Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Suffixes, Determinatives, and Words

Author(s): Walter Petersen


Source: Language , Mar., 1928, Vol. 4, No. 1 (Mar., 1928), pp. 7-17
Published by: Linguistic Society of America

Stable URL: https://www.jstor.org/stable/409498

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to


Language

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
SUFFIXES, DETERMINATIVES, AND WORDS

WALTER PETERSEN
UNIVERSITY OF FLORIDA

In the recently published third volume of his Indogermanische Gram


matik Professor H. Hirt, in his strikingly original and suggestive way
has unfblded a panorama of the history of the formation of IE noun
which differs so radically from the conception of the majority of scholars
both in many details and in his general point of view, that it will be
of the utmost importance to examine the credibility of his theories an
the reliability of his methods. This is true all the more because h
presentation, as elsewhere, through its clarity, through the interest which
it arouses, and through the confidence with which it is presented, is apt to
fascinate the minds of those who have not yet formed an opinion of their
own, but are seeking orientation in the problems of IE linguistics fo
the first time.
The outstanding peculiarity of Hirt's opinion is this, that while he
believes in adaptation as far as inflectional endings are concerned, an
insists on the identity of these with the word-forming suffixes, yet the
latter themselves are explained altogether in the manner of Bopp and
his followers. It is true that he admits that new suffixes arise by mis-
dividing words, so that the suffix is increased at the expense of the primi-
tive word, and that he devotes a whole chapter to this process, but t
him this simply means taking a formative part of the word and addin
to a suffix which already was there (226), as in the example of Gr. (epv
Kbs by misdivision as pev1-LKbs giving rise to the suffix -LKOS. He ignore
the possibility that parts of roots and unanalyzed words may become
suffixes through word contamination and subsequent associations of
old and new forms, e.g. that IE *9him-osl Skt. him-ds 'cold' brought it

1 To the speaker of the language *^him-o-s was the correct analysis, i.e. he did
not think of the m as being a suffix because there was no associated word withou
m with which to compare. For the probable earlier history of the word cf. Brug-
mann, Gr. 2. 12. 135. Of course it is equally true that words which were very
clearly derivatives to the most superficial analysis would also often or usually be
felt as a unity. These considerations show how illogical it is to accept 'mis-
7

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
8 WALTER PETERSEN

m to the opposite *#huor-mo-s


-mo- then revealed itself by ass
corresponding e.g. to Gr. O'pos S
The suffixal system as such, th
However, the most original aspect
division of suffixes into two cla
he speaks of 'determinatives', th
determinatives of IE grammar, w
origin of which he seeks in numer
those added to pronouns, as *s
those suffixes which display a ta
independent words, so that he h
the late Professor Fay, whom, ho
the obvious similarity of opinio
Since Hirt denies any other ulti
and independent words (85), we
on determinatives (V) and suffi
meant to give an adequate gen
suffixes arose, even though he nat
an example and will not offer a
examining therefore his method
it will be well to apply to his ent
tions and principles, so as to give
or false in essence.
To the general point of view that there is no ultimate source of suffixes
other than composition, be it with particles or 'independent' words, it
can merely be answered that Hirt has made an incomplete disjunction.
Not only is at least one other such source conceivable, but he has, in
the opinion of the writer, omitted from consideration the most impor-
tant one of all, sc. word-contamination, irradiation, congeneric assimila-
tion, and influence of otherwise associated words, which processes can
cause not only redividing between words and suffixes already there, but
can create a new suffix altogether. Cf. Am. J. of Phil. 37. 176 ff., 255 ff.
Whether Hirt has never heard of these additional possibilities, or

division' as a process which rearranges the length of the various formative ele-
ments, but to reject the same occurrence when it clips off something from the
supposedly sacred and inviolable root. To the mind of the speaker it did not
make a particle of difference whether the part clipped came from a root or another
formative, since he did not and could not know what the original form of the root
might have been, and the whole process was unintentional anyway.

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
SUFFIXES, DETERMINATIVES, AND WORDS 9

whether he has purposely refrained from mentioning the


affect the fact that he has based far-reaching conclusions o
tion that lacks a most important alternative.
Not only this, but there is a definite reason why Hirt's sk
history of IE word formation must be essentially false, i.e. f
large majority of words. It totally neglects to draw the nec
clusion from the psychological fact that in ordinary speak
rarely conscious of the analysis of words into distinct elements
the vast majority of mistakes (which are the source of lingu
are not due to consciously (though wrongly) dividing words
elements at all, but rather to associations of which we are ei
scious or dimly conscious. Cf. the books of R. Meringer, V
und PIerlesen and Aus dem Leben der Sprache. We have ever
assume that these same influences which cause new word formations in
the offhand speech of today are those which caused new words to be
formed in the past. Surely it would be wrong to deny both that we ever
analyze to any extent our spoken words as well as that suffixes ever
arose by composition, but ordinarily we certainly do not while speaking
analyze our words, and the analogy of the mental processes involved in
momentary mistakes forces us to conclude that most of the linguistic
changes, including the formation of words, must have been unintentional
processes, and this can mean only that most suffixes at all times were
formed by word contamination which was unknown to the speaker on
its first occurrence.
On the other hand, not only the assumption that all words were
formed by the conscious process of composition, with the exception of
a small number of root words, but also the amount of analysis often
presupposed for the individual word, is manifestly impossible. Thus
Gr. ~aXbs Lat. similis is supposed to have been formed by adding the
determinative 1 (139), followed by the independent word o (187) to the
stem of o5bs which, since it has the thematic vowel, must itself be a
compound of *som- and the word o, followed by the determinative -s.
To analyses of this kind there is absolutely no limit whatever. How
far e.g. Hirt would go in the case of Gr. vepE'X Lat. nebula 'cloud',
no one could profess to know. But we can apply his method and get
the following result: The root or better primitive word *ne is followed
either by the determinative bh (133) or the independent word bh < *bhfz
or *bha (224); next comes the independent word e or o (184 ff.) followed
by the determinative 1 and the determinative d (108). Result: *ne-bh-
2 Cf. Wundt, Sprachpsychologie3 1. 584 ff.

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
10 WALTER PETERSEN

e-l-a, consciously analyzed, at least


though not one of these has a tang
Still worse than the case of compo
tion with particles, which Hirt cal
the meaningless suffixes. The indi
generally compounded with particl
pronominal case-endings, cannot be
nouns had the same habit in earlie
for the emphatic and contrasting p
not hold good for nouns. The ever present contrast between I and
thou, we and you, he and she, this and that, and many others, is a contin-
uous inducement to employ all the emphasizing elements which
language possesses. But this motive is present only occasionally for
adding emphasizing particles to substantives and adjectives, nor is
there any evidence in historical periods that particles were added to
nouns in large numbers. Hirt's own explanation of the reason for
employing them is that they had already become a sort of suffixed article
in IE times (164 f.), comparable to the various articles in certain Bul-
garian dialects. But we are squarely up against a destructive dilemma.
If these particles really had developed into articles, they would have
been inflected, but there is no evidence that any such IE articles de-
veloped from particles, or that there was any IE article at all, even one
developed from pronouns, like those which individual IE languages
know. Even the use in this function of the pronominal stem *to- in
Greek and Germanic can be shown to have been an innovation of these
languages. Cf. Brugmann-Thumb, Gr. Gr4 485 f. If, on the other
hand, we assume that these particles were added as such. and had not
developed into articles, this assumption of the addition of an immense
number of particles to nouns conflicts with the rareness of such an
addition in historical times and the absence of any frequent motive for
such an addition.
When therefore Hirt himself is astonished that he has found so many
particles suffixed to nouns, we might add that he is so no more than we-
until we looked into his method of finding them. These 'determina-
tives' are for the most part either single sounds (i, u, , 5, a, 2 , Ik, g, t, d, dh,
th, p, bh, r, 1, n, m, s) or else a combination of only two sounds (e.g.
te yo, ghe gho). They thus comprise a large part of the sound material
out of which the language is made, including the simple vowels except
ai, and o e, which is claimed as an independent word, and most of the
common consonants. There are not many words which do not contain

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
SUFFIXES, DETERMINATIVES, AND WORDS 11

at least one of these sounds, and we may pertinently ask wha


Hirt has to distinguish between determinative and other us
same sounds. Certainly not the meaning, since they are dec
meaningless, or at least are supposed to explain meaningless
Nor can we appeal to their position, for it is an arbitrary a
that a meaningless element placed at the end of a word just b
inflectional endings ever had any more of a separate existenc
same sounds in the beginning or middle of the words. Th
ranted assumption is made still worse by the fact that these
tives, being particles, must have been uninflected originally
the process by which e.g. German dieser diese, opposed to E
these, has shifted the inflection from middle to end, in th
not an occasional occurrence, but took place in at least half
stantives, or at least in half of those which existed when th
took place. Once more that which took place occasionally in
times is supposed to have been the normal thing in a format
Once more Schleicher's point of view is adopted against that
'Junggrammatiker', who maintain that the linguistic proce
present and the past are identical.
We may discard absolutely and unconditionally the whole
'determinatives'.3 It has almost everything against it, an
nothing for it. It is unnecessary to go into details and to sh
inconsistencies and absurdities it leads to when applied to in
words, since the total absence of criteria lets it depend on the

s The assumption that determinatives, i.e. particles, were promiscuou


to nouns is also a blemish on Hirt's otherwise admirable suggestions on
of nominal inflections (VI). That inflectional endings, as far as no
may have originated as pronouns or prepositions, we may well be
there were any particles among them, except when a pronominal endin
ferred to nouns secondarily, is highly improbable. That there were
Hirt assumes, is outside of the realms of possibility. The difference
endings of nouns and pronouns resolves itself just exactly into this:
tively pronominal endings, of the origin of which there is any clue, ar
while of the nominal case endings just the reverse is true. In the li
above paragraph that is what we are bound to expect.
4A peculiarly whimsical example of Hirt's, bearing on inflection r
word formation, is his assumption that e.g. *iug-om nom. ace. neut. is
analysis rather than *iugo-m. From the fact that only the neuters o
clension have the m, and that therefore no -m occurs alongside of -o
neuters, Hirt draws the astonishing conclusion (88) that the -om of the
not subject to the workings of the accent, and must therefore be a lat
particle of course. He is sure it must be separated from the accusative

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
12 WALTER PETERSEN

the individual's analysis whether he


tive', or a 'real' suffix derived by com
able stem. Two or three examples
most meaningless of all suffixes, it
word (185 ff.) because it often forms a
it is used parallel to o in almost ever
oftener than the latter because of its g
of abstract nouns and because it also forms feminine derivatives from
masculines, is placed among determinatives (108), although in terms
that would better fit its classification with the real suffixes. In case of
Gr. ?pei&Ai, the abstract of pel6oiaL 'spare', the i is declared a deter-
minative (139) which does not have any influence on meaning', but the
suffix -men-, which forms similar abstracts e.g. in Gr. vei a Lat. niZmen
'nod': v:ebw nuo, has such an important meaning that it probably must be
an independent word (202). Or in case of the suffix -tat(i)-, in spite
of the parallelism of the series of equivalent suffixes -t-, -ta-, -tat-, -tdti-,
which suggests gradual accretion by 'misdivision', the function of
forming adjectival abstracts is declared so important that it must needs
be an independent word. The suffix -mo- is surmised to have been
mn 'measure' in Lat. minimus, analyzed mtnu-mo-s 'having a small
measure', while e.g. Goth. dams 'judgment' is said to have the deter-
minative m, although it is just as easy to force it into the category of
me 'measure' as being 'the measure of what is being done' (cf. Gr.
riOerLy). No further comments are necessary. This is enough to show
that even if the whole doctrine of determinatives were not fundamentally
wrong, no reliance could be placed on judgments of individual instances.
Let us now turn our attention to the chapter on suffixes, in which those
examples are cited which Hirt surmises or believes to have been inde-

in -o-m, as well as from all the o cases of the neuters themselves (all except nom.
acc. pl.), but that it does belong to the ending of pronouns as Skt. ah-dm 'ego,'
and of adverbs like Gr. /#ab6'p.
6 It is true that alongside of <pe~rwXj we find 0pabe& in the same sense, but it would
be rash to maintain that when one abstract suffix is added to another the later one
is meaningless. Surely one diminutive suffix preceded by another is not called
meaningless, but is considered to have been added for emphasis. Cf. Brugmann,
op. cit. 674. To the apperception of the Greek speaker -wXh apparently was felt
as a single suffix, as is shown by the analogical ebx-wX74: Ei:xo/AaL and repr-wX7:
T-p7rw, for which Hirt assumes abstracts *ebx& and *Tep7'r as primitives. That in
this combination the X had any less to do with the sense than the w, there is nothing
to prove.

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
SUFFIXES, DETERMINATIVES, AND WORDS 13

pendent words. Here he does not stand alone, but is doing what o
have done before, e.g. Prellwitz in Germany and Fay in America
large part of the objections to his procedure will therefore be the
as those applying to the others. The writer has stated these at gr
length in his article in the Am. J. of Phil. 37. 173 ff., and can r
them briefly here only to show that Hirt is falling into the same pit
as the others, and is making the same mistakes, in spite of the fac
the class of determinatives has been invented to receive the discards
from the other category.
When the word from which the Isuffix is ostensibly derived is a general,
vague, or abstract word, as 'standing', 'appearing', 'being', there is
indeed a large proportion of derivatives which can be forced into the
mould, but they are all unconvincing because logically the meaning of
any word ending in any other suffix could be analyzed in exactly the
same way, because there is nothing about any particular word which
points to such an origin, and because the whole idea presupposes that in
an early state of language abstract words were so incessantly in use that
they were regularly added to complete words without changing their
meaning, and so that it was preferable to say e.g. 'what has become a
deer' to simply 'deer' (EaX" os 'deer' supposedly related to *bhji 'become').
Similar examples from Hirt are the derivation of -istho- from *sthd-6
'stand', so that e.g. I8Twros 'sweetest' must have been 'standing as
sweet', and that of the suffix -es- -os- from *es- 'to be' (19', though
doubtfully), so that Skt. grdvas Gr. KXEo~ must have been the 'being of
fame' instead of 'fame'.
When the supposed source of the suffix is a concrete word, or a w
of small extension of use, there are other elements of improbabilit
is then necessary to assume that one or very few of the derivatives
show a mere trace of the original meaning, and that the rest have w
dered so far astray that no connection could be discovered even by
penetrative intelligence of the inventors of such etymologies
would be necessary to believe that in hundreds of examples the suf
-/o- shows no trace of its origin from i 'to go', but only a mere han
like Gr. 6iyptos 'wild' Skt. ajryds7 'belonging to the plain' still prov

6 So Fay, Am. J. of Phil. 34. 15.


It is a specious argument used by Hirt and others that part of the example
of a suffix in such cases must have had a different origin from the rest, and that
not permissible to want to bring everything under one category. As a gen
principle this certainly cannot meet with any objection: it is its application
is not convincing. If &,ypLos could be interpreted in no other way, or if the o

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
14 WALTER PETERSEN

to have originated thus neverthele


meaning 'going in the field' upon
occurrences. We are asked to accep
from the basis *teyM 'be strong'
others, one can force Gr. vebrTS 'y
youth' even though the identical Lat.
it did not mean 'youth', at all, but
primitive adjective v4os shows 'ne
the original meaning of IE *neyo-s
uses (Am. J. of Phil. 37. 181 f.), is
meaning only in a handful of word
and briima < *brevimd 'having a
when -yent- is derived from yen 'wi
is at hand (200), so that we do not
proof. That a mere picking out of
a certain suffix, even with the aid
from being even a presumption in
various considerations. Every suffix must look like some word or
other, since there are only a limited number of sounds and syllables, of
which all occur in some word or other. Often, moreover, a suffix looks
like several different words, and one can then usually make out as good
a case for one as the other. Thus Fay, CP 6. 315 ff., saw the root
*ed- 'eat' in the Latin suffix -Edon-, e.g. iirado 'blast', 'blight' was
'eating to burn', dulcedo 'sweetness' was 'sweet taste'. The writer
remarked (loc. cit. 175) that the suffix with as good a right could be
referred to *d&- 'give', and that e.g. rubado 'redness' might just as well
have been 'the giving of redness', and frigado 'cold' was just as con-
vincing as 'the giving of cold'. He also remarked that by the same
process one might spot Lat. ventus 'wind' in the IE suffix -yent-, e.g.
Gr. -vyco-fPer- 'having wind blasts'. It is no more fantastic than
Hirt's derivation of the same suffix from yen 'win', and has at least this
advantage that a pattern could be found. Finally, the assumption that
a few words showed the original concrete meaning of a suffix, and that
this became attenuated in course of time, is in conflict with the history

from i 'go' were self-evident and did not bear the stamp of artificiality, we could
readily subscribe to its etymology. But &LypLos is much more convincingly taken
as 'belonging to the field', and thus is exactly like the great number of denomina-
tive i(i)o adjectives. Where both form and meaning coincide with the normal,
it puts an exceedingly onerous burden of proof on him who would separate in
origin.

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
SUFFIXES, DETERMINATIVES, AND WORDS 15

of suffixes where it can be controlled; for all words like diminut


which show concrete meanings of suffixes, are found side by side
others in which the same suffix has no meaning or an extremely at
uated meaning, and it can in every instance be shown how the for
developed from the latter. To say that these must be separate
having a different origin, could be correct only if this was not the reg
occurrence but only an isolated phenomenon. Furthermore, suffi
which are demonstrably derived from words, e.g. German -heit, -t
begin with a narrow meaning and branch out gradually, as all suff
must have done if they were derived from words.
Two further considerations detract from the impressiveness of H
long list of suffixes from words. Though it is a point in his favor
not against him, that he is very doubtful about a number of his
suggestions, yet this does lessen the persuasiveness of the whole cha
if its object is to show that all suffixes with meanings were origin
words. This applies as well to those for which he has suggested patt
which do not convince himself, as to others for which he cannot even f
possible patterns. Altogether valueless is his classification in t
chapter of certain suffixes, e.g. -ent- (190 f.) and -yos- (200), as pos
coming from words because they look like words to himself, presum
because they are longer than the determinatives, although he has b
unable even to suggest a word that is superficially similar.
In the chapter is included one other class of words which could g
a mistaken impression of the frequency of suffixes derived from word
namely those which form a small circle with word ends of narrow mea
ing which can with certainty be traced to independent words. Th
mostly would not be regarded as suffixes at all, but as final membe
compounds. Not many would think of the -oiq of Gr. alooj 'gleam
or olvoqt 'wine-colored' as a suffix, since the resemblance of this sy
both in form and meaning to the verbal root b7r- undoubtedly kep
association between them intact. The only reason that Lat. atrax a
ferox, if their final member is identical, are no longer felt as compoun
is the loss of the independent word, so that there was no longer anythi
with which to associate. However, it is to be observed that th
Latin words did not cause the suffix to be added to other words
miscuously. Even in spite of its being an obscure final member, it
not yet a true suffix. Nor could we justly speak of a suffix -gena in
zndz-gena 'born within, native', nor would the last member of e.g.
prathama-jd-s 'first born', or of Gr. vio-yv6-s 'newly born' be righ
considered as suffixes in spite of Hirt's classification (206), for the

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
16 WALTER PETERSEN

ciation with the verbs from which


one. Only in a few examples lik
nature', would the final member beg
of connection with the verb. St
plu-s 'simple' (223 f.) and the iden
are still so close to the original me
that they no doubt were felt as such
verb root became impossible. On t
able or possible instances mention
speak of a suffix having developed
Gr. 7repL-o6-s < * 7rWpL-Kqo-S, of w
identified with the root of KELJa
around', and the -teno- in Lat. dif
'stretch' as 'stretching out, i.e.
served, however, that such plausib
productivity, and that not one of
satisfactorily related to any word.
This is indeed a very small residu
found in the chapter discussed, an
with Hirt himself (225) at the num
The reasonable examples are very
a large granary. Moreover, before
validity of Hirt's doctrines, it mu
on word formation are not a com
comparable to that of Brugmann i
merely those suffixes which seemed
ing his point of view, and again hi
survey its extant uses and meaning
to corroborate his theory of dete
composition. If we think of the man
were not mentioned because they
addition to those which were men
is not persuasive, we will rather b
convincing material found in the
absence of persuasiveness in the ch
This failure to carry conviction
exuberant and suggestive flood of
mind to the rigid censorship of sel
He has mistaken for a real proof
feels for his own thoughts, and re

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms
SUFFIXES, DETERMINATIVES, AND WORDS 17

his experience everything which conflicts-linguistic material,


trary opinions, general principles. The exposition of the history
noun-formation remains a description of how the system possibly cann
have originated as a whole, and offers only a few individual explana
that give promise of being true. The present writer, after reading
chapters here discussed, is more convinced than before that compos
cannot have been the main source of the IE system of suffixes, but
these, like the root-determinatives,8 are due principally to irradiat
word contamination, mutual influence of associated words.

8 The theory of Bloomfield, Am. J. of Phil. 4. 66 ff., that root determina


arose by word contamination, is infinitely more capable of explaining these el
elements in their large majority, than their identification by Hirt with 'dete
natives' or particles which he has found in noun formation. The objection r
by Persson, Beitrage 523 ff., that the identity of suffixes and root-determina
disproved Bloomfield's theory, is met by the answer that suffixes arose in the
way-an idea which to Persson seemed inconceivable.

This content downloaded from


189.173.145.71 on Sat, 05 Nov 2022 02:00:36 UTC
All use subject to https://about.jstor.org/terms

You might also like