The Origin of the Infinitive in Greek Indirect Statement

Author(s): A. C. Moorhouse
Reviewed work(s):
Source: The American Journal of Philology, Vol. 76, No. 2 (1955), pp. 176-183
Published by: The Johns Hopkins University Press
Stable URL: http://www.jstor.org/stable/292253 .
Accessed: 06/03/2013 08:24

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
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].

The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to The
American Journal of Philology.

http://www.jstor.org

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions
 THE ORIGIN OF THE INFINITIVE IN GREEK
INDIRECT STATEMENT.
What is the origin of the infinitive construction in Greek, to
represent an indirect statement? It is commonly explained as
a development of the accusative and infinitive used after verbs
of ordering. Thus, KEXEEOere e'vat (cf. P 30 f.: aXXa ' eyo/y
avaXwoprravTa KEXA.W es icvac) would be the starting point:
rXrA7r/Ov
in this phrase the accusative o-ewas originally the direct object
of the verb KEXEv1o,
and the infin. ievaLwas a dative form, having
here a sense of purpose. So the original sense was " I order you
for the purpose of going." But this phrase was then analysed
in a different way, so that the accusative and the infin. together
were regarded as the object of KEXEVW:and re now became, in this
new connection, the subject of Ievat. The next step was to extend
the use of this construction so that it was governed by a verb
of saying; and the result was k/ul oae 1evat"I say that you are
going."
This explanation is the one widely accepted in standard gram-
mars not only of Greek, but also of Latin (for its similar con-
struction).1 The formation of the construction is generally
thought to be the same in both of the classical languages: and
it is a notable point that Greek and Latin are the only two
IE languages which have the construction in a developed form,
and use it as a normal expression for indirect statement. But
Professor Adelaide Hahn 2 has recently called the explanation
into question so far as Latin is concerned. She argues that it
is difficult to bridge the gap between " I order him to go " and
"I say that he goes," since in the latter not only is the accusative
no longer the object of the main verb but-what is more diffi-
cult-the infin. is no longer used to express purpose. She claims
that, on the analogy of iubeo hunc abire, we should expect dico
hunc abire to mean " I tell him to go," as an indirect command:
1 Schwyzer, Griech. Gramm., II, pp. 372-3; Chantraine, Gramm.
Horn., II, p. 312; Wackernagel, Vorles. iiber Syntax, I, pp. 263-4;
Kiihner-Gerth, II, pp. 27-8. Further references are given in the article
by Professor Hahn (next note).
2 " Genesis of the Infinitive with
Subject-Accusative," T. A. P. A., 1950,
pp. 117-29.
176

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions
 NINFINITIVE IN INDIRECT STATEMENT. 177

but in fact it cannot possibly mean that. She therefore looks

elsewhere for the origin of the Latin indirect statement, and
with the aid of Ilittite parallels she finds it in phrases where an
accusative object has a predicative modifier, as dicant te be-
nignum, faenus creditum audio. A past participle in such a
sentence is turned into an infin. by the simple addition of esse,
and indeed such a participle form can be understood as repre-
senting an infinitive. Eventually the two constructions, iubeo
hunc abire and dicant te benignum, coalesced.
Professor Hahn excludes Greek from her consideration, ad-
mitting that the development in Greek may have been quite
different. And so it must have been, if her account of the Latin
process is correct: for we do not find in Greek the necessary
basis of a participle which is changed into the corresponding
infin. by the addition of the verb to be. We can find many par-
allels in Homer to dicant te benignum, both with an infin. (T
95 f.: TOV 7rEp IptoTrovvavsp v O aav &a`'fyEvaL) and without
8E
(? 153: Jl Trp yap a' .. c. . qe).
"EKTrp 7y KaKoyV Beside these
we can set the two comparable expressions possible in oratio
recta, both with and without the verb to be (A 169: ere -j7roXv
cEpTrEPOv EtaL, B 204: OVK ayaOov 7roAvKotpavlrj). We might be

tempted to argue that the two oratio obliqua constructions repro-

duce the antithesis of those in the oratio recta, and that, having
thus started with the infin. of the verb to be, the oratio obliqua
then went on to use the infin. of other verbs. But that would
be to leave unexplained exactly why it was an infin. which was
used: why not, for the sake of argument, a participle, which
seems a far more natural expression ("they speak of him as
being the best"), and which was in fact used after verbs of
knowing and perceiving ?
The objections raised by Professor Hahn to the supposed
development from iubeo hunc abire to dico hunc abire apply with
equal force to the similar explanation of the Greek construction.
The transition from oe
KCXEV(O tevaLto vai . . . is not properly
explained, in view of the great differences between them. Typical
is the remark of Wackernagel who, after describing the use of
the accusative and infin. after verbs of ordering and requesting,
merely states " hievon ist im Grunde derjenige Fall nicht so sehr
verschieden, wo bei einem Verbum des Sagens ein Ace. c. Inf.
steht."

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions
178 A. C. MOORHOUSE.

In addition, two other features in Greek oratio obliqua must

be mentioned. First, in Greek, but not in Latin (save in poetical
imitations of Greek), when the subject of the dependent infin.
is the same as the subject of the main verb, it is itself regularly
omitted altogether, while attributes of it are left in the nomina-
tive case. So A 397 f.: 'rqora . . . or ... oyoLyvadvvat, E 473 f.:
sq . . . 7ro'Xv Cet4/ev . . olos. If the origin of the construction
was in sentences like or''EKr7Wp KaKOV raEL, it is hard to explain
first, why the subject of the infin. is omitted, and secondly
(and this is more serious), why the nominative is used for the
attribute. Both these features argue against that sort of origin.
And they take us further, for they also argue against the deriva-
tion from the type of KEXeVZoe
l'va?.3
There is a second point of difference between Greek and Latin.
Professor Hahn, as we have seen, comments that, if the tradi-
tional derivation of the Latin accusative and infin. is right, we
should expect dico hunc abire to mean "I tell him to go,"'
whereas in fact it does not. But in Greek we do find precisely
the sense that Latin does not allow. O 57 f.: 'aro-t IIooaeLSaov
avaKTt 7ravoCaaLeVoV 7roAe.Loto . . . tKercatL,
" tell Poseidon to cease
from fighting and return," 0 433 f.: TE7-tStwjf,rtv EwE7reva/CpQ 7rvpl
aTrycrat Tpt'7roa, y 427 f.: eti7rare 8s' EWt 8fi(r(a t . . . SaLra TEveOat.
The connection of this construction with that used after KEXAv
is clear: just as KEaEVdoorElevat was originally "I order you for,
with a view to going," so Aeyw aot iEvat was " I speak to you for
going," and the only difference lies in the case, accusative or
dative according as it represents direct or indirect object. The
existence of this construction with Greek verbs of saying does
not, however, facilitate the explanation of the construction of
indirect statement. Rather the reverse: for we might more easily
believe in a transition from "order to do" to "say that," if a
verb of saying is incapable of being used with the infin. for an
order, as in Latin. Greek on the other hand shows us verbs of
saying followed by the infin., but used with two very different
meanings. We must surely expect to find that these came from
two different sources.

3 Schwyzer, op. cit., pp. 373-4, finds an explanation of the nominative

in quite a different source, the construction after verbs of wishing
(0 40: eOe'w 8e rot `7rTos elvat).

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions
 INFINITIVE IN INDIRECT STATEMENT. 179

This expectation is increased when it is observed that there

is a distinction in the choice of the verbs of saying themselves,
when used with the two types of dependent infin. clause. Nor-
mally an order is introduced by Eirov,but a statement by 4^du.
(a) Orders. In Homer the verb used for orders is T7rov.4
Later, ip( and Acywhad the same construction: Hdt., III, 89:
iv; Soph., O. C., 840: XaXavXcyw aoc.
Trola . . . JpqTro . . . 7rayLve
As a further development, the accusative of the person ordered
was used instead of the dative, as at Soph., Phil., 101: Xeyo a-e
...XaapEv; ae . . . fJL`evetv;clearly this con-
O. T., 350: evv&rW
struction is modelled on the analogy of that with KEXevW.The
infin. may occur alone, without expression of the person ordered,
as at Soph., Tr., 480: Olr' CTE KpV7TrrTv.But the use of the
infin. alone in this way is uncommon;5 this is a further point
of distinction from the construction after </tpz with indirect
statement, where the simple infin. is very frequent.
to introduce an order. Fournier
It is extremely rare for r%Utl
indeed (p. 24) quotes one Homeric example, Y 365: cadro8' tpevat
avr' 'AXtXLro,but it is far more likely that this is a statement,
"he declared that he would go." The earliest use of vutdwith
an order to be quoted by L. S. J. is from Lysias (16, 13: &-v
Tr(5Op6of3ov'XoweaXelqat' uE)."

(b) Statements. The infin. construction is very common in

Homer after vllt, totalling as many as 191 examples. But the
examples with other verbs of saying in Homer are rare: only
three with elrov, and one each with avaSw and jiv0cotua.7 There
are other verbs of saying which do not take the infin. construc-

4 Twelve examples in Homer: H. Fournier, Les verbes "dire" en grec

ancien, p. 131.
5
Fournier, op. cit., p. 132.
6 It is found with a
compound form of /rAtdin Homer, A 577: FLTrpi
..rapadq5)rt.. r. . erpa epetv At. But the meaning of the verb is
altered by the composition, and so it takes the construction appropriate
to verbs of urging, advising. The nominal form is seen in Soph., O. T.,
474-5: E\XaII/e fpdiua . . o. dbv' 6XO)\v iz8pa radvr' 1tXevetlv.
7 Statistics
from Fournier, op. cit., p. 138. The example of aisaw
is not unequivocal: K 47 f.: ob 'yap 7ir 1650,u7v, ou6' 'K\Xov av3aacavros,
avsp' eva rorcrade ... . . I7riaaOat. The verb 'KXvov may here share in
governing the accus. and infin., or even be solely responsible for it:
CiKGo can take the construction, as at Q 543: . . adKovoAiev
i'\iXpo elvaz.

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions
180 A. C. MOORHOUSE.

tion at all in Homer, such as ayopev'e, *Ev'7ro, 3agow. In the post-

Homeric language we find that when et-rovand ayopev' introduce
indirect statement, they adopt the ort-construction.
The reason for this distinction is a difference of meaning, as
has been established by Fournier and others. Eirov with its asso-
ciates is objective in sense: it means " speak, utter, tell," it
stresses the physiological act of producing speech, it is concerned
with external relations with other persons. 4yue is basically sub-
jective, "affirm, maintain, declare, think," it gives expression
to a judgment or opinion. It is significant that vldtincludes the
meaning of "think, believe," which may not lead to an act of
speech at all.8 Hence it is understandable that dirovshould have
been used for speaking which leads to an act committed by
others, i. e. for an order. On the other hand, the infin. construc-
tion after <qrutis most closely paralleled by that after 6''o,
otooaCat think."
To return now to the question of the origin of the infin. con-
struction after ,rIt, this must be seen in the use of a simple
infin. as the object of the verb 4vul; and originally there was
no subject expressed for the infin. The use of the infin. as an
object (syntactically equivalent to an accus. noun or pronoun)
is widespread and well recognized in Homer, though it cannot
have been a very ancient one.9 It is found after verbs denoting
wish, desire, be able, understand, etc. The origin of the usage
was a final-consecutive infinitive, accompanying an accusative
object (noun or pronoun). So at A 112 f.: /3ovAo/uaav rv oIKo
the pronoun was originally the object of /pov'XAoat," I wish
XECLV,
her for keeping"; but the infin. had instead clearly come to be
regarded as the object of /povXoaLC(and therefore as itself govern-
ing avTrjv), as may be seen in line 116 of the same speech of
Agamemnon, aAXa Kat 0 efeOXwo SocJvat 7raXtv(where to supply or
understand acrTjv as an object of e0eXw, if that were possible,

8 See also Buck, Dictionary of Selected Synonyms, pp. 1253 f. el7rov,

rt. *wekw-, Lat. vox, Skt. vatc- "voice": the primary sense is "give
vocal utterance, speak," used of the actual speech activity. The meaning
is
"say," with emphasis on the result rather than on the action,
as
secondary. For 0v[i, the derivation is probably from the same root
in fpatvw "make clear."
9 Schwyzer, Griech. Gramm., II, p. 365; Chantraine, Gramm. Horn.,
p. 304; Monro, Horn. Gramrm., pp. 199 f.

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions
 INFINITIVE IN INDIRECT STATEMENT. 181
would produce nonsense). The two infinitives (only the first
with a noun) are instructive for comparison in A 18 f.: Viv
sOLEV
ILZVBOol ... . K The
e 8' ocKaS'lKEaeaL.
KTEpaL lIptdauoto 7roALV
simple infinitive (without a subject expressed) after verbs of
ordering may also be included here: A 386: KEXoprv 8OEviXacaKea-
0aL. It is of course clear in all these uses that the original case
sense of the infinitive, dative, locative, etc., had been lost from
sight, and that the infinitive was regarded as a case-less noun.
The use of an infinitive after r,qtLIis of the same kind.10
We find ,qvu used with a direct object in the accusative to
denote the thing spoken of, or thought about (in distinction
from the words, or the thoughts, themselves). E 184: el 8' o'y
avyp ov <rkul, "if he is the man of whom I am speaking (or,
thinking)," B 81: t'evSo' Kev ail/teV, "we should speak of (think
of) a lie." The construction is often misconceived, by under-
standing an omission of elvaL. Similarly ? 153: e' e KaKOv fr)J-eL,
' speak of you as cowardly"; compare Soph., O. T., 362: rov'a
ae q/_ud;Xen., Hell., III, 5, 12: Koptvti'ovs 8 . . . rit (L?ev
(wrongly Underhill, "the construction requires some infinitive
to be supplied "); Eur., Phoen., 312: rt 4O a-ce;
[e. g. yLyvWT-KEtv]
Pindar has some clear examples, which have no attributive
adjective or noun, so that it is not possible to propose the
omission of elvaL or other infinitive. 0. 13, 103: rd r' raolpeva
/Ev ev Kovla XepacrWTa 8
rTr' av a4alrv raaes', N., 9, 42 f.: 7roXXAa
ye7roL 7roVTrW darotfLat. Compare N., 4, 91 f. The construction is
similar with o't -ot aL: so K380: i rtva' rov MXov XAAovoea,
N 283: K7Jpas o'iouELVW.
I have analysed all the examples of infinitive in oratio obliqua
in the first ten books of the Iliad, and find that after ytp? the
subject of the infin. is as often omitted as expressed: out of a
total of 41 examples, there are 21 with no subject in the
dependent clause. (a) Subject not expressed, 21 cases; of which
(i) 15 have identity of subject in principal and subordinate
clause, (ii) 6 have change of subject between the two clauses.
(b) Subject expressed, 20 cases; all with change of subject. It
is usual to omit the dependent subject when it is identical with
10 Since I am
presupposing that here too we have the infinitive treated
as case-less, this will support the view that the Greek infin. oratio
obliqua construction is a relatively late creation, and independent of
the Latin.

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions
182 A.. . MOORHOUSE.

that of the main verb; when it is expressed, it is for the special

purpose of emphasis,ll 0 221: eAC'(PrtU7roXv7rpobpepP(TEpov CvaL.
So with other governing verbs: H 198 f.: ov8' we vViS6ay' oVr(tw
XfAroLaL . . . evEcraL, I 682 f.: avros 8' 'rr'XrEctvre . . a. va
iEXKE,L (avrod must be taken with the infin., being pointless
with the main verb). What is perhaps more surprising is the
omission of the subject in as many as 6 cases with 4ut' in Books
I-X where there is not identity, and where there might be fear
of ambiguity. A 351: 7ra 8$iS' s 4roXfeoto EGte/Ev;, "how cal
you say that we avoid the fight?", I 35: '(as Cev aJrrroXeT/'o Kal
adVaKLta, "(you) saying that I am . . ." Similarly E 652 fE.:
aro 8' iyo) evad8e fr]/pt . . .t ' {r5 erTa evXos c/uoL 8a(c,tv
sovpr
8atp
(notice the lack of agreement between oot and the following
accus.); also r 44, 220, A 375. Also after other verbs, H 310:
aeX7rTrovTrearoov JevaL; compare M 66: oi;8rpt;aracOat O't, ' I think
that they . . ." and o 278 &SoKEegvaLyap o, " I . . . they." 12
There is a contrast here with Latin, in which the subject of
the infin. is expressed regularly, both with identity and with
change of subject. A notion of the comparative scarcity of omis-
sion for early Latin may be gained from Bennett (Syntax of
Early Latin, I, pp. 367 ff.). After dico he says that there are
several hundred examples with subject expressed. He claims to
quote the most important instances of omission (p. 383); and
after dico 15 are given. aio: over 200 with subject, 14 without.
censeo: over 100 with subject, 6 without. Examples of omission
are:-(a) with same subject: Plaut., Asin., 806: si forte velle
dixerit; Ter., Andr., 394: dic patri velle; ibid., 353: ait tibi
uxorem dare hodie; (b) with change of subject: Plaut., Amph.,
345: iam faciam ut verum dicas dicere (sc. me); Pacuv., 369 R.:
insanam autem esse aiunt (sc. fortunam).
The contrast supports the idea that in Greek the subject of the
infin. is a later addition to the construction; whereas in Latin
the accusative is an integral part of the original construction,

1 Chantraine, Gramm. Horn., II, p. 312.

12 This
aspect is incorrectly treated in Schwyzer, II, pp. 373 and 376.
In the first passage he says that the simple infin. is restricted to cases
where there is identity of subject; in the second, that the simple infin.
is indeed found with a different subject, but that it is confined to cases
where the subject of the infin. is indefinite (with two examples from
Xenophon). For Homer at any rate this is certainly untrue.

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions
 INFINITIVE IN INDIRECT STATEMENT. 183

both according to its traditional explanation, and in Professor

Hahn's version. In Latin we have an accusative and infinitive;
in Greek we have in essence a simple infinitive construction, of
which the accusative (and nominative) and infinitive is a
development. So the first stage in Greek was a sentence such as
,edtefeuEv. The next stage arose when the need for
7troXAfjoto
greater precision led to the expression of the subject of the
infinitive; but this was normally not required when it was
identical with the main subject. Verbs of wishing could simi-
larly add a dependent subject in the accusative to the infinitive
which followed them, when there was a change of subject: so
A 117 : 3povXo/L' ? 210 f.: OVKav oywy' eGEXot/t
iyo Xaaovar v !/JAevat,
Au Kpovowvt,dXeOatu cas rov' aAXXovs.The choice of accusative
for the case of the dependent subject may be variously explained.
Possibly the model for it was found in the accusative and
infinitive of KEXEV'WCae vat. Wackernagel would explain it as
arising from the use of the accusative to denote the person
spoken about.13 Or, looking further, we might compare it with
the use of the accusative and infinitive after 7rptv (7rapos) and
;aorc-all constructions of different origin, but agreeing in the
possession of a dependent infinitive. We may visualise a number
of constructions producing a situation in which the accusative
was looked upon as the fitting case for the subject of an
infinitive.14
A. C. MOORHOUSE.
UNIVERSITY COLLEGE, SWANSEA.

13 Vorles. iiber Syntax, I, p. 264.

14A brief version of this paper was communicated to the Classical
Association, at its meeting in London, April 1954.

This content downloaded on Wed, 6 Mar 2013 08:24:25 AM

All use subject to JSTOR Terms and Conditions