FRAGMENTS FROM AN EARLY KARAITE

GRAMMATICAL TREATISE
Geoffrcy Khan
Numerous texts have come down to us attesting to the grammatical activities of the
Kaaites in the Middle Ages. These are all concerned with the Hebrcw language of
the Bible, which occupied a central place in Karaite scholarship. In this paper I
present an edition of some hitherto unpublished fragments of a Kaaite grammatical
teatise. lt is great pleasure for me to devote it to Tapani Harviainen, who has made
many important contributions to scholship both in the field of Kaaite studies and in
that of Hebrew grammar.
The grammatical studies of the medieval Karaites is divided into two phases, which
can be labelled 'the early tradition' and the 'the classical tradition'. The eady
tradition mainly appeafs in texts datable to th tenth century. The most extensive of
these that has been preserved is the Diqduq of Ytisuf ibn N, which is a grammatical
commentary on the Bible, Fragments of various other texts tht are very close in
methodology to the Diqduq are extrant. The classical tradition is represented by the
grammatical works of the eleventh century scholar
'Ab
al-Faraj Hrn ibn Faraj and
a number of dependent treatises written by other scholars in the eleventh century. The
works of
'Ab
al-Faraj continue some elements of the earlier tradition, but mark a
radical break from it in numerous features.l
Ttrc Diqduq of lbn Ntl and other texts belonging to the early grammatical tradition
are mainly concerned with the analysis and explanation of word structure. The
pronunciation of the language and its syntactic structures are only of marginal
concern in these works. Particular attention is given in the texts to problematic
grammatical issues, which are generally refened to as mas'il (singular maala
For the Diqduq of Ysuf ibn N see G. Khan, The Early Karaite Traditon of
Hebrew Grammatcal Thought: Including a Critical Edition, Translalion and
Analysis of the Diqduq of
'Ab
Ydqb Ysuf lbn Nt on the HagioSrapha (Bll,
Leiden, 2000). For'Abi al-Faraj Hn see G. Khan, Maria Angeles Gallego
and Judith Olszowy-Schlanger, The Karaite Tradition of Hebrew Grammatical
Thought in ts Classical Form: A Critical Edition and English Translation of
al-Kitb al-Kfi fi al-Lufa al-(Ibrniyya by'Ab al-Farai Hdrn ibn al-Farai
(Brill, Leiden, 2003) and the literature cited there.
I
Sndia Orientalio 99 (2004), pp. l2l-l16
t22 GEOFFREY KHAN
'question'). These usually emerge from a comparison of similar words and forms that
differ from one another in small details, a feature of methodology that is reminiscent
of that of the Masoretes in their annoations to the biblical text. This activity of
concentrating on the fine details of the language of the Bible was refened to in the
early texls by the term Diqduq, which did not yet have the sense of 'gammar' in the
more inclusive sense that it acquired at a later period. The early grammarians
discussed the mas'il at considerable length. They frcquently cited various altemative
explanations for a particular issue. some of these may have reflected the ditrering
opinions of scholars who were astive in the Karaite grammatical circles in the tenth
cenhrry. It is likely, however, that in many cases the proposal of such altematives has
a pedagogical purpose and the writer ofthe text is using them to invite the reader to
explore a variety of possibilities. They are sometimes presented as responss to
anonymous sttements introduced by the formula 'If somebody were to say ...' (rltt
qta qtilun).2
The fragments published below, which come from an eady Karaite grammatical
treatise, are concmed with the sFucture and derivation of segholate nouns. The
treatise resembles very closely, in structue and methodology, the fragmentary texts
that I published in the volume Early Karaite Grammatical ?exls (Allanta, 2000) as
Text I and rext 2 (henceforth refened to as EKGTI and EKGT2). These two texts
concern verbal conjugations and noun pattems respectively. In the aforementioned
volume, I noted the similarity between them in format and methodology and
suggested that they may have belonged originally to a single grammatical work. The
passage presented below, which comes from a manuscript that I discovered in the
second Firkovitch collection (The National Library of Russia, st.
petersburg,
II
Firkovitch Evr. Aab. 12771) after EKGTI and EKGT2 were published, appears to
confirm this. It does not overlap with EKGTI and EKGT2, but seems to belong
together with them. It somes from the beginning of the section on nouns, which is
missing in the published fragments. What is of particular interest is that in the extant
manuscript the section on nouns is preceded by the end of a section on verbal
conjugations. This provides the missing link between the two previously published
texts and indicates that they indeed originally belonged to a single work.
2, For further details see the works cited in n.l and also G. Klan 'The Karaite
tradition of Hebrew grammatical thought', in W. Horbury (ed.), Hebrew Stu
from
Ezra to Ben-Yehuda. 1999, Edinburgh, pp.l86-206;
.The
early Karaite
grammatical tradition' in J. Tgona Bonris and A. senz-Badillos (eds.),.Iewis
studies at the Turn of the Twentieth century.
proceedings
of the 6th EAJS
Congress, Toledo, July, 1998. Volume I: Biblical, Rabbinical, and Medieval
Studies, 1999, Leiden: Brill, pp.72-80.
Fragments
Jrom
an Early Karaite Grammotical Trealise t23
The author of the work is refened to in EKGTI as Sa(ld. This is likely to be the
scholar Sa(id Shn, a grammarian who was pupil of Ysuf ibn N.r There arc
indeed many similarities in methodology between this work and the Diqduq of [bn
Nh.' The arrangement of the material in lhe work, however, differs from the Diqduq,
Whereas the Diqduq is presented as a biblical commntafy, arranged according to the
order of the biblical books, the work attributed to Sa'id is ananged according to the
morphological pattems of words. Verbal conjugations and nouns are categorized
according to their pattern and presented in a series of chapters refened to as
'alphabets'. These contain an exhaustive, nor neaf exhaustive, inventory of verbs or
nouns with the pattem in question together with sampte paradigms of inflected forms
and a discussion of problematic gfammatical issues. It is the exposition of the
problematic issues (msa-if) that particularly preoccupies the author. Some chapters
contain more discussi on of masdil than others. In the fragments published below the
author identies various such issues and examines these with considerable prolixity.
The fact that such attention is given to masdil indicas that the work was not
intended as an elementary gammaf of Hebrew. The exposition of these issues may
reflect debates that took place among the circles of Karaite grammarians. They were
also intended to stimulate the reader to be inquisitive and independent minded in the
investigation of grammar.
One of the main theoretical concepts that is reflected by the passag below is the
notion of morphological base. Inflected forms of nouns are derived from
morphological bases. These are not abstractions, but real morphological forms. The
Karaite grammrians did have a notion of an abstract root consisting of radical letters,
but only used this to categorize words in the lexicon. The primitive base of any
derivational morphology was a form with vowels and consonants which was used, or
at least could potentially be used, as a real word in the language. The morphological
base of a noun is generally the singular form with no pronominal suffxes. Inflected
forms of the same morphological pattern would generally be expected to be derived
from bases with the same pattem. In some cases this analogy leads to the positing of a
noun base that is not attested in the Bible, even where another singular form is
a6ested. The base of the plural form nlllil, for example, is said to be ;11il, which
is not attested in the Bible, rather than the attested form
f,li,
since this is analogous to
n1it?
-
nll'iiD!, etc. (fol. l0 recto). Likewise, the forms DD? and o1)1i@rov.
28:6) arc said to be derived from n!1 and
!'!J,
rather than from n!': and
':1:,
by
3. See Khan, Early Karaite Grammalical Texls,p'17.
4. See Khan, Early Karaite Grammatical fe.rtr, pp. 13ff. and 175ff. for futher
details.
t24 Gtorrnrv Kruu
analogy with t, which is derived from
l)f
(fol. 9 recto).'
The manuscript is paper, containing between l8 and 20 lines on each page. various
sigla are used to fill in spaces at the end of lines to
justi!
the left margin. on a few
occasions the final letter of the line is extended horizontally. The end of sense units,
usually consisting ofa series ofclauses, are marked by the soph pasuq sign ofbibtical
manuscripts (:). Most Hebrew words in the manuscript are fully vocalized. The
vocalization is generally accurate, though the scribe from lime to time vocalizes
incorrectly, e.C.
'PF
for
tjl![
(fol. I recto), or at least in a way that deviates from
standad Tiberian vocalization, e.g. llJ for t;tJ (fol. 3 recto),
ti??n
for
rfi, r
for
'4
(fol. I recto).
Texl
II Firkovitch Evr. Arab.l277l
Fol. I verso
nr NN
'n
i1'tyf
1n "'rxt
ll!NynN |JNN
ril yDN
N
iN
:NilnflNt NI'ND
yital
nnnxxnx
'l? Tl D9 Ng
B
rglt lF r?l n?l
u,]
a
l9l n]n
op Dg? r
rlr
yql:13
DB? ?T p't r'lp r? y p ntg rrg

y
n?
l?a
pr?
or:r? g?
rrP,l xnJ N'PDN
r
trn! NrroNN ilt
rl: lx
rN :l?0 t
9
?g
y
r
fNn)x Nl
r
nnlnD Ti'r r r ]N:ntnN xD
r
''
Ntt :t
!:
xn:nt
'f
rrl
n'r' xx
:rntl
lNrN
"
nr) r{N:N:N Nln 'i
.'t
N:N tit
?'tl? r'll?
'll?
orf o'.rl l1l
l?.ll l1l1
oll fi
l?
o?Jl t.r
l?'l? l''!l? l?'ll lll?
:D'l'lll r'll? l'll?
5. For further details conceming this pect of grammatical theory in the early
Karaite tradition see C, Khan, The Early Karaite Tradition of Hebrew
Grammal ical Thought, p.39ff .
6. This scribal practice, which is bonowed from Bible manuscripts, is found in
several other maruscripts of Karaite grammatical texts; see G. Khan, The Early
Karaite Tradition of Hebrew Grammatical Thought, pp.l52-153.
Fragmenlslrom an Early Karaile Grammatical Treatise t25
Fol. 3 recto
Nn
l
rr
l FDpl l>'N
n) 'rr Nrx
:
rn xnb>
N
N:p N n xrr
n)N
1tirN
p)' nn nn
r
:
fr: )r
Ni'N
':
'rx
N
t)x
rv' xi''x
!:
"rm
:D) fln rp'Dr
lxn
nrtnr: xnn ilD) iltnxr nryf
Nr'r nr xrrrt xtrnx' l{tg > 'xp rrr
xo
:p nn xrnr

n x nn
rr n Nir
ll!' fi
nn" ir xr' :1rp
1ru'
Nlil
fpt
gl n'urr ilnN
r)
!r :,?l'rx
lr
'rr
i:
'nx
rgx
n
p: xr
Tt
:il Nc'$ , ny"rx
fnl
:'t l? D-r u na nno nvn n xupr
r
I
N > ox ilt
la lN
v xn r xnlir ly p: yB oyrt y
llrl
oyt p ntt l:r
n
tt
:
D:-1
iD rt xat :o?
1
onx n'uyr pr o! n:n
''r
[? xx xrrn
1nx
xrn
'
ornluln p> on n xlx
xx
:o"nnnr n p Dn o['t
lD
Nt'ul x nn o
lil:[
Fol.3 verso
l?y 'r?y
bnn or1p) D[tl
n m
rr :orir'!

l o'l ry; n] xr rr na on
l
n'o or nn: yr n nx )t
r
m
na >r nJ ug nnr o|f o!
fnp)
x'x nx:
INr
I'n nnnnt
yT yg
r?I nn nrr nlt, nl n
1x
nn
n
nrx uot
'
N r nn':o ilil
lilN
nyp y il'il,t iut ngt.r n$ n
;p tnn'J? n nn: nx DNt
l)'
rxuxx
:vr
1:1s
Nnt upr NIN
lf
'
rn?
nn r nnr n
'ix
Nilnnr nlul nnax
y
lpt tr it'
u
'Dt
nn rn nn'
'rN
r1r nn I'nl oln il?y
llly?
{lTE? irl}rr-.l1,
r
1n nn xnoxx iltil
lt
:1r na 'n) N
'lyl
uyn
n
'Pft?'?n ''l
.-nj r1,
1or
rlt
u
'Drt
)
'Ilty
nn urr: nj iij
a
nn:x r:x
xnnr :U
':r
'r
')r
n
ryl
:"1
'.lrp 'ptf
nn ngp l;1 Nr Nir
ngp n5 ii
n
nn:N l!
r
'nx
I'n)
Fol. 7 recto
1vr
rvr
u'rr
l)r
n\rt
"
N'rN';!p bnn
n)'
NN
'ix
r xx Nlr
'
on)xl
nn u'l )ayr )
')n'ra
nn n5 ii
n
n)N
r
xr: y' rl :lr'
n'Nl N.tl :rurJ' lN xr
'y
n>r ttrr':lly.i:)l'?l tIp,:r
"l,'pi?'??
5
t26 GEqFFRE,y I(HN
n
r"ll9
xt 'rrx
nno.' rhr pvx np;.rlx nr
l
ir NuN nD
r.rp
f,,nn
n
''r:
m N'rr
'
oxr :u)rr n ox xrn
J
nne*ix
'?,'1gD
1'r'o
:xur :ilN{
1x:n
ur tll: ol nnr or :x l)N
r
ni ii p
fr lnx
i n'o nro tl'rl :n ii
n lrnn
i il' lNs N ilDJl Ny
lll'
nnr
nx
ri
nli
1xb::
nor xy ni iii
n
qrnn i rul nr
n)' lt{
n' n ii:
nnnl u'n rn'ir tlDtl ?lr
1x::
nor: 1lD
1>:
nnnr "'.1
rrr nJN nr irl
1x::
urr: nr : i! ign tlln lngp toyl tnpt tl
x
olxt :r:xx ry: rnxrx n5 ii
f, lnx
i n' lry'rx r nrn
1n::
ox nr Ntt'tp
Fol. 7 verso
nnn ilnN nurr: au'Nt p ltpr i'y nrnyl nn:x ilNtrn nr NtN
NnNnDx.t i:tt 1t 1nt llrrt tyyl 1lU lnur
,hrrx
nn nnx
n
rxrr
1rn
:tltDt rrnr xnnr
r)il
')JN ')l
ir..! 1: ttlt nn ny'
;x
oozlxr :r'n:
ir
r
"rn
rr"ln
'z?tt'??
n?Tf p
'?$
nn uzl
i:
au'
ix
oozxr
xp
tto
:r
H
nil rxnnnr l"n xnxnnxl r,ll
'L?y '.Jy
u1y'r1 ur
nr ilr
yrnx'l
lrprx
n )nr" RitNl x )'xonr'o x lr ltrr rNp
lp
{rN oyn"nn pr
pnr
xa noan uiu
lrur
x
ar
nux
1rt
rn noan
xrrt'u
1rr
Nil :il xx oyl il:e nnnN noax
yyu
y
no:xx
:$rlt oyx li onx oux
x
rx oy oyl r>x
'u
il:'y nnnJn tp
fJ
nnl 1vlt'YlB 'rxv
'rP xr
x fNx
:rre nxr
t>'
noan
lrvr
x
ar
Nn' D' nnt ynil
rN
ilt lD: )tln'p?X xrn n )xpr '
rrx
un'lx n
n:x oxl'y
rrx
l$x
;lirr
lD
'xp rp
l
nn: Nnill urr
Fol. I recto
'y
o'b'ix mrn xi'xl nno xnn o'r qxr xnn lnv tJ i )x2,' nnr tg
na u:: i: rnr
1nx
ox'y'ix x'xl l)l 1lr nn nn rnr
1x
noxr
'
)N
p:n na: n'r
1)N
N'rt
'
rn o'
1:x
orrxr ?lf o??U p
'!$
nn xx nnx ox rrn x nn Nnrl
' 'nx
na > >l :irnf Niuo.l
;r' o'f iyB
'rpl'h'L
nn
1ru:n
tri nno: lN t1r: rn in nn qru:n
'lltt
''P?F
'?
nn non>n
'J 'r'ti'r
vn xoxr
'y 'nn
nn>n :t):
Fragments
from
an Early Karaite Grmmatical Treatise
t27
'lPF
Pi?D rPln
Nil'N): 1!t'?TIr ??? u l:10 :or::l
''lltl
N):,1
rp ir Nniyt iln) nDp Nirtt n ' NtN y
rnn
on op?y
l??y
1rJ)9 tngl nNlt
' '?D '?q
ttl
nn nn Nnyl prpr
g
na
na pyn o':lx Dl
'it
rnx
ox :rl :mx Nn
r
oa x'm '?pl
'?1xnnnt
E't ln' ;'!n'r
1x>tt
;"t?'19f!
'l?l
l')?{ 'lgn l'.lt? l'.t?f
'lS 'lgf!
l''tg4 l'l?
l"l?y
lN
olnr :1r ;'1u
nn'n
ll
y' or
D'!t
'it 'nx
:'!r
E
lg4
o'll? o'.rr na
n,n
xnnnaxr
Fol. 8 verso
l"lgl l"ll?
nn
1!r
nxr o'|r p, nrNr lr o:!r,l
n
rn
: D'?l
fJ
'n
no
x 1tt 1nx
rn o'l
''l
l1 1":
l'rl
lonr rlbrrx Nt'Dn
lt)
t'
l
tn
1r
nar
1'n l
ln) ny' x :'!nr
:1'?n xi'n ry'l
fnx
nu'fll
bttr rh :rn
ylx
rux
zi
lryx yp r'
lN xx
nr.r

)N ny'x
1x>n
rn 'lryr ilyD )1xrx
x
F
l
lntN rx lr urur rx rr.r ur-lr
tN{ ;llrlN';l no' rrr
yr
rurr xr NIN ntN
1l
.rr$x
ily: t.' x x
'tNtN tfN
'l
on v:lx l rux
ltril FltD
nnnt N ilytN
rit
r{rN
al'
'
t'!r :'fr
nn
x
x:r
xx r
xit rr
rnu
ngp rnr nnnr
r:n
x rryr xyn r x x'rn;'!1r?f o:?tf
>xh
rnx rr !r1 r
rn:x n
I'n)n
''t
r:yx xnyn l Rrlix o:t
n
nir5xn xnr lr
l
nllt
)t' ND r:y xx iln: lnxl
l
nN
l'?f
ltlf o:.t.1n N.1" n' nl x.rxl
nm rr
rnn
ND;l I'Jlp ltl':rzr xnnnr :'tlt
frlnrx.i'r'ix.lrylx
NrJityD
n)H
Fol.9 recto
lr : ot :n':rp xi'x nynr n:rrr nun 'lt
lN
{rru,
,t'll?
n:xr
n:n
p
nxuxx
'trn
r)?
'n
ir:tl o':lxr

xnsrrn)x r':p
x frr
nrnr arrn
l'nJn
l
ln)N
ln
n o'17
n
n'nX
I'J'l?t :ltrNyl
I'nJn lD
nyn r'::
nn
?)'lp
lN
'y
'rxl :lnxt{ llx
'n 'n)x
r';17 nJ
'n
IrJn 'l
:t?
tN
nftin n':ux
l
nnr nnxil nlnl: 'ltpn o11: r)r:xn>x rz: rrn xnn
7. The MS has nyrl afler this word, which ppears to be an enoneous copy of
what appears in the next line.
128
GEoFFREY KIAN
yl?
lrJ '1
?yl! n'N N'rNt 'l
lr
'!n nn
nP n
rit
o:li? n:llnx ntv
'J?
llp l
orr
1x'1rr
nrrt i
n
.'in
I'll?
1x
o)tn p'illt :,lt?
lD
o'p na
?.:!
ln p
na
P
rm
Ri'tl't rl'D
'rn
'll?
n
nn r''p nun orlP
lt't
1'J? Nil)t
rn
rnn
NDit D'1? x'ur xi'r$ :!'!! rl,?
l? l'
I'J?nlE
l?.'!
l
Dn :
Tt't nr
n!1ltt Nrtit no') ol)ll lnh nrl tllx 'l'lrP l{n
fnN
1)nn :1.'! o:?:! xrxr
1 ar
n!1 pr Nx) rr op
7
nn :11 pt n!1
D}, t
e?
?? oltltr llP o?ll
Fol, 9 verso
xzn :xn:N
r
nln) o: x:n n xrr
11
xbr nf1 xt xa 'xP xp
xo
n *rrr Nrl r'nr 'x xll rzt :xn:bx
r
ru' x xrn xnl ot:r
Iu
rr[rf'r:tr] 'iPr rzr o':r n xrrr:iy?
nD:Pl
p> o9 nn:xn:x
r
':
'pt n'r'op nn p x'xr
n n
o'r!n r;p
l
'l?P na og; o'fl
rt
fNt
:!!t
'l'n'vn rzo o
n
orp: mrrr
x
x)x o'9n 9? o'l?'l ?'!
ll:
xr nnx
r
vp NtJ ogr? np
Fr
:NpN
'
D9? n rrpr ND oP?
n
xtlr
na
1p
x: o:
r
tnr
rpp:
orr
'l 'n'N
)rx x:x ovm rf i: x'n urt
lN'? :!p nn'g:
IN)'rN ''tl
u; nn'Vt
lN)
D3?
li x:
rt'19? V?
lllJ lh )na o'p1p ui' nn'rt';
x:
Di{tlr
lD n>
rt
'Z?tt
p?n nn'ni:
'pJt! u1h nn E''g? irn o':tx
t:nr
o':rx H:ll Nt x:rl ou5
x
xx
ir)r
r"rx
t1 nn orl; x :xn:x'o vPt xnt
'pit
x'xll :orJP 't
li?')nnr:? na r
''l
oig xt
''''
Fol. l0 recto
'rz xi'n ntn o':rxl vpl :p oir
xxr
'''!pl
olp l'i?l ??
'ir
nn Nnl
"rr
'
NJ'ulrx
r J
r xn x:'rrl xl nilinr x:'ul xi'Nl !!tN tr'lPl t:> uzt
ni:1n:
r1
1r1 or.1pr rprr o'! i nn o': il
'I'
o':lx :rnx
n
l)r
o':rxr t: urN :n'illilr nJlirl nltllitl, ilJ'lt na n:'in:
:'
lzt ot1tr .:Nl NIN
1r>r
rt'n:
'1r
nnr o'rP: xn:
'
Nlll RJl 'lPl
on:
lr
Dt?
llp D
tl!
llp l
olt? vlP
l
n o'tJ)
fr:t.;111
n:N
N'rr Nl tlg NJI ogx x::rt ND ol9S
llr
!lN i'tlD Nt
lir
m:rtr :1n
Ir,
Nv o 'xp xz
xo 'n
x:rn N or:n? x:rrtr
9h
lrl Nl t'p xi'rtt
9i
Fragmenlstrom an Early Karaile Gramnatical Treatise
129
Tp
N<'rt
l'tx lli{ IrJ
l
il'N lnt
ln tJ 1U pr rgJB Nr ul
ntPt
'nx
'r'lll
1'ilt
'Tl
tlxl ;r xr'
'try
lNlrilt]p
xi:xr
! hl
lri2o
ti
i1 tr:l$
'rDn
l$
lt,
il'n)ltrr N]rN 'u
lt
'9lp l'9lp'lp
x
Irxx l
n)x l ur' 51
Fol. l0 verso
nn r'ry
ll)r
t
l
nrx )rr NN Nlilt rtN
'rlxo 'i'r
'llry
'It
'!f
n
rn
norh ll$
l';1l';!S
rtll't'Jtft
o'pJp
lD'il'qJp
l'glp
t'gl?
tx
.'tplp
fr
on nxr
Jll'r?Jp
x':T{
f
n)on n:x N) rpJp
ltt
rt'nN
o'Tf
:l'l!T
l'r
on n:x
lr: 1''y
rx
17't t
noax nlt Nt'Ju
rn rr1
n':
lx
r
':vrN
Nln nn p1 nuN Nln rty nonx nnn tp l'gJp
rN
r
lD
rnn
tyt
Nne irun DilrNon
lN
itril ilnty n1t t rBlr
Nrit Nit't)x
" 'nN
n: n't
1bn
Nit
lN
ouxr
tu: n
xnu: oyn' $r'o
',1
il)N
.'
'r$ N ly
fJS
nn nn: iroxt
ll)r
i'nn
a ln
tnx Niloln
r
nm yit
Fi
nly,? }: nfn_: nl ilpU }U nnar nno: ltitr itil Ny
n?lp 'lp n?_J t_r na uzn rrnn
n
'u
R'tot
'o
xl xnoxl l pl x'nx
ilnn nn ;?'1i? rli7 nfu o)ut
Translalion
Fol. I verso
(The treatment) of imperatives and conjugations ends here. What follows concen$
nouns, giving an explanation of their differences and identi$ing problematic
issues
relating to them and offering solutions for them.
fJl,
?l,og?,ogl, !l,T,l9'1, ?,1,''lg,l, nll, ?1,:J[t,tg[t,1Jg, Nlg,
yU],
rT:,trJp,g?,ol?,p?,
l??,
nl?, r, rJ,
y!?,
nlg, lp, y, x??, 1J?,pJT,
r?T,Dgp, :lp,
yu"l, y1,
ur?9,9, y-t,
l?r,{.
Take note thar there are many
nouns with this pattern in Scripture, some with five points (i.e. containing
ere
and
seghoQ and some with six points (i.e. containing two segols), and everything that we
say relating to the nouns that we mention here applies also to nouns that are not
9
130
GE0FFREY KNN
mentioned in this chapter of the book or other chapters when they have a similar
pattern. Now, from If; is derived
'Jl,
l, ?l?,
lJl,
tll, Jl,
TU?,
l?Jl,lll.
The plural is D'llP,
"!l?,
l]ll?, 0?'-l.l,
!'Jl?,
'Jl?, 0,1'll, :ll?,
l?'11',1'J?' l,l'-1..
(Fol. 3 recto) All words with five points do nol have qame when they become
disjoined and if anybody disputes this claim, the response to him would be that this
applies also to words with six points, for, indeed, we fnd hnd words with six points
that have lhe same form when conjoined and when disjoined,t as in, for exampte, ll)
hni
1t
Sam. 27:10), which is conjoined, and lg '3'$p (Psa. 126:4), which is
disjoined.e Nouns with six points, however, are also inflected in another wa as is the
case with
ftN,
which also has the form
l?S.
If somebody wee to say 'We find all
instances of {tp with five points, but we find l{tp
{t'Py
nl'r$'P (2 Sam. 12:12)
with qame and this contradicts your claim,' the response to him would be: Perhaps
this is like l'll, which is attested in a form with frve points and also in a form with six
points, and r{tp? is derived from't[B with six points and not from the form with five
points.
Take note that all words of this tlpe that have in medial position the letters il, Il or I)
have forms containing pa ta, for e xample r?, D, ]?, f ll,
fD,
tn:-1, lrll, D,
lF9, [t, nt], ly, o!t,1!t, y, Dltlr, p, !:, p), and others. You will find that
all such words have this form with the exception of [, from which is derived Dll,
as in o[rf

oix u'lPpl (Szet. 4:9).
re
do not find in addition to this the form
F, in the way that we hnd in addition to DFJ also ['!, as in biti441 J (Jud.
5:30). Now, if you find nf,, as in -'!tJ D'Irq (lsa. 46:3), this is derived from
DDl,
just
as p,f is derived from p), and not from ['!. If it were derived from ['!, it
would be D[tf,, like Jf;
-
]?P, on?
-
on]? and like tt?;
-
P.
Any word that has the letters h or It in final position is formed with pata, there being
8. The terms 'conjoined' (mudfl and 'disjoined' (mut) are used in the early
Karaite texts to refer to 'context' and 'pausal' forms respectively as well as to
'construct' and 'absolute' forms; see G. Khan, The Ear Karaite Tradition of
Hebrew Grammalical Thought, p. l 1 2ff.
9, The syntax of the beginning of this sentence is rather diffrcult, It is not made any
easier by assuming that a preceding leaf is missing from the manuscript, so I am
assuming that it is continuous with the text in fol. l. Fol. 2 of the manuscript is
from a different chapter of the text.
f'ragnenlsfrom an Early Karaite Grammatical Treatise
l3l
no exception to this, for example: nJl, nlp, n,n,
y?1, y?9, y:T,
and many simila
forms. The same applies to forms that have
ere,
like htlt ilpf
ryl
(Jer. 3:3), 1
i4
np-t
@xod.
28:38) and the like. Any form that contains the letters ir, n or n
in medial position, when it has inflection, i,e. a pronominal suffix, there is pta at
beginning of the word, for examplc:
'F,
tn?,
?[?.
In some cass the para is
followed by a shewa that is vocalized with pata. Examples of forms that are not
vocalized with pata are:
'[,
r"tTr-, ll (Job 39:20), hlPDA (l Sam. 17:51),
F[t (Ezek. l9:4),'ltf?
fu,
s:4), iT]l (Deut. 24:4), oJf (Esther 9:2), and
many similar cases. Examples of forms (that have a shewa) that is vocalized with
patatt: *-\t!f
@ev.
16;12),
{4J
'Jil)
(Job 20:17),
rrfqJ gou
16:8),
!p 1esa.
77:19), ulf!! (Deut. 26:7), 1t'lt{
'l}r}
(Job l8:7), and many similar cases.
you
find
among these nouns some lhat end in bgdf letters with dagesh, such as l-?f
(Isa. 5:10), t{
'q
(Cant. 6:8),
lBl' ?l
(Cen. 50:23), hTnf
'.elp
(Eznk. t7:9),
and some that have ropheh, such as n9?,-.tp91 (Psa. 76:4), n}l
t?p
@sa.
l:3),
rp!
trt (Job l8:7), n^l,r?
?lt!
(2 Sam. 2:22), :
'll
(oU 2l:33) and many simitar cases
of words endinbgdt withrapheh,
(Fol. 7. recto) The problematic issue in this chapter is the following.
rhen
a noun
such as':p is given a pronominal sufx it has dagesh, i.e. it has forms such as
t?,
l*. It does not have forms such as
'?!,
t>n
@xh
rapheh) as are the b7dt
letters at the end (of its base form), but rather has dagesh, for example
r*, gi?,
t:190,
tP,
tl?1,
l!?1,
t-11,
tTfP, All nouns of this category are inflecred in this
way. If you find
'1ry
:tFllt (2 Sam. 22:37) with rapheh, the rcason for this is that
the
'ayin
has the vowel pata, If it did not have pata, lhe form would be
rT!9,
like
tlPn, t?l,
for a noun ofthis type thathas patah (on the second letter) does not have
dagesh. The problematic issue in this chapter is the form
tJl,
which has one of the
bgdt letters at the end of the word and does not have pata,but does not have a
dagesh in contrast to other nouns of the same category. A possible explanation is that
it has rapheh since it contains three of the bgdt letters. The other nouns do not
contain tfuee bgdt letters and it is possible that it is due to the presence in it of three
bgdt letters that it has rapheh in contrast to the other nouns. (This form of the word
is anested) in the verse
'^?'yr
rJ?
(Eaa9:3) and also in'ti
l!)f
(Haggai 2:12),
which have rapheh, in contrasr to t]p}] pN (Jer. 5l:23), which has dagesh, and
likewise ll, 'inp'1, ll!,
'intp,1!lF,
lrgE, 1l':1, all with dagesh in contrast to
t32
GEAFFREY KI,IN
ttl. So (as remarked) it is possible that it has a form that conhasts with the others on
ccount of the fact that it contains tfuee bgdt letters, one after the other.
Take note that we find that (fol. 7 verso) when these wods have pronominal suffixes,
they have different forms. These fall into three calegories. One category is where a
single point (bireq) occurs under the frrst letter of the word, for example l?{,1n[l,
'lll,iy}, l:l,1n,'lg, t and many similar cases' The second category has
forrns such as l}'tg, 1l, l!l':!,
'P?8,'l'!-t,
rlf[
and many similar cases with patafr.
The third category has forms with ttuee points (segol), for example pr
1o6
38:28), nX (cen. 39:10),
?,1
(s".. 7:15),
rfi
(Psa. 16:15), 11?t
'l?n
(Hosea
13:13),11
t:F? osa.
40:17), ?[,'][, tD]P,
rl'ly,
\'
l
'HY
0sa.
?:20), and
many similar cases. This requires an explanation. If somebody were to say 'What is
the advantage of (identi$ing) problems without their solution?', the response would
be: Becoming awar of the reason why there is a problem is half the labour. If
somebody is not aware of the reason why there is a problern, he should not ask any
questions until he leams why. When somebody becomes aware of the reason why
there is a problem, his eyes are opened when he knows that it is a problem, If he
discovers the solution, his eyes are opened to something else and he learns something
in addition, for knowledge of the problem is not knowledge of its solution. Somebody
who does not know the problem, however, is doubly ignorant. Now, if somebody
were 1o say that the forms'$'1,'lYB have pata on account of the res that occurs in
it, rhe response would be: we find 9 l
\p?X
(Jud. 6:15), \X[F
l;lX
(Exod.
38:3) with pata. T-hey do not contain res but nevertheless they have pata, lf
somebody were 10 say that it is on account of the
'aleph
at the beginning of the word
(fol. 8 recto) that it is formed with pata, the response would be: tf and 1n9 have
'aleph
but do not contain pata. We find, moreover, words without irulial
'aleph
ttll
contain pat, such as l!fl, 'il':!, and words with initial
'aleph
that contain three
points (segol), such as p-try (Job 38:28), ory
1Jer.
3l:13) and ;ty$ (aen.
39:10). Take note that you do not find at all words in this category that have medial
'aleph.
All words that have medial il, n,
y
you find have short pata in initial
position, for example'ia, 1r, ity, or have pata that is not short, for example
tL,J),'fnl,
'rqj
(oU l6:s),'lyp (Ezek. l7:20).'o There are no exseptions to this'
10. In the medieval Tiberian reading tradition a pata in an unstressed closed
syllable was pronounced short whereas a pata in an uns[essed open syllable,
such as the pata in the lirst syllable of forms such as
'lil, 'Rl,
etc., was
pronounced long; see G. Khan, 'Vowel length and syllable structure in the
Tiberian tradition of Biblical Hebrew' Journal of semitic studes )ooflvl
Fragmenls
from
an Eotly Karctite Grammatical Trecttise
133
Words that have initial il, n or
y
we find to be of vaious forms, for example
r:n
(Ecc. ?:15),'B?;t (pr". 16:5),
rJtD
(Psa. 70:6),'r and in contrast to these r
\ll
(Isa. l:23), tp?: t?l (Psa. 7:13), oep,i (tsa. 29:16),
'yr,
\.19 (Deut. 3:t l), and
different srill
'plF
(Psa. l8:2),r'ppn (r. 5:15),
r1:rl
(Jud. 5:16),
lp?y
(Jer.49:4),
ofBlt (Jer. 47:5). Words that do not have il, rl or tr as initial letters are also attested in
different forms, some having three points (segol, like
l[lpl
(Micah 6:14),lffpl
(Psa. 38:2), some having patalike
'lf, T?
(Hosea Z:7),yJ4 (Hosea l0:4),
rfip
(Psa. l:3) and some others differing from these, for example 1ntl, tt?,
t91
Gsa.
76:4).
Another problematic issue in this category (of noun) is as follows. we find the plural
of such words with forms such as
ItJ?I,
:l'Jl?,
ltJgq,T?$,
t?y, r'!p[,'.lp{
and
in contrast to these we find
'?'!,
wtrictr is a plural, and likewise
tfl,
which do not
have the rorms
ltll,
'll
ttce
I'J?
and
l.lJ{.
Take note that
'l?,;l'JT, IrJgl,
''!91!,
tS
and the like are derived from forms such as D'J?!l, Etll?, Orlgq whereas
tl,
wtrct is plural, (fol. 8 verso) is nor derived from r!f" but rather from o1!1.tr
lf it were derived from ot|"1, it would b
l
like
:r'!, ;l'Jgt!,
and it would be
admissible to translate it 'your legs
1.)'.
The translation of
h
withyod, on the
other hand, is 'your two legs' and not 'your legs (pl,)'. This is because
ltll
r"y
have the sense of'your legs (pl.)' and also 'your two legs', but
1'!'1
does not denote
more than 'your two legs'. l,ikewis. l'?t does not denote more than 'his two legs'.
Now, if a number is menrioned wirh it, as in rrll DplN? .9$ (Exod. 25:26), it
would also be admissible to say '!l a or r'!l nirr!, or more, since if a number
be mentioned with it, it is diflerent from a form that does not have a number
mentioned with it. This is because when it says rr'1
ylN?
.9\, it does not mean
'four pairs' but rather 'four units'. A similar case is
y.!'f
1?-9S
ll,1 fU
)tl
l^?l (Lv, I l:23), which denotes four units. Another similar case is tt?t1
'Itg
t:tpp
(2 Sam. 9: l3), which denotes two units. Now, we have said that
T?1,
t'!1,
'1
can
have the sense of 'two' and can also have the sense of 'legs
1.)',
whereas or'1,
lt?1,
?ll,
when a number is not mentioned with them, cannot denote more than a
(1987), pp.23-82.
I l. These words are vocalized with patah in the MS.
12. The scribe has vocalized the word with sere in the MS, which would be a plural
construct form of
tpln.
ft appears that the author intended the noun form
pIF
'my strenglh' (Psa. l8:2).
13. Vocalized Dlr in the MS.
t34 GE2FFREY KIIN
pair, since they are derived from o1r1, without a number mentioned with it, which
denotes no more than two. If you look carefirlly, you will not find o:?l ltl
?lt
ever being used to denote (the legs of) more than one penon, unless a number is
mentioned with them rhar is more than two. A similar sas is ttlP bfl
ltlp.!
(Deut.
33:17), which denotes 'two (homs)'. If they were more (fol. 9 recto) than two, the
word would have the form t!l'!?, though ')lif may have the sense both of
,his
homs
1.)'
and 'his two homs'. Not every instance of rJ'!J has the sense of
,his
two
homs" This is because'Dt'!p, which has the base
f'p
and (is derived from) the plural
of this, which is tl'!!, the form with a referentiat pronoun being.Pl.!?, has the sense
of two and above. The form Jljf, which is derived from t!'!J is no more than two.
This is because Er)]? denotes 'two' and from it is derived l'lp, which denotes 'a
single pair'. The proof that ?)? denotes a pair is the passage in Scripture Olt"lp 1?l
t\RS? n)'y n,i5$l n'iu,lT q$ hnt,:,1 nln5 oltyl (Dan. 8:3). rhe form
OllrJ is derived from
f'l!,
just
as E?l is derived from )1. If you see the form
o?91P, which is derived from tr'!p, as 0ll? is derived from
lfp,
you should know
that rl'!? has thee possible derivations. This is because the plural of
f'l!
is r)'!f and
from 'J.J is derived rl'!p, just
as from i'!ff is derived t'!f!. The form rtl? may
also be derived from
f'1p
with the pattern of JJ
-
rl?J
-'t''lJ':!,
and so likewise
l?,
r)?, ltl'!?. We find Oll'lJ, which denotes 'two (horns)' and from this is derived
ll,?, which denotes 'his two horns'. lf you see :Dh, O1!1:r.' (Prov. 28:6), indeed
these are not derived from n!1 and from';l']J but rather from n!1 and from 1T, like
1l?
-
Drl!. lf they were derived from n!1 and from
IJ-1,
rhey would have the form
lD?:! and D:!'1': like l
-
o:lf,:ly
-
D:
jly,
??
-
o:??p, oy?
-
:gp.
(fol. 9 verso) lf somebody were to say 'We do not have (in Scripture) the forms nlJ
and
lJtl',
the response to him would be: We find many words in Scripture that are in
the plural but we do not find the corresponding singular form in Scripture. Likewise,
we find a singular form without finding its plural in Scripture. Take, for example,
pp as in o_t41p71
@xod.
30:23). We frnd a plural of this word, for example
l orPPl (Exod. 30:23). The form Ertjg is the plural of oipf
,
jusr
as o:"lJif is
derived from'tJp and or:! from:!B. The form o'V? may also be derived from a
form such as org?, with the patem of l?:l
-
orl?1,9
-
o'9. However, if we
posit that 'tJp is derived from ifrp, both the plural and the singular are attested,
whereas if we posit that it is derived from Op|, the form oin does not occur in
Scripturc. lf you say that o does not occur in Scipture and so its plural likewise
Fragments
trom
an Early Karaite Grammatcal Treatise 135
does not occur, note that we frnd
'9?-ol,
\11
rr
(Cant. 5:l), which is derived
from p9 without doubt, tike ll9?
-'lgt.
If it were derived from opg, it would be
'pl,
like
'r?
-'|1,
or it would be'g, like ;fp
-'?,
or't?, like p
-
'i??F.
If it were derived from opll, it would be'9?, like tD1
-
'JP and like
ttlh
-
lt-ll, except that we frnd olpi bul we do not find its plural. The plural
would be orrpig like lh
-
'PJR and uiJ
-
'qJP. We find
9
(cant. 5:l),
but there is no occurrencc of the form Ep, like JJ, from which is derived'J:!, nor
the form Ei?, like :{t} and P?, (fol. l0 recto) from which are derived , ??
and lpl, lPl,'lpl. Now, opf occurs and its plural also occurs, for example o'9?
tx\ (Exod. 30:23). We also find the form nlfh! (Jud. l:6) but we do not find its
singular, The plural of the form
f,lb,
which we find in Scripture, would be Drfrlg, like
,lh
-
o'h$, y
-
o'l?f . The singular of n1!'l would be nltil?, like nJ'ln9
-
nlt1,T9, nlfb?
-nllltt!.
We find lb, the plural of which would b orlE'?, but we
do not find in Scripture r''!pJ. There are many cases simila to these. Likewise, when
we s O'?J"!, we say that it is derived from';l'!'!, and similarly o:p'!) is derived from
tJ?, D:? from
f"!f,
rpl from
1l?,
l' from':lJ?, o:l from 1. we frnd
flit
and from this is derived Dtlll, with the pattem of lpp$, but we do not ftnd pi\. We
frnd'tfpg but we do not find
fpx.
We find also l'fP[ but we do not frnd ph. We find
lt| but we do not find
f[tb.
If somebody were to say'Why is it that from t1 is derived l'pJR and from 1U is
derivedl'9'1,l,and ui') has the pattern of
tt,
but from
'X
is derived rPll$,
so we
hnd:tf1p in contrast to
I'll
and also'tlp in contast to'l!{?', the response to him
would be: The forms
':Tl,
l';l$, l')t$, which occur in contrast to
'JP, tgJP,
lo9JP, come about on account of the fact that they are derived from Oll!$. The
translation of O?l$ is only 'two ears'. The form is not used to denote more than two,
(fol. l0 verso) The forms'l!$,
't!S,1'l$
are nouns denoting a pair. If the intention
was to refer to more than a pair, the lorm would be r)lN, like TpJP, for tp1p,
;ltfJp, ttlp
are derived from tl'rt, whereas TlS,
"lS,
lt;!l
are derived from
ol)!l,t and not from o'!$ with the pattem of 'UJP. Just as there is a problematic
issue regarding the opposition between'll$ and
t'p,
there is likewise a problematic
issue regarding the opposition between'll and
r'!f!|,
Just as there is a problematic
issue regarding the opposition between
tht
*
l'J?lt,
there is likewise a
problematic issue regarding the opposition between
Itll$
and
ltplp.
I have solved
136 GFoFFfTKIUII
the poblem in thc foregoiqg cxplnation. I have, in fact, pcscnted
thc sone
explanation n tbc alphabet of l4tr anl h, but I dcsidd to rcpeat it he sime the
problanatic iggues of thc two soction art similer, so tht a readcr ca lara abort
both in caph sction.
Tak notc that every word in this alphabet (ofthe cunent rection) th has as addcd
firsl fo and hs an initial ll, it, n or ! has s
Ntab
at its beginnng. For onplc,
\4'hen
nf
fu lh addd to its n4 it bcoomes nnf \dth lrltfr itr bcginnl,,g
Similrr crges orc tp
-il'lyp,
hrt
-n,.rn -n1l:,
fil:
-filtr:.A
wod rht
docs not have any of the leers t, il, ll or
g
in initial or mdiat poriton
has s fcn
likc r
-
n#, E
IB
-
ip?, E?t
-
n?p (rros. 6:9), ?
-
rqlp (Jrd. 4:9),
rul-il?{rP.
1
i
I
(
I
.
I
i
{
I
I
:
{
,
t
I
{
)
(
)
I
(
I
I
I
)
i
I

\
I
I
\
\
\
I
I