An Attempt to Quantify “Literalness” in New Testament Translations
R. Grant Jones, 14 June 2010
(Modified 17 Dec 2010, with the addition of the revised NIV)
In his A User’s Guide to Bible Translations, David Dewey presents a graphic (Table 2.6, page 66) depicting “Bible Versions Organized by Translation Philosophy.” In order of more to less literal, Dewey’s arrangement begins as follows:
1. New American Standard Bible (NASB)
2. New King James Version (NKJV) / King James Version (KJV)
3. English Standard Version (ESV)
4. Revised Standard Version (RSV)
5. New Revised Standard Version (NRSV)
6. New International Version (NIV) / Today’s New International Version (TNIV)
7. New American Bible (NAB)
8. New Jerusalem Bible (NJB)
9. New English Bible (NEB) / Revised English Bible (REB)
10. Jerusalem Bible (JB)
He also includes far less literal translations (e.g., the New Living Translation, the Good News Bible), with the Message as least literal.
At its web site, the Zondervan publishing company provides a chart plotting different translations. To the left are those that employed a “word for word” translation philosophy. To the right are listed more “thought for thought” versions. Arranged from most literal to least, the Zondervan order begins:
1. Interlinear [by which is meant, apparently, any interlinear translation]
3. Amplified (AMP)
8. Holman Christian Standard Bible (HCSB)
Others follow. As in Dewey’s list, Zondervan presents the Message as the most “thought for thought.”
Being familiar with most of these translations, I believed Dewey’s list to be more accurate. But I wondered whether I could produce a methodology to quantify “literalness” - the degree to which a translation is “word for word.” My goal was not simply to rank order translations, but to answer the question, “How much more literal is translation X than translation Y?”
I chose to score the following New Testament translations:
the King James Version (also known as the Authorized Version),
the American Standard Version (ASV) of 1901,
the New American Standard Bible of 1977 (NASB77),
the New American Standard Bible of 1995 (NASB95),
the New King James Version (copyright 1982, but incorporating subsequent changes – effectively, the 2010 manifestation of this version),
the Revised Standard Version of 1971,
the English Standard Version of 2007,
the New Revised Standard Version of 1989,
the New International Version of 1984 (NIV84),
the New International Version to be published in 2011 (NIV11),
the New American Bible 1986,
the New English Bible of 1972,
the Jerusalem Bible of 1968,
and the New Jerusalem Bible of 1985.
For those who just want the bottom line, I present the results first. The ASV has the most literal New Testament translation of those I scored. The NKJV and KJV follow closely. The NASB of 1977 is less literal than those, and the NASB of 1995, with its propensity to drop conjunctions and insert proper names for pronouns, is less literal still. (To this point, all the translations indicate supplied words through oblique or italic type.) Next come the ESV and the RSV. There is a broad gap between the RSV and the next translation, the NRSV. The latter is quite close to the NAB. The two NIV editions stand in a class by themselves1 – they far freer than the NAB, and far more literal than the NJB. The NEB and the JB complete the list.
The following bar chart shows the order from most to least literal, as described above. It also compares each translation numerically to the most literal of the group, the ASV. The numeral above each bar is indicative of the “freeness” of the translation as compared with the ASV. For instance, the NKJV is 1.15 times as free as the ASV, while the JB is 9.3 times as free.
I will explain what I mean by a “liberty” in the following section.
It is more common to compare translations along a single axis, as Zondervan does at the site linked above. My results display as follows:
I chose 200 distinct verses at random from the New Testament. I performed no mathematical analysis to support that number. The choice was subjective, 200 being small enough to allow me to score 13 translations2 within a few months, working an hour or so a few evenings a week. The mechanics were straightforward: using a random number generator, I obtained 200 distinct numbers between 1 and 7956. I then mapped those numbers to New Testament verses. I used the same 200 verses for all 13 translations.
The fourteen translations listed above were chosen for many reasons – popularity, reputation for literalness (or the opposite), and my familiarity with them, among others. Translations were excluded on various grounds. Based on experimental attempts to score the NEB, I determined that my method would be difficult to employ with the least literal versions (e.g., the Good News Bible, the New Living Translation, the Message). I omitted the Holman Christian Standard Bible, the Revised English Bible, and Today’s New International Version because I do not own paper copies of them. Without a paper copy, I had no way to ensure an electronic version available on the Internet was free from error.3 I excluded other versions (e.g., the Modern Language Bible, Goodspeed’s translation) because they are not widely used, and still others (Rheims (1582), Confraternity (1941)) because they were based on the Latin, not the Greek.
My purpose was to quantify “literalness” in New Testament translations, so it was important to grade each translation against its proper Greek text. Consequently, the KJV and the NKJV were compared with the Received Greek (Textus Receptus), while the others were graded against the Nestle-Aland Greek-English New Testament, eighth edition, and the textual variants as shown in the notes. In other words, if a translation reflected any Greek text that could be constructed from any consistent set of readings, no points were scored. In one instance, one of the verses chosen at random was omitted by a translation. In that case, however, the version provided a translation of the verse in the margin, and I scored that.
A “liberty” is any variation from a word-for-word translation of the Greek into syntactically acceptable English. To clarify, if a translation simply arranged the English counterparts of Greek words into a normal English word order, no liberty was scored. Examples of liberties include the omission or insertion of conjunctions, the supply of proper names or the use of a pronoun where a proper noun is given, the insertion of explanatory words or phrases, the resolution of ambiguities, the failure to reflect tense or mood, and the omission of verbs (such as “saying”). All liberties were scored equally and given a value of one point.
Supplied words that appear in oblique or italic text, as in the KJV, ASV, NKJV, and NASB, were not scored as liberties. Present tense verbs translated as past tense but marked by an asterisk, as in the NASB, were not scored as liberties. Footnotes providing alternate or more literal translations were not considered.
For each translation, the point value of liberties for each verse was recorded and summed across the 200 verses scored. The ASV, as the lowest scorer, was chosen as the point of comparison. I divided the cumulative score for each translation by the ASV’s cumulative score.
The left-hand (more literal) side of the “Translation Continuum” is dominated by translations of the Tyndale family, the KJV and its descendants.
The major differences between Dewey’s order and my own involve his placement of the NASB (as more literal than the KJV/NKJV) and the NIV (as more literal than the NAB). It may be that the NASB brings out a subtlety in the Greek to which I am blind and which I consequently failed to score.4 (I did take note of the Greek imperfect tense and score accordingly.) Dewey may have considered the extent to which a translation reflects the order of words in the original language. (In general, I did not.) Or it may be that the NASB and NIV are more literal than their rivals in the Old Testament.
Zondervan’s list places the KJV and NKJV toward the middle of the pack, rather than near the top, and it sets the NIV after the NJB. It may be that Zondervan compared each translation with the eclectic (Nestle-Aland) text in all cases. If so, that would tend to move the KJV and NKJV toward the “less literal” end of the spectrum, but not by very much. It is true that the NJB is more literal than its predecessor, but it inherits too much of the JB’s unconstrained character to rank as more literal than the NIV.
I have no insight into the methods others employed to develop their rankings. My approach has the advantage of assigning a numerical value for the “literalness” of a translation, but it should be kept in mind that my scoring method was arbitrary. If one were to assign different values to the various classes of “liberties,” different scores would certainly result. (For instance, one could score omissions of conjunctions as 0, or 0.5 (or any other number, for that matter), rather than 1 as I did.) It may be that a different ordering of the translations along the continuum would result.
1The NIV11 scored as slightly less literal than the NIV84. The revision is more literal in places (e.g., Luke 17.7, John 8.25), but those are balanced by new liberties that improve readability (e.g., Philippians 2.15) or impose gender neutrality (e.g., 1 John 5.16).
2 The 13 translations originally included in this project excluded the NIV11, which was not available at the time.
3 I made an exception for the NIV11, since no paper copies were yet available.
4 Foundation Publications advertised the NASB update (NASB95) with the phrase, “the most literal is now more readable.” My assessment does not support the claim that the NASB is the most literal modern English translation. Foundation Publications’ website (http://www.foundationpublications.com) makes the following statement: “Originally produced in 1977, the NASB has been widely embraced as ‘the most literally accurate English translation’ from the original languages.” It may be that “accurate” is an allusion to the eclectic Greek text. I concur that the NASB is the most literal English translation from the Nestle-Aland text of those I examined.
How would my scoring of the NASB95 and NASB77 have changed had I considered marginal notes (i.e., not credited the NASB with a liberty if a marginal note provided a literal reading)? The NASB95’s normalized score would drop from 1.96 to 1.74. The NASB77’s score would move from 1.54 to 1.33, slightly above the KJV’s 1.20.
Is the New American Standard the most literal translation?
Is the NASB the most literal translation?
Comparing Bible translations
Which translations are most literal?
Which translations are most accurate?
Is the ESV more literal than the NIV?
Is the 2011 NIV more literal than the old NIV?
Bible comparison charts