Agglutination and Adaptation. 9781463221812

Edwin W. Fay uses the process of agglutination and adaptation to explain the base patterns of a variety of languages and

184 106 5MB

English Pages 65 [69] Year 2009

Report DMCA / Copyright


Polecaj historie

Agglutination and Adaptation.

Citation preview

utination and Adaptation.

A n a l e c t a Gorgiana

343 Series Editor George Anton Kiraz

Analecta Gorgiana is a collection of long essays and


monographs which are consistently cited by modern scholars but previously difficult to find because of their original appearance in obscure publications. Carefully selected by a team of scholars based on their relevance to modern scholarship, these essays can now be fully utili2ed by scholars and proudly owned by libraries.

Agglutination and Adaptation.

Edwin Fay

gorgias press 2009

Gorgias Press LLC, 180 Centennial Ave., Piscataway, NJ, 08854, USA Copyright © 2009 by Gorgias Press LLC Originally published in All rights reserved under International and Pan-American Copyright Conventions. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise without the prior written permission of Gorgias Press LLC. 2009 ^ &

1 ISBN 978-1-60724-597-1

ISSN 1935-6854

Extract from The American journal of Philology 15,16,17 (1894,1895,1896)

Printed in the LTnited States of America




V O L . X V , 4.


I—AGGLUTINATION AND ADAPTATION.1 I. For a lustrum or two the science of linguistics has advanced on the hypothesis that there are no exceptions to phonetic law. As an a priori contention this is no better nor worse than all things a priori. Phonetic laws as we have them are the result of our own inductions ! The belief iij their inviolability depends on our granting a priori several impossible conditions. I can do no better than quote the words of Br6al on this point (Transac. Am. Phil. Assoc., 1893, p. 21): " T h e phonetic laws act blindly if we admit a set of conditions that are never realized anywhere; viz. a perfectly homogeneous population coming into no contact with the outside world, learning everything by living and oral tradition, without any books, without any monuments of religion,—a population in which every one should be of the same social condition, in which there should be no differences of rank, of learning, nor even of age or sex." Not but that exceptions to phonetic laws are granted: analogy is allowed to be a centrifugal force to this centripetal influence. Dialect variation is called into play also to explain differences of phonetic treatment. It were easier, in my opinion, to allow phonetic variation in many cases than dialect mixture, as in Lat. bovern for a theoretical *vovem. 1A brief abstract of part of this study was read before the American Oriental Society in New York (March 31, 1894); cf. Proceedings (1894, p. cxl). On my return from New York I found in Bezzenberger's Beiträge, X X , p. 81 sq., an article on the Sk. dat. by Johannson which has a certain likeness to my own speculations (cf. infra, p. 425).






I especially note as an objection to a rigid belief in phonetic inviolability t w o p o i n t s : i s t . W i t h i n the same person's l a n g u a g e two forms o f expression or pronunciation sometimes obtain. T y p i c a l for this is C i c e r o ' s use in early life of abs te, in later life of á ie. In the p a s s a g e of w o r d s from one p h o n e t i c value to another there must a l w a y s be a longish period w h e n both forms obtain, a n d both f o r m s m a y indefinitely persist in the same dialect, and be finally a d a p t e d to different uses. T h e R o m a n g r a m m a r i a n s had a clear tradition of w o r d s that retained a prerhotacistic -i. T h e doublet quaero || quaeso used to b e e x p l a i n e d in this w a y , and possibly this was a correcter view than the interpretation from *quaesso} 2d. Linguistic science has failed to note the importance o f the difference between familiar and unusual w o r d s , in r e g a r d of their phonetic treatment (cf. Primer, A . J. P. I I , p. 201). I refer at this point particularly to W h i t n e y ' s ' E x a m p l e s o f S p o r a d i c and Partial Phonetic C h a n g e in E n g l i s h ' ( I F . I V , p. 32 sq.). T a r b e l l ( T r a n s a c . A m . Phil. A s s o c . , 1886, p. 1 sq.) w a s the first to raise the objections noted b y W h i t n e y . T h e e x a m p l e s p r e s e n t e d there o u g h t to be c o n c l u s i v e : an ounce of fact is w o r t h a pound of theory. T h e s e points m i g h t be indefinitely increased b y insisting on the categories of h a l l o w e d w o r d s (alluded to b y B r é a l in the p a s s a g e j u s t cited), technical w o r d s , differentiation of sense c o u p l i n g with phonetic differentiation, etc. T h e delicate interplay o f analogies is also not sufficiently r e c k o n e d with. T h e source of analogical influence m a y itself h a v e been lost, or a subsequent d i v e r g e n c e of m e a n i n g m a y p r e v e n t our e v e r tracing the a n a l o g y . Thus u n d e r a too rigid construction of p h o n e t i c law the linguist will k e e p separate what o u g h t to b e b r o u g h t together. When D i a l e c t s must e v e n t u a l l y root in individual variations. s u c h a r e fostered b y g e o g r a p h i c a l separation, dialect ensues. C a s t e distinctions are as potent as g e o g r a p h i c a l remoteness to 1 A s long as there is no good etymology of quaero, I propose the f o l l o w i n g : quae-sivi is, l i k e po-sivi, a compound, and meant originally ' put-whats.' The original compound started possibly with quaestor, as a contemptuous designation of a person a l w a y s a s k i n g questions. W e could then see in quae-sivi the source of fel-ivi and lacess-lvi, a n d , subsequently, audlvi. I note the E n g l i s h w o r d ' q u i d n u n c ' a s comparable in m e a n i n g ; comparable formations are ' n o n p l u s , ' ' w h a t - n o t . ' W h o shall say that b e h i n d the S k . roots I . i t ' c l a s s i f y , punish,' 2. ci ' o b s e r v e , ' at ' p e r c e i v e ' and cint ' t h i n k , ' a pronominal stem *qi does not lurk with a primary sense ' to ask w h y , i n v e s t i g a t e ' ?





keep alive dialect in the same district. T h e caste distinction may pass away and leave no trace but the survival of some word in an unusual phonetic value. Caste may be of no wider extent than a single family. 1 LINGUISTIC SCIENCE U N D E R T H E REIGN OF PHONETIC L A W . —

This is a seductive working theory. T h u s the science becomes an exact science with sharp critical possibilities. T h e analogies of linguistics are not, however, with the exact, but with the natural sciences. In biology one must reckon with variation from type quite as much as with conservation of type. T h e greatest trouble with the results of linguistic science up to this time is that they do not harmonize. Schrader's Urgeschichte has demonstrated for the Aryans a meagre civilization.2 Brugmann's Grundriss, on the other hand, gives them an extremely high development of language. These results are irreconcilable to a degree. Grammatical potentialities greater than the Greeks had are an inconceivable possession for a primitive and unlettered people. A primitive people must have had a primitive language. I illustrate from 'mixed cases': the Greek genitive, for instance, is regarded as a sarcophagus in which bones of dead cases repose. From the biological standpoint it ought to be regarded as the representative of an undifferentiated embryo out of which the differentiated cases have come. Quite early in the study of Aryan linguistics under the influence of the Sanskrit Dhatupatha (Root-Book), words were regarded as developments of primitive monosyllables called roots. This tenet has been of late years called in question, and the claim is specious enough that the sentence and not the word is the unit of expression. But if, as the biologist sees in the embryo the traces of previous development, the linguist can see in a child's language a repetition of primitive conditions, then a little child can teach us that the monosyllabic word, excluding mere reduplicated cries like mama, which is the first stage, comes before the dissyllabic word, and before the sentence as well. 1 1 am acquainted, for instance, with a family that has maintained the tradition of ' w o u n d ' for'woond,' in spite of all their neighbors. It m a y b e answered t h a t ' w o o n d ' is an Irish pronunciation re-introduced into English by the Duke of Wellington (for there is such a story), but the illustration serves to show that two pronunciations may both be current in the same geographical and social environment. 2 For a convenient summary I refer to Clark's Manual of Linguistics, p. xxiii sq.






G r a n t e d our m o n o s y l l a b i c words, w h e t h e r imitative (' b o w - w o w ' ) or interjectional ( ' p o o h - p o o h !'), or r e v e r b a t o r y like ouch! ( ' d i n g d o n g ' ) , the p a s s a g e to dissyllables remains still to be traced. If a n y considerable n u m b e r of the original s t o c k of w o r d s w a s m o n o s y l l a b i c , then dissyllables must h a v e c o m e b y composition or agglutination. D e l b r ü c k (Einleitung 3 , p. H I ) thus formulates the result o f his critique of the t h e o r y of a g g l u t i n a t i o n : " A u c h jetzt noch k ö n n e n wir nichts weiteres behaupten als w a s o b e n b e h a u p t e t w u r d e , dass das Princip der A g g l u t i n a t i o n das einzige sei, w e l c h e s uns eine verständliche E r k l ä r u n g der F o r m e n g e w ä h r t . " I now u n d e r t a k e to locate s o m e o f t h e a g g l u t i n a t i v e processes o f the A r y a n l a n g u a g e out of which the inflections o f the d e r i v e d l a n g u a g e s d e v e l o p e d . PRONOMINAL STEMS.—I note first that the A r y a n s possessed a considerable n u m b e r o f pronominal stems m a d e up, in the main, of stop-consonants plus a vowel, 1 or of a v o w e l alone. I note t h e following, using a as a s y m b o l for a v o w e l undifferentiated b e t w e e n a, e, o2: a-, ta-, ka-, sa-,ya-, Wa-, na-, r^a-, a n d these w e r e further c o m b i n e d with one another into g r o u p s , t h u s : a-ia, a-sa etc., a-ra, fya-,




tr%a-, ta-r^a-





o f these stems will display t h e m s e l v e s later on. T h e y were at first o f v e r y free e m p l o y m e n t , a sort of uninflected interjection, a c c o m p a n i e d doubtless b y g e s t u r e . In addition to these interjections w e r e a class of m o n o s y l l a b l e s t o w h i c h more definite m e a n i n g had been attached. Into their further e m b r y o g e n y I p r o p o s e to g o in a second essay. I a s s u m e as s u c h early A r y a n m o n o s y l l a b l e s bhar and ad, m e a n i n g respect i v e l y ' b e a r i n g ' and 'eating.' T h e y were originally neither v e r b s nor nouns, but a m o r p h i c centres out of which v e r b s a n d nouns e q u a l l y d e v e l o p e d , such as w e call action-nouns, but with a n 1 T h i s phenomenon meets us in almost every other language as well. I refer to the paper of D r . Brinton, the A m e r i c a n ethnologist and linguist, in the Proc. of the A m . Or. Soc. for 1894, p. cxxxiii. 2 1 put myself on the footing of Merlo's essay : " R a g i o n e del permanere dell' A e del suo mutarsi in E (O) fin dall' età p r o t o a r i a n a " ; that is to say, though è and possibly ö had developed by the end of the A r y a n period out of ä, there was a time w h e n many roots that now appear with / f ö had ä as their original vowel. T h e r e is no intrinsic improbability in ascribing àya and ISa to roots originally ag- and ad-, and the reason for the change to ed- before the close of the primitive period is to be sought, in my opinion, in the consonantal environment (infra, p. 425 sq.). In constructed forms showing this a I shall f e e l at liberty to omit the construction-symbol (*).





element o f agent-nouns, and directly c o m p a r a b l e with the E n g l i s h w o r d s I h a v e used in translating t h e m . IMPERATIVE-VOCATIVE.—Starting f r o m s u c h bases as actual w o r d s of speech, w e m a y posit as their simplest inflectional forms bhar-a ( G r k . ep-e) and ad-a ( L a t . ed-e). T h e s e f o r m s are d u b i o u s as soon as w e c o m e to their interpretation. T h e y m a y be imperatives o r t h e y m a y be vocatives. T h e form is o n e ; the d e v e l o p m e n t of m e a n i n g is subsequent. H e r e a g a i n w e m u s t r e g a r d the differentiated as later than the undifferentiated. The v o c a t i v e is b y nature an imperative, and this force m a y be felt even now, after l a n g u a g e has been h i g h l y d e v e l o p e d . When l a n g u a g e was in e m b r y o , the probability o f this relation must h a v e been m u c h stronger. T h e elliptical in l a n g u a g e t o - d a y is a continuant of the elliptical in primitive l a n g u a g e , and the primitive man was doubtless in addition m u c h m o r e gesticulative. If one's c a r r i a g e is in waiting before the house, but not directly at the step, the call ' D r i v e r ' is e n o u g h to h a v e it b r o u g h t to the step. In the primitive period, before individual n a m e s had d e v e l o p e d , it w o u l d not be necessary to call to a h e r d s m a n ' D r i v e r , drive,' for ' D r i v e r ' was s u m m o n s e n o u g h . N o little child misunderstands t o - d a y if her father s a y s ' C o m e , m y slipper-bringer,' or ' W h e r e ' s m y s l i p p e r - b r i n g e r ? ' T h e c o m m a n d is inferred e v e n without the father's pointing to his slippers. Cries like ' F i r e ! ' ' M u r d e r !' ' Police !' s p e a k volumes, a n d ' Police !' is certainly a vocative. ' W a i t e r , two s a n d w i c h e s ' is h a r d l y an ellipsis: when actor and action w e r e fused in o n e m o n o s y l l a b l e , w h e n bhar meant ' b e a r i n g ' and ' b e a r e r , ' to say bhar to a person w h o was bhar and point at a thing was c o m m a n d e n o u g h . F o r the identification of impv. a n d voc., a n d t h e subsequent d e v e l o p m e n t o f the v e r b from the impv., the c o m m o n enclisis o f v o c . and principal v e r b s p e a k s most s t r o n g l y (cf. infra, p. 416). A s a first enlargement, then, of bhar ' bearer, bearing,' I r e g a r d bhar-a, and s e e in the -a a demonstrative, a suffixless interjection, m e a n i n g ' n o w , ' ' h e r e , ' w h i c h is still p r e s e r v e d in G r e e k e-neivos, ¿•Hi,1 L a t . e-quidem, etc. A combination bhar-a w o u l d thus mean ' Bearer, here,' a n d the s u m m o n s was equivalent to a c o m m a n d to b e a r ; but bhar-a w a s doubtless liable also to the interpretation ' this o n e bears,' for -a has b e e n retained in its 3d personal signification in the perf. 3d sg.; S k . ved-a, G r k . foifi-t. T h i s was 1 Very probably, too, in the interjection it, where the reduplication gave an especial emotional tone.






termination enough where but two persons were concerned, but given a speaker and two others, inferiors, and a further specification was necessary. Here it was necessary to call in another set of demonstratives, sa and ta, the former being used perhaps of a nearer and the latter of a remoter person addressed. These forms as thus employed are directly comparable in use with iste and ille in Latin. In Vedic these stems are used of all persons. To the stem bhar- we thus get the groups bhar-Sa 'bearing this one' and bhar-ta 'bearing that one,' and to the stem bhar-a- the groups bhar-a-sa bearing here this one,' bhar-a-ta 'bearing here that one.' Another element of enlargement to the verb comes from the pronominal stem of the ist person, m-a-1| a-m-.1 When this is added to the stem bhar- we get a form — 'bearing I.' The forms bhar-a-sa and bhar-a-ta in collocations with words of vowel initials were elided to bhar-a-s and bhar-a-i. I say elision, for this is altogether a simpler and more natural process than the mystical one of gradation for earlier linguistic stages.2 COMBINATIONS OF D E M O N S T R A T I V E S . — I n Lithuanian sztái we are taught to see a combination of two demonstrative stems, viz. ko and to (Brugmann, Gr. II, §409). A similar phenomenon is doubtless to be seen in Latin i-s-te. The Sanskrit representative of i(J)te reverses the order of the two stems, viz. syá-, tyá-, 1 I n E n g l i s h the commonest word for the 1st person is a nasal grunt represented b y such spellings as humph, ugh, etc. It is properly a vocalic m, and is an interjection of the 1st person. Some one m a k e s a r e m a r k : I grunt in r e p l y m\ it means that I am listening. I am asked a question: I answer m, and this m, with a rising inflexion, expresses surprise or interrogation: I express assent by a double m h m, protracting and accenting the final m. I express dissent by protracting and accenting the first m. T h e A r y a n s had all these uses, for this nasal grunt is at the base of Grk. /J^, vr¡°, Sk. ma, net, L a t . ne negative, Grk. vi] (vai), L a t . ne affirmative, Grk. ¡ia affirmative or negative (reinforced by vai or ov), L a t . -ne interrogative. I n this grunt I would find the origin of the i s t person pronoun stem m-a-. It is to-day, when emphatic, prefaced or followed by an h that is almost sonant, as witness the popular orthography of humph and ugh. Is this sound identical, perhaps, with Sk. h in aham ' I ' ? 2 T h a t gradation as a conscious mode of form-making had been developed before the close of the A r y a n period is perhaps indubitable, but only three propositions seem to me reasonable and provable in this regard as to the e/a series: i s t . e is accentual, 2d. 0 is post-accentual, and 3d. complete disappearance of the vowel is pre-accentual. T h e s e formulae do not sufficiently explain the words of w h i c h Grk. 6poc is the type (infra, p. 426); and I am not at all certain that the e/o variation is an accentual phenomenon.





T h e s e stems were also A r y a n , as will be presently shown. We thus have the triplet bhar-Sa, bhar-a-sa, and bhar-a-sya ; bhar-ta, etc. GEN.-[ABL.]










remember, now, that we are dealing with nominal concepts undifferentiated between action and agency, we are entitled to assume a sentence of the following type, paratactic and without a copulative v e r b 1 : bhar-a-s' ad-a-sa, primitively 'bearer this, eater this,' or ' bearing this, eating this.' Now, the potentialities of hypotactic meaning resulting from this collocation are numerous. 2 1st. ' T h o u bearest, thou eatest„' which passes into ' T h o u , the bearer, eatest.' Here bhar-a-s is, speaking anachronistically, an ¿-stem 3 ; 2d. ' < w h a t > thou bearest thou eatest,' whence the subsequent -esstems were d e v e l o p e d ; 3d. in certain cases the shading became ' < a f t e r > bearing thou eatest,' then ' < f r o m > bearing thou eatest,' or ' < o f w h a t > thou bearest thou eatest,' and thus the gen.-abl. was developed. Simpler than bhar-a-s ad-a-sa we may assume bhar-s ad-a-sa, taking bhar-s after the 1st interpretation. Thus i ° bhar-s || bhar-a-s are active (nomina agentis)-, 2° bhar-as- is passive (nomen actionis), and 3 0 bhar-as4, is a gen.-abl. to a nomen agentis. T h i s t y p e we can illustrate ( i ° ) in S k . vac, L a t . voc-s, Lat. °zIOCO-S ' s p e a k i n g ' ; (2°) S k . vac-as-, G r k . Fin-os- ' s p o k e n ' ; ( 3 0 ) Lat. voc-is, S k . vac-as ' o f the speaking.' A confusion of active and passive stems is seen in G r k . ayyeXo-j beside S k . dngiras-. W e are prepared, after what has been origin of the other genitives in this same °sya yag-a-sa ' o f what thou bearest, thou instance of how they might be in actual

said, to recognize the group. In bhar-a-sa || sacrificest' we see an use. In bhar-a-sa we

1 T h e copulative verb has no warrant to pass for extremely primitive. The earliest literatures could always dispense with it altogether! 2 T o justify the shifts of meaning out of which the various case uses are developed, I compare the absolute constructions in the individual languages: Sk. loc. and instrum. absolute, G r k . gen. and acc., L a t i n abl., etc. T h e s e are all remnants of the paratactic stage of linguistic development, and represent original verb-nouns. Caesare duce (infra, p. 416 n. 2, for the ) Dixit may well be a development from *Caesar ducet vivit' Caesar rules, that one lives.' 3 T h i s is Streitberg's nomenclature for the usual ' o-stem,'and seems to me very convenient. 4 W e shall presently recognize in this ending (infra, p. 418) what I call a ' t h e m a t i c ' ending. I had independently recognized this valuation and adopted this terminology before being aware that Streitberg, in I F . I 91, had preceded me by some years.






have the Aryan beginning out of which the Germanic genitives sprang (Brugmann, Gr. II, §228), while in bhar-a-sya we have the prototype of the Greek and Sanskrit forms.1 ABL.-[GEN.] = 3D SG.—bhar-a-t ad-a-ta or bhar-t ad-a-ta = 1st. 'he bears, he eats,' 2d. ' < w h a t > he bears he eats,' and 3d. ' < o f w h a t > he bears he eats.' Typical of the xst value are such words as Grk. 6tj-s, gen. 6r)-r-6s 'workman,' Lat. sacer-dd-s, °dd-t-is 'sacrifice-doer,' tege-s teg-e-t-is 'cover-ing.' Of nouns with passive meaning, few examples are quotable. In Greek a-yva-i, gen. a-yva-T-as 'ignorant,' 'unknown' both active and passive senses obtain. Lat. seg-e-s, gen. seg-et-is 'field,' 'crop' seems also to contain both senses. T h e survival of this suffix as abl.-[gen.] is plainly to be seen in Avestan yimap beside S k . y a m a d (cf. Bartholomae, Altir. Dial., §238), and perhaps in Latin modd(d), bene(d), etc., which Brugmann explains as instrumentals (Gr. II, § 2 75)- 2 I n Sanskrit also this ending is preserved in such words as dev-di-as, where in -at- we are to see one abl. sign and in -as another, as in Latin fund-il-us; but dev-di-as was perhaps felt as a 3d pers. stem (devdt-as), like seget-is. That this suffix had also, at least in its dithematic form (cf. infra), the value of a genitive is seen from Lith. vilko, O.Bulg. vluka 'lupi' (Brugmann, Gr. II, §228). W e are not to regard this as a confusion of a differentiated genitive and ablative, but as a survival from their undifferentiated state. THE ENCLISIS OF THE VOC.-IMPV.—We can assume, however,

a still more primitive sentence bh&ra yaga,

subordinated under

-oio from -osyo, a n d those -oso. I t c a n n o t b e proved that reXeio | reMo are e a r l y a n d late forms of *Ttktayo-. T h e a d j e c t i v e s TS'AUOQ | TDIEO( m a y b e l o n g to an ¿-stem just CEt0 as w e l l a s the pair xpi' S II Xf""!EOi- ( J o h a n n s o n also to the same effect, B B . X X , p. 1 0 0 , note.) It has not b e e n proved that the difference b e t w e e n 6eo( 1

G r e e k had p e r h a p s both t y p e s : the genitives in




a n d SeloQ is a n y t h i n g more than orthographic, with the passing v o w e l b e t w e e n e a n d 0 indicated

in one case a n d not i n d i c a t e d in the other.

I t is m e r e

h e d g i n g with the phonetic l a w s to write TroZof out of *TTOI-IOC, as B r u g m a n n does.

O n e is e v e r y w a y justified in s e e i n g in

m a d e nominative, like the L a t i n pronoun V I 4 3 1 sq.), from the genitive. is from 2




a g e n . *N010 out of


(cf. K i r k l a n d in C l a s s . R e v .

I ask if the possibility is e x c l u d e d that

*qud-zyo- ?

h a v e w r i t t e n the abl. sign i n d i s c r i m i n a t e l y -1 or -d.


O f course t a n d d

are but v a r y i n g forms of the dental stop-consonant, due only to their e n v i r o n ment as


T h e setting aside of -t for v e r b a l a n d of -d for n o m i n a l use

w a s m e r e l y a later adaptation. below, p. 4 2 1 .

F o r a different interpretation of

bene see





one accent (cf. Wh. 2 , 314 d ) when equal to two vocatives in apposition, thus: 'beärer-sacrificer.' But as soon as differentiation of verb from noun sets in, then both words are accented, bearer, sacrifice.

I c o m p a r e S k . agrutkarna


h&vam ' O -

thotì-of-listening-ears, hear our call ' (Wh. 2 , 594 a). If, however, both words have verb value, as in Pàpa, aga yäga 'Papa, come sacrifice,' it is the second that receives the accent. I cite asmdbhyamje§iyótsi ca 'for üs conquer and fight.'1 T h e reason for the accentual treatment in the last case is obvious. T h e second impv. is not to be considered initial in a new sentence, as Delbrück takes it, after the native grammarians ( S F . V , §23, 1), but is to be considered as dependent on the ist impv.; thus, 'come and sacrifice' = 'come to sacrifice' nearly. T h e approximation of 'and' to a result-purpose particle, as in the English doublet ' g o and see' || ' g o to see,' is shown in Greek also. Compare Horn, a 284 A0è Kai e'lpeo ' g o and ask' with the normal construction as in Ar. Ran. 326 «'Atfè ^optio-ai/.2 1 W e are able to get a glimpse here of the reason for the accentuations C O T E , èldé, laße in Homer. Their accent came from their use as the second of a pair of imperatives. W e are justified in reconstructing a pair àye eliré (so far as .the order is concerned) by y 332 ( à l l ' aye ràfivere), 8 14g ( a l l ' àye Keipqoai). In the Odyssey Homer has elsewhere the order àye -(-intervening words impv. There are, however, five cases of ein-' àye (0 347, ip 261, T 192, I 673, K 544), a position doubtless due to metrical considerations, àye . . . e'm'e occurs nineteen times. Homer uses èldé five times with impv. or infin. ( = impv.), once in the order of "t 770 nlvBi, Bea, àyadi] F101 èmppodof; èldè KOÓOÌIV, and uses laßi three times with another impv. (infin.), once in the phrase 7rapéf£o Kai laße (A 407). T h e phrase àye èWé is to be inferred for Greek from Homer's àye .. . elaeWe (w 25), àye . .. loßev (p 190). By Sk. accentual laws a phrase *à?X àye laße would accent only \aßk of the two impvs. In Grk. aye laße the accent of aye is the secondary substitute for enclisis, according to Wackernagel's famous law ( K Z . X X I I I , p. 457 ; Bloomfield, A . J. P. I V , p. 21), whereas laße would represent the original Aryan accent of a thematic aorist impv. (cf. laßüv), when it follows directly another impv. (cf. yótsi in the Sk. example cited above). 2 Connection can possibly be established by this line of reasoning between Sk. utd ' a n d ' and Lat. ut of purpose-result. Ut belonging to tita, fell into confusion with the rei. *quut, *cut (cf. Sk. kü-tra) J *put (?) in the Italic period, whence the loss of qu-, c- \f- in uter, ubi, etc. In the Vedas utd seems a simple ' and,' but in Brähmana it is about equivalent to itaque, and amounts to the affirmation of a result ; it is construed with the opv. as well as the indie. While itaque never became a particle of result, the precisely equivalent particle &are reached that stage.






THEMATIC AND DITHEMATIC.—Grammatical t e r m i n o l o g y h a s

heretofore practically restricted the term 'thematic' 1 to verbs. After the reduction of verb and noun to a common basis, it is necessary to extend the terminology to nouns also. If now in Avestan yim-a-p we see a thematic abl., then for Sk. yam-lid I propose the term 'dithematic.' 2 The origin of the dithematic forms was this : to a stem capable of functioning alone as a nominative, there was doubtless a 2d pers. nom. in -s, and also a 2d pers. thematic nom. in -a-s; thus to bhar the forms bhar-s and bkar-a-s. There was also a voc.-nom.3 bhar-a-. Now, when in the upgrowth of inflection bhar-a was conceived as a stem and -as was conceived as an ending, by the syncretism of the two there resulted -as.* For the ablative likewise there was a form -at. THE DISTINCTION OF NUMBER.—No language has ever entirely differentiated singular and plural in the 2d person. Our English you is in line with linguistic phenomena all over the Aryan field. It is fair to extend this fact to the primitive period. Thus in our gen.-abl. form bhar-a-s, which we saw was also a 2d pers. nom. sg., we may see the nom. plur. bhar-as\ and as this bhar-as (conceived finally as a stem) was used as object in the sg. (Grk. ytvos), so, doubtless, it was conceived as object in the plural. Jn the 3d declension nom.-acc. -esb of Latin we may see an example of this in a dithematic form. A c c . SG. = IST PERS.—I now take for illustration the sentence bhar-am, ad-am ' I bear, I eat.' This comes to mean, ist. ' I bearing eat,' 2d. ' < W h a t > I bear I eat,' 3d. ' < O f what> I bear I eat.' The manifest accusative of 2° will be clear to all. In 3° no vital genitive survives of the thematic form, but the di-thematic form is to our hand as the Aryan GEN. PLUR. in -dm. In agentx Streitberg seems to use the term ' t h e m a t i c ' freely, as in I F . I, p. 91, but Johannson (BB. X X , p. 100) prints with inverted commas ' thematische.' Brugmann (Gr. Gr, 2 , p. 91) speaks of -0- as a nominal suffix, but so far as I know does not recognize the division of noun-stems as I suggest. 2 T h i s term has been used already of the rjn-stems, know, of the endings with long thematic vowel.

but not, so far as I

3 Represented by Grk. Imz6ra, etc., in Homer, and possibly also in made of the Latin proverb made virtute esto, which may be, however, for macte virtute (supra, p. 416), that is to say, an abl. of quality. 4 Such a dithematic nom. sg. is, in my opinion, the of Greek -ef-stems. I note especially Sk. dngiras to Grk. ayyeXot;. ' E x p l a i n e d by Brugmann (Gr. I I , §325, 1) as being derived from -ens, spite of Latin ensis from *nsis.



t h e f o r m i ° s e e m s o n t h e f a c e o f it n o t

A s to nominatives

l i k e bkar-a-s

















to h a v e

s t e m s l i k e bha.r-a.s-, bhar-at-








a n d bhar-am-, -m-a-





and to nomi-

and so we



we may





( B r u g m a n n , Gr. II, §72). DEVELOPMENT indubitable


PTC. S T E M IN -ant.—It

that, g i v e n a s t e m

i t a b y e - f o r m bkar-an



explanation the



pers., infra, p. 432). bhav-a-tas parable

we with

have the


have to



is 2 d It



pers. and by

an A r y a n


(nom. sg.).












1 s t p e r s . (? o r 3 d



m a y , instead, b e

(supra, p. 416).


W e thus reach





as the p r o d u c t of sentence euphony.1

i n i ° w e c a n s e e t h e S k . p t c . bhdr-an bhavan:


we should



Sk. com-



1 B r u g m a n n (Gr. I I , §325) implies that the group -ms in A r y a n w o u l d be p e r m a n e n t and not b e c o m e -ns, on the ground, I infer, of certain B a l t i c forms in -ms, -mt. I n this I c a n n o t b e l i e v e he is right. T h e persistence of -ms in the L i t h . dat. plur. kurems, for instance, must h a n g with the dat. sg. kurie'm, ins. sg. kuHu-mi. Spite of the loss of the L i t h . correspondent of L a t i n decimus w e are to see in deszimtas ' tenth,' deszimt' t e n ' the influence of A r y a n *dekmmo-s. B e c a u s e of L a t . decimus, S k . da^amd-, w e cannot conclusively set up an A r y a n form *dekm-to. G r k . Senaro^, Goth, taihunda, etc., m a y w e l l h a v e b e e n c a l l e d separately into b e i n g from EKTOI, saihsta, just as S k . saptdthas, O H G . sibun-to probably were. B r u g m a n n (Gr. I I , §186) cites O.Prus. deiwans (deos) as proof that the A r y a n a c c . plur. was -ns and not -ms. E v e n granting the v a l i d i t y of the contention that B a l t i c -ms represents A r y a n -ms, this example w i l l prove nothing. T h e B a l t i c paradigm w o u l d h a v e had an acc. sg. represented b y deiwan (Brug., G r . I, §217), b e s i d e w h i c h a plur. *deiwam-s could hardly h a v e b e e n m a i n t a i n e d . T h i s can be p r o v e d b y Pruss. mans 'nos.' I n the A r y a n acc. ns (Goth, uns) I see ms; there w a s also an a c c e n t e d form ma-s, doubtless. O u t of i n t e r p l a y b e t w e e n ma-s and ns a b y e - f o r m na-ns d e v e l o p e d , w h e n c e a plur. stem na- w a s abstracted. Pruss. mans represents a still more primitive syncretism. N o t h i n g more definite than m/n ought to b e written for a final nasal in the A r y a n period. T h i s w o u l d save a good deal of analogical j u g g l i n g about novem, for instance. 2 T h e s t r o n g stem has p e n e t r a t e d from the nom. into all but the I n d i r a n i c group, the C e l t i c , and possibly the Italic (but here the influence of the g e r u n dive ndo, cf. A . J. P. X V , p. 317 sq., can account for the vocalism). Proof is furnished b y O.Irish car-it (dat.), car-at (gen.), w h i c h m a y represent a n original -ct or -ot. A d d i t i o n a l proof is furnished b y G r k . E'ISOT-OQ, S k . vidvdt-, for u n d e r a n y theory of the perf. s u f f i x — S c h m i d t ' s van-s or B r u g m a n n ' s vas (CI. R e v . V I I I , 455 n. 2 ) — w e have to seek an analogical source for -vat-, and that source w a s doubtless the pres. ptc. w i t h a stem-form still represented in Irish -at-.






the -t- out of bhavatas over the rest of the inflection is simple enough. In this way rajan- ' k i n g ' is to be interpreted as an older form of raj-ant- 'ruling.' If, however, we t a k e a ist-person form bhar-am with euphonic bye-form bhar-an, it was liable to be made 2d person bhar-an-s, or 3d person bha.r-a.n-t-, thus in S k . raja (nom. sg.) ' k i n g ' may lurk a 2d pers. *rdjanz, from *rajan-s, and in rajan ' ruling' a 3d pers. rajan-t. THE INSTRUMENTAL SG.—I resume my sentence in the form bha.r-m adat11 bear, he eats.' T h i s is liable to the interpretation ' b y my bearing he eats.' W i t h consonantal stems Brugmann's argument for an instrumental suffix -a rests on forms like S k . prati-bhidy-a 'with splitting,' G r e e k prepositions like ncSd 'with,' and Latin ped-e 'with the foot.' E v e r y one of these forms may represent ah A r y a n -rti} T h e instrum. ending -mi of the Baltic languages speaks for this conception, and is to be equated in the verb with the primary 1st sg. -mi. It remains to discuss the instrumentals to e-stems of which S k . vrka is a type. I note, in the first place, that if the assumption of a suffix -m above is right, and of a suffix -mi — ist-pers. vb. suffixes, then we may see in vrka the correspondent of the 1st pers. in -0, e. g. ep-a>.2 Another explanation of this case involves no phonetic difficulties. W e know that the t y p e bkar-aSa is impv. in G r e e k (cf. itreo) in its verb-function, and in Latin impv. or indie, (cf. sequere). T h e t y p e bhar-a-ta is indie, ( ' m i d d l e ' ) in G r e e k (fa'pero), impv. in the S k . 'injunctive' (bhdrata), and impv. in Latin (tegito 3 ). T h e n 1 C e r t a i n l y in such early Latin as the epitaphs of the Scipios such forms as omne for omnem, aide for aidem appear. It may w e l l be that -e is -e and the normal representative of final m. T h e n in omnem w e are to see a restoration from domom, etc. In decern the force of compounds like decemviri accounts for the form. It cannot be denied, however, that 1st- and 2d-declension accusatives also lose the final m. It is not necessary to explain pede as instrum. of the 1st person. It may w e l l be instrum. of the 3d, d e v e l o p i n g from a sentence bhar-ad adasa ' he bears, thou eatest,' which g i v e s ' by his b e a r i n g thou eatest.' 2 T h e relation of -0 to -om, primary and secondary 1st sg., has not been explained. C a n w e c o n c e i v e of -0 as -g arising from -om in certain cases of sentence e u p h o n y ? T h i s seems to me the interpretation of H o m e r i c $Z> •home,' which I take to be for dom-, a neuter non-thematic stem. N o t e the suffixless A v e s t . loc. dam. A n o t h e r explanation is given below (p. 421). 3 1 am not oblivious of the form tegitod, but I regard -itod as syncretic, just as the abl. -itos (fund-it-os). B a c k of tegitod lie three f o r m s : *teget\*tegod (dithem.) and *tegeto\ tegitod is *tegitd reinforced from *tegdd, or, more simply, the failing abl. sense of tegito was reinforced, and the result was tegitd-d.

A GGL UTINA in the doublet bhar-a for bhar-a



|| bha.r-a.-sa

(supra, p. 4 1 3 ) .





w e can infer the indie, function

In a sentence



bearest, I eat,' the sense ' b y t h y bearing I e a t ' c o m e s



T h u s in the instrum. -a w e m a y see a dithematic continuant of bhar-a

in the typical sentence g i v e n .

T h e r e is also no reason w h y in s u c h L a t i n a d v e r b s as bene should not see this earliest extension of the "stem. then explain the vocalization of bene




: bonos b y r e g a r d i n g bon- as

infected from -os (infra, p. 4 2 6 ) . A n o t h e r source from w h i c h the instrum. m a y s p r i n g is the 1 s t s g . -âu || -a, and the loss of all trace of ûu m a y be d u e to differentiation from the dual (infra, p. 4 2 9 ) . T h e r e is also an instrumental suffix in -bhi, P 477).

G r k . - thou eatest,' i. e. a gen., or ' by my bearing thou eatest,' i. e. an 1 O n the origin of these infinitives and the nature of the final diphthong, I refer to my note on the Lat. gerundive (A. J. P. X V , p. 317). 2Cf.

Sk. as-ail, Grk. av-re, etc., infra, p. 429.





instrum. A s a relic of the gen.-dat. w e must r e g a r d S k . forms in -ayai ( W h . 2 , 365 d) ; the G r k . dat. is really a dat.-instrum. T h u s , o n c e a g a i n the ' m i x e d ' case p r e c e d e d the differentiated. In consideration of the Indiranic d a t i v e s in aya, for which I g i v e below a specific explanation as s e c o n d a r y , w e cannot e x c l u d e the possibility that the dative - a y is related to -aya in just the s a m e w a y that the g e n . -as is related to -aSa. It d o e s not seem to me, h o w e v e r , that b y this e x p l a n a t i o n w e are b r o u g h t so close to the solution of the optative p r o b l e m w h i c h is, I believe, b o u n d u p in the c o m p l e x of dat. = 1st sg. mid. (cf. infra, p. 439). Johannson, in B B . X X , p. 98, has indeed already seen in S k . -aya an A r y a n -o-yo.

I h a v e now d e v e l o p e d the typical A r y a n cases for the s g . remains to s p e a k of s o m e individual p h e n o m e n a in S a n s k r i t . T H E SK. DATIVE IN -aya.—The *devai

|| *devay;

S k . dative is

t h e h i s t o r i c a l f o r m devaya



is d u e t o t h e i n f l u e n c e

of the g e n . devdsya, and it is not i m p r o b a b l e that the g e n . before v o w e l s was elided in A r y a n to -sy', w h i c h would render this a n a l o g y easier. T H E S K . INSTRUM. IN -ena.—We are entitled to assume (supra, p. 420) that there was a S k . 1st pers. instrum. of the t y p e *dev-am thematic, or *dev-am dithematic, 1 and to assume the b y e - f o r m *devan in sentence e u p h o n y (supra, p. 4 1 9 n. 1). N o w this *devan m a y h a v e a d d e d once m o r e the consonantal stem-suffix a (from m), g i v i n g *devana\ cf. A v e s t . instrum. sg. ka-na and S k . a d v e r bial forms like cand. In devtna w e can explain -na in the w a y s u g g e s t e d , and see in deve- a thematic dat.-instrum. (supra, p. 424), as in deve-bhyam (instrum. dual). T H E S K . GEN. PLUR. IN -anam.—Here I find a syncretic form, the result of the e u p h o n i c d o u b l e t *devam || devan. F r o m these sources -n- b e c a m e a regular inflective element in S k . C H A N G E S OF T H E V O W E L A; G E N D E R . — A c c o r d i n g t o t h e v i e w

stated a b o v e (footnote to p. 412), e a n d o d e v e l o p e d out of a primitive n, o w i n g to the consonantal environment. T h u s in a 2d pers. nom. to an ¿-stein w e s h o u l d e x p e c t -e-s, the dental v o w e l before the dental sibilant, a n d so in the 3d pers. abl. -e-t. In the 1st pers. nom.-acc. an original -a-m g a v e us o-m, a c h a n g e d u e to the r o u n d i n g of the lips preparatory to their closure for the -msound. T h u s , and not b y g r a d a t i o n , I w o u l d explain the vari' F o r instrumentals in -am cf. Brug., Gr. I I , §896, A n m . , and the articles there cited.






ation of e and o in masc. thematic inflexion. N o gradation theorywill genetically explain, for example, the -o-s of thematic genitives to monosyllables, which was always under the accent, and no gradation theory explains the invariable -e of the voc. sg., which was never under the a c c e n t ; no explanation of the. stems ped-1| pod- as due to gradation can pretend to be adequate. Particularly cogent for the view I have stated is the fact that in verb-inflexion the thematic vowel -o appears only before m (»). B y this view an explanation of the type opos is in our reach. T h e r e were conflicting ist and 2d pers. noms. in o-m and e-s. Out of this conflict came o-s; but this was not all: when *bher-e-s and *bker-o-m created the new t y p e *bher-o-s, the infection went further to *bhor-o-s. B y subsequent differentiation came accs. of the •ycVos-type (supra, p. 415) and noms. like op-os. T h i s explanation is obviously applicable to the doublet represented by the G r k . gen. woSos || Lat. pedis, i. e. *pedes.1 B y this explanation Lat. bene represents an older vocalization than bonos. T h e same principle of explanation is applicable to initial vowels and root-finals. L e t us take the tj ag ' drive.' Is there any phonetic reason w h y a was the vowel-shade in this root ? Y e s , a v e r y g o o d o n e : a and g are both gutturals. In the same w a y the V a d ' e a t ' became ed- b y assimilation. It is noticeable that in a l a n g u a g e as copious as the G r e e k there is no root-word beginning with iy-, for in ¿-yeip-a the so-called prothetic ¿- is the ' a u g m e n t , ' j u s t as in ¿-6ik- : fdeXa as in any w a y different from that of Ktivos : eWvos. T h e reason that in ¿8t\a> ' w i s h ' the e- became attached to the entire vb. was doubtless that it had a slight emotive force (cf. the interjection « ) , which helped the connection to be made. T h e use of the unaugmented past tenses in both H o m e r and the 1 I n Lat.pedem w e may see a thematic acc. affected b y the gen. *ped-es. Sk. pad-am may be also a thematic acc. persisting in n&n-thematic inflexion. T h u s in the L a t . instrum. pedi we can see -e = -»2 (supra, p. 420), and in Sk. pad-a (with secondary lengthening) -a, = -m. Apropos of G r k . jroiJtif, I suggest that the doublet ira; || irovg is based primarily, by external phonetic similarity, on the doublet /3«f J/3oic; oiioif ' t o o t h ' has doubtless affected the accent of vovf, while the accent of iraf (for v a c , Bloomfield, A . J. P. I X 15) shows the effect of /?£>f. T h e extension of the form ttovq was doubtless aided b y the other part of the body, btSoiig (Bloomfield, A . J. P. X I I 2). I note also the spelling oif : L a t . auris (cf. Mod. L a n g . Notes, I X , col. 262).





V e d a s as r e g u l a r pasts ( a l o n g with the ' i n j u n c t i v e ' m e a n i n g ) s h o w s that the a u g m e n t was not definitely g i v e n a past v a l u e till in the separate life of G r e e k a n d Sanskrit. T h e only other w o r d is iy-a, w h o s e e- is d u e to the e- in o t h e r w o r d s of the p a r a d i g m (cf. G r . fy6' äp' 'c/ie?/.ev (cited b y A u t e n r i e t h , 1. c. x ) : " in d e r C o m p o s i t i o n steht ro z w i s c h e n P r ä p o s u n d V e r b a l f o r m " ( W i n d . , 1. c. 8 ) ; n o t e /cor' äp s&to

( A . 6 8 ) : ro is u s e d a f t e r co-n ' in o r d e r

t h a t ' ( W i n d . , 1. c. 4), hp' is u s e d a f t e r c a u s a l p a r t i c l e s ( A u t e n . , 1. c. 3).


i d e n t i f i c a t i o n o f ro w i t h irp6, g o o d e n o u g h p h o n e t i c a l l y , suffers f r o m a l a c k o f a n a l o g y in its s y n t a c t i c a l v a l u e : w h e n w e c o n c e i v e ro as a d e m o n s t r a t i v e , t h e L a t . use o f iam ( G i l d e r s l e e v e ' s G r a m . , §230) as a t e n s e - f o r m i n g p a r t i c l e , a n d t h e S k . use o f sma ( W h . 2 , § 7 7 8 b a n d c) a r e c a s e s d i r e c t l y a n a l o g o u s . w i t h p r o n o u n s a n d c o n j u n c t i o n s , t h e e q u a t i o n w i t h ttp6 is w o r t h l e s s .

F o r ro






makes the Ir. gen. athar come from * pater os, or *patr-os, but, after all, that does not make the -ar clear, so far as I can see. I suggest instead that in the gen. mathar we have the same form as in Sk. mdtur. In both I see an Aryan -tr, not -tr-s, before vowels -trr (Gr. I, §285), whence Sk. -tur and Irish -tar (cf. ib., §298, 3). Thus in Ir. 3d sg. deponent -thar, and in Lat. 3d sg. dep. -tur, I would see an Aryan -trr, Sk. 3d dual -tur. This state of things allows us to see at once that in Ir. 3d sg. - a r 1 we have a form directly comparable with Sk. 3d plur. -ur. I note also Umbr. 3d sg.fer-ar (=fer-a-tur). The correspondence is certainly striking, that the abl. gen. sg. is again found similar to the 3d pers. sg. of the verb. A word remains to be said on the nature of the compound stem tara- with thematic bye-form tr-a-. It is, like sya (supra, p. 422), and sva, below, a compound of two demonstrative stems, ta- + ra-. As we actually have it before us, we may regard the ending -tur as an abl. 3d sg. in -/+an abl. 3d sg. in -rr, the result of elision of -ra, as -t is of -ta. T H E S U F F I X ter.—The difficult question of the relation of the past ptc. suffix in -to to the agential suffix in -tar here comes to a solution. In Avestan the ptc. and the agency noun often conflict in meaning (cf. my 'Studies in Etymology,' A . J . P. X I I I , p. 477). It was noted above (p. 416) how the 3d pers. consonantal stem T- is act. or pass, in meaning. Its thematic extension a-yvtorowas prevailingly passive; a further compound with the weak stem -r-, giving -tar-, was again prevailingly active. I reserve to a later point the discussion of the development of the Italo-Celtic deponent-passive, and of the Sanskrit perf., merely remarking for the present that what is probably the earliest of all the perfs., the only one, I believe, widely diffused in the subsequent literatures, is that represented by Sk. vida, Grk. Folia, which is neither reduplicated nor 'perfect.' OTHER





stem ve-'1 also entered into verb-inflexion; this stem was finally adapted to the 2d person, and is the base of Lat. zids. It appears in Sanskrit verb-inflexion as a 2d sg. (mid.) impv. in the compound form -sva, directly comparable with -sya (supra, p. 422). 1

This is the ending for conjunct inflexion. In the form ber-ir beside dober-ar I would see a *ber.r,o' with metathesis of *ber-ri because of dober-ar. 2 From this point on the writing of stems with -a will be occasionally given up in favor of the current theories.





ke, ske (Brugmann's Classes X X I I - X X I I I ) . — A n o t h e r stem that made its way into verb-inflexion was ke-. Just like e-, this was doubtless added first to the root-nouns to form impvs. Its deictic value for verbs was just what it was for pronouns. The -k became so thoroughly identified with the root that examples covering the Aryan field do not appear. I find one in Latin face || fac, precisely comparable in point of formation with hl-ce || hi-c. In Greek the pf. sign is doubtless to be ascribed to this source, as also the three aorists in -«e. But if the proposition is incapable of proof that -ke- was added to the Aryan root, it is very clear that the compound -ske- was. I note Horn, fido-nf1 (always with elision Poo-*') and Vedic gâcha. This type was widely diffused. Interesting is the distribution of these suffixes between pres. and aor. in Grk., e. g. \à-a