Language Log

« previous post |

Following upon our enthusiastic, productive discussions on the main East Asian word for "slave" (奴隷 J. ドレイ M. núlì) a few weeks ago and Chau Wu's drawing of parallels with the corresponding Greek word for a person of that status several days after that, I've become deeply interested in Greek δούλος ("slave").  (See the first three items in "Selected readings".)

What particularly intrigues me is its great time depth — Mycenean, Homer, Classical, Koine, down to Modern Greek — and its breadth of correspondences in other languages:

Related to Mycenaean Greek (do-e-ro /⁠dohelos⁠/), possibly from Canaanite *dōʾēlu “servant, attendant” (compare Late Babylonian (daggālu, “subject, one who waits on another, does their bidding”).

According to Parpola,  the word δοῦλος is related to the ethnonym Dahae (found as Δάοι, Δάαι, Δαι or Δάσαι in Greek sources) and thus related to Sanskrit दस्यु (dasyu, “bandit, brigand”) and Sanskrit दास (dāsa) which originally meant 'demon' and later also 'slave' or 'fiend'.

(Wiktionary)

Asko Parpola, “The Coming of the Aryans to Iran and India and the Cultural and Ethnic identity of the Dāsas”, in Studia Orientalia‎, volume 64 (1988; Helsinki), 195–202.

In a personal communication to me, Asko noted:

I have nothing to add to the suggestion that Iranian dasa might come from PIE *dos- and be related to Greek doûlos < Mycenaean do-e-ro < *dohelos < *doselos. I was thinking then that this might refer to the people of  the earlier of two waves of Greek immigration, who became slaves < war-captives for the people of the second wave. Sanskrit dāsa became to mean ’slave’ from the ethnic self-appellation of the enemies of the Rgvedic people, taken war-captives.

Don Ringe:

Note that the original meaning of dāsá- seems to have been "enemy", and the forms with -h- are specifically Iranian and probably cognate with it.  Mayrhofer, Etymologisches Wb. (1992) s.v. gives a concise but accurate rundown for Indo-Iranian.  Frisk believes that it's a loanword in Greek (thus not really cognate with the Skt. word), and notes suggestions that it was borrowed from Lydian or Carian or Northwest Semitic without endorsing anything.  In the Nachträge he notes that Heubeck thought the Greek word is cognate with the Skt.

Nicholas Sims-Williams:

In addition to the ethnonym Dahae, relatives of OInd. dasyu- etc. are common in Iranian. The simplest form is Khot. daha- "male" = Sogd. PN δx, and *dahyu- "land, country, etc." is widely attested.

Stephanie Jamison:

Mayrhofer (in the Etymologisches Wörterbuch des Altindoarischen, the current standard etymological dictionary) does point out that a connection of δούλος with both dásyu and dāsa is frequently suggested (see s.vv.). Though he falls short of endorsing it, he doesn't dismiss it out of hand.

I'd stay away from "brigand" for dásyu. It's cognate with well-attested Aves. dax́iiu– 'country [or what passes for such in antiquity], inhabitants thereof' and Old Persian dahyu 'id.' The negative sense it has in Vedic probably comes from referring to the inhabitants of some other place than ours. Sim. dāsá-. Although Joel and I originally translated both as 'barbarian' in our Rigveda tr., in the end we went with Dasyu, Dāsa, since the exact referent is never clear. I don't think they mean 'demon' at least directly.

Melanie Malzahn:

The etymology of δούλος is indeed interesting. While the Greek time depth is not particular for this term (many words have a continuity from Mycenean to Greek), the standard view connects the Greek form with the ethnonym that then also entered Vedic as the name of a foreign, later demonized people. The proto-form of Myc. do-e-ro /dohelos/ should be *dós-e-lo- with the suffix *lo being used here as diminutive, i.e., showing the same degradation as we find in Vedic (the o-grade in Greek also fits nicely with the long a in Vedic daasa-).

I am not sure whether the Akakdian forms really belong to the same ethnonym, but from first look I doubt it.

Doug Adams:

I'm afraid Tocharian isn't going to be very helpful here.  The most basic word for 'slave' is käryau, lit. 'the bought one.'  This word is pure Tocharian.  Also there is mañiye (m.)/mañiya (f.) '(household?) servant/slave' borrowed fom Iranian, cf. Old Persian māniya– 'household servant,' a derivative of māna– 'house.'   I'm not exactly sure what a cognate of doulos might look like in Tocharian as the fate of word-initial *d– is not altogether secure (t-, ts-, zero-?).  But nothing that fits any of those bills is attested.

Thanks for bringing the possible Canaanite antecedent to my attention.  I knew only about dasyu-, etc.

J. P. Mallory:

Robert S. P. Beekes, Etymological Dictionary of Greek, 2 vols. Leiden: Brill, 2009.

Miriam Robbins Dexter:

There is an ameliorative feminine form of the Greek word, δούλα (“a woman who assists women during labor and after childbirth", from Modern Greek (first used 1975-80), so still “female slave” in origin, but more of an idea of “helper > help in childbirth” in the modern era.

Julie Wei:

Old Irish  dóer "slave" curiously resembles Myc. Greek do-er-o.

dóer, Wiktionary:

Etymology

Analyzable as do- +‎ fer.

Pronunciation

• • • IPA: /ˈd̪oːi̯ɾ/

Old Irish

Adjective

dóer

1. 1. 1. 1. servile
         2. unfree

Noun

dóer m or f (genitive doír, nominative plural doír)

1. 1. 1. 1. serf

Old Irish  do- , Wiktionary:

Old Irish

Etymology 1

From Proto-Celtic *dus-, from Proto-Indo-European *dus-.

Alternative forms

• • • du-

Prefix

do-

1. 1. 1. impossible, extremely difficult
      2. ill, evil 

Old Irish fer "man", Wiktionary

This is not to assert consanguinity between Ancient Greek and Proto-Celtic (Brythonic, Old Welsh, etc.) or any other language mentioned in this post.  During the next few months, in Sino-Platonic Papers and on Language Log, as I have in the past, I will publish a considerable amount of comparative data between Old Celtic and other languages of the same time frame.  Readers may make of it what they will.  (Here is not the place to rehearse / rehash the abundant archeological data for trans-Eurasian exchange [bronze, wheat, horses, chariots etc.] during the Late Neolithic, Bronze Age, and Iron Age.)

Bottom line (for now):  Greek (δοῦλος [doûlos]), Latin (servus), and English (< "Slav") have completely separate words for "slave", and none of them point to a Proto-IE root meaning "slave(ry)" per se, leading me to believe that, if there were an IE root meaning "slave", it was something else entirely different from these common, widespread European terms.

The Proto-Indo-European (PIE) language may well have had a word for "slave" or "servant", reconstructed as \*h₃órbʰos, meaning "orphan" or "servant". This term was the source of words like the Proto-Slavic *orbъ, from which the modern word "robot" derives, and highlights the interconnectedness of status and servitude across Indo-European languages writ large.

 

Selected readings

• "Correspondences between Ancient Greek doȗle (voc.) 'slave' and 奴隷 Jpn dorei / Tw lô·-lē" (10/9/25)
• "Sino-Japanese n- / d- initial interchange" (10/1/25)
• "Stay hyDRAEted" (9/29/25)
• "Slaves and clients; Arabic Mamluks and mawlas: a fishy Turkic tail" (5/11/21)
• "Kunlun: Roman letter phonophores for Chinese characters" (2/16/21)
• "Kunlun: the origins and meanings of a mysterious place name " (2/24/21)
• "An early fourth century AD historical puzzle involving a Caucasian people in North China" (1/25/19)
• "European slaves in the year 1000 " (5/5/20)
• "Slavs and slaves" (1/17/19)
• "Cool slave / guy / tofu / whatever" (5/12/18)
• "The bearded barbarian" (8/26/15) — especially this comment
• "Misogyny as reflected in Chinese characters" (12/25/15)
• Julie Wilensky, "The Magical Kunlun and 'Devil Slaves':  Chinese Perceptions of Dark-skinned People and Africa before 1500", Sino-Platonic Papers, 122 (July, 2002), 1-51.  free pdf
• Sarah Forsdyke, Slaves and Slavery in Ancient Greece (Cambridge: Cambridge University Press, 2021).
• Dimitris J. Kyrtatas, "The vocabulary of slavery", in A.-F. Christidis, ed., A History of Ancient Greek: From the Beginnings to Late Antiquity, (Cambridge:  Cambridge University Press, 2007), pp. 1056-61.
• Benedicte Nielsen Whitehead, "Hiding in Plain Sight? The Enigma of the Linguistic Remains of Prehistoric Slavery", in Kristian Kristiansen, Guus Kroonen, and Eske Willerslev, eds., The Indo-European Puzzle Revisited:  Integrating Archaeology, Genetics, and Linguistics (published online by Cambridge University Press:  29 April 2023).

[Thanks to Nicholas Sims-Williams, Jim Mallory, Doug Adams, Craig Melchert, Stephanie Jamison, Don Ringe; Asko Parpola, Melanie Malzahn, Hannes Fellner, Miriam Dexter, Ralph Rosen, Julie Wei, Marcel Erdal, Robert Drews, Michael Carasik, and Denis Mair for the data they have generously shared.  None of them are responsible for the comparisons I have indicated.]

October 24, 2025 @ 7:53 am · Filed by Victor Mair under Borrowing, Etymology, Language and history, Language and society

Permalink