Some Problems in Determining the Vorlage of Early Syriac Versions of the NT

P. J. Williams

Tyndale House, Cambridge, CB3 9BA, England

This paper has been published in New Testament Studies 47 (2001) 537-43, published by Cambridge University Press. © 2001 New Testament Studies. My apologies that I have been unable to solve some problems of font wrap and spacing. For footnotes see the published version.

At present serious study of the level of certainty or uncertainty there is in determining the Vorlage of the early Syriac versions of the NT is still in its infancy. The Nestle-Aland Novum Testamentum Graece (27th edition; hereafter NTG), by virtue of its wide distribution, tends to make itself the consensus opinion of how such versional evidence should be used. This work is a monumental achievement and an indispensable tool by which many scholars achieve their main access to the early versions. However, while the edition in many ways represents an improvement over previous editions of the Greek NT in its references to Syriac evidence, it still leaves a lot to be desired in its representation of the earlier Syriac versions. The stated policy of the edition is that the versions only appear in the apparatus when ‘their underlying Greek text can be determined with confidence’ (NTG, Introduction, p. *63). This policy is not here disputed. However, the method used to deduce the Vorlage of the Syriac versions is still not rigorous. For a high degree of certainty to be achieved in the citation of evidence it is necessary for concordance checks of Syriac to be performed to ensure that variations in Syriac are not due to natural preferences in that language. Now that a concordance of the NT of the Peshitta (P) is available and also an analytical key to the gospels, although further tools are still desiderata (for instance, a concordance of the Old Syriac gospels), it is obvious that NTG’s references to Syriac evidence need to be revised. This will be illustrated with several examples below.

The degree of care that needs to be taken with an apparatus can be appreciated by reference to variants related to the word karpo/j, ‘fruit’. While in the text of NTG there are 50 occurrences of karpo/j in the singular, the plural occurs a mere 16 times. On the other hand, the Syriac word )r)P ‘fruit’, the standard equivalent of karpo/j whether singular or plural, occurs 77 times in P, all but 3 of these in the plural. There is thus an overwhelming idiomatic tendency for Syriac to have the plural where Greek uses the singular. The three singular occurrences in Syriac are (1) the expression ‘fruit of the womb’ in Luke 1.42, referring to the unborn Jesus, evidently singular, and (2) the botanical phenomenon mentioned in Mark 4.28 and 29. Despite P’s evident preference for the plural, NTG cites P as witness for the plural when Greek witnesses show variants on four occasions (Luke 3.9; 6.43 twice; Phil 1.11). It also cites Old Syriac witnesses S (Sinaitic) and C (Curetonian) in Luke 3.9 and S in Luke 6.43. To use the Old Syriac, however, is just as precarious as to use P, since it like P shows an overwhelming preference for the plural of the word ‘fruit’. The situation is more complex in the Old Syriac since not all plurals receive the plural marker seyame. Without suffixes or syntactic markers the singular and plural forms of )r)P are indistinguishable at a consonantal level. However, the absence of seyame in S on forms morphologically marked as plural (e.g. Luke 6.44; John 15.16), along with the frequent absence of seyame throughout S, suggests that it is possible to read all the forms in the Old Syriac as plural with the exception of Luke 1.42. Thus that in S ))YGS )r)P in John 15.5 with seyame occurs so close to same phrase in v. 2 without seyame suggests that the occurrence in v. 2 is also plural, and means ‘many fruits’ not ‘much fruit’. Whether or not visually ambiguous cases in the Old Syriac are plural, the Old Syriac’s tendency to pluralise can be unambiguously observed when all nine of the singular examples of karpo/j in Matt are, as shown by seyame or their context, read by C and S whenever they are extant as plural (S is not preserved for Matt 7.19; 21.19). The consistent Syriac idiom makes it questionable whether one can adduce the Syriac versions in the way NTG does. It might be objected that in Luke 3.9 and 6.43 the Syriac witnesses show agreement with ‘Western’ textual witnesses and that this confirms the retroversion in NTG’s apparatus. However, the consistent Syriac idiom undermines the certainty which the Introduction of NTG supposedly requires for versional authorities to be quoted in the apparatus. It is entirely possible that a Greek plural underlies these occurrences in Syriac of the plural of )r)P. However, there is simply no way of knowing whether or not this is the case. Further confirmation that a Syriac translator of the NT might well have translated a singular in his Vorlage by a plural is found in the Syriac OT. Whereas Hebrew yrp, ‘fruit’ is invariably singular, its cognate )r)P in Syriac is almost invariably plural.

The problem is not that NTG’s apparatus informs readers that here is yet another agreement between Syriac witnesses and ‘Western’ ones. This is indeed useful to know. The problem is the statement in the Introduction of the purpose of citation of versional witnesses in the apparatus as giving the ‘underlying Greek text’. Various hypotheses can be raised to explain common elements between ‘Western’ witnesses and Syriac versions, one of which is naturally that Syriac witnesses were translated from a text with a plural as in the ‘Western’ text. However, this is merely one hypothesis. One can as readily posit that ‘Western’ witnesses have been influenced by Syriac or that the Syriac plural derives from the Diatessaron (which in turn could have had a singular or plural as its Vorlage), or that the Syriac and ‘Western’ witnesses agree by independent convergence, the latter being assimilated to a parallel passage (as Luke 6.43 D may have been assimilated to Matt 7.17-18). Yet none of these other possibilities would justify recording the Syriac in the apparatus according to the stated intention of NTG.

Another example of an idiom where Syriac witnesses need to be treated with greater care is that of the phrase ‘all these things’, pa/nta tau~ta or tau~ta pa/nta (by which is meant the neuter plural ‘near’ demonstrative pronoun in any case, not modifying a noun, but modified by pa~j). The order of these two words varies considerably and thus we find that NTG marks a variation of word order in its textual notes for nine locations (Matt 6.32; 19.20; 23.36; 24.8, 33, 34; Mark 13.30; Luke 24.9; Acts 7.50; there is a further variant in Luke 16.14, but not about word order). In its main text the order tau~ta pa/nta predominates (20 occurrences versus 12). In its apparatus for 6 of the variants NTG cites Syriac versions as part of the evidence for or against a variant (Matt 6.32; 24.8, 33, 34; Mark 13.30; Luke 24.9; see also Luke 16.14). In the first of these C is quoted as supporting the order tau~ta ga\r pa/nta as opposed to the text’s pa/nta ga\r tau~ta, the witnesses for which are not cited. In 3 cases in Matt 24 (vv. 8, 33, 34) where the main text reads pa/nta (…) tau~ta, P is quoted as supporting the opposite order, and alongside it in the first and third occurrences S is quoted. The order tau~ta (…) pa/nta is printed in the text in Mark 13.30 (with the support of SP), and in Luke 24.9 (with P). The point to note here is that in fact on every occasion where the Syriac is quoted as a witness to the text, whether on behalf of what is printed by NTG or against it, it supports the order tau~ta (…) pa/nta. This in itself should give us serious pause about the use of Syriac witnesses at this point. Scepticism about this use of witnesses only increases when the readings of the Syriac versions are considered in all the cases of tau~ta with pa/nta in the NT. In fact, in all cases where the Old Syriac witnesses have an equivalent of both pa/nta and tau~ta they have the order with the demonstrative preceding ‘all’ (at least one of the Old Syriac witnesses does not have a direct correspondence with the Greek words in Matt 4.9; Mark 7.23; 10.20; 13.4; Luke 18.21; 21.36; 24.9). In the NT of P, when the demonstrative oYLh ‘these’ does not qualify a noun, oYhLK (‘all’) follows the demonstrative 34 times, whereas the reverse order occurs a mere 6 times. It seems then that there was a strong preference for the idiom with ‘all’ following the demonstrative, and that therefore the Syriac translators may have inverted the elements of their Greek Vorlage in translation.

Further confirmation of the possibility that a Syriac translator might invert the order of ‘all’ and the demonstrative for reasons of idiom is found in the Syriac translation of the OT. In the OT the phrase hl)-lk, ‘all these’ (not an exclusively neuter phrase), has a fixed order and occurs 53 times in the MT. However, on the majority of occasions where this phrase is translated into Syriac the demonstrative precedes the word ‘all’. The inversion occurs in 25 cases as opposed to 18 which retain the Hebrew order. The 10 other examples do not display a direct enough correspondence between the Hebrew and Syriac to be used in this discussion. When both elements, the demonstrative and ‘all’, are represented, there are some interesting word order patterns in Syriac, which suggest that it cannot be used as a direct witness to the word order of its Vorlage. In general ‘all’ follows the demonstrative in the Syriac OT when it is the object, or follows a preposition.

A final example where the Vorlage of a Syriac version is less obvious than appears at first is in the name of Satan, or the devil. In a single variant in Luke 4.2, NTG quotes S alongside D pc e as supporting tou~ satana~ instead of the text’s tou~ diabo/lou. This is problematic since, although Syriac versions often use the form )crQ lK), ‘accuser’, to represent Greek dia/boloj, )crQ lK) is far from a uniform equivalent. The Peshitta represents dia/boloj by )N+S, ‘Satan’, in Matt 13.39; John 6.70; 13.2; 1 Tim 3.6, 7; 2 Tim 2.26; Heb 2.14; Jas 4.7; 1 Pet 5.8; 1 John 3.8 (three times), 10, and perhaps also in Luke 4.5 where dia/boloj is omitted in the text of NTG. When we consider the picture in the Old Syriac gospels it is even more striking. For the 15 occurrences of dia/boloj in the gospels the Old Syriac witnesses (where extant) show similar numbers of occurrences of )crQ lK) and )N+S, 4 and 5 respectively, with the witnesses split in a further 2 cases: Matt 4.1, with C reading )N+S and S )crQ lK), and Matt 4.8, with the witnesses reading the reverse. In Matt 13.39 and John 8.44 )$YB, ‘evil one’, is used, and in Luke 8.12 )BBdL(B, ‘enemy’. Given the variety of Syriac phrases used where there are no widely attested Greek textual variants it seems unwise to take a single occurrence of a Syriac phrase as representing a textual variant. It seems that there was no fixed equivalent of dia/boloj in the earliest Syriac, or even at a later stage when P was made. This stands in contrast to satana~j which is rendered )N+S in Peshitta and the Old Syriac whenever extant with the exception of Luke 13.16 C (see below). Not all variety can be explained, and the reason for greater consistency with satana~j may be that the close equivalent )N+S suggested itself so readily. However, an explanation for the variety of correspondences in Syriac may be along the following lines. Syriac has no definite article and partially compensates for this by representing definiteness by an intricate system of anticipatory suffixes, object markers, demonstratives, and so forth. Sometimes these are not available and a literal translation of the Greek would create an ambiguous text. This is the case with )crQ lK), ‘The Accuser’, or equally ‘an accuser’. The title )N+S is less ambiguous than )crQ lK). Sometimes the diabolic character of a figure mentioned is contextually obvious (as in Matt 25.41) and so )crQ lK) can occur. When the identity of the accuser is obvious )crQ lK) can even occur as an equivalent of o9 satana~j (Luke 13.16 C). But apart from these examples the Old Syriac has no examples of )crQ lK) used when the devil has not already been introduced as a character, except in Matt 4.1, where as noted above the witnesses of the Old Syriac are split. The ambiguity that would be created in some situations is only too obvious if we consider how Syriac might translate John 6.70 (kai\ e0c u9mw~n ei[j dia/bolo/j e0stin). How could a translator who used )crQ lK) in this context avoid the possible understanding of what he had written as ‘and one of you is an accuser’? The solution was simple: to use )N+S. The same tendency to avoid ambiguity seems to be present in P, which also uses )N+S as an equivalent of belia/r in 2 Cor 6.15, and to avoid saying ‘resist an accuser and he will flee from you’ (there being no clear contextual indication that the devil was meant) explicitly mentions Satan in Jas 4.7.

To conclude, this brief survey of three areas of use of Syriac as a textual witness has shown weaknesses in the way the apparatus of NTG was compiled if citation of Syriac evidence is truly meant to reflect a high degree of confidence about its underlying Greek text rather than merely striking agreement between diverse witnesses. The work of the International Greek New Testament Project avoids such a criticism by its own caveat on the use of its apparatus, though for instance it lists S as supporting the plural karpou/j in Luke 6.43. It must be asked, however, whether even that project is not liable to give unsuspecting readers a false impression of how widespread a Greek reading may have been by failing to alert readers to the fact that a version may have a particular translation tendency. The Biblia Hebraica Quinta project for the OT avoids this problem by pointing out where variant wordings are likely to have arisen by the process of translation. However, it may err too far the other way in advising readers against positing a variant Vorlage from MT. That said, it is easier to choose to ignore the advice of the Biblia Hebraica Quinta than it is to become aware of translation tendencies, which are usually only known amongst specialists on a particular version. It is arguably therefore the task of those preparing critical editions and suggesting versional parallels more than it is the task of the general user to note such translational issues. Thus whenever a version is considered as a witness the word or construction analysed must be considered in all its occurrences in that version (or better still, in its occurrences in wider corpora). This has clearly not been done comprehensively for the Syriac witnesses in preparation of critical editions of the Greek NT to date; given the previous lack of concordances and the sheer volume of the task one can easily understand why this sort of verification was not undertaken. The production of editions like NTG already represents a vast amount of work, but still more needs to be done to improve on its standards. At present a widely agreed standard is that versional evidence should only very rarely be treated as evidence for a variant when there is no support from Greek authorities. Thus )N+S in S is treated as a variant in Luke 4.2 (when it is supported by D pc e), but not in Luke 4.5. However, even this standard fails to consider a version in its own right, and does not examine mechanisms whereby a version could arrive independently at a translation which parallels variant Greek texts. Of course a relationship between Syriac texts and the ‘Western’ text is undeniable, and thus the Syriac in many cases is taken to support a variant attested in Greek witnesses. However, even this small survey has shown that the origin of some such variants can be well explained as due to Syriac idiom or preferences. This in turn might give weight to the proposal that some ‘Western’ readings originated from Syriac.

The conclusion that NTG is deficient in its representation of Syriac evidence is not in any way meant to reflect badly on those who undertook the work, but merely notes the provisional nature of their conclusions and the need for us now to move on to a better method.