Some Alleged Confusions in Translation from Hebrew to Greek*

Trevor V. Evans

I. The Phenomenon of Confused Verbal Renderings 

Two remarkable passages in the Greek translation of Numbers have recently been identified by Anssi Voitila. These are Num 9,16-23 and 10,11-25. They show clear influence from Hebrew verbal forms on the translator's choices of Greek verbal forms. In both cases this type of influence extends to the degree of overriding the semantic indicators of the broader context. Confused translations result. Voitila's isolation of these passages is an important contribution to our understanding of the Greek translators' sensitivity to the Hebrew verbal system1. He in fact goes somewhat further, describing more than a dozen other instances of 'confusions' in the Greek Pentateuch where indicative tense-forms are alleged to be contextually inappropriate. Influence from the underlying Hebrew verbs is presented in all cases as the cause of confusion. Here we enter debatable territory. While the Numbers examples already cited are certain, all the others seem to me open to alternative interpretations. There is scope for reassessment of the individual cases, and also of the broader implications of the data. If we find Voitila's additional examples persuasive, the phenomenon is a general feature of translators' habits and will demand a general explanation. If we do not find them convincing, however, the Num 9,16-23 and 10,11-25 passages take on a special appearance. They will have to be seen as isolated anomalies in the work of a particular translator2.

The problem of how we are to prove translators' mistakes will be crucial to the analysis. Voitila is able to demonstrate numerous instances of influence from specific Hebrew verbal forms on Greek verbal choices. But do such identifications in themselves indicate confusion? To isolate an error of the type under investigation it surely has to be shown that influence from Hebrew text components motivates a Greek verbal form which fails to agree with the sense of its context. This must not simply mean the underlying Hebrew context, as it is explained by modern interpreters. Where the ancient interpretation offered by the LXX translators yields a rendering different from what might be suggested today we need to consider the Greek on its own terms. In my view, we must decide whether a particular translation makes sense within its particular Greek context before labelling it an example of confusion. In addition, Voitila's interpretation of the demands even of the original Hebrew context does not always convince. And where more than one modern interpretation of a passage is possible, we will be on shaky ground if we insist on what the ancient Pentateuchal translators ought to have done.

Before turning to the debatable examples, let us consider the nature of the phenomenon manifested in Numbers 9 and 10. As is well-known, the indicative forms of the Greek and Hebrew verbal systems function in markedly different ways. There are essentially only two indicative forms in Biblical Hebrew. These, unlike the range of seven Greek indicative forms, do not in themselves convey temporal values3. The Hebrew perfect and its practical equivalent the consecutive imperfect usually occur within past time-frames, but may appear within present or future time-frames. Conversely, the Hebrew imperfect and its practical equivalent the consecutive perfect usually occur within future time-frames, but may appear within present or past frames4. For these two pairs of forms (and for most other forms of the Hebrew verb) the translators of the Greek Pentateuch developed regular renderings, according to the natural suitability of these renderings for the usual contexts in which the Hebrew forms occur. This was presumably a quite unconscious process, at least in its origins. The habit normally yields a sensible Greek rendering.

Thus, Hebrew imperfects and consecutive perfects, since they most frequently occur in future time-frames, regularly motivate Greek future indicatives in all the Pentateuchal books5. In Num 9,16-23 and 10,11-25, however, these Hebrew forms occur within past time-frames. The Numbers translator usually handles such contexts effectively, translating by means of a Greek past tense, such as the Greek imperfects in Num 11,5-96, the aorist ei]pen for rm)yw in Num 10,35 (MT 10,36), and the imperfect e!zh for yxw in Num 21,9. Thus we find in Num 9,15 h]n for hyhy (not mentioned by Voitila), in v. 16 e)gi/neto for hyhy and e)ka/lupten for wnsky, in v. 17 a)ph=ran for w(sy, e!sth for Nk#$y, and parene/balon for wnxy. In all these instances context would appear to have been a more potent influence than particular Hebrew forms.

Yet in vv. 18-23 there is an odd switch to the characteristic future indicative rendering of Hebrew imperfects and consecutive perfects, before a change back to the aorist e)fula/canto for the perfect wrm#$ (another characteristic equivalency) in v. 23. Contextually inappropriate renderings result, shifting the temporal setting abruptly from past to future and then back again. The translator has surely been misled by the specific influence of the Hebrew tense forms, mechanically employing the regular translation equivalents. The case of Num 10,17-25 is similar. In that passage Voitila identifies a series of six consecutive perfects within a past time-frame translated by future indicatives, producing an abrupt temporal switch in the Greek after a series of aorist indicatives neatly rendering Hebrew perfects and consecutive imperfects7.

These instances, to which we shall return, are by no means the first to be isolated of influence from Hebrew verbs yielding Greek verbal renderings at odds with their contexts. A number of examples, mainly from Psalms, are discussed in Barr's earlier investigation of the choice of Greek tenses in past and future referring contexts8. There it is argued that general content and context are far more important factors in determining Greek tense forms than the underlying Hebrew tenses and that confused renderings characteristically arise in contexts which are 'semantically ambiguous' with regard to the past/future contrast, especially poetic literature. The special significance of Voitila's examples rests in the fact that they occur within historical narrative. Semantic ambiguity, in Barr's terms, does not seem to be a factor.

 

II. A Reassessment of the Data 

Voitila identifies confusions of the type described above in all the Pentateuchal books except Leviticus, and especially in the work of the Numbers translator. The examples are introduced below according to the order of his 1996 treatment9.

Gen 44,20: kai_ ei!pamen tw=| kuri/w|  1 Estin h(mi=n path_r presbu/teroj kai_ paidi/on gh/rwj new/teron au)tw=| kai_ o( a)delfo_j au)tou= a)pe/qanen, au)to_j de_ mo/noj u(pelei/fqh th=| mhtri_ au)tou= o( de_ path_r au)to_n h)ga/phsen

 dlyw Nqz b) wnl-#$y ynd)-l) rm)nw
wyb)w wm)l wdbl )wh rtwyw tm wyx)w N+q Mynqz
wbh)

As Voitila points out, wbh) is a Hebrew stative verb, a type traditionally interpreted as expressing a continuous state and conventionally translated into modern European languages by present tense forms10. Thus the RSV has here 'and his father loves him'. Voitila regards the present indicative as the correct Greek translation and believes the context itself indicates this on the grounds that '[b]oth persons under discussion, the younger brother who is loved and the father who loves, are still very much alive'11. The translator's choice of the aorist indicative as a rendering (as in Gen 22,2) may indicate ignorance of the semantics of Hebrew statives. Whether it produces an inappropriate interpretation of the context is another matter.

The verb a)gapw= is lexically stative in this sense of parental love. The aorist indicatives of Greek stative verbs often have an ingressive force12. In Gen 44,20 Judah is reminding the unrecognized Joseph of his brothers' earlier report of Jacob and Benjamin. The aorist h)ga/phsen completes a sequence of aorist indicatives and is used, I suggest, to express the natural consequence of the death of the elder brother (a)pe/qanen), leaving the younger brother as the only remaining son of Jacob's favourite wife (u(pelei/fqh). If this was the translator's interpretation of the context, the fact that both father and son are still alive is beside the point. The focus of the Greek sentence is on the causes and origin of Jacob's special affection for Benjamin. I would translate 'and we said to the master (i.e. you) "we have an elderly father, and he has a child of his old age, rather young, and his brother died, and he alone was left behind for his mother, and his father (as a result) came to love him"'.

Gen 22,2: kai_ ei]pen labe_ to_n ui(o/n sou to_n a)gaphto/n o$n h)ga/phsaj

tbh)-r#$) Kdyxy-t) Knb-t) )n-xq rm)yw

 

The example of Gen 22,2 involves the same two verbs and essentially the same problem, h)ga/phsaj rendering tbh). Influence on choice of the aorist indicative from the underlying Hebrew perfect does not make the Greek sense 'and he said "Take your beloved13 son, whom you have come to love"' inappropriate to the context14.

Gen 1,21: kai_ e)poi/hsen o( qeo_j ta_ kh/th ta_ mega/la kai_ pa=san yuxh_n zw/|wn e(rpetw=n a$ e)ch/gagen ta_ u#data kata_ ge/nh au)tw=n

 Myldgh Mnynth-t) Myhl) )rbyw
 wcr#$ r#$) t+mrh hyxh #$pn-lk t)w
Mhnyml Mymh

 

An allegedly similar case is the rendering of wcr#$ by e)ch/gagen in Gen 1,21. Here the difference in lexical values is important. The Hebrew verb means 'teem with'. The Greek verb means 'bring forth'. As a result the sense of the entire sentence is decisively altered in the Greek, where the relative clause can be taken to express consequent occurrence. I translate the Greek 'God made the great sea-monsters and every living thing among creeping animals, which the waters (then) brought forth according to their kinds'.

Num 25,17-18: 'Exqrai/nete toi=j Madihnai/oij kai_ pata/cate au)tou/j, o#ti e)xqrai/nousin au)toi_ u(mi=n e)n dolio/thti, o#sa doliou=sin u(ma=j dia_ Fogw/r, kai_ dia_ Xasbi_ qugate/ra a!rxontoj Madia_n a)delfh_n au)tw=n, th_n peplhgui=an e)n th=| h(me/ra| th=j plhgh=j dia_ Fogw/r

.Mtw) Mtykhw Mynydmh-t) rwrc
 Mhylknb Mkl Mh Myrrc yk
Mkl wlkn-r#$)
 ybzk rbd-l(w rw(p-rbd-l(
 hkmh Mtx) Nydm )y#&n-tb
rw(p-rbd-l( hpgmh-Mwyb
 

 

Voitila's explanation that the present indicative e)xqrai/nousin in Num 25,18 is motivated by the Hebrew participial expression Mh Myrrc and in turn encourages the use of the present doliou=sin for the perfect wlkn is satisfying15. Difficulties arise with the comment that what precedes is a narrative passage. Immediately before the sentence in question we in fact find a series of imperatives, la/lhson (a Greek plus) in v. 16, e)xqrai/nete for the imperatival infinitive absolute rwrc and pata/cate for the consecutive perfect Mtykhw in v. 17. So it is hardly surprising to find the translator taking Mh Myrrc to refer to the present and translating with e)xqrai/nousin. As Voitila recognizes, it is context which then motivates the choice of doliou=sin for wlkn. The Greek makes sense as a general expression of the effects of Midianite influence in the matter of the local god Phogor, which is a continuing danger. Only mention of the matter of Chasbi introduces a specific reference to a past event and this may be read as an amplifying addition regarding the deceit of the Midianites. One might translate 'to the extent they are deceiving you through Phogor, and (have done) through Chasbi' if it is felt necessary to spell out the temporal sequence (in any case Chasbi is to some extent a symbol of an ongoing temptation — there are other Midianite women). Once again the rendering distorts the sense of the Hebrew, but does not have to be interpreted as indicating lack of sensitivity to the broader context. The Greek version may be translated 'be enemies to the Midianites and smite them, for they are enemies to you in their deceit, to the extent they are deceiving you through Phogor, and (have done) through Chasbi daughter of the ruler of Midian, their sister, the woman struck down on the day of the plague on account of Phogor'.

Num 14,13-14: kai_ ei]pen Mwush=j pro_j ku/rion kai_ a)kou/setai Ai!guptoj o#ti a)nh/gagej e)n i)sxu/i sou to_n lao_n tou=ton e)c au)tw=n, a)lla_ kai_ pa/ntej oi( katoikou=ntej e)pi_ th=j gh=j tau/thj a)khko/asin o#ti su_ ei] ku/rioj e)n tw=| law=| tou/tw|

 w(m#$w hwhy-l) h#$m rm)yw
 .wbrqm hzh M(h-t) Kxkb tyl(h-yk Myrcm
 w(m#$ t)zh Cr)h b#$wy-l)
wrm)w
hzh M(h brqb hwhy ht)-yk

 

Num 14,24: o( de_ pai=j mou Xale/b, o#ti e)genh/qh pneu=ma e#teron e)n au)tw=| kai_ e)phkolou/qhse/n moi kai_ ei)sa/cw au)to_n ei)j th_n gh=n, ei)j h$n ei)sh=lqen e)kei=, kai_ to_ spe/rma au)tou= klhronomh/sei au)th/n

  trx) xwr htyh bq( blk ydb(w
 wyt)ybhw yrx) )lmyw wm(
hn#$rwy w(rzw
hm#$ )b-r#$) Cr)h-l)

The rendering of Myrcm w(m#$w by kai_ a/kou/setai Ai!guptoj in Num 14,13 raises two relevant questions. First, potential problems lurk in attempts to rationalize rhetoric. As Voitila observes, the Egyptians may be presumed at this point in the narrative to be well aware of the removal of the Israelites from Egypt16. Moses, however, has the present and future in mind. The central idea developed in vv.13-16 is that if God destroys his chosen people after bringing them by strength out of Egypt, other nations will assert that it is because he has lacked the power to deliver them into the promised land. Therefore, I take the future a)kou/setai to produce a clause illogical in strict terms of chronological sequence, but hardly out of keeping with Moses' rhetorical point. This is not to suggest conscious nicety of style on the part of the translator, though such features can be identified in the Greek Pentateuch17, but rather a simple feeling for the perceived flow of Moses' argument. It does not necessarily indicate blindness to the demands of the context. I translate the cited Greek passage of vv. 13-14 'and Moses said to the Lord "And Egypt will hear that you have brought up through your strength this people from among them, but also all those dwelling in this land have heard that you are lord among this people"'. In v. 15 we have the future indicatives e)ktri/yeij for htmhw and e)rou=sin for wrm)w. The form a)kou/setai is aligned with these.

Second, the Greek text manifests distortions which go well beyond renderings of particular verbal forms. It appears in part to misunderstand the syntax of the Hebrew construction, and in v. 14 to follow a Vorlage different from and inferior to the MT. The Masoretic punctuation indicates that hzh M(h-t) Kxkb tyl(h-yk Myrcm w(m#$w in v. 13 is to be taken 'and the Egyptians will hear, for you have brought up in your strength this people'. The Numbers translator has rendered yk by o#ti and understood the following words as an indirect statement depending on a)kou/setai. So the chronological question raised by Voitila in relation to the Greek is not a factor in the Hebrew text. Then in v. 14 the initial expression Cr)h b#$wy-l) wrm)w 'and they will tell the inhabitants of this land' of the MT is unlikely to be identical to the Vorlage actually underlying a)lla_ kai_ pa/ntej oi( katoikou=ntej e)pi_ th=j gh=j tau/thj18. These matters complicate assessment of the translator's alertness to form and context.

In the case of Num 14,24, the sense according to Voitila 'is surely to describe the present state of Caleb'19. This conclusion seems particularly unconvincing, imposing an unnecessary interpretation on the Hebrew text even before we come to the Greek. The RSV, for instance, translates the Hebrew '[b]ut my servant Caleb, because he has a different spirit and has followed me fully, I will bring into the land...', and wm( trx) xwr htyh could as effectively be rendered in English as 'a different spirit has been in him', i.e. as demonstrated in the events immediately precipitating God's threat to destroy the Israelites. So it is difficult for me to accept confusion as an explanation for the Numbers translator's choices of e)genh/qh for htyh, e)phkolou/qhsen for )lmyw, and especially ei)sa/cw for wyt)ybhw. In the preceding verses we have been told that none of those who have in the past seen the glorious works of God will be permitted in the future to see the promised land. Caleb will receive different treatment in the future because in the past (albeit the immediate past) he has behaved differently. I translate the Greek 'but my servant Caleb, because a different spirit has been in him and he has obeyed me (just now), I shall also bring into the land, into which he has (already) gone, and his descendants will inherit it'.

Exod 3,17: kai_ ei]pa a)nabiba/sw u(ma=j e)k th=j kakw/sewj tw=n Ai)gupti/wn

Myrcm yn(m Mkt) hl() rm)w 

Exod 4,22-23: su_ de_ e)rei=j tw=| Faraw/ ta/de le/gei ku/rioj ui(o_j prwto/toko/j mou 'Israh/l: ei]pa de/ soi e)capo/steilon to_n lao/n mou i#na moi latreu/sh|

 .l)r#&y yrkb ynb hwhy rm) hk h(rp-l) trm)w
yndb(yw ynb-t) xl#$ Kyl) rm)w

 

Voitila's rationalizing objection to the aorist ei]pa for the consecutive imperfect rm)w in Exod 3,17 and 4,23 has obvious attraction. The present tense would be highly appropriate to both — very similar — contexts. We may also compare le/gei, which belongs to a developing translational formula20, for the perfect rm) in Exod 4,22. Past tense forms seem odd, arguably suggesting something mentioned earlier in the text21, and here again it is hard to doubt as a motivation the translator's awareness of underlying consecutive imperfects.

Given the unequivocal examples of mistaken verbal renderings in Numbers 9 and 10, it would not surprise to find additional instances. Perhaps these two Exodus examples are such. They are certainly difficult cases. But are the Greek translations nonsensical? Here too a different explanation seems possible. As in the case of Num 14,13 a)kou/setai we should consider the issue of rhetorical style. Voitila has captured the crucial point in noting 'God tells what he has just decided'22. God's utterances in these Exodus examples represent his decisions for the future. Thus, 'I have said' in both cases approaches the meaning 'I have decided', indicating divine intention 23. I argue that we need to think in terms not of a previous utterance, but of a decision in the immediate past, which is itself focused on the future time-frame of the following direct speech. If this is correct, the Exod 3,17 instance may be translated: '...and my word is "I will bring you up out of the oppression of the Egyptians"'. The Exod 4,22-23 instance may be translated:

...and you will say to Pharaoh, 'thus says the Lord, my first-born son is Israel, and my word to you is "send forth my people, that it might serve me"'.

I shall pass briefly over some instances of allegedly problematic future indicatives which are less persuasive than those of Num 9,18-23. Voitila identifies additional cases of futures rendering Hebrew imperfects against the sense of their contexts, namely Num 21,27 e)rou=sin for wrm)y, Deut 2,20 logisqh/setai for w)rqy, Deut 2,11 logisqh/sontai for wb#$xy, and Deut 3,13 logisqh/setai for )rqy. None of them can in my opinion be proved. He states, for instance, that the 'correct translation' of wrm)y in Num 21,27 would have been the present indicative, as in Num 21,14 le/getai for rm)y24. Yet the Greek translator's choice of a future indicative can hardly be said to be nonsensical within its own context here or in the other examples. The future indicative has a jussive quality in such cases25.

Num 16,5: kai_ e)la/lhsen pro_j Ko/re kai_ pro_j pa=san au)tou= th_n sunagwgh_n le/gwn 'Epe/skeptai kai_ e!gnw o( qeo_j tou_j o!ntaj au)tou= kai_ tou_j a(gi/ouj kai_ proshga/geto pro_j e(auto/n, kai_ ou$j e)cele/cato e(autw=| proshga/geto pro_j e(auto/n

 wtd(-lk-l)w xrq-l) rbdyw
 wl-r#$)-t) hwhy (dyw rqb rm)l
byrqhw #$wdqh-t)w
wyl) byrqy wb-rxby r#$) t)w wyl)

Let us lastly consider Num 16,5, where the translation of tenses is complicated by the translator's difficulties with the Hebrew text. The noun rqb is understood as a verb in the perfect and is rendered by the Greek perfect e)pe/skeptai. The jussive (dyw is rendered by the aorist indicative e!gnw. The following consecutive perfect byrqhw and imperfects rxby and byrqy are rendered by the aorist indicatives proshga/geto, e)cele/cato, and proshga/geto respectively.

Voitila's argument that this sequence demonstrates the translator understood the sentence as narrative and did not know the future referring context of the next verses is unacceptable. The Greek perfect indicative is a primary tense — in my opinion essentially a stative present26 — so here e)pe/skeptai means 'observes'. The aorist e!gnw means 'has perceived', i.e. 'knows'27. In other words the time-frame is present. The following aorist indicatives are then chosen against the influence of the specific Hebrew verbal forms which they render and according to the translator's interpretation of the context. The translator has taken God's selection of particular individuals to be represented by Moses as a fait accompli. The ensuing punishment of Kore and his supporters merely demonstrates the truth of Moses' statement in v. 5. A possible motivation for this interpretation may in fact have been v. 28 e)n tou/tw| gnw/sesqe o#ti ku/rioj a)pe/steile/n me for ynxl#$ hwhy-yk Nw(dt t)zb. Thus, the difficulty of this passage appears to lie in the translator's misinterpretation of rqb as a verb. But it is precisely the context (including the following context), not specific Hebrew forms, which shapes the translation. The Greek version of v. 5 makes good sense in its own terms if we render 'and he talked to Kore and to his whole group, saying "God observes and knows those who are his and who are holy, and he has drawn them to himself, and those whom he has chosen for himself he has drawn to himself"'.

The preceding treatment raises a variety of objections to Voitila's various identifications of confused renderings. In a few cases our disagreements arise inevitably from conflicting ideas on the grammatical semantics of both the Greek and Hebrew verbal systems. Most often, however, the different interpretations revolve around evaluation of the Greek contexts. In my view Voitila's response to the Greek version is too heavily dictated by the sense of the Hebrew original28. My general argument is that the thirteen passages discussed cannot be proved to contain confusions in verbal renderings, since they are effectively integrated within their Greek contexts. Voitila has addressed the problem of proof in relation to possible confusion between present and future time-frames 29. It can be a difficulty of similar proportions for trying to identify muddles between past and future and also between present and past.

 

III. Causes of Confusion in Numbers 9 and 10 

We may now turn to the question of why the translator became confused in rendering the Num 9,16-23 and 10,11-25 passages, the oddities of which were introduced in §I above. For Voitila, as we have seen, these sequences are representative examples of a widespread phenomenon among the Pentateuchal translators regarding verbal renderings. He applies the theory of short segment translation as developed by Soisalon-Soininen, according to which LXX translators worked on only short units of Hebrew text at a time, without reference to the surrounding context. Anacoluthic or quasi-anacoluthic phenomena of various sorts (such as apodotic kai/) have previously been cited as proof. They allegedly 'show that the translators were seldom conscious of the following context, which had not yet been translated, and were better informed on the part of the text they had just translated'30. Voitila, the first writer to seek evidence in the translation of tense forms, suggests the translators 'were not always aware of the previous context either'31. Yet, whatever one's general response to Soisalon-Soininen's arguments32, I hope to have shown already that the Numbers 9 and 10 passages are isolated instances of confusion in verbal renderings. They thus lend weaker support to the idea of segmentation.

How, then, are these passages to be explained? Given that all the Pentateuchal translators usually chose effectively in rendering Hebrew verbal forms, I suggest that we are dealing with cases of temporary aberration. A possible cause for this lies in the broader context of the two passages. In both places we find repeated switches between narrative and direct speech. These may well have caused the confusions.

Numbers 9 displays a past narrative framework within which are set God's commands to Moses in vv. 2-3.10-14, and the direct speech exchange between certain Israelites and Moses in vv. 7-8. It seems plausible that the translator has slipped in vv. 18-23 by mistaking the narrative there for further direct instruction, taking the lines to be spoken by God. The thread of the narrative is then recovered in v. 23, though perhaps more by accident than design, since the aorist e)fula/canto translates a Hebrew perfect, as we saw in §I.

The structure of Numbers 10 is roughly similar, with detailed instructions from God to Moses in vv. 2-10, followed by a narrative account of the departure from Sinai, which runs down to v. 28. The last eight verses of the chapter show interchange of narrative and direct speech. Once again, I suggest that the translator, lulled by the consecutive perfects in the Hebrew, has slipped back into the construction of divine instructions in vv. 17-25. In fact the confusion in this second passage is even worse than Voitila has indicated. In the midst of his sequence of futures there we find the aorist e)ch=ran translating the consecutive perfect (snw in v. 18, following e)carou=sin for w(snw in v. 17 and preceding further instances of future indicative for the consecutive perfect of the same verb in vv. 21.22.25.

So I agree with Voitila that in these two places the translator has lost awareness of the general context and has produced confused Greek renderings. The value of his contribution in identifying them needs to be stressed. It enables us to nuance the argument put earlier by Barr, sharpening our understanding of translators' practices regarding verbal forms in the LXX. On the basis of Voitila's identifications it would not surprise to discover further instances of such confusion in the work of one or another of the Pentateuchal translators.

On the other hand, it has been argued in the present study that none of Voitila's other examples can be proved (his strongest case seems to me to relate to the Exod 3,17 and 4,23 instances). Though it is accepted here that they manifest influence on the translators from Hebrew text components — including specific Hebrew verbal forms — the resulting translations need not be seen as 'confused'. They provide sensible enough renderings when assessed on their own terms. The Num 9 and 10 cases thus emerge as isolated slips in the work of (presumably) an individual translator. I suggest that they manifest temporary confusion rather than the effects of a regular habit of segmentation in translating. Indeed, it seems to me that intrusive thoughts of the broader context are likely to have caused the confusion, rather than that the translator has 'forgotten' that context. These passages are of course far from the only places where such switches between narrative and direct speech occur, but that should not weaken the case. We are simply observing at Num 9,16-23 and 10,11-25 a particular translator losing the thread of the context for relatively brief spaces. What the phenomenon suggests is a lack of revision within the translation process, at least in these instances. After going astray, the translator has not corrected the mistakes, and they have remained to perplex us today.

A point of broader significance is brought out by this study. Analysis of LXX language involves numerous delicate problems. Our responses to them will be dictated by the guiding assumptions with which we approach these enigmatic texts. In the matter of treating verbal renderings one's particular approach to the volatile debate on both the Greek and Hebrew verbal systems allows considerable scope for divergent interpretations. The key difference, however, between Voitila's response and my own to 'confusions' in translation of indicative tense-forms is our interpretation of Greek contexts. In judging probabilities Voitila places greater weight on the underlying Hebrew contexts, I rather less. This difference reflects contrasting concepts of the degree to which the Greek Pentateuch's original function was dependent upon its Vorlage (cf. n. 28 above). How one views that function will establish which of our interpretations is most satisfying.

NOTES

* It is a pleasure to express my thanks to Dr John Lee, Tamara Neal, and the editorial board of Biblica for helpful suggestions, and especially to Dr Anssi Voitila, both for his research and discoveries which stimulated the ideas presented in this paper and for generous discussion of our different responses to relevant data.
1 See A. VOITILA, "What the Translation of Tenses Tells about the Septuagint Translators", SJOT 10 (1996) 186, 188, 193-194; ID., "The Translator of the Greek Numbers", IX Congress of the International Organization for Septuagint and Cognate Studies, Cambridge 1995 (ed. B.A. TAYLOR) (Atlanta, GA 1997) 111-113.
2 On the opinion reflected here, that each of the Pentateuchal books was the work of a separate translator, see e.g. J.W. WEVERS, "The Göttingen Pentateuch: Some Post-partem Reflections", VII Congress of the International Organization for Septuagint and Cognate Studies, Leuven 1989 (ed. C.E. COX) (Atlanta, GA 1991) 57-59.
3 This is inevitably a controversial statement with regard both to Hebrew and Greek. The current theoretical debate on each of these verbal systems is volatile. The approach adopted in the present study is that the Biblical Hebrew system is based on aspectual distinctions, with morphologically encoded temporal reference at most only nascent. For a recent discussion see P.G. GENTRY, "The System of the Finite Verb in Classical Biblical Hebrew", Hebrew Studies 39 (1998) 7-39. In relation to the Greek verb, Porter's denial that temporal values are encoded even in the indicative mood is not accepted here. The most elaborate statement of his theory appears in S.E. PORTER, Verbal Aspect in the Greek of the New Testament, with Reference to Tense and Mood (New York 1989); cf. T.V. EVANS, Verbal Syntax in the Greek Pentateuch: Natural Greek Usage and Hebrew Interference (Oxford 2001) 40-50.
4 Throughout this paper the terms perfect, consecutive imperfect, imperfect, and consecutive perfect will be employed for the Hebrew indicative forms. Voitila uses qatal, wayyiqtol, yiqtol, and weqatal respectively.
5 The statistics are treated in EVANS, Verbal Syntax, 121-122.
6 On these cf. VOITILA, "Translation of Tenses", 189-190; ID., "Translator of Numbers", 117-118, n. 11.
7 See Voitila's discussion of the passages as cited in n. 1 above.
8 J. BARR, "Translators' Handling of Verb Tense in Semantically Ambiguous Contexts", VI Congress of the International Organization for Septuagint and Cognate Studies, Jerusalem 1986 (ed. C.E. COX) (Atlanta, GA 1987) 381-403.
9 Omitted as not strictly relevant are cases discussed by Voitila in which he accepts translators' sensitivity to the broader context: Gen 47,25 eu#romen for )cmn, Exod 1,18 e)poih/sate for Nty#&(, Exod 4,11 e)poi/hsen for Mw#&y, Exod 9,15 pata/cw for K)w and e)ktribh/sh| for dxktwv, Exod 16,21 dieqe/rmainen for Mxw and e)th/keto for smnw, and the string of imperfects for consecutive perfects in Num 11,8. The example of h]san for wyh in Gen 46,32 will be passed over for a different reason. Voitila took this to be an instance of confusion in VOITILA, "Translation of Tenses", 185, but has modified his opinion so as to remove it from the debate in ID., Présent et imparfait de l'indicatif dans le Pentateuque grec. Une étude sur la syntaxe de traduction (Helsinki – Göttingen 2001) 14.
10 This traditional characterization misses the full complexities of the Hebrew statives. For a recent treatment, applying modern theories of aspect and Vendlerian classification, see F.W. DOBBS-ALLSOPP, "Biblical Hebrew Statives and Situation Aspect", JSS 45 (2000) 21-53, especially 35-40, 42-43 for bh).
11 VOITILA, "Translation of Tenses", 184.
12 B.M. FANNING, Verbal Aspect in New Testament Greek (Oxford 1990) 137-138. Fanning, however, classifies a)gapw= itself slightly differently, as an "activity", not a "state" (ibid., 145). The issue of abstract lexical classification of verbs, highly problematic, cannot be pursued here.
13 For the meaning see J. CHADWICK, Lexicographica Graeca. Contributions to the Lexicography of Ancient Greek (Oxford 1996) 32-34.
14 Contrast VOITILA, "Translation of Tenses", 185, n. 5.
15 Ibid., 186-187. One point should be clarified. In stating that the Hebrew participle's "most usual equivalent" is the Greek present indicative Voitila has in mind only those examples where it is rendered by a finite verb.
16 VOITILA, "Translation of Tenses", 187; ID., "Translator of Numbers", 113.
17 Cf. EVANS, Verbal Syntax, 195-196.
18 Incidentally, it is suggested in the apparatus criticus to BHS that the MT is corrupt here, on the evidence of the Greek and other later versions. This seems unlikely. The MT is surely less problematic than the Greek.
19 VOITILA, "Translation of Tenses", 187; cf. ID., "Translator of Numbers", 113.
20 J.W. WEVERS, Notes on the Greek Text of Exodus (SCS 30; Atlanta, GA) 53.
21 VOITILA, "Translation of Tenses", 191.
22 Ibid., 190.
23 Cf. WEVERS, Notes on Exodus, 36 (on the 3,17 instance).
24 VOITILA, "Translation of Tenses", 192.
25 H.W. SMYTH, Greek Grammar. Revised by Gordon M. Messing (Cambridge, MA 1956) §1917.
26 See EVANS, Verbal Syntax, 26-32, 147-166.
27 Cf. LSJ, s.v. gignw/skw.
28 This point introduces the fundamental question of the original purpose of the Greek Pentateuch. Was it intended primarily as a translation, i.e. as a separate entity in its own right (this is my view), or as an exegetical resource? For discussion see J.A.L. LEE, Review of J.W. WEVERS, Notes on the Greek Text of Deuteronomy, JSS 45 (2000) 178-179.
29 VOITILA, "Translation of Tenses", 184.
30 Ibid., 186.
31 Ibid. Voitila does not mean to imply that the translators lacked a reading knowledge of the Hebrew text, which seems inherently unlikely for the Torah, but that they did not have a "word for word" appreciation of the following context, and tended to forget what preceded (private communication).
32 For the work of Soisalon-Soininen and the Helsinki School on segmentation as a criterion of translation technique see the references collected ibid., 185, n. 6. In positing especially heavy dependence of the translators on Hebrew text components their approach has similarities with Pietersma's new "interlinear" theory of LXX translation; cf. A. PIETERSMA, The Psalms (A New English Translation of the Septuagint and the Other Greek Translations Traditionally Included under That Title) (New York – Oxford 2000) ix-x.