Use the following persistent identifier: http://nrs.harvard.edu/urn-3:hul.ebook:CHS_Benveniste.Indo-European_Language_and_Society.1973.
Chapter 5. The Slave and the Stranger
The free man, born into a group, is opposed to the stranger (Gr. xénos), that is to say, the enemy (Lat. hostis), who is liable to become my guest (Gr. xénos, Lat. hospes) or my slave if I capture him in war (Gr. aikhmálōtos, Lat. captivus).
A stranger by necessity, the slave is designated in the Indo-European languages, even modern ones, either by a foreign word (Gr. doûlos, Lat. servus), or by the name of a foreign people (slave < Slav).
The notion of slave is not designated by a single word, and this is true both of the Indo-European languages as a whole and for quite a number of dialects.
In the ancient civilizations, the status of a slave puts him outside the community. The word for the slave has this negative aspect.
There are no slaves who are citizens. They are always introduced into the city from outside, in the first instance as prisoners of war. In the primitive Indo-European society, as in the ancient non-European societies (Sumero-Akkadian, for instance), the slave is a man without rights, reduced to this condition because of the laws of war.
A little later, a slave may be acquired by purchase. To the great markets of Asia Minor slaves flowed in abundance, coming from all regions, but their state was due after all to their being prisoners of war or people carried off in raids. Asia Minor supplied large contingents of them, to judge by the nicknames of slaves which are often ethnics: Phrygian, Lycian, Lydian, Samian, etc.
Given the conditions, we can understand why the slave was identified with the stranger, and why they were called by specific names of places. Apart from this, certain qualifications define them as captured or bought. There are two series of designations which can sometimes coincide; that of “prisoner of war” and that of “slave” properly called as such.
Let us consider, in the first instance, the “prisoner of war.” His condition is often expressed by various words denoting “taken” (e.g. French prisonnier < pris); this is the case in Latin with captus, captivus, with Gr. aikhmálōtos (αἰχμάλωτος), Homeric douríktētos (δουρίκτητος), Gothic frahunþans, Old Slavic plěnĭnikŭ (Russ. plennyj). Greek aikhmálōtos must be looked at a little more closely, not because the sense of “taken at the point of the spear” is obscure; the composition of the word was clear to the Greeks themselves, a proof of this being the doublet—douríktētos which was formed with the word dóru ‘lance’ as first element. But the interpretation of aikhmálōtos is not as obvious as it seems: –alōtos does not simply mean “taken”; this is rather a rough translation. The root of halískomai (ἁλίσκομαι) conveys the idea of being suddenly seized, being taken unawares, without any possibility of defense, whether it is applied to a city or a person: from this comes the sense of the perfect hḗlōka (ἥλωκα) ‘I am lost’, which is one of the rather irregular forms attached to halískomai. This notion of surprise, which eliminates the power of resistance, makes aikhmálōtos a quite different expression from captus, captivus, which is derived from capio ‘to take with the hand’.
The substantive aikhmḗ (αἰχμή) must also be considered. It designates the “point of the spear”; then, by extension, the whole of the weapon, a spear, pike or javelin, etc. What we must note is that aikhmḗ is the weapon par excellence of the Homeric warrior, so much so that the derivative aikhmētḗs (αἰχμητής) is the poetical term for warrior; and, further, that in Homer it has always an elevated value. Thus, to put an end to the fight between Ajax and Hector, Talthybius says to them: “Zeus loves you both… you are both valiant warriors,” amphotérō gàr sphō̂ï phileî … Zeus, / ámphō ďaikhmētá (ἀμφοτέρω γὰρ σφῶϊ φιλεῖ…Ζεύς, / ἄμφω δ’ αἰχμητά), (Il. 7, 280-281). The weapon called aikhmḗ is therefore that which specifies the warrior, without which he loses his status and, as a consequence, his power in battle.
In Iranian, the designation for “prisoner of war” reflects a different image: Middle Iranian dast–grab, literally “taken with the hand.” This time it is the hand which is the instrument of capture, which is also suggested by captivus and High German hafta, taken from a root corresponding to Lat. capio. The Iranian verb grab– ‘to take’ is used in the Persian Achaemenid inscriptions of Darius in the sense “to take prisoners of war.” Dasta ‘hand’ relates to the same notion: “he put him in my hand,” says Darius of Ahura Mazdā, with reference to an enemy. Thus dasta and grab- pool their respective senses in the expression for a prize of war. Similarly, one sees in Armenian jerb–a–kal ‘prisoner of war’ (literally “taken with the hand”), a calque on the Middle-Iranian dast–grab; this is further evidence of the Iranian influence on Armenian.
All these compounds depict the prisoner of war according to the manner of his capture. But these are not the only terms. We must mention also Old Iranian banda(ka), Skt. bandhin, which define the prisoner as he who is “bound.” We find in Gothic frahunþans, a participle of frahinþan ‘to make a prisoner, Gr. aikhmalōtízein’, cf. hunþs ‘capture in war, aikhmalōsía’, Old English hunta ‘hunter’, huntian ‘to hunt’ derivatives from a root not attested elsewhere which have become specialized in the terminology of the hunt and of war. The same notion inspired Old Slavic plěnŭ ‘booty’ (Russ. polón), whence plěniti ‘take prisoner’ and plěnĭnikŭ ‘prisoner’, to which the Lithuanian pel̃nas ‘advantage, gain’ and Skt. paṇa ‘(gambling) stake’ correspond; these words can be linked up with the root *pel– of Gr. pōleîn ‘to put up for sale’ (cf. Book One, Chapter Ten) and this would associate the idea of “spoils, prize of war” with that of “economic profit.”
We must now turn to the word for the “slave.”
The best-known Greek term, doûlos (δοῦλος) is the usual one in the Homeric period. Although it does not appear in Homer, some derivatives are already Homeric, such as the feminine doúlē and the adjective doúlios (δούλιος) in expressions like doúlion ē̂mar (δούλιον ἦμαρ) ‘day of servitude, condition of a slave’ (see especially Il. 6, 463).
There are in Homer other words, such as dmṓs (δμώς) and also to some extent oikétēs (οἰκέτης), although with the latter word it is difficult to draw the line between “servant” and “slave.” We leave these two terms aside; they are derivatives from the word for “house” (cf. Book Three, Chapter Two). Virtually equivalent is Lat. famulus, although the idea behind it is different. From famulus the collective noun familia was coined. What constitutes the familia is, etymologically speaking, the whole of the famuli, the servants who live in the same house. The notion does not coincide with what we understand by “family,” which is restricted to those connected by kinship.
It seems that we can associate the term doûlos with this notion of “house”; the specific word for slave, if we accept the testimony of Hesychius, who glosses doûlos as oikía ‘house’, while a compound dōlodomeîs is glossed oikogeneîs ‘born in the house’. Consequently, doûlos would be close in sense to oikétēs, whatever Greek dialect it may first have belonged to.
But now doûlos has appeared in Mycenaean in the form of do-e-ro (do-e-lo), which presupposes a prototype *dowelo– or *doselo-. This greatly complicates the origin of this term, which has thus been in use in the Hellenic world at least since the twelfth century BC. Only two hypotheses compatible with this situation shall be discussed. An ancient *doselo– could be compared, for its root, to the Indo-Iranian term dāsa– which, as we have seen, has taken on in Indic the sense of “barbarian, slave.” But we have also seen that dāsa-, in the Indo-Iranian period, was probably merely the name for “man” (cf. Book Three, Chapter Two). It is difficult to see how the correspondent could have acquired from the most ancient Greek onwards, under the form of *doselo-, the sense of “slave.” Thus we can only suppose, as scholars have done before, that doûlos was taken from a non-Indo-European language of the Aegean basin. But the borrowing must have taken place much earlier than was thought, and must have entered Greek in the form represented by Mycenaean doelo. The chances of finding the origin diminish the further back in time the term in Greek recedes. There are other pointers which suggest that doûlos is a foreign word. First, we have the geographical distribution of proper names in doulo-, which indicates an Asiatic origin, although we are unable to specify the language of Asia Minor that acted as the source. Lambertz has collected the ancient examples of doûlos and the numerous proper names composed with doûlos.  Most of these names are attested in Asia Minor, so much so that it seems probable that doûlos comes from Asia Minor.
Besides, it would not be surprising if Greek employed a foreign term to designate the slave, because—and this is frequently the case with this term in Indo-European—the slave is necessarily a stranger: the Indo-European peoples only knew what we may call “exodouly.”
The same is true of the Latin word servus.  It is impossible to consider servus as a derivative of the verb servare and to imagine that it was the function of the servus to “guard.” The verb servare has a clear Indo-European etymology: Avest. harva ‘who watches’, Gr. horân (ὁρᾶν) ‘to observe, consider’. But servus indicates the legal and social condition of a slave and not a specific domestic function. Surely the servus was not obliged to servare.
Since no citizen could be a slave in Rome, it is probably outside Rome and the Latin vocabulary that we must look for the origin of the word servus. Now there is considerable evidence from proper names to show that the root existed in Etruscan in the form serui-, serue-. We find also among Latin proper names some of Etruscan formation, such as Servenius, Servena, Servoleni, with the suffixes which characterize Latin names of Etruscan origin. It is therefore probable that servus is an Etruscan term, although it has not yet been found in any Etruscan inscriptions which we are in a position to interpret. Thus, in very different historical circumstances, we find for servus the same initial situation which is very probable in the case of doûlos.
We can also recall the modern French word esclave ‘slave’: it is properly the name for the Slavs in the South Slavic form (Serbian or a related dialect), an ethnic Slověninŭ . From Slověninŭ is derived the Byzantine Greek form Sklavēnoí (Σκλαβηνοί) (Italian Schiavoni) which, being regarded as a derivative, produced the ethnic Sklávoi (Σκλάβοι). This was the source in the whole western world of the word esclave and its related forms. We find another parallel in the Anglo-Saxon world, where wealh ‘slave’ properly means “the Celt,” the subject people.
We can point to yet another parallel, this time a medieval one; it concerns not the slave, but the vassal, who has an inferior and subject status: vassus (from which comes vassalis) is in the Latin of the period a borrowing from the Celtic form represented in Irish by foss, Welsh guas, both meaning “servant, slave.” Thus, each language borrows from another its designation for “slave.” A people even designates the slave by the name of its neighbors, if they have been subjected by it. Here we see the emergence of a profound semantic correlation between the expression “free man” and its opposite “slave.” The free man designates himself as ingenuus, as “born in” the society in question, hence endowed with full rights; correlatively, the one who is not free is necessarily someone who does not belong to this society; he is a stranger without rights. A slave is something more: a stranger captured or sold as prize of war.
The notion of stranger is not defined in the ancient civilizations by fixed criteria, as he is in modern societies. Someone born elsewhere, provided that he has certain conventional links, enjoys some specific rights, which cannot be granted even to citizens of the country: this is shown by the Greek xénos ‘stranger’ and ‘guest’, that is to say, the stranger who benefits by the laws of hospitality. Other definitions are at hand: the stranger is “he who comes from outside,” Lat. advena, or simply “he who is outside the limits of the community,” Lat. peregrinus. There is no “stranger” as such: given the diversity of notions, the stranger is always a particular stranger, who carries a distinct status. In short, the notions of enemy, stranger, guest, which for us form three distinct entities—semantically and legally—in the Indo-European languages show close connections.
We have studied above (Book One, Chapter Seven) the relations between hostis ‘enemy’ and hospes ‘guest’; Latin hostis ‘enemy’ has a correspondent elsewhere in the Gothic gasts ‘guests’. In Greek xénos designates the “stranger” and the verb xeinízō refers to “hospitable behavior.”
This cannot be understood except by starting from the idea that the stranger is of necessity an enemy and correlatively that the enemy is necessarily a stranger. It is always because a man born elsewhere is a priori an enemy that a mutual bond is necessary to establish between him and the EGO relations of hospitality, which would be inconceivable within the community itself. This dialectic “friend-foe,” as we have seen, is already operative in the notion of phílos: an enemy, even one’s adversary in battle, may become temporarily a phílos, as the result of a pact concluded according to the rites and customary pledges. In the same way, in the early history of Rome, the stranger who becomes a hostis enjoys pari iure cum populo Romano, legal rights equal to those of the Roman citizen. Rites, agreements and treaties thus interrupted this permanent situation of mutual hostility which existed between peoples or cities. Under the protection of solemn conventions and by means of exchange arrangements, human relationships could develop, and as a result the words for agreements or legal status came to denote sentiments.
[ back ] 1. Glotta V, 1914, p. 146, n. 1.
[ back ] 2. The demonstration has been published in volume X of the Revue des Etudes Latines (1932), pp. 429ff.