Vol.

xlvii] Subject azd Predicate I3

II. - Szubject and Predicate

By PROFESSOR LEONARD BLOOMFIELD

UNIVERSITY OF ILLINOIS

IT is remarkable and perhaps characteristic of the progress

of investigation into the more habitual and socialized of our
mental processes, that linguistic theory is by no means clear
as to the nature of subject and predicate in language, in spite
of the fact that our speech-feeling seems to distinguish quite
clearly between predicating and non-predicating utterances.
The prevalent view, expressed in our practical handbooks
(e.g. Goodwin's Greek Grammar, I897, P. i96 f.) and many of
our scientific manuals (such as Paul's Prinzipien der Sprach-
geschiclzte4, chap. 6), is that " every sentence contains two
parts, a. subject and a predicate." 1 With this view the
speech-feeling often enough comes into conflict, and we then
resort to auxiliary hypotheses and forced interpretations of
various-kinds, saying, for instance, that one or the other of
two parts is left unexpressed in exclamations such as ouck !
or fire!, or that the two are contained in one word in such
Latin sentences as cantat2 or p/nit,3 or, worst of all, we deny
the name of sentence to such utterances as yes or to an-
swers such as-yesterday.4
There is a psychologic principle coming to be more clearly
recognized by students of language, which shows the way to
a better interpretation of the process of utterance and of the

I This notion arose in the seventeenth and eighteenth centuries; it appears in

the Port-Royal Gramnazaire generale et raisonnee (I676), in Wolf, Philosophia
rationailis (1732), in Hermann, De ernzedanda ratione Graecae grammaticae
(I80I), all quoted by Delbriick in the introduction to his Syntax (Brugmann
and Delbriuck, Grundriss, iII), and in 13ernhardi, Anfangsgruinde d. Sprachwis-
senschaft (1805), quoted by Delbriick, Einl. in d. Studium d. indogermi.
Sprachen5, p. 34 ff.
2 So even Delbriick, Grundriss, v, Io: "Bei der ersten und zweiten Person
des Verbums steckt das Subjekt in der Verbalform."
3 So even Wundt, VIuker-psychologie, II3, 227 (" indefinite subject "').
4 Wundt, op. cit. 24I. The standard view is presented in my Introduction to
the Study of Language.

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
 14 Leonard Bloomfeld [19I6

speech-feeling. This principle is to the effect that the men-

tal phenomena must be viewed as they actually occur and
not as their products or a record of their occurrence may be
interpreted by an observer after the f act.5 It is this principle
which has led to Kretschmer's definition of the sentence as
the linguistic expression of an affect - of a single rise and
fall of the emotion prompting to speech.6 It is this principle
which makes it clear that a single word can express only one
separately apperceived element: that it is wrong, for instance,
to interpret a form like cantat as coIitaining two such ele-
ments. A leisurely student may reinterpret such a form into
a logical judgment predicating the act of singing of a certain
person, but the logical judgment is not present in the
speaker's mind when the sentence-word canftat is spoken;
for what we mean by saying that cantat is felt as a single
word (and not as two words) is exactly this, that it contains
no opportunity for an apperceptive (and hence for a logical)
act of division. The speaker's experience is simply that of a
known and definite person's singing; his expressive reaction
is a habitual unit, cantat, and such morphologic structure as
we find in this word is mnerely associative; it exists only by
virtue of the parallelism and contrast of other forms and is
not explicit in the utterance itself. Finally, pursuing the
same principle, I have suggested (T.A.P.A. XLV, 65 ff.)that
even where there is a word-boundary, there is in most in-
stances of utterance no apperceptive division; that, for
instance, the English sentence, she is singing, is usually
spoken in much the same way as the Latin, Italian, or Slavic
one-word equivalent, and differs from the latter primarily
only in being occasionally used for an attentively discrimi-
nating statement, sk6 is sizging, which corresponds to Italian
el/a canta, not to Italian canta, Latin cantat.
If we keep this principle in view, it is not difficult, I believe,

5 See James, Psychology, I890, I, I66 f., 274 f. (the " psychologist's fallacy ")
and Wundt, Grundriss d. Psychologiell, I3 (the fallacy is " die eigenen Reflexi-
onen des Psychologen uiber die Tatsachen in diese selbst zu verlegen "), Logik,
III, I50 f., and, generally, the works of the latter author.
6 In Gercke and Norden, Ein. in d. Altertumswissenschaft2, I, 516.

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
 Vol. xlvii] Subject and Predicate I5

to reach a clearer understanding of the nature of predication

in language.7

I. We may consider first a type of sentence about which

there can be little question. If, in the course of a philosophi-
cal discussion, there occurs the statement, homo mortalis est,
it is obvious that this sentence may well be the linguistic ex-
pression of a logical judgment. To the logical subject, that
talked about and underlying the predication (To- V7rOKECL4EvOv),
corresponds the word lhomo, and to the logical predicate, that
said about the subject (To' Kar7TyopovIievov), corresponds the
phrase mortalis est. It is a natural transference of terms
but we must not forget that it is a transference of terms -to
call the linguistic element corresponding to the logical subject
a grammatical or linguistic 'subject' and to speak similarly
of a grammatical or linguistic ' predicate.'
Another type of sentence differs from this by the absence
of the verb: beatus ille /omo; the division into subject and
predicate is, however, no less clear. This type, entirely lack-
ing in English, is in Russian, for instance, the only form for
non-narrative statements of a certain kind: mnuz'ik be'den
'peasant poor', i.e. ' the peasant is poor.'
Such examples as these have played an unduly important
part in the development of syntactic theory. A studetnt con-
fronted by the task of analyzing his speech enters into a state
of abnormally careful attention; this attention he exercises
not only in the analysis, but also, inappropriately, in forming
his examples, which, in consequence, are logically constructed
statements of the type we have described, rather than casual
phrases.
We may, however, take an utterance of this very type, such
as he is a lucky fellow, and, with a different distribution of
pause, duration, pitch, and stress, utter it not as the expres-
sion of a deliberate judgment, but as an enthusiastic exclama-
7 In accordance with this principle the process of sentence-utterance has been
most vividly and exactly described by James in his Psychology, I, 260-263 ; the
consequences for linguistic theory were not in this connection (nor, so far as I
know, in any other place) drawn by this great philosopher. See also Morris,
T.A.P.A. XLVI, 103 f., esp. 110.

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
 i6 Leonard Bloomfield [I9I6

tion, he's a lucky ferlow!, or we may half plaintively, half

enviously mutter, he's a lucky fellow ! In these instances
the speaker's frame of mind is far removed from that of logi-
cal predication. He is expressing primarily an emotion, and
his speech comes forth without any apperceptive jointings.
Though it is easy enough, once the words are spoken and
remembered, to interpret the sentence, in cold blood, into a
judgment, yet our task is not to interpret what the speaker
may or should have meant to say, but to analyze the expres-
sion itself. It is an expression of emotion at a certain state
of affairs, and lacks logical structure.
Yet there is a reservation. Although our ejaculation of
wonder or envy differs in accentual features from the calm
judgment, he is a lucky fellow, the two utterances are the
same so far as distinctive word-form is concerned; and,
what is more, the casual ejaculation is accompanied by a
peculiar feeling-tone, a subtle and indescribable sense of com-
pleteness or roundedness, whose presence we are wont to
signal by calling the statement a 'complete predication' or a
'complete sentence.' This appears clearly when we con-
trast he's a lucky fellow! with the otherwise equivalent lucky
fello ! which lacks this tone of completeness.
This circumstance bears its explanation on the face of it:
thze language from which our example is taken uses for many
non-logical utterances the samne distinctive word-forms asfor
the expression of a logicaljudgment. If we ignore - as per-
haps we have the right to ignore temporarily -certain fea-
tures of duration, pitch, and stress, then we may say that the
expression of a logical judgment (predicating, e.g., a state of
happiness of a known person) is often the same, in English,
or Latin, and probably in most languages, as a rather ex-
plicit exclamation (e.g. of envy or surprise at the circumstance
of his happiness). If we wish to keep the terms 'linguistic
subject' and 'linguistic predicate,' we must therefore define
them not straightway as the linguistic expressions of a logical
subject and predicate, but rather as linguistic elements whichi
can be used in this function, but are used also in other utter-
ances, as components of a habitual sentetnce-type.

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
Vol. xlvii] Subject and Predicate

IT. The type of sentence we have so far examined is in

Latin and in Russian confined to non-narrative statements
and therefore relatively often used as the expression of a
logical thought-content. Of other types this is less true.
When we say, thzen Mary bought a hat, we are usually in a
narrative frame of mind; a concrete and colorful picture
floats past the 'inner eye,' and from logical judgment we are
far removed. To make a sentence of this type express a
logical judgment we must postulate some rather strained sit-
uation, in which moreover the accentual features of the utter-
ance will be entirely different. Outside of such unusual
situations our sentence is by no means the utterance of a
logical predicating-experience; yet it presents the character-
istic structure which allows us to analyze it into a linguistic
subject and predicate.
These linguistic predications of the narrative type differ in
Latin and in Russian from those of equational type (hzomo
mortalis est, beatus ille) because these latter in Russian
always and in Latin optionally lack the finite verb. In Eng-
lish, German, and French the two types are merged.
Now, it is a fundamental principle of linguistic study that
we have no right to inject into our analysis of a language
distinctions not expressed in the language. If, therefore, we
borrow the technical terms 'subject' and 'predicate' from
logic for such a sentence as man is m orta/, we are bound to
keep them also for the structurally similar Mary boughit a hat,
and consequently to distinguish between the use of these
terms in logic and their use in linguistics.
No doubt the extensive use in our languages of linguistic
subject and predicate in non-logical utterances has contributed
to the induration of the traditional rationalizing view, which
tries to see in every sentence of language the expression of a
logical judgment. We are now in a position to clear up some
of the difficulties to which this view has led. For instance,
Wundt, who strictly identifies linguistic subject and predicate
with those of logic, is forced to make the following inconsis-
tent statement ( Vo/kerps. I13, 270): " If the speaker ever for the
nonce grammatically chooses a subject different from that

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
 i8 Leonard Bloomfield [I9I6

which logically he might intend to make his subject, then he

has given his thought an inadequate form; in which case, to
be sure, other than purely logical motives, such as euphony
and the rhythm of speech, may at times excuse the devia-
tion." Now, it is not for us to make excuses for a speaker or
to heap humiliation and reproach upon him if he fails to
accord with our theories of syntax. The situation Wundt
describes is merely this, - and it is in our languages a very
frequent one - that the linguistic subject and predicate would
not, under a logical reinterpretation of the sentence, produce
a correct logical judgment. Or, more exactly: if we write
down the sentence and then read it with logical intonation,
we may find the subject and predicate poorly chosen for the
logical purposes of the situation. If I say, The hat was
priced at five dollars. A woman went in and bought it,
my second sentence, under logical interpretation, would pre-
sent a poorly chosen subject, for it is the hat and not the
unknown woman that ought, logically, to be the subject of the
new statement; I should say: The hat was priced at five
dollars. It was bought by a woman who had entered the
store in order to buty it. This deviation of linguistic sub-
ject and predicate from a logical norm is, however, not, as
Wundt's words suggest, a rare or occasional feature, but will
be founid extremely common in our languages.

III. While in modern English, German, or French the

great majority of narrative sentences exhibits the structure of
linguistic predication, this is not true of the older stages of
these languages, of Latin, Italian, Greek and Slavic, or, in
general, of the older type of Indo-European speech. These
latter languages possess a kind of narrative sentence in which
a linguistic subject and predicate cannot be found: the
simplest instances are sentences of one word, such as cantat.
Of whatever parts such an utterance may consist, they are
not separated from each other by any apperceptive analysis,
such as that of logical predication; if they were, we should
speak of several words, not of one word. In English, for
instance, a word such as stones contains two associatively

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
 Vol. xlvii] Subject and Predicate I9

joined elements; if we attentively separat

use a single word, but speak of several s
expressing the plurality by a separate word. So a Latin
speaker, if he apperceptively analyzed the experience into an
actor and an action, would no longer say cantat, but illa
cantat (Italian ella canzta, Russian ona poJot). The analysis
into object and number in stones, into actor and action in
cantat is never explicit; the word as a whole corresponds to
the experience as a whole; this experience is associated with
other partially unlike experiences, which are expressed by
similar words, such as stone, stony (same material element) or
trees, rivers (same element of number), cantds, cantabat (same
element of action) or saltat, dor-mit (same element of actor);
but the analysis involved in the existence and association of
these parallel words is merely implicit and associative. We
have therefore no right to speak of a linguistic subject and
predicate in a sentence like can/at.
Two factors have led to the forced interpretation which
sees in cantat a subject and a predicate. One is the obvious
similarity between such a one-word sentence and an English
sie is singing. The two might be used by a bilingual speaker
of English and Italian of one and the same experience. To
those who see in the English sentence the expression of a
logical judgment, the obvious similarity of the English and
the Latin-Italian sentences is a motive for seeking in the
latter also a logical predication. For us, however, the simi-
larity between the two types confirms the conclusion that
normally the linguistic subject-and-predicate structure of the
English sentence does not express any apperceptive analysis
of the experience, but is merely a habitual formality.
The second factor is this: the duality of elements in cantat
does correspond in a striking way to the duality in sie is
singing or elle chante. This correspondence has, however,
in principle nothing to do with subject and predicate, and
consists only in this, that Latin, like the modern languages,
analyzes the situation into actor and action - though, to be
sure, by a merely implicit analysis, whereas the English,
French, or German sentence is at least capable of attentive

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
 20 Leonard Bloomfield [i9i6

separation. When an author is persuaded that the Latin

word "contains a subject," he is mistaking an actor for a
subject, a fallacy induced by the circumstance that in Eng-
lish, German, and French the subject is always viewed as an
actor. To use the term 'subject' for 'acting person or
object' would be an unwarrantable extension of the term
which could only create confusion. Both the Latin cantat and
the English she is singing are expressions of actor and action,
but only the latter contains a (linguistic) subject and predicate.
One class of sentences of the ancient one-word type has
received special attention from linguistic students, that of
utterances about the weather, such as pluit. Both Paul (op. Cit.
I30 f.) and Wundt ( Volkerps. II3, 227 f.) see in these a subject
and predicate.8 From our standpoint there are two reasons
why this cannot be true. Psychologically it is not correct to
attribute an act of logical judgment to a speaker who merely
says plhiit or piove or it's raining. His act of apperception is
by no means an analytic one: he takes the experience as a
whole without breaking it up into an underlying element and
a predication about it. Linguistically, we mean, when we
call pluit or piove a single word, that it is not capable of
expressing more than one apperceptively grasped element of
experience.
The ancients were able to make a logical extension of such
a sentence as p1uit, when they did so, they said Ze1s VEt or
Luppiter tonat. Strepsiades asks, aX'Xa rAs i E; and his
answer is, in burlesque form, the ancient view. As our
analysis, today, when we devote attentive thought to mete-
orological phenomena, is rather akin to that of Socrates in
the Clouds, we are forced, at such times, to diverge far from
the usual utterance, it's raining.9 The linguistic subject and
predicate in the English it's raining thus give us an example
of a linguistic predication which never represents a logical
j udgment.

8 Delbriick at first rejected this view, but later hesitates; cf. Grundriss, V, 37
9 This way of talking about the weather has come to us from of old, when the
it still represented a concrete actor. Traces of the old state of affairs in Ger-
manic in Grimm, Deutsche Grammatik, IV, 228.

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
Vol. xlvii] Subject and Predicate 21

If these examples have made clear to us the general nature

of linguistic predication, especially in its divergence from logi-
cal predication or judgment, we may, in conclusion, briefly
note a few of the features of linguistic subject and predicate
that appear in our languages.
We have already had occasion to see that in English, Ger-
man, and French the linguistic subject is looked upon as an
actor and the linguistic predicate as an action performed by
this actor: Mary bought a hat, she is singing, and even it is
raining. This, indeed, is universal, no matter how inappro-
priate the identification may seem when we reflect upon it:
I hear az noise, Mont Blanc is highi, the house was butilt, the
house is being buiilt, and so on.
This is not true in Slavic or in Latin. Both of these
forms of speech add a second type of linguistic predication,
in which subject and predicate are viewed as equated terms:
beatus ille.
Latin has a third type, in which the linguistic subject is not
an actor, but an object fully affected or produced by the action-
predicate: domus struitur. As this construction is known as
the 'passive,' we may define this term in accordance with the
conditions in Latin: in a language which employs a construc-
tion (morphologic or syntactic) of actor and action, a parallel
construction in which some other feature is coupled with
action, is a passive.
This somewhat obvious definition is worth formulating
because there has been some uncertainty as to the applica-
tion of the term. Most writers find in the Philippine lan-
guages three ' passives' (so the Spanish writers and with
them H. C. von der Gabelentz, A bh. Sachs. Gesell. VIII, 48I),
but Wilhelm von Humboldt (Kawi-Sprache ii = Abh. B6rl.
Akad. I832, 3. Teil, I50) refused to apply this term to the
Philippine constructions; he is followed in this by the best
of our Philippine grammars, the late Dr. Seidenadel's de-
scription of the Bontoc Igorot language. Under our defini-
tion the Philippine constructions will receive the name of
passives, as may be seen from a few examples taken from
Tagalog. There is an actor-and-action construction, e.g. sya

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
 22 Leonard Bloomfield [I9I6

y sumu'/at nay liham, 'he wrote a

is a sentence-type in which the (linguistic) subject is the
object fully affected or produced, somewhat as in the Latin
passive: sinu'iat nya ay li'liam, 'was-written by-him the
letter,' ie. 'he wrote the letter'; this we may call the 'direct
passive.' Secondly, there is a 'local passive,' in which the
subject is the person, thing, or locality less fully affected
by the action-predicate, as though an Indo-European dative
or locative should become the subject of a passive con-
struction: sinuldtan nya ako, 'was-written-to by-him I,' i.e.
' he wrote to me.' Finally, there is a construction which we
may call the 'instrumental passive,' in which the subject is
the means or instrument or that given forth -in part some-
what as though an Indo-European instrumental could become
the subject in a passive construction: isinziat nya ay kwento,
' was-written-down by-him the story,' i.e. ' he wrote down the
story.' It may be worth mentioning that these languages
have also the type of sentence in which the experience is not
viewed as an action: mabu'ti sya or sya y mabte, 'bonus
ille.'
For the sake of completeness we may refer to another
feature which has been confounded with subject and predi-
cate. G. von der Gabelentz (Die Sprachwissenschaft2, 369 ff.)
invented the expression " psychological subject," which he
used to name what we should call (with Wundt) the emo-
tionally dominant element of the sentence: e.g., in today is
mI birthday the " psychological subject " is my. Wundt
(VFokerps. I 268 ff.) shows conclusively that the terms
'subject' and 'predicate' are here entirely inappropriate.
Whatever the exact relation may be between an emotionally
dominant element and the apperceptive processes, such as
underlie logical judgment, a confusion of terms can have
only bad results. It is interesting to see that in certain
languages, namely Celtic and French, there is a tendency
to identify the emotionally dominant element with a linguis-
tic predicate: compare the Irish-English fondness for such
constructions as it's he that did it.

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms
Subject and Predicate
Author(s): Leonard Bloomfield
Source: Transactions and Proceedings of the American Philological Association , 1916,
Vol. 47 (1916), pp. 13-22
Published by: The Johns Hopkins University Press

Stable URL: https://www.jstor.org/stable/282823

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 [email protected].

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

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and
extend access to Transactions and Proceedings of the American Philological Association

This content downloaded from

186.84.22.59 on Mon, 01 Feb 2021 17:01:49 UTC
All use subject to https://about.jstor.org/terms