NJPSV versus ESV: which translation adheres more strictly to MT?
In the Preface to the 1985 JPS Edition, we read, “Like the translation of The Torah, the present translation of the prophetic books adheres strictly to the traditional Hebrew text.” The ESV also trumpets its adherence to the traditional text: “The ESV is based on the Masoretic text of the Hebrew Bible as found in Biblia Hebraica Stuttgartensia (2nd ed., 1983) . . . The currently renewed respect among Old Testament scholars for the Masoretic text is reflected in the ESV’s attempt, wherever possible, to translate difficult Hebrew passages as they stand in the Masoretic text rather than resorting to emendations or to finding an alternative reading in the ancient versions.”
ESV, analysis shows, makes a defensible claim. The Preface to the 1985 JPS Edition does not. Ill-fitting and difficult-to-defend expressions in the traditional Hebrew text (hereafter MT) are reproduced in ESV, not emended. NJPSV, on the other hand, departs from MT fairly often in such instances – and others as well - without alerting readers to the fact.
My sympathy lies with NJPSV. To put it more precisely, ESV’s definition of “wherever possible” is too broad. The result: ESV translates too many improbable expressions found in MT which are, on the basis of sound text-critical principles, subject to correction. Other recent translations based on an emended text in places where MT sees to have suffered wear and tear include REB, NAB, NJB, NRSV, and NET.
But – and it’s a big but: it would have been better if NJPSV footnoted every instance in which it departs from MT.
Is it really true that NJPSV departs from the traditional Hebrew text without alerting readers to the fact? Is it really true that it does so more often than ESV does? It is. Let me prove it. Isaiah 38:9-20 may serve as a test case.
The results of my analysis are as follows. Across Isa 38:9-20, NJPSV silently emends – or at least owes it readers an explanation as to why it fails to ‘adhere strictly to the traditional Hebrew text’ - in 15 cases; ESV, in 6 cases. For the purposes of this post, I concentrate on cases in which a sequence of consonants or a vowelling of the consonants at odds with MT is presupposed. I ignore instances in which one or the other translation transposes entire phrases from place to another in the text, revises elements of grammar and syntax of the source text in non-standard and quite unnecessary ways, or ignores the phrasing of the text implied by the masoretic accents. If those instances are factored in, the count is more lopsided in favor of ESV as the more strictly MT-adherent translation.
Silent Emendation #1
Isaiah 38:9
9 *מִכְתָּב* לְחִזְקִיָּהוּ מֶלֶךְ־יְהוּדָה בַּחֲלֹתוֹ וַיְחִי מֵחָלְיוֹ
NJPSV
9 *A poem* by King Hezekiah of Judah when he recovered from the illness he had suffered.
ESV
9 A writing of Hezekiah king of Judah, after he had been sick and had recovered from his sickness.
MT מכתב, a term used to refer to an ‘inscription’ (Ex 32:16; 39:30; Deut 10:4), a rescript = ‘edict’ (2 Chr 35:4; 36:22 = Ezr 1:1), or a rescript = ‘letter’ (2 Chr 21:12), has often been emended toמכתם , a term thought by some to mean ‘song,’ ‘canticle,’ or ‘poem.’ מכתם occurs six times within superscriptions similar to Isaiah 38:9 (Ps 16:1; 56:1; 57:1; 58:1; 59:1; 60:1). LXX translates στλογραφία ‘inscription on a stela’ in these instances; the Targum, גליפא תריצא ‘upright sculpture.’ מכתם may indeed be a genre label, originally, for something written on a raised stela; in the cases at hand, a prayer or psalm. An unrelated genre label in Akkadian is similar: narû literature; Sumerian na-rú-a means ‘erected stone.
To be sure, one must be careful about genre labels. They do not always tell us much about what goes under that label. For example, the etymology of 'epigram' and its equivalent in ancient Greek cannot be taken to imply that the distinctive feature of the item so labeled is that it was ‘written down.’
It is possible that מכתם and מכתב are byforms of a single term. It is perhaps more probable that the obscure מכתם was replaced by מכתב in Isa 38:9 in the course of textual transmission. OG Isaiah προσευχὴ is more readily explainable on the hypothesis that מכתם stood in its Vorlage. If מכתב stood in its Vorlage, it is not clear why γράμμα or the like was not chosen as the equivalent in translation.
Translations which do not render מכתב as they otherwise do may be presumed to emend to מכתם based on a widely adopted conjecture that goes back to Franz Delitzsch. The CTAT committee judged Delitzsch’s conjecture to be very probably correct. It did not adopt the conjecture, because it is just that: a conjecture - it is not supported by primary evidence.[1] Such resolute rejection of conjectural emendation is understandable, but nonetheless exaggerated. CTAT remarks that TOB (‘poème’) duly notes that the translation is based on a conjectural emendation, whereas RL (‘Lied’) and NEB (‘poem’) do not. Other silently emending translations: NAB ‘song’; NJB ‘canticle’ (in a French edition of the Jerusalem Bible, the conjecture on which ‘cantique’ is based is duly noted); NET ‘prayer’ ( = OG προσευχὴ); HCSB and NJPSV ‘poem.’ It might be claimed that ‘poem’ or ‘prayer’ is simply a context-sensitive translation of מכתב in NJPSV, HCSB, and NET, respectively. Perhaps, but the evidence points the other way. מכתב elsewhere is never used in reference to a ‘poem’ or ‘prayer.’
Translations which strictly adhere to MT: RSV, ESV, NRSV, NASB95, and (T)NIV.
ESV’s Personal Name king of Geographical Name calques the Hebrew word order. In this context, the word order sounds somewhat out of place. ‘After’ instead of the literal ‘when’ (so KJV, NASB95) is a felicitous choice. NJPSV offers a free translation of the whole, which captures the sense of the Hebrew with equal or greater accuracy.
Silent emendation #2
11 אָמַרְתִּי לֹא־אֶרְאֶה *יָהּ יָהּ* בְּאֶרֶץ הַחַיִּים
NJPSV
I thought: I shall never see Yah,
Yah, in the land of the living.
ESV
11 I said, I shall not see *the Lord,
the Lord* in the land of the living.
BHS, whose textual apparatus has long been the bane of biblical scholars, says:
prb l c 2 Mss σ'S יְהוָה cf G
Unpacked: Probably read, with 2 manuscripts, Symmachus, and Syriac, יְהוָה. Compare the Septuagint.
Despite BHS, a perusal of HUBP suggests that there are no manuscripts that attest to the supposed reading. The Septuagint, which reads τὸ σωτήριον τοῦ θεοῦ ‘the salvation of God,’ implies on the contrary that it also had יה יה in its Vorlage (cf. HUBP). BHS fails to note that 1QIsaa reads יה once only (by haplography of one יה), whereas Aquila, Theodotion, and the Vulgate indubitably reflect the double יה preserved in MT.
The text as it stands in MT is a nice example of staircase parallelism - abc1 // c2de, where the two c’s are identical – a phenomenon discussed by Wilfred Watson.[2] MT preserves the more difficult reading. It should not be emended away. In particular, it should not be emended away without alerting readers to the fact.
Silently emending translations, in accordance with BHS: RSV (‘the Lord’ once), NRSV (ditto), REB (ditto), NAB (ditto), and NJB (‘Yahweh’).
Is there anything to be said for ESV’s decision to translate יה with ‘the Lord’? At least tradition is on its side (KJV ‘the Lord, even the Lord’). Other “two Lord” translations: NASB95, HCSB, and (T)NIV; However, ESV would have done better to offer a footnote in this case - as it does at Isa 12:2 - Hebrew Yah, // Yah.
Silent emendation #3
11 לֹא אַבִּיט אָדָם עוֹד עִם־יוֹשְׁבֵי *חָדֶלׂ*
NJPSV
Or ever behold men again
Among those who inhabit *the earth.*
ESV
I shall look on man no more
Among the inhabitants of *the world.*
חדל is a hapax legomenon. Nor is a noun of this root, with the meaning of ‘cease’ as a verb, attested in cognate languages. Mitchell Dahood, who stuck to the consonantal text at all costs, thought it might be a term of the underworld: ‘(land of) cessation.’ But, as Michael Barré has shown, one expects a synonymous parallel to ‘in the land of the living’ here.[3]
For that reason, and in light of an idiom attested in Ps 49:2, it is best to assume that the common mechanical error of metathesis lies behind חדל. The assumption of a mechanical error, however, still amounts to an emendation. Yefet ben Eli, Saadya, and many other medieval Jewish exegetes took חדל to mean the same thing as חלד. Barthélemy notes that Judah ben Qoreish and Abraham ha-Babli, followed by others, pointed out the letter permutation. Barthélemy thinks of חלד and חדל as byforms, like שִׂמְלָה and שַׂלְמָה ‘clothing.’ However, the sequence חדל is “taken” in Hebrew (it is used as a verb meaning ‘to cease’), so the suggestion seems improbable. It is more likely, as Barré points out, that חדל is a ghost word of the kind textual critics are used to running into on rare occasions.
Aquila took חדלwith the following דורי and translated ἐπαύσατο ‘ceased.’ Theodotion, the Vulgate, and the Targum (which has a double reading) attest to חדל as well. Cf. LXX, which translates חדל דורי, it would appear, with κατέλιπον τὸ λοιπὸν τῆς ζωῆς μου ‘I have left behind the rest of my life.’
The manuscript evidence almost universally attests to חדל, not חלד. A very interesting and often overlooked biblical manuscript, Codex Babylonicus Petropolitanus, constitutes an exception: it has חלד, but its Masorah Qetanna states, חדל ק̇ ‘read חדל.’ חלד in P most probably represents a copyist’s error for חדל. Nevertheless, חדל before that was probably a copyist’s error for חלד. חדל represents a copyist’s error for חלד in some mss at Ps 49:2. חלד / חדל confusion is also evident in the manuscript tradition at Ps 89:48, though it is less clear in that instance which reading is to be preferred.
חלד means duration, eon, world ( = αἰών, not κόσμος) (cf. BDB).
Silently emending translations: besides NJPSV and ESV: RSV, NRSV, NJB (in a French edition of the Jerusalem Bible, the conjecture on which ‘monde’ is based is duly noted), NAB, NASB95, and TNIV.
Two exceptions: REB, which footnotes as follows: ‘world: so some MSS; others . . . underworld’; and NIV, which footnotes as follows: “A few Hebrew manuscripts; most Hebrew manuscripts in the place of cessation.” It is unfortunate that TNIV omits the footnote.
HCSB translates חדל as ‘what is passing away.’ HCSB is to be commended for its reluctance to emend. But a description of this world as that which ‘is passing away’ is out of place here. With Barthélemy, it might be claimed that חדל is a byform of the more common חלד. But that explanation cannot be taken for granted, so if that is the reasoning behind the translation offered, it ought to be noted.
Silent emendation #4
12 כְּאֹהֶל *רֹעִי*
NJPSV
Like a tent of *shepherds*
ESV
like a *shepherd*’s tent
כְּאֹהֶל רֹעִי, as vocalized, would be ‘like the tent of my shepherd,’ which does not fit the context. Two of the more plausible solutions to this textual crux are the following:
(1) Assume that רעי is a mistake for רעים ‘shepherds.’ This is the route taken by a Targum variant, Symmachus (as corrected by Ziegler), the Vulgate, the Syriac, the Arabic, and NJPSV.
(2) Assume that רֹעִי is a mistake for an oddly spelled רֹעֵי ‘shepherd,’ or in any case equivalent to רֹעֶה in standard biblical Hebrew. Targum, Symmachus (uncorrected), and most recent translations translate accordingly, including ESV.
In both cases, an emendation is presupposed. To be sure, Barré suggests that רֹעִי so pronounced may have meant ‘shepherd,’ but in that case, one must assume that רעי ‘my shepherd’ was concurrently pronounced rōʿiya or the like. That may have been the case at one time, but probably not as late as the 8th-7th cent. bce, from which this prayer and its preserved orthography, at least in part, might derive.
Silently emending translations: besides NJPSV and ESV: RSV, NRSV, REB, NAB, NJB (in a French edition of the Jerusalem Bible, the fact that its translation with a plural is based on ancient versions is duly noted), NASB95, HCSB, and (T)NIV.
Silent emendations #5 and #6
12 *קִפַּדְתִּי כָּאֹרֵג* חַיַּי
NJPSV
My life *is rolled up like a web*
ESV
Like a weaver I have rolled up my life
MT has קִפַּדְתִּי כָאֹרֵג ‘I drew together / bundled up like a weaver.’ Two of the more plausible solutions to this textual crux are the following:
(1) Vocalize: קֻפַּדְתִּי כְּאֶרֶג‘I was drawn together / bundled up like a woven cloth’ and understand the following word, חַיַּי ‘my life,’ as a double subject. This is the route taken by Symmachus, Theodotion, Vulgate, Syriac, the Targum, and NJPSV. As Barré points out, the double subject construction this interpretation implies is discussed in an important article by W. Randall Garr.[4] Parade examples include Ps 57:5; 73:21.
(2) Vocalize קֻפַּדְתִּי ‘I was bundled up / packed to go’ and associate (against MT) with the preceding two words ‘like a shepherd’s tent.’ The double subject construction would no longer obtain; instead, the subject in the main clause is picked up by an equivalent object in an embedded unmarked relative clause. Vocalize כְּאֶרֶג ‘like a woven cloth,’ and treat it and חַיַּי ‘my life’ as a stich unto itself.
LXX also seems to have taken קפדתי with the preceding, but its construal of the whole is problematic. 1QIsaa reads ספרתי, perhaps סֻפַּרְתִּי ‘I was cut off’ (a meaning of D ספר in post-biblical Hebrew). Curiously, KJV translated as if it read סִפַּרְתִּי: ‘I cut off.’
REB and NAB, without saying so, emend to קִפַּדְתָּ חַיַּי כָאֹרֵג ‘you have rolled up my life, like a weaver . . .’ REB and NAB phrase חַיַּי with קִפַּדְתָּ and כָאֹרֵג with the following stich. But כָאֹרֵג can govern the words following, ‘from the thrum he will cut me off,’ if and only if כָאֹרֵג and חַיַּי are transposed. Barré so emends.
MT continues on as follows: מִדַּלָּה יְבַצְּעֵנִי ‘from the thrum he will cut me off.’ God might seem to be the subject, as often in 38:12-13. But the verb’s subject is not God in LXX, Theodotion, and the Syriac. The verb also lacks an object in these versions. The omission of ‘me’ is understandable as an accommodation to the style of the target language.
NJPSV, REB, and NAB follow the ancient versions just listed in not construing God as the subject and in the omission of ‘me’ as the verb’s object. NJPSV ‘and cut from the thrum’ passivizes the construction in line with its handling of the preceding. It does not necessarily depend on an emendation in so doing. REB translates ‘like a weaver when he cuts the web from the thrum’; similarly, NAB. as previously noted, these translations depend on a transposition of two words.
Silently emending translations: NJPSV (twice); REB and NAB (twice).
Besides ESV, RSV, NRSV, NJB, HCSB, NASB95, and (T)NIV do not emend. They translate ‘like a weaver I have rolled up my life’ with variations regarding punctuation and word order.
However, it is not clear how ‘like a weaver I have rolled up my life’ is appropriate to the context. KJV ‘I have cut off like a weaver my life’ fits the context somewhat better, but is not justifiable without emendation from a philological point of view.
Silent emendation #7
13 *שִׁוִּיתִי* עַד־בֹּקֶר
NJPSV
*I cried* out until morning
ESV
*I calmed myself* until morning
The meaning שִׁוִּיתִי would have here is not clear. KJV: ‘I reckoned.’ HCSB: ‘I thought.’ NASB95: ‘I composed my soul.’ (T)NIV: ‘I waited patiently.’ None of these attempts at making sense out of the Hebrew is particularly convincing. If you know that God is about to finish you off and break all your bones like a lion, you do not calm yourself or wait patiently.
NJPSV describes this and much of 38:12-13 as of “uncertain” meaning. That may be true, but it also does not say enough. In effect, NJPSV silently emends MT. Its translation, ‘I cried out,’ presupposes an oft-proposed emendation: שִׁוַּעְתִּי. The Targum, as ESV remarks in a footnote – I cried for help, based on the Targum, is its alternate reading – seems to have read likewise. RSV noted that its ‘I cry for help’ is based on a correction of MT. NSRV is also explicit: ‘I cry for help,’ with the following footnote: “Cn: Meaning of Heb uncertain.” NAB and NJB translate likewise, but without an explanatory footnote. REB renders ‘I am racked with pain.’ A footnote makes it clear that a Scroll (1QIsaa, understood to read שפותי, vocalized שַׂפּוֹתִי) is being followed.
Silently emending translations: NJPSV, NAB, and NJB (in a French edition of the Jerusalem Bible, the conjecture on which the corresponding translation is based is duly noted).
Silent emendation #8
14 כְּ*סוּס עָגוּר* כֵּן אֲצַפְצֵף
NJPSV
I piped like *a swift or a swallow*
ESV
Like *a swallow or a crane* I chirp
As Barré points out, a uniquely important piece of evidence with respect to סוּס עָגוּר is contained in the Deir ʿAlla texts:
כי.ססעגר.חרפת.נשר
For the ססעגר reproaches the griffon vulture.
This text makes it certain that a single ornithological species is in view. Note the word dividers. Barré builds a careful argument for the conclusion that the swallow is the referent (so, already, LXX). NAB and NJB render accordingly. REB translates ‘swallow’ also, but does so, according to a footnote, by omitting עגור which it thinks refers to ‘a wryneck.’
NJPSV and ESV, on the other hand, silently emend Isa 38:14 to accord with Jer 8:7 וסוס ועגור. That made sense at the time perhaps, but the assimilation to Jer 8:7 should have been duly noted. It is now clear that Jer 8:7 needs to be emended to accord with Isa 38:14 and the Deir ʿAlla text.
Silently emending translations: besides NJPSV and ESV: RSV (a footnote says “Heb uncertain”) NRSV (a footnote says “Meaning of Heb uncertain”) and (T)NIV. This is one of many examples in which the modesty which characterizes RSV / NRSV is missing in NJPSV and ESV. HCSB emends likewise, but places ‘or’ between brackets, and thus alerts the reader to what it is doing. NASB95 reads ‘Like a swallow, like a crane, so I twitter.’ So far as I know, however, ellipsis of the kinds these translations assume is not otherwise attested.
To be continued.
[1] Dominique Barthélemy, ed., Critique textuelle de l’Ancien Testament 2: Isaïe, Jérémie, Lamentations (OBO 50/2; Fribourg / Göttingen: Éditiones Universitaires / Vandenhoeck & Ruprecht, 1986) 263-64.
[2] Wilfred G. E. Watson, Classical Hebrew Poetry: A Guide to its Techniques (2d ed.; JSOTSup 26, Sheffield: Sheffield Academic Press, 1995 [1st ed. 1984]; corr. repr. with suppl. bibliog.; London: T & T Clark, 2005) 150-156.
[3] Michael L. Barré, The Lord Has Saved Me: A Study of the Psalm of Hezekiah (Isaiah 38:9-20) (CBQMS 39; Washington: Catholic Biblical Society of America, 2005).
[4] W. Randall Garr, “The Grammar and Interpretation of Exodus 6:3,” JBL 111 (1992) 385-408.
John, your analysis requires extraordinary credulity.
Example 1: The ESV copies from the RSV word for word. There is no intellectual contribution of the ESV here at all.
Example 2: As you point out, here the NJPS follows the MT, and the ESV does not.
Example 3: The ESV copies from the RSV word for word. There is no intellectual contribution of the ESV here at all. (I am amused by your implication that Bibles are required to replace every instance of a hapax legomenon with a blank or lengthy footnote -- that would certainly make the book of Job uncomfortable reading.)
Example 4: The ESV copies from the RSV word for word. There is no intellectual contribution of the ESV here at all.
Example 5: The ESV copies from the RSV word for word. There is no intellectual contribution of the ESV here at all.
Example 6: The ESV copies from the RSV word for word. There is no intellectual contribution of the ESV here at all.
Example 7: Here is a good example to illustrate my point. The RSV and the NJPS say clearly that the Hebrew text is uncertain (in which case, in the NJPS, always means that they have varied from the original text, a point you completely elide over). So, your claim that the NJPS "slightly emmends" is simply false. However, the ESV magically claims certainty, claiming an alternative meaning from the Targum, which is certainly intellectually dishonest.
Example 8: Another good example to illustrate my point. The ESV copies from the RSV word for word. There is no intellectual contribution of the ESV here at all. However, the ESV omits a footnote in the RSV which explains that this text is uncertain. Your claim that the Deir Alla texts should be translated here rather than the MT strains credulity: the Deir Alla texts were not released in full until 1984 and the NJPS translation of the prophets appeared in 1978. Apparently, you expect time travel of the translators.
In summary, your arguments are unsupported and deceptive.
Posted by: Iyov | April 23, 2008 at 12:39 PM
Iyov,
your insinuation that ESV merely follows RSV word-for-word is not born out by the facts. Of course it does so on many occasions: it is a revision of RSV after all, as the ESV preface clearly and accurately states.
As you have said on your blog, and I encourage readers to take a look, you prefer RSV to ESV because, it would seem, hard-core Calvinists give you the creeps. The fact that some of those who contributed to ESV, particularly on the oversight committee, are about as ecumenical in their approach to cooperative scholarship and editing, and open-minded on questions like the ordination of women, as the people at ArtScroll are, also bothers you. I am touched by your concern for the intramural woes of Calvinist Christianity.
On a number of details you plain misread my comments. There is, however, one matter you get wrong which cannot go unanswered, your claim that:
JPS say[s] clearly that the Hebrew text is uncertain (in which case, in the NJPS, always means that they have varied from the original text, a point you completely elide over).
In the 1985 JPS Preface, the opposite is stated. I quote:
Like the translation of The Torah, the present translation of the prophetic books adheres strictly to the traditional Hebrew text; but where the text remains obscure and alteration provides marked clarification, *a footnote is offered with a rendering of the suggested emendation.* [asterisks mine]. . . . In *all* cases [asterisks mine], the emendation is given in a footnote, which may be disregarded by those who reject it on either scholarly or religious grounds. The *only* [asterisks mine] exceptions involve such changes in grammatical form as those, say, from second person to third or from singular to plural. *In such rare instances, the change is incorporated in the text, and the traditional Hebrew is translated in a footnote.* [asterisks mine]
I rest my case. For anyone who has not read what else I've written about NJPSV, let me repeat: despite its failure to live up to its own preface, I recommend NJPSV over any other English translation of the Hebrew Bible on the market, in particular in The Jewish Study Bible edition.
Posted by: JohnFH | April 23, 2008 at 01:55 PM
Repeated comment from elsewhere:
I note that you did not compare the ESV and the RSV. Why don't you compare those? That is the element of non-novelty.
By and large, the substantive revisions of the ESV over the RSV are to impute New Testament readings into the OT. That may or may not be a valid way of reading the Hebrew Bible (depending on your confessional stance -- I gather for you it is) but it is certainly not a "word for word" translation of the Hebrew Bible, as the ESV literature claims.
You cannot even quote the preface to the Tanakh correctly. Here is what it says (from my Libronix edition -- I note that you said you had this edition.) I have added italics to emphasize:
"In preparing the translation of The Prophets, the translators faced a recurring problem that deserves special mention. The prophetic books contain many passages whose meaning is uncertain. Thus, in order to provide an intelligible rendering, modern scholars have resorted to emending the Hebrew text. Some of these emendations derive from the ancient translators, especially of the Septuagint and the Targums, who had before them a Hebrew text that sometimes differed from today’s traditional text. Where these ancient versions provide no help, some scholars have made conjectural emendations of their own. Many modern English versions contain translations of emended texts, sometimes without citing any departure from the traditional Hebrew text."
That seems to me to be the model of intellectual honesty, unlike your post, which requires that translators be capable of time-travel, takes unrepresentative examples, and fails to not the careful footnoting in the RSV and NJPS, unlike the ESV.
Posted by: Iyov | April 23, 2008 at 02:29 PM
I refer the reader to my reply found on the "More Silent Emendations" thread.
Posted by: JohnFH | April 23, 2008 at 02:44 PM
Just an amateur's curiosity, but what do you think of "when I am among them that have ceased to be" (as found in the 'ol RV's margin "Or,...")?
Posted by: John | April 26, 2008 at 03:16 AM
John,
That's a good question.
The trouble with the "them that have ceased to be" translation, apart from the fact the phrase is nowhere else attested, is that one expects at this place a parallel to "the living," not the dead. Furthermore, the "along with" phrase should modify "human" immediately preceding, not the subject of the sentence.
Posted by: JohnFH | April 26, 2008 at 07:38 AM