This is the hinge of Galatians. Everything from 1:11 has been building a case that Paul’s gospel came by revelation and was recognized, not conferred, by the Jerusalem leadership; everything from 2:15 onward is the doctrinal argument. Between them stands this verse, where the narrative collapses into direct speech and Paul quotes himself confronting Κηφᾷ in public.

Four features carry the weight:

1. ὀρθοποδοῦσιν — a word otherwise unattested in the NT and effectively unattested before Paul, picturing a straight-footed walk. The deviation is named as a gait, not as a doctrine.
2. ἔμπροσθεν πάντων — the rebuke is as public as the conduct was, which is the whole point of the episode’s inclusion.
3. ἐθνικῶς … Ἰουδαϊκῶς — a matched pair of adverbs, both hapax, setting two manners of living side by side.
4. Ἰουδαΐζειν — the verb that gave English its word ‘Judaizer’, used here of what Gentiles are being pressed to do, not of what the pressers are doing.

The syntactical crux is where Paul’s quoted speech ends. The Greek carries no quotation marks, and the decision to close at v. 14, v. 16, or v. 21 is entirely editorial. It is recorded here, not resolved.

Proposed — pending reviewF. F. Bruce, The Epistle to the Galatians: A Commentary on the Greek Text, New International Greek Testament Commentary, Grand Rapids: Eerdmans, 1982.

A compound of ὀρθός (‘straight, upright’) and πούς (‘foot’): to walk a straight course, to go with level feet. It is a NT hapax and is not securely attested in Greek literature before Paul, which raises the real possibility of Pauline coinage — though the morphology is so transparent that coinage would be unremarkable, and later Greek uses it freely.

Two things follow for the rendering.

First, the foot stays. Galatians runs a sustained walking-and-motion field — τρέχω (2:2; 5:7), στοιχέω (5:25; 6:16), περιπατέω (5:16), ἄγομαι (5:18) — and this verb belongs to it. Rendering ‘they were not acting consistently’ or ‘not behaving in line with’ discards the image at the point where it is doing the work: Paul does not say the men held a false doctrine, he says their feet had gone crooked.

Second, the present tense is retained from the direct discourse Paul is reporting. What he saw was a deviation under way, not one concluded.

Proposed — pending reviewJ. B. Lightfoot, Saint Paul's Epistle to the Galatians, 10th ed., London: Macmillan, 1890.

πρός + accusative after a verb of walking is open between two forces, and the choice shades the charge.

The directional sense gives ‘walking straight toward the truth’ — the truth as the destination the gait should be aimed at. The normative sense, common enough in Koine, gives ‘in line with the truth’ — the truth as the standard the gait fails to match.

Note that Paul had κατά + accusative available for an unambiguous norm-phrase, and he uses it four words into v. 2 of this chapter (κατὰ ἀποκάλυψιν). He did not use it here.

The phrase occurs twice in Galatians and nowhere else in Paul in this form (2:5, 14), bracketing the Jerusalem and Antioch episodes. In both places it names what is at stake when practice is at issue — not a proposition to be affirmed but something that can be walked toward or away from.

The genitive is left as a bare genitive. Three construals are live: possessive (the truth belonging to the gospel), epexegetical (the truth which the gospel is), and objective-ish (the truth about the gospel). ‘The gospel truth’ collapses the genitive into an adjective; ‘the truth of the message’ substitutes a different noun. English holds a bare genitive, so it does.

What is worth marking is how concrete the phrase is in context. The ἀλήθεια here is not violated by a false statement about Christ — nobody at Antioch said anything false about Christ — but by who eats with whom. Paul treats table practice as a matter on which the gospel’s own truth is either held or surrendered.

ἔμπροσθεν is an improper preposition with the genitive, ‘in front of, in the presence of’. The genitive plural πάντων is formally ambiguous between masculine and neuter; the setting (a meal, a withdrawal, a confrontation) fixes the personal reading.

The phrase must be rendered in full and must not be softened. Paul could have said he spoke to Cephas; he records that he said it in front of everybody. Per §6, the hardness of the text is not shaded toward comfort: this is a public rebuke of the most prominent apostle in the Jerusalem church, and the verse is included in the letter precisely because it was public.

Supply no identification of the ‘all’. Whether the referent is the whole Antiochene assembly, the Jewish believers who had withdrawn, or those present at the disputed meal is not stated, and ‘before the whole church’ adjudicates.

A §5 flashpoint, and the translation rules require that the range of serious positions be flagged rather than silently adjudicated.

The term spans at least: descent from the people of Judah; residence in or origin from Judea; and adherence to the ancestral way of life (the Ἰουδαϊσμός of Gal 1:13-14). In the Gospels and Acts the civic-geographical sense is sometimes dominant (‘Judean’), and some scholars press that rendering broadly across the NT.

The corpus policy already keeps Ἰουδαῖος lexically distinct from Ἰσραήλ / Ἰσραηλίτης (the covenant-insider name) — a distinction Paul himself exploits at Rom 9:4 and 2 Cor 11:22. Here he uses the civic-ethnic designation, and the choice is preserved.

Proposed — pending reviewJames D. G. Dunn, The Epistle to the Galatians, Black's New Testament Commentary, London: A. & C. Black, 1993.

ὑπάρχω rather than εἰμί: per the standing glossary entry the verb denotes inherent, settled being, and the choice is deliberate. The being-a-Jew is presented as what Cephas constitutionally is, over against the ζῇς that follows.

The participle carries no concessive marker. Greek has καίπερ and καίτοι for that purpose, and Paul uses καίπερ at Phil 3:4. Here he wrote a bare adverbial participle, and per the standing policy (ἐν μορφῇ θεοῦ ὑπάρχων, Phil 2:6; ὤν, Gal 4:1) the literal tier keeps the bare participle and any ‘though’ is confined to the smooth tier and flagged.

The concessive reading is contextually strong — the sentence is built on a mismatch between what Cephas is and how he lives — but ‘though you are a Jew’ states as grammar what is in fact an inference. ‘Being a Jew’ in the literal tier leaves the reader able to see the difference.

A first-class condition: εἰ with the present indicative. The protasis is entertained as real for the sake of the argument. Per the standing policy (Phil 2:1; Jas 2:8; 1 Pet 4:14) it is not rendered ‘since’, which closes the rhetorical gap the conditional deliberately leaves for the hearer to fill. Paul is inviting Cephas to grant the premise, not informing him of it.

The expressed σύ is the marked element. The verb ending already carries the person, so the nominative pronoun is emphatic and deictic — a finger pointed across the room. Per the standing policy on expressed nominative pronouns it must survive in English and must not be absorbed into the verb.

ζῇς is a durative present: a standing manner of life. This matters for the historical reconstruction, since Paul is not describing what Cephas did at one meal but a settled practice he had been following — which is what makes the withdrawal of v. 12 a reversal rather than a policy.

Both adverbs are NT hapax legomena and both are rare in Greek generally; Ἰουδαϊκῶς appears to be a Hellenistic-Jewish formation, and the pairing may well be Paul’s own construction for this sentence. They stand in perfect formal symmetry (‑ικῶς / ‑ϊκῶς) with the negative attached to the second.

The symmetry must survive in English. ‘Live like a Gentile and not like a Jew’ carries it; ‘live as the Gentiles do, rather than following Jewish custom’ breaks it by varying the construction between the members.

Two exclusions. ἐθνικῶς does not mean ‘in a pagan manner’ or ‘irreligiously’ — the ἔθν‑ root is the LXX outsider-designation (גּוֹיִם), not a word for impiety, and importing a verdict here would make Paul accuse Cephas of apostasy rather than of inconsistency (§5). And the adverbs describe a manner of living, not a set of beliefs: ‘live by Gentile standards’ tips toward principle where the Greek names practice.

What the practice consisted of is left unstated. Table fellowship with uncircumcised believers is the obvious referent given vv. 11-13, but Paul does not specify, and the rendering should not.

Already fixed in the glossary as ‘compel / press’: coercion short of physical force but leaving no honourable exit — social, ecclesial, theological pressure. Galatians uses the verb three times of exactly this (2:3, 2:14, 6:12), and its internal consistency is the control on the meaning.

The present tense is the live question.

Note what the verb does not say: it names no demand, no ultimatum, and no stated requirement. On the most natural reading of vv. 12-13, the compulsion is exerted by the withdrawal itself — the Gentile believers can recover fellowship only by becoming Jews. Paul treats a social fact as a coercion, which is the sharpest thing in the verse.

NT hapax. A denominative verb on Ἰουδαῖος with the ‑ίζω suffix, which in Greek regularly forms verbs meaning ‘side with, adopt the manner of, speak the language of’ (cf. ἑλληνίζω, μηδίζω, λακωνίζω). The one secure LXX occurrence is Esth 8:17, where Gentiles, in fear of the Jews, περιετέμνοντο καὶ ἰουδάιζον — were circumcised and adopted the Jewish way of life. Josephus and later Greek use it the same way, of Gentiles taking on Jewish practice.

The decisive lexical point is who the subject is. The verb is used of Gentiles adopting Jewish practice, never of Jews pressing them to. The English noun ‘Judaizer’ — meaning one who urges Torah-observance on Gentiles — is a post-biblical coinage that reverses the Greek direction of the verb, and the apparatus should say so plainly (§9, §11): it is reception history, not lexical evidence.

Rendering options:

• ‘to Judaize’ — retains the single verb and the Greek’s compactness, at the cost of opacity for a modern reader, and at the risk of importing the reversed modern sense.
• ‘to live like Jews’ — transparent, and matches Ἰουδαϊκῶς ζῇς earlier in the sentence, which is a real argument: the sentence’s own vocabulary glosses the verb.
• ‘to become Jews’ — overshoots, specifying conversion and circumcision where the verb names a manner of life.

The reading text uses ‘live like Jews’ so the mirror with Ἰουδαϊκῶς ζῇς is audible; the literal tier keeps ‘Judaize’ so the single lexeme stays visible, with the warning attached.

Proposed — pending reviewHans Dieter Betz, Galatians: A Commentary on Paul's Letter to the Churches in Galatia, Hermeneia, Philadelphia: Fortress, 1979.

Per the standing entry, the LXX equivalent of גּוֹיִם: outsiders to the covenant people rather than an ethnic category. ‘The nations’ in the literal tier keeps the Israel/nations frame visible; ‘Gentiles’ in smooth registers.

Two notes. The noun is articular here and in v. 12 (μετὰ τῶν ἐθνῶν συνήσθιεν), pointing at the specific Gentile believers at Antioch rather than at Gentiles in general — the article is anaphoric. And the word must receive the same English rendering as the identical plural in vv. 8, 9, 12, and 15, since the chapter runs the term as a thread.

Never ‘pagans’ or ‘heathen’ (§5): the referent is Christian Gentile believers, and either rendering would make Paul describe fellow members of the assembly in terms of their former religion.

Greek manuscripts carry no quotation marks. The decision about where Paul’s reported speech to Cephas stops is entirely a modern editorial judgment and must be reported as such, never as manuscript evidence.

The decision is not trivial: it determines whether vv. 15-21 are a report of what was said at Antioch or a fresh address to Galatia, and therefore how directly the justification argument is aimed at Peter.

Proposed — pending reviewJ. Louis Martyn, Galatians: A New Translation with Introduction and Commentary, Anchor Bible 33A, New York: Doubleday, 1997.

Κηφᾷ. The Byzantine tradition and a scattering of other witnesses read Πέτρῳ. The Aramaic form is strongly supported by the earliest witnesses and is the harder reading in a Greek-speaking transmission environment; the substitution runs in one direction only, toward the familiar Greek name. Per the standing glossary entry, Paul’s own distribution in this chapter — Κηφᾶς at 2:9, 11, 14 against Πέτρος at 2:7-8 — is itself data about the sources behind the passage and must not be levelled.

ὀρθοποδοῦσιν. The present is secure. A few witnesses offer variants in spelling without semantic consequence.

Ἰουδαϊκῶς / Ἰουδαΐζειν. The iota-diaeresis is a modern editorial convention; the majuscules write the letters without it.

No variant in this verse materially affects the sense.

This verse generated one of the sharpest exegetical disputes of the patristic period, and the shape of the dispute is instructive.

Origen and Chrysostom, followed by much of the Greek tradition, read the confrontation as staged: Peter’s withdrawal was a deliberate accommodation, and Paul’s rebuke was a piece of agreed theatre for the benefit of onlookers, so that neither apostle is convicted of error. Chrysostom develops this at length, arguing that two men so closely allied could not genuinely have been at odds and that the public form of the rebuke is evidence of its pedagogical purpose.

Jerome took over this reading in his commentary on Galatians. Augustine attacked it directly in their famous correspondence, on the ground that it makes Paul’s οὐκ ὀρθοποδοῦσιν and his κατεγνωσμένος ἦν (v. 11) into untruths, and that a Scripture in which the apostles dissemble cannot be trusted anywhere. Jerome eventually conceded. Augustine’s position — that Peter genuinely erred and Paul genuinely rebuked him — became the dominant Western reading.

The exchange is worth recording not because it settles the exegesis but because it shows what was felt to be at stake: whether the text may be read against its surface to protect the standing of its characters. Augustine’s instinct is the one this project follows (§6, Working Principle 6): if the text is hard, the translation is hard too.

The verse was also a load-bearing text in the Reformation, where Peter’s fallibility and the public correction of an apostle by an apostle were pressed into service in a quite different controversy. That is reception, not evidence.

Proposed — pending reviewJohn Chrysostom, Commentary on the Epistle to the Galatians, in Nicene and Post-Nicene Fathers, First Series, vol. 13, ed. Philip Schaff, Buffalo: Christian Literature Company, 1889. · Augustine, Letters 28, 40, 75, 82 (the correspondence with Jerome on Galatians 2:11-14), in Nicene and Post-Nicene Fathers, First Series, vol. 1, ed. Philip Schaff, Buffalo: Christian Literature Company, 1886.

Three things the verse does not say, and which the rendering must not supply.

It does not say what Cephas demanded. The verb ἀναγκάζεις names a compulsion; it does not name circumcision, dietary observance, or any stated requirement. Gal 2:3 and 6:12 make circumcision the issue elsewhere in the letter, and the inference is reasonable — but it is an inference drawn from the letter’s wider argument, not from this clause.

It does not say whether anyone yielded. The present tense is open between the conative and the actual.

It does not say how Cephas answered. Paul records his own speech and stops. The absence of an outcome is conspicuous in a narrative otherwise so concerned to establish what the Jerusalem leaders conceded (2:6, 9), and readers have filled the gap in both directions for nineteen centuries. The text does not.

What the verse does establish, and establishes without qualification, is that for Paul the ἀλήθεια τοῦ εὐαγγελίου is something a gait can deviate from — that the gospel’s truth is held or surrendered at a table as really as in a confession.