronounced like sh of English ship; x^ like ch of German ich. tc, td, dj (like j of PJnglish jam), ts, and ts! are affricatives (stop plus corresponding spirant; no simple stops correspond to -g''-. Just as g^ and w are related, so there is reason to believe that dj and y are related, though there is perhaps not quite as con- vincing internal evidence at hand. See Type VIII of plural formations for such evidence. Moreover, with Comox djidis "tooth" compare KwantlEn yenis "tooth;"^ with Comox djicin* "foot" compare Siciatl yicin.^ * See C. Hill-Tout, Ethnological Studies of the Mainland IlalkomilBm, a division of the Salish of British Columbia, Report of British Association for the Advancement of Science, 1902, Ethnological Survey of Canada, p. 65. ' See F. Boas, Kwakiutl, Handbook of American Indian Languages, Bulletin 40, Bureau of American Ethnology, 1911, p. 447. ' C. Hill-Tout, ibid., p. 64. * F. Boas, Comparative Vocabulary of Eighteen Languages spoken in British Columbia, Report of British Association for the Advancement of Science, 1890, 6th Report on the North- western Tribes of Canada, p. 148. ' C. Hill-Tout, Ethnological Studies of the Mainland HalkomilBm, a division of the Salish of British Columbia, Report of British Association for the Advancement of Science, 1902, Ethnological Survey of Canada, p. 86. ' F. Boas, Comparative Vocabulary of Eighteen Languages spoken in British Columbia, Report of British Association for the Advancement of Science, 1890, 6th Report on the North- western Tribes of Canada, p. 14' Eight: Wakashan and Salishan Languages 389 9 II. NOUNS NORMALLY REDUPLICATED. A considerable number of Comox nouns always appear in reduplicated form, reduplication in these cases being of no grammatical significance, but belonging to the noun as such. Many of them are animal names, and of these some are quite evidently onomatopoetic. Ten fairly distinct types of redupli- cation seem to be illustrated in the rather limited material available. Very likely others exist. Type I. Completely Reduplicating. hd^mho^m blue grouse xop'xop' humming-bird k'^dck'^ac bluejay tsU'xHsHx" fish-hawk pok'^'pok"'' liver gH'gH'- panther qe'n'qen"^ duck qwVqwi' sea-gull "Duck" and "sea-gull" have both syllables with vowels alike in quality but with short vowel in the second. Type II. Completely Reduplicating with e. te'ltoV small butter-ball duck hd'ihei' arrow "Arrow" belongs perhaps rather with Type I. Both of these nouns lose a glottal stop in the reduplicating syllable. Type III. Reduplicating Syllable: cvci.' titctitcV c owl kiva'kwd'^djo' grey squirrel t.'Aq't.'Aqdi dog-wood Type IV. Reduplicating Syllable: ce. mf'mau cat k^It'k^Idk^! crow td'itca.iq' salt-water hunter In "salt-water hunter" reduplicating icl- is broken into tcl'i-. ' In these formulae c represents first consonant of stem, v first vowel, ci accond consonant of stem, vi second vowel, and so on. V represents any long vowel, <■ any shortened vowel. 390 VI American Indian Languages 2 10 Type V. Reduplicating Syllable: ci. Only one or two certain examples have been found of this type. They differ from the preceding in that the vowel of the redupli- cating syllable is short. qwi'^qwdH! .dd'^^k' butterfly we'wdlos young man (form probably diminutive in). Possibly also: — e'ddjam' young woman Type VI. Reduplicating Syllable: cd or ca. LdLdpx pocket-knife qwdqumi^s marten xdxe'* nit mdmstco^m mink Type VII. Reduplicating Syllable: cv. ts!ats!dwicin' hail tdatddH'.dn" mouse xwdxwadjo'm fly (word probably diminutive in form). q.iq'td'amas game with wooden ball^ qoqgwVm^ down (of bird) Type VIII. Reduplicating Syllable: cv'. qd''^qa'^ rush mat djd'''dja' tree Type IX. Reduplicating Syllable: eo. Only one example has been found of this type: — tofxHal necklace Type X. Reduplicating Syllable: cec. Of this very peculiar type (doubly reduplicating consonant, otherwise like Type IV) also only one example has been found: — qliqlq.'d'adje'uk''^ butter-ball duck ' Formed from q'td'ahas "wooden ball covered with spruce-roots." There were two sides in the game, with the same number on each. Each side had a goal consisting of a little pit, which was guarded by one man. All but the two guards gathered in the centre. One man threw up the wooden ball and everyone tried to catch it, run with it to the goal of the opponents, and put it into the pit. Those of the other side tried to take the ball away from the one that had it. The side that 6rst made ten goals won the game. After four goals had been made, the game was suspended for a while and a general free-for-all fight took place. Eight: Wakashan and Salishan Languages 391 11 Here may also be given: — qldq.'tux'* big fire (form is augmentative?) : scattered around. cf. q!6tix'' fires III. REDUPLICATED PLURALS OF NOUNS. By far the larger number of Comox nouns form their plural by reduplication, in a few cases different stems are used for singular and plural, while still other nouns seem to form no plural. The most persistent type of plural reduplication is that in which both first and second consonants of stem are repeated, though 'ess numerously represented types also occur. Type I. Reduplicating Syllable: cvci lAkoni"^ beaver kuvidqin^ sea-lion qwAdPs humpbacked whale qwASAm woolly grouse xop'xdp' humming-bird ts!oxd"° codfish LlAxwd'"' dog salmon sd^arC cohoe salmon qlwAt'lHcin" humpback salmon xd'd big clam Lp'Am' cockle xApd'^ red cedar qo'^'a''^ hemlock q!dp!xwai oak p'.tHxdi alder t.'t'ibdi wild cherry bush 'dwdk''' tobacco qfwA'ix wood xd'a.idatc stump plural tlAk^HlAkom"^ kumkumdqirC qwxd'qw Adit's qwAsqwASAm xop'xop'xop^ ts!6xts!oxd'° lIaxlIaxwo''- sd^'^sa'an" qlwAVq'.wAVlHcirC xd'^xa'd Lp'Lp'Am'^ (type viii?) xAp'xApd^"^ qo'^qd'^'a'' q!ap!q!dp!exwai (with lengthening of first stem-vowel; -e- is in- organic) p.'e'p.'t'ixdi (type viii?) HeH.'t'ibdi (type viii?) 'au'dwdk'^ many bun- ches of tobacco qlwAiqfwA'ix xd'''xa'a.idaic 392 VI American Indian Languages 2 12 mAqsm nose djiciti^ foot djidis tooth dikuiriKS heart xAucirC bone k^it! little finger tslAmdla" index finger qlwdt'Am river pdxai' creek Lfdqe'nac spring kupu^'mi^x'' hill Lldxai' old man qAl'q! warrior L.'Ams house ajiSAm box kwd'am coiled storage basket L.'pdtjl basket bag q!dk'" board k^Iik^dyu oar SAq'Ak'" war-club lAq!" bow tcHCqdmin knife sipfAvitn" shinny stick lAq'.AS mountain-goat blanket L.'pi'ts.'d''^ yellow-cedar q.'As'Addi buckskin shirt Lldq.'acin" moccasin pdq'dos white-eyed tdxdos red-eyed plural niAqimAqsin' djicdjicirC djiddjidis L.'^k'L.'ikuinAS xAuxAUcirC kHV.kHt! ts! Amis! AmaW q!wdVq!waVAm pdxpaxai' ildq'L.'aqe'^nac kup'kupumt"x" (with shortening of second stem-vowel) LfdxLlaxai' qAl'qAlq! LfAmL.'Ams xAsxASAm kwd'^kwa'am l! Ap' l! Apdtil qlak'^'qlak"" k^Hk^'k^Hk^dyu sAq'sAqAk'"" lAq.'HAq!" tcHVtdit' qdmin" siplsiplAmi^n' lAqUAq'.AS L.'Ap^LfApitsfd'" q! Asq! As' addi LlAq/LlAqfadn'^ pdq'paq'dos tdxtcixdos Type II. Reduplicating Syllable: cac. This type differs from the preceding in that, while both first and second stem-consonants are reduplicated, the stem vowel between these consonants is not, but is replaced by an inorganic A-vowel. If the vowel is followed or broken by a glottal stop, or if there are two successive vowels, the second consonant is Eight: Wakashan and Salishan Languages 393 13 repeated just the same, the glottal stop being neglected in the reduplicating syllable. Thus, tc'.e'dd- and L.'d'al- reduplicate as tdin- and l!aI- respectively. Several nouns with stem-A and reduplicating-A, listed under Type I, should perhaps belong here. Three sub-types are to be recognized, according to whether a remains as such (sub-type a), is palatalized by s, tc, td, k^, I, or y to i (/) (sub-type 6), or is labialized by xw to u (sub-type c). Sub-type II a. mi'xdl bear L/d^ard'm' wolf qld^L.' land otter qld^sa" sea otter xd'^'wa fur seal dsx" hair seal k!6°ddt!° porpoise pIdqiAddtc goose qen^qerC duck hew^qen' swan qt'w'^x steel-head salmon td^qfwa' devil-fish mdt'.di horse clam sd^^ba" mussel mdHdirC louse dsd'i huckleberry bush xwdsAbdi soapberry bush t'.e'^^de'qwai salmon-berry bush t/d'abuxwdi gooseberry bush qex" ring finger bld^qlwdi fish-gill sopAdatc tail tsldmuql cloud plural mAxmi'xdl LlAiLld'aVd'nC qlALlqld^L! qlAsqld^sa' xAuxd'wa ^As^dsx'' klwAd'k.'wd^dot!'? (with shortening of second vowel of stem) plAqlp'.dq'.Addtc qAd' qeri' qerC hAuhew^qerC qAuqt'w^x tAqftd^qlwa' mAt.'mdHldi (with length- ening of first vowel of stem) SAmsd^^ba^ mAtdimdHdirC ^ As'dsdH xwAsxwdSAbdi t!Ant!e'''de'qwai t! Amt! Amuxwdi (with re- duction of d'a of stem to a) qAx^qex^ L!Aq!''L!d''q!wdi sAp'so'^pAdatc tslAmits'.dmuql 394 VI American Indian Languages 2 14 tld'^qfaV mountain st'qeV dug hole, well td'mic man xd''p! baby basket t.'o'mV paddle waxdHsU pipe toVxHal necklace q'td'ahas wooden ball used in game mitdli beaver-tooth die plural t!Ag!t!d''q!aC sAq'se'qeV tAmto'mic XAplxd^p! V.AmV.Q^mV wAxwaxdHsli tAt'oVxHal qAfq^td^ahas mAfmiHdli (with length- ening of first stem vowel) qAq'qd^^qa" lAqfHdq.'wdinop' qd'^qa rush mat Idq.'wdinop cedar-bark mat L.'dxe oldest dAXhldxe Lldlsdmi strong LfAlLldlsdmi An irregular example of this sub-type is: — sdts'.Am tyee salmon SAmsdHs! Am Here the first and third, instead of first and second, consonants are reduplicated. Sub-type II 6. tde'ddg dog k^dck^dc bluejay l6"''honC small clam tsfdtc.'ilbai spruce sdsin'^ mouth sdpdxos horn ko^sAd' star ydxai^^ pack-basket tcHntde'ddo k^ick^dck^dc lnnld'^°hoirC taHtc'ts.'dtcHlbai {tsUtc"- instead of tsUtc!-) sissosirC (with shorten- ing of second stem- vowel) sipsdpdxos kwiskosAd' yixiydxai^^ Irregular examples of this sub-type are: idatddH.'dn" mouse Itx^sal tongue tclUHddH.'dn" (for tdiC- instead of tdit.'-see "spruce" above) Hsiix"sal Eight: Wakashan and Salishan Languages 395 15 In the first of these the plural is built not on the already redupli- cated simplex (as e.g., in "bluejay" above), but on a simpler unreduplicated stem abstracted from it. In the second example the first and third, unstead of the first and second consonants, are reduplicated (cf. "tyee salmon" above). Sub-type II c. Only one example is available: — xwdtoqo'm "falls" plural xuCxwdtoqo^m Type III. Reduplicating Syllable: co or co. Nearly all of these nouns have g" as their second consonant, representing, as we have already seen, original w. These nouns could be considered a sub-type of Type II, were it not that they form their reduplicating syllable not in -au, as might perhaps be expected (cf. xAuxAUciii'^ under Type I), but in -o- (-w- after dj- and g^-) or -o- (probably due to contraction of original -yiw-). Two sub-types can be recognized, according to whether the reduplicating vowel is short (sub-type a) or long (sub-type 6). Sub-type III a. td^ag^ax"^ fern td'agHn salmon spear Sub-type III b. Id^g^et!" herring pleg^di halibut gT/figvp panther t.'cg^em sun, moon Mg^os chief djigHn' song Id^dak'" skin totd'ag^ax'' tqtd'agHn Idla^gyet!" p!d'^p!eg^di *g^ug''Vg"l^ (not obtained as such, but implied in diminutive plural gyVg^ugH'gH' "pan- ther cubs") tldH.'eg^em sun and moon hd'^he'gvqs djudjig^in' loHd^dak'" It is not clear why "skin" should reduplicate with o-vowel. 396 ^I American Indian Languages 2 16 Type IV. Reduplicating Syllable: cv; Syncope of First Stem Vowel. Only one example has been found of this type. As it begins with g"-, the stem -g^ of the plural, coming immediately before another consonant, reverts to it;, uniting with preceding a to form au. g^dq'dhas married woman plural g"duq'dhas That *wdq'dhas is to be presupposed is corroborated by com- parison with KwantlEn s-wd-wskus "married woman. "^ Type V. Reduplicating Syllable: cvc. Nouns belonging to this group have long stem-vowels and differ from Type I in that the reduplicated vowel is shortened, though it keeps its quality. xdug"as grizzly bear xduxdug^as qd'um' eye qduqd'um' qd^'mai" snow on ground qiirnqo"' mai' }6kd°min bailer luk'loko'^min Type VI. Reduplicating Syllable: caci. ti'hd''ddn' chief's wife tdhtihd''ddn' heq'sd^min pole for poling canoe hdq'heq'sd°min' 6lqai'' snake 'dVolqai'' (with shorten- ing of first stem- vowel) aL leggings 'dt^aL "Leggings" may, of course, just as well belong to Type I. Type VII. Reduplicating Syllable: cv. q!6a'dda ear q!oq!oa'dda tt (L.'.-ims) big (house) titl (LfAvis) big (houses) dx" snow-flake a'dx" falling snow ' C. Hill-Tout, Ethnological Studies of the Mainland HalkomilBm, a division of the Salish of British Columbia, Report of British Association for the Advancement of Science, 1902, Ethnological Survey of Canada, p. 89. Eight: Wakashan and Salishan Languages 39ri 17 Type VIII. Reduplicating Syllable: ce. According to varying phonetic circumstances we have either I or e, the latter occurring after q, g! and x. The examples of this type obtained are: — qldik''' eagle kwudjak''^ trout ft'x" yellow cedar djd^'dja' tree sd'idJA^ leaf tcdyac hand sayd^ada neck qd^ya' water s&'yal lake xd'adjaic stone tcu"i child k.'oyokobPn (or -mVd) fisherman sidjdqo'p' basket hat IdidatdAn woman's cedar-bark skirt plural q!t'q!dik'" kivPkwudjdk''^ tltPxwai' (may belong also to type vii; note -ai' in plural) *djidjd'''dja" (not ob- tained as such, but implied by diminu- tive plural djedjidjd- '"dja) stsa'idJA' tcitcdyac stsayd'ada qtqd'ya^ stsd'yal xtxd'adjaic tcttcu'i kfwtk! oy okomi 'n stsidjdqo'p' ItldidatctAn Eliminating "yellow cedar," which, as was pointed out, may just as well be reckoned as belonging to Type VII (there is reason, however, to believe that il'x" goes back to *tiyix"; see diminutive type I b and diminutive plural type II f), all these plurals may be plausibly explained as cases of Type II, redupli- cating -I- or -e- being the contracted result of -Ay-. It will be observed that the stems of these nouns contain either i- diph- thongs, including broken groups (-di-, -a'l-, -u'i-), vowel plus y {-ay-, -ay-, -a'y-, -oy-), or vowel plus dj {-udj-, -d'^dj-, -d^adj-, -idj-) ; dj, as we saw above, is probably a resultant of original y. 398 VI American Indian Languages 2 18 Type IX. Reduplicating Syllable: cd (or ca). Sub-type IX a (with d). tdd rain plural tcldtclel qd'"qwai speaker qiuaq6'"qwai ylp'i'x" hole ydyipl'x'* Sub-type IX b {with a). tct'itca.iq^ salt-water hunter tcatcVitca.iq* Type X. First Stem-vowel Changed to e. These nouns are reduplicated to begin with, and substitute for plural reduplication a change of the first stem-vowel to e (long and open). The few examples are: — we'wdlos young man we"wdlgs e'ddja7n' young woman e"ddjam' ky.'t'ky.'dky! crow k«!e''k«!dk«! Type XI. Reduplicating Syllables: cdCAC. Only two examples have been found of this doubly reduplicat- ing type of plural formation. In the first, the a, coming after g", is palatalized to i; in the second, the reduplicating -Ay- becomes -I- (see Type VIII). g^d^di'm slave gyagndg^ddPrn tdyac killer-whale idtltdyac Irregular Plurals. Several plurals listed above are somewhat irregular, but there has been no difficulty in assigning them to definite types. The two that follow are quite irregular. The second shows not only reduplication but breaking of -a- to d'a-. djd'°dja' tree djadj'id'm inAVq" fawn mamd' aliq' " Eight: Wakashan and Salishan Languages 399 19 A few nouns change the stem entirely in passing from singular to plural. Such are: — sQ.lV'^ woman plural nig'-^dp'tai sa'as?r" girl (diminutive of ssiir") ninig^ap'lai (diminutive of nig^dp'tai) Involving this same change of stem is: — sdHux'^ married man nig^dpHahai' Rather different, presumably, is: — tatfndtcap^ leg tcuklu'ndtcap' which keeps the same suffix in the plural, while changing the stem. Nouns without Plurals. Quite a number of nouns were secured which form no plural. Some of these are reduplicated to begin with, and there is clearly a feeling, though one by no means consistently applied, against re-reduplication in forming plurals. Others, however, are such as might easily be reduplicated, were it usage to do so. It is possible that reduplicated plurals might have been given for some of these by other informants. Reduplicated nouns that form no plural are: — gypgyp panther qivdqumi^s marten titditcVc owl mdmstco'm mink is'.i' xHs'.ix'' fish-hawk qwVqwV sea-gull qwi^qwdH! Ald^'k" butterfly xivdxwadjo'77i' fly (probably diminutive: dim. plur. is found) pok^^pok'"^ liver xdxe'i nit LaLdpx"^ knife hdiihei' arrow Non-reduplicated nouns for which my informant would give no plurals are: — mdyos raccoon pVk! ground-hog qlt'etc elk p.'d'alats!" skunk v!dxd'° \ ^dmaxHdjo'" ant ^ ■■ Vraven ,. • •" p. 'ah ) qeix salmon-egg tdeq" robin mo'os head For "robin," tdAq'tcfeq', which might well enough l)e expected as plural, was explicitly denied. If necessary to express plurality in these nouns, qax or q.ix "many" can be juxtaposed before any of thorn. 400 VI American Indian Languages 2 20 IV. REDUPLICATED DIMINUTIVES OF NOUNS. Diminutives in Comox, as in other Salish languages, are formed by means of reduplication. Reduplicated diminutive forms, however, differ from reduplicated plurals in that the reduplicat- ing syllable rei:>eats the first consonant of the stem, never also the second. Moreover, the vowel of the reduplicating syllable is formed according to different rules from that of the redupli- cating syllable of plural forms. Further complications result from the internal changes to which the stem is often subjected, so that altogether a large number of more or less distinct types of diminutive formations may be recognized. It will be advan- tageous to list in a purely analytical way the various features that are found in diminutives, so that ready reference may be made to them when discussing the types as^such. Diminutivizing characteristics are: — {I. ) Reduplication of initial consonant of stem, followed by a. Short e (i or i). Two types of c- reduplication may be recognized, according to whether e is or is not accented. Thus, mimosas from yno^os "head"; qeqn^ya' from qd^ya^ "water.' b. Long e (i or i), always accented. Thus LllLlAXWd'* from lIaxwo''- "dog-salmon." c. e, always accented. Thus qfe'^q'.e^L! from q.'d'^L!" land- otter." d. V, which may or may not be accented. Thus, lolko°min from loko^min "bailer." e. V, which is regularly accented. Thus, k!6k!gddf!° from k!6"ddt!'^ "porpoise." f. Short a, accented or not. Thus, LdLV'tm'^ from lV' attC "cockle." g. Long a. Thus, djddjd''gHn' from djig^in' "song." h. Long d'a. Thus, sd^aslV^ from sd/f" "woman." i. Short 0. Thus, LloLld'ami's from dAms "house." (2.) Glottal stop inserted in stem. This may occur as a. Breaking of (non-final) yowel or diphthong. Thus, tcitcd^'^yac from tcdyac "hand." b. Glottalizing of final consonant (generally m or n); this should probably include breaking of vowel when final. Thus. Idtbo'm' from 16'^^hom' "small clam." Eight: Wakashan and Salishan Languages 401 21 (3.) Quantitative vocalic changes (increments). These include a. Lengthening of (last) stem voweL Thus, tdtig'^dx" from td'agyax"" "fern." b. Change to wd or wa of u of stem. Thus, diminutive plural kwikumkwd'''mdqin' from plural kumkumdqin" "sea-lions." c. Lengthening of inorganic a (or i. e) to i. Thus, xexsim' from XASAm "box." Less often, full a is changed to i (cf. 4b), as in kwekwi'im' from kwd'am "coiled storage basket." d. Insertion of i. This is probably but another form of 3c, inorganic a and absence of vowel being perhaps con- sidered as phonologically equivalent. Thus, qeqAli'q! from qAl'q! "warrior." e. Insertion of short vowel (a, i) before syllable with length- ened vowel. Thus, xt^xigHctn^ (note second i) from xAucin^ "bone." f. Lengthening of a or a (non-final) to d. Thns, qlwdq.'wd''^- djix from q'.wAHx "wood." (4.) Qualitative vocalic changes. These include a. Umlaut of a to short e (i). Thus, xexd'adjeHc from xd'adjaic "stone." b. Umlaut of a {or d), rarely o, to long e {i, i). Thus, qle^qle^k'^ from qfak^"- "board." c. Change of stem vowel to a'a. Thus, tot&'amic from td'mic "man." (5.) Vocalic reduction. Under this head may be grouped a. Shortening of stem vowel before syllable with lengthened vowel (or inserted i). This shortening before lengthening is doubtless due to quantitative rhythm. Thus, qicrl''- qwi'qwV* (note second i) from gmV/iw' "sea-gull. Such shortened syllables regularly lose their glottal stop, if there is one present, as in qeqawem" from qd'um' "eye." b. Syncope of stem vowel after reduplicating syllable with accented vowel. Long vowels may thus fall out quite as readily as short ones. Thus, se''sp'xos from sdpdxos "horn." These twenty-two diminutivizing features occur in various combinations, so that a large number of possible types of 402 VI American Indian Languages 2 22 diminutive formation maj' result. A considerable number of such types can be constructed from the available material, but this need not exemplify all that actually occur. As to which of the features listed are fundamental to Salish and which merely secondary in Comox or several Coast Salish languages, it is useless to speculate. Adequate comparative data are necessary. A few points of a comparative nature will be brought forward at the end of the paper. The various diminutive types will now be taken up in order, the main stress being laid on the form of the reduplicating syllable. Type I. Reduplicating Syllable: ce. Various sub-types occur, according to whether the stem vowels remain unmodified or are subjected to comparatively slight changes. Sub-type I a. Diminutive feature la only: — mo' OS head diminutive mimo'os qtaabas wooden ball qeq'td'abas q.'ga'dda ear q!weq!oa'dda 'dwdk^'' tobacco 'e'awdk'" qwAdi^s whale qweqwAdi^s qo'^'a'^ hemlock qioiqd'^'a'' sidjdqo'p' basket cap sisidjdqo'p' qeix salmon-egg qeqeyix {-eyi- prob- ably merely variant of -ei-) Sub-type I b. Diminutive features la, 3c (or d) : — mAqsirC nose mimAqsi^n' djidis tooth djldjidVs ^I'x" yellow cedar {