Doulos: Why Your Bible Says 'Servant' When Paul Wrote 'Slave'

Open Romans 1:1 in almost any English Bible and you’ll read that Paul called himself a “servant” of Christ Jesus. The Greek word underneath that translation is doulos (δοῦλος) — singular. Philippians 1:1 opens with the same word, but there it’s plural, douloi (δοῦλοι), since Paul is speaking for himself and Timothy together. For most of the history of English Bible translation, that word has been rendered “servant” — a choice so consistent across denominations and centuries that few readers ever stop to ask whether it’s accurate. In recent decades, a number of New Testament scholars have argued it isn’t, and that something real gets lost in the softening.

What “Servant” Erases

The standard Greek lexicon for New Testament study, BDAG (Bauer-Danker-Arndt-Gingrich), glosses doulos plainly as “slave,” treating “servant” and “bondman” as euphemistic alternatives. In the ancient Mediterranean world, a doulos was the legal property of a master — someone who could not simply resign, who could be bought, sold, or inherited, and who had no independent legal will.

Murray J. Harris makes the case at length in Slave of Christ: A New Testament Metaphor for Total Devotion to Christ (New Studies in Biblical Theology 8; InterVarsity Press, 1999). Harris found that among twenty major twentieth-century English translations, only one — Edgar J. Goodspeed’s 1923 The New Testament: An American Translation — consistently rendered doulos as “slave.” Harris’s argument is not merely lexical. He reads Paul’s self-designation as “slave of Christ” as a deliberately extreme image: not modest self-deprecation, but a claim of total ownership and unconditional submission, with no room for a volunteer’s freedom to walk away. That reading reached a much wider popular audience through John MacArthur’s Slave: The Hidden Truth About Your Identity in Christ (Thomas Nelson, 2010), which built directly on this scholarship for a general readership.

How “Servant” Became the Default

The euphemistic translation tradition is old. William Tyndale rendered doulos as “servant” in his sixteenth-century English New Testament, and the Geneva Bible and King James Version followed his lead. One piece of the explanation lies further back, in the Latin Vulgate, where doulos became servus — a word the Reformers, working from memory of the Vulgate, carried over into English as “servant.” By Tyndale’s era, chattel slavery had effectively disappeared from England, so “servant” carried no connotation of ownership; a servant could, at least in principle, quit. The English word “slave,” by contrast, was already a term of contempt and low status at the time — reason enough, translators reasoned, to avoid it in a text meant to be read aloud in church. Later English translations largely inherited this choice rather than re-deriving it from the Greek.

Why Roman Slavery Isn’t Atlantic Slavery

Here the picture needs to be balanced rather than simply reversed. The moment a translation says “slave,” English-speaking readers tend to import the images of the nineteenth-century American South — hereditary, race-based chattel slavery with almost no path to freedom. Scot McKnight, in his commentary The Letter to Philemon (NICNT; Eerdmans, 2017), devotes roughly twenty-two pages to first-century Roman slavery specifically to guard against two opposite errors: minimizing it as “not that bad,” and collapsing it into a direct equivalent of modern slavery. Roman slavery was not organized around race. Manumission — the formal freeing of a slave — was a routine, institutionalized practice, and some slaves held real social and economic standing as household managers, tutors, or business agents. John Byron’s survey of the scholarship, “Paul and the Background of Slavery: The Status Quaestionis in New Testament Scholarship” (Currents in Biblical Research 3.1 [2004]: 121–145), likewise cautions against reducing the origin of Paul’s slavery language to either Greco-Roman social practice or Jewish “servant of God” idiom alone — the evidence, he argues, resists a tidy single source.

What Modern Translations Have Done With It

The debate is not merely academic; it has shaped real translation decisions. In the summer of 2010, the ESV Translation Oversight Committee met at Tyndale House in Cambridge specifically to reconsider how doulos and the Hebrew ’ebed should be rendered, finalizing changes in spring 2011. The outcome was a compromise: the committee voted 9–3 to revert four instances of “slave” in 1 Corinthians 7 back to “bondservant.” The 2021 Legacy Standard Bible, published by the Lockman Foundation, went the other direction, adopting “slave” as its consistent rendering of doulos throughout the New Testament. Two serious translation committees, working from the same Greek text, reached opposite conclusions — which is itself worth noticing. There is no single settled answer here, only a translation choice that carries real theological weight either way.

References

• Murray J. Harris, Slave of Christ: A New Testament Metaphor for Total Devotion to Christ (NSBT 8; Leicester: IVP, 1999)
• John MacArthur, Slave: The Hidden Truth About Your Identity in Christ (Nashville: Thomas Nelson, 2010)
• Scot McKnight, The Letter to Philemon (NICNT; Grand Rapids: Eerdmans, 2017)
• John Byron, “Paul and the Background of Slavery: The Status Quaestionis in New Testament Scholarship,” Currents in Biblical Research 3.1 (2004): 121–145
• Walter Bauer, Frederick W. Danker, William F. Arndt, and F. Wilbur Gingrich, A Greek-English Lexicon of the New Testament and Other Early Christian Literature (BDAG), 3rd ed. (Chicago: University of Chicago Press, 2000), s.v. δοῦλος
• Justin Taylor, “The ESV Translation Committee Debates the Translation of ‘Slave,’” The Gospel Coalition (2011)
• “Why does the LSB translate ‘doulos’ as ‘slave’?” Legacy Standard Bible FAQ (lsbible.org, 2021)

Comments

Loading comments…

Leave a comment

Comments are reviewed before being published. Only your name and comment are stored — no personal data collected.

Name (required)Write a comment (max 1,000 characters)

Submit comment