God Didn't Say That

Bible Translations and Mistranslations

Top Translation Traps: Relying on Structure

Perhaps the biggest translation mistake I’ve seen is relying too closely on word-internal structure to figure out what words mean. We saw this last week with toldot and in a comment regarding etymology.

I call this the trap “word-internal structure” (even though it applies to phrases, too).


As usual, we can look at modern languages to see how poorly internal structure reveals the meaning of a word.

Two examples from my recent And God Said include “hostile,” which doesn’t mean “like a host,” even though the pattern of “infant” and “infantile” would suggest otherwise; and “patently,” which means “obviously” even though a patent by definition must be non-obvious. We see that even with something so simple as adding “-ly” to a word, we can’t rely on structure to tell us what a word means.


Also from And God Said comes this example about phrases:

A more detailed example highlights the issue. English has a verb “pick” and two words “on” and “up” that can be added to verbs. “Pick” (as in “pick a lock”) means, “open stealthily without a key.” “Up” means “away from gravity” and “on” means “touching and located in the direction of open space.” (All of these definitions are approximate. That isn’t the point here.) This knowledge, however, doesn’t explain why “pick on” means “annoy,” “pick up” means “increase” (as in, “pick up the tempo”), and “pick up on” means “discern.”

This demonstrates the important fact that phrases, like words, don’t always get their meanings from their parts. (Another favorite example is “drive-through window.”)


We’ve already seen one clear case where internal structure leads us astray. The internal structure of the Hebrew word toldot suggests that it specifically has to do with “birth,” or maybe “generations” or “descendants.” But we saw that it does not.

Another example comes from the Hebrew phrase “spy after” in Numbers 15:39. The verb there is tur, which means “spy” or “explore.” And the preposition is acharei, “after.” But — just as with “pick up” and “pick on” — it’s a mistake to assume that we can understand the phrase just by knowing its parts. In this case, the phrase occurs nowhere else, so we’re stuck with a problem. The full sentence — important enough in Judaism to be included in the m’zuzah that adorns doorways and the t’fillin that serve as ritual prayer objects — is this: “this will be your tassel. When you see them, you will remember all of Adonai’s commandments and do them. Do not ??? your heart and your eyes, after which you lust.”

(Two notes are in order: “heart” is misleading here, as is “lust.” Also, t’fillin enjoys the utterly useless English translation “phylacteries.”)

Translations for the literal “spy after” include “follow after” (ESV), which I don’t think is even an expression in English; “[go] wantonly astray after” (NAB); “going after the lusts of” (NIV); and “follow” (NRSV). Except for the NRSV, all of these translations (wrongly, in my opinion) insist on putting the word “after” in the translation. (The LXX gives us diastrafisesthe opiso, while the Vulgate has the single word sequantur, from sequor, “to follow.”)

Hebrew word-internal structure is complicated, and — depending on personal constitution — either immensely enjoyable or the ultimate barrier to learning Hebrew. Either way, it’s hard to ignore Hebrew’s rich word-internal structure, but sometimes translation demands that we do.

By way of further example, we can consider the Modern Hebrew word m’sukan. It is the passive of the active m’saken. The active means “endanger.” So word-internal structure points us to “endangered” for a translation of the passive. But that’s wrong. The word means “endangering.” In other words, the passive means almost the same thing as the active. “Dangerous” is the usual translation.


When I discussed energeo (responding to discussions by J.R. Daniel Kirk and on BBB — then BBB followed up, as did T.C. Robinson), one comment noted that I “miss[ed] the distinction between the active in Matthew 14:2, Galatians 3:5 etc. and the middle or passive in Galatians 5:6 and James 5:16.” I think we see from the discussion here that, while the active/passive/middle distinction is not to be ignored, neither can we rely on it to tell us what words mean. It’s possible (as we just saw in Modern Hebrew) for a passive form not simply to indicate the passive of what the active form indicates.


It seems to me that two lessons are important.

First, word-internal structure, while sometimes helpful and often fun, is an unreliable way to figure out what a word means.

Secondly, phrases are just like individual words in this regard.

So when we look at a word or a phrase, I think it’s important not just to look at its formal structure.


March 8, 2010 - Posted by | Bible versions, translation theory, Translation Traps | , , , , , , , , , ,


  1. Terrific. A lot to chew on here.

    Comment by David Ker | March 8, 2010

  2. One of the best examples I’ve seen regarding the difficulty of “connecting” the internal structure of phrases with the meaning is from Ron Moe. This is in reference to building dictionaries:

    “‘On the other hand we wore our washing machine out.’

    “The phrase ‘on the other hand’ has the internal structure of a preposition[al] phrase but functions as a conjunction syntactically and semantically. It is a fixed phrase in that it cannot be varied in any way. None of the constituent words can be replaced or inflected and no other words can be inserted in it. The phrase ‘washing machine’ is actually a compound noun. It can be pluralized, but otherwise is fixed. The phrase ‘wear out’ can have other words inserted in it and the verb ‘wear’ can be inflected. Glossing sentences like this are going to be a challenge. Developing dictionary entries for different kinds of phrases will also require insight.”

    Comment by Mike Sangrey | March 8, 2010

  3. I still think “legacy” might work for toldot.

    I love the drive-through window example.

    And with “dangerous/endangered” in Hebrew, I could imagine some comical situations in a zoo based off that misunderstanding.

    Comment by Gary Simmons | March 8, 2010

  4. […] phrase-level issue is pretty close to internal structure, which I discussed last […]

    Pingback by Top Translation Traps: Myopic Translations « God Didn't Say That | March 15, 2010

  5. I subscribe to Merriam-Webster’s “Word A Day” email. As I was reading the word for today (“archetype”) I was struck with the zigzag they provide from what was often a very “true-to-structure” usage to more modern day “not-so-close-to-structure” usage. Here’s the link, if anyone is interested in subscribing:


    Comment by WoundedEgo | March 16, 2010

  6. >>>…Secondly, phrases are just like individual words in this regard…

    Idioms… for example, in Spanish…

    “Hay que tener cuidado”

    would literally be “There is that to have been cared” or something like that. In actuality, it means “Be careful!”

    Thus, I have a saying for those of us who may be wound too tightly about literalness and formal equivalence (two values I prize highly):

    “Hay que tener cuidado”

    Comment by WoundedEgo | March 16, 2010

  7. You use English as ‘proof’ that we cannot know phrases by there parts. But aren’t these two completely different languages? That’s a Non-sequitur argument. You haven’t demonstrated anything. You pick some random phrases from English and somehow this proves that we cant know the whole by its parts in Hebrew. Well, the problem with this is: You would have to know what something means at the start to prove the other method is wrong. You can say with certainty “drive-through-window” doesn’t actually mean I’m about to drive through a window because its known at the start that’s not what it means.But what good does that do us in Hebrew? One of your arguments is that we cannot know what the words mean… So on your premise You cannot demonstrate that the parts do not make sense of the whole because in order for the argument to work you have to know what the words mean in order to demonstrate that the said method is fallacious. So you either know the words or you don’t. And if you don,t, it would be better to just close up shop and go home!!!

    Comment by jamie | June 22, 2010

  8. You cannot (logically) use word meanings to prove that words don’t have meaning. Your whole argument about words not having meaning is built on the assumed premise that what you are saying is being communicated clearly and that we understand what you mean!! Its like the person who says “there is no absolute truth’, yet his statement is an absolute truth claim, so it is logically unsound. His statement disproves his statement.

    Comment by jamie | June 22, 2010

    • Jamie: no languages are completely different, so long as both are used by humans. And if they’re (proper spelling!) not completely different, then they obviously have some things in common. If for no other reason, then simply by virtue of the fact that they are used by people.

      Example: “by virtue of” is a prepositional phrase that uses the archaic sense of virtue as in “power.” On its own, virtue today means roughly the same thing as “integrity.” Greek, actually, has a similar expression but uses charis (grace) rather than arete (virtue).

      Among English, Greek, German, and Hebrew: prepositions and other words strongly affect the meanings of verbs. It’s not that words have no meaning. It’s just that a word’s expected meaning may be suppressed or overwritten completely by the phrase it is found in. Often, though not always, the whole is greater than the sum of its parts.

      Comment by Gary Simmons | June 26, 2010

  9. […] What makes his example work is that the meaning of “pick on” doesn’t come from the meanings of “pick” and “on.” More generally, phrases, like words, are not the sum of their parts. (I have more here.) […]

    Pingback by Always Pick On The Correct Idiom « God Didn't Say That | September 22, 2010

Sorry, the comment form is closed at this time.