The Pi'W1 in Hebrew.-By Louts B. kins University, Baltimore, Md.

WOLFENSON,

Johns Hop-

IN the Semitic languages the great majority of words are derived from triconsonantal roots. There are a number of pluriconsonantal roots,' e. q., O (impf., Qal with suffix Ps. 80, 14), ' eat off,'

.Mt

'frog'; Arab.

qam'al ' sprout;

be chief'; Syr. *osh. 'hasten'; Eth. OMMflanbasa (YES ' lion,' etc.; but these are in nearly all cases 'anbas from ues)
derived from triconsonantal roots in various ways.' According to some authorities all triconsonantal roots are in their turn derived from biconsonantal roots.' The biconsonantal roots, however, they regard as altogether prehistoric, and all words in the historic stages of the languages are looked upon as derived from triconsonantal stems. The shorter biconsonantal form of the verbs I"I'3 and p7"V are thus considered to be the result of elision and contraction of triconsonantal ones.
'The number of pluriconsonantal roots in Hebrew is small; in Syriac the number is larger, while in Arabic and especially in Ethiopic they are relatively numerous. 2 For the ways in which these formations are developed cf. GeseniusKautzsch27, ? 30, p. q.; Dillmann, Aeth. Grain.2. Leipzig, 1899, g? 71-73, 77, 78; N6ldeke, Syr. Gram.2, Leipzig, 1898, ? 180; and the special treat-

bischen. Leiden, 1878. 3 Cf. Ed. K6nig, Lehrgebaude der Ilebr. Sprache, Leipzig, 1881, 1895, IL', ? 119, 3 b), c) (p. 370 ff.); and contrast Gesenius-Kautzsch27, p. 99, n. 1. It is most likely that originally all roots were not biconsonantal, but that there were also triconsonantal ones; cf. Delitzsch, Studien ilber Indogermanisch-Semitische Wurzelverwandtschaft, Leipzig, 1873, p. 70. 4 Verbs tj8r is used as a convenient symbol meaning verbs mediae u, mediae i, following K6nig, who uses also tl"O similarly.

Erster Theil; Bildungen durch wiederholung des letzten Radicales am Schluss und des ersten nach dem zweiten has appeared]; Siegmund Fraenkel, Beitrage zur Erklarung der Mehrlautigen Bildungen im Ara-

lautigen Thatw6rter der Ge'ezsprache, Leipzig, 1871; Martin Hartmann, Die Pluriliteralbildung in Semitischen Sprachen, Halle, 1875 [only the

ises of F. G. Schwartzlose, De Linguae Arabicae VerborumPlurilitterorum Derivatione, Berolini, 1854; Stade, Ueber den Ursprung der Mehr-

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

304

L. B.

Tfolfenson,

[1906.

This view of these verbs is that of the national Arabic grammarians, and it is supported by the many secondarily regular In Hebrew, however, which is in some forms in their language. respects more primitive than Arabic,' the conditions are different, and the earliest Hebrew grammarians and lexicographers of They believed that in the Middle Ages did not hold this view. Hebrew there are biconsonantal2 and even uniconsonantal roots etc. The explanain the case of certain weak roots like tion that the shorter forms of the verbs 11"37and V7";7are contractions, is based on the Arabic view, and was first introduced in Hebrew by the grammarian and lexicographer Ha -yy j4 living at Cordova, Spain, in the latter half of the 10th century and early part of the 11th, who spoke and wrote Arabic, and applied to Hebrew the principles of the Arabic language and the methHis view of these verbs preods of the Arabic grammarians. vailed until the last century, and is held even at the present time by such a prominent grammarian as Ed. Konig,5 as well as by others of less note.
n

Thus, e. g., in of their lexicons. the lexicon of Menahem ben Saruk we find the root Xty treated under the biconsonantal heading By; fl?ea and ?nW both under ft; frt, under showing that the ultimate root of many soAdd, and
ro:

1 Cf. Gesenius-Kautzsch21, ? 1, n. 2 This is shown by the arrangement

ro;

See the edition of called weak roots was considered biconsonantal. Menahem's Lexicon by Herschell Filipowsky, Antiquissirnum Linguae . . Hebraicae et Chaldaicae Lexicon . . . A Menahem ben Saruk. Londini et Edinburgi MDCCCLIV, pp. [1681a, [t7h]b, [14418, and cf. J. Filrst, Zur Geschichte der lebrdischen Lexicographie, the Introduction to his Iebrdisches u. Chald. Handworterbuch, Leipzig, 1863 (2d'ed.), p. xx. cf. also p. ; 3 Cf. the Lexicon of Menahem, p. [103]b, under 1 for root of [123]3, and [1271b for other examples of uniconsonantal roots

kajDroot ofND)
4 His views

on this subject are expounded in the two treatises called

the Kitdb al-'Af

at dawdt Ijurff al-Lin, and the Kitab al-'Af dl dawdt

al-Mithlain; see the edition by Morris Jastrow, Jr., published under the .ayyuj," title "IThe Weak and Geminate Verbs in Hebrew, by.... Leide, 1897, Preface, p. xi, and cf. FArst, 1. c., p. xxiv. 5 See his Lehrgebdude, I, ? 34 (p. 320 ff.), where the verbs 1?";7 are treated under the heading of Contracted Verbs. Cf. also Vorrede VII, and pp. 479-81 with pp. 451-53. Of course the question of the ultimate origin of these shorter verbs is not affected by this opinion. Thus both

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

Vol. Xxvii.]

The Pi'let in, Hebrew.

305

However, beginning with J. FTirst' and Ewald,2 there has been a constantly growing number of scholars who have regarded the verbs tlop and r"r not as contractions of triconsonantal forms, but as developments of biconsonantal roots which were not expanded to the triconsonantal forms as in other cases. Thus N61deke,3 Buttcher,4 A. Miffler,' Stade,6 deLagarde,7 Friederich Delitzsch,8 Zimmern,9 Kautzsch,'0 Wellhausen," and others have K6nig (cf. p. 303, n. 3), and Mayer Lambert, who believes that the verbs T and pep are contractions of triconsonantal forms (cf. his article " La Trilitteralite des Racines VIT et IV"J," in Revue des Etudes Juives [REJ], Tome xxxv, 1897, p. 203ff.), consider that these verbs MIT and Ah'IVas well as all other triconsonantal verbs are derived from original biconsonantal roots. Cf. Mayer Lambert's paper in Semitic Studies in Honor of Alex. Kohut, Berlin, 1897, p. 354-62, but contrast GeseniusKautzsch27,p. 99, n. 1. 1 Cf. Lehrgebdude der aramdischen Idiome, Leipzig, 1835, ?? 91 (p. 81), 153 (p. 158). As far as I can find, no credit has been given Fiurst for postulating the theory that the verbs t"r and p7p are biconsonantals, as his name is omitted in everything on this subject which I have seen, Ewald and Bottcher being the first scholars mentioned as holding this view. 2 Cf. Lehrbuch der Hebrdischen Sprache8, G6ttingen, 1870, ?? 112, 113. 3 In a review of Olshausen's grammar in Benfey's Orient u. Occident, I, 1862, p. 760 ff.; cf. Manddische Gram., Halle, 1875, ? 87, and Beitrdge zu semit. Sprachwissenschaft [BzsS.], Strassburg, 1904,p. 46. 4 Lehrbuch, 1866-68, ?? 1116 f.; 1127f. 5 In ZDMG. 33, 1879, pp. 698-700. 6 Hebr. Gram., 1879, pp. 109 ff., 138 ff. 7 Cf. Orientalia, II., Gdttingen, 1880, p. 6; Ubersicht, Gdttingen, 1889, pp. 26, 27. 8 Assyrian Gram., Berlin, 1889, ? 61, 1); ? 115. Vergl. Gram. d. semit. Sprachen, Berlin, 1898, ? 50 b, c.; ? 51 b, c.
10 Gesenius-Kautzsch,

Hebr. Gram.27, ?? 67, 72. 11Skizzen u. Vorarbeiten, VI, Berlin, 1899, p. 250-255. This article

Wellhausen says he wrote to explain especially the impf 's. of the socalled verbs ~"37, e. g. D j De Lagarde, however, had previously stated that these verbs were biconsonantal just as the verbs In 1880 in his Orientalia, II., p. 6, he says: " die wurzeln called l'p. 3 und By halte ich garnicht fur dreiconsonantig, sondern-seit jaren habe ich dies 6ffentlich gelehrt-fur zweiconsonantig," and in his Ubersicht, p. 26, 27: "Ich glaube, dass es zweikonsonantige Wurzeln mit [....]. urspritnglich langem Vokale gibt: AU-mit ai Ich filge jetzt hinzu,- dass ich
VOL. XXVII.

mit
21

parallel setze, etc." Apparently no notice Wellhausen does not refer to

has been taken of these statements.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

306

L,. B. Wolfenson,

[1906.

It is A. Mifller, Stade, considered these verbs as biconsonantal. and Wellhausen especially, who have most consistently carried out the biconsonantal explanation. According to their explanation, these verbs are derived from original biconsonantal roots with a short characteristic vowel between the two radicals, corresponding to the characteristic vowel between the 2d and 3d radicals of triconsonantal verbs. Under the influence of the prevailing triconsonantal types, the biconsonantal forms of the verbs 1"r and 17"? were usually amplified: the former, by lengthening the short characteristic vowel between the two radicals,' e. g., in MI the A is from
,Lagarde, nor does N6ldeke in the reprint of his paper Die Verba by im Helbrdischen in BzsS., p. 34 ff., although Lagarde refers (Uebers., p. 26 below) to N8ldeke's article (first published ZDMG. 37, 1883, p. 525 ff.), and N81deke accepts the explanation of Wellhausen (BzsS., p. 46). Lagarde's explanation, however, is based on the assumption that the verbs tl"Y had an originally long vowel between the two radicals (cf. above). This same view is held by Ewald, Delitzsch, and Zimmern (cf. 11. cc.). Wellhausen, on the contrary, correctly explains these verbs as derived from biconsonantal roots with an originally short characteristic is lengthened (under the influence of the vowel, so that the i in DOWN p longer, predominant triconsonantal forms) from i, just as the A in is from fi, and the 6 for d in NIf from d. ' This lengthening takes place usually in forms in which the characteristic vowel stood originally in an open syllable, e. g., Arabic I A ,eLS, Da0 i ?>Al~i, etc., from original qania, qdmdt, qdmfi; Syriac Ecu, , etc. But in Hebrew this lengthening

did not take place in forms of the Qal perf., trpn,, VZ), etc., being for qdm, qdmd (met, bs), with tone-long vowels, and so really= qam(a), qdmd (mnat,biks), etc., with heightening (not lengthening) in the tone. The forms of the Qal act. part. are the same as those of the 3d masc. sing. perf., hence qam, met, b6s, although the vowels are Cf. F. R. Blake in JAOS. vol. xxii, 1901, p. 51, n. 3; unchangeable. Wellhausen, Skizzen u. Vorarb. VI., p. 252; and contrast GeseniusKautzsch2, ? 72 g. Inboth Hebrew and Arabic an originally short characteristic vowel was retained without lengthening when it occurred in a closed syllable. Thus in the jussive and apocopated forms the original short vowel was not lengthened because in a form expressing a command or the like it was the corresponding desirable to have as short a form as possible, e. g., later tone-long 6 from jaqu'm,

C? =0qom with

p1=
I

uaiiaqm
1 q

in which the original it

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

Vol. xxvii.]

The Pi'lWl in Hebrew.

307

an original A, in t,: the i is from an original i, and in Rllr the o is for 4 from 4; the latter by doubling (not repeating) the e. g., DID from an original sdbdt, etc., the doubling being secondarily omitted in Hebrew in forms in which the final vowels are dropped, e. g., :I for sabb, cf. Arabic ? farra, etc.
with tone-long e from pagl, etc.; Arabic O Aazdl,etc.

second radical,

appears as 6, ,:z=1ag.l iaqMl, A,)

asir,

Also in forms having an afformative

beginning with a consonant, the characteristic vowel, occurring in a closed syllable with a second consonant immediately following the final radical, was not lengthened, no doubt on account of the firmness resulting from the juxtaposition of two consonants without intervening

vowel, e. g., Arabic ;

(with 6 heightened

qflmta, <>n binta; Hebrew FVj,

JaZ:
J

from U' in the tone): impf. .yJa iaqu'lta, iasirna, Ad tazdlna; Hebrew MnXjll with 6 heightened lengthened) from ui (if it were lengthened it would become A as in

(not

from iaq~m), :. with 6 from I, and n with 6 (not 6 as is T T T T ~~ stated, Ges.-Kautzsch2, ? 72 k) obscured from a which was lengthened from an original 6 on account of the quiescing of the 8 in an original taba'na. (cf. Ges.-Kautzscheq, ? 76 g) the [In the rarer J is also o for a; but here the d-as jltxp-arose well as the A in 1-711nl and the X in of an original short vowel,
T *

through the lengthening

6 (U, 1), under the influence of the prevailing triconsonantal type, since the root syllable is.no longer closed when intervenes before the afformative 71 .] In such forms as these, in which the characteristic vowel occurs in a closed syllable, some (e. g., A. Miller, ZDMG. 33, p. 699) are inclined to think that this vowel was first lengthened and then shortened again in a closed syllable, so that J.1 e. g., is shortened from *qdmta, which arose from qdmta. This is apparently supported by Syriac ia, etc., Ethiopic '1Yth qnmka, etc., with long vowel in a closed syllable. In Ethiopic, however, the long vowel in the closed syllable is contrary to rule. - Cf. Praetorius, Gram. Aethiopica, 1886, ? 15. The long vowel here must be explained. In both Ethiopic and Syriac the long vowel is best explained as due to the analogy of other forms in which the long vowel occurs regularly in an open syllable, e. g., Za.0 ia etc.; 7iao qoma, ;FaPir qom6t, 4jOW qomft, etc. The Hebrew forms like are then to be explained as preserving the originally short vowel unchanged, and are therefore more original.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

308

L. B.

Wolfenson,

[1906.

This biconsonantal theory is the most natural explanation of these classes of roots, and is the one most in accord with philological principles. For if the prevailing triconsonantal type of root is in considerable part a development from a biconsonantal state,' it is more than likely that remains of this former state should be preserved in the stages with which we are familar. In language a new order of things is a growth, the older existing at least for a time beside the new, and it is not introduced by unanimous agreement, as it were, of those using it. In all languages in which a growth can be observed a certain number of, older forms are preserved. These older forms appear irregular in comparison with the prevailing types. To consider the This conclusion is supported by the corresponding Arabic forms Ad * qgmta, .ix binta, etc., in which the vowels are also short. The shortness is original. Their quality, however is secondary. One would expect to find a in the root syllable, as in Hebrew. Wellhausen has correctly explained the , and I as due to the characteristic vowels A and i i Cabin Of course in the case of intrans. iaqam, in the impf. i 'cease' (. zilta), it should JAlo be long' ( .Jlo tflta), Jo) occasion no surprise that the original intrans. characteristic vowels are retained. N61deke, however, has questioned (BzsS., p. 46, n. 2) in this connection: Why, if J zilta is the intrans. form, do we not find verbs like

zdla? By way of answer it will be recalled that such instead of zAlla intrans. forms are actually found dialectically; cf. Wright-deGoeje, Arabic Gram.3 L, 1896, p. 83 D. In general, however, this form became the passive in the case of trans. verbs (cf. F. R. Blake's paper, The Internal Passive in Semitic, JAOS., vol. xxii, p. 51 ff.), and when this took place the act. form JA prevailed also in the case of intrans. verbs in forms in which the characteristic vowel occurs in an open sylqdmA, etc,). In Hebrew also the trans. type, lable (mold qdmdt, { prevailed in many verbs that must have been originally
intrans. Only lno, ei . Bi .and :5 occur as intrans. forms in the perf. The trans. form prevailed to such an extent that we find the trans. vocalization in the case of forms of nl, having an afformative and not ; beginning with a consonant, e. g., 1rrItD All the forms of the verbs tj"r May thus be satisfactorily explained on the biconsonantal theory. l Cf. n. 8, p. 303.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

Vol. xxvii.]

The Pi'lWl in Hebrew.

309

shorter forms of the verbs IYT and '7' 1 as contractions of are original biconsonantal regular ones is unphilological2-they forms preserved in the historical stages of the Semitic languages.'3 Accordingly, forms of these verbs with three distinct radicals In fact, in case of the roots are a relatively later development. 1'7 in Hebrew, verbal forms with consonantal' I and t are very
1 I retain the symbols t11' and pry as being customary and convenient, although they are, of course, inaccurate since there was properly no radical in these verbs corresponding to V in b;7. 1In the case of the verbs *18'y it is even impossible. For in the verbs that actually have I as 2d radical we find the I preserved as a radical consonant which does not suffer contraction, and that too in just those situations in which the advocates of the triconsonantal explanation of the verbs t'y, say that t and I were elided or underwent contraction e. g., nJJ I Sam. I6. 23, beside forms like in verbs 1 Anne,

as

beside forms like

t0 ?,

etc., etc. Of verbs with) as 2d radical there

occur (not including verbs

7 , lu, , and ;71P. Cf. below, n. 4. Accordingly there n is no reason why a contraction should have taken place in verbs J";7 if I had been present originally 'as 2d radical any more than in the above verbs. We must therefore conclude that no I was present. These verbs (i. e., the so-called verbs *5";7) were originally biconsoxiantal. 3 As N6ldeke pointed out as early as 1862, saying (Orient u. Oceid. I, p. 760): " Wir betrachten eben Wurzeln wie q4m, sab als werthvolle Uberreste einer Zeit in welche die Dreikonsonatigkeit noch nicht bestand." Accordingly these roots have only two radicals. In his more recent statement (BzsS., p. 46 below, 47), however, viz., "Alle historischen semitischen Sprachen behandeln hier doch die Vokalbuchstaben I und t als Radicale," there is not a little inconsistency. According to this remark, there are practically three radicals. This is open to grave objections in fact, as N6ldeke himself must needs grant, since he admits that the Arabic forms cited by him in support of his statement may be considered secondary. 4 Of course, forms of triconsonantal roots with l as second radical are formed regularly, and in these I appears as a consonant throughout, e. go Is. 42, 11; etc., cf. n. 2. The gutturals 7; M, T : - MlY, IM ~~IT: in these roots have nothing whatever to do with the retention of the I as a consonant, as Kdnig (Lehrgeb. I, p. 453) followed by Mayer Lambert (REJ., xxxv, 1897, p. 211) supposes, since we find a large number of roots 1"3<which also have gutturals as radicals, and yet no consonantal 5 appears, as one would expect according to K6nig, if it had once been

r n

n"5) in Hebrew P1

f,

Mm

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

310

L. B.

Wolfenson,

[1906.

few, being limited to Pi'els occurring only in the latest literature, so that they are really Aramaic forms and not Hebrew, e. g., 88p Ps. IIg, 61; Amp Est. 9, -21, 29, 31; IM-lp Esth. 9, 27, 31; -n-IM Dan. I, 10. In case of the verbs

7'T the tri-

consonantal form with repeated second radical appears regularly in Hebrew in the 3d sing. mas. and fem., and 3d plu. of the Qal perf. as the trans. form, while the more original biconsonantal form is used as the form with intrans. meaning,' e. g., 1TI 'make narrow,' but :9 'be narrow.' Also a considerable number of regular Pi'Vl forms from the amplified triconsoas we saw, are practically of non-occurrence nantal stem-which, in verbs V"V-are formed from roots V"Y, e. g., A

I-T,n. 1:D ApA,

etc.

It is clear from the foregoing that originally no Pi'6l intensive stem could be formed in the case of the biconsonantal verbs

and 3"p. since the Pi'e1 requires three radicals for its IV"37
It is not until these roots have been fully assimiformation. lated to the triconsonantal form that the Pi'dl can be made. In actual fact the Pi'el of verbs 11"r, as was stated above, does We find in its stead the Pi'lel. not properly occur in Hebrew. Also in the verbs mediae gerninatae the Pi'lll occurs as the intensive stem, although some regular Pi'els are found, e. g., etc. With regard to the origin of the Pi'ldl there have been a In general, opinions as to its number of different explanations. origin may be divided into two main currents according as it is supposed to have arisen independently in the verbs l"r and p7V ,2 the agreement in final form being then accidental, or it if I was second radical the presence of a guttural in the root did not can, thereprevent contraction as is supposed. Roots like r ( fore, be explained only as a separate class distinct from the roots lyp We cannot otherwise account for the difference in meaning between two roots, otherwise identical, like bl3, ('!lu7 Is. 26, 10) 'act unjustly,

present,e. g,,

l11j (two gutturals!), etc., provingthat .37fit,

'suckle '-the former is tricorruptly,' and blr (part. plu. fem. jllt) biconsonantal. latter the middle radical, consonantal with I as ICf. Ges.-Kautzsch27, 6 67a, ? 2; K6nig, Lehrgeb., 1, pp. 320, 321. 2 Thus Bottcher, Lehrgebdude, ? 1016, ? 1030, 2; Olshausen, ? 251 b, 252, 254; K6nig, Lehrgeb. I, pp. 451 and 349.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

Vol. Xxvii.]

The Pi'lMl in Hebrew.

311

arose in the one class and was transferred to the other by analogy. The latter view, that the Pi'lMl arose in one class and was transferred to the other, is undoubtedly correct whatever It is the view followed by explanation is otherwise adopted. Ewald,' Hartmann,2 Stade,3 and Barth,4 Ewald and Hartmann believing that the Pi'lMl arose in the verbs P"P, while Stade and Barth, although differing in other respects in their explanations of the form, believing that it arose in the verbs 1"rStade's explanation, that the Pi'ld arose from the Qal stem qama by reduplicating the final radical in order to indicate the intensive stem, producing qa4rnma, qdmrim, qomem, is untenable especially because the Qal stem is not qama, but qdma. The long a becoming 6 in qomem is thus unaccounted for. Barth's explanation, which is based on the triconsonantal theory of verbs l"Y, has been accepted by Kautzsch (GeseniusKautzsch, Hebrew Gram.27, ? 72 m), although he follows the

biconsonantal explanation of the verbs T'? (and r"r,

? 67).

There are, however, certain difficulties in Barth's exposition, apart from the fact that it is based on the triconsonantal explanHis ation, which render it impossible to accept his theory. explanation is briefly as follows. A weighty indication that the Pi'lWl (P6o'll) did not arise in the verbs P"7 is the fact- that they can and do form a regular , etc., in Hebrew as in the other languages. Pi'el, e. g., In the verbs V'T, Pi'el forms do not properly occur because of the difficulty of pronouncing an intervocalic [my italics] sharpened waw [as if a doubled I could be anything but intervocalic!]; only the Pi'lWl with reduplicated final consonant, as in is found. The reduplication of this final consonant in this stem of roots T'. is supported by a similar reduplicaIn Arabic tion in certain nominal forms in Arabic andc Hebrew. no verbal form with this reduplication is made in verbs I"p,only nominal forms occur. These nouns, in which no intensive
5, no _

meaning is present, are the peculiar inf 's. like Ad.

I
2 3 4

' go away '

Lehrbuch8,?? 121a, 125a.

Op. cit., p. 2, 3.

Hebrew Gram., ? 155 c, d. Die Poll-Conjugation und die P6lal-Participien in Semitic Studies

in Honor of Alexander Kohut, Berlin, 1897, p. 83-93, especially p. 84.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

312 (i c)
ios

L. B. Wolfeiison, Ned.i), ' rule', G 'be' 'be

[1906.

LS med. u), etc.; the two infs. pregnant'; and the broken plurals

'Cpregnant' and I.ye 'barren for a long time (eamels),'

all from roots med. waw and med. yod. In the case of roots med. geminatae no corresponding formations occur. In Hebrew TVVs 'spark' (Arabic c)ts the nouns mans pleasure', (V n) Only have this same reduplication. 'emit fire') from roots l-VW root a from is This from V"P. 'sparkle' 'spark' Except may be formed on the analogy of its synonym ni'y.

r,

rm

for this single instance, formations in Semitic corresponding to the Pi'ldl are always from roots 11"7. The same result is, according to Barth, arrived at from a consideration of the Hebrew participles :1')e utr, with which he says DOntl and the uncertain Chid are connected as regards On '~Je he lays little stress, since its meaning, and formation. These participles are not intensive hence its root, is obscure. in meaning, but are simply Qal. Trans. in force, although having an apparently pass. or intrans. vowel - in the second syllable, they are really qattd forms represented in Hebrew by nor etc., in strong roots; by 1fI, All in roots med. i. :zs Accordingly from roots med. u of which no qattdl form occurs A " sharthe original form of these participles was qawwdm. " " sharpened a than more in Hebrew w avoided " being pened reduthe by was replaced being doubled w up, the given j [?], plication of the following radical, so that qawwdmn became qawmdm, qdmdm, the change of aw to 6 being similar to that

in 1111 from IVry

From Like these participles the Pi'lel is to be explained. for X with been have originally 4 must the qawwam Dlj1 qittel The intervocalic sharpened w was avoided in the first syllable. by substituting the reduplication of the final radical so that from qawwmrn, qawmam becoming q6nzma arose. That In this explanation there is little that is convincing. etc., have in Arabic a reduplithe nominal forms like x cated final consonant like the Pi'lVl proves nothing for this verbal stem in Hebrew since the origin of these forms is obscure

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

Vol. xxvii.]

The Pi'lll

in Hebrews.

313

and a matter of dispute. Barth's explanation of them is no better than that of the grammarians which he rejects. He says' e. g., arose from beside which it occurs, that R * This was because the phonetic sequence -upi was unpleasant. obviated by inserting a consonant identical with the last one after the i so that we get 1inilnru. In the first syllable of this form, A was changed to a, and thus b1ingnat arose. The changes that Barth assumes here are all unsupported by similar phenomena elsewhere, and are therefore entirely gratuitous. It may be that the forms like arose from an analogical com-

like bination of the two regular infs. be true there is no organic reduplication

? and If this in these forms. At

any rate there is nothing in forms like X about the origin of which nothing is really known, that is like the Pi'lll, except the reduplicated final radical. Similarly the isolated forms
,iaJo2

Ajax prove nothing.

In Hebrew the nominal forms inns nine. riser are entirely too few to base any conclusions upon. Since iSj is from a root 7"P and there are only two other examples of this formation, it is just as possible to conclude that the formation originated in roots Y37'3 and were transferred to those med. u, especially since the root of 8TVT' which occurs only in Job. 41, 11, does not occur as a verb in Hebrew. '2' DVtLikewise the few forms l l, prove j and '2'2U T T T T nothing. Barth himself attaches no importance to Chits; It is most likely derived from a root . is usually in" D explained as having the adverbial ending D-T cf. Ges.-Kautzsch", ? 100 g; Gesenius-Buhl"4, s. v.2 There is no compelling reason for regarding DC~Vet as anything but an adverb in the three passages that it occurs. In DtOt fOl Hab. 2, 19 we must connect DtOVl with what follows, according to the suggestion made in the latest (14th) edition of Gesenius' Dictionarv. In Lam. 3' 26 it is difficult to see how not to make DOV-1 an adverb. AN.
1 Die Nominalbildung in den Semitischen Sprachen, Leipzig, 1889, 1891, pp. 211, 212. 2 In the 13th ed. Barth's explanation was given.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

314

L. B.

Wolfenson,

[1906.

translates, " It is good that a man should both hope and wait In Is. 47, 5, DtIOl is clearly an adverb. quietly (D?!T), etc." Moreover the change of unexplained.

I to .l in DC1nJT which Barth assumes is and 93J21 This leaves only the two forms nitt

This would be precarious upon which to base any conclusions. even if Barth had correctly explained them.- But granting that they were originally qattdl formations as he says, there is no support for the supposition that the " intervocalic sharpened w," which must originally have been present, -e. g., qatowdm, in verbs " med. w," was any more unpleasant to the ear or difficult to pronounce in Hebrew in case of forms. " med. w"

than in forms " med. j," e. g., AlT

V

. We find a considere. g., '2Tb' 1p -

able number of forms with doubled l,

etc., and in verbs 3'0 the first radical l is regularly doubled in the NViph.imp/, imv., and inf., e. g., po7d, etc. In the case of forms like AVp, 1VIy etc., Barth tries to obviate this difficulty by the remark' " Wurzeln mit durchweg cons. behandelten w, There is, however, no reason gehbren nicht hierher." . . . w gehbren nicht hiermit . . cons. . "die Wurzeln why her." There can be no difference between original w in qawwdm, if such there was (which the biconsonantal theory denies; cons. I that appears in roots l".V is secondary), and that in .pW I5M, etc., where I appears everywhere as a conso2Wp, nant, cf. Konig, Lehrgebadude I, p..453. The case of the Xiph's. Moreover even though a like V37-1 Barth does not consider.
-T-

doubled w were objectionable in Hebrew, as the preceding shows it was not, there is no parallel for reduplicating a radical The in compensation for the lack of doubling in another. approved method of compensation for the omission of doubling is to heighten the short vowel preceding the doubled consonant, as is done in countless instances in the case of the article, the Niph. impf., imv., and inf. of verbs primae gutt., e. g., T 71, etc. Accordingly it is impossible to assume that an original qawwdrn became qawmem. Similarly the Pi'lel stem cannot be explained as coming from an original qdwwem (qiwwum) becoming qomem. In fact it is ' Die P6ll-Conjugation, p. 90, u. 3.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

Vol. xxvii.]

The Pi'ldl in Hebrew.

315

absolutely impossible to assume such a verbal form as qawwam in roots ITT, as it was shown above that these roots had no middle radical. Consequently no form like qdwwem, etc., ever existed in Hebrew; the forms 813p, D' j etc., with doubled I andy, are not Hebrew. We must, therefore, look for the origin of the Pi'lll elsewhere, and it is in the verbs 7"r that it is found, as Ewald and Hartmann saw. It is not necessary to assume with Hartmann that the Pi'lll represents the III form of the Arabic. Ewald explained the o in | e. g., as due to the obscuration of d4. This d arose from 4 in compensation for the difficult doubling of the second radical in N To this explanation Barth objects that it presupposes an 4 after the first radical of the root, which does not occur in the Hebrew period, the form being always sibbab, although he assumes 4 for i in his own explanation of q4wwam for qiwwarn. See above, p. 312, ?t 2. In the imperfect, howvever, as well as the forms agreeing with it in structure, viz., the imv. and inf., the regular forms are '=D', etc. Here the original 4 after the first radical is retained throughout. iesdbbabeaccordingly would become iesdb1b, with a lengthened from 4 in compensation, as soon as the doubling of the second radical is given up. The d is then obscured to o, as frequently in Hebrew, and hence the form D From the 6 impf. the was then transferred to the perf., e. g., nn *2 That there is a tendency to avoid the occurrence of three identical consonants in two successive syllables as in the Pi'al of verbs Z".V, not only in Hebrew but in other languages, is seen from the fact that in classical Arabic beside such forms as
.ir,

Q5A

etc., with doubled 2d radical identie, ;5,

cal with the 3d, we find ma.>LJ 3d radical replaced by the diphthong

etc., with the of the

4W, on account

1 So also KEnig, Lehrgeb. I, p. 349, in the case of verbs Amp; Bickell, ? 116; and Land ? 55 (two latter quoted by K6nig). 2 In this explanation I have followed the principles established by Prof. Haupt, viz., that the impf. is older than the perf. (cf. his article in Jour. Royal Asiatic Soc., New. Series X, 1878, pp. 244-252), and that the origin of verbal forms is to be sought in the impf. as the more original form.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions

316

Wolfenson,

The Pi'ld

in Hebrew.

[1906.

Cf. Fleischer, Kleinere Schrif ten "heaping up " of consonants. I, p. 138; Wright-deGoeje, Aralbic Gram.3 I, p. 69 C. In modern Arabic even the simple Qal forms like Kayo with repeated

consonant are given up, and forms like ~x.Jo only are used.' In fact modern Arabic goes even farther in the case of such forms, using the form of
_

verbs
o~
_

tertiae ;
_@

instead

of those

for %zXio. mediae geminatae,' e. g., vA of is therefore formed on the basis of the Pi'l The Pi'll verbs 37'37;the doubling of the second radical is given up on account of the tendency to avoid a succession of three identical consonants in two successive syllables, and the preceding short d is lengthened in compensation to 4, and this is further obscured to o. The corresponding passive form, the Pi'lNl, has 4 in the N)O. The indication of the dissecond syllable, e. g., :N'J tinction between act. and pass. by i (a in Hebrew) and a, respectively, is regular in Arabic in the impf., not only of the intensive stems II, III, but also of the IV, VII, VIII, and X
, 1 "",9 o _.,

forms, e. g., II form

a act., J

pass., etc. to the verbs

From the verbs V".r the Pi'ldl was transferred

IVP.
1 Cf. Spitta, Gram. des arab. Vulgardialectes von Aegypten, Leipzig, 1880, p. 216.

This content downloaded from 146.232.129.75 on Thu, 27 Jun 2013 18:36:36 PM All use subject to JSTOR Terms and Conditions