You are on page 1of 26

[JAB 3 (2001) 75-101]

ISSN 1462-3153

The Translations of Proto-Onqelos and the


Palestinian Targums∗

Paul V.M. Flesher

University of Wyoming

According to Eduard Yechezkel Kutscher’s linguistic analysis, Targum


Onqelos achieved its final form in two stages. Its initial translation
occurred in Palestine prior to the Bar Kokhba Rebellion in 135, perhaps
in the first century. This Proto-Onqelos translation was then taken to
Babylonia where it was later revised into what we now know as Targum
Onqelos.1 This portrayal has become the dominant scholarly view over
recent decades, and many new pieces of evidence, both literary and
linguistic, have strengthened it.
As this picture of Targum Onqelos has gained greater certainty, it
has given rise to a further question. If Proto-Onqelos was an important
translation, did it remain in Palestine and retain its influence, rather
than just disappear into Babylonia? In a recent article, I argued that it

I wish to dedicate this paper to Michael L. Klein, who befriended me many years
ago when I was a young graduate student at Oxford. His work on the Palestinian
Targums has been a constant source of learning and inspiration to me since then.
Our frequent conversations in recent years often left me with a respect for his
courage and his dedication to the study of the targums. His responses to my own
work always helped improve them. As a scholar and a friend, he is missed.
I want to thank Stephen A. Kaufman of the Comprehensive Aramaic Lexicon
project (CAL) for offering his advice, support and the resources of CAL. I also wish
to thank Jerome Lund, also of CAL, with whom I have had many conversations
about the analysis of translations. The advice of both helped improve the quality
of this paper. Any errors in it, however, are my own.
1
See E.Y. Kutscher, ‘The Language of the “Genesis Aprocryphon”: A Prelim-
inary Study’, Scripta Hierosolymitana 4 (1957), pp. 1-35, as well as the oft-cited
essay by M.H. Goshen-Gottstein, ‘The Language of Targum Onqelos and the Model
of Literary Diglossia in Aramaic’, JNES 37 (1978), pp. 169-79 (esp. n. 22). The
general discussion by B. Grossfeld is also helpful. See his The Targum Onqelos to
Genesis (AB, 6; Wilmington, DE: Michael Glazier, 1988), pp. 10-11.
76 Journal for the Aramaic Bible 3 (2001)

Figure 1: Expansions

did.2 Examining Genesis 28–50, I discovered that 35 per cent of the ex-
pansions of the Proto-PT Source, the expansion source shared by all the
Palestinian Targums, were based upon Proto-Onqelos. That is, those
Proto-PT expansions used added words and other non-translational
material drawn from Proto-Onqelos.3 This could only have happened
if Proto-Onqelos remained in Palestine during the rabbinic period. Fig-
ure 1 depicts the results of this analysis.
While I left the door open in the above literary analysis of these tar-
gums and sources to permit the influence to go either way—from Proto-
Onqelos towards the Palestinian Targums or vice versa—advances in
our understanding of the Aramaic dialects effectively close one direc-

See ‘Is Targum Onkelos a Palestinian Targum? The Evidence of Genesis 28–50’,
2

JSP 19 (1999), pp. 35-79 (44-45).


3
In the article, I referred to the initial stage as the Old Palestinian Targum,
which I saw as equivalent to Proto-Onqelos. I still think this identification is correct,
although in this article, I will use the term Proto-Onqelos since it emphasises the
link to Targum Onqelos.
Flesher Proto-Onqelos and Palestinian Targums 77

tion.4 The dialect of Proto-Onqelos was that of Jewish Literary Ara-


maic, a Middle Aramaic dialect which apparently went out of produc-
tive use in Palestine by the middle of the second century ce. The Jewish
Palestinian Aramaic of the Palestinian Targums is not evidenced prior
to the middle of the second century, and the Late Jewish Literary Ara-
maic found in Targum Pseudo-Jonathan is even later than that.5 So
the linguistic evidence places Proto-Onqelos at the beginning of these
dialectal stages, rather than at the end.6
Given this situation, did Proto-Onqelos’s importance in Palestine
extend beyond its initial creation? Did it influence the Palestinian Tar-
gums not only in their expansions, but also in their direct translation
of the Hebrew text? This is a much harder case to demonstrate, for
many similarities between the two translations can be attributed to
the common base text they translated. However this question is worth
exploring, for if Proto-Onqelos remained in Palestine and influenced
the Palestinian Targums, then the targumim and their interpretation
of scripture stand prior to the midrashic interpretations found in rab-
binic literature. Rather than being derivative of the rabbinic texts, the
targumim stood in a position to influence them.
This paper will be applying three different approaches to deter-
mine whether the Palestinian Targums drew their translation from that
of Proto-Onqelos. It will explore the methodological characteristics of
these approaches and apply them to a few verses to test whether they
might be useful in more extensive explorations of this question. To fore-
shadow the results, the experimental application of these three meth-
ods indicates their viablity as approaches to addressing the question of
translation dependency. The results also suggest that the Palestinian
Targums were influenced by Proto-Onqelos’s translation.
Before we begin, two preliminary matters must be addressed. First,
as Stephen Kaufman and others have argued,Targum Pseudo-Jonathan
does not belong to the classification of Palestinian Targums.7 It has
a composite character that differentiates it from the true Palestinian
4
See Flesher, ‘Targum Onkelos’, p. 74.
5
On the Aramaic dialects of Palestine, see Kutscher, ‘The Language of the “Gen-
esis Apocryphon”’, and the works by Kaufman and Cook, referenced below (n. 7).
6
Thus Geza Vermes’ idea that Targum Onqelos comprises a version of the Pales-
tinian Targum with its expansions significantly reduced, a position that has long
appealed to me, becomes difficult to fit into the linguistic evidence. See G. Vermes,
‘The Targumic Versions of Genesis 4:2-16’, in idem, Post-Biblical Jewish Studies
(Leiden: E.J. Brill, 1975), pp. 92-126. See further discussion of his views at the end
of this essay.
7
See S.A. Kaufman, ‘Dating the Language of the Palestinian Targums and their
Use in the Study of First Century CE Texts’, in D.R.G. Beattie and M.J. McNamara
78 Journal for the Aramaic Bible 3 (2001)

Targums, a character that is revealed both by source criticism and by


linguistic analysis. If the Palestinian Targums are represented by Tar-
gum Neofiti, the Fragmentary Targums, and the Palestinian Targum
fragments from the Cairo Geniza, they share three features: they are
composed in Jewish Targumic Aramaic (a sub-dialect of Jewish Pales-
tinian Aramaic),8 they share the expansions of the Proto-PT source
where extant,9 and their translations seem to derive from a common
original.10 Pseudo-Jonathan, by contrast, brings together three diver-
gent parts: the Proto-Onqelos translation primarily composed in Jew-
ish Literary Aramaic, the Proto-PT source composed in Jewish Tar-
gumic Aramaic, and an enormous number of expansions found in no
other targum that are written in the dialect of Late Jewish Literary
Aramaic.11 Pseudo-Jonathan’s composite character indicates that it is

(eds.), The Aramaic Bible: Targums in their Historical Context (JSOTSup, 166;
Sheffield: JSOT Press, 1994), pp. 118-41. The observation of Pseudo-Jonathan’s
composite nature has been made by a number of scholars in the past. These include
Vermes, ‘The Targumic Versions of Genesis 4’, p. 110, and E.M. Cook, ‘Rewriting
the Bible: The Text and Language of the Pseudo-Jonathan Targum’ (doctoral dis-
sertation, University of California, 1986), whose discussion on pp. 2-5 gives some
history of the scholarly views on this matter.
8
See Kaufman, ‘Dating’, and idem, ‘Languages (Aramaic)’, in ABD, IV, pp.
173-78.
9
These are the results of the source-critical analysis of the Palestinian Targums
which I directed in the early 1990s with three of my then doctoral students: B.P.
Mortensen, R.M. Campbell, and L. Simon. Mortensen and I are moving our stud-
ies towards publication, but some results have already been published. See P.V.M.
Flesher, ‘Mapping the Synoptic Palestinian Targums of the Pentateuch’, in Beattie
and McNamara (eds.), The Aramaic Bible, pp. 247-53; idem, ‘Exploring the Sources
of the Synoptic Targums to the Pentateuch’, in P.V.M. Flesher (ed.), Targum Stud-
ies. I. Textual and Contextual Studies in the Pentateuchal Targums (University
of South Florida Studies in the History of Judaism, 55; Atlanta: Scholars Press,
1992), pp. 101-34; R.M. Campbell, ‘A Fragment-Targum without a Purpose? The
raison d’être of MS. Vatican Ebr. 440’ (doctoral dissertation, Northwestern Univer-
sity, 1994); B.P. Mortensen, ‘Targum Pseudo-Jonathan: A Document for Priests’
(doctoral dissertation, Northwestern University, 1994); idem, Pseudo-Jonathan and
Economics for Priests’, JSP 20 (1999), pp. 39-71; idem, ‘Pseudo-Jonathan’s Tem-
ple, Symbol of Judaism’, forthcoming.
10
This general observation is apparent to any scholar who has compared the
translations of the Palestinian Targums. See also the important study by S.A. Kauf-
man and Y. Maori, ‘The Targumim to Exodus 20: Reconstructing the Palestinian
Targum’, Textus 16 (1991), pp. 13-78.
11
See Kaufman, ‘Dating’. Also, in a paper delivered to the Aramaic Studies Sec-
tion of the Society of Biblical Literature in 2000, I explored the dialectal character
of the sources identified by my ongoing source-critical analysis of the Palestinian
Targums. I found that the dialectal differences in Targum Pseudo-Jonathan out-
lined by Kaufman and Cook largely coincided with the sources of this Targum.
Flesher Proto-Onqelos and Palestinian Targums 79

not a Palestinian Targum like the other targums, which makes it an


unsuitable test case for whether Proto-Onqelos remained in Palestine.
Second, in this light, Targum Neofiti constitutes the most suitable
representative of the Palestinian Targums for our purposes. It contains
the shared Palestinian Targum translation and is the only complete
targum. For individual passages, it is sometimes not the ‘best’ ver-
sion, but the targum permits sample analysis from across the Penta-
teuch, rather than limiting it to passages extant through selection or
accident—many of which are not direct translation of the source text.
More in-depth investigation of the Palestinian Targums’ translations,
however, will need to look at all Palestinian Targums.

Approach 1: Non-Exact Translations


The key question for comparing translations is this: when examining
two translations of the same base text, how do you show whether one is
literarily dependent on the other or whether their similarities stem from
rendering the same base text? P.E. Dirksen answers this succinctly. In
studying whether the Peshitta depends upon the targumim, Dirksen
identifies three different types of connection which can be used to test
for possible literary dependency between two translations. He says:
In the main we can distinguish among three levels of relationship: (1)
that of language, (2) that of exegetical tradition, and (3) that of literary
dependency.12
These three levels move from the least significant to the most, and the
scale is designed to be used step-by-step. In other words, if two transla-
tions seem to be similar, one should first determine whether language
usage could account for it, and if not, then one should ascertain if
the two texts share an exegetical tradition. If either level can explain
their similarities, there is no case for literary dependence. Only if these
first two levels fail, can literary dependency then be examined and, as-
suming no other impediment appears, can be argued. Let me explain
further.
The first level of relationship asks what similarities between trans-
lations result from their composition in the same language, whether in
the same or similar dialects. Translations into English, for example, use
many common words. These range from ‘the’ and ‘a’, the definite and
indefinite markers, to pronouns, and even to common verbs and nouns.
12
See P.E. Dirksen, ‘Targum and Peshitta: Some Basic Questions’, in Paul V.M.
Flesher (ed.), Targum Studies. II. Targum and Peshitta (University of South Florida
Studies in the History of Judaism, 165; Atlanta: Scholars Press, 1998), pp. 3-14 (4).
80 Journal for the Aramaic Bible 3 (2001)

To what extent do two translations simply use such words because the
language gives few other choices? The extent to which this issue bears
on this study becomes clear in a comparison of Aramaic translations
to Numbers 22. If we compare only the word-roots employed and ig-
nore matters of orthography and morphology, we find that Targum
Pseudo-Jonathan and Targum Onqelos share 84 per cent of the same
roots, while Targum Neofiti and Targum Onqelos share 83 per cent.
Since Targum Pseudo-Jonathan frequently copied Proto-Onqelos, the
closeness of Neofiti and Onqelos appears significant—just one percent-
age point less than Targum Pseudo-Jonathan and Targum Onqelos.
However, a comparison of Onqelos to Numbers 22 with the Samaritan
Targum and the Peshitta reveals results not much lower; the Samari-
tan Targum has 76 per cent of Onqelos’s roots, while the Peshitta has
68 per cent. These two translations with no dependency on Onqelos
are thus only slightly less parallel to Onqelos than Pseudo-Jonathan,
the targum which frequently copied Onqelos, when measured as a per-
centage of parallel word roots. This establishes a fairly high standard
to show one translation’s dependence upon another, and the difficulty
of getting above Dirksen’s first level of language.
Dirksen’s second level is that of exegetical tradition. This includes
matters that often fall under the term ‘translation technique’, as well
as different ways of rendering the divine name, matters of the an-
thropomorphic presentation of God, and parallels in meaning but not
wording, especially when those parallels can be found in other texts,
such as those of rabbinic literature.
It is only at the third level where the question of literary dependence
comes into play. Even here, however, the question must be approached
with caution. As Dirksen observes:
Only after these two levels of relationship have been taken into account
may there remain agreements which suggest the third level, that of liter-
ary dependence. Agreements of this type must not only deviate clearly
from the Masoretic Text but also be of a literary character. . . . There
can, however, be no automatism here. In every case, we must ask whether
a literary relationship is the most obvious explanation.13

The application of Dirksen’s three principles means there is really


only one way to identify literary dependence between two translations.
That is by identifying passages where the earlier translation contains a
phrase that deviates from a direct translation of the Hebrew text and
the later translation follows it. This could happen if the later transla-
13
Dirksen, ‘Targum and Peshitta’, p. 5.
Flesher Proto-Onqelos and Palestinian Targums 81

tion essentially copies the earlier one, or if the later translation exegetes
the earlier translation rather than the Hebrew text.
Using this approach, we find evidence of the Pentateuchal Targums’
dependence upon Proto-Onqelos’s translation. For Targum Pseudo-
Jonathan, this evidence is extensive, while for Neofiti and the other
Palestinian Targums there is less, but still significant, evidence. Even
excluding the expansions, Pseudo-Jonathan reproduces between a third
and a half of all Proto-Onqelos’s non-literal renderings. By contrast,
Targum Neofiti follows the translation of Proto-Onqelos much less fre-
quently, perhaps only one of every four or five instances. But even
these are sufficient to indicate Neofiti’s literary dependence on Proto-
Onqelos. To get a flavor of the parallels between the translations of
Targum Neofiti and Proto-Onqelos, let us examine three examples.
At the beginning of Gen. 44.1, the Hebrew text reads, ‘And he
[Joseph] commanded he who was over his house (wtyb l[ r!a ta wxyw)’.
Targum Onqelos seems to have sensed a lack of specificity in the term
‘he who was (r!a)’, which it addresses by adding a verb, ‘he who was
appointed (anmmd) [over his house]’. Neofiti picks up on Targum Onqe-
los’s rendering and expands it by adding three words, ‘And he com-
manded the one (÷ml) who had been appointed (anmm hwhd) epitropos
(swpwrfypa)14 over his house’. The first and second added words build
directly on Onqelos’s addition. Thus it is clear that Targum Neofiti
exegetes Onqelos’s rendering in this verse.
Similarly, in Exod. 19.6, the Hebrew text reads, ‘you shall be for me
a priestly kingdom and a holy nation’, with the two words for ‘priestly
kingdom’ standing in a construct formation, µynhwk tklmm. Targum On-
qelos changes the meaning to ‘kings, priests [and a holy nation]’, by
dropping the construct state and changing ‘kingdom’ to ‘kings’, ÷yklm.
Neofiti follows Proto-Onqelos’s rendering with just one alteration; it
adds a waw to make it ‘kings and priests’.
Finally, in Deut. 3.5, the Hebrew text has twrxb, which means ‘for-
tified’, as an adjective describing ‘cities’. Targum Onqelos renders this
as ÷kyrk, a passive, plural participle meaning ‘fortified’, but then adds
another word ÷pqm, meaning ‘surrounded’. Targum Neofiti has Targum
Onqelos’s added word, as ÷pyqm.
These three verses illustrate that Neofiti contains passages in which
it draws from the translation found in Proto-Onqelos. This method thus
reveals links between the translations of Neofiti and Onqelos. It might
be objected that the passages examined consist of only a word or two,
rather than extended phrases which would provide more solid evidence.
14
This is an Aramaic rendering of the Greek word ejpivtropo" meaning ‘manager’.
82 Journal for the Aramaic Bible 3 (2001)

This is a valid point. However, the analysis focuses on short phrases or


just a couple of words because the Palestinian Targums usually make
larger phrases into expansions. In my work, this means that they have
already been analyzed as expansions linked to Targum Onqelos.15 To
study them again as translations would provide no new information.16

Approach 2: Conflations

While Neofiti sometimes follows Proto-Onqelos’s deviations from the


Hebrew text, it more often does not. Frequently Neofiti translates the
Hebrew text directly when Proto-Onqelos deviates. The question then
arises whether Neofiti knew only the Hebrew text or whether it con-
sciously made a choice between Onqelos and the Hebrew text. While we
cannot answer that question in most cases, one translation technique
provides a glimpse into what Neofiti’s translator did, namely, the phe-
nomenon of conflation. This phenomenon has been well documented for
Targum Pseudo-Jonathan, but is less well-known for Targum Neofiti.17
However, it does exist in Neofiti, where Neofiti provides both Onqelos’s
rendition and a more literal translation of a Hebrew word, often in the
form of a Hebraism. Here are three examples of that phenomenon.
In Num. 22.18 the rsv translates part of the verse as ‘I could not
go beyond the command of the Lord my God.’ The Hebrew word which
the rsv renders as ‘command’ is actually ‘mouth’, yp. Onqelos trans-
lates the word like the English does, rendering it as ‘decree’, tryzg.
Targum Neofiti renders the Hebrew word directly as µwp and then adds
Onqelos’s tryzg. Thus, Neofiti produces a conflate of the Hebrew and
Onqelos.
Similarly in Gen. 28.11, Targum Neofiti renders both the Hebrew
text and Targum Onqelos’s translation of it. The Hebrew text has ‘and
he set it [= a stone] under his head’. Targum Onqelos translates the
Hebrew ‘head’ (!ar) as adsya which means ‘pillow’ or ‘side (of the
15
See, for example, Flesher, ‘Targum Onkelos’.
16
The twelve examples of additional material in Targum Onqelos adduced by
Geza Vermes in his well-known article ‘Haggadah in the Onkelos Targum’, for in-
stance, become the basis for expansions in the Palestinian Targums, not of transla-
tion. See Vermes, ‘Haggadah in the Onkelos Targum’, in idem, Post-Biblical Jewish
Studies, pp. 127-38. Examples where this happens in Gen. 28–50 can be found in
Flesher, ‘Targum Onkelos’.
17
Both Donald Splansky and Edward Cook have analyzed the conflates in Targ.
Ps.-J. See D.M. Splansky, ‘Targum Pseudo-Jonathan: Its Relationship to Other
Targumim, Uses of Midrashim and Date’ (doctoral dissertation, Hebrew Union
College–Jewish Institute of Religion, 1981), esp. pp. 22-40; and Cook, ‘Rewriting’,
pp. 46-48.
Flesher Proto-Onqelos and Palestinian Targums 83

head)’. Targum Neofiti brings the two together with h!ar ydsa, which
should be understood as ‘side of his head’ or ‘pillow for his head’.
Finally, in Exod. 19.5, Targum Neofiti follows Targum Onqelos’s
one-word change of the Hebrew text by creating a conflate with the
Hebrew word and Onqelos’s Aramaic one. The Hebrew text reads ‘you
will be to me a special possession (hlgs) from all the nations’, which
Onqelos renders as ‘you will be beloved (÷ybybj) before me from all the
nations’. Neofiti brings these two versions together as ‘you will be to my
name a beloved (÷ybybj) people, as a special possession (hlgs) from all
the nations’. In all three of these examples, Neofiti provides a rendering
that demonstrates it knows and draws from both the Hebrew text and
Proto-Onqelos’s translation of it.18
What we see in these examples is that Targum Neofiti occasionally
indicates that it knew both the Hebrew text and Proto-Onqelos’s ren-
dering of it. The analysis of conflates shows Neofiti’s composer strug-
gling with the choice between the Hebrew text and Proto-Onqelos,
and rather than choosing between them, deciding to use both—thus
indicating the role which Proto-Onqelos’s translation played in the
construction of Targum Neofiti’s translation.

Approach 3: Clausal Parallels

So far the two approaches that have been applied to the analysis of
possibly dependent targum translations suggests that Neofiti used On-
qelos’s translation in creating its own. But this is not the end of the
story. Despite my agreement with Dirksen’s scheme and its inherent
common sense and usefulness, there is one goal it does not accom-
plish in this new application, namely, it does not actually evaluate the
literary dependency of translations but rather the dependency of non-
translations. That is, it is only when one translation deviates from a
literal rendering of the base text that we can evaluate a second trans-
lation’s dependency on it. If the earlier translation translates exactly,
then the later translation’s possible dependency upon it cannot be de-
termined.
Why is this? One answer is suggested by Dirksen’s evaluation of
Wernberg-Møller’s claim that the Peshitta occasionally consulted Tar-
gum Onqelos.19 Dirksen states:
18
Of course, this essay has only provided a few sample analyses and examples,
which are suggestive but not definitive for all of Targum Neofiti. Further work needs
to be done.
19
See P. Wernberg-Møller, ‘Some Observations on the Relationship of the
84 Journal for the Aramaic Bible 3 (2001)

Perhaps the most important issue concerns the type of agreement: if the
translator of the Peshitta made use of Targum Onqelos, we would expect
him to do so in linguistically or theologically difficult passages, and not,
at least not primarily so, where such difficulties do not exist.20
This is a reasonable expectation, and it works well for analyzing the re-
lationship between Targum Onqelos and the Peshitta, which are both
essentially literal translations. However, for studying the Palestinian
Targum’s use of Proto-Onqelos’s translation, two problems arise. First,
the Palestinian Targum uses expansions to address linguistic and the-
ological difficulties in the Hebrew text. Thus the places where Dirksen
expects agreement become the very locations where the Palestinian
Targums either elaborate or ignore the rendering of Proto-Onqelos.
Second, if an earlier translation became prestigious—gaining a quasi-
sacred or revered status—then a later translation which drew from it
might wish to use it in an obvious fashion. And when we turn to Tar-
gums Pseudo-Jonathan and Neofiti, that is what we find. While both
texts use expansions to solve difficulties, the places where no inter-
pretive problems exist become the locations where the most extensive
agreement with the translations of Proto-Onqelos can be found. Indeed,
with Pseudo-Jonathan it almost seems as if the targumist wanted to
concentrate on the expansions, and used Proto-Onqelos’s prestigious
translation to fill in between them.
The problem for the scholarly interpreter lies in the difficulty of de-
termining whether the Palestinian Targums followed Proto-Onqelos’s
translation or whether they inadvertantly created translations with the
same wording. Is it possible to design an approach that can evaluate
translation, and not just deviations from translation? I want to suggest
a method here.
I call this method the search for ‘clausal parallels’.21 The goal is to
compare two translations of a single base text and determine whether
one of them is literarily dependent upon the other. By ‘clause’, I mean a
Peshitta Version of the Book of Genesis to the Palestinian Targum Fragments
Published by Professor Kahle, and to Targum Onkelos’, ST 15.2 (1961), pp. 128-
80.
20
Dirksen, ‘Targum and Peshitta’, p. 11. Dirksen discusses two other principles
here which are also useful for studying translation, but are not directly relevent to
the question here.
21
This is similar to what I imagine P. Wernberg-Møller’s term ‘phraseological sim-
ilarity’ might mean; see p. 255 of his essay, ‘Prolegomena to a Re-Examination of the
Palestinian Targum Fragments of the Book of Genesis Published by P. Kahle, and
Their Relationship to the Peshitta’, JSS 7 (1962), pp. 253-66. However Wernberg-
Møller provided neither a definition nor examples, so it has never been clear what
he intended. Rather than impose a definition on his term, I have created my own.
Flesher Proto-Onqelos and Palestinian Targums 85

number of words in a row, the more the better, since shorter parallels
between two translations could occur as chance or just similarity of
language, as Dirksen observed.22 The clauses should not just consist of
minor words, but should contain words carrying a passage’s meaning,
such as key verbs and nouns. A solid argument in favor of literary
dependency using this method derives not from a single or occasional
parallel clause, but must be based on a number of parallel clauses
found in a limited set of verses. Indeed, the primary strength of this
method is its ability to analyze the translation of a limited number of
verses—ten, twenty, or a single story perhaps—and derive clear results.
These results will indicate whether the two translations to the passage
in question are literarily dependent; this in turn will be suggestive,
but by itself not conclusive, about the relationship between the two
targums across the entire text.
This analysis is complicated by the fact that Onqelos and the Pales-
tinian Targums were written in different dialects of Aramaic. Onqelos
was composed in Jewish Literary Aramaic while the Palestinian Tar-
gums were written in Jewish Targumic Aramaic, a sub-dialect of Jewish
Palestinian Aramaic. Given this dialectal difference, there can be no
expectation of exact identity between the targums. If there was, then
they would be in the same dialect.23
Finally, the analysis does not seek to determine whether the two
translations are the same, but to ascertain whether one translation
knew and used another. This means that words which a later transla-
tion inserts into its rendering of the earlier one do not interfere with
determining whether the later translation falls into the categories of
this method. In our targums, the translations will often be interspersed
with expansions and translations that find no parallels.
There are four categories of clausal parallelism which indicate dif-
ferent degrees of relationship, from direct literary dependency to no
links at all.
(1) Exact quotation. No changes in form or dialect, although oc-
casional minor spelling variation is acceptable. This constitutes direct
copying, bringing the earlier text’s rendering into the latter’s without
concern for consistency within the latter text.
(2) Quotation with minor changes. In this case, the variation from
an exact quote may come from the later text translating the dialect of
the earlier text into its own. Other minor or occasional changes may

22
As a rule of thumb, I used only passages of eight or more words in the analysis.
Several of the parallel clauses analyzed here contain as much as sixteen words.
23
For further discussion of dialects, see Kaufman, ‘Languages’.
86 Journal for the Aramaic Bible 3 (2001)

be due to morphological or clarification needs, such as the unbundling


of suffixes or prefixes. Occasionally, there will be slight differences in
stylistic matters. Thus, while there is a desire to cite the earlier, pres-
tigious rendering, it is adapted to fit the character of the target text.
Occasionally, a later targum provides a near citation of a verse from
an earlier targum, at either category 1 or 2, but will make a single
significant change in a word-root or a construction. Assuming a verse
of sufficient length, the verse will be classified in category 2.
(3) Citation with major changes. In this category, the link between
the two translations appears mainly in two features: use of common
word-roots, and a common approach to the words and particles link-
ing phrases and words. This category permits one or two exceptions
in an average-length verse, but more than that places the verses into
category 4. With regard to roots, as long as the two translations share
the roots, their orthography and even morphology may differ and still
be seen as belonging to this category. By ‘a common approach to the
words and particles linking phrases and words’, I intend to signal not
that the two translations use the exact same words, but that within
their dialect and style, they use the same type of word(s), prepositions,
or particles in the same manner. For example, do both texts indicate
the construct with a dalet or with nothing? Do they signal the direct
object with related particles (l[ and l), unrelated particles (ty vs. l),
or nothing? The goal is not to find identity in the terms themselves,
but to use them to ascertain whether the translated sentences have
the same structure, as indicated by words that govern the syntacti-
cal links within a sentence, such as prepositional phrases, subordinate
clauses, and other links among multiple words. It is the translation’s
attempt to replicate the structure of the source text that remains im-
portant, even when the form of the individual words themselves may
vary. Clauses in this category are suggestive rather than definitive in
identifying dependency.
(4) No demonstrable links. Verses fall into this category when they
fail to meet the requirements of category 3. Thus, regular use of differ-
ent roots, especially in key verbs or nouns, or different approaches to
linking clauses and words indicates that the two translations indepen-
dently follow the Hebrew text, rather than one drawing upon another.
If there is a significant quantity of the first category of clausal par-
allelism, then it is clear that the latter translation uses—essentially
copies—the earlier one. Category 2 is only slightly less substantial in
indicating literary dependency. If there is a significant number of this
category of parallel clause, then these also indicate literary dependence.
Flesher Proto-Onqelos and Palestinian Targums 87

Category 3 is a transitional category. It generally does not indicate


literary dependence by itself. However, if there is extensive material
in this category, and material in the surrounding passage falls into a
higher category, then category 3 is supportive. Category 4 indicates no
dependency at all.
The ten verses of Gen. 28.11-20 will serve as an initial test of the
method. The main object of the test will be to determine whether Tar-
gum Neofiti draws from Onqelos’s translation. However, since this is
also a test of the method, we will also study the well-known relation-
ship between Targum Pseudo-Jonathan and Targum Onqelos to see
whether the method confirms that knowledge. We shall also compare
the Samaritan Targum to Targum Onqelos to see how the method per-
forms when faced with two targums that are not literarily dependent.24
In the ten verses of Gen. 28.11-20, there are no verses in which
Targ. Neof. directly quotes Targ. Onq.’s translation, the first category
of the clausal parallel method. There are four verses where Targ. Neof.
renders Targ. Onq. in a category 2 classification, that of quotation with
minor changes: Gen. 28.12, 16, 18, 19. It is dialectal differences by and
large that prevent these verses from belonging to category 1.25 None of
Neofiti’s verses fall into category 3, but five of them belong to category
4, that of insufficient demonstrable links: Gen. 28.11, 13, 14, 15, and 20.
One verse cannot be examined, that of Gen. 28.17, because it contains
a Proto-PT expansion that is inseparable from the translation. This
expansion is tied to the additional material in Targ. Onq., but since
it is an expansion it falls outside our interest in translation.26 Let us
begin with the verses where Targ. Neof. belongs to category 2.

Genesis 28.12 (note: Proto-PT expansion not shown)

hmym!h [ygm w!arw hxra bxm µls hnhw µljyw Heb.


aym! tyx d[ yfm hy!yrw a[rab Åy[n amlws ahw µljw Onq.
aym! tyx d[ yfm h!arw a[rab [bq µls ahw µljw Neof.
ayym! tyx d[ yfm y!yrw a[rab [ybq amlws ahw µljw Ps.-J.
hymw!l yfm h!yrw a[ra µ[q µls haw µljw Sam.
24
I use the J ms of the Samaritan Targum, as published by A. Tal in his The
Samaritan Targum of the Pentateuch: A Critical Edition (3 vols.; Tel Aviv: Tel
Aviv University, 1980). The text reproduced below is from pp. 108-10 of the first
volume. The J ms respresents the oldest known form of the Sam. Targ. See Tal,
The Samaritan Targum, III, p. 54.
25
While it is standard to observe that Jewish Literary Aramaic final alephs are
indicated by hes in JPA, I will not draw upon that dialectal indicator. On the one
hand, it is too common, and on the other hand it is too subject to copyist alteration
and corruption.
26
See Flesher, ‘Targum Onkelos’, pp. 44-45, 75.
88 Journal for the Aramaic Bible 3 (2001)

wb µydryw µyl[ µyhla ykalm hnhw Heb.


hyb ÷ytjnw ÷yqls ywyd aykalm ahw Onq.
hb ÷ylktsmw ÷ytjnw ÷yqls yyyd hykalm ahw Neof.
hyb alktsml ÷ytjn yyyd ay!ydq aykalm ra! ÷ykb Ps.-J.
hb µyt[nw µyqls µyhla ykalm haw Sam.

Even though Gen. 28.12 contains a Proto-PT expansion, that expan-


sion separates easily from the translation, thus enabling us to analyze
the remaining translation. Targum Neofiti’s translation falls into cate-
gory 2 when compared to that of Targum Onqelos. This is largely be-
cause of a single word, Targ. Neof.’s rendering of [bq for Targ. Onq.’s
Åy[n. Otherwise this would be a category 1, for Neofiti follows the rest of
Onqelos’s entire verse—some sixteen words—with only minor adjust-
ments to spelling and word form. For example, instead of Onqelos’s
amlws, Neofiti has µls—the base word for the dialects of both targums
being µlws. Instead of spelling the masculine possessive suffix in a mele
form (hyb, hy!ar) as does Targ. Onq., Targ. Neof. spells it in a plene
form (hb, h!ar), a standard practice for Targ. Neof. Neofiti even copies
Onqelos’s two additional words, tyx d[, which render the locative suf-
fix the Hebrew text places on µym!. But although Targ. Neof. nearly
quotes Targ. Onq.’s translation of this verse, the word change requires
it to be placed in category 2.
Pseudo-Jonathan follows Targ. Neof.’s and then Targ. Onq.’s ren-
dering in the first half of the verse, with only minor alterations in
spelling and Targ. Neof.’s single different word. In the second half,
however, Targ. Ps.-J. deviates a little further from both Targ. Onq.
and Neof., following Targ. Neof.’s rendering generally, while dropping
one of Targ. Onq.’s words and changing two others and some particles.
Thus while the first part of Targ. Ps.-J.’s rendering easily falls into
category 2, the second part brings the verse into category 3.
For the Samaritan Targum, two types of differences from the Jewish
targums indicate that it is based upon the Hebrew text, rather than
the other targums. It thus should be classifed as category 4. The first
comprises the two main word differences, probably due to dialectal dif-
ferences: µ[q instead of [bq or Åy[n, and µyt[n instead of ÷ytjn. The second
comprises Sam. Targ.’s adherence to the Hebrew text rather than than
the targums in its use of small particles linking words or phrases. For
instance, the bet which the other targums insert before a[ra does not
appear in Sam. Targ.; the two words tyx d[ found in the Jewish targums
inserted before ‘the heavens’ do not appear in Sam. Targ.. Likewise,
it lacks the dalet before God’s name. In all of these, Sam. Targ. fol-
lows the Hebrew text. So here, this method differentiates between the
targums which have links of dependence and those that do not.
Flesher Proto-Onqelos and Palestinian Targums 89

Genesis 28.16
hwhy !y ÷ka rmayw wtn!m bq[y Åqyyw Heb.
yr! ywyd arqy af!wqb rmaw hytn!m bq[y r[taw Onq.
hyr!d yyyd hnyk! rqya af!wq ÷m ah rmaw hytn! ÷m bq[y r[taw Neof.
yr! yyyd htnyk! rqy tya af!wqb rmaw hykmdm bq[y r[tyaw Ps.-J.
hwhy tyy ÷rb rmaw htn!m bq[y ry[taw Sam.

yt[dy al yknaw hzh µwqmb Heb.


ady ytywh al anaw ÷ydh artab Onq.
ady hna tyl hnaw ÷ydh artab Neof.
[dy tywh al anaw ÷ydh artab Ps.-J.
tmkj al anaw ÷dh hrtab Sam.

Targ. Neof. to Gen. 28.16 comprises a near quotation, but must be


classified as category 2. Neofiti parallels Onqelos’s translation fairly
closely with only a few quite minor changes, such as separating hytn!m
into hytn! ÷m, use of min instead of a bet before af!wq, spelling Onqelos’s
arqy as rqya, and adding a he and a dalet to yr!. Another indication
that Targ. Neof. follows Targ. Onq.’s translation is that like Targ. Onq.,
Targ. Neof. fails to reproduce the Hebrew text’s !y. It is in the final
phrase where Targ. Neof. deviates noticably from Targ. Onq.’s render-
ing. Targ. Neof. does not follow Targ. Onq.’s ytywh, the second-to-last
word, instead using hna; and gives tyl instead of Targ. Onq.’s al for
the word before it. Although none of these differences carry great sig-
nificance, taken together they put Neofiti’s rendering of Onqelos into
a category 2 instead of category 1.
Pseudo-Jonathan’s treatment of Gen. 28.16 also falls into category
2, but with different set of variations from Targum Onqelos. The fur-
thest away from Targ. Onq. is Targ. Ps.-J.’s substitution of hykmd for
Targ. Onq.’s hytn!. However, for Targ. Neof.’s differences with Targ.
Onq. in the rest of the verse, Targ. Ps.-J. follows Targ. Onq. Targ.
Ps.-J. follows Targ. Neof.’s addition of htyk!/htnyk! before the divine
name, however. And it reproduces the Hebrew text’s !y with tya, which
is missing from both Targ. Onq. and Targ. Neof. So, with the exception
of the word change, Targ. Ps.-J. reproduces Targ. Onq.’s translation
with category two adherence. It is also clear, however, that Targ. Ps.-J.
is drawing from both the Palestinian Targum and the Hebrew text as
well.
The Samaritan Targum consistently shows that it provides a close
translation of the Hebrew text, and is thus not related to the Jewish
targums—earning a classification of category 4. From the use of tmkj
instead of yt[dy as the verse’s final word, to its careful echo of the He-
brew text’s hwhy !y ÷ka, Sam. Targ. clearly is based upon the Hebrew
90 Journal for the Aramaic Bible 3 (2001)

text. This last-mentioned three-word Hebrew phrase the Sam. Targ.


renders word-for-word in a parallel three-word phrase, while Targ. Onq.
sets the stage for the other targums by adding two new words and giv-
ing ÷ka as af!wqb. This clear contrast leaves no doubt that the method of
clausal parallels can distinguish between dependent and non-dependent
translations.

Genesis 28.18
wyt!arm µ! r!a ÷bah ta jqyw rqbb bq[y µk!yw Heb.
yhwdsya yw!d anba ty bysnw arpxb bq[y µydqaw Onq.
h!ar ydsa yw! yd anba ty bsnw arpxb bq[y µdqaw Neof.
ywdsya yw!d ÷m anba ty bysnw arpxb bq[y µydqaw Ps.-J.
htwhmdqm hb!d hnba ty bsnw hrpxb bq[y µdqaw Sam.

h!ar l[ ÷m! qxyw hbxm hta µ!yw Heb.


h!yr l[ aj!m qyraw amq hty yw!w Onq.
h!ar l[ j!m qyraw amyyq hty yw!w Neof.
h!yr l[ aj!ym qyraw hmq hty yw!w Ps.-J.
h!yr l[ j!m qraw hm[q hty hb!w Sam.

Targum Neofiti at Gen. 28.18 is again almost a direct quotation of


Targum Onqelos, but not quite. So it falls into category 2. There are
a few minor orthographic variations—such as µdqaw instead of µydqaw,
yw! yd instead of yw!d, and j!m instead of aj!m—but the main difference
between Targ. Neof. and Targ. Onq. is dialectal. The different words
for ‘pillar’ in the two targums—Targ. Onq. has amq, whereas Targ.
Neof. has amyyq—indicate the two targums’ different dialects, for neither
word ever appears in the other targum.27 Targ. Neof. also also adds a
conflate of a Hebrew and Onqelos rendering which we discussed earlier,
h!ar ydsa. So Targ. Neof. reproduces Onqelos’s translation but with a
dialectal alteration that places it into category 2.
Targ. Ps.-J.’s rendering clearly constitutes a direct quotation of
Targ. Onq.’s translation, at category 1. It follows Targ. Onq.’s transla-
tion closely, with the exception of an orthographical change at ywdsya,
and the added word ÷m.
The Samaritan Targum provides a close rendering of the Hebrew
text, at category 4, even though it coincidently renders the first six
words of the verse the same as Targ. Onq., with only a couple of
spelling differences. However, it becomes clear that Sam. Targ. did not
copy Targ. Onq. (or any other Jewish targum) when it understands
the Hebrew wyt!arm as meaning ‘before’ or ‘in front of’ and renders it
27
See Michael Sokoloff’s Dictionary of Jewish Palestinian Aramaic (Ramat Gan:
Bar Ilan University Press, 1990) for a listing of the verses where this word appears.
Flesher Proto-Onqelos and Palestinian Targums 91

htwhmdqm, as it did in Gen. 28.11. There are several other orthographi-


cal differences, the most obvious of which are hm[q and hb!, which may
stem from dialectal differences.
Genesis 28.19
la tyb awhh µwqmh µ! ta arqyw Heb.
la tyb awhh artad hym! ty arqw Onq.
latyb awhh artad hm! ty arqw Neof.
latyb awhh artad am! arqw Ps.-J.
la tyb awhh arta µ! ty arqw Sam.

hn!arl ry[h µ! zwl µlwaw Heb.


÷ymdqlm atrqd hm! zwl µrbw Onq.
atymdqb atrqd hm! hwh zwl µrbw Neof.
and tmdq ÷m atrqd am! zwl µrbw Ps.-J.
whmdql htrq µ! hzwl µwraw Sam.

Targ. Neof.’s rendering of Gen. 28.19 follows Onqelos’s quite closely,


with the exception of the final word, which places this verse into a
classification of category 2. Most of the differences are orthographic:
making Bethel one word rather than two and missing Targ. Onq.’s yod
in hym!. Targ. Neof. also adds a copula, as it regularly does. The one
obvious variance is Targ. Neof.’s atymdqb for Targ. Onq.’s ÷ymdqlm, and
the different prefix and suffix that require a category 2 classification
rather than a category 1.
For Targ. Ps.-J., its rendering of Gen. 28.19 follows that of Targ.
Onq. quite closely, but, like Targ. Neof., its rendering of the verse’s final
word places it into category 2. The key differences are Targ. Ps.-J.’s use
of aleph instead of Targ. Neof.’s and Targ. Onq.’s he, the dropping of
the ty, and its lack of Targ. Neof.’s added copula. For Targ. Onq.’s final
word, Targ. Ps.-J. drops Onq.’s lamed and separates the mem-prefix
into an independent preposition, and ends by adding and to finish the
sentence after using a different word ending, tmdq– instead of Targ.
Onq.’s ÷ymdq–.
In the Samaritan Targum, Gen. 28.19 is rendered much more like
the Hebrew text than the Aramaic found in the other versions—again
a category 4. Rather than using a dalet to create the construct state
between the third and fourth words, Sam. Targ. joins them together as
if they were Hebrew: arta µ!. It does the same for htrq µ! later in the
verse. For the last word, the Sam. Targ. gives whmdql, which follows the
Hebrew text’s hn!arl, rather than the more prepositional renderings of
the Jewish targums. So it is clear here that the Sam. Targ. created its
own translation rather than following one of those found in the other
targums.
92 Journal for the Aramaic Bible 3 (2001)

Now that we have seen the Neofiti verses in category 2 and how
those classifications compare to the translations of Targ. Ps.-J. and
Sam. Targ. with Targ. Onq., let us now turn to the verses that are
classified as category 4 for Neofiti.

Genesis 28.11
!m!h ab yk µ! ÷lyw µwqmb [gpyw Heb.
a!m! l[ yra ÷mt tbw artab [r[w Onq.
÷mt a!m! hyl t[mf µwra ÷mt tbw artab ylxw Neof.
a!my! [mf µwra ÷mt tbw a!dqwm tyb rtab ylxw Ps.-J.
h!m! l[ alh ÷mt tybaw artab [gpaw Sam.

awhh µwqmb bk!yw wyt!arm µ!yw µwqmh ynbam jqyw Heb.


awhh artab byk!w yhwdsya yw!w arta ynbam bysnw Onq.
awhh artab hl ûmdw h!ar ydsa twjt yw!w artad ywnba ÷m bsnw Neof.
awhh artab byk!w ywdsya yw!w !ydq rta ynbam h[bra bysnw Ps.-J.
awhh artab ûmdw htwhmdqm hb!w arta ynba ÷m bsnw Sam.

Targ. Neof.’s rendering of Gen. 28.11 falls into category 4, that of no


demonstrable links. This is because for three key verbs Targ. Neof.
uses different words from Targ. Onq.: ylx instead of Targ. Onq.’s [r[;
ûmd instead of Targ. Onq.’s byk!; and t[mf instead of Targ. Onq.’s l[.
This last difference is interesting, for it seems that Targ. Neof.’s au-
thor reacts against a rendering like that of Targ. Onq., which could
be understood as ‘the sun rising’, taking l[ from yl[ rather than from
ll[. Targ. Neof. then recasts the phrase by adding two words and
using the term [mf, which can only mean ‘to set, sink’. This differ-
ence, while perhaps done in reaction to Targum Onqelos’s transla-
tion, is sufficient to require the method to classify this verse as cate-
gory 4.28
Targ. Ps.-J.’s rendering of Gen. 28.11 falls into category 2. In terms
of orthography, it follows Targ. Onq. much more closely than does Targ.
Neof. There are a couple of dialectal changes: µwra for Targ. Onq.’s yra,
and the pronomial suffix yw– for Targ. Onq.’s yhw–. The key differences
are two points at which Targ. Ps.-J. follows Targ. Neof. rather than
Targ. Onq.: ylx instead of Targ. Onq.’s [r[, and [mf instead of Targ.
Onq.’s l[. If it was not for the close adherence of the remainder of
Targ. Ps.-J.’s verse to Targ. Onq., then these would probably put the
28
Note also Targ. Neof.’s conflate of one word from the Hebrew text and Targ.
Onq.’s translation of it. This suggests that Targ. Neof. is following Targ. Onq.’s
rendering. So the analysis of conflates suggests that Targ. Neof. knows and uses
Targ. Onq. here, even though the clausal parallels method cannot take that into
account.
Flesher Proto-Onqelos and Palestinian Targums 93

verse into category 4. But the otherwise general adherence to Targ.


Onq. enables its classification in category 2.
The rendering of Gen. 28.11 found in the Samaritan Targum again
suggests that it created its own independent translation of the Hebrew
text, a category 4 classification. The opening verb is unrelated to any
of the Jewish targums, [gpaw, while the Sam. Targ. chose three other
words with clear dialectal differences: ûmd, hb!, and alh. However, as in
Gen. 28.18, discussed above, the Sam. Targ. understands the Hebrew
wyt!arm as meaning ‘before’ or ‘in front of’, unlike the Jewish targums.
This interpretation makes clear that the Sam. Targ. translates inde-
pendently of the Jewish targums.

Genesis 28.13
ûyba µhrba yhla hwhy yna rmayw wyl[ bxn hwhy hnhw Heb.
ûwba µhrbad hyhla ywy ana rmaw yhwwl[ dt[m ywyd arqy ahw Onq.
ûwba µhrbad hhla yyy ana rmaw ywl[ µyyq yyy ahw Neof.
ûwba µhrbad hyqla yyy ana hyl rmaw ywly[ dt[m yyyd arqy ahw Ps.-J.
ûwba µhrbad hhla hwhy yna rmaw wybl[ µ[q hwhy haw Sam.

û[rzlw hnnta ûl hyl[ bk! hta r!a Årah qjxy yhlaw Heb.
ûnblw hnynta ûl hl[ byk! tad a[ra qjxyd hyhlaw Onq.
ûynblw hty ÷ta ûl hl[ ûmd ta a[ra qjxyd hhlaw Neof.
ûnblw hnynta ûl hl[ byk! tnad a[ra qjxyd hyqlaw Ps.-J.
û[rzlw hnnta ûl hyl[ ûmd htad ha[ra qjxyd hhlaw Sam.

Targ. Neof.’s rendering of Gen. 28.13 shows that it does not follow
Targ. Onq.’s translation. It lacks the word Targ. Onq. added before
the divine name and twice it uses key verbs different from those in
Targ. Onq.: µyyq instead of dt[m, and ûmd instead of byk!. So in this
verse, Targ. Neof.’s relationship to Targ. Onq. can only be considered
that of category 4.
Targ. Ps.-J. to Gen. 28.13, by contrast, follows Targ. Onq. by quot-
ing it directly, a clear category 1. Targ. Ps.-J. adds one word, hyl, makes
two orthographical changes: ywly[ for yhwwl[, and tna for ta.29 The only
other differences lie in the two targums’ different spelling of the divine
names. In this verse, then, Targ. Ps.-J. comprises an almost exact copy
of Targ. Onq.’s translation.
The Samaritan Targum to Gen. 28.13 again follows the Hebrew text
rather than suggesting any link to the Jewish targums, which earns it
a category 4 classification. It lacks Targ. Onq.’s added word, and it
gives the opening two words of God’s speech in an obvious Hebraism,
29
The addition of the nun in the last term, tna, is a standard practice of Targ.
Ps.-J.’s Late Jewish Literary Aramaic dialect. See Cook, ‘Rewriting’, pp. 126-29.
94 Journal for the Aramaic Bible 3 (2001)

hwhy yna. For the verse’s final word, it follows the Hebrew text directly
rather than the rendering found in the Jewish targums.

Genesis 28.14
hbgnw hnpxw hmdqw hmy txrpw Årah rp[k û[rz hyhw Heb.
amwrdlw anwpxlw ajndmlw abr[ml ¹qttw a[rad arp[k ÷yaygs ûnb ÷whyw Onq.
ajndmw amwrdw anwpxw abr[m ¹qttw a[rad arp[k ûynb ÷wwhyw Neof.
amwrdlw anwpyxlw ajnydmlw abr[ml ¹qtytw a[rad arp[k yh ÷yaygs ûnb ÷whyw Ps.-J.
hmwrdw hnwpxw h[ndmw hbr[m ytptw a[ra rp[k û[rz yyw Sam.

û[rzbw hmdah tjp!m lk ûb wkrbnw Heb.


ûnb lydbw a[ra ty[rz lk ûlydb ÷wkrbtyw Onq.
ûynb tyy[rzbw a[rad atyy[rz lk ûtwwkzb ÷wkrbtyw Neof.
ûnb twwkz ÷ygbw a[ra yswjyy lk ûtwwkz ÷ygb ÷wkrbtyw Ps.-J.
û[rzbw a[ra ynrk lk ûb ÷wkrbtyw Sam.

Targum Neofiti to Gen. 28.14 falls into category 4, that of no demon-


strable relationship to Targ. Onq. To start with, Targ. Neof. lacks the
three words that Targ. Onq. has added into its rendering of the Hebrew
text: ÷yaygs, ûlydb, and lydb. It also renders the list of cardinal direc-
tions in a way that more closely parallels the Hebrew text than Targ.
Onq.’s translation. So, according to the clausal parallel method, there
is no link between Targ. Neof. and Targ. Onq. in this verse. However,
if we switch to the two previously discussed methods, dependence of
Targ. Neof. on Targ. Onq. could be argued. The common rendering
of Hebrew text tjp!m as ty[rz suggests a link between the two tar-
gums. It is also possible to understand Targ. Neof.’s final two words
in this verse as a conflate rendering the Hebrew text and Targ. Onq.’s
translation.
In the opening clauses of Gen. 28.14, Targ. Ps.-J. follows Targ.
Onq.’s rendering rather closely. Targ. Ps.-J. has Targ. Onq.’s added
word and it gives the list of cardinal directions as it appears in Targ.
Onq. However, in the final clause, Targ. Ps.-J. seems to draw from
Targ. Neof.’s rendering and then goes its own way. So the relation-
ship between Targ. Onq. and Targ. Ps.-J. in this verse must also be
characterized as a category 4.
Finally, once again the Samaritan Targum receives a category 4
classification by following the Hebrew text much more closely than it
does any of the Jewish targums. It is the only targum to translate the
Hebrew text’s ‘seed’ exactly rather than metaphorically, and it does so
twice. It renders the list of directions in a way that closely reflects the
Hebrew. And, finally, it provides it own rendering of tjp!m, namely,
ynrk, from ÷rk meaning ‘clan’.
Flesher Proto-Onqelos and Palestinian Targums 95

Genesis 28.15

tazh hmdah la ûytb!hw ûlt r!a lkb ûytrm!w ûm[ ykna hnhw Heb.
adh a[ral ûnybytaw ûhtd rta lkb ûnyrfaw ûd[sb yrmym ahw Onq.
adh a[ral ûty rwzjaw lztd lkb ûty rfnaw ûm[ yrmymb hna ahw Neof.
adh a[ral ûnybytaw ûhtd rta lkb ûnyrfyaw ûd[sb yrmym ahw Ps.-J.
adh a[ral ûnrz[w ûhtd lkb ûnrfaw ûm[ hna haw Sam.

ûl ytrbd r!a ta yty![ µa r!a d[ ûbz[a al yk Heb.


ûl tylylmd ty dyb[ad d[ ûnyqb!a al yra Onq.
ûl tylylmd ty dyb[ad d[ ûm[ yrmym qwb!y al µwra Neof.
ûl tylylmd ty dyb[a yd ÷mz d[ ûnyqb!a al µwra Ps.-J.
ûl tllmd ty db[a ÷a d[ ûnqb!a al alh Sam.

Targum Neofiti to Gen. 28.15 does not demonstrably follow Targum


Onqelos here, preferring to produce its own translation of the Hebrew
text. In the first three phrases, it has different words from Targ. Onq.
for three terms: ûm[, lztd, and rwzjaw. In the second half of the verse,
Targ. Neof. follows Targ. Onq. a little more closely, with only the di-
alectal change of µwra for Targ. Onq.’s yra. But the other targums,
including Sam. Targ., also closely parallel Targ. Onq. and, more im-
portantly, the Hebrew text, as does Targ. Neof. So in the end, Targ.
Neof.’s rendering of this verse must be classified as a category 4.
Again in Gen. 28.15, Targ. Ps.-J. provides a near-exact quotation of
Targ. Onq., and so belongs in category 1. Targ. Ps.-J. has only a couple
of orthographical differences with Targ. Onq., and the standard dialec-
tally different rendering of the Hebrew yk, Targ. Ps.-J.’s µwra against
Targ. Onq.’s yra, which is so common as to be almost unnoticed.
The Samaritan Targum to Gen. 28.15 clearly does not draw from
Targ. Onq. or its copy-cat, Targ. Ps.-J. It contains a few similarities
with Targ. Neof., but these can be traced back to the Hebrew text.
Thus the Sam. Targ. again falls into category 4. Indeed, the Sam.
Targ. consistently provides its own rendering of the Hebrew text. It is
the only targum to follow the Hebrew text exactly in the verse’s first
phrase as well as to reproduce the Hebrew text’s ‘if’ in the second to
last phrase.

Genesis 28.20

ynrm!w ydm[ µyhla hyhy µa rmal rdn bq[y rdyw Heb.


ynnrfyw yd[sb ywyd armym yhy µa rmyml µyq bq[y µyyqw Onq.
yty rfnyw yd[sb yyy ywwhy ÷a rmyml rdn ?bq[yÀ rdnw Neof.
ynnyrfyw yd[sb yyyd armym yhy ÷ya rmyml µyyq bq[y µyyqw Ps.-J.
ynrfyw ym[ µyhla yhy µa rmyml rdn bq[y rdnw Sam.
96 Journal for the Aramaic Bible 3 (2001)

hzh ûrdb Heb.


adh ajrwab Onq.
ajrwab Neof.
÷ydh ajrwab atyyr[ yywlygw arkwn anjlwpw yakz µda twkyp!m Ps.-J.
÷dh h[rwab Sam.

!bll dgbw lkal µjl yl ÷tnw ûlwh ykna r!a Heb.


!blml wskw lkyml µyjl yl ÷tyw lyza anad Onq.
!blml !wblw lkyml µjl yl ÷tyw hb ûlhm ana yd Neof.
!wblyml wskw lkyml µyjl yl ÷tyw lyza anad Ps.-J.
!blml !wblw lkyml µjl yl ÷tyw lza hnad Sam.

There appears to be no relationship between Targ. Neof. and Targ.


Onq. in this verse, Gen. 28.20. Not only does Targ. Neof. have two
words different from Targ. Onq. in the first four (with Targ. Neof.
closer to the Hebrew text), but Targ. Neof. leaves out two words found
in Targ. Onq. and has different words for two key verbs: ûlhm and !wbl.
So all in all, the relationship between Targ. Neof. and Targ. Onq. in
this verse falls into category 4, namely, no link.
For Gen. 28.20, Targ. Ps.-J. follows Proto-Onqelos word-for-word,
with the exception of its seven-word insertion. In most words, Targ.
Ps.-J.’s orthography is that of Targ. Onq., with only four exceptions.
Targ. Ps.-J.’s relationship to Targ. Onq. is that of category 1.
The Samaritan Targum’s opening phrase makes it clear that it does
not follow Targ. Onq. or Targ. Ps.-J. The second phrase shows that
Sam. Targ. adheres closely to the Hebrew text rather than Targ. Neof.,
having two of the four Hebrew words exactly replicated and provid-
ing a distinctive rendering of the four words. In the last two clauses,
Sam. Targ. follows the Hebrew text closely and consistently, while its
choices agree now with one Jewish targum and now with another. No
indication here that the Sam. Targ. is anything other than a transla-
tion straight from the Hebrew text, thus earning it a final category 4
classification.
The results of the method of clausal parallelism analyzing the rela-
tionship of Targum Neofiti’s translation to that of Proto-Onqelos indi-
cates that four of the nine verses analyzed belong to category 2, thus
revealing regular evidence for Targum Neofiti’s use of Onqelos’s trans-
lation. The fact that its translation is category 2 rather than category
1 indicates that Targum Neofiti holds to its own dialect, that of Jewish
Targumic Aramaic, and alters the orthography, morphology and word
choice of Proto-Onqelos’s Jewish Literary Aramaic to fit its needs, even
as it tries to follow Proto-Onqelos’s renderings. The five verses in which
Targum Neofiti’s translation is not demonstrably linked to that of Tar-
gum Onqelos are just that, verses with no demonstrable link. That is,
Flesher Proto-Onqelos and Palestinian Targums 97

there is nothing in these verses to disprove a Targ. Neof./Targ. Onq.


link; indeed, there is much in each verse that is compatible with the
conclusion that Targum Neofiti drew from Proto-Onqelos’s transla-
tion. In Gen. 28.14, in fact, the application of the other two methods
demonstrated in this article reveals links of dependency between Tar-
gum Neofiti and Targum Onqelos.
Targum Pseudo-Jonathan provides a different set of results, one
which reflects its already recognized close relationship to Targum On-
qelos. For four of the ten verses of Gen. 28.11-20, Targum Pseudo-
Jonathan’s translation provides a direct quotation of Onqelos’s trans-
lation; these are category 1 renderings. The verses in question are Gen.
28.13, 15, 18, and 20. Three additional verses, Gen. 28.11, 16, 19, reveal
a category 2 relationship between the translations of Targum Pseudo-
Jonathan and Targum Onqelos. So seven of the nine verses analyzed
demonstrate Pseudo-Jonathan’s close and extensive use of Onqelos’s
translation as its own. Of the remaining verses, Gen. 28.12 falls into
category 3, which supports the conclusion of direct literary dependency,
while Gen. 28.14 alone belongs to category 4.
In comparison to the results for Targ. Neof./Targ. Onq. relation-
ship, the Targ. Ps.-J./Targ. Onq. relationship is much clearer and bet-
ter supported. Pseudo-Jonathan’s use of Targum Onqelos comprises
the defining characteristic of its translation. By contrast, Targum Ne-
ofiti’s use of Targum Onqelos is not only less extensive, but also of a
different character. If Pseudo-Jonathan’s relationship to Targum On-
qelos can be characterized as quotation, then Targum Neofiti’s rela-
tionship to Targum Onqelos constitutes a dialectal translation—the
rendering of Targum Onqelos’s Jewish Literary Aramaic translation
into the dialect of Jewish Targumic Aramaic.
For the Samaritan Targum, the results are clear and straightfor-
ward. The method of clausal parallels placed all nine verses into cat-
egory 4, indicating that the Samaritan Targum consistently based its
translation on the Hebrew text, and had no links to any of the Jewish
targums. Just as we expected.
Finally, with regard to the clausal parallel method, its proper ap-
plication enables the precise delineation of translations which are lit-
erarily dependent and those which are not. Rough distinctions can be
made simply by identifying the use of different words. But even when
word roots are the same, as in Gen. 28.19, careful reading shows where
each targum made different choices about which text to render (the
Hebrew text or another targum) and how to present that rendering.
Thus, although the method cannot evaluate every feature of a transla-
98 Journal for the Aramaic Bible 3 (2001)

tion, when used in conjunction with the other two methods, it provides
a solid evaluation of the relationship between two translations.

Conclusions
So what have we learned? Most importantly, we have identified three
methods which can be used to evaluate the relationships between two
translations of the same text, and to determine whether one transla-
tion draws from another or whether both draw only from the original.
The first method analyzes the translations for places where they do not
produce a literal rendering of the source text, but rather where they
differ from it. It then determines whether the translations are the same.
This approach identifies locations where the shared renderings of the
translation cannot be attributed to the original base text. The second
method comprises the identification of conflations. Here one targum
provides a rendering of the Hebrew text, a rendering which may per-
fectly legitimate or non-literal. A second targum takes that rendering
and combines it with one closer to the Hebrew text. This may be done
using a Hebraism, where the first targum did not, or by translating
the Hebrew text with a different term reflecting a different translation
approach to the original.
The third method, that of clausal parallels, provides an approach
to evaluating parallel translations where the targums do not vary dras-
tically from the Hebrew text. Instead it attempts to identify passages
where one targum quotes another, or quotes it with only minor differ-
ences. Here the search for parallels is not for a word or two, but rather
for a long clause or phrase where one translation follows the other.
This usually requires multiple clauses of ten to fifteen words or more,
for shorter clauses of only five or even seven words can occassionally be
similar by pure chance or by language similarities (as in the compar-
ison of Onqelos and the Samaritan Targum of Gen. 28.18). This new
method must be applied with a high degree of care and caution, for if
improperly used, it could confuse dependence on the original text for
dependence on a translation of it.
The tentative conclusions from the test application of these meth-
ods suggest important avenues for further investigation. All three meth-
ods indicated that Targum Neofiti drew from the translation found in
Proto-Onqelos. Although Pseudo-Jonathan clearly was more closely
tied to Proto-Onqelos’s translation, Neofiti also knew and used Proto-
Onqelos’s translation as the basis for its own. It regularly altered the
earlier translation to be consistent with its own dialect of Jewish Tar-
gumic Aramaic, and followed Proto-Onqelos’s non-literal interpretive
Flesher Proto-Onqelos and Palestinian Targums 99

Figure 2: Expansion and Translation Links

renderings in its own translation. Thus Proto-Onqelos may stand as the


base of the Palestinian Targums as well as Pseudo-Jonathan, as illus-
trated in Figure 2.30 Proto-Onqelos seems to have stayed in Palestine
rather than simply disappearing into Babylonia.
The impact of these results, if borne out by further research, be-
come rather suggestive when seen in the light of past scholarship, not
just that of Kutscher, as mentioned above, but also that of Kutscher’s
contemporaries and predecessors. G. Vermes identified three different
explanations for the relationship among the Palestinian Targums, Tar-
gum Onqelos, and Targum Pseudo-Jonathan in his essay, ‘The Targu-
mic Versions of Genesis 4:3-16’, namely, two previous scholarly posi-

30
The diagram includes one result not directly relevent to this study, but which
requires explanation, namely, the indication that Targ. Ps.-J. copied its translation
from Proto-Onqelos rather than from the finished Targ. Onq. This conclusion stems
from the extensive amount of additional material in Targ. Onq. that Targ. Ps.-J.
does not use. This suggests that Targ. Ps.-J. drew from Proto-Onqelos and that
the additions it does not follow were composed in Targ. Onq.’s second stage in
Babylonia. See Flesher, ‘Targum Onkelos’, pp. 68-69.
100 Journal for the Aramaic Bible 3 (2001)

tions and his own new suggestion.31 The results of this paper’s anal-
ysis suggest that these positions may all be correct, although in ways
their proponents probably did not imagine. The first scholarly position
Vermes identified was that Pseudo-Jonathan ‘consists of O[nqelos] sup-
plemented by extracts from the PTs’. This is generally correct, except
that Pseudo-Jonathan consists of Proto-Onqelos, supplemented by ex-
tracts from the PTs (and a larger amount of non-PT additional ma-
terial).32 The second scholarly position discussed by Vermes was that
Pseudo-Jonathan is ‘a Palestinian Targum whose original text has been
reshaped after O[nqelos]’.33 This holds generally correct, except that
Targ. Ps.-J. is a Palestinian Targum whose translation text has been
shaped in dialogue with Proto-Onqelos (and by the non-PT additional
material). Vermes suggested his own interpretation, namely, ‘that it is
O[nqelos] that depends on [Pseudo-Jonathan], this dependence being
either direct . . . , or indirect (O[nqelos] and [Pseudo-Jonathan] deriving
from a common targumic source)’.34 The results of this paper’s analy-
sis suggest that Vermes’ ‘common targumic source’ was Proto-Onqelos.
Of course, Vermes understood this source to be more expansive than
Onqelos, rather than the less expansive character of Proto-Onqelos. In
the end, however, it becomes possible to paraphrase another observa-
tion by Vermes’ and suggest that Proto-Onqelos is as fundamentally
Palestinian as Targum Pseudo-Jonathan, the Fragmentary Targums,
Targum Neofiti and the Palestinian Targum Fragments from the Cairo
Geniza.35

31
G. Vermes, ‘The Targumic Versions of Genesis 4’.
32
The PT extracts being the Proto-PT source identified in my studies referenced
above, and the additional material of Targ. Ps.-J. consisting of the PJ-unique source
identified by Beverly Mortensen. See above.
33
Vermes, ‘The Targumic Versions of Genesis 4’, pp. 109-10.
34
Vermes, ‘The Targumic Versions of Genesis 4’, pp. 110-11.
35
Vermes, ‘The Targumic Versions of Genesis 4’, p. 111.

You might also like