Use the following persistent identifier: http://nrs.harvard.edu/urn-3:hul.ebook:CHS_Benveniste.Indo-European_Language_and_Society.1973.
Chapter 6. Cities and Communities
The Western dialects of Indo-European (Celtic, Italic, Germanic, Baltic) have preserved the word *teutā, derived from a root *tew– ‘to be swollen, powerful’, to designate “the people” as a full development of the social body. Quite naturally, this term, which supplied national ethnics among the Germans (Teutoni, deutsch) acquired the opposite meaning when Slavic borrowed it from German: Old Slav. tŭždĭ means “stranger.”
The Greek pólis and the Latin civitas, which were closely linked in the development of Western civilization, provide a good illustration of the phenomenon of convergence in institutional expressions: nothing could be more different at the outset than the old Indo-European word for “citadel” (cf. Gr. akró–polis) and the Latin derivative civitas ‘the whole body of citizens’. Arya, which signifies “people” (= my people) in Indic and was the source of the name of Iran ( < aryānām) is the common ancient designation of the “Indo-Iranians.” Isolated in Iranian, arya can be analyzed in Sanskrit as a derivative from arí; the latter seems to designate, in contrast to the stranger, the man of my people; perhaps more precisely, the relation by marriage, the member of the other exogamic moiety.
We have analyzed, by means of the terms which express it, the condition of the free man, born and integrated within a society and enjoying full rights that belong to him by birth.
But how does this man imagine the society to which he belongs and how can we form a picture of it ourselves? Do we know of a “nation,” dating from the time of the Indo-European community, which is designated by a single and constant term? To what extent could an aggregate of tribes conceive of itself as a political entity and call itself a “nation”?
Let us state straight away that there is no term, from one end of the Indo-European world to the other, which designates an organized society. That is not to say that the Indo-European peoples did not evolve this concept; we must guard against concluding that a deficiency in the common vocabulary implies the absence of the corresponding notion in the dialectal prehistory.
In fact there are a whole series of terms which encompass the whole extent of territorial and social units of varying dimensions. From the beginning these territorial organizations appear to be of great complexity, and each people presents a distinct variety.
There is nevertheless a term which is attested in the western Indo-European world over a considerable area. In Italic, excluding Latin, this term is represented by the Umbrian word tota, which means “urbs” or “civitas,” “town” or “city.” In the great lustration ritual called the Iguvine Tablets, which contain a detailed list of sacrificial rites, processions, and prayers, carried out in order to secure the favors of the gods for the city and territory of Iguvium, the formulae totaper iiouina, tutaper ikuvina ‘for the city of Iguvium’ often recur. No distinction is made between the town and the society: it is one and the same notion. The limits of the habitation of a given group mark the boundaries of the society itself. Oscan has the same word in the form touto ‘city’ and Livy (xxiii, 35, 13) tells us that the supreme magistrate in Campania was called meddix tūticus ‘iudex publicus’.
We find *teutā also in Celtic, in Old Irl. tuath ‘people, country’, in Welsh tud ‘country’ (Breton tud ‘people’) and in the Gaulish proper names Teutates, Teutomatus, etc.
The corresponding term in Germanic is Gothic þiuda ‘Gr. éthnos (ἔθνος), people, nation’, an important term because of its date and because it is constant from the oldest Germanic text onwards, important also because of its extent and persistence. We have seen above (Book Three, Chapter Two) its important derivative þiudans ‘chief’. From the Old High German form deot ‘Ger. Volk’, there was formed by means of the very frequent suffix –isc– the adjective diutisc (transcribed in Middle Latin as theodiscus), which developed to German deutsch. This derivative at first designated the language of the country, the popular language as opposed to the learned language, Latin; then it became the ethnic for a part of the German people—those who called themselves “those of the people,” to be understood as “those of the same people as we, those of our community.” Another ethnic formed from the same root is Teutoni. It is as well to note that, in the evolution which has produced the ethnic deutsch, it was the language to which this description first applied. A curious testimony to the peculiarity of use survives in the shape of the German word deuten, which is traced to the same origin as deutsch. In fact deuten, Old High German diuten, comes from a Germanic *þeudjan, a verb derived from þeudō– ‘people’; its meaning would then have been “to popularize, to make accessible to the people” (the message of the Gospels), then generally “to explain, interpret.”
In this dialectal area Baltic is also included; Lith. tautà ‘people, race’, Old Prussian tauto ‘country’. Here Old Slavic shows an interesting divergence vis-à-vis Baltic, both in the form and the sense of the adjectives tŭždĭ and štŭždŭ, which signify “foreign” (Russian čužoj). In reality the Slavic forms which represent *tudjo– and *tjudjo– do not come from an inherited root; they are derivatives from a Germanic loanword, and this explains the sense of “foreign.”
It is easy to understand, says Meillet, that an adjective coined from a foreign word signifying “nation” should become the word for “stranger”; the Germanic nation was for the Slavs the foreign nation par excellence: the němĭcĭ, that is the dumb, the βάρβαρος, is the German. It is incidentally curious that Lettish tauta at an early date meant mainly a foreign people.  Thus the form and sense of Slavic tŭždĭ confirms that the term *teuta characterized the Germanic peoples, in particular in the eyes of the neighboring Slavs.
Apart from Italic, Celtic, Germanic and Baltic, it seems that we must include Thracian and Illyrian among the languages which possessed the word *teutā, to judge by the Illyrian proper names Teutana, Teuticus, Thracian Tautomedes, a fact which extends this lexical area towards Central and Eastern Europe. But contrary to a widely held view, we must exclude the Hittite tuzzi-, which signifies “camp,” and refers only to the army. Some scholars proposed a different solution and traced back to *teutā– the Latin adjective tōtus ‘entire, all’. This connection has a certain appeal, for it would relate the notion of “totality” to that of “society”; it is all the more attractive because another adjective meaning “all,” Skt. viśva-, Av. vispa-, has been adapted to viś– ‘tribe’. But this origin for tōtus is not admissible except at the cost of a number of indemonstrable hypotheses: (1) that the ō of tōtus, instead of the expected *tūtus, is to be explained as a dialect form; (2) that the feminine *teutā directly produced in Latin an adjective *teutus, which later disappeared without a trace, whereas in the languages in which *teutā remained alive, it never produced a derivative indicating totality. Thus this affiliation is hardly probable. It seems that tōtus must be connected in Latin itself with tōmentum ‘stuffing’ and that the first sense of tōtus was, more vulgarly, “stuffed full, compact,” which developed to “complete, entire.”
The formation of the social term *teutā is clear. It is a primary abstract in *-tā made from the root *teu– ‘to be swollen, mighty’. This root was very productive. Notably, it has given rise in Indo-Iranian to the verb “to be able,” Av. tav-, and numerous nominal forms with the same sense: Sanskrit tavas– ‘strength’, taviṣī– ‘might’, Old Persian tunuvant– ‘mighty’, etc. *teutā may therefore be explained roughly as “plenitude,” indicating the full development of the social body. An analogous expression is found in Old Slavic plemę ‘tribe’ (Russ. plemja ‘tribe, people’), which is derived from the root *plē– ‘to be full’, like Gr. plē̂thos ‘crowd’, and perhaps Latin plebs.
The group of dialects which have *teutā (Celtic, Germanic, Baltic, Italic) form a continuous zone in Europe, from which Latin and Greek are excluded to the south and Slavic, Armenian and Indo-Iranian to the east. This dialect distribution apparently implies that certain ethnic groups, those which were to become the Indo-Iranians, Latins and Hellenes, had become separated from the community before the term *teutā came into use among a certain number of peoples who became established in the center and west of Europe. In fact in Latin, Greek and Indo-Iranian different terms are in use to denote the respective societies.
We must take the Greek term pólis (πόλις) and Latin civitas together. Intrinsically they have nothing in common, but history has associated them first in the formation of Roman civilization, in which Greek influence was paramount, and then in the development of modern Western civilization. They are both the concern of a comparative study —which has not yet been attempted—of the terminology and political phenomenology of Greece and Rome. For our purposes two points must be stressed: the Greek pólis, even in historical times, still shows the sense of “fortress, citadel,” as Thucydides said: “the akrópolis (citadel) is still today called pólis by the Athenians” (II, 15). This was the prehistoric sense of the word, to judge by its Vedic correspondent pūr ‘citadel’ and Lithuanian pilìs ‘castle, stronghold’. We thus have here an old Indo-European term, which in Greek, and only in Greek, has taken on the sense of “town, city,” then “state.” In Latin things are quite different. The word for “town,” urbs, is of unknown origin; it has been conjectured—but without proof—that it may come from Etruscan. But it is a fact that urbs, in the sense of “town,” is not correlative with the Greek pólis, but with ástu (ἄστυ); its derivatives came to have senses which were calques of the corresponding Greek word, e.g. urbanus ‘of the town’ (as opposed to rusticus ‘of the country’), which came to mean “fine, polished” after the Greek asteîos. To correspond to Gr. pólis, Latin has a secondary term civitas, which literally indicates the entire body of cives ‘fellow-citizens’. It follows that the connection established in Latin between civis and civitas is the exact reverse of that shown in Greek between pólis ‘city’ and polítēs ‘citizen.’ 
In the principal eastern group of Indo-European, in Indo-Iranian, a term of quite a different kind may represent the notion studied here, but in the ethnic aspect rather than the political one: this is ārya-, which was at first a social qualification before becoming the designation of the community; it was in use both in India and in Iran from the earliest records.
All terms of an ethnic character were in ancient times differential and oppositional. The names which a people gives itself expresses, either clearly or otherwise, the intention of setting itself off from neighboring peoples; it affirms that superiority inherent in the possession of a common, intelligible language. This is why the ethnic often forms an antithetic pair with the opposed ethnic. This state of affairs is due to the little noticed difference between modern and ancient societies with regard to the notions of war and peace. The relation between peace and war was once exactly the reverse of what it is today. For us peace is the normal condition, which is interrupted by a state of war; for the ancients, the normal state was war, to which peace puts an end. We have little understanding of anything about the notion of peace and of the vocabulary which designates it in ancient society if we do not grasp that peace intervenes as a sometimes accidental and often temporary solution to a quasi-permanent state of hostility between towns and states.
The problem of the word ārya is of interest because, in the region defined as Indo-Iranian, it is a designation which free men apply to themselves as opposed to slaves, and also because it is the only word which comprises a common nationality, that of those whom we must call “Indo-Iranians.”
For us, there are two distinct entities, India and Iran. But seen in the light of evolution from the Indo-European parent language, the distinction between “India” and “Iran” is inadequate. The word “India” has never been used by the inhabitants of the country; whereas the Iranians do call themselves “Iranians.”
This difference is due precisely to the uneven survival, between one region and the other, of the ancient word ārya. The Greeks, to whom we owe our knowledge of India, themselves first knew India through the mediation of Persia. An evident proof of this is the form of the root Indía (Ἰνδία), generally Indikḗ (Ἰνδική), which in fact corresponds to the name of the river and of the province called “Indus,” Skt. Sindhu. The discordance between the Greek and the Sanskrit is such that a direct borrowing of the indigenous form is out of the question. On the contrary, everything is explained if the Persian Hindu was the intermediary, since the initial h– corresponds regularly to s– in Sanskrit, while the Ionian psilosis accounts for the root ind– (ἰνδ-) with loss of the initial aspirate. In the Persian inscriptions of Darius, the term Hindu only applies to the province which is today called Sindh. Greek usage has extended this name to the whole country.
The Indians, at an early date, gave themselves the name of ārya. This form ārya is used in Iranian territory as an ethnic term. When Darius lists his ancestry, “son of Vištāspa, grandson of Aršāma,” he adds to characterize himself arya ariyačissa ‘Aryan of Aryan stock’. He thus defines himself by a term which we would now express as “Iranian.” In fact it is arya– which, in the genitive plural form aryānām, evolved in a more recent phase of Persian to the form ērān, later īrān. “Iranian” is thus the continuation of ancient ārya in Persian territory proper.
Very far away, towards the northwest, encircled by peoples of Caucasian speech, there is an Iranian enclave in the shape of a people called Ossetes, descendants of the ancient Alani, who were of Sarmatian stock. They represent the survival of the ancient Scythian peoples (Scythians and Sarmatians) whose territory once comprised the whole of south Russia as far as Thrace and the Balkans. The name of Alani goes back to *Aryana-, which is yet another form of the ancient ārya. We thus have a proof that this word is an ethnic description preserved by several peoples belonging to the “Iranian” family.
In Iranian, arya is opposed to anarya “non-arya”; in Indic ārya serves as the antithetic form to dāsa– ‘stranger, slave, enemy’. Thus the term confirms the observation made above that there is a fundamental difference between the indigenous, or the “self,” and the stranger.
What does ārya mean? This is a very difficult problem which is seen in all its complexity if it is given its place in the Vedic vocabulary; for Arya is not isolated in Sanskrit, as it is in Iranian (where it appears as a word not amenable to analysis, serving only to describe those who belong to the same ethnic group). We have in Vedic a coherent series of words, proceeding from the form which is at once the most simple and the most ancient one, arí; the group comprises no fewer than four terms: arí, with its thematic derivatives árya and aryá, and fourthly, with lengthening of the root vowel, ārya. The difficulty is to distinguish these forms by their sense and to recognize their relationship. The basic term, arí, presents itself in so confused and contradictory a way that it admits flatly opposed translations. It is applied to a category of persons, sometimes only to one, designated sometimes in a friendly and sometimes in a hostile way. Often the author of the hymn decries the arí, from which we may conclude that he regards him as his rival. However, the arí as the singer offers sacrifice and distributes wealth; his cult is addressed to the same gods with the same ritual gestures. This is why we find arí translated in the dictionaries by “friend” and by “enemy” concurrently.
The German Indologist P. Thieme devoted a detailed study to this problem in 1938; it is entitled Der Fremdling im R̥gveda, because at the end of a long analysis, the author believes he can translate the root arí– as “stranger.” The two contradictory senses “friend” and “enemy” may be compared, he suggests, to the two senses of *ghosti-: on the one hand Lat. hostis ‘guest’, Got. gasts ‘guest’, on the other Lat. hostis ‘enemy’. Similarly, arí is “the stranger, friend or enemy.” Based on arí, the derivative arya would signify “he who has a connection with a stranger,” hence “protector of the stranger, German gastlich ‘hospitable’,” and also “master of the household.” Finally, from arya– the secondary derivative ārya would literally mean “belonging to the guests”; hence “hospitable.” The ārya called themselves “the hospitable ones,” thus contrasting their humanity with the barbarism of the people who surrounded them.
Following this study, there appeared from 1941 on a number of works by M. Dumézil, who proposed other interpretations which tend to establish the social sense and then the ethnic sense of this family. 
On the whole our views are close to those of Dumézil. But it will not be possible to justify them here in detail. The examples involve, for the most part, detailed questions of Vedic exegesis, and the discussion would require a whole book of its own. We shall limit ourselves to a few observations and a summary definition.
In such matters, philological criteria must not run counter to intrinsic probabilities. To define the Aryans as “the hospitable ones” is a thesis remote from all historic reality; at no time has any people whatsoever called itself “the hospitable ones.”
When peoples give themselves names, these are divided, as far as we can understand them, into two categories; if we exclude names of a geographical character, they are either (1) an ethnic consisting of a complimentary epithet, e.g. “the valiant,” “the strong,” “the excellent,” “the eminent” or (2) most often they simply call themselves “the men.” If we start with the Germanic Ala–manni and follow the chain of peoples, whatever their origin or their language, to Kamchatka or the southern tip of South America, we encounter peoples by the dozen who call themselves “the men”; each of them thus presents itself as a community of the same language and the same descent, and implicitly contrast themselves with neighboring peoples. In a number of connections we have occasion to insist on this character which is native to many societies.
In these circumstances, to imagine that a people, in this case the Aryas, called themselves “the hospitable ones” would run counter to all historical probability. It is not in this way that a people affirms its individuality vis-à-vis its neighbors, who are always presumed to be hostile. We have already seen (Book One, Chapter Seven) that the relationship of hospitality is not established either between individuals or between groups except after the conclusion of a pact under special circumstances. Each time a specific relation is established. It is thus inconceivable that a people should proclaim itself as “the hospitable ones” in general and towards everybody without distinction. We must always determine by precise contexts the original sense of institutional terms such as “hospitality,” which for us has only a moral or sentimental sense.
Without going into the details of the very numerous examples, the exegesis of which is sometimes difficult, we may stress certain features which help us to define the status of the arí or the arya.
The connotations of the word arí, which are sometimes favorable and sometimes unfavorable, do not affect the true sense of the word. It designates a man of the same people as the one who speaks about him. This man is never considered as the member of an enemy people, even if the singer is enraged with him. He is never confused with a barbarian. He takes part in all the cults, he receives gifts which the singer may envy him, but which put him on the same footing. He may be generous or avaricious, friendly or hostile—but it is always a personal hostility. At no time can we perceive that the arí belongs to a different ethnic group from the author of the hymn.
Further, the arí are often associated with the vaiśya, that is to say the members of the third social class, which confirms that the arí is not a stranger. There is more precise testimony to the social position of the arí in the complaint of the daughter-in-law of Indra (Rig Veda X, 28, 1): “All the other arí have come (to the sacrifice); only my father-in-law has not come.” Indra is thus counted among the arí of his daughter-in-law. If we took the expression in the most literal sense, we should conclude that the arí formed the other moiety in an exogamic society. Nothing contradicts this inference, and some facts seem to confirm it. In this way we could understand why the arí are sometimes in a relationship of friendship, sometimes of rivalry, and that they together form a social unit: the expression “all the arí (or ă̄rya)” often recurs in the Rig Veda; it is also known in the Avesta, so that it is an inherited item of Indo-Iranian phraseology.
We must also pay attention to the name and role of the god Aryaman, who belongs to the Indo-Iranian pantheon. This name is a compound of arya–man– ‘of the spirit of arya’. Now the god Aryaman in Vedic mythology establishes friendship and, more particularly, he is the god of marriages. For the Iranians, too, Aryaman is a friendly god, but in the different guise of a healer. As a noun, aryaman– in the Zoroastrian Gāthās designates the members of a religious confraternity. In the Persian proper name Aryarāmna ‘who gives peace to the arya’, we again find the communal sense of arya.
In summary, we can disentangle from the brief mentions and often fleeting allusions in the Vedic texts some constant features which enable us to form a probable idea of what the word meant: the arí or arya (we cannot always distinguish the two forms) form what was doubtless a privileged class of society, probably entering into the relation of exogamic moieties, and maintaining relationships of exchange and rivalry. The derivative ārya, which at first designated the descendants of the arí (or the arya), indicated that they belonged to the arí, and it soon came to serve as a common denominator for the tribes who recognized the same ancestors and practiced the same cults. These comprise at least some of the components of the notion of ārya, which among both the Indic people and the Iranians, marks the awakening of a national conscience.
It remains to determine what the stem of ari, arya– properly signifies, and to decide whether the form ari– belongs to the Indo-European vocabulary or whether it is limited to Indo-Iranian. Scholars have often suggested that ari may be connected with the prefix ari-, which in Sanskrit denotes a degree of excellence and may correspond to the Greek prefix ari– (ἀρι-), which also indicates excellence; and since this Greek prefix ari– probably connects up with the group of áristos ‘excellent, supreme’ this would suggest for ari-, arya– some such sense as “eminent, superior.” But these etymological connections are far from certain. In any case, to return to our point of departure, the idea of mutual behavior (whether friendly or hostile) is more strongly felt in the uses of ari-, arya– than any suggestion of eulogy. Only a more profound analysis based on new facts would permit us to make any pronouncement on the etymology.
[ back ] 1. Meillet, Etudes sur l’étymologie et le vocabulaire du vieux-slave, Paris, 1902-1905, p. 175.
[ back ] 2. This point is developed in an article contributed to a collection of Mélanges offered to C. Levi-Strauss.
[ back ] 3. Theses and antagonistic interpretations: on the one hand, P. Thieme, Der Fremdling im R̥gveda, 1938; Mitra und Aryaman, 1958; on the other, G. Dumézil, Le troisième souverain, 1949; L’idéologie tripartite des Indo-Européens, 1958, p. 108ff.