(The following is from the paper I read at the annual ETS meeting last week. The entire paper can be downloaded from BillMounce.com/papers.)
Much of the current misunderstanding about translation theory is due to putting Bible translations in the wrong category, or putting two translations together that should be kept separate. I believe there are five, not two or three, categories of translations, and critiques specifically of the NLT ("natural language") should not be applied to the NIV (functional equivalence").
1. "Literal"
The category of "literal" translations should only include interlinears, and in fact I don't like the term "literal" at all since we use the word in a way that is contrary to its actual meaning. The word "literal," in any English dictionary, literally means "without embellishment," and it should never be used in a discussion of translations.
There is no such thing as a literal meaning of a word — what does λόγος "literally" mean? — no such thing as a literal translation of a verse, and therefore there is no such thing as a "literal translation" or even an "essentially literal" translation. Even interlinears are technically not literal but are, to some degree, interpretive. The minute you translate τοῦ θεοῦ as "of God," you are no longer literal but interpreting a genitive noun construction with a prepositional phrase and dropping ὁ, a word that actually has no precise equivalent in English.
No competent translator should say that their translation is literal. The problem is that the folks in our churches mistakenly equate "literal" with "word-for-word" and think that means "accurate." It is our responsibility to help people see the error in this thinking. [See Van Leeuwen on the KJV. "I prefer not to call it 'literal' because translations always add, change, and subtract from the original. The only literal Bible is written in Hebrew and Greek" ("We Really Do Need Another Bible Translation," Christianity Today, 45 no 13 (October 22, 2001).]
2. Formal equivalent
"Formal equivalent" refers to translations that show a strong preference for replicating the form of the Greek and Hebrew, and only move to meaning when translating words doesn't make sense. This category includes the NASB, ESV, CSB [to some extent], KJV, RSV, and NRSV (except for gender language). I also use the term "direct translation" for this category.
The ESV has invented a category called "essentially literal," which only means they are more willing than others to leave the original words and translate meaning. However, when Grudem includes the NASB and RSV in this category, I wonder if there are any translations left for the category of "formal equivalent." I suspect that "essentially literal" is an attempt to abandon the linguistic baggage of formal equivalence. For the purpose of this paper, I place the ESV, along with many other fine translations, in the category of "formal equivalent," acknowledging their commitment to translate the meaning of every word and not just every word.
3. Functional equivalent
"Functional equivalent" refers to the translation process that places primary emphasis on the meaning of each of the original words understood in context. These translations are more willing to move to meaning more quickly than formal equivalent translations, but they still try to honor the structure of the original if possible. This is where the NIV and NET fit.
Some people include the NLT in this category, but that is far from accurate. Whatever terms you use, the NIV and the NLT are fundamentally different and must be kept in separate categories.
4. Natural language
"Natural language" translations are those that follow the teaching of Eugene Nida and his emphasis on the reader's response. For the sake of clarity, and since Nida created the term "dynamic equivalent," I use the term "dynamic" for natural language translations, not functional equivalent translations. This is where the NLT belongs, and much of the criticism of functional equivalent translations really belongs in this category. [The Theory and Practice of Translation, Eugene A. Nida and Charles R. Taber (Leiden: Brill, 1974). From One Language to Another: Functional Equivalence in Bible Translating, Jan de Waard anf Eugene Nida (Nashville: Nelson, 1986).]
5. Paraphrase
I have no term for the fifth category other than perhaps "paraphrase," but even that is an improper title since the word "paraphrase" refers to the simplification of a text in the same language. So the original Living Bible is a true paraphrase of the ASV, and I also put The Message here as well.