SMITHSONIAN INSTITUTION

BUREAU OF AMERICAN ETHNOLOGY

BULLETIN 40

HANDBOOK OF AMERICAN INDIAN LANGUAGES

BY FRANZ BOAS

PART 2

WITH ILLUSTRATIVE SKETCHES

By EDWARD SAPIR, LEO J. FRACHTENBERG, AND WALDEMAR BOGORAS

WASHINGTON

GOVERNMENT PRINTING OFFICE 1922

BM+

ncrou Library

LETTER OF TRANSMITTAL

SMITHSONIAN INSTITUTION, BUREAU OF AMERICAN ETHNOLOGY,

Washington, D. 0., February 20, 1911.

SIR: I have the honor to submit for publication, subject to your approval, as Bulletin 40, Part 2, of this Bureau, the manuscript of a portion of the Handbook of American Indian Languages, prepared under the editorial supervision of Dr. Franz Boas.

Yours, respectfully,

F. W. HODGE,

Ethnologist in Charge. Dr. CHARLES D. WALCOTT,

Secretary of the Smithsonian Institution*

in

CONTENTS

Page

The Takelma language of southwestern Oregon, by Edward Sapir 1

Coos, by Leo J. Frachtenberg 297

Siuslawan (Lower Umpqua), by Leo J. Frachtenberg 431

Chukchee, by Waldemar Bogoras 631

THE TAKELMA LANGUAGE OF SOUTH WESTERN OREGON

BY

EDWARD SAPIR

3045°— Bull. 40, pt 2—12 1 1

CONTENTS

Page

1 . Introduction 7

§ 2-24. Phonology 8

§ 2. Introductory 8

§§ 3-11. Vowels 10

§ 3. General remarks 10

§ 4. System of vowels 10

§ 5. Stress and pitch-accent 15

§§ 6-11. Vocalic processes 22

§ 6. Vowel hiatus 22

§ 7. Dissimilation of u 24

§ 8. I- umlaut 24

§ 9. K-sounds preceded by -w-vowels 27

§ 10. Inorganic a 28

§ 11. Simplification of double diphthongs 29

§§ 12-24. Consonants 31

§ 12. System of consonants 31

§ 13. Final consonants 35

§§ 14-17. Consonant combinations 36

§ 14. General remarks 36

§ 15. Initial combinations 36

§ 16. Final combinations 38

§ 17. Medial combinations _ 39

§§ 18-24. Consonant processes _ 41

§ 18. Dropping of final consonants 41

§ 19. Simplification of doubled consonants 42

§ 20. Consonants before x 44

§ 21. Dissimilation of n to / and m 45

§ 22. Catch dissimilation 47

§ 23. Influence of place and kind of accent on manner of articula- tion 48

§ 24. Inorganic h 51

j 25-115. Morphology 52

§ 25. Introductory 52

§§ 26-32. Grammatical processes 55

§ 26. General remarks. 55

§ 27. Prefixation 55

§ 28. Suffixation 56

§ 29. Infixation 56

§ 30. Reduplication 57

§ 31. Vowel-ablaut 59

§ 32. Consonant-ablaut 62

§§ 33-83. I. The verb _ 63

§ 33. Introductory 63

§§ 34-38. 1. Verbal prefixes 64

§ 34. General remarks 64

§ 35. Incorporated nouns 66

§ 36. Body-part prefixes 72

§ 37. Local prefixes 86

§ 38. Instrumental wa- 91

3

THE TAKELMA LANGUAGE OF SOUTHWESTERN

OREGON

BY EDWARD SAPIR

§ 1. INTRODUCTION

The language treated in the following pages was spoken in the southwestern part of what is now the state of Oregon, along the middle portion of Rogue river and certain of its tributaries. It, together with an upland dialect of which but a few words were obtained, forms the Takilman stock of Powell. The form "Takelma" of the word is practically identical with the native name of the tribe, Daagelma'£n THOSE DWELLING ALONG THE RIVER (see below, § 87, 4) ; there seems to be no good reason for departing from it in favor of Powell's variant form.

The linguistic material on which this account of the Takelma language is based consists of a series of myth and other texts, pub- lished by the University of Pennsylvania (Sapir, Takelma Texts, Anthropological Publications of the University Museum, vol. n, no. 1, Philadelphia, 1909), together with a mass of grammatical material (forms and sentences) obtained in connection with the texts. A series of eleven short medicine formulas or charms have been pub- lished with interlinear and free translation in the Journal of Ameri- can Folk-Lore (xx, 35-40). A vocabulary of Takelma verb, noun, and adjective stems, together with a certain number of derivatives, will be found at the end of the " Takelma Texts." Some manu- script notes on Takelma, collected in the summer of 1904 by Mr. H. H. St. Clair, 2d, for the Bureau of American Ethnology, have been kindly put at my disposal by the Bureau ; though these consist mainly of lexical material, they have been found useful on one or two points. References like 125.3 refer to page and line of my Takelma Texts. Those in parentheses refer to forms analogous to the ones discussed.

8 BUREAU OF AMEKICAN ETHNOLOGY [BULL. 40

The author's material was gathered at the Siletz reservation of Oregon during a stay of a month and a half in the summer of 1906, also under the direction of the Bureau of American Ethnology. My informant was Mrs. Frances Johnson, an elderly full-blood Takelma woman. Her native place was the village of Dak'tslasin or DaldanVY, on Jump-off-Joe creek (Dip!olts!i'lda), a northern affluent of Rogue river, her mother having come from a village on the upper course of Cow creek (Hagwal). Despite her imperfect command of the English language, she was found an exceptionally intelligent and good-humored informant, without which qualities the following study would have been far more imperfect than it necessarily must be under even the very best of circumstances.

In conclusion I must thank Prof. Franz Boas for his valuable advice in regard to several points of method and for his active interest in the progress of the work. It is due largely to him that I was encouraged to depart from the ordinary rut of grammatical description and to arrange and interpret the facts in a manner that seemed most in accordance with the spirit of the Takelma language itself.1

PHONOLOGY (§§2-24)

§ 2. Introductory

In its general phonetic character, at least as regards relative harsh- ness or smoothness of acoustic effect, Takelma will probably be found to occupy a position about midway between the characteristically rough languages of the Columbia valley and the North Californian and Oregon coast (Chinookan, Salish, Alsea, Coos, Athapascan, Yurok) on the one hand, and the relatively euphonious languages of the Sacramento valley (Maidu, Yana, Wintun) on the other, inclining rather to the latter than to the former.

From the former group it differs chiefly in the absence of voice- less Z-sounds (L, Z,2 L!) and of velar stops (q, g, q!) ; from the latter,

i What little has been learned of the ethnology of the Takelma Indians will be found incorporated in two articles written by the author and entitled Notes on the Takelma Indians of Southwestern Oregon, in American Anthropologist, n. s., ix, 251-275; and Religious Ideas of the Takelma Indians of Southwestern Oregon, in Journal of American Folk-Lore, xx, 33-49.

8 In the myths, I is freely prefixed to any word spoken by the bear. Its uneuphonious character is evi- dently intended to match the coarseness of the bear, and for this quasi-rhetorical purpose it was doubtless derisively borrowed from the neighboring Athapascan languages, in which it occurs with great frequency. The prefixed sibilant «• serves in a similar way as a sort of sneezing adjunct to indicate the speech of the coyote. Owi'di WHERE? says the ordinary mortal; Igwi'di, the bear; s-gwi'di, the coyote.

§ 2

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA

in the occurrence of relatively more complex consonantic clusters, though these are of strictly limited possibilities, and hardly to be considered as difficult in themselves.

Like the languages of the latter group, Takelma possesses clear- cut vowels, and abounds, besides, in long vowels and diphthongs; these, together with a system of syllabic pitch-accent, give the Takel- ma language a decidedly musical character, marred only to some extent by the profusion of disturbing catches. The line of cleavage between Takelma and the neighboring dialects of the Athapascan stock (Upper Umpqua, Applegate Creek, Galice Creek, Chasta Costa) is thus not only morphologically but also phonetically distinct, despite re- semblances in the manner of articulation of some of the vowels and consonants. Chasta Costa, formerly spoken on the lower course of Rogue river, possesses all the voiceless Z-sounds above referred to ; a peculiar illusive g/, the fortis character of which is hardly as prominent as in Chinook; a voiced guttural spirant 7-, as in North German Tage; the sonants or weak surds dj and z (rarely) ; a voiceless interdental spirant $ and its corresponding fortis tp!; and a very frequently oc- curring ti vowel, as in English HUT. All of these are absent from Takelma, which, in turn, has a complete labial series (6, p\ p!, m), whereas Chasta Costa has only the nasal m (labial stops occur appar- ently only in borrowed words, Itdpi' CAT <pussy). The fortis Jc!, com- mon in Takelma, seems in the Chasta Costa to be replaced by ql; the Takelma vowel u, found also in California, is absent from Chasta Costa; r is foreign to either, though found in Galice Creek and Shasta. Perhaps the greatest point of phonetic difference, however, between the Takelma and Chasta Costa languages lies in the peculiar long (doubled) consonants of the latter, while Takelma regularly simpli- fies consonant geminations that would theoretically appear in the building of words. Not enough of the Shasta has been published to enable one to form an estimate of the degree of phonetic similarity that obtains between it and Takelma, but the differences can hardly be as pronounced as those that have just been found to exist in the case of the latter and Chasta Costa.

This preliminary survey seemed necessary in order to show, as far as the scanty means at present at our disposal would allow, the phonetic affiliations of Takelma. Attention will now be directed to

the sounds themselves.

§ 2

10 BUEEAU OF AMERICAN ETHNOLOGY [BULL. 40

Vowels (§§ 3-11) § 3. General Remarks

The simple vowels appear, quantitatively considered, in two forms, short and long, or, to adopt a not inappropriate term, pseudo- diphthongal. By this is meant that a long vowel normally con- sists of the corresponding short vowel, though generally of greater quantity, plus a slight parasitic rearticulation of the same vowel (indicated by a small superior letter) , the whole giving the effect of a diphthong without material change of vowel-quality in the course of production. The term PSEUDO-DIPHTHONG is the more justified in that the long vowel has the same absolute quantity, and experi- ences the same accentual and syllabic treatment, as the true diph- thong, consisting of short vowel + i, u, I, m, or n. If the short vowel be given a unitary quantitative value of 1, the long vowel (pseudo-diphthong) and ordinary diphthong will have an approxi- mate value of 2 ; while the long diphthong, consisting of long vowel + i, u, I, m, or n, will be assigned a value of 3. The liquid (Z) and the nasals (m and n) are best considered as forming, parallel to the semi-vowels y (i) and w (u), diphthongs with preceding vowels, inasmuch as the combinations thus entered on are treated, similarly to i- and u- diphthongs, as phonetic units for the purposes of pitch- accent and grammatic processes. As a preliminary example serving to justify this treatment, it may be noted that the verb-stem lilw-, Hlu- JUMP becomes bilau- with inorganic a under exactly the same phonetic conditions as those which make of the stem Tctemn- MAKE Jcleman-. We thus have, for instance:

bilwa'£s jumper; Wfafuk* he jumped Jc!emna,'es maker; Tcfema^rik* he made it

From this and numberless other examples it follows that au and an, similarly ai, oil, and am, belong, from a strictly Takelma point of view, to the same series of phonetic elements; similarly for e, i, o, and u diphthongs.

§ 4. System of Vowels

The three quantitative stages outlined above are presented for the various vowels and diphthong-forming elements in the following table :

§§ 3-4

BOAS]

HANDBOOK OF INDIAN LANGUAGES TAKELMA

11

I. Short.	II. Long.	Short diphthong.	III. Long diphthong.
a	da, (a)	ai, au, al, am, an	ai, au, aal, aam, aan
e	e«,(i)	ei, eu, el, em, en	ei, bu, eel, eem e*n
i	»*, (*)	iu, il, im, in	lu, HI, iim, tin
0, («)	6«, (6)	oi, ou, ol, om, on	oi, ou(w), oul, oum, dun
		(o«) (uZ) (MTO) (un)
u	««, (fi)	ui, uw, ul, urn, un	ui, uu(w), uul, Hum, uun
		(uu)
u	uu, (U)	iii, uw, iii, iim, un	iii, iiu(w), uul, Hum, Hun

It is to be understood, of course, that, under proper syllabic con- ditions, i and u may respectively appear in semivocalic form as y and w; thus ou and uu appear as ow and uw when followed by vowels ; e. g., in Tc!uwuu- THROW AWAY, uw and uu are equivalent elements forming a reduplicated complex entirely analogous to -del- in Tielel- SING. Similarly ai, au, ai, and au may appear as ay, aw, aay, and aaw; and correspondingly for the other vowels. Indeed, one of the best criteria for the determination of the length of the first element of a diphthong is to obtain it in such form as would cause the second element (i or u) to become semi-vocalic, for then the first vowel will adopt the form of a short vowel or pseudo-diphthong, as the case may be. The following phonetic (not morphologic) pro- portions will make this clearer:

6iZiuZV I jump: bttiwat* you jump = hee£m he went away from him: 7ieee^wifen I went away from him

</aiF he ate it : g&yawafsn I ate it = </aiF he grew: geiaya'£t* he will grow

gay&n he ate it : gaya,wa'£nl ate it = Jiant'gsiU over land : Latg'st&waf£

one from Lat'gau [uplands]

Sometimes, though not commonly, a diphthong may appear in the same word either with a semivowel or vowel as its second element, according to whether it is or is not followed by a connecting inor- ganic a. A good example of such a doublet is 7M,yefiwa'xdaada or hayeuxdaada IN HIS RETURNING (verb stem yeu-, yeew- RETURN). It is acoustically difficult to distinguish sharply between the long vowel or pseudo-diphthong ou and the it-diphthongs of o (both ou and ou are often heard as du), yet there is no doubt that there is an organic difference between ou, as long vowel to o, and ou = ou, ou. Thus, in Io7iouna'£n I CAUSE HIM TO DIE, and lohona'n I SHALL CAUSE HIM TO DIE, ou and o are related as long and short vowel in parallel

§ 4

12 BUEEAU OF AMERICAN ETHNOLOGY [BULL. 40

fashion to the aa and a of yaa?iaY YOU WENT, and y&nada'5 YOU WILL GO. On the other hand, the ou of p'oup'au- (aorist stem) BLOW is organically a diphthong (ou), the ou of the first syllable being related to the au of the second as the iu of Yiuk^au- (verb stem) BRANDISH is to its au. Similarly, the -ou- of s'ofuelc'6p'- (verb stem) JUMP is organic shortened ou, related to the -owo- of the aorist stem s*o'wo£fc'op'- as the -6*- of Jie'i£x- (verb stem) BE LEFT OVER is to the -eye- of Jieyesx- (aorist stem). A similar acoustic difficulty is experi- enced in distinguishing up, (uu) as long vowel from the u- diphthongs of u, (u).

Examples of unrelated stems and words differing only in the length of the vowel or diphthong are not rare, and serve as internal evidence of the correctness, from a native point of view, of the vowel classification made:

gal- eat, but gai- grow

verb-prefix dda- ear, but da- mouth

waaxa his younger brother, but wa'xa at them

It may happen that two distinct forms of the same word differ only in vocalic quantity; y&ada/£t' HE WILL SWIM, y&da'£t' HE SWIMS.

It is, naturally enough, not to be supposed that the long vowels and diphthongs always appear in exactly the same quantity. Speed of utterance and, to some extent, withdrawal of the stress-accent, tend to reduce the absolute quantities of the vowels, so that a nor- mally long vowel can become short, or at least lose its parasitic attachment. In the case of the i- and u- diphthongs, such a quan- titative reduction means that the two vowels forming the diphthong more completely lose their separate individuality and melt into one. Quantitative reduction is apt to occur particularly before a glottal catch; in the diphthongs. the catch follows so rapidly upon the second element (i or u) that one can easily be in doubt as to whether a full i- or u- vowel is pronounced, or whether this second vowel appears rather as a palatal or labial articulation of the catch itself. The practice has been adopted of writing-such diphthongs with a superior i or u before the catch: ai£, au£, eus, and similarly for the rest. When, however, in the course of word-formation, this catch drops off, the i or u that has been swallowed up, as it were, in the catch reasserts itself, and we get such pairs of forms as:

nag&n£ he said ; but naga,'idas when he said

sgele/u£ he shouted; but sgele'uda£ when he shouted § 4

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 13

On the other hand, vowels naturally short sometimes become long when dwelt upon for rhetorical emphasis. Thus ga THAT sometimes appears as gaa:

gda loho'£ee in that case I shall die

ga'a gafcfl for that reason

As regards the pronunciation of the vowels themselves, little need be said. The a is of the same quality as the short a of German MANN, while the long da (barring the parasitic element) corresponds to the a of HAHN.

A labial coloring of the a (i. e., 6 as in German VOLL) frequently occurs before and after Yw:

gfthoY™ planted, sown

ik?wafaY<wdY he woke him up But there were also heard:

se¥a¥w shot

malaYwa he told him

The e is an open sound, as in the English LET; it is so open, indeed, as to verge, particularly after y, toward a.1 Also the long vowel ee is very open in quality, being pronounced approximately like the ei of English THEIR (but of course without the r- vanish) or the e" of French FETE; ee, though unprovided with the mark of length, will be always understood as denoting the long vowel (pseudo-diphthong) corresponding to the short e; while e will be employed, wherever necessary, for the long vowel without the parasitic -e. The close e, as in German REH, does not seem to occur in Takelma, although it was sometimes heard for i; in the words laale? HE BECAME, ldalet'am YOU BECAME, and other related forms, e was generally heard, and may be justified, though there can be small doubt that it is morphologically identical with the I1 of certain other verbs.

The i is of about the same quality as in English HIT, while the long I* is closer, corresponding to the ea of English BEAT. Several monosyllables, however, in -i, such as gwi WHERE, di interrogative particle, should be pronounced with a close though short vowel (cf. French FINI) . This closer pronunciation of the short vowel may be explained by supposing that gwi, di, and other such words are rapid pronunciations of gwl*j dl1, and the others; and indeed the texts sometimes show such longer forms.

iThe word yewe'i* HE KETUENED, e. g., was long heard as yawe'tf, but such forms as ytu RETURN! show this to have been an auditory error.

§ 4

14 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

The o is a close vowel, as in German SOHN, as far as the quality is concerned, but with the short quantity of the o of VOLL. This close- ness of pronunciation of the o readily explains its very frequent interchange with u:

lts'!o'p*al sharp-clawed dets' luguY sharp-pointed

and also the u- quality of the parasitic element in the long close vowel ou. The short open 6, as in German VOLL, never occurs as a primary vowel, but is practically always a labialized variant of a. Thus in Takelma, contrary to the parallelism one ordinarily expects to find in vocalic systems, e- vowels are open in quality, while o- vowels are close.

The vowel u is close, as in the English word RUDE, the long mark over the u being here used to indicate closeness of quality rather than length of quantity. The u is not identical with the German u, but is somewhat more obscure in quality and wavers (to an un- Indian ear) between the German short u of MUTZE and u of MUSS ; sometimes it was even heard with the approximate quality of the short o of GOTZ. The long #fi' is, in the same way, not exactly equivalent to the long u of the German stiss, but tends in the direc- tion of uu, with which it frequently varies in the texts. It is some- what doubtful how far the two vowels u and u are to be considered separate and distinct; it is quite possible that they should be looked upon as auditory variants of one sound. Before or after y or w, u is apt to be heard as u, - - ~k!uwu'£ THEY RAN AWAY, uyu'£s- HE LAUGHED, lguyuglfisi, HE KEEPS NUDGING ME, otherwise often as u.

The only short vowel not provided for in the table is fa (as in Eng- lish SUN), which, however, has no separate individuality of its own, but is simply a variant form of a, heard chiefly before m:

7iee£rte'me£xtim he killed us off (for -am) xfam in water (for xam)

The. absence of the obscure vowel E of indeterminate quality is noteworthy as showing indirectly the clear-cut vocalic character of Takelma speech. Only in a very few cases was the E heard, and in the majority of these it was not a reduced vowel, but an intrusive sound between m and s:

dak'?be'e£k'fbagamES he tied his hair up into top-knot (in place

of -ams). § 4

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 15

Even here it may really have been the strongly semantic quality of the m in contrast to the voiceless s that produced the acoustic effect of an obscure vowel. The exact pronunciation of the diph- thongs will be better understood when we consider the subject of pitch-accent.

§ 5. Stress and Pitch- Accent

Inasmuch as pitch and stress accent are phonetic phenomena that affect more particularly the vowels and diphthongs, it seems advisable to consider the subject here and to let the treatment of the conso- nants follow. As in many Indian languages, the stress-accent of any particular word in Takelma is not so inseparably associated with any particular syllable but that the same word, especially if consisting of more than two syllables, may appear with the main stress-accent now on one, now on the other syllable. In the uninterrupted flow of the sentence it becomes often difficult to decide which syllable of a word should be assigned the stress-accent. Often, if the word bears no particular logical or rhythmic emphasis, one does best to regard it as entirely without accent and as standing in a proclitic or enclitic relation to a following or preceding word of greater emphasis. This is naturally chiefly the case with adverbs (such as Tiesne THEN) and conjunctive particles (such as gariehi* AND THEN; agas'is AND so, BUT THEN) ; though it not infrequently happens that the major part of a clause will thus be strung along without decided stress-accent until some emphatic noun or verb-form is reached. Thus the following passage occurs in one of the myths:

ganehi£ dewenxa laale hono£ p*elefxa£, literally translated, And then to-morrow (next day) it became, again they went out to war

All that precedes the main verb-form p'de'xa? THEY WENT OUT TO WAR is relatively unimportant, and hence is hurried over without any- where receiving marked stress.

Nevertheless a fully accented word is normally stressed on some particular syllable; it may even happen that two forms differ merely in the place of accent :

naga1 '-ida£ when he said, but naga-4dafs when you said

The important point to observe, however, is that when a particular syllable does receive the stress (and after all most words are normally

§ 5

16 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

accented on some one syllable) , it takes on one of two or three musical inflections :

(1) A simple pitch distinctly higher than the normal pitch of unstressed speech (-).

(2) A rising inflection that starts at, or a trifle above, the normal pitch, and gradually slides up to the same higher pitch referred to above (=^).

(3) A falling inflection that starts at, or generally somewhat higher than, the raised pitch of (1) and (2), and gradually slides down to fall either in the same or immediately following syllable, to a pitch somewhat lower than the normal (-).

The "raised" pitch (^) is employed only in the case of final short vowels or shortened diphthongs (i. e., diphthongs that, owing to speed of utterance, are pronounced so rapidly as to have a quanti- tative value hardly greater than that of short vowels; also sec- ondary diphthongs involving an inorganic a); if a short vowel spoken on a raised pitch be immediately followed by an unac- cented syllable (as will always happen, if it is not the final vowel of the word), there will evidently ensue a fall in pitch in the unaccented syllable, and the general acoustic effect of the two syllables will be equivalent to a " falling" inflection (-) within one syllable; i. e. (if — be employed to denote an unaccented syllable),

(-) 4 =(-)• The following illustration will make this clearer:

YOU SANG is regularly accented helelaY, the ar being sung on an interval of a (minor, sometimes even major) third above the two unaccented e- vowels. The acoustic effect to an American ear is very much the same as that of a curt query requiring a positive or nega- tive answer, DID HE GO ? where the i of DID and e of HE correspond in pitch to the two e's of the Takelma word, while the o of GO is equiva- lent to the Takelma a\ The Takelma word, of course, has no interrogative connotation. If, now, we wish to make a question out of JielelaY, we add the interrogative particle di, and obtain the form Jielela'^idi DID HE SING? (The % is a weak vowel inserted to keep the f and d apart.) Here the a' has about the same pitch as in the preceding word, but the I sinks to about the level of the e- vowels, and the di is pronounced approximately a third below the normal level. The Takelma interrogative form thus bears an acoustic resemblance to a rapid English reply: so HE DID GO, the o of so and

§ 5

BOAS] HANDBOOK OF INDIAN LANGUAGES - TAKELMA 17

e of HE corresponding in pitch to the unaccented e- vowels of the Takelma, the i of DID resembling in its rise above the normal pitch the a', and the o of GO sinking like the i of the interrogative particle.1 If the normal level of speech be set at A, the two forms just considered may be musically, naturally with very greatly exaggerated tonal effect, represented as follows :

lie- le- la^t' he- le-

The "rising" pitch (^) is found only on long vowels and short or long diphthongs. The rising pitch is for a long vowel or diphthong what the raised pitch is for a short vowel or shortened diphthong; the essential difference between the two being that in the latter case the accented vowel is sung on a single tone reached without an inter- mediate slur from the lower level, whereas in the case of the rising pitch the affected vowel or diphthong changes in pitch in the course of pronunciation; the first part of the long vowel and the first vowel of the diphthong are sung on a tone intermediate between the normal level and the raised pitch, while the parasitic element of the long vowel and the second vowel (I or u) of the diphthong are hit by the raised tone itself. It is easy to understand that in rapid pronuncia- tion the intermediate tone of the first part of the long vowel or diph- thong would be hurried over and sometimes dropped altogether ; this means that a long vowel or diphthong with rising pitch (a, al] becomes a short vowel or shortened diphthong with raised pitch (a\ a?i).2 Diphthongs consisting of a short vowel + Z, m, or n, and provided with a rising pitch, ought, in strict analogy, to appear as an, al, am; and so on for the other vowels. This is doubtless the correct repre- sentation, and such forms as :

nanli he will say, do

gwalt* wind

dasmayain he smiled

wulx enemy, Shasta

were actually heard, the liquid or nasal being distinctly higher in pitch than the preceding vowel. In the majority of cases, however,

1 It is curious that the effect to our ears of the Takelma declarative helela't' is of an interrogative DID YOU SING? while conversely the effect of an interrogative helela't'idi is that of a declarative YOU DID SING. This is entirely accidental in so far as a rise in pitch has nothing to do in Takelma with an interrogation.

8 A vowel marked with the accent =* is necessarily long, so that the mark of length and the parasitic vowel can be conveniently omitted.

3045°— Bull. 40, pt 2— 12 - 2 § 5

18 BUREAU OF AMERICAN ETHNOLOGY . [BULT,. 40

these diphthongs were heard, if not always pronounced, as shortened diphthongs with raised pitch («X ay/, axm). The acoustic effect of a syllable with rising pitch followed by an unaccented syllable is neces- sarily different from that of a syllable with falling pitch (^), or of a syllable with raised pitch followed by an unaccented syllable, because of the steady rise in pitch before the succeeding fall. The tendency at first is naturally to hear the combination — ^ — as — - — , and to make no distinction in accent between yewe'ida£ WHEN HE RETURNED and yewelfe* i RETURNED; but variations in the recorded texts between the rising and falling pitch in one and the same form are in every case faults of perception, and not true variations at all. The words tlomom HE KILLED HIM and yawalt'e£ i SPOKE may be approxi- mately represented in musical form as follows : ^- -^

a.

t!o- md-um ya- wa- i- t'e*

The falling pitch ( - ) affects both long and short vowels as well as diphthongs, its essential characteristic being, as already defined, a steady fall from a tone higher than the normal level. The peak of the falling inflection may coincide in absolute pitch with that of the rising inflection, though it is often somewhat higher, say an interval of a fourth above the ordinary level. The base (lowest tone) of the fall is not assignable to any definite relative pitch, the gamut run through by the voice depending largely upon the character of the syllable. If the accent hits a long vowel or diphthong not immedi- ately followed by a catch, the base will, generally speaking, coincide with the normal level, or lie somewhat below it. If the long vowel or diphthong be immediately followed by an unaccented syllable, the base is apt to strike this unaccented syllable at an interval of about a third below the level. If the vowel or diphthong be immediately followed by a catch, the fall in pitch will be rapidly checked, and the whole extent of the fall limited to perhaps not more than a semitone. As soon, however, as the catch is removed (as often happens on the addition to the form of certain grammatical elements), the fall runs through its usual gamut. The words

Ywede'i his name

yewe'ida5 when he returned

yewe'i£ he returned

will serve to illustrate the character of the falling pitch. § 5

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 19

k'we- de' - i ye- we' - i - da- ye - we is

The pronunciation of the diphthongs is now easily understood A shortened diphthong (cfi, a'i£) sounds to an American ear like an indivisible entity, very much like ai and au in HIGH and HOW; a diphthong with falling pitch (a'i) is naturally apt to be heard as two distinct vowels, so that one is easily led to write naga'-idas instead of nagafida£ WHEN HE SAID; a diphthong with rising pitch (al) is heard either as a pure diphthong or as two distinct vowels, according to the speed of utterance or the accidents of perception. All these interpretations, however, are merely matters of perception by an American ear and have in themselves no objective value. It would be quite misleading, for instance, to treat Takelma diphthongs as " pure" and " impure," no regard being had to pitch, for such a classi- fication is merely a secondary consequence of the accentual phenomena we have just considered.

One other point in regard to the diphthongs should be noted. It is important to distinguish between organic diphthongs, in which each element of the diphthong has a distinct radical or etymological value, and secondary diphthongs, arising from an i, u, I, m, or n with pre- fixed inorganic a. The secondary diphthongs (ai, au, al, am, an), being etymologically single vowels or semivowels, are always unitonal in character; they can have the raised, not the rising accent. Con- trast the inorganic au of

&iZa\iF ( =*&iZwv^r',1 not *6iZauF) he jumped; cf. bilwa'£s JUMPER with the organic au of

gayau he ate it; cf. gay&waf£n I ate it Contrast similarly the inorganic an of

Jdems^nk* (=*&/6mnvF, not *&/6maiiF) he made it; cf. Jc!emna/es maker

with the organic am of

dasmay&m he smiled ; cf . dasmaysLma/£n I smiled Phonetically such secondary diphthongs are hardly different from shortened organic diphthongs; etymologically and, in consequence, in morphologic treatment, the line of difference is sharply drawn.

» Non-existent or theoretically reconstructed forms are indicated by a prefixed asterisk.

§ 5

20 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

It was said that any particular syllable, if accented, necessarily receives a definite pitch-inflection. • If it is furthermore pointed out that distinct words and forms may differ merely in the character of the accent, and that definite grammatical forms are associated with definite accentual forms, it becomes evident that pitch-accent has a not unimportant bearing on morphology. Examples of words differ- ing only in the pitch-accent are:

se'el black paint, writing; sel kingfisher

la'ap* leaves; (1) lap' he carried it on his back, (2) lap* become (so

and so) !

sa'at* his discharge of wind; sat* mash it! will'* his house; will house, for instance, in daYwill on top of

the house 1ie'el song; hel sing it!

Indeed, neither vowel-quantity, accent, nor the catch can be consid- ered negligible factors in Takelma phonology, as ,,shown by the following :

waya* knife

waya'a his knife

waya'£ he sleeps

wayan he put him to sleep

Jc!wa£ya^ ( = Jc!walsa^ just grass

It is impossible to give any simple rule for the determination of the proper accent of all words. What has been ascertained in regard to the accent of certain forms or types of words in large part seems to be of a grammatic, not purely phonetic, character, and hence will most naturally receive treatment when the forms themselves are dis- cussed. Here it will suffice to give as illustrations of the morphologic value of accent a few of the cases :

(1) Perhaps the most comprehensive generalization that can be made in regard to the employment of accents is that a catch requires the falling pitch-accent on an immediately preceding stressed syllable, as comes out most clearly in forms where the catch has been second- arily removed. Some of the forms affected are:

(a) The first person singular subject third person object aorist of the transitive verb, as in :

t!omoma'£n I kill him t!omoma'ndas as I killed him § 5

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELJVCA 21

(&) The third person aorist of all intransitive verbs that take the catch as the characteristic element of this person and tense, as in : ya'£ he went ya/ada£ when he went

(c) The second person singular possessive of nouns whose ending for this person and number is -et*s as in:

t!i'i£t' your husband elafH* your tongue

Contrast :

tUFTc" my husband elcffY my tongue

There are but few exceptions to this rule. A certain not very nu- merous class of transitive verbs, that will later occupy us in the treat- ment of the verb, show a long vowel with rising pitch before a catch in the first person singular subject third person object aorist, as in:

Jc!emesn I make it dit!ugusn I wear it

The very isolation of these forms argues powerfully for the general correctness of the rule.

(2) The first person singular subject third person object future, and the third person aorist passive always follow the accent of la:

doumcb'n I shall kill him tlomomafn he was killed

Contrast :

xouma^n he dried it Like lc!emesn in accent we have also:

fclemen it was made

(3) The first person singular possessive of nouns whose ending for that person and number is -£Y shows a raised or rising pitch, according to whether the accented vowel is short or long (or diphthongal) :

Ywedelt'Y my name p!ant*k* my liver tUbagwcfnfJc* my pancreas

Contrast :

Ywede'i his name p!a'an£ his liver tlibagwa'n his pancreas

22 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

(4) The verbal suffix -aid- takes the falling pitch:

sgelewa'ldaen I shouted to him

sgdewaflt* he shouted to him Contrast :

gwalt* wind

Many more such rules could be given, but these will suffice at present to show what is meant by the " fixity " of certain types of accent in morphological classes.

This fixity of accent seems to require a slight qualification. A tendency is observable to end up a sentence with the raised pitch, so that a syllable normally provided with a falling pitch-accent may sometimes, though by no means always, assume a raised accent, if it is the last syllable of the sentence. The most probable explanation of this phenomenon is that the voice of a Takelma speaker seeks its rest in a rise, not, as is the habit in English as spoken in America, in

a fall.1

Vocalic Processes (§§ 6-11)

§ 6. VOWEL HIATUS

There is never in Takelma the slightest tendency to avoid the com- ing together of two vowels by elision of one of the vowels or con- traction of the two. So carefully, indeed, is each vowel kept intact that the hiatus is frequently strengthened by the insertion of a catch. If the words ya'pla MAN and afnl£ NOT, for instance, should come together in that order in the course of the sentence, the two a- vowels would not coalesce into one long vowel, but would be separated by an inorganic (i. e., not morphologically essential) catch yap la £a'm£. The same thing happens when two verbal prefixes, the first ending in and the second beginning with a vowel, come together. Thus:

de- in front

xaa- between, in two

+ 1- with hand generally appear as:

de£l-

xaa£l- respectively. The deictic element -a\ used to emphasize preceding

i Those familiar with Indogermanic phonology will have noticed that my use of the symbols (-), (-), and {^) has been largely determined by the method adopted in linguistic works for the representation of the syllabic pitch-accents of Lithuanian; the main departures being the use of the (-) on short as well as on long vowels and the assignment of a different meaning to the (-).

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 23

nouns, pronouns, and adverbs, is regularly separated from a pre- ceding vowel by the catch :

ma'£a^ but you, you truly bou£a^ nowadays indeed

If a diphthong in i or u precedes a catch followed by a vowel, the i or u often appears as y or w after the catch :

k!wa£ya^ just grass (= Jc!wdl + -a^) a'£ya^ just they (== ai- they + -a^) Jia£wl- (= lia-u- under + 1- with hand)

If the second of two syntactically closely connected words begins with a semivowel (w or y) and the first ends in a vowel, a catch is generally heard to separate the two, in other words the semivowel is treated as a vowel. Examples are:

ge'£ wok* (=ge' +wok*) there he arrived

bee£ waadlri ( = ~bee + waadlfi) day its-body = all day long

ge£ yafahi (=ge+yd'ahi) just there indeed Such cases are of course not to be confounded with examples like:

me£woY HE ARRIVED HERE, and

me£yeu COME HERE !

in which the catch is organic, being an integral part of the adverb me£ HITHER; contrast:

me£gini'£lc' HE CAME HERE, with

ge ginif£¥ HE WENT THERE.

The same phonetic rule applies even more commonly when the first (element is a noun or verb prefix:

hB,twinl'ida inside of him; but hafoc6imv at noon

de£wiliwiafu£ they shouted; but dexebe'£n he said so

Sibsii£wa£yewenTii he returned inside with him; but abai#im/£&' he went inside

wi£wa my younger brother; but wiJicfm my father It Is interesting to note that the catch is generally found also when the first element ends in Z, m, or n, these consonants, as has been already seen, being closely allied to the semivowels in phonetic treatment :

al£waadide to my body; but als'dumcfl to the mountain

sl£yowo/£ he looked; but alxl'tJc* he saw him

6aage'Fi/o he lay belly up; but ge\kliyi'elc* he turned to face him

gweii£wat*geits'!lk*wa, his (head) lay next to it; but gwenZimZa/M£ he looked back

yiwin£ wo'lc*i£ (=yiwin speech +wo'k'i£ without) without speech

§ 6

24 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

It goes without saying that the catch separates elements ending in Z, m, or n from such as begin with a vowel :

s'in£ilats!agi'£n I touch his nose al£tt*baga't*bak* he struck them

§ 7. DISSIMILATION OF u

A diphthong in u tends, by an easily understood dissimilatory process, to drop the u before a labial suffix (-yw-, -p\ -la5). Thus we have :

wahawax^gwa^n I rot with it, for *xiugwa'£n Compare:

7iawaxi'u£ he rots wahawaxiwigwa'n I shall rot with it

Similarly :

T)iliVw he jumped having it, for *lilillk*w (stem biliu-) wiliYw he proceeded with it, for *wiliuTc*w (stem wiliu-)

Observe that, while the diphthong iu is monophthongized, the orig- inal quantity is kept, i being compensatively lengthened to li. In the various forms of the verb yeu- RETURN, such dissimilation, wherever possible, regularly takes place:

yelc'w he returned with it, for *yeuYw( = yeu- gw- V) me£yep* come back! (pl.)> but sing. me£yeu yeebaf£ let us return! for *yeuba'£

It is interesting to note how this u- dissimilation is directly respon- sible for a number of homonyms :

ye¥w bite him! (al)yep* show it to him!

A similar dissimilation of an -u- after a long vowel has in all proba- bility taken place in the reduplicating verb laaliwi'£n i CALL HIM BY NAME (leela'usi HE CALLS ME BY NAME) from * lauliwi'£n (* UuLa'usi) .

§ 8. I- UMLAUT

Probably the most far-reaching phonetic law touching the Takelma vowels is an assimilatory process that can be appropriately termed "i- umlaut." Briefly stated, the process is a regressive assimilation of a non-radical -a- to an -i-, caused by an -i- (-$*-) in an immediately following suffixed syllable, whether the -i- causing the umlaut is an original -i-, or itself umlauted from an original -a-; the -i- of the §§ 7-8

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 25

pronominal endings -bi- THEE, -si- HE TO ME, -xi- HE ME, fails to cause umlaut, nor does the law operate when the -i- is immediately preceded by an inorganic h. The following forms will make the applicability of the rule somewhat clearer :

waJc!ayayinif£n I caused him to grow with it (but k!ayayanaf£n I caused him to grow, with preserved -a-, because of following -a'sn, not -i'sri)

waJdeyeya'nxi he caused me to grow with it

wak!ayaya'nxbi£n I caused thee to grow with it

iyulu'yili£n I rub it (from -yali£n)

iyulu'yalJii he rubs it

It should be carefully noted that this i- umlaut never operates on a radical or stem-vowel, a fact that incidentally proves helpful at times in determining how much of a phonetic complex belongs to the stem, and how much is to be considered as belonging to the grammatical apparatus following the stem. In:

wdagiwi'£n I brought it to him (from -awif£n; cf. waagaf^bi£n I brought it to you)

the -a- following the g is shown to be not a part of the aoristic stem waag- by the i- umlaut that it may undergo; on the other hand, the corresponding future shows an un-umlauted -a- :

wagawi'n I shall bring it to him

so that the future stem must be set down as waga-, as is confirmed by certain other considerations.

It would take us too far afield to enumerate all the possible cases in which i- umlaut takes place ; nevertheless, it is a phenomenon of such frequent recurrence that some of the more common possibilities should be listed, if only for purposes of further illustration:

(1) It is caused by the aoristic verb suffix -I1- denoting position:

s'as'inl he stands (cf. s'afs'an£aa he will stand) tlobigl he lies as if dead (cf . future t!obaga'sdaa)

(2) By an element -i- characteristic of certain nouns, that is added to the absolute form of the noun before the possessive pronominal endings :

buubiniYlc* my arm (cf. buubcfn arm) tfgafltfgilixdek* my belly (for * t'galt'gali-)

(3) By the common verbal " instrumental " vowel -i-, which, for one reason or another, replaces the normal pre-pronominal element

§ 8

26 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

-a-, and often serves to give the verb an instrumental force. This instrumental -i- may work its influence on a great number of preceding elements containing -a-, among which are :

(a) The -a- that regularly replaces the stem-vowel in the second member of a duplicated verb :

al£lt*l>agaft'bigi£n I beat him (cf. -fbaga'fbdk* he beat him) ts'!ele'ts-!ili£n I rattle it (cf. ts'lele'ts'lalhi he rattles it) Ismili'smiltfn I swing it (cf. Ismi'lsmal swing it!)

(5) The causative element -an-: wap!dagini'£n I cause him to swim with it (cf. p!daganaf£n I cause

him to swim) See above:

wak!ayayini'en I cause him to grow

(c) The element -an- added to transitive stems to express the idea

of FOR, IN BEHALF OF :

wat!omomini'£n I kill it for him with it (cf. t!omomana'£n I kill it for him)

(d) The pronominal element -am-, first personal plural object: alxlfiximi£s one who sees us (cf . alxlfixam he sees us)

4. By the suffixed local element -dll ON TOP OF added to the demon- strative pronoun ga THAT to form a general local postposition :

gidl* on top of it, over (so and so) Compare the similarly formed :

gada^Y above

gada^l among and others.

5. By the pronominal element -ig- (-ik*) , first personal plural subject intransitive :

tJomoxinik* we kill each other (cf . t!omoxa£n they kill each other) daxinigam we shall find each other (cf . daxan£t* they will find each other)

This list might be greatly extended if desired, and indeed numerous other examples will meet us in the morphology. Examples of a double and treble i- umlaut are :

Io7iounininif£n I caused him to die (i. e., killed him) for him (cf.

lohdunana'nhi he killed him for him) Iklumininini^nk1 he will fix it for him (compare lkluumcfn he

fixed it)

§ 8

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 27

The semivowel corresponding to i, namely y, is also capable, under analogous circumstances, of causing the i- umlaut of a preceding non- radical a. Examples are:

daxoyo'xiyafn ( = -xaya£n) I scare them around ; daxoyo'xi ( = -xiy

= -xay) he scares them around al£lt*ge'it*giyak*w (=-t*gay-) rolled up alliuyu'Jiiix ( = -hiyx = -hayx) he used to hunt saniycf (=sanaya^) to fight him

doum¥wiya (=-k'waya] to kill him; and numerous other infini- tives in -Jc'wiya ( = -Ywaya)

§ 9. K- SOUNDS PRECEDED BY TJ- VOWELS

An u- vowel (o, u, il, and diphthongs in -u) immediately preceding a Jc- sound (i. e., g, F, &/, x) introduces after the latter a parasitic -w-, which, when itself followed by a vowel, unites with the &- sound to form a consonant-cluster (gw, Jc'w, lc!w, xw}, but appears, when stand- ing after a (word or syllabic) final Y, as a voiceless -w. The intro- duction of the excrescent w simply means, of course, that the labial rounding of the u- vowel lingers on after the articulation of the Jc- sound, a phonetic tendency encouraged by the fact that the produc- tion of the guttural consonant does not, as in the labials and dentals, necessitate a readjustment of the lips. A few examples will illustrate the phonetic process :

gelgulugwaf£n I desire it

gelgulu^w he desires it (contrast gelgulcfY he desired it, without

the labial affection of the -F because of the replacement of the

-u- by an -a-) guxwlfi his heart du^gwi't'gwa her dress duYw woman's garment yduk!waa his bones

As also in the upper Chinook dialects (Wasco, Wishram), where exactly the same process occurs, the w- infection is often very slight, and particularly before u- vowels the -w- is, if not entirely absent, at least barely audible :

yok!woyaf£n I know it

yo'Wyan I shall know it

In one very common word the catch seems to be treated as a Jc- sound in reference to a preceding u when itself followed by an -I- :

s'u^wili he sits; but

s'uf£aWaa he will sit

28 BUKEAU OF AMEKICAN ETHNOLOGY [BULL. 40

The first form was, for some reason or other, often heard, perhaps misheard, as s'isul%.

§ 10. INORGANIC a

It frequently happens in the formation of words that a vowel present in some other form of the stem will drop out, or, more accu- rately expressed, has never been inserted. Consonant-combinations sometimes then result which are either quite impossible in Takelma phonetics, or at any rate are limited in their occurrence to certain grammatical forms, so that the introduction of an " inorganic" -a-, serving to limber up the consonant-duster, as it were, becomes neces- sary. Ordinarily this -a- is inserted after the first consonant; in certain cases, after the two consonants forming the cluster. The theoretical future of gini'Yde* i GO SOMEWHERE should be, for example, *gink*dee; but, instead of this somewhat difficult form, we really get gina'lc*dee. That the -a'- is here really inorganic, and not a characteristic of the future stem, as was at first believed, is clearly shown by the imperative gVrik* (all imperatives are formed from the future stem) . Similarly :

lc!iyaflc'dee I shall go, come; aorist, Tcliyi'Yde*

alxikla'Ihik* ( = theoretical * cdxikJliJc*) he kept looking at him;

aorist first person alxik!rJM£n I keep looking at him Iclema'n make it! ( = theoretical *Jc!emn); cf. Tclemna'n I shall

make it baislyeewa'n drive out sickness!; aorist, -yewen he drove out

sickness sgela'u£ee I shall shout ( = theoretic *sgelwt*ee)', aorist second

person, sgelewaY you shouted

As an example of an inorganic -a- following a consonantic cluster may be given:

wisma'£ee I shall move (stem wism-) ; aorist, wits' !imt*e£ I moved1 The exact nature of the processes involved in the various forms given will be better understood when stem-formation is discussed. Here

iSuch an -a may stand as an absolute final; e. g., ba-imasga? START IN SINGING! (stem masg-), aorist third person, -mats.Wk'. The form masga* well illustrates the inherent difficulty of delimiting the range of a phonetic law without comparative or older historical material to aid in determining what is due to regular phonetic development, and what is formed on the analogy of other forms. The final cluster -sk' does occur in Takelma; e. g.,dink!a^sk' (long object) lay stretched out; so that a phonetic irregularity must exist in one of the two forms. Either we should have *ma^sk', orelse^dink.'asa^k' or *d^nk.rasga^ is to be expected. On closer examination it is found that the -fc' in forms like dink.Wsk' is a grammatical element added on to the future stem dink.'as-; whereas in masga* the -g- belongs in all probability to the stem, and is no added suffix; at least is not felt as such. It seems evident, then, that the quasi-mechanical juxtaposition of grammatical elements does not entirely follow the same phonetic lines as organic sound-complexes.

§ 10

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 29

it will suffice to say that there are three distinct sorts of inorganic or secondary a- vowels : the regular inorganic a first illustrated above, inserted between two consonants that would theoretically form a cluster; the post-consonantal constant a of certain stems (such as wism- above) that would otherwise end in more or less impracticable consonant clusters (this -a appears as -4 under circumstances to be discussed below); and a connecting a employed to join consonantal suffixes to preceding consonants (such suffixes are generally directly added to preceding vowels or diphthongs). The varying treatment accorded these different secondary a vowels will become clearer in the morphology.

§ 11. SIMPLIFICATION OF DOUBLE DIPHTHONGS

By a double diphthong is meant a syllable consisting of an ordinary diphthong (long or short) followed by a semivowel (y, w) or by I, m, or n. Such double diphthongs are, for instance, aiw, aiw, auy, auy, ain, din, alw, dalw; those with initial short vowel, like ain, have, like the long diphthongs (e. g. aan), a quantitative value of 3 morae, while those with initial long vowel, like din, have a quantitative value of 4 morae and may be termed over-long diphthongs. Double diph- thongs may theoretically arise when, for some reason or other, a con- necting or inorganic a fails to lighten the heavy syllable by reducing it to two (see particularly § 65 for a well-defined class of such cases). Double diphthongs, however, are nearly always avoided in Takelma; there is evidently a rhythmic feeling here brought into play, a dislike of heavy syllables containing three qualitatively distinct sonantic elements.

In consequence of this, double diphthongs are regularly simplified by the loss of either the second or third element of the diphthong; in other words, they are quantitatively reduced by one mora (the simple double diphthongs now have a value of 2 rnorae, the over- long diphthongs 3 morae like ordinary long diphthongs), while qualitivetatly they now involve only two sonantic elements. An exception seems to be afforded by double diphthongs in -uy (e. g. -any), which become dissyllabic by vocalizing the y to i, in other words, -any becomes -awi:

Tfc' he ran fast; cf. ' fe/a-uya/es fast runner, (aorist) you ran fast

6 1 shall talk; cf. yawayaY (aorist) you talked

30 BUKEATJ OF AMERICAN ETHNOLOGY [BULL. 40

The -awi- ( = theoretic -awy-) of these forms is related to the -away- of the aorist as the -ilw- of bilwa/£s JUMPER to the -iliw- of the aorist MliwaY YOU JUMPED.

Such double diphthongs as end in -w (e. g. -aiw, -dalw) simply lose the -w:

gal eat it! (=*gaiw); galk* he ate it (=*galwV); compare ga-iwa'n I shall eat it

Other examples of this loss of w are given in § 18, 2. All other double diphthongs are simplified by the loss of the second vowel (i, u) or consonant (I, m, ri); a glottal catch, if present after the second vowel or consonant, is always preserved in the simplified form of the double diphthong. Examples of simplified double diphthongs with initial short vowel are :

gelJiewe'Ji^n ( = *-^au£n) I think; compare gelJiewe'Ji&u he thinks imi'7i&£i\. (=*-^am£n) I sent him; compare imi'Tisun he sent him mo^ma^n (=*malen) I stir it up; mo'Z£man ( = *-maln) I shall

stir it up; compare parallel forms with connecting a: mo'los-

mala£n, mo'Z£malan, and third person aorist mo'?o£mal maanmsifeii ( = *-man£n) I count them; compare damdanmmi'£n

(umlauted from -man-i'%) I counted them up Jc!emxa't*ee (=*Jc!e>nmxa't*ee) I shall make; compare &/emna/ff&

maker and fc/ema'n make it! (with inorganic a because accent

is not thrown forward) Examples of simplified over-long diphthongs are :

da&ldi'n ( = *da,i(di'ri) I shall go to him for food; compare

d&lt*ee I shall go for food el t'gelxl* (=*t^g^l]xli) wagon (literally, rolling canoe); compare

t*ge*ycflx it rolls

dat!agsi£u (=*t!ag&isn} I build a fire; compare datlag&i he builds a fire

n (=*fc/emei£n) I make it; compare Jclemei he makes it £n ( = *o7/on£n) I give it; compare third person oyon he gives it

In the inferential, less frequently passive participle and impera- tive, forms of the verb, double diphthongs, except those ending in w, generally fail to be simplified. If coming immediately before the inferential -Y- the double diphthong is preserved, for what reason is not evident (perhaps by analogy to other non-aorist forms in which the last element of the double diphthong belongs to the following syllable) :

HANDBOOK OF INDIAN LANGUAGES TAKELMA

31

fe*/aimF (but also ts'!a,yam F) he hid it; compare ts'!&-ima'n I

shall hide it oinF he gave it; compare oma'n I shall give it

If the inferential -&'- does not immediately follow, an inorganic a seems to be regularly inserted between the second and third elements of the diphthong:

gelts'!a,yafmxamk*na£ since he concealed it from us Examples of other than inferential forms with unsimplified double diphthong are:

oln give it! (yet ts'I&ya'm hide it! with inorganic a)

Consonants (§§ 12-24) §12. System of Consonants

The Takelma consonant system is represented in the following table :

	Aspirated tenuis.	Voiceless media.	Fortis.	Spirant.	Lateral.	Nasal.
Labial . .	n	b	D!	v. unv. w -'«*		nt
Dental .	f	d	t!		I	n
Sibilant			is!, ts\'	•fr
Palatal				y	(0
Guttural	It'	g	k!	X
Faucal .			t	h

The spirants have been divided into two groups, those on the left- hand side of the column (labeled v.) being voiced, while those on the right-hand side (labeled unv.) are unvoiced. The rarely occurring palatal lateral I (see § 2, footnote) is also voiceless. Every one of the consonants tabulated may occur initially, except the voiceless labial spirant -*w, which occurs only with Jc at the end of a syllable. Prop- erly speaking, -¥w should be considered the syllabic final of the labialized guttural series (k*w, gw, Tc!w)', a consideration of the consonant-clusters allowed in Takelma shows that these labialized consonants must be looked upon as phonetic units. The catch (£) as organic consonant is found only medially and finally; the I only

§ 12

32 BUKEAU OF AMERICAN ETHNOLOGY [BULL. 40

initially. In regard to the pronunciation of the various consonants, w, Sj y, h, I, m, and n do not differ materially from the corresponding sounds in English.

The first two series of stops — tenuis Op', f, Y) and media (&, d, g) — do not exactly correspond to the surd and sonant stops of English or French. The aspirated tenues are, as their name implies, voiceless stops whose release is accompanied by an appreciable expulsion of breath. The voiceless mediae are also stops without voiced articula- tion ; but they differ from the true tenues in the absence of aspiration and in the considerably weaker stress of articulation. Inasmuch as our English mediae combine sonancy with comparatively weak stress of articulation, while the tenues are at the same time unvoiced and pronounced with decided stress, it is apparent that a series of con- sonants which, like the Takelma voiceless mediae, combine weak stress with lack of voice will tend to be perceived by an American ear some- times (particularly when initial) as surds, at other times (particularly between vowels) as sonants. On the other hand, the aspirated tenues will be regularly heard as ordinary surd-stops, so that an untrained American ear is apt to combine an uncalled-for differentiation with a disturbing lack of differentiation. While the Takelma tenuis and media are to a large extent morphologically equivalent consonants with manner of articulation determined by certain largely mechanical rules of position, yet in a considerable number of cases (notably as initials) they are to be rigidly kept apart etymologically. Words and stems which differ only in regard to the weak or strong stress and the absence or presence of aspiration of a stop, can be found in great number:

daan- ear; t'aan squirrel

bou now; p'ou- to blow

ga that ; Ya what

dl*- on top ; fl*- to drift

boud- to pull out hair; p'oud- to mix

daag- to build fire ; daag- to find ; t'aag-to cry

gai- to eat; Yai- thing, what x

i These two series of stops are not at all peculiar to Takelma. As far as could be ascertained, the same division is found also in the neighboring Chasta Costa, a good example of how a fundamental method of phonetic attack may be uniformly spread over an area in which far-reaching phonetic differences of detail are found and morphologic traits vary widely. The same series of stops are found also in Yana, in northern California. Farther to the east the two series are apparently found, besides a series of true sonant stops, in Ponca and Omaha (J. O. Borsey's p, t, k, and d , ?, q). The Iroquois also (as could be tested by an opportunity to hear Mohawk) are, as regards the manner of articulating the two series, abso- lutely in accord with the Takelma. A more accurate phonetic knowledge of other languages would doubt- less show a wide distribution in America of the voiceless media.

§ 12

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 33

The fortes (p!, t!, Jc!, ts! [ = ts'l], and £, which has been put in the same series because of its intimate phonetic and morphologic rela- tion to the other consonants) are pronounced with the characteristic snatched or crackly effect (more or less decided stress of articula- tion of voiceless stop followed by explosion and momentary hiatus) prevalent on the Pacific coast. From the point of view of Takelma, p!, t!, and Jc! are in a way equivalent to p£, ts, and Jc£, respectively, or rather to &£, d£, and g£, for the fortes can never be aspirated. In some cases it was found difficult to tell whether a fortis, or a voice- less stop followed by a glottal stricture, was really heard :

yap!a^ and yap-a" man ga'pfini* and ga'p£ini^ two

In fact, a final tenuis + a catch inserted, as between vowels, to pre- vent fhonetic amalgamation, regularly become, at least as far as acoustic effect is concerned, the homorganic fortis:

aklcf he indeed (=aY he + deictic ecf; cf. ma'ecf you indeed) saklelt* you shot him ( = saY he shot him + (e)elf you are) mapta^ just you [pi.] (= map1 you [pi.] 4- sa?)

Nevertheless, p£, t£, Jc£ are by no means phonetically identical with p!j t!, Jc!; in Yana, for instance, the two series are etymologically, as well as phonetically, distinct. One difference between the two may be the greater stress of articulation that has been often held to be the main characteristic of the fortes, but another factor, at least as far as Takelma (also Yana) is concerned, is probably of greater mo- ment. This has regard to the duration of the glottal closure. In the case of p£, t£, and Jc8 the glottis is closed immediately upon release of the stop-contact for p, t, 'and Jc. In the case of p!, t!, and Jc! the glottis is closed just before or simultaneously with the moment of con- sonant contact, is held closed during the full extent of the consonant articulation, and is not opened until after the consonant release; the fortis p!} e. g., may be symbolically represented as £p£ (or £b£, better

£

as £&£, i. e., a labial unaspirated stop immersed in a glottal catch). As the glottis is closed throughout the whole extent of the fortis articulation, no breath can escape through it; hence a fortis conso- nant is necessarily unaspirated. This explains why fortes are so apt to be misheard as voiceless mediae or even voiced mediae rather than as aspirated tenues (p!, e. g., will be often misheard as & rather than p). The cracked effect of the fortes, sometimes quite incorrectly 3045°— Bull. 40, pt 2—12 3 § 12

34 BUKEATJ OF AMERICAN ETHNOLOGY [BULL. 40

referred to as a click, is due to the sudden opening of the closed cham- ber formed between the closed glottis and the point of consonant contact (compare the sound produced by the sudden withdrawal of a stopper from a closed bottle) ; the hiatus generally heard between a fortis and a following vowel is simply the interval of time elapsing between the consonant release and the release "of the glottal closure.1 That the fortis consonant really does involve an initial glottal catch is abundantly illustrated in the author's manuscript material by such writings as:

dulti'H!ilien=dulu't!ilien I stuff it

du'let!iUn==du'U!ilin I shall stuff it

leme'£Tc!ia-uda£ = leme'lc!ia-uda£ as they go off

Many facts of a phonetic and morphological character will meet us later on that serve to confirm the correctness of the phonetic analysis given (see §13, end; also §§ 30,4; 40,6; 40,13a, p. 113; 40,13b). Here it is enough to point out that p!, t!, Tc!, ts'! are etymologically related to ~b, d, g, s' as are i£, us, £l, £m, £n to i, u, I, m, n.

There is no tenuis or media affricative (ts — dz; ts', tc — dz', dj) corre- sponding in Takelma to the fortis ts!, ts'!, though it seems possible that it originally existed but developed to x (cf. yegwexi they bite me [upper Takelma yegwe'tci]', ts'Ii'xi dog [from original *ts* tits' it2]) . Morphologically ts!, ts'! stand in the same relation to s, s' that p!} tl, and Tc! stand in to &, d, g. For example, Aorist stems :

tfomom- kill, ptugug- start (war, basket), Tdolol- dig — are related

to their corresponding Future stems :

doum-, bu^g-, goul-, — as are the Aorist stems:

ts'tadad- mash, ts'Ielel- paint — to their corresponding Future stems:

s'aad-, s'eel-

Of the other consonants, only x, -*w, and s, s' call for remark, x is equivalent to the ch of German DACH, though generally pronounced further forward (x). It frequently has a w tinge, even when no it-vowel or diphthong precedes, particularly before i'} examples are M'pxwi CHILD and Tiaxwiyc? (ordinarily Tiaxiya?) IN THE WATER. -Fw,

1 Doctor Goddard writes me that an examination of tracings made on the Rousselot machine leads to substantially the same phonetic interpretation of the fortes as has been given above.

2 See Notes on the Takelma Indians of Southwestern Oregon, American Anthropologist, n. s., ix, 257.

§ 12

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 35

in which combination alone, as we have seen, -*w occurs, is the aspirated tenuis Y followed by a voiceless labial continuant approxi- mately equivalent to the wh of English WHICH, more nearly to the sound made in blowing out a candle, s is the ordinary English s as in SELL; while s' is employed to represent a sibilant about midway in place of articulation between s and c (= sJi in English SHELL), the fortes is! and tsm! corresponding, respectively, in place of articulation to s and s\ The two sounds s and s' have been put together, as it is hardly probable that they represent morphologically distinct sounds, but seem rather to be the limits of a normal range of varia- tion (both sal- WITH FOOT and s'al-, e. g., were heard). The only distinction in use that can be made out is that s occurs more fre- quently before and after consonants and after £:

s'a's'ant'e* I shall stand

ogu's'i he gave it to me, but ogu'sbi he gave it to you

lous'lfi his plaything 110.6

llasgi'n I shall touch it

leepsi" feathers

yols steel-head salmon

Jia-uJiana'£s it stopped (raining)

§ 13. Final Consonants

By a "final" consonant will always be meant one that stands at the end of a syllable, whether the syllable be the last in the word or not. Such a final position may be taken only by the aspirated tenues, the voiceless spirants, the catch, the liquid (Z) , and the nasals, not by the voiceless mediae, fortes, and semivowels (y and w) ; Ti occurs as a final only very rarely :

Icfh excrement

lohlalicfnY he always caused them to die

A final semivowel unites with the preceding vowel to form a diph- thong :

gayati he ate it (cf . gayawa'£n I ate it)

gal grow! (cf. gaayaf£t* he will grow)

A final voiceless media always turns into the corresponding aspirated surd; so that in the various forms of one stem a constant alternation between the two manners of articulation is brought about:

seeba'£n I roasted it; sep* he roasted it

xebef£n he did it; xep'ga* I did it

xuduma'lda£n I whistle to him; xuduma'W, xuduma'lt" gwa he whistles to him

t!ayaga'£n I found it; U.aycfY he found it, dak*na£ since he found it

§ 13

36 BUKEAU OF AMEKICAN ETHNOLOGY [BULL. 40

A final fortis also becomes the corresponding aspirated surd (-ts! becoming -£s), but with a preceding catch by way of compensation for the loss of the fortis character of the consonant. This process is readily understood by a reference to the phonetic analysis of the fortes given above (§ 12). Final p!, for instance, really £Z>(£), is treated in absolutely parallel fashion to a final &; the final media implied in the p! must become an aspirated surd (this means, of course, that the glottal closure is released at the same time as the stop, not sub- sequently, as in the ordinary fortis), but the glottal attack of the £b still remains. Examples are:

wasgdfp!in I shall make it tight; wasgaf£p* make it tight Yap!afYap'na£n I throw them under (fire, earth) ; future, Va£p*-

baaxd't!an I shall win over him; t>aaxd'£t* win over him! baaxo'£t*gas

I won over him alxtfJclin I shall see him; alxl'£Y see him! (contrast alxl^g^n I

saw him; cblxlfiY he saw him) Jia£wlhaf nts !in I shall cause it to stop (raining) ; Ji,(L£wlha'n£s

make it stop raining!

no'tslafgwan next door to each other; no'u£s' next door 7iaelmift8ladan t!eimi'£s six times 100; Tia*imi'£s six

Consonant Combinations (§§ 14-17}

§ 14. GENERAL REMARKS

Not all consonant combinations are allowable in Takelma, a cer- tain limited number of possibilities occurring initially, while a larger number occur as finals. Medial combinations, as we shall see (§17), are simply combinations of syllabic final consonants or permissible consonant combinations and syllabic initial consonants or permis- sible consonant combinations.

§ 15. INITIAL COMBINATIONS

If, as seems necessary, we regard gw as a single labialized consonant, the general rule obtains that no combinations of three or more con- sonants can stand at the beginning of a word or syllable. The fol- lowing table shows all the initial combinations of two consonants possible in Takelma, the first members of the various combinations being disposed in vertical columns and the second members, with which the first combine, being given in horizontal lines. Examples fill the spaces thus mapped out. Inasmuch as the mediae and fortes,

§£ 14-15

BOAS]

HANDBOOK OF INDIAN LANGUAGES TAKELMA

37

the liquid, nasals, semivowels, and 7i never appear, or with very few exceptions, as the first members of initial combinations, it was not con- sidered necessary to provide for them in the horizontal row. Simi- larly the tenues and fortes never occur as second members of initial combinations. A dash denotes non-occurrence.

	p'	'	fcf	6'	X
b	— •	t'bdag- hit	•	sbln beaver	?
d	—			s-do'is-dagwa- put on style	xdeU' flute
Q	—	t'geib- roll		sgi'si coyote
gw	—	t'gwa* thunder		sg wini^ raccoon
J }	—
I	—			?	xliwi war feathers
m	—	t'mila^px smooth		sma-im- smile	?
n	—	—		s-na mamma!	xnW acorn mush
y	—
w		t'waplat'wap'- blink	[k' waa gw- awaken]	swat'g- pursue	?

It will be noticed that only t* (p' and Y were given mainly for contrast) and the two voiceless spirants s and x combine with fol- lowing consonants (Fw- is not to be analyzed into Y +w, but is to be regarded as a single consonant, as also gw- and Jc!w-, both of which frequently occur as initials) ; furthermore that s, x, and y never com- bine with preceding consonants. The general law of initial combi- nation is thus found to be: tenuis (?) or voiceless spirant (s, x) + media (6, d, g) or voiced continuant (Z, m, n, w).1 Of the combina- tions above tabulated, only t*b- t*g-, sb-, sg-, and perhaps sgw- and sw-, can be considered as at all common, t*m-, t*w-, sd-, sn-, xd-, xl-, and xn- being very rare, si-, sb-, xm-, and xw- have not been found, but the analogy of xl- for the first, and of si)-, sm-, and sw- f or the others, make it barely possible that they exist, though rarely ; there may, however, be a distinct feeling against the combination x + labial (b, m, w).

Only two cases have been found of fortis or media + consonant :

tlweple'tlwapx they fly about without lighting; future dwep'- dwa'pxda*

This may possibly serve to explain why the aflricative is- (to correspond to is-.') is not found in Takelma.

§ 15

38

BUREAU OF AMERICAN ETHNOLOGY

fBULL. 40

§ 16. FINAL COMBINATIONS

Final consonant combinations are limited in possibility of occur- rence by the fact that only aspirated tenues and voiceless spirants (p\ t\ Y, Jc'v, s, and x) can stand as absolute finals after other con- sonants. The following table will give examples of all final combi- nations of two or three consonants that have been discovered in the available material.

	P'	f	fc1	/	m	n	s	X
p'		eit'p' ye are	-	&?lp' swan		s-a's-anp' stand! (pi.)
t'			-	sgelewa'lt' he shouted to him	ts.'dela'mt' he paints it	p/d'ant'his liver
k'	xZp'k'hedidit	p'j?«aTk' my sal- mon	-	a Ik' silver-side salmon	xaxmk' grizz- ly bear	do«77ic^nk' he will kill him	mz/avsk' he loved her	k'wd'a-xk' he's awake
k'w			-	fgweW* rat	?	yarik'* he took it along
p'k-		'	-	*-M'£alp'k' he sat		se'nsanp'k' he whooped
fV			-	do«mavlt'k' my testicles	zda/a'mt'k' my urine	bilga^nVk' my breast
s	Zaxps blanket		-	blls moss	gums blind	pfe^ns squirrel		•
	t'geya^px round		-	t'geeya^\x i t rolls	ya'mx grease	6i5nx hun- ger
xk"	des-lpxk' i t closed		-	gii'lk.'alxk' it was blazing	date-/dvmxk' it hurt	ugwa^nxk' he drank
px			-	sgiUpx warm your back!		?

No examples of -mVw and -npx have been found, but the analogy of -Ipx makes the existence of the latter of these almost certain (Z'and n are throughout parallel in treatment) ; the former (because of the double labial; cf. the absence of -mp') is much less probable, despite the analogy of -IV w and -nk'w. It is possible also that -IsV, -msk\ and -nsV exist, though their occurrence can hardly be frequent. Of final clusters of four consonants -nt*p*Y has been found in s'a's'anVp^Y HE STOOD, but there can be small doubt that the -t- is merely a dental tenuis glide inserted in passing from the dental nasal to the labial tenuis; compare the morphologically analogous form se'nsanp^Y HE WHOOPED. However, the combinations -IpxY and -npxY (if -npx exists), though not found in the available material, very probably ought to be listed, as they would naturally be the terminations of morphologically necessary forms (cf. des'lpxk*). Most, if not all, of

§ 16

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 39

the preceding final combinations may furthermore be complicated by the addition of £, which is inserted before the first tenuis or voiceless spirant of the group, i. e., after a possible liquid or nasal:

ti'i£s'V he laughed ¥o'£px dust, ashes. ts'!u'n£s (deerskin) cap

As compared to the initial combinations, the table of final clusters seems to present a larger number of possibilities. It is significant, however, that only those that consist of Z, m, or n + single consonant can ever be looked upon as integral 'portions of the stem (such as xcfmk* and t'gwe^llc'w) ; while those that end in -s can always be sus- pected of containing either the verbal suffix -s ( = t + x), or the noun and adjective forming element -s. All other combinations are the result of the addition of one or more grammatical elements to the stem (e. g., s'u'scilpW = s'ue(ilr+p* + lc*). Further investigation shows that only two of the combinations, -t*p* (second personal plural sub- ject aorist) and -W (first personal singular possessive) are suffixal units ; though -t*p* might be ultimately analyzed into -t* (second per- sonal singular subject aorist) + -p\ It is interesting to note that these clusters are at the same time the only ones, except t*gw-, allowed initially, t*b- and t*g-. The constitution of the Takelma word-stem may thus be formulated as

tenuis (or voiceless spirant) + media (or voiced continuant) + vowel (or diphthong) + liquid or nasal + stop (fortis or media — tenuis) ,

any or all of the members of which skeleton may be absent except the vowel; h may also be found before the vowel.

§ 17. MEDIAL COMBINATIONS

A medial combination consists simply of a syllabically final com- bination or single consonant + an initial combination or single con- sonant, so that theoretically a very large number of such medial combinations may occur. Quite a large number do indeed occur, yet there is no morphologic opportunity for many of them, such as V-l, np'-m, and numerous others. Examples of medial combinations are:

t!omoma'n-ma£ when he was killed

helk*-na£ when he sang

dak*-t*gu'uba£n I put hollowed object (like hat) on top (as on head)

§ 17

40 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

The occurrence of such clusters as -Yn- must not for a moment be interpreted as a contradiction of the non-occurrence of the same clus- ters initially or finally, as they are not, syllabically speaking, clusters at all. Had such combinations as, say, -t'gn- (in which -t* would be the final of one syllable and gn- the initial of the next) occurred, we should be justified in speaking of an inconsistency in the treatment of clusters; but the significant thing is, that such clusters are never found. A Takelma word can thus ordinarily be cut up into a definite number of syllables :

ga$lc*na£ when he ate it ( ='galY-na£) yo'Yyan I shall know it ( = yo'Y-yan}

but these syllables have only a phonetic, not necessarily a morpho- logic value (e. g., the morphologic division of the preceding forms is respectively gai-Y-na£ and yok*y-ari). The theory of syllabification implied by the phonetic structure of a Takelma word is therefore at complete variance with that found in the neighboring Athapascan dialects, in which the well-defined syllable has at least a relative morphologic value, the stem normally consisting of a distinct syllable in itself.

One important phonetic adjustment touching the medial combina- tion of consonants should be noted. If the first syllable ends in a voiceless spirant or aspirated surd, the following syllable, as far as initial stops are concerned, will begin with a media (instead of aspi- rated surd) or aspirated surd + media; i. e., for a cluster of stops in medial position, the last can be a media only, while the others are aspirated surds. As also in the case of single consonants, this adjust- ment often brings about a variation in the manner of articulation of the final consonant in the cluster, according to whether its positi'on in the word is medial or final. Thus we have:

xep'gas I did it; xep*Y he did it Contrast, with constant -V- :

alxl'£Ya£ I saw it; alxi'fy*1 he saw it

the -g- of the first form and the -Y of the second being the same mor- phological element; the -p* of both forms is the syllabically final & of the stem xeel- DO, so that xep'ga£ stands for a theoretical *xelik*a£, a phonetically impossible form. Other examples are :

1 This form is distinct from alxi'ek' LOOK AT IT!, quoted before. The imperative theoretically = *alxl'k! the text form

$ 17

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 41

ga-iwa't*ba£ ye shall eat it ; gayawaYp* ye ate it di'n£xga£ I (as long object) was stretching out ; di'n£xk' long object was stretching

Consonant Processes (§ § 18-24)

§ 18. DROPPING OF FINAL CONSONANTS

There is a good deal to indicate that the comparatively limited number of possible final consonant-clusters is not a primary condi- tion, but has been brought about by the dropping of a number of consonants that originally stood at the end.

1 . The most important case is the loss of every final -t* that stood after a voiceless spirant or aspirated surd. Its former presence in such words can be safely inferred, either from morphologically par- allel forms, or from other forms of the same stem where the phonetic conditions were such as to preserve the dental. Thus gwidWw HE THREW IT represents an older reduplicated *gwidi^wt^ ( = gwid-i-gwd-) , as proven by the corresponding form for the first person, gwidi'Jc*wdaen i THREW IT and gwidi'Vdagwa HE THREW HIM (122.13). Similarly all participles showing the bare verb stem are found to be phonet- ically such as not to permit of a final -t\ and are therefore historic- ally identical with the other participial forms that show the -C :

saV shooting ( = *sak*t*) dox gathering ( = *doxf) Tia-tlulY following in path ( = **/flZfcY) sana^p* fighting ( = *8(mcfp*t*) Compare:

yanaY going lohoY dead sebe^t* roasting domt* having killed se'nsant* whooping yi*W copulating with

The combinations -Ywt'¥ (-Vwt'g-) and -Jc*wt*x-, however, seem to lose, not the -£'-, but the -&'">-, whereupon -£'&' (-t*g~) remains, while -t'x- regularly becomes -s- (see § 20, 2) :

7ieeSgwidaW ( =*gwidaWwf-k' , inferential of gwidik*wd-)he lostit hee£gwida't*ga£ (=*gunda'Fwt*-ga£) I lost it

xamgwidi'sgwide£ (=*gwidi'lc*wt*-x-gwi- or possibly *gwidi'Ywt*- gwi-) I drown myself

§ 18

42 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

2. Somewhat less transparent is the former existence of a -w after consonants. The following examples have been found in the material at disposal:

lal she twined basket ( = *lalw) ; cf . laalwa'£n I twine it (that -w

really belongs to the stem is shown by the forms laawa'n

I shall twine it ; letixi twine it for me !) Jclel basket bucket ( = *7c!elw)', cf. klelwi'* her bucket Val penis ( = *Yalw) ; cf . Valwlfi his penis. sgelel£ (=*sgelel£w) he keeps shouting; cf. sgelewaY you shout,

sgelwa'We* I shall keep shouting alsgalYa? (=*sgalwVa£) I turned my head to one side to look at

him; cf. alsgaalwifn I shall turn my head to look at him alsgelelxi ( =*sgelelwxi) he keeps turning his head to one side to

look at me; cf. alsgalaaUwi'£n I keep turning my head to look

at him, future alsgalwalwi'n

This process, as further shown by cases like gal EAT IT! (=*gaiw), is really a special case of the simplification of double diphthongs (see § 11). Perhaps such "dissimilated" cases as Za°- and lee- (for Idu- and leu-), see § 7, really belong here.

Other consonants have doubtless dropped off under similar condi- tions, but the internal evidence of such a phenomenon is not as satisfactory as in the two cases listed. The loss of a final -n is probable in such forms as lh,egwe'JiaYw HE WORKS, cf. ihegwe'7iaTc*wnasn i WORK, and Ihegwe'halc^nanaW WE WORK. Certain verb-forms would be satisfactorily explained as originally reduplicated like gwidiW™, if we could suppose the loss of certain final consonants:

gini'£Y he went somewheres ( = ^gin-i'-^Yri)

gelgulu"¥w he desired it (=(!*-gul-u^-Jc'wl)

In the case of these examples, however, such a loss of consonants is entirely hypothetical.1

§ 19. SIMPLIFICATION OF DOUBLE CONSONANTS

Morphologically doubled consonants occur very frequently in Ta- kelma, but phonetically such theoretic doublings are simplified into single consonants; i. e., k*+g become F or g, and correspondingly for other consonants. If one of the consonants is a fortis, the simpli- fied result will be a fortis or aspirated surd with preceding catch, according to the phonetic circumstances of the case. If one of the

1 Many of the doubtful cases would perhaps be cleared up if material were available from the upper dialect, as it shows final clusters that would not be tolerated in the dialect treated in this paper; e. g. Kii'toiufks-e RELATIVES (cf. Takelma k'winaxde MY KIN).

§ 19

BOAS] HANDBOOK OP INDIAN LANGUAGES TAKELMA 43

Jc- consonants is labialized, the resulting Jc- sound preserves the labial affection. Examples of consonant simplification are:

mo'VeY my son-in-law (= mo'f- + -dele')

latfwoY he gave him to eat ( = lag- + -Jc'woJc')

deJc!iya'Jc'i£ if it goes on (= dekliya'g- + -Jc'i£)

Itfgwa'n I shall fetch them home ( = lllg- + -gwari) ; cf . aorist

ligigwa'sn dtfhila'Tc !wemesn I make him glad (= TiilcfY™ glad + Jc!eme£n I

make him)

A good example of three fc-sounds simplifying to one is : ginak'wis if he comes (= ginag-Yw-Yi£}

The interrogative element di never unites with the -f of a second person singular aorist, but each dental preserves its individuality, a light % being inserted to keep the two apart :

xemda/Mdi do you wish to eat ? ( = xemelaY + di)

The operation of various phonetic processes of simplification often brings about a considerable number of homonymous forms. One example will serve for many. From the verb-stem saag- SHOOT are derived :

1. Imperative sak' shoot it!

2. Potential saV he can, might shoot it

3. Participle saY shooting ( = *sa£Y)

4. Inferential saV so he shot it ( = *sag-k*)

The corresponding forms of the stem yana- GO will bring home the fact that we are here really dealing with morphologically distinct formations :

1. yana^ go!

2. yanaf£ he would have gone

3. yanaY going

4. yanaW so he went

Another simplification of consonant groups may be mentioned here. When standing immediately after a stop, an organic, etymo- logically significant Ji loses its individuality as such and unites with a preceding media or aspirated tenuis to form an aspirated tenuis, with a preceding fortis to form an aspirated tenuis preceded by a glottal catch (in the latter case the fortis, being a syllabic final, cannot preserve its original form). Thus, for the Tc- series, g or Y +7i becomes V , ~k! (or £F) +h becomes £F; gw or Yw -\-Ji becomes Jc*w, lc!w (or £lc'w) +7i becomes *Vw. Under suitable conditions of accent

§ 19

44 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

(see § 23) the contraction product V or Yw may itself become g or gw, so that all trace of the original h seems to be lost. Examples for the Tc- sounds 'are:

t'gunuJc*is (=CgunuY + quotative -hi£) it became warm, it is said

nag<ma'alc'i£ ( = nagana'a£lc' + quotative -hi£; see § 22) he always said, it is said

gwen-he'Ywaagw- (= reduplicated he'gw-hdagw-) relate; with ac- cent thrown forward gwen-hegwa'agw-an-i- (=hegw-hafagw-}; compare, with preserved h} gwen-hegwe'hagw-an-i tell to

s'o'wo£Y6p" ( = s'ofwo£Y-hap* =*s'o'wok!-hap*) he jumps (6 = wa; see § 9) he jumps; compare s'owo'lcfanafn I cause him to jump

Similarly, d or t* +h becomes t*, t! (or £t*) +7i becomes ££'; 5 or p* + 7i becomes p*, p! (or £p*) + h becomes £p* :

gana'fi (=gana?t* + emphatic -hi) of just that sort

yo't"i (=yoY being + emphatic -hi) alive; compare plural

yot'i'hi hee£sgu'u£t'6¥w (=sgu'ut!-Jiak'w) cut away; compare Jieessgo'ut!an

I shall cut it away

s' and x also generally contract with Ji to s' and x, e. g. : ndus'i/e(=nouss'+-hi£) next door, it is said.

§ 20. CONSONANTS BEFORE x

No stopped consonant or spirant may stand before x, except p. The dentals, guttural stops, and sibilants all simplify with x into single sounds; the fortes (including ts!) following the example of the ordinary stops and of the s, but leaving a trace in the vicarious £.

1. All Jc- sounds (F, g, Jc!, Yw, gw, Tc!w) simply disappear before x without leaving any trace of their former existence, except in so far as lc! and Jc!w remain as £; if x is followed by a vowel, the w of the labialized fc-sounds unites with x to form xw:

alxlfixi he saw me ( =al-xlrig-xi) ; cf. alxlHgi£n I saw him Ywd'axde£ I awoke (=Jc'wdfagw-x-dee)'} cf. lVwofagwi£n I woke

him up gelgulu'xbi£n I like you ( = -gulu'gw-x-bi£ri) ; cf . -gulugwa,rsn I

like him baadini'£x (clouds) spread out on high ( =-diniflc!-x) ; cf. di'nik!a£n

I stretch it out

lu£xwa? to trap ( =luk!w-xa?) ; cf. lo'~k!wan I shall trap (deer) yexwinY ( =yegw-xink') he will bite me; but yexda6 ( =yegw-x-da£)

you will bite me § 20

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 45

2. tx always simplifies to s, t!x to £s. Whether the combination tx really spontaneously developed into s it is naturally impossible to say; all that can safely be stated is that, where we should by mor- phologic analogy expect t + x, this combination as such never appears, but is replaced by s. Examples are numerous:

lebe'sa* she sews (^ebe't-xa^] cf., for -t' of stem, lebeY she

sewed it, for suffix -xa£, lobo'xa* she pounds sgelewa'lsi he shouts to me ( =sgelewa'ld-xi) ; cf. sgelewoflda£n I

shout to him d&ibodobd 'sasn they pull out each other's hair, with reduplicated

stem bodobad- + x- xaat*lefe£¥t*bagams it is all tied together (=-fbagamt-x)'j cf.

xda£bd'agamda£n I tie it together hansgd'u£s he cut across, lay over (road) ( =-sgo'ut!-x) ; cf .

Jiansgd'ut!an I shall cut it across

This change of tx to s is brought about constantly in the course of word-formation, and will be incidentally exemplified more than once in the morphology.

3. sx simplifies to s, ts!x (=ssx) to £s. Examples are:

yimi's'a* he dreams ( =yimi's'-xa£, with suffix -xa£ as in lobo'xa*

above Tia-uhanaf£s it stopped (raining) (=*-Jiana'£sXj stem ~hanats!- +

-x)

§ 21. DISSIMILATION OF n TO /AND m

If a (generally) final n of a stem is immediately followed, or, less commonly, preceded by, a suffix containing a nasal, it dissimilates to Z. The following examples have been found:

yalalaTiaY you lost it . (cf . yalnanada's you will lose it, with n

preserved because it forms a consonant-cluster with Z) 7ia-gwdal-a^m in the road (cf. gwan road)

Dldalcfm Grant's Pass (probably =over [drir] the rocks [da?n]) cca°ZaWF my urine; xalcb'xam£e£ I urinate (cf. xdn urine) la-is'in-xirlik!wi£n I blow my nose, with Z due to -n of prefix

s'in- nose (cf. xln mucus) s'inp*i'l£s flat-nosed, alongside of s'inp*i'n£s

The possibility of a doublet in the last example shows that the prefix s'in- is not as thoroughly amalgamated with the rest of the word as are the suffixes; probably, also, the analogy of forms m-p'in£s with other prefixes not containing an n would tend to restore an anomalous-sounding s'inp*i'l£s to -p*i'n£s.

§ 21

46

BUREAU OF AMERICAN ETHNOLOGY

[BULL. 40

A suffixed -(a)n dissimilates to -(a)l because of a preceding m in the stem:

s'imi^l dew (cf. such nouns as pfiyi^n deer) dak*-s'duma^l on the mountain (s'om mountain) douma'lt'LV my testicles (doum testicles)

With these compare:

dda-ts!aawa^n by the ocean (tsldu deep water)

In xda-gulma^n AMONG OAKS, the Z immediately preceding the m seems to have prevented the dissimilation of the -an to -al.

It is practically certain that the -am of Jiagwaala^m, Dldala^m, and xaala'mCY is at bottom phonetically as well as functionally identical with the suffix -an (-al), seen in xaa-gulma^n (gulifm OAK) and daV- s'duma^l, and rests on a second dissimilation of the nasal lingual (n) of the suffix to a labial nasal (m), because of the lingual (Z) of the stem. The history of a word like Jiagwaala^m is in that event as follows: An original *hagwaana^n IN THE ROAD (stem gwaan--+ nominal characteristic -an) becomes first *liagwaala^n by the dissimilation of the first n because of the following n, then Jiagwdala^m by the .dissimi lation of this second n because of the preceding Z. Similarly Dldala^m and xaala^m£V would go back to *Didana^n and *xaancfnt*Jc* respec- tively ; with the second form compare the reduplicated verb xala'xam- ( = *xanaxan-) URINATE. The probability of such a dissimilation of n to m is greatly strengthened by the fact that nearly all nouns with an evidently suffixal noun-forming element -(a)m have an Z in the stem as compared to an -(d)n of nouns not so affected. Contrast:

-n

7keelavm board (cf. dtfhe'liya

sleeping on wooden platform) <7elavm river

hail (cf. stem tslel- rattle)

sick, ghost

ts'Iu^lm wart * habi\B?m empty

frog

turtle

wigm red lizard

pliyi^n deer (-n here as suffix

shown by pliya^x fawn) yutlu^n white duck (cf. yutl-

u'yidi£n I eat it greedily) yu'xg&n trout xdan. eel (cf . ha£-xdd'axdagwa£n

I throw something slippery

far away) woup!\m- eyebrows

* No other example of final -lm is known, so that this form was probably misheard for (ct. gulu*m OAK).

§ 21

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 47

yulvfm eagle (also 7/tdaNm is dda- n- ear

found)

guWm oak beb&n rushes

Jc'ulum fish (sp. ?) #a'&/an house ladder

\egem- kidney gwitlm- wrist

It should not be concealed that a few words (such as Jiultin OCEAN, tlaga^m LAKE, and yuk!um-a- BONES) do not seem to conform to the phonetic law implied by the table ; but more exact knowledge of the etymology of these and similar words would doubtless show such disagreement to be but apparent. It is probable that in delga^n- BTJTTOCKS, ~bilgcfn- BREAST, and do'Win-i- ANUS, the g, (F) im- mediately following upon the I prevented the expected dissimila- tion of n to m; in le'Ywan- ANUS the dissimilation was perhaps thwarted by a counter-tendency to dissimilate the two labials (k*w and m) that would thus result. *yalan-an- LOSE (tr.), dissimilated, as we have seen, to yalal-an-, fails to be further dissimilated to *yalal- am- because, doubtless, there is a feeling against the obscuring of the phonetic form of the causative suffix -an-. The great probability of the existence of a dissimilatory tendency involving the change of n to m is clinched by the form do'Wim-i- ANUS alongside of •do'Win-i-.

A dissimilation of an original Z to n (the reverse of the process first described), because of an I in the stem, is found in

yiltfnma^n I keep asking for it ( = original *yillilma'£n [ I inserted as repetition of stem -I- in iterative formation from yilima'£n I ask him])

leeba'nxde£ I am carrying (object not specified) (= original *lee- ba'lxde£) ; cf . identical suffix -al-x-, e. g., gayawa'lxde* I eat.

In uugwa'nxde£ I DRINK (stem ugw-), it hardly seems plausible that -an-x- is at all morphologically different from the -al (-an) -x- of these words, yet no satisfactory reason can be given here for a change of the I to n.

§ 22. CATCH DISSIMILATION

If to a form with a glottal catch in the last syllable is added a syn- tactic (conjunctive) element, itself containing a catch, the first catch is lost, but without involving a change in the character of the pitch- accent ; the loss of the catch is frequently accompanied by a length- ening of the preceding vowel (or rather, in many cases, a restoration of the original length) . This phonetic process finds its most frequent

22 §

48 BUEEAU OF AMEBICAN ETHNOLOGY [BULL. 40

application in the subordinate form of the third person aorist intransitive :

ya'ada£ when he went (cf . ya'£ he went) gimfiYda£ when he went to (cf. ginif£Y he went to) yawa'ida£ when he spoke (cf . yawa'is he spoke) loTio'ida* when he died (cf . loho'i£ he died)

The connectives -hi£ IT is SAID, and -s'i£ BUT, AND are, in regard to this process, parallel to the -da£ of the preceding forms:

nagcb'ihi£ he said, it is said (cf. naga'i£ he said) nous'i'£ but, so (he went) next door (cf. no'u£s' next door). a'nls'is but not (cf. a'nl£ not)

£l's'is'i£ but no matter how (often) (cf. £i's'i£ even if) dal£wlfis'i£ but some (cf. dalswi'£ sometimes; -wlfis'i£ is related to -uri'£ as is ya'ada£ to ya'£)

§ 23. INFLUENCE OF PLACE AND KIND OF ACCENT ON MANNER

OF ARTICULATION

The general phonetic rule may be laid down that an aspirated surd, when not immediately followed by another consonant, can, with com- paratively few exceptions, be found as such medially only when the accent immediately precedes, provided that no consonant (except in certain circumstances Z, m, and ri) intervene between the accented vowel and the aspirated surd; under other conditions it appears as a media. This phonetic limitation naturally brings about a con- stant interchange between the aspirated surd and the correspond- ing media in morphologically identical elements. Thus we have as doublets -da and -t*a, third person possessive pronoun of certain nouns :

&emt'aa his stick se'eZt'aa his writing wila'ut'si* his arrow #a'Zt'aa his bow mo't'aa his son-in-law; but da'gaxda, his head

and numerous other nouns with -x-. This consonant in itself, as we have seen, demands a following media. Another pair of doublets is -de£ and -t'e£, first person singular subject intransitive aorist (-dee and -t'ee to correspond in future) :

p*ele'xade£ I go to fight; p'elxa't*ee I shall go to war yant*e£ I go ; yana't'e* I shall go nagalt*e£ I say; raz/t'e6 1 shall say § 23

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 49

but:

wits' !lsmade£ I keep moving; future wits' /e'smade6 (contrast

wits'!imt'e£ I move and wisma't*ee I shall move) Other examples of interchange are :

sgout'sga'Vi he cut them to pieces; sgo'ut'sgidi£n I cut them to

pieces ts'!umttmt*a£n I boil it, s'Umt'an I shall boil it (stem s'uum-t'a-) ;

s'omoda'£n I boil it, s'omda'n I shall boil it (evidently related

stem s'om-d-) S'as'inlp*ik* we stand; 6cbirF we are

This phonetic rule must not be understood to mean that a media can never appear under the conditions given for the occurrence of a surd. The various grammatical elements involved are not all on one line. It seems necessary to assume that some contain a surd as the primary form of their consonant, while others contain an organic media. The more or less mechanical changes in manner of articula- tion, already treated of, have had the effect, however, of so inextri- cably interlocking the aspirated surds and mediae in medial and final positions that it becomes difficult to tell in many cases which manner of articulation should be considered the primary form of the consonant. Some of the medially occurring elements with primary tenuis are:

-£'a, third person possessive

-t'a, exclusive (as in JcIwa'Wa young, not old; younger one)

-£V, first person intransitive aorist (future, -t'ee)

-t*elc*, first person singular possessive (as in gafWeY my bow)

Such elements show an aspirated consonant whether the preceding accent be rising or falling; e. g., bemt'a like Jie'elt'a. Some of those with primary media are:

-da, third person possessive with preceding preposition (corre- sponding not to first person -t'ek*, -dek', but to -de) -a'ld- and -a'md- indirect object -dae, subordinating element

This second set regularly keep the media whether the accent imme- diately precedes or not. The first two of these generally, if not always, require the preceding accent to be a falling one:

dak* will' tda on his house hat'gd/ada in his country xaasa'lda between his toes xaaliafmda on his back 3045°— Bull. 40, pt 2—12 4 § 23

50 BUREAU OF AMERICAN ETHNOLOGY [BULL/40

liawa'nda under him sgelewa'lda£n I shout to him ts!elela'mda£n I paint it

The third retains its primary character as media when the preceding verb form has the falling accent :

yewe'idas when he returned naga'-ida£ when he said baxa'mda£ when he came Tiele'lda* when he sang xebe'nda* when he did it

On the other hand it appears as an aspirate tenuis when preceded by the rising accent :

laalet'ae as it became s'as'init*a£ when he stood

The rule first given, when interpreted in the light of a reconstructed historical development, would then mean that a rising accent preserved an immediately following aspirated surd (including always those cases in which I, m, or n intervened), and caused the change of a media to an aspirated surd; while a falling accent preserved a simi- larly situated media or aspirated surd in its original form. That the change in the phonetic circumstances defined of an original media to an aspirated surd is indeed conditioned by a preceding rising accent, is further indicated by such rather uncommon forms as hadedll-t'a EVERYWHERES. Here the -fa is evidently the same as the -da of liawillfida IN HIS HOUSE, and the difference in manner of articulation is doubtless in direct relation to the difference of accent.

A modification of the general phonetic rule as first given remains to be mentioned. After Z, m, or n an original aspirated tenuis retains its aspiration even if the accent falls on the preceding syllable but one; also after a short vowel preceded by Z, m, or n, provided the accented vowel is short. Examples are :

alweflc!a\t*ee I shall shine; alwe'1c!alp*igam we shall shine; alw&'-

Tclalk."wa to shine

fcVp'alt'e6 1 shall be absent; Jc*e'p*alk'wa to be absent wuln'hamt*ee I have menstrual courses for the first time xala,'xamt*ee I urinate I'm^amk'am he was sent off (I is short, though close in quality:

contrast domJiigam, he was killed) Imi'Jiamk'wit* he sent himself § 23

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 51

t8'!umu't8'lamt*a*n I always boil it (cf. s'omoda'£n I boil it) s'a's.ant'6e I shall stand; s'a,'s'anp*igam we shall stand; s'aVan-

k'wa to stand

sene'sant*e£ I whoop; se'nsant^e* I shall whoop de£iwifigank*wide£ I spread (it) out for myself dasg&'\it*aa (grain) will lie scattered about

With -t*aa and -t'e£ above contrast the morphologically identical ele- ments -daa and -de£ of the following examples, in which the same accentual condition prevails but with a consonant other than Z, m, or n preceding the affected dental :

t*ge'its'!id(ia (round object) will lie (there) s'u'Vdi&aa (string) will lie curled up

dak*t"eJc!efxade£ I smoke (but future -xafVee because of immedi- ately preceding accent)

§24. INORGANIC h

Whenever two morphologically distinct vowels come together within the word (verbal prefixes and postposed particles, such as deictic -a\ are not considered as integral parts of the word), the first (accented) vowel is separated from the second by an " inorganic" -h-\

lt!ana'Jii£n I hold it (aorist stem tlana- + instrumental -i-), but future IHani'n (stem t!an-)

dak'-da-7ialafhin I shall answer him (future stem Jiala- + instru- mental -i-), but aorist daV-da-liaali'£n (stem M°Z-)

This inorganic h is found also immediately following an m, n, or Z preceded by the accent:

wayanha£n I put him to sleep (cf. same form with change of

accent wa-yaana'£n)

daa£agdn?ii£n I used to hear about it (cf . -agani'£n I hear it) liwllhaut* ''e£ I kept looking (cf . liwilofuCe£ I looked) xar-i£ glHt* gaflhi he broke it in two (cf. with identical -i- suffix

xaPsalt* gwi'lt* gwili he broke [somebody's arm] by stepping) I'mhamYam he was sent off (also in aorist stem imiham-) wadomhik* he killed him with it (stem doum- + -i-)

It will be observed that the insertion of the h is practically the same phonetic phenomenon as the occurrence of an aspirated tenuis instead of a media after an accented vowel. The vowel, nasal, or liquid may appropriately enough be considered as having become aspirated under the influence of the accent, just as in the case of the mediae.

§ 24

52 BUKEAU OF AMEKICAN ETHNOLOGY [BULL. 40

MORPHOLOGY (§§ 25-114) § 25. Introductory

Takelma conforms to the supposedly typical morphology of Amer- ican languages in that it is thoroughly incorporating, both as regards the pronominal, and, though somewhat less evidently, the nominal object. If by " polysynthetic " is merely meant the introduction into the verb-complex of ideas generally expressed by independent ele- ments (adverbs or the like), then Takelma is also polysynthetic, yet only moderately so as compared with such extreme examples of the type as Eskimo or Kwakiutl. The degree of intimacy with which the pronominal objective elements on the one hand, and the nominal objective and polysynthetic (instrumental and local) elements on the other, are combined with the internal verb-structure is decidedly different. The former combine as suffixes to form an indissoluble part, as it were, of the verb-form, the subjective elements of the transitive verb, though in themselves absolutely without independent existence, being secondarily attached to the stem already provided with its pronominal object. The latter vary in degree of independ- ence ; they are strung along as prefixes to the verb, but form no integral part of its structure, and may, as far as grammatical coherence is concerned, fall away entirely.

The polysynthetic character of the Takelma verb (and by discuss- ing the verb we touch, as so frequently in America, upon the most vital element of the sentence) seems, then, a comparatively accidental, superimposed feature. To use the- term " polysynthetic " as a catch- word for the peculiar character of Takelma, as of many another American language, hardly hits the core of the matter. On the other hand, the term " incorporation," though generally of more value as a classificatory label than "polysynthesis," conveys information rather as to the treatment of a special, if important, set of concepts, than as to the general character of the process of form-building.

If we study the manner in which the stem unites in Takelma with derivative and grammatical elements to form the word, and the vocalic and consonantic changes that the stem itself undergoes for gram- matical purposes, we shall hardly be able to find a tangible difference

§ 25

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 53

in general method; however much the details may vary, between Takelma and languages that have been dignified by the name "inflec- tional." It is generally said, in defining inflection, that languages of the inflectional as contrasted with those of the agglutinative type make use of words of indivisible psychic value, in which the stem and the various grammatical elements have entirely lost their single indi- vidualities, but have "chemically" (!) coalesced into a single form- unit; in other words, the word is not a mere mosaic of phonetic materials, of which each is the necessary symbol of some special concept (stem) or logical category (grammatical element) .

In support of the actual existence of this admired lack of a one- to-one correspondence between a grammatical category and its pho- netic expression is often quoted the multiplicity of elements that serve to symbolize the same concept; e. g., Lat. -4, -ae, -a, -es, -us, all indicate that the idea of a plurality of subjects is to be associated with the concrete idea given by the main body of the words to which they are attached. Furthermore, variability of the stem or base itself is frequently adduced as a proof of its lack of even a relative degree of individuality apart from the forms from which by analysis it has been abstracted; e. g., German bind-, band-, bund-, band-, bund-. These two characteristics are very far indeed from constituting anything like a definition of inflection, but they are often referred to as peculiar to it, and hence may well serve us as approximate tests.

As regards the first test, we find that just such a multiplicity of phonetic symbols for the same, or approximately the same, concept, is characteristic of Takelma. The idea of possession of an object by a person or thing other than the speaker or person addressed is expressed by -xa, -a, -da (-fa), -t\ or -,all of which are best rendered by HIS, HER, ITS, THEIR (the ideas of gender and number do not here enter as requiring grammatical expression). Similarly, the idea of the person speaking as subject of the action or state predicated by the main body of the verb is expressed by the various elements -*V (-de£), -t*ee (-dee), -sn, -n, -¥a£ (-ga£), all of which are best ren- dered in English by "I." -t*e£ is confined to the aorist of intransi- tive verbs; -t'ee is future intransitive; -sn is aorist transitive; -n is future transitive; and -Vas is used in all inferential forms, whether transitive or intransitive.

54 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

As for the second test, it soon appears that the Takelma stem may undergo even more far-reaching changes than we are accustomed to in German or Greek. As examples may serve :

doum-, du^m-j tlomom- (t!omou-), t!umu'ii- kill naag-, nee-, naga^, nege- say to

The first form in each of these sets is the verb-stem, properly speak- ing, and is used in the formation of all but the aorist forms. The second is employed in non-aorist forms when the incorporated object of the verb is a first person singular, and in several derivative forma- tions. The third is characteristic of the aorist. The fourth is used in the aorist under the same conditions as determine the use of the second form of the stem in other groups of forms. It needs but a moment's thought to bring home the general psychic identity of such stem-variability and the "ablaut" of many German verbs, or the Latin stem-variation in present and perfect :

frang- :freg- break da- : ded- give

If the typical verb (and, for that matter, noun) form of Takelma is thus found to be a firm phonetic and psychic unit, and to be charac- terized by some of the supposed earmarks of inflection, what is left but to frankly call the language "inflectional" ? " Polysynthetic" and " incorporative " are not in the slightest degree terms that exclude such a designation, for they have reference rather to the detailed treatment of certain groups of concepts than to morphologic method. Everything depends on the point of view. If chief stress for purposes of classification is laid on the relative importance and fulness of the verb, Takelma is polysynthetic ; if the criterion of classification be taken to be whether the verb takes the pronominal object within its structure or not, it is incorporating; if, finally, stress be laid on the general method of building up the word from smaller elements, it is inflective. Not that Takelma is in the least thereby relegated to a peculiar or in any way exceptional position. A more objective, un- hampered study of languages spoken in various parts of the world will undoubtedly reveal a far wider prevalence than has been gener- ally admitted of the inflectional type. The error, however, must not be made of taking such comparatively trivial characteristics as sex gender, or the presence of cases, as criteria of inflection. Inflection has reference to method, not to subject-matter. 5 25

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 55

Grammatical Processes (§§ 26-32) § 26. General Remarks

There are four processes employed in Takelma for purposes of grammatical modification and word-formation: affixation (pre-, in-, and suffixation), reduplication, vocalic change (ablaut), and conso- nant change (consonant ablaut). Pitch-accent is of grammatical importance, but is most probably a product of purely phonetic causes. Of the processes mentioned, suffixation is by far the most important; while the presence of infixation will have to be allowed or denied according to the definition given of it.

§ 27. Prefiocation

Prefixation is either of the loose polysynthetic type already referred to, or of the more firmly knit inflective type. Loose prefixation is extremely common, nominal objects, instruments, and local ideas of one kind or another finding admittance into the word-complex, as we have seen, in this manner. Examples of such loose prefixation are :

gwen-£a'l-yowo£ he looked back (gwen- in back; al- is difficult to define, but can perhaps be best described as indicative of action away from one's self, here with clear implication of sight directed outward; yowof£ he was, can be used as independent word)

s'in-l-lats!agi'£n I touched his nose (s'in- nose; I- with hand; lats!agi'sn I touched him, as independent word)

gwenfge^m black necked (gwen- nape, neck; t'ge^m black)

The first example shows best the general character of loose prefixa- tion. The prefixed elements gwen-, al-, s'in-, and I- have no separate existence as such, yet in themselves directly convey, except perhaps al-, a larger, more definitely apperceived, share of meaning than falls to the lot of most purely grammatical elements. In dealing with such elements as these, we are indeed on the borderland between independent word and affix. The contrast between them and gram- matical suffixes comes out strongest in the fact that they may be entirely omitted without destroying the reality of the rest of the word, while the attempt to extract any of the other elements leaves an unmeaning remainder. At the same time, the first example well illustrates the point that they are not so loosely attached but that they may entirely alter the concrete meaning of the word. Pre- fixation of the inflective type is very rare. There is only one

§§ 26-27

56 BUREAU OP AMERICAN ETHNOLOGY r BULL. 40

such prefix that occurs with considerable frequency, wi-, first person singular possessive of nouns of relationship :

my father your father

§ 28. Suffixation

Suffixation is the normal method employed in building up actual forms of nouns and verbs from stems. The suffixes in themselves have for the most part very little individuality, some of them being hardly evident at all except to the minute linguistic analyst. The notions they convey are partly derivational of one kind or other. In the verb they express such ideas as those of position, reciprocal action, causation, frequentative action, reflexive action, spontaneous activity, action directed to some one, action done in behalf of some one. From the verb-stem such adjectival and nominal derivations as participles, infinitives, or abstract nouns of action, and nouns of agent are formed by suffixation. In the noun itself various suffixed elements appear whose concrete meaning is practically nil. Other suffixes are formal in the narrower sense of the word. They express pronominal elements for subject and object in the verb, for the pos- sessor in the noun, modal elements in the verb. Thus a word like tlomoxiniV WE KILL ONE ANOTHER contains, besides the aorist stem Homo- (formed from doum-), the suffixed elements -x- (expressing general idea of relation between subject and object), -in- umlauted from -an- (element denoting reciprocal action [ -x-in- = EACH OTHER, ONE ANOTHER]), and -i¥ (first personal plural subject intransitive aorist). As an example of suffixation in the noun may be given tlibagwa^n-VY MY PANCREAS. This form contains, besides the stem Hiba-, the suffixed elements -gw- (of no ascertainable concrete signifi- cance, but employed to form several body-part nouns; e. g., tlibaW" PANCREAS 47.17), -an- (apparently meaningless in itself and appear- ing suffixed to many nouns when they are provided with possessive endings) , and -tW (first personal singular possessive) .

§ 29. Infijcation

Infixation, or what superficially appears to be such, is found only in the formation of certain aorist stems and frequentatives. Thus the aorist stem mats lag- (from masg- PUT) shows an intrusive or

§§ 28-29

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 57

infixed -a- between the s (strengthened to ts!) and g of the stem. Similarly the aorist stem wits' !im- (from wism- MOVE) shows an infixed i. Infixation in frequentative forms is illustrated by: yonoina'£n I always sing (aorist stem yonon-) tslayalY he used to shoot them (cf. tslaya^Y he shot them) On examination it is found that the infixed element is invariably a repetition of part of the phonetic material given by the stem. Thus the infixed -a- and -i- of mats lag- and wits' lim- are repetitions of the -a- and -i- of the stems masg- and wism-; the infixed -i- of yonoin- and tslayaig- are similarly repetitions of the y- of yonon and -y- of tslayag-. It seems advisable, therefore, to consider all cases of infixation rather as stem-amplifications related to reduplica- tion. An infixed element may itself be augmented by a second infixation. Thus we have:

Verb stem Aorist stem Frequentative

Tiemg- take out Jiemeg- 7iemeemg-

ts!a-im- hide tslayam- tslaya-im-

masg- put mats lag- mats!aasg-

yawl- talk yawa-i- yawa-iy-

baxm- come baxam- l>axaaxm-

§ 30. Reduplication

Keduplication is used in Takelma as a grammatical process with surprising frequency, probably as frequently as in the Salish languages. The most interesting point in connection with it is probably the fact that the reduplicating increment follows the base, never, as in most languages (Salish, Kwakiutl, Indo-Germanic) , precedes it. It is, like the infixation spoken of above, employed partly in the formation of the aorist, partly to express frequentative or usitative action. Some nouns show reduplicated stems, though, as a process, redupli- cation is not nearly as important in the noun as in the verb. Some verbs, including a number that do not seem to imply a necessary repetitive action, are apparently never found in unreduplicated form. Four main types of reduplication, with various subtypes, occur :

1. A partial reduplication, consisting of the repetition of the vowel and final consonant of the stem :

aorist helel- (from heel- sing)

aorist tlomom- (from doum- kill)

The reduplicated vowel is lengthened in certain forms, e. g., lieleel-, t!omoum-.

§ 30

58 BtJEEAIT OF AMEKICAN ETHNOLOGY [BULL. 40

1 a. A subtype of 1 is illustrated by such forms as exhibit an unreduplicated consonant after the reduplicated portion of the word, the second vowel in such cases being generally long

aorist ts'IumvPmt^Or- (from s'vPmfa- boil)

usitative aorist tlulvFlg- (from verb stem tfvPlg-, aorist t!ulug-

follow trail) usitative aorist ginllng- (from verb stem ging-, aorist ginig- go to ;

ging-j ginig- itself is probably reduplicated from gin-)

2. A complete reduplication, consisting of the repetition of the entire base with a change of the stem-vowel to a:

aorist tleutla'w- (from tleu- play shinny) aorist bofbad- (from loud- pull out one's hair) aorist laa- sal- xo(x)xag come to a stand (pi.) ; aorist sal-xog-V- stand (pi.)

3. A complete reduplication, as in 2, with the addition of a con- necting vowel repeated from the vowel of the stem:

aorist yuluyal- (cf . verb stem yulyal- rub)

aorist frequentative Jiogohag- keep running (from Jioug- run)

aorist frequentative s'wilis'wal- tear to pieces; verb stem s'wil-

s'wal- (from aorist s'wtfls-wal- tear; verb stem s'wtfl-) If the stem ends in a fortis consonant, the reduplicating syllable regularly shows the corresponding media (or aspirated tenuis) :

sgotlosgad- cut to pieces (from verb stem sgout!-, aorist sgoud- cut) 3 a. A subgroup of 3 is formed by some verbs that leave out the -a- of the reduplicating syllable:

gwidiVwd- throw (base gwid-}

4. An irregular reduplication, consisting of a repetition of the vowel of the stem followed by -(£)a- + the last and first (or third) consonants of the stem in that order :

frequentative aorist Homoamd-, as though instead of *t!omo- t!am-; cf. non-aorist doumdam- (from aorist Homom- kill)

frequentative aorist Jc!emesamg- (from Jc!eme-n- make; verb stem Jc!em-n-)

frequentative aorist p!uwueaug-, as though instead of *p!uwup!aug- (from aorist pluwuk!- name)

It will be noticed that verbs of this type of reduplication all begin with fortis consonants. The glottal catch is best considered a partial representative of the initial fortis; in cases like Jc!emeeamg- an original § 30

BOAS] HANDBOOK OP INDIAN LANGUAGES TAKELMA 59

-k!am (i. e., -£gam) may be conceived of as undergoing partial meta- thesis to -samg.

Other rarer reduplications or stem-amplifications occur, and will be treated in speaking of aorist formations and frequentatives.

§ 31. Vowel-Ablaut

Vowel-ablaut consists of the palatalization of non-palatal stem- vowels in certain forms. Only o and a (with corresponding long vowels and diphthongs) are affected; they become respectively u (u) and e. In sharp contradistinction to the i- umlaut of an original a to i, this ablaut affects only the radical portion of the word, and thus serves as a further criterion to identify the stem. Thus we have weega'si HE BROUGHT IT TO ME (from stem waag-, as shown also by waag-iwi/en i BROUGHT IT TO HIM), but wege'sinY HE WILL BRING IT TO ME (from stem waga-, as shown also by waga- wi'n I'LL BRING IT TO HIM), both i- umlaut and stem-ablaut serving in these cases to help analyze out the stems. Vowel-ablaut occurs in the following cases :

1. Whenever the object of the transitive verb or subject of the passive is the first person singular :

mele'xi he told it to me 172.17, but malafxbi£n I told it to you

(162.6)

nege's'i he said to me 186.22, but naga'sam he said to us (178.12) dUmxinas I shall be slain (192.11), but domxbina* you will be slain

(178.15) gel-luliuigwa' si he avenges me, but -lohoigwa'^n I avenge him (148. 3)

Not infrequently vowel-ablaut in such cases is directly responsible for the existence of homonyms, as in yeweyagwa' 'si HE TALKS ABOUT ME (from yaway-talk) , and yeweyagwa' si HE RETURNS WITH ME (from yew ei-return) .

2. With the passive participial endings -aF% -iYw:

waseegWw wherewith it is shot (from saag- shoot) me'xaY™ having father (from ma'xa his father) wa£-i^duxiYwdeY my gathered ones (=1 have been gathering

them) (from doux- gather) dals^wa-p'H't!iJc'w mixed with (from p'ot!- mix) 178.5

3. In some verbs that have the peculiar intransitive-forming suffix -x-, by no means in all:

geyewa'lxde* I eat (136.15) (cf. gayawa'en I eat it 30.11) le*lcfnx he carries 178.6 (stem laab~)

§ 31

60 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

didaat*be'e£lc't*bag-ams (= -amtx) they had their hair tied on sides of head (from base t^aag-) 142.17; cf. -t*ba'agamda£n I tie his hair (27.1)

No satisfactory reason can be given why most verbs in -x- do not show this stem-palatalization. It is quite possible that its occurrence is confined to a restricted number of such verbs; at any rate, there is some limitation in its employment, which the material at hand has not been found extensive enough to define.

4. In nouns ending in -x-ap* (-s-ap* = -t-x-ap(), probably derived from such verbs in -x- as were referred to under 3 :

xdale'esap' belt (cf. xaala'ada£n I put it about my waist) halu'u£x6p' (= -xwap") shirt (cf. Jialo'u£V she put on [her dress])

5. In verbs provided with the suffix -xa-, which serves to relieve transitive verbs of the necessity of expressing the object:

lu'£xwagwadinin ( = luk!-xa-) I'll trap for him (stem lok!w-) llu'pxagwanV she shall pound with (stone pestle) (cf. lobcfp* she

pounds them) Jc!edeixade£ I was out picking (cf. lc!add£n I pick them, Tcladai he

picks them) ts!eye'mxade£ I hide things (cf. ts!ayamaf£n I hide it)

6. In reflexive verbs ending in -gwi- or -Viva- (-gwa-~) :

Jcletfgwtfp* pick them for yourself! (stem lfc!aad-)

alts leyelc' wit' he washed himself with it (cf. alts! ay ap* he washed

his own face)

llets!ek'wide£ I touch myself (cf. llats!agi'£n I touch him) k!edelk'wa£n I pick them for myself (aorist stem Jc/adai-) alnu'u¥wa he painted his own face (stem nougw-)

Yet many, perhaps most, reflexive verbs fail to show the palatal ablaut :

plagank'wif he bathed himself

t'gwdaxa'nt'gwidee I shall tattoo myself (but lu'ugwanCgwide£ I

trap deer for myself) xda-sgd/ut'gwide£ I cut myself igaxagafxgwa£n I scratch myself

We have here the same difficulty as in 3. Evidently some factor or factors enter into the use of the ablaut that it has not been founp possible to determine.

7. Other cases undoubtedly occur, but there are not enough of them in the material gathered to allow of the setting up of further groups. All that can be done with those cases that do not fall

S 31

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 61

within the first six groups is to list them as miscellaneous cases. Such are:

gwel-lelsdee I shall be lame (cf. gwel-la'is Idemnafn I shall make

him lame

leepsV wing (if derived, as seems probable, from stem laab- carry)

t!emeycL'nwiaus people go along to see her married 178.1 (cf.

t!amayana'£n I take her somewheres to get her married [148.5])

Palatal ablaut, it should be noted, does not affect the -a- of the

second member of reduplicated verbs :

t*gaalt*gcfl it bounced from her 140.8

t*geeltg*a'lsi it bounced from me

The connecting vowel, however, of verbs reduplicated according to the third type always follows the stem-vowel:

daVda-tiele'lialxade* I am accustomed to answer (stem ~7iaal-) It is difficult to find a very tangible psychic connection between the various cases that require the use of the palatal ablaut, nor is there the slightest indication that a phonetic cause lies at the bottom of the phenomenon. If we disregard the first group of cases, we shall find that they have this in common, they are all or nearly all intransi- tives derived from transitives by means of certain voice-forming ele- ments (-X-, -xa-, -gwi-, -Ywa-), or else nominal passives or derivatives of such intransitives (-ak*w, -x-ap') ; -k*wa-, it is true, takes transi- tive pronominal forms; but it is logically intransitive in character in that it indicates action in reference to something belonging to the subject. The only trait that can be found in common to the first group and the remaining is that the action may be looked upon as self-centered; just as, e. g., a form in -xa- denotes that the (logically) transitive action is not conceived of as directed toward some definite outside object, but is held within the sphere of the person of central interest (the subject), so, also, in a form with incorporated first per- son singular object, the action may be readily conceived of as taking place within the sphere of the person of central interest from the point of view of the speaker. No difficulty will be found in making this interpretation fit the other cases, though it is not conversely true that all forms implying self-centered action undergo palatalization. The explanation offered may be considered too vague to be con- vincing; but no better can be offered. In any event, the palatal ablaut will be explained as the symbolic expression of some general mental attitude rather than of a clear-cut grammatical concept.

§ 31

62 BUKEAU OF AMERICAN ETHNOLOGY [BULL. 40

Besides these regular interchanges of non-palatal and palatalized vowels, there are a number of cases of words showing differing vowels, but whose genetic relationship seems evident. These vocalic varia- tions have not been brought into the form of a rule; the number of examples is small and the process apparently touches rather the lexical material than the morphology. Variations of this character between a and e are:

g&l&-b-a'£n I twist it; pfV-wa-gele-g-i^n I drill for fire with it (88.12), dii£al-golegal-a'mda£n I tie his hair up into top-knot (172.2)

dda-d&l&-g-a' 'mdaen I pierce his ear (22.1); daa-dele-b-i'£n I stick it through his ear

Zax' excrement 122.2; le'-k'w-an-t*¥ my anus

Variations between o (u) and u are :

s'omoda'£n I boil it (58.10); ts'!umumt*a£n I boil it (170.17) xumcf food 54.4; x\rni\\.'Yde£ I am sated (130.18)

An a — u variation is seen in :

hau-Ji&n&'£s it stopped (raining) 196.8; p!ai-7iunu/u£s he shrank 33.16

Variations between a and i are:

y&w&lt*e£ I talk (132.3) ; yiwiya'ut*e£ I keep talking, I converse (194.5); yiwin talking, (power of) speech 138.4

lab&'n I shall carry it (124.5); libin news (what is carried about from mouth to mouth[?]) 194.9

Of o (u) — e variations there have been found:

Io7iolt*e£ I die 184.18; leJieit'e£ I drift dead ashore (75.5) xaa-kuk!\\'lidk'na£n I breathe; xaa-hvgQrliciYna£n I breathe (79.2) £/os-6'u little 180.20; aZ-£/eeHY little-eyed 94.3

An e — i variation is found in the probably related:

p!eyent'e£ I lie 71.5 (future p!e't*ee [146.9]) ; gwen-p!iyi'nk*waen

I lie on pillow (future gwen-p HYwari) t*geeya^lx it rolls; a'l-t*gl[ycflx tears rolled from (his) eyes 138.25

§32. Consonant- Ablaut

Consonant-ablaut, ordinarily a rare method of word-formation, plays a rather important part in the tense-formation (aorist and non- aorist) of many verbs. The variation is in every case one between fortis and non-fortis; i. e., between p!, t!, lc!, is!, and Z>, d, g, s, respec- tively. Three main types of grammatical consonant change are to be recognized:

§ 32

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 63

1. An initial fortis in the aorist as opposed to an initial media in non-aorist forms:

aorist Iclolol- (stem goul- dig) aorist tlebe- (stem deeb- arise) aorist tlayag- (stem ddag- find)

2. A medial fortis followed by a vowel in the aorist as opposed to a medial tenuis followed by a consonant in non-aorist forms :

aorist loplod- (stem lop'd- rain, snow, or hail) aorist latslag- (stem lasg- touch)

3. A medial media in the aorist as opposed to a medial fortis in the remaining forms :

aorist nuud- (stem nuut!- drown) aorist wtfg- (stem wlk!- spread)

Needless to say, this consonant-ablaut has absolutely nothing to do with the various mechanical consonant-changes dealt with in the phonology.

A few examples of consonant-ablaut not connected with regular grammatical changes have also been found:

s'omod- boil; ts'Iumu^mt'a- boil

Jiau-gwen-yut \uyad-i- swallow down greedily (like duck or hog) 126.10; hau-gwen-yunu£yau-i- dit.

The second example illustrates an interchange not of fortis and non- fortis (for n£ is related to n as is t! to d), but of non-nasal stop and nasal.

I. The Verb (§§ 33-83)

§33. Introductory

The verb is by far the most important part of the Takelma sen- tence, and as such it will be treated before the independent pronoun, noun, or adjective. A general idea of the make-up of the typical verb-form will have been gained from the general remarks on mor- phology; nevertheless the following formula will be found useful by way of restatement:

Loosely attached prefixes + verb-stem (or aorist stem derived from verb-stem) + derivational suffixes + formal elements (chiefly pronominal) + syntactic element.

This skeleton will at the same time serve to suggest an order of treatment of the various factors entering into verb morphology.

§ 33

64 BUKEAU OF AMERICAN ETHNOLOGY [BULL. 40

Before taking up the purely formal or relational elements, it seems best to get an idea of the main body or core of the word to which these relational elements are attached. The prefixes, though not entering into the vital grammatical structure of the verb, are impor- tant for the part they play in giving the whole verb-form its exact material content. They may, therefore, with advantage be taken up first.

1. Verbal Prefixes (§ § 34-38)

§ 34. GENERAL REMARKS

Verbal prefixes may be classified into four groups when regard is mainly had to their function as determined largely by position with respect to other prefixes: incorporated objects, adverbial (including local) elements, incorporated instrumentals, and connective and modal particles. These various prefixes are simply strung along as particles in the same order in which they have been listed. Inasmuch as the exact function of a prefix is to a considerable extent determined by its position, it follows that the same prefix, phonetically speaking, may appear with slightly variant meanings according as it is to be interpreted as an object, local element, or instrument. Thus the prefix I- always has reference to the hand or to both hands; but the exact nature of the reference depends partly on the form of the verb and partly on the position of the prefix itself, so that I- may be trans- lated, according to the circumstances of the case, as HAND(S):

l-plV-nd'uYwtfn I warm my hands WITH THE HAND:

l-soudini'£n I hunt for it with the hand ( = I am feeling around for it)

IN THE HAND:

p*im-l-7idugwagwa/£n I run with salmon in my hand

In the first of these three examples the I- as object precedes the incorporated instrumental pfl* FIRE, so that the form means literally i WARM MY HANDS WITH FIRE. In the third form the I as local ele- ment follows the incorporated object p'im SALMON. Such a triplicate use is found only in the case of incorporated nouns, particularly such as refer to parts of the body. These incorporated elements are to be kept distinct from certain other elements that are used in an

§ 34

BOAS]

HANDBOOK OF INDIAN LANGUAGES TAKELMA

65

adverbial sense only, and regularly occupy the second position. The line between these two sets of prefixes is, however, difficult to draw when it comes to considering the place to be assigned to some of the prefixed elements. It is doubtful whether we are fully justified in making absolutely strict distinctions between the various uses of the body-part prefixes ; at any rate, it is certainly preferable, from a native point of view, to translate the three examples of I- incorpora- tion given above as :

I-hand-fire-warm(-as-regards-myself)

I-hand-hunt-for-it

I-salmon-hand-run-with

leaving in each case the exact delimitation in meaning of the element HAND to be gathered from the general nature of the form. The fol- lowing examples will render the matter of position and function of the various prefixes somewhat clearer:

Object.	Locative adverb.	Instrument.	Modal.	Verb proper.
bem- sticks	wa- together	£l- hand		t!oxo'xi£n I gather (them) (=1 gather sticks together)
	he#- away	wa- with it		waagiwi'n she is bought (= she is brought with it) 176.17
gwan- road	ha- in		yaxa- continuously	t!uliiulga'sn I follow (it) (=1 keep following the trail)
dan- rocks	bda~ up	el- hand		sget!e'sgidisn I lifted (them) (=1 lifted up the rocks)
	han- across	waya- knife		swilswa'lhr}\Q tore him (=he tore himopen with a knife)73.3
	dak'- above	da- mouth	walaftsina- truly	haali'nda& I answering him (= I did answer him)
j	10- between, in two	I- hand	ml' iswa- probably	sgl'ibien I cut him (=I'll prob- ably cut him through) 31.13

If two adverbial (local) elements are used, the body-part prefix follows that which is primarily adverbial in character ; thus :

ba-ide'£didi'nik!at* did you stretch it out? ( = Zxz-i- out + de-lip, in front + di interrogative particle + di'niklat' you stretched it)

In general it may be said that instances of a body-part prefix pre- ceding a primarily adverbial element (like 6o-i-, baa-} hee£-, and others) are rare or entirely lacking.

From what has been said it might seem that the connective and modal elements (like yaxa, mi'i£wa, and di) are more closely associated with the verb form than are the other elements, yet this is only apparently the case. Properly speaking all these modal elements are post-positives that normally attach themselves to the first word of 3045°— Bull. 40, pt 2—12 5 J 34

66 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

the sentence, no matter what part the word plays in the sentence. Thus in a form like me'£-di-ginigaY DID YOU COME ? ( = me£- HITHER + di- interrogative particle + ginigaV YOU WENT TO) , the modal (inter- rogative) element di regularly stands nearest the verb ; but as soon as another word is introduced before the verb, the interrogative particle shoves back a step, and we have a form of sentence like, e. g., hoida'£s di me'£giniga,y DID YOU COME AS SINGER, i. e., TO SING? From this it becomes f airly evident that the di in the first example is not prop- erly a verbal prefix at all, but merely a post-positive particle depend- ing upon the preceding me'5, in the same way that, in the second example, it depends upon the noun 7ioida'£s SINGER. This inference is clinched by a form like giniga't'tdi DID YOU GO (SOMEWHERE)? for here the di is evidently an enclitic element, not a prefix.

In sharp contradistinction to such movability, the body-part and adverbial prefixes occupy rigidly fixed positions before the verb; they therefore belong to a class quite distinct from the modal parti- cles. These latter are verbal prefixes only in so far as then1 post- positive tendency may force them to become embedded in the verb-complex, in which case they seem to cut loose the incorporated object, adverbial prefix, and instrumental element from the verb. Diagrammatically the last form tabulated may be represented by xa-l- [ml'i£wa] -sglribi£n. We may then dismiss the modal elements from our consideration of verbal prefixes, to return to them when speaking of connective and adverbial particles.

§ 35. INCORPORATED NOUNS

It may seem strange at first sight to interpret in the examples given above such elements as bem STICKS, gwan ROAD, and da^n ROCKS as incorporated objects, when they occur as absolute nouns in that form as well, though a faint suggestion of incorporation is given by gwan-7ia-yaxa-t!ulu'iilga'£n i KEEP FOLLOWING THE TRAIL, in that the modal post-positive yaxa follows not gwan, but rather Tia-, as though the direct object were not quite felt to be an element inde- pendent of the verb. Without laying particular stress on this latter point, there are, it would seem, good reasons for considering the nouns referred to as incorporated, though in any event the incor- poration must be called a loose one, and not at all comparable with the Iroquois usage.

§ 35

BOAS] • HANDBOOK OF INDIAN LANGUAGES TAKELMA 67

1. In the first place it is evident from such examples as l-pft*- no'uYwa£n i WARM MY HANDS and Jian-waya-swilswa'lhi HE TOKE HIM OPEN WITH A KNIFE, that nouns (in these cases p!l* FIRE and way a KNIFE) occur as incorporated instrumental, for such elements as l- and Tian- can not possibly be isolated from the verb (Jian- does not occur as independent adverb, but only as prefix; I- is inconceivable as independent noun) ; furthermore, if, in the forms just quoted, p!l* and way a be looked upon as absolutely independent nouns, they lose all semblance of grammatical form, there being, indeed, nothing but a definite position in a verb-complex that could here suggest the notion of instrumentality. It is also possible to isolate waya, but that would involve considerable readjustment of the verbal structure. To be stamped as an instrumental, waya must in that case be fol- lowed by a postposition wa WITH, so that the sentence then reads, Jian-swilswa'lhi wa'ya wo" (the phrase wa'ya wa* may also precede).

If we wish to incorporate the instrumental idea into the verb, and yet keep the noun outside of the verb-structure, we may let the wa, which seems properly to denote WITH IT, occupy the place of the incor- porated waya, which, as an appositive of wa, then either precedes or follows the verb-form, wa'ya 7ian-wa-swilswa'lhi, or han-wa-swilswa'lhi waya* HE-ACROSS-WITH-IT-TORE-HIM (it, i. e.), THE-KNIFE. This con- struction is identical with the well-known appositional structure of Nahua or Chinook (e. g., I-IT-KILLED THE-DOG), except that the incor- porated element is here instrumental and not objective in character. The noun and its representative can not both be incorporated in the verb, such a form as Jiarwvaya-wa-swilswa'lhi, for instance, being quite impossible.

It becomes clear, therefore, that an incorporated instrumental noun like wa'ya is quite analogous to an instrumental body- part prefix like i- HAND, with the difference that wa'ya may be isolated in that form, while I- must, when isolated, be provided with a possessive pronominal element. The form han-l- swilswa'lhi i TORE HIM OPEN WITH MY HAND is strictly analogous to han-waya-swilswa'lhi; the sentence iuxdeW harwva-swilswa'lhi MY- HAND I-ACROSS-WITH-IT-TORE-HIM corresponds to wa'ya 7ian-wa-swil- swa'lhi; and, finally, Jian-swilswa'lhi iuxde'Jc* wa* I-ACROSS-TORE-HIM MY-HAND WITH (-IT) is parallel to Jian-swilswa'lhi wa'ya wa*. What- ever is true morphologically of i- must be true of wa'ya; the evident

§ 35

68 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

incorporation of I- involves the incorporation of wa'ya in the analogous form.

As the incorporation of the noun as an instrument seems a rather important trait of Takelma, a number of further examples may be given :

xaa-bee-nd'uYwaen I warm my back in (really = with) the sun

(foesun); cf. 188.20

lieeS-xi-le'me£Yi he destroyed them with water (xi water) Tiee£-p!li-lemef£Yi he destroyed them with fire (pH* fire) 98.12 xa-dan-t' 'gtflt' 'ga'lhi he broke it with a rock (dan rock) 24.4 gwen-waya-sgo'ut*i he cut their necks off with his knife (wayafa wcf

with his knife, apart from verb-structure) 144.5, 22 xaa-beem-lc!wdut'Jc!widi£n I broke it with a stick (beem stick) daa-heel-yebebi'£n I sing for him, literally, I engage (?) his ears

with song (7ieel song; al-yebeb-i- show to) daa-t*muugal-lewe'£liwi£n I shake my ears with twisted shells

(attached to them) (t'muugal twisted shell) 122.1 dV-Yal-p^ili'p^iltfn I squash them with my penis (FaZ penis) 73.14 de-ye't*-baxamagwanaW we came crying, literally, we came hav- ing (our) mouths with tears (yet* tears) yap!a-dauy(La-ts!aya'Yi he shot people with his shaman's spirit

(dauya'aVwda his shaman-spirit, apart from verb-structure);

cf. 164.14

All these, except the last, begin with elements (xda-, 7iee£-, gwen-, daa~, dl*-, de) that can not be isolated from the verb.

Instrumental, whether nouns or body-part prefixes, can occur only in transitive verbs. The forms noxwcf yana-wa-lobobif*n i POUND ACORNS WITH A PESTLE and noxwa^-l-loboxagwaf£n i POUND WITH A PESTLE, as compared with lobo'xade* i POUND, will serve to illustrate this. The first sentence reads, when literally translated, PESTLE (noxwa^) I-ACORNS (ycbno") -WITH-IT-POUND. The logical instrument (noxwcf) stands outside the verb-complex and is in apposition with its incorporated instrumental representative (wa-), yana? being the direct (incorporated) object. The form lobo'xade* i POUND is made intransitive by the element -xa- (hence the change in pronominal form from transitive -£n to intransitive -de£), and allows of no instrumental modification ; a form like l-lobofxade£ could hardly mean i POUND WITH THE HAND; at most it could signify i POUND IN THE HAND. If we wish, however, to express the logical instrument in some manner, and yet neglect to specify the object, we must get around the difficulty by making a secondary transitive of § 35

HANDBOOK OF INDIAN LANGUAGES TAKELMA 69

the intransitive in -xa-. This is done by the suffixed element -gw- HAVING, ATTENDED BY. The grammatical object of a transitive verb in -gw- is never the logical object of the action, but always dependent upon the comitative idea introduced by this suffix. Hence the sec- ond form is not provided with a true instrumental (WITH A PESTLE), but takes the logical instrument (noxwa^) as a direct object, while the I- is best rendered by IN THE HAND; to translate literally, the form really means i POUND HAVING A PESTLE IN THE HAND.

It sometimes happens that a verb form has two instrumenta-ls, one, generally I- WITH THE HAND, expressing indefinite or remote instrumentality, the second, a noun or demonstrative, expressing the actual instrument by means of which the action is accomplished. In such cases the second instrument is expressed outside of the verb- complex, but may be represented in the verb by the incorporated wa WITH IT following the first instrumental element (I-} . Examples of such double instrumentals are:

gwalt* baa-£l-wa-xd'ut'i wind he-up-hand- with-it-caused- them-to-

fall, i. e., he caused them to fall by means of a wind (that he

made go up) 168.2 go, £i-wa,-molo£ma'lhi that she-hand- with-it-stirs-it-up, i. e., she

stirs it up with that (incidentally, of course, she uses her hand

too) 170.16 dan (object) Iclama (instr.) p!ai-£i-wa-sga'ak*sgigi£n rocks tongs

down-hand-with-it-pick-up, i. e., I pick up the rocks with the

tongs (and put them) down

2. The noun as instrument has been shown to act in a manner entirely analogous to the instrumental body-part prefix. The latter can, without phonetic change, become the direct object of the verb by occupying the proper position :

s*in-i-lats!agi'£n I touched his nose with my hand (s'in- nose) but, theoretically at least,

l-s'in-lats!a,gi'£n I touched his hand with my nose If we bear in mind that such elements as s'in- and I- are really nothing but nouns in their stem form (with possessive pronoun: s'in-i-x-da HIS NOSE; I'-u-x-da, HIS HAND), the parallelism with such noun- objects as 6gm and gwdn (see examples on p. 65) becomes complete. The fact that they may occur independently, while s'in- and l- never do, is really irrelevant to the argument, as a body-part noun must necessarily be associated with some definite person. Entirely

§ 35

70 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

analogous to the nominal elements -l{-x- and -u-x- of s'inlxda and I'uxda is, e. g., the -am- of gwdal-a^m-t*k* MY ROAD. Just as they drop off when the body-part nouns are incorporated, whether as object or instrument, into the verb, so, also, the -am- of gwaal-am- ( = gwdan-an-) drops off when the noun is used without pronominal or prepositional modification. That the -am- has nothing per se to do with the pronominal affix, but is really a noun-forming element added to the stem, is proven by forms like ha-gwdala^m IN THE ROAD. Thus:

object &em, in ~bem-wa£-i-t!oxo'xi£n I gather sticks, is related to object s'in-, in s'in-i-lats!agi'£n I touch his nose, as instrument &em, in xda-beem-k!wdut*lc!widi£n I broke it with a

stick, to instrument s'in-, in s'in-t!ayagi'sn I find it with my nose ( = 1

smell it)

In view of the complete parallelism of noun and body-part element and the transparent incorporation of the noun as instrument, nothing remains but to look upon the simple noun without pronominal affixes, when placed immediately before the local and instrumental prefixes of the verb, as itself a loosely incorporated object. Exam- ples of noun-objects in such form and position are to be found in great number; in fact, the regularity with which the object is put before the verb, as contrasted with the freely movable subject, argues further for the close relation of the noun-object to the verb.

A few further examples of incorporated noun-objects are given by way of illustration:

Jieel-gel-gulugwa'sn I desire to sing (literally, I-song-breast-desire ;

7ieel song)

7ieel-yununa'sn I sing a song (106.7) wili-wa-l-t!a'nida£ you shall keep house (literally, you-house-

together-hand-will-hold ; will house) 28.13 abai£ xuma-Jc!emna'£s cook (literally, in-the-house food-maker;

xuma food) 54.3 wai-s'ugu's'uxgwa£n I am sleepy (literally, I-sleep-am-confused ?-

having; wai sleep)

plV-da-Uagdl he built a fire (p!l* fire) 96.17 p!li-lda-ydnk'w he picked up the fire (literally, he-fire-up-went-

having) 96.25

xi-£ugwa^nk* he will drink water (xi water) 162.17 s'ix-ligi^Y'u> he brought home venison (s'lx venison) 134.4 § 35

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 71

In none of these would the placing of the object after the verb- form be at all idiomatic; in some (as in heel-gel-gulugwa'sn and wai- s-ugu' s'uxgwasn) it would be quite inconceivable. The incorporation must be considered particularly strong in those cases in which the object is what might be called a root-noun identical in form with a verb-stem of corresponding significance:

wai1 sleep, to sleep

Jieel- song, to sing

seel- black paint, to paint

likewise where the object gives special color to the verb, deter- mining the concrete significance of the form, as in xuma-Jc!emna,'ss and wili-wa-l-t!a'nida£.

3. Besides being used as instrumentals and direct objects, a few incorporated nouns are found employed in set phrases, apparently as subjects. Such are:

baa-bee-7c!iylfiJc'da£ forenoon (literally, up-sun-going, or when-it- goes) (&a°- is never used as independent adverb, so that bee- sun must here be considered part of the verb-complex)

nou-bee-Jc!iyl'iJc*dae afternoon (literally, down-river [i. e., west]- sun-going)

mot'-woW as son-in-law he visits wife's parents ( = mot'- son-in- law + w&Y, probably identical with woY he arrived) 17.13, in which mot'- must be considered an integral part of the verb, because unprovided with pronominal affix (cf . mo't"aa his son- in-law), and, further, because the whole form may be accom- panied by a non-incorporated subject (e. g., bo'mxi mofwoW Otter visited his wife's parents, literally, something like : Otter son-in-law-arrived)

4. Several verb-forms seem to show an incorporated noun forming a local phrase with an immediately preceding local prefix; in such cases the whole phrase must be considered an incorporated unit, its lack of independence being evidenced either by the fact that it is itself preceded by a non-independent verbal prefix, or else differs in phonetic form from the corresponding independent local phrase. Examples are:

daa-ts'!elei-sgalawi'£n I looked at them out of the corners of my eyes (literally, I-alongside-eye-looked-at-them) 2 ; cf. dda-ts'!e- leide alongside my eyes

*wai- indeed could not be obtained as an independent noun, its existence as substantive being inferred from forms such as that cited above.

2 It may be, however, that this form is to be interpreted as I-ASIDE- ( WITH-THE-) EYE-LOOKED-AT-THEM, ts'lelei- being in that case an incorporated instrumental noun.

§ 35

72 BUKEATJ OF AMERICAN ETHNOLOGY [BULL. 40

7ia-fgaa-gwidiWv> he threw it into the open (literally, he-in-earth-

threw-it) ; cf . Jia-t'gau in the earth 'ba-i-daY-wili-t!dadi'£n I ran out of the house (ba-i- out, adverbial

prefix + dak"- on top of + will house) 24.13; cf. dak*-vrill on

top of the house Jia-yau-£ge'nets!<isn I put it about my waist (literally, I-in

[ under ?]-rib-put-it-about) ; cf. Jia-yawade inside my ribs

Such verbs with incorporated local phrases are naturally not to be con- fused with cases in which a local prefix is followed by an incorporated (instrumental) noun with which it is not, however, directly connected. Thus the 1m- of ha-tgaa-gwidWw is not directly comparable to the Jia- of a form like:

Jia-p!li-ts'!urluk!i£n I set it on fire (pli* with fire) 73.9 Here Jia-pft*- cannot be rendered IN THE FIRE.

Some verb-forms show an evidently incorporated noun that has so thoroughly amalgamated with the stem that it is difficult to make out its exact share in the building up of the material content of the verb. For example:

s'omlohoya'ldasn I doctor him as s'omlo7io'lxa£s

doubtless contains the incorporated noun s'om MOUNTAIN; but the implied allusion is not at all evident, except in so far as the protecting spirits of the s'omloho'lxtfs are largely mountain-spirits. The verb itself is probably a derivative of the verb-stem loho- DIE (aorist lohoi-} .

§36. BODY-PART PREFIXES

Having disposed of the modal prefixes, which on analysis turned out to be verbal prefixes only in appearance, and of incorporated nouns, which one would hardly be inclined to term prefixes in the narrower sense of the term, there remain for our consideration two important sets of genuine prefixes, body-part elements and adverbial, chiefly local, prefixes. The former will be taken up first. By " body- part prefix" is not meant any body-part noun in its incorporated form (many of these, such as ts'Ielei- EYE, tliba- PANCREAS, not differing morphologically from ordinary incorporated nouns), but only certain etymologically important monosyllabic elements that are used to indi- cate in a more general way what body-part is concerned in a particular action, and which may be regarded as in some degree verbal classifiers. With the exception of I- HAND and s'in- NOSE, classed with the rest

§ 36

BOAS]

HANDBOOK OF INDIAN LANGUAGES TAKELMA

73

because of their very extended use, they differ fundamentally from other body-part nouns in that they have, besides their literal, also a more formal, local value ; in this capacity they are regularly employed, also, as the first element of noun and pronoun local phrases, and, some of them, as the second element of local postpositions. In the fol- lowing list the second column gives the literal body-part significance ; the third, the generalized local meaning; the fourth, the correspond- ing independent noun (in a few cases, it will be observed, there is no such corresponding noun); and the fifth column, an example of a local phrase:

Prefix.	Body.	Local.	Noun.	Phrase.
dak'-	head	over, above	da'g-ax- dek' my head	dak'-wiU over the house
{da-, de-	mouth, lips		dex- dek'
de-		in front		del' gwa in front of himself
dda-	ear	alongside	dda- n- x- de*k'	daa-gela^m along the river
s-in-	nose		S'in-ii-x-de^k'
gwen-	neck, nape	in back,behind	[bo'k' dan-x- de'k']	gwen-t'gaiiou. east side of the
				land
i-	hand		l-u-x- de'k'
xaa-	back, waist	between,in two	xan-ha^m-t'k'	xaa~ gweldS between my legs
dH-	back	on top of		dli-ludZ over my hand
gelr	breast	facing	gSl-x-dek', [Mlg-an-x-deW}	geldS facing, in front of me
dV-	anus	in rear	[delg- a*n- t'k']	dis-t'gau on west side of the
				land
ha-	woman's pri-	in	hau-x-dek'	ha-xiya* in the water
	vate parts
gwel-	leg	under	gwel-x-dek'	gwel-xiya* under water
la-	belly		llaa- excrement	La-t'gdu Uplands ( = ? front
				of the country)
sal-	foot	down, below	sal-x-de*k'
al-	eye, face	to. at	[ts- ! eld- t'k' my eye]	al- s- ou ma'l to the mountain
			[li'ugw- ax- dek' my face
dli*al-	forehead ( =		dlWl-t'k'	dHsa'lda at his forehead
	above eye)
gwenha-u-	nape (=neck		gwenha-u-x-de^k'	gwenha-udZ at my nape
	under)

The last two are evidently compounded; the first of dl*- ABOVE and al- EYE, FACE, the second of gwen-NECK and probably adverbial prefix Jia-u- UNDER. The noun Jiau-x- WOMAN'S PRIVATE PARTS may possibly be connected with this prefix ha-u-, though, in view of the fact that Jia- appears as the incorporated form of the noun, it seems more probable that the resemblance in form and meaning is acci- dental. It is possible that other rarer body-part prefixes occur, but those listed are all that have been found.

In not a few cases, where the body-part prefix evidently has neither objective nor instrumental meaning, it may yet be difficult to see a clearly local idea involved. This is apt to be the case particularly

§ 36

74 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

with many intransitive verbs, in which the share of meaning con- tributed by the body-part prefix is apparent enough but where the logical (syntactic) relation of its content to that of the verb proper is hardly capable of precise definition. Thus, from yowo/s HE is are formed by means of body-part prefixes :

al-syowo/s he-eye-is, i. e., he looks 62.6

daa-£yowo/£ he-ear-is, i. e., he listens, pays attention 96.9

~baa-gel-£yowo'£ he-up-breast-is, i. e., he lies belly up 140.5

In these cases it is obviously impossible, yowo- being an intransitive verb not implying activity, to translate aZ-, daa-, and gel- as instru- mental (WITH THE EYE, EAR, BREAST) ; nor is there any clear idea of location expressed, though such translations as AT THE EYE, EAR, BREAST would perhaps not be too far fetched. In many verbs the body-part prefix has hardly any recognizable meaning, but seems necessary for idiomatic reasons. In a few cases prefixes seem to interchange without perceptible change of meaning, e. g., al- and dak' in:

aldemxigam we shall assemble (186.7) d&\zdemxiaust* people (indef.) will assemble (136.11)

Where two body-part prefixes occur in a verb form, they may either both retain their original concrete significance, the first prefix being generally construed as object, the second as instrument (e. g., s'al-£l-lats!agi'£n I-FOOT-HAND-TOUCH-HIM, i. e., i TOUCH HIS FOOT WITH MY HAND) ; or the first prefix may have its secondary local signifi- cance, while the second is instrumental in force (e. g., de-£l-wlfigi£n I-FRONT-HAND-SPREAD-IT, i. e., i SPREAD IT OUT); or both prefixes may have secondary local or indefinite significance (e. g., gwel-ge'l-

£yoWO£ HE-LEG-BREAST-IS, i. 6., HE FACES AWAY FROM HIM); rarely

do we find that two body-part prefixes are concrete in significance and absolutely coordinated at the same time (see footnote to 12 below).

To illustrate the various uses of the body-part prefixes it seems preferable to cite examples under each separate prefix rather than to group them under such morphologic headings as objective, instru- mental, and local, as by the former method the range of usage taken up by the various prefixes is more clearly demonstrated. The examples are in each case divided into two groups : (a) literal signifi- cation (objective, instrumental, or local) and (&) general adverbial (local) signification.

§ 36

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 75

1. dak'"

(a) HEAD, WITH HEAD, IN HEAD!

dak.*ts!ayap*de£ I washed my head (literally, I washed in

my head

dsik"t'la'agamt* he tied together (their head hair) 27.1 dsik^lats!agi/£n I touched top of his head d&k.'Jiagdlt'e£ I felt thrill in my head (as when sudden cold

tremor goes through one)

aldsik'saamscfm he bumped (with) his head against it 79.7 do^iYiwirY(iuYwcbsn I brandish it over my head

(6) ON TOP OF, ABOVE :

da]zt*gu/uba,£n I put rounded scooped-out object (like hat or

canoe) on top (of head) (61.9)

d&]zt'eJc!e'xade£ I smoke (literally, I raise [sc., tobacco- smoke] over [one's head]) (96.23) dsik'limlmxgwaf it (i. e., tree) falls on you (108.12) dsik'waagaf£n I finish it (literally, I bring it on top) (110.17) will dak?yaangwa/en I pass house (? literally, I go with house

above me) (150.8)

dak'daMali'£n I answer him (61.6; 180.18) dak?tlem8xik* we assembled together (43.9; 136.11) do^iJieneedafsn I wait for him

The last three or four examples can hardly be said to show a transparent use of daY-. Evidently the meaning of the prefix has become merged in the general verbal content, becoming unrecognizable as such; cf. UNDER in English UNDERSTAND, UNDERGO.

2. da-, de-

It seems possible that we have here two distinct prefixes to begin with, dor INSIDE OF MOUTH (cf. d&ts!ayap* HE WASHED HIS MOUTH) and de- LIPS (cf. deets!ayap* HE WASHED HIS LIPS and noun dee-x- LIPS), from the second of which developed the general local significance of IN FRONT; contrast also AadaY- gwa IN HIS OWN MOUTH with def gwa in front of himself. The strict delimitation of the two, however, is made difficult by the fact that da-} alone in this respect among non-radical verbal elements, undergoes palatal ablaut (thus becoming de-) whenever the stem shows a palatal vowel, whether primary or itself due to ablaut; observe also the stem-change from da- to de- in Jiada,'t*gwa 170.2 and Jiadede IN MY MOUTH. These

§ 36

76 BIJBEATJ 'OF AMERICAN- ETHNOLOGY [BULL. 40

apparently secondary de- prefixes will be listed together with and immediately following the da- prefixes, while the true, chiefly local, de-, (da)- prefixes will be put by themselves.

(a1) da-, (de-) MOUTH, IN MOUTH, WITH MOUTH, LIPS, TEETH, TONGUE :

Jda£0#o2Mhegavehim to eat (lit., he mouth-gave him) (186.25)

{fafugu's'i he gave me to eat 186.2

d&t!aya/i£ he went to get something to eat 75.9

dada'FdaW sharpen your teeth! 126.18; 128.23

da,ts!ala'ts!ili£n I chew it

ald&t!ele't!ili£n I lick it

d&lats!agi'£n I taste it (literally, I mouth-touch it)

aldo,p'dp'iwi£n I blow at it (194.1)

d&dama'£x he was out of wind 26.5

d&smayamaf£n I smile

Jiad&£yowo'uda£ (creek) going into (river) (literally, in- mouth- being)

td&ldul£ he lied (literally, he mouth-played) 110.23; 156.14

[deltinhixi he lied to me

d&yuwo'£s he suddenly stopped (singing, talking) (literally, he mouth-started, as in fright) 138.23

\da~k* da,Jiaali'£n I answer him (180.18)

{dak'deTielsi he answers me

(a2):

Tieedsle'lek!i£n I finished (story, talking) 50.4

delumu'sgade5 1 tell truth (184.3)

dexebenaY you said it (literally, you mouth-did it) 14.10; 15.6

aldets'!u'luk!i£n I suck it

dedets'!u'luk!i£n I kiss her (first de- as object, her lips; sec- ond de- as instrument, with my lips)

deJiememi'£n I taste it (cf . iJiemem- wrestle)

ba-idehenenaY you are through eating (literally, you are out-mouth-done) (136.16)

deligia'lda£n I fetch it for him to eat (130.9)

dehe'yelc!i£n I left food over

da- can not stand before I- HAND, because of the palatal timbre of the latter. Examples of desl-:

&eelda'mk!irik* it will get choked

deellats!agi'£n I touched his mouth (de- =da- as object; I- as instrument. Contrast above da-lats!agi'£n I tasted it, with da- as instrument) Similarly other palatal non-radical elements cause a change of

da- to de-: § 36

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 77

de-his-gulu-gwa'£n I want it in my mouth ( = I desire to eat

[his = trying])

(6) d&-,(da-) IN FRONT, AHEAD, AT DOOR OF HOUSE!

deelk!(dafJc!iUn (house) was scratched on door 154.1, 2, 3

dtflse'eY he opened door of house (cf. alse'eY he bowed to him) 63.12

de£lp*owo'£Jc* he bent it

'baad&/syeweya^w he started traveling again (literally, he up-ahead-went-again-with it) 22.4; 24.9; 25.6

dewiliwa'lsi she is fighting me 27.3

he stuck (threw) it into (fire) 27.8

tfn I brandish it before my face (172.12)

gasaflhi de'hitslaaga'e8 fast stepper (literally, quickly ahead- stepper)

ba-ide£di'nixiau£ they marched by in regular order (literally, they out-ahead-stretched) 144.14

dG£lwlHgi£n I spread it out (120.1)

t" gaa dofJii lc!iya'~k*i£ if the world goes on (literally, world ahead-goes-if) 146.4

dsimatslaW he put it point foremost (into their eyes) 27.8

As in the case of dale'-, so also here, not a few forms occur in which the meaning of the prefix da-, de- is far from being clearly in evidence:

d&t!aga£n I build a fire (96.17) [aU&tclu'lu^V he caught fire 98.3 [aldetdu'lu^xi I caught fire degulii'ldalx it glows (142.1) ; 188.15 aldsit*guyu'i£si (fire) blisters my face (25.11) .deetfa,'mak!isn I put out the fire d&£amcbf£x the fire goes out d&t!abaga'£n I finish it (176.6) d&sgayanaf£n I lie down

As the first seven of these examples show, da-, de- sometimes imply a (probably secondary) reference to fire.

3. dan-

(a) EAR, WITH EAR (referring to hearing), IN EAR, CHEEK, SIDES OF HEAD:

d&^tstayap* he washed his ear

d&lts' lamcfJc* he squeezed his ears

d&£ilats!agi'en I touched his ear, cheek

dsLaeagani/£n I heard it (55.3; 108.16)

d&&daagi'£n I am able to hear it (literally, I can ear-find it)

(100.12)

§ 36

78 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

d&&le€lagwa'£n I listen to him (55.1; 96.2; 146.5)

da,ats'!emxde£ I hear big noise 90.21

am£ ge d&'£yowo£ he did not listen to it (literally, he» not there

ear-was) 96.9

dsi&sgeJc!el7ia£n I kept listening (102.3) da,ayehel he went where he heard (noise of people singing or

gambling) 106.10.

da,adele'p*i he stuck it across his ear d&adalaga,'mt* he made holes in his ears

e' eeJc*t*1>aga,ms they had their hair tied on sides of head a- probably as incorporated phrase, over ears) 142.17 idify I pull out his hair (from side of head) (194.7) (6) ALONG, ON SIDE:

wi'lau dsi^watfba'vgamdiTia* arrows shall be tied along (their length) with it (i. e., sinew) 28.1

4. s'in- NOSE, IN NOSE, WITH NOSE:

s'iYflgile'£sgw(L he scratched his own nose 14.11 ; 15.7

s'int!ayagi'£n I smell it (literally, I nose-find it) (160.20)

s'mdalaga'mt* he made holes in septum (cf. under dda-) 22.1

s'inZo'wfc'i he stuck it into nose

s'in.deele'p*gwa he stuck it up into his own nose

S'mgeycfn he turned away his nose

s*myuwo'£s he dodged with his nose (as when fly lights; cf.

under da-)

s'mt*uwulc*de£ I feel warm in my nose s'mxi'ntfxanp^de5 1 sniff s'mwilifiJc*ap*de£ I blow my nose als'inld'uxa£n they meet each other (24.12)

5. gwen-

(a) NECK:

gwepiSgd'uda£n I cut his neck (14402, 3, 5, 22) gweuts!ayaga'£n I washed his neck 7ia-ugwen.yunu'£yini£n I swallow it greedily (cf. 126.10) gwenZo/wFi he stuck it in his throat (cf. under s'in-) 25.4 gwen£llats!agi'£n I touched back of his neck gwemvayanaganhi he swung his knife over their necks 144.2

(&) BACK, BEHIND:

gwe'n£alyowo£ he looked back gwenyewelt*es I went back (152.13; 188.19) gwe'nZiwiZaw£ he looks back (on his tracks) 59.14; 94.9 gwenhegwa'agwarihi he related it to him 17.11

In gwena-ia'£s GOOD SINGER, the part played by the prefix is not

clear. § 36

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 79

6. t- HAND, IN HAND, WITH HAND

No body-part prefix, except perhaps al-, is used with such fre- quency as v, the scrupulousness with which verbs implying action with the hand incorporate it seeming at times almost pedantic. Only a small selection out of the great number of occurrences need here be given:

Its lay ap* he washed his hand

Ipltfno'uYwafn I warm my hands

wilcfu £ihoyodagwa'£n I dance with arrow in hand

nax iheleelagwa/£n I sing with pipe in hand

lglrina he took it 15.1; 31.8; 44.8; 47.9

jJc*wsi'agwi£n I woke him up 16.4

lgaxagixi'£n I scratch him

lgis'igis'i'£n I tickle him

lJiegwefJiaYwncb£n I am working

xa£lts'!iwiY he split it open 26.6

iheme^m he wrestled with him 26.11; 27.10,11

lyono'u£Y he pulled it

lguyu'£Y she pushed her 55.14

s'eleVw lluf pxagwanY she shall pound with acorn pestle 55.9

Jiee£llemef£Y he killed them off 55.1; 144.6.

lt!a'ut!iwi£n I caught hold of her (29.12; 140.15)

ifwtfyili^n I make it whirl up

al£lyulu'yili£n I rub it

lt*gwanye'egit* you enslaved her 16.14

In some cases one does not easily see the necessity for its use : wi£l£ge'ye£xi they are round about me (48.5) cillwulu'u£xbi he ran away from you

7. xaa~, (oca-)

(a) BACK, WAIST:

x&Ats!ayap* he washed his back p!l* x.&adat*guyufi£sgwa his back got blistered 25.11 xa,£ilats!agi'£n I touched his back x.o,a'p!iind'v'k*wa he warmed his back 188.20 ^a,&la/ada£n I put (belt) about my waist

(b) BETWEEN, IN TWO (in reference to breaking or cutting) :

^Si&p!a-its'!iudifn I shall split it by throwing (stone) down

on it (140.7)

x.a,awlsaa go-between (in settling feuds) 178.11, 13, 18 £n I cut, saw it (21.2, 4)

(bodies) cut through 21.2

§ 36

80 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

'lhi he broke it with rock 24.4 ' eeJc*ffiagams it is all tied together 27.13

wili he broke it by stepping on it 31.4, 5 yL&*lee7Yik!dut~k!idi£n I broke it with stick

In x.&hege'TiaYna£n i BREATHE (79.2) and x&huk!u'Jiak'na£n i BREATHE, the xd- may refer to the heaving motion up from the waist.

(a) BACK:

The local uses of xaa- and dl{- (IN MIDDLE, BETWEEN, and ABOVE, respectively) would indicate that, in their more literal signifi- cation, they refer respectively to the LOWER BACK about the waist and the UPPER BACK, though no direct information was obtained of the distinction.

dlHsfayap* he washed himself in back of body d^hax his back is burning

diifbdulc!a'lxde£ I have warts on my back 102.20 d^duUgwcfnY she will wear it (i. e., skirt) 55.9 (Z>) ABOVE, ON TOP:

d^Jie'liya sleeping on board platform 13.2 dldaat*l)a'agamt*gwide£ I tie my hair on sides of my head (see

under daa-) (140.11; 142.17)

dli£algelegalafmda£n I tie his hair up into top-knot (172.2) dl£uyufts!amda£n I fool him (aorist uyutsl- laugh) dPJiinxd'ugtfn I scare him dPmas (earth) is lit up (78.1) d^TiiWgwaf^n I am glad 22.2

dl*- is used in quite a number of verbs of mashing or squeezing, the primary idea being probably that of pressing down on top of something:

dlip*ili'p*'ilien I squash (yellow-jackets) (74.3); contrast gel-b8m-plili'p'ili£n I whip him on his breast (literally, I-breast-stick-whip-him) (cf. 76.1, 2, 3) dlWiytfsi^n I mash them ba-idigwibl'Wwap* it popped all around 27.14 diWgumu'fgimi'n I squeezed and cracked many insects (such

as fleas)

In many cases, as in some of the forms given above, the primary signification of dtf- is greatly obscured. It is not at all certain but that we are at tunes (as in dl'uyu'tslamdafri) dealing really with the phonetically similar prefix dle- REAR. 36

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 81

9. gel-

(a) BREAST, WITH BREAST (mental activities) : geltslayap* he washed his breast ge\£ilats!agi'£n I touched his breast

baage'\£yo lie down with belly up! (lit., up-belly-be!) 140.4 ge\gulugwa'sn I desire, want it 32.5, 6, 7 gelhewe'Jiau he thought 44.11; 124.3; 142.20 gelloJioigwa'£n I avenge him (apparently = I breast-die- with

him) (146.8; 148.3)

gelttayaW they thought of it (see under s'in- and daa) 152.10 g£\yalaxaldi£n I forgot him (lit., I breast-lost him) (77.10) gelts'laya'mxamk* she hid (certain facts) from us 158.7 ge\dulu'Fde£ I am getting lazy ge\heye'£x he is stingy (literally, he breast-leaves-remaining =

keeps surplus to himself) 196.8 (&) FACING:

ge\t!anafhi she pushed him (? literally, she held him [away]

facing her)1 (25.10) gelwayan he slept with her (literally, he caused her to sleep

facing him) 26.4; (108.3; 190.2) wa't'gwan ge\£yowo'£ they faced each other (literally, to

each other they breast-were) 26.15 gelk!iyif£¥ he turned around so as to face him 170.2

10. dle-

(a) ANUS:

dl£ts!ayap* he washed his anus

ba-idl~fgats!a't*gisi£n I stick out my anus (164.19; 166.1)

dlsJiax his anus is burning 94.13

dl£hagalt'e£ I feel ticklish in my anus (as though expecting

to be kicked) (cf. under dak'-) 166.1 di£xo'us (food) is spilling out from his anus, (acorns) spill out

from hopper 94.2, 4, 5

(b) IN REAR, BEHIND:

dlssalyomo'Jiin I shall catch up with him in running bee dl'eJc!iyi/€Tc* afternoon came (lit., sun went in rear) (124.15) da£o'l diehiwilitit*ee I ran close behind As happens more or less frequently with all body-part prefixes,

the primary meaning, at least in English translation, of dl£-

seems lost sight of at times:

abaidleyowd'uda£ coming into house to fight (a&oi-into house; yowo'uda£ being) 24.14

1 Though perhaps better SHE HELD HIM WITH HER BREAST, taking gel- as instrument. 3045°— Bull. 40, pt 2—12 6 § 36

82 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

p!a-idi£hana'£s it stopped (wind, rain, snow, hail) 152.16 In a number of verbs dl£- expresses: felling, digging under, or erecting a tree or stick, the fundamental notion being probably that of activity at the butt end of a long object:

di£sgot!olha bem he was always cutting down trees 108.8

di£k!olola'n (tree) was dug under 48.5

dlelsguyu/uJclin (tree) was made to fall by being dug under

48.7, 8, 12 p!a-idl£lofugwa£n I make (stick, pestle) stand up (by placing

it on its butt end) (116.18; 176.1, 2) p!a-idl£sgimi'sgam they set (house posts) down* into ground

11. ha-

(a) WOMAN'S PRIVATE PARTS:

h&ts!ayap* she washed her private parts ha,sllats!agi'£n he touched her private parts ha,£iwesga'hak*w she spread apart her legs 26.4

'(&) IN:

(danxdagwa) h&ts!ayak' he washed inside (of his ear) (dexda) h&lo'uJc'i he stuck it into (his mouth) (s'inixda) h&dele'p*i he stuck it up into (his nose) hsiloJidn he caught them in trap (literally, he caused them

to die in) (100.8)

(gwari) hsit!ulugwa'£n I follow in (trail) (96.8,9) haZo/MF she put on (her dress), they put on (their skins,

garments) 160.6

ha,slhu'luuhal they skinned them 160.5 hB/yar-u£ge'nets!aen I put on (my vest)

As the last examples show Tia- sometimes conveys the special notion of putting on or taking off a skin or garment.

12. gwel-

(a) LEG, IN LEG, WITH LEG:

gwe\ts!ayap* he washed his legs gwelle'yeesde£ I am lame

put on (your leggings)!

i/£?i I beat him in running (lit., I-leg-left-him) (184.14) gwe\salt!eyesnaen * I have no fat in my legs and feet 102.22

(&) UNDER, AWAY FROM VIEW:

gwelmate/avF they put (food) away (sc., under platforms)

124.22; (132.8) gwelge'l£yowduda£ he having his back to him (literally, facing

him away from view) 122.7

1 This form is an excellent example of the rather uncommon coordinate use of two body-part prefixes (gwel- LEG and sal- FOOT).

§ 36

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 83

13. la-

(a) FRONT OF BODY (probably BELLY as contrasted with gel-

BREAST) :

\a,ts!ayap* he washed himself in front of body (6) BURST, RIP OPEN:

laf Wax it burst 24.17

la,£lt*l>afaJc!it*1)a£ you (pi.) shall rip them open (like game

after roasting) 118.5

leisalt'~ba'agi£n I burst it with my feet (140.22) \a,£wayat*bd/agi£n I rip it open with knife (waya knife, as

incorporated instrument)

14. sal-

(a) FOOT, WITH FOOT:

s&llats!agi'£n I stepped on it (instrument sal-: I foot-touched

it) (196.18)

s&\£llats!agi'£n I touched his foot (object sal-; instrument 1-) s&\ts!ayap* he washed his feet s&lxugl they are standing 63.2 ~h,ee£$&\t'gUn kick him off ! (24.17) als&lt'ba'a¥ he kicked him 86.16,17,18 gelbam s&\gwi't*gwat* kick it way up! s&\yuwo'£s he suddenly lifted up his foot (as when frightened)

(cf. under da- and s'in-) s&\p!tfnd'ulc'wa£n I warmed my feet

15c al- FACE, WITH EYE, TO, AT

This is in all respects the most difficult prefix in regard to the satisfactory determination of its exact meaning. In a large number of cases it seems to involve the idea of sight, not infre- quently adding that concept to a form which does not in itself convey any such implication. In most of the verb- forms, however, many of which have already been given under other prefixes, the al- seems to have no definitely ascer- tainable signification at all. In some cases it may be consid- ered merely as an empty element serving as a support for a post-positive modal particle. For example:

&\-Tiis-gulugwa'£n I am desirous of something

where Jiis TRYING can not occupy an initial position

B\-di-yolc!oyaY did you know him?

Here &lyolc!oyaY in itself hardly differs in content from yoTdoyaY YOU KNEW HIM. The most satisfactory definition

§ 36

84 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

that can be given of al- in its more general and indefinite use is that it conveys the idea of motion out from the sphere of the person concerned, whether the motion be directed toward some definite goal (object) or not; an approximate translation in such cases would be TO, AT. The correctness of this interpretation is borne out by the fact that al- at times replaces a more definite local phrase, as though it were a substitute for it, of the same general formal but weaker material content.

waada lougwa'£n to-him I- thrust-it, where wdada definitely ex- presses a local pronominal idea TO, AT HIM. Compare :

alld'ugwi£n I stretched it out to him

where the exact local definition of the action is not so clearly expressed; the direct object of the verb being here not the object thrust, but the person aimed at, while the indirectness of the action is interpreted by means of al- as an adverbial or local modification of the verbal content. The change of vowel in the ending, a — i, is closely connected, as we shall later see, with this change of "face" in the verb. The first form may be literally translated as TO-HIM I-IT-THRUST; the second, as I-HIM-TO-THRUST (IT) . Similarly, in &\£ilats!agi'£n i TOUCHED HIS BODY, the al- is probably best considered as a general directive prefix replacing the more special prefixes (such as sal-, s'in-, and so on) that indicate the particular part of the body affected, or, as one might put it, the exact limit of motion. The use of al- in local phrases shows clearly its general local significance: o\s'duma^l AT, TO THE MOUNTAIN; ga£a^l TO THAT, as postposition equivalent to TO, FOR, FROM. (a) FACE, EYE:

&l£oudini'£n I look around for him (cf . ouda'£n I hunt for him) (92.27)

&\xlfigi£n I see, look at him (-xtfg- never occurs alone) 186.7; 188.11.

&\gaya^n he turned his face

&\yebebi'£n I showed it to him (77.8)

si[yowdt'e£ I looked (cf. yowot'e£ I was) (64.3)

&lts!ayaga'£n I washed his face (64.5)

manx &lnu'uk'wa he painted his (own) face § 36

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 85

W he found, discovered it (literally, lie eye-found it;

cf. under s'in-, daa~, and gel-) 47.10; 92.27; 194.13 a\sgalaaliwi'sn I looked at them (moving head slightly to

side)

&\t'~bdu'k!aflxde£ I have pimples on my face (cf. 102.20) &lt*wap!a't*'wap*na£n I blink with my eyes 102.20 &\we'Jc!ala£n I shine xa£&rlt!anaTii they watched it (literally, they-between-eye-

held it; xa-£al as incorporated local phrase[?]) 136.8

(6) TO, AT:

It is at least possible, if not very probable, that al- TO, AT, and al- EYE, FACE, are two entirely distinct prefixes. As many preceding examples have incidentally illustrated the local use of al-, only a few more need be given:

slp'oup'auTii he blew on it 15.1

&\Myuxde£ I go hunting (42.1; 58.14; 70.2; 126.21)

&lgesegasa'lt*e£ I was washing

&lhemeW they met him 24.11

%\.slxlep!e'xlap* he mashed it up into dough-like mass 94.11

&\£lts' ! o'udi£n I touch, reach it

&\se'egi£n I bowed to him (172.10)

16. dlual- FOREHEAD:

dli£&\ts!ayap* he washed his forehead

dli£Si\gelegala'ms he tied his hair up into top-knot 172.2

dl^alFa'^V^ ne Put (dust) on his forehead 136.28

17. gwenha-u- NAPE:

gwei\h.Sif-uts!ayaga£n I shoot off nape of neck

gwenha-uZ' be'egams he has his hair tied in back of his head

It will have been noticed that several of the body-part prefixes have developed special uses that almost entitle them, at times, to being considered verbal in function. Thus xda- BACK, BETWEEN has been seen to develop, from its latter local use, the more strictly verbal one of cutting, splitting, breaking, or rending in two; the ideas of BETWEEN and of DIVISION IN TWO are naturally closely associated. The specialized semiverbal uses of some of the prefixes may be thus

listed :

da-, de- activity in reference to fire (burn, set on fire, glow) xaa- rend in two (cut, split, break) al1- crushing activity (mash, squeeze) dl£- fell, erect (long object) Jia- dress, undress

§ 36

86

BUREAU OF AMERICAN ETHNOLOGY

[BULL. 40

la- burst, rip open al- look, see

The resemblance between this use of the Takelma body-part prefixes and the Siouan use of verb prefixes denoting instrumental activities (e. g., Ponka l>a- BY PRESSING WITH THE HAND, ma- BY CUTTING, $a- WITH THE MOUTH, BY BLOWING) is not far to seek, although in Takelma the development seems most plausibly explained from the local, rather than the instrumental, force of the prefixes. Neither the employment of Takelma body-part nor of Siouan instrumental prefixes with verb stems is in any morphologic respect comparable to the peculiar com- position of initial and second-position verb stems characteristic of Algonkin and Yana. The same general psychic tendency toward the logical analysis of an apparently simple activity into its com- ponent elements, however, seems evident in the former as well as in the latter languages.

§ 37. LOCAL PREFIXES

The purely local prefixes, those that are not in any way associated with parts of the body, are to be divided into two groups:

(1) Such as are used also in the formation of noun and pronoun local phrases or of postpositions, these being in that regard closely allied to the body-part prefixes in their more general local use; and

(2) Such as are employed strictly as verbal prefixes, and are inca- pable of entering into combination with denominating elements. The following table gives all the common prefixes of both groups, examples of noun or pronoun local phrases being added in the last column :

Prefix.	Translation.	Local phrase.
han-	across, through	hanwaxga^n across the creek
ha-u-	under, down	hawandS under me
hees-	away, off	hetes-oumaV beyond the mountain
dal-	away into brush, among, between	dan gada^l among rocks
Jiafya-	on both sides	ha'tyade on both sides of, around me
Aoof-	yonder, far off
m<£-	hither
wl-	around
hawi-	in front, still
wa-	together
baa-	up
ba-i-	out, out of house
p!a-i-	down
aba-i-	in house, into house
bam-	up into air
xam-	in river

§ 37

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 87

Of these, the first five belong to the first group, the last nine to the second. The position of liaa£- and me£- is somewhat doubtful; but the fairly evident etymological connection of the former with 7iaa£ya and the correlative relation in form and meaning between me£- and he e£-, make it probable that they are to be classed with the first group. While some of these prefixes (such as dal- and Jian-) are inconceivable as separate adverbial elements, others (particularly aba-i, which is apparently composed of demonstrative element a- THIS -f 6a-i) are on the border-land between true prefix and inde- pendent adverb. me£- and liee£-, though they are never used alone, stand in close etymological relation to a number of local adverbs (such as eme£ HERE and ge THERE), which also, though not so rigidly as to justify their being termed prefixes, tend to stand before the verb. The difference between local prefix and adverb is one of degree rather than of fundamental morphologic traits; in any case, it is rather artificial to draw the line between me£- in such forms as me£yeu COME BACK! and ge in, e. g., ge £yowo'£ THERE IT is. Sometimes, though not frequently, two local prefixes, neither of them a body-part element, occur in a single verb form. See, e. g., p!ai-hau- under 2 below, also abai-baa- 62.1.

1. han- THROUGH, ACROSS:

h&nyada't'e£ I swim across

he threw it across 120.22 ' he looked through it h&nyewe'ie he went back across 178.16

gwan-h&i].sgd'usde£ I lie stretched across the trail (literally, I- road-across-cut) (148.8)

2. ha-u- UNDER, DOWN:

h&-ugwenyut!u'yidi£n I swallow it down greedily, making grunting

noise (126.10)

ha-usaF™ he paddled him down river (baa- up river) ha,-uyowo't*e£ I sweat (literally, I-under-am) ei p!a-iha,'-\it'guupx canoe upset 60.8 h&-uhana/£s it stopped (raining) 196.8

3. hee£- OFF, AWAY:

he£ileme'£F he killed them off 14.13; 110.21; 144.6 h&e£sgofuda£n I cut it off (44.4); 72.10; (92.14,16) hee£gwidiWw he threw it away heesiuk*wa he went away from him (23.12; 146.18) hee£salfgunfgini£n I kick him off (24.17)

§ 37

88 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

hee£l7iu'lup !i£n I beat off bark (with stick) h&tflc^apIaWibfrn I chipped them off (92.3) hee£waaga'£n I buy it (literally, I carry it off) (176.17) he*et*guyu'i£s it is blistered

4. dal- INTO BRUSH, AMONG:

dalyewefi£ he ran off into brush 14.6; 110.10 d&\gwidiWw he threw it into brush d&lp*d'udi£n I mix it with it (178.5) dalxabili'u£ he jumped between them 106.20

5. ha£ya- ON BOTH SIDES:

haeya^m'£F they passed each other

h&£ya,wat!emexiau£ they assemble coming from both sides 144.23

6. haa£ FAR OFF:

h&'a£yewei£ they returned going far off 146.22; (47.4; 188.1) ha,£xda'axdagwa£n I threw something slippery way off This prefix is evidently identical with the demonstrative stem Jiaas seen, e. g., in haf£ga THAT ONE YONDER.

7. me£- HITHER:

* he came here 146.24 (ge gini'£¥ he went there 77.7) iniW they come from across (note two local prefixes; Jiangintflc" they go across)

me£^ come back! (y£n return!) (23.11,12,13,14; 96.5); 59.5 me£hiwili'u£ he came running this way

Not infrequently me£- conveys the fullar idea of COME TO ,

as in: me£bep*xip' come (pi.) and chop for me! 90.16

8. wl- AROUND:

wielt*ge'yesxi they are surrounding me (48.13; 190.14) wtfge'yee*k*i they put it round about 176.14

9. hawi- IN FRONT, STILL:

fhawM/anZV I go in front

\h&wiyana/£s front dancer

hawi&am/£m still they come, they keep coming 146.1

~bou h&wideguflk!alxdaa after a while it will blaze up (bou = now)

10. wa~ TOGETHER: waJdoyoxiniJc* we go together

w&slts' !o'meJc* squeeze (your legs) together! (26.5)

baawsiwilik*w he traveled up along (river) (literally, he went up

having it together with him) 21.14 w&yanlc*w he followed him (literally, he went having him together

with him) 23.11 § 37

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 89

wa,t!emexiau£ they are assembling together (110.3); 144.23

wsi£lt!oxofxi he gathered them, together 112.6

WB,t!illJc*ni she gave them one each 130.4

w&himiY he talked to him 59.16; 63.10

da'gaxdeV w&£alt*geye'fgiyi£n I tied it about my head (literally,

my-head I-together-to-surround-it) p!das wsJc!eewa'lxgwa snow is whirling around

Sometimes wa- seems to indicate simultaneity of activity, as in: wa2dala'u7ii she kept twining basket (while talking) 61.51

In many cases the adverbial meaning of wa- is hardly apparent, and one is sometimes in doubt whether to look upon it as the prefix here discussed or to identify it with the instrumental element wa- WITH, WITH IT; the two may indeed be at bottom identical.

11. baa- UP (55.16; 59.10; 60.11; 63.6,12):

b&adini'£x (clouds) were spread out in long strips (literally, they

stretched up) 13.3 b&H!ebe'£e£ I get up 186.14; (196.1) b&awadawayaWw he flies up with it b&yarik*" he picked it up 15.9; 24.3; 59.15 Jcliytfx baaitfdF smoke comes out (literally, up-arrives) 29.3 (danxda) ba,£algwili^s he turned up (his ear) (dak'will) \)&*gini'eTc* he went up (on top of house) 30.6 bsias'd'£s' stand up!

b&ayewe'i£ he got better (literally, he-up-returned) (15.2) b&ahawa'£Jc* she dipped up (water)

12. ba-i- OUT, OUT OF HOUSE, OUT OF WATEK TO LAND, FROM

PLAIN TO MOUNTAIN: b&-iyewe'i£ they went out again • b&-ixodo'xat' she took off (her garment) 13.4 borisili'xgwa he lands with (boat) 13.5 ba-isaFw he came to land b&-i£a'lyowo£ he looked outside b&-ihimima'£n I drive him out ba-i gwidWw he threw it out 92.15,16; (haxiya'daf) b&-igwidi*k*w

he threw it (from in the water) on to land (31.2) bsi-ibiliwaY you jumped out of house 24.15; (46.6) Qiadede) b&-iyeweyini'£n I took it out (of my mouth) (literally,

I-out-caused-it-to-return) bsi-ideJienenaV you are through eating (literally, you-out-mouth-

are-finished) (132.14) b&-it!ixi'xi he pulled (guts) out 92.17

(dak*s'dumafl) ba-iwofc' he got up (on the mountain) 124.4; (60.9)

§ 37

90 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

In certain idiomatic turns the primary signification of 'ba-i- is as

good as lost:

(7ieel-)l>ei-imats!aW he began to sing (lit., he-song-out-put) 102.17 }>a,-ik!iyi'£V he comes 92.1, 2; 156.24; 168.13

13. p!a-i- DOWN:

pla,i£tt!ana'Jii£n I held him down

pldL-igwidiWw he threw it down

pl&-iwaya'£ he went to lie down, to sleep (lit., he down-slept) 25.9

p \a,-ilo7iolt*e£ I fell down (literally, I down-died)

pl&-iyewe'i£ (arrow) fell down back 22. 5; 48.14

p \&-i£a'lyowo£ he looked down 26.14

p\&iyowo's they sat down (literally, they down-were) 56.2

p \&-isgaya'pxde£ I lay down

14. aba-i- IN HOUSE, INTO HOUSE

It would perhaps be best to consider this an independent adverb (demonstrative pronoun a- THIS + ba-i-, formed analogously to eme£ HERE [ = demonstrative adverb e- HERE + me£]) ; its correlative relation to ~ba-i- makes it seem advisable to give examples of its occurrence here:

abai#im'£F he went inside 25.8; 27.7,13; 64.3

&b&ihiwili'u£ he ran inside 16.12

aba-iwoF they went into house 29.6; (44.7); 160.19

aba-h/owctfV I stay at home

abaife/a°Ffe/e&/€F he stepped into house 31.3

15. bam- UP INTO AIR

This prefix occurs often with preposed elements gel- or di*- as gelbam- or dtfbam-, which would seem to mean respectively

WITH BELLY SIDE UP and WITH BACK SIDE UP, Or IN FRONT OF

and DIRECTLY OVER one : w he threw it up w he threw it up w he threw it up gelbamsaF™ he shot it up 22.5

he looked up he was sitting up (in tree) 48.7

16. xam- IN RIVER, INTO WATER, FROM MOUNTAIN TO PLAIN: xa,malts!ayap* he washed himself in river xa,mgwidWw he threw it into river (33.6) ; 108.5 xa,mUwili'u£ he ran to river 29.13; 94.16

they became in river ( = were drowned) 166.16 he looked down from top of mountain 124.4 (con- trast p\a,i£aflyowos he looked down from ground 26.14) § 37

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 91

§ 38. INSTRUMENTAL wa- it is somewhat difficult to classify this prefix, as it does not belong either to the body-part or the purely local group. Strictly speaking it should be considered the incorporated form of the demonstrative pronoun in its instrumental function. As was seen above, it may represent an instrumental noun, but, while the noun may itself be incorporated to denote the instrument, this is not the case with the demonstrative pronoun. For example:

ga wede yap!a-wa-domJiigas that not I-people-with-shall-kill ( = 1 shall not kill people therewith)

In other words, it would seem likely that such a form as ga al£wa- ts!ayagif£n i WASH HIM WITH THAT is related to an al£wats!ayagi'£n i WASH HIM WITH IT as, e. g., xi al£wats!ayagi'£n i WASH HIM WITH WATER, to the form alxits!ayagi'£n i WATER- WASH HIM, i. e., the wa- rn al£wats!ayagi'£n is to be regarded as an incorporated ga THAT, IT (such forms as *algats!ayagi'£n have never been found to occur). It will be noticed that the verb-forms with incorporated wa- are nor- mally characterized by a suffixed -i- or -hi- ; as soon, however, as the verb loses its instrumental "face," this -i- is replaced by the normal -a-. Thus:

wilau wats!ayagif£n arrow I-shoot 1 -him- with-it (with incorpo- rated wa-, wila^u ARROW being outside the verb-structure and in apposition with wa-) but:

ts!ayaga'£n wi'lau wa^ I-shoot-him arrow with (in which also wa- stands outside the verb-complex, acting as an instrumental postposition to wila^u) Examples of instrumental wa- are:

(salxdeJc')sal£wsilats!agi'£n I touched him with my foot (literally, my-foot I-foot-with-it-touched-him)

(xli}^N^suugwarnhi I drink (water) with it

(yap!a)wa,t!omomi'sn I kill (people) with it (but yap!a t!omoma'£n I kill people)

alw&ts!eyeFwides I washed myself with it

ga Ids doumia gelwa,gulugwi'£n I try to kill him with that (literally, that trying killing-him I-with-desire-it)

seel-wa,ts!elelamda£n I write with it

(iuxdeW)w&gaya-iwi'£n I used to eat with (my hands)

1 Aorist ts.'ayag- SHOOT and aorist ts.'ayag- WASH are only apparently identical, being respectively formed from stems scfaQ- and tstcLiQ-

§ 38

92

BUREAU OF AMERICAN ETHNOLOGY

[BULL. 40

(p*im)wa,sana'hink* they will spear (salmon) with it 28.15 (cf.

sancfnY they will spear it)

Although, as was suggested before, the prefix wa- as instrument may be ultimately identical with the adverbial wa- TOGETHER (the concepts of DOING SOMETHING WITH, BY MEANS OF IT and DOING SOMETHING TOGETHER WITH IT are not very far removed), the two can not be regarded as convertible elements. This is clearly brought out in such forms as bem wa,elwa,t!oxo'xisn i PICKED THEM TOGETHER WITH STICK. Literally translated, this sentence reads, STICK i- TOGETHER-HAND-WITH-IT-PICKED-THEM; the first wa- is the adver- bial prefix; 1-, the general instrumental idea conveyed by the character of the verb (GATHER WITH ONE'S HANDS) ; and the second J the incorporated representative of the more specific instrument

STICK. If preferred, I- may be interpreted, though less prob- ably, as a local element (-Iwa- = with it in hand) .

2. Formation of Verb-Stems (§§ 39, 40)

§ 39. GENERAL REMARKS

By a verb-stem will be here understood not so much the simplest possible form in which a verb appears after being stripped of all its prefixes, personal elements, tense-forming elements, and derivative suffixes, but rather the constant portion of the verb in all tense and mode forms except the aorist. The verb-stem thus defined will in the majority of cases coincide with the base or root, i. e., the simplest form at which it is possible to arrive, but not always. Generally speaking, the aorist is characterized by an enlargement of the base that we shall term " aorist stem," the other tense-modes showing this base in clearer form; in a minority of cases, however, it is the aorist stem that seems to coincide with the base, while the verb-stem is an amplification of it. Examples will serve to render these remarks somewhat clearer:

Aorist stem	Verb-stem	Probable base
t.'omom-	downi'	doum- kill
naga-	naag-	ndag-(nag-~) say to
haal-	hala-	hdfllr answer
dud-	odo	dud- hunt for
lohoi-	loho-	loh- die
yuluyal-	yulyal-	yul- rub

39

BOAS]

HANDBOOK OF INDIAN LANGUAGES TAKELMA

93

By far the larger number of verbal bases are monosyllabic. Where the simplest radical element that can be analyzed out remains dis- syllabic (as in dawi- FLY, agan- PERCEIVE, yimi- LEND), the proba- bility is always very great that we have to reckon either with ampli- fications of the base, or with suffixes that have become so thoroughly amalgamated with the base as to be incapable of separation from it even in formal analysis; in some cases the dissyllabic character of the verb-stem is due to a secondary phonetic reason (thus dawi- is for dawy-, cf. dauy-\ while in agan- the second a is inorganic, the real stem thus being *agn-) . Most bases end either in a vowel or, more frequently, in a single consonant; such as end in two consonants (as yalg- DIVE, s'omd- BOIL, bilw- JUMP) may often be plausibly suspected of containing a petrified suffixed element.

The few examples of verb and aorist stems already given suffice to indicate the lack of simple, thorough-going regularity in the forma- tion of the aorist stem from the base. Given the verb-stem, it is possible only in the minority of cases to foretell the exact form of the aorist stem. Thus, if doum- had followed the analogy of the pho- netically parallel naag-, we should have in the aorist not Homom-, but dome-; similarly, the phonetic similarity of odo- and loho- would lead us to expect an aorist stem louJi-, and not lohoi-, for the latter. Nor is it safe to guess the form of the verb-stem from a given aorist stem. Thus, while the aorist lohoi- corresponds to a verb-stem loho-, yewei- corresponds to yen- RETURN; nagai-, to na- SAY, DO; and Jclemei-, to 7c!emn- DO, MAKE. Mere phonetic form has, indeed, com- paratively little to do with determining the relation of the two stems. This is clearly evidenced by the following cases of homony- mous but etymologically distinct bases with corresponding aorist stems.

Verb base	Meaning	Aorist stem
heem-	1. mock 2. wrestle	hemeham- hemem-
Jieegw-	fl. work (2. relate 1. be finished	hegwehagw- 1iegw(h)aagw- , Jiegwe- Jiagw- henen-
heen-	2. wait for	henee-
daag.	1. find 2. build fire	t.'ayag- t.'agAi-

39

94

BUREAU OF AMERICAN ETHNOLOGY

[BULL. 40

The signification of the verb-stem gives almost no information as to the form of the aorist stem, the various types of aorist forma- tion being each exemplified by a heterogeneous array of verbs, as far as any discernible similarity of meaning is concerned. It is true that, in a comparatively few cases, certain types of aorist formation can be shown to be characteristic of intransitive verbs; but in these the formation of the aorist stem involves the addition of a distinct pho- netic element that has every appearance of being a worn-down suffix.

Not the least remarkable feature of tense-formation lies in the fact that the most frequently used of the tense-modes, the aorist (equivalent to immediate future, present, and past) , generally shows the derived or amplified form of the base; while the far less important tense-modes, the future, inferential, potential, and present and future imperatives employ the generally more fundamental verb-stem. In its naked form the aorist stem appears as the third person subject third per- son object aorist transitive. For example:

tlomom he killed him nagcf he said to him -hal he answered him o'uC he hunted for him

The bare verb-stem appears as the second person singular (third per- son object) present imperative intransitive and transitive. For ex- ample :

doum kill him ! odo^ hunt for him! no? say! do!

and as the first element of the periphrastic future, that will later receive treatment.

In striking contrast to the extensive use in Athapascan of distinct and unrelated stems for the singular and plural, only a very few such cases have been discovered in Takelma; and even in these the singu- lar stem may, it seems, also be used in the plural.

Sing, verb-stem	PI. verb-stem	Sing, form •	Tl. form
s-as'- stand	sal-xogw-	s'as'im he stands	sal-xogwl they stand
		baa-saasa'sdes (= saas- sas-) I come to a stand	baasal-xo'xiginak' (= xog-xag-} we come to a stand
s~u£al- sit	al-xalii	s-uswiUt'ee (= s'usall-) I am seated	al-xattyanaW we are seated

39

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 95

It is interesting to observe that, while STAND and SIT are intran- sitive in the singular, the plural stems sal-xogw- and al-xall*- make transitive forms with a third personal object (-ancfY first person plural aorist transitive, -W intransitive; cf. tlomomancfY we kill him, but s'as'inlp^iY we stand and s'uswillp'iJc' we are seated, dwell, stay).

The great majority of verb-stems are either necessarily transitive or intransitive, or are made such by appropriate suffixes. Only a few cases occur of verbs that are both transitive and intransitive, the respective forms being kept distinct only by the varying pro- nominal suffixes. Such are:

moyugw-a'n-t*e£ I am spoiled, and moyugw-an-a'£n I spoil him ligl-n-t*es I rest, and Ugtf-n-a^n I rest him

~k!uwu'£ they ran away in flight, and Jcluwu he sowed, threw them about

Certain forms are alike for both transitive and intransitive; e. g., second person plural subject: Jc!uwuwaYp\

§ 40. TYPES OF STEM-FORMATION

In looking over the many examples of verb and corresponding aorist stems obtained, it was found possible to make out sixteen types of stem-relations. Of this large number of types about half are of frequent occurrence, while of each of the rest but few examples have been found. It is not claimed for a moment that all of these types should be regarded as being exactly on a par, but merely that they have the value of forming a convenient systematization of the some- what bewildering mass of methods of radical or base changes encoun- tered. It is very probable that some of these are ramifications of others, while some types show more or less petrified suffixes that for some reason or other became specialized in certain tenses. As com- parative linguistic material is entirely lacking, however, we can not make a genetic classification of types; a purely descriptive classifi- cation must suffice.

In the following table of types of stem-formation, c means conso- nant; v, vowel; d, the fortis correspondent of c; clf c2, and so on, other consonants; vv denotes pseudo-diphthong; other letters are to be

literally interpreted.

§ 40

96

BTJBEAU OF AMERICAN ETHNOLOGY Table of Types of Stem-Formation

[BULL. 40

Type No.	Formula verb-stem	Formula aorist stem	Example verb-stem	Example aorist stem
1	v+c	w+c	06- dig up	oub-
2	tH-(c)	v+c+v	j/o-be	yowo-
3	V+C+Ci	v+c!+v+ci	<uits!- laugh [masg- put	uyuts!- matsfag-
4a	VV+C	v+c+v+i	t'aag- cry	t'agai-
46	v+c+v	v+c+v+i	loho- die	lohoi-
5	v+c+v	vv+c	yana- go	yaan-
6	w+c!	vv+c	p'ot!- mix	p'oud-
7a	C+W+Ci	C.'+V+Ci+V	de«b- arise	t.'ebe-
76	C+W+Ci	c!+v+c\+v+i	duugw- wear	t.'ugui-
8	C+W+Ci	C.'+V+Ci+V+Ci	goul- dig	k.'olol-
9	C+VV+Ci	c!+v+y+v+ci	daag- find	t.'ayag-
lOa	C+V(+Ci)	C+V+C(+Ci)	Zo«- play	loul-
106	C+V+Ci	C+V+Ci+C(+V)	sana- fight	saans-
11	C+V+Ci+C	C+V+Ci+V+C	yawy-t&lk	yawai-
12	C+VV+Ci	c+w+ci+c+a+ci	t!bu- play shinny	t.'tut.'au-
13a	c+v+Ci+c+a+ci	c+v+Ci+v+c+a+ci	sensan- whoop	senesan-
136	c+v+Ci+c!+a+c\	c+v+c.i+v+c!+a+ci	dultlal- stuff with	dulutlal-
13c		C+V+Ci+V+C+Ci		lobolb- be accustomed to pound (also lobolab-)
14	v+c	v+c+v+n	xeeb- do	xeben-
(15a	—	-H	s-as-an- stand	s-as-inli-)
(156	-as	-V	dink. 'as- lie spread out	dinkfli-)
(16	v+c+ci+i	V+C+V+Ci	k.'alsi - be lean	k'alas-)

Not all forms find an exact parallel in one of the sixteen types here listed. There is a considerable number of more or less isolated cases left, particularly of frequentative or usitative forms, that it is difficult to classify ; but on closer examination some at least of these are seen to be secondary developments. Verb-stem al-sgalwal(w)-

KEEP LOOKING BY TURNING HEAD SLIGHTLY TO SIDE, as Compared to

aorist stem al-sgalaal(a'w)-, looks anomalous because of its apparently inserted first -w-; but these two forms become explicable as frequen- tative developments, according to Type 8, of their corresponding simplexes, verb-stem al-sgalw- LOOK BY TURNING HEAD TO SIDE and aorist stem al-sgalaw-. It will be convenient to dispose of such anomalous and difficult cases under such headings as allow them to appear as at least comparatively regular formations. It should not be supposed that a particular yerb-stem always and necessarily involves a fixed aorist stem in all possible derivations of the verb, though in probably the larger number of cases such a fixed parallel- ism may be traced. As examples of the occurrence of more than one aorist stem to match a verb-stem may be mentioned : § 40

BOAS] HANDBOOK OF INDIAN LANGUAGES — TAKELMA 97

verb-stem -xlk!- see; aorist Type 6 -xVg- and Type 2 -xlk!i-xa- see (without object)

verb-stem yeu- return; aorist intransitive Type 4 yewei-, causa- tive Type 2 yewee-n-, and, according to Type 8, yewew-ald- go back for some one

There are few if any verbs whose verb and aorist stems absolutely coincide. If in nothing else the two differ at least in the quantity of the stem vowel, the aorist stem always tending to show a long vowel. In some cases the two (dissyllabic) stems seem identical in phonetic form because of the persistence of an inorganic a in the second syllable of the verb-stem and the presence of a repeated radical a in the second syllable of the aorist stem. Sometimes only certain of the forms built on the verb-stem exhibit the inorganic a; in such cases the secondary character of the a is directly proven by the forms that lack it. A case in point is:

aorist stem ts'layam- hide; verb-stem ts'!ay[a]lm- and ts'!a-im-

Other verbs, however, are phonetically so constituted as to require the presence of the inorganic a in all forms derived from the verb- stem. Such are:

aorist stem agan- feel, hear; verb-stem ag[a]n- aorist stem plahan- be ripe, done; verb stem p!ah[a]n-

Under such circumstances ambiguous forms may result; e. g.; M?aeaganiY may be construed either as an aorist (YOU FELT IT) or as a potential (YOU WOULD FEEL IT) derived from the stem ag[a]n-. But evidence is not lacking even in these cases to prove the inor- ganic character of the second a in the non-aorist forms. One test has been already referred to in another connection — the incapability of a secondary diphthong (a diphthong involving an inorganic a) to have a rising accent. Thus:

aorist Ja°£agan (-agavn) he heard it; but imperative cZaa£ag[a']n

hear it!

A second test is the failure of inorganic a to become ablauted to e. Thus:

aorist p!ehen- a'nxi he causes me to be done; but future p!eh[a]n- a'nxinlc he will cause me to be done

The various types of stem-formation will now be taken up in the order of their occurrence in the table.

1 Brackets indicate an inorganic element. 3045°— Bull. 40, pt 2—12 7 § 40

98 BUREAU OF AMERICAN ETHNOLOGY [BULL. 40

TYPE 1. Verb-stem v + c; aorist v v + c. In this type are embraced partly monosyllabic and partly dissyllabic verb-stems that either seem to undergo no change at all in the aorist or merely lengthen the stem-vowel. The number of verbs that follow the type does not seem to be very great. Examples:

Verb-stem Aorist stem

woga/£f he will arrive (196.20) wouk' he arrived 47.15

oba'n I shall dig it up oubafen I dug it up (48.7)

yixl^ copulating 86.5 yllla'en I copulated with her 26.3

ugwa'n I shall drink it (162.17) uugwa'$n I drank it 186.3

hogwana'n I shall make him run hougwana'£n I made him run

(138.2) (79.2)

7iinex-mwa/es coward 76.5 ; 1uinex-i&wa'*n I was afraid (17.7)

(160.19)

wife6 1 shall travel (178.11) wife* I traveled (90.1)'

t Wla'mxade e I shall go fishing t ftla'mxade* I went fishing

yimi'Am I shall lend it to him yi1miya'*n. I lend it to him

(98.14) (98.15)

huli'itfV I shall be tired out huuli'n£V I was tired out (102.1)

hagaif ee I shall have a cold thrill hagaif e£I had a cold thrill 166.1

lohona'n I shall cause him to die lohouna'£n I caused him to die

(100.8)

aZ-ge'y &nde e I shall turn my face al- gey ano/£?i I turned my face

As regards the accent of the stem syllable, the examples show that, whenever accented, it takes the rising pitch when long, the raised pitch when short (and final). Compare further:

oup" he dug it up 124.5, 12 hin£x-nlu he was afraid

Uk*w he drank it 162.20 al-geycfn he turned his face

TYPE 2. Verb-stem v + c; aorist v + c + v. If, as seems probable, the second consonant of verbal bases ending in two consonants is in many cases really a petrified suffix, a very large proportion of those verbs that might be listed under Type 3 really belong here, thus making Type 2 probably the most numerously represented of all types. In some forms it is possible to detect the derivative character of the second consonant by a comparison of etymologically related forms that lack it; e. g., in ts'Ielm- RATTLE (aorist ts'Idem-), the -m- is shown to be a suffix, though of no determinable signification, because of its absence in the corresponding frequentative ts'lelets'lal-. A corroborative phonetic test lies in the treatment of the first con- sonant of the cluster, in so far as verbs following Type 3 show a fortis in the aorist as against a media or tenuis in the verb-stem, while those

§ 40

BOAS] HANDBOOK OF INDIAN LANGUAGES TAKELMA 99

of Type 2 suffer no change in this respect; e. g., verb-stem wism- MOVE has aorist according to Type 3, wits' !im-, as contrasted with verb-stem t'gism- GET GREEN with aorist of Type 2 t'gisim- (t'gism- should therefore be analyzed as base t'gis- + suffix -m-). This criterion enables us to pick out an otherwise unsuspected suffix in verbs like t!ap*g- FINISH, aorist tlabag- (not Type 3, *l!ap!ag-}, but can be applied only where the first consonant of the verb-stem is s, ?>, d, or g. A more general phonetic test would seem to be the position occupied by the inorganic vowel -a-. In those cases in which we have most reason to consider the second consonant as part of the base, this -a- follows the cluster as " constant" a; while otherwise, and indeed in the majority of cases, it is inserted between the two consonants: wismaft"ee i SHALL MOVE (base wism-\ but t*gisa'mt'ee i (AS PLANT) SHALL GET GKEEN. An application of these