📄 The Biblical Hebrew idiom ‘lift the face’ in the Septuagint of Job

Summary

Idioms are hard to translate because their meaning cannot be reliably assembled from their words. Mangum asks how ancient translators dealt with the Biblical Hebrew idiom "lift the face" (nasa panim). Its meaning ranges from proper respect and acceptance to improper favouritism, and context decides which reading applies. He compares how the Septuagint, the Peshitta and the Targums render its occurrences, and he gives special attention to the Greek Job, a book whose translation is known for literary freedom. Drawing on Translation Studies and cognitive linguistics, he sorts each rendering into four strategy types: formal, semantic, blended, and idiomatic. Most versions mostly copy the Hebrew wording and adjust the verb to fit the context, while plain explanation of the meaning is rare. The Job translator stands apart: he prefers natural Greek idiomatic phrasing and uses the idiom's polysemy to shift theological emphasis. The study refines the common view that ancient Bible translators were mechanically literal.

Key points

• The idiom occurs 28 times in the Hebrew Bible. Across the versions, roughly 60% of renderings are word-for-word, and the Targums go even further in that direction.
• The LXX keeps the noun "face" in 80% of its renderings, the Peshitta retains it at a rate of 93%, and the Targums preserve it every time (100%).
• The LXX has two stock equivalents:
  • λαμβάνω τὸ πρόσωπον ("take the face"), which uses the usual verb for nasa;
  • θαυμάζω τὸ πρόσωπον ("honour the face"), which is not lexically equivalent.
• The Aramaic versions mostly use cognate calques, alongside variants meaning "raise" or "brighten the face".
• The strategy typology extends Joosten's Types A/B/C with a new Type D (target-language idiom). LXX Job favours Type D.
• In Job 42:8 the Greek turns God's acceptance of Job's prayer into forgiveness of the friends' sins. It also has the friends speaking against Job rather than about God.

Interesting ideas

• I1. Each rendering of a figurative phrase is classified on a single axis: keep the form, give the meaning, blend both, or substitute a target-language figure. This makes translation choices countable and comparable across versions. — For Mova (back): Generation and translation through the meta-world could tag each idiom output with a strategy type, so the system can explain whether it preserved form, explained meaning, or substituted an equivalent idiom.
• I2. One idiom can carry a shared core sense (favourable disposition) while surrounding cues decide its polarity (proper respect vs improper bias). This separates lexical meaning from contextual evaluation. — For Mova (frames): Idiom frames in prag could store a core relation plus an evaluation slot filled from context, so "lift the face" and English analogues like "play favourites" get explainable polarity.
• I3. Departures from a stock equivalent are treated as evidence that the translator was reading the context. Deviations from the default carry the signal. — For Mova (learn): When witness parsers or teachers depart from their usual output for a construction, Mova could flag those spots as context-sensitive cases worth targeted review or repair rules.
• I4. Retention rate of one key component (the noun "face") serves as a simple measurable proxy for how literal a version is. — For Mova (learn): Evaluating the text → graph → text compressor could track how often key lexical anchors survive the round trip, a cheap literalness metric beside meaning checks.
• I5. Several independent ancient versions are compared as parallel witnesses to the same source. Their agreements show the stable reading and their disagreements show the ambiguity. — For Mova (world): A compiled idiom lexicon could record multiple attested paraphrases per idiom, with disagreements marking senses that the world model must resolve from context.