Use the following persistent identifier: http://nrs.harvard.edu/urn-3:hul.ebook:CHS_Benveniste.Indo-European_Language_and_Society.1973.
Chapter 3. The Free Man
Although the opposition “free/slave” is common to all Indo-European peoples, a common designation of the notion of “liberty” is unknown. The fact that this designation evolved along parallel lines in two groups of languages merely serves to bring out better the specific content of the notion.
In Latin and Greek the free man, *(e)leudheros, is positively defined by his membership of a “breed,” of a “stock”; proof of this, in Latin, is the designation of (well-born) children as liberi; to be born of good stock is to be free; it comes to the same thing.
In Germanic, the connection which is still felt, for instance, between German frei ‘free’ and Freund allows us to reconstitute a primitive notion of liberty as the belonging to a closed group of those who call one another “friends.” To his membership of this group—of breed or of friends—the individual owes not only his free status but also “his own self”: the derivatives of the term *swe, Gr. idiṓtēs ‘individual’, Latin suus ‘his’, but also Greek étēs, hetaîros ‘ally, companion’, Latin sodalis ‘companion, colleague’, show that the primitive *swe was the word for a social entity, each member of which realizes his “self” only in the “inter-self.”
The general framework of Indo-European society and the great divisions it comprises are already “institutions.” To bring greater precision to our study, we shall now investigate the fundamental notions which inform the structure of these institutions.
Each of the Indo-European societies is pervaded by a distinction founded on free or servile condition. One is born free or born a slave. In Rome we have the division between liberi and servi. In Greece, the free man, eleútheros (ἐλεύθερος), is opposed to doúlos (δούλος).
In Germany, according to Tacitus, society comprised nobiles, ingenui, liberti and servi. It is clear that nobiles and ingenui, with the distinction of nobility and birth, are the equivalent of the liberi; on the other hand, the servi form a group with the liberti, former servi. Thus the division of society, evidenced by these four terms, is much the same. In India, the ārya (the name by which the Indo-Iranians called themselves) are opposed to dāsa (slaves and foreigners).
Despite innovations of terminology the same institution is maintained. But we have at least one term common to two or more languages: Lat. līber/Gr. eleútheros. There is a perfect correspondence; the two terms can be superimposed and traced back to an ancient form *(e)leudheros, which is found in a third language, in Venetic.
There is in fact a Venetic goddess Louzera, the Latin equivalent of which would be Libera, the feminine consort of the god Liber, who is identified with Bacchus. Furthermore, we have a case form louzeroφos, interpreted as liberibus, with a root diphthong –ou-, which is accounted for by the ablaut alternation e/o, as in Faliscan loferta (= liberta) and Oscan Luvfreis (gen. sing.) (= Līberī), as contrasted with *(e)leud–heros, Lat. līber.
The etymological analysis brings to light in liber a complex of relationships. First and foremost, we must decide whether there is one word liber or several. For are the adjective liber and Liber, the name of a divinity, one and the same words? There are also liberi ‘children’, which is apparently something different again. What complicates the question in another way is that the root from which liber and eleútheros are made, that is, *leudh-, produces in Old Slav. ljudŭ, ‘the people’, ljudĭje ‘gens’; in Germanic, in OHG liut, OE leod, modern German Leute ‘people’. Finally, apart from these adjectives and nouns, the verbal root supplies in Gothic liudan ‘grow’; in Indo-Iranian, Skt. rudh-, Av. rud– ‘grow, develop’.
The relationship between these forms is easy to establish, but what are we to make of the variety of meanings? These are so peculiar that at first sight they seem irreconcilable. How can we explain by a root *leudh– ‘to grow, develop’ a collective term for “the people,” then the adjective “free,” and, locally in Latin, a divine name Liber and a noun liberi ‘children’?
We have here a fairly frequent model of the relationships to be studied: at one extremity of the chain (in the case of Rome), the term refers to institutions, whereas elsewhere it forms part of other structures and designates different things.
Let us begin with the simplest forms, the verbal ones: Gothic liudan means “increase, grow” and it is used of a plant which reaches fullness of growth. In fact this verb liudan also gives rise to laudi ‘figure’, and –lauþs in the compound jugga–lauþs, literally “of young stature”; sama–lauþs ‘of the same growth, equal’. Similarly, in Indo-Iranian we have Skt. rudh-, Av. rud-, raod– ‘grow’, and the Av. noun raodah– ‘growth, stature, figure’.
We now see how the image of accomplished growth, culminating in “stature” and the human figure, has produced elsewhere a collective notion such as “stock, breed,” or “growth group” to designate an ethnic group, the totality of those who have been born and grown up together. The social sense of a noun such as *leudho– favored the transition to the sense of “people” (as in Old Slavic ljudĭje ‘people’ and in Germanic leod ‘people’). From this noun *leudho– (or *leudhes-) it was easy to form the adjective *(e)leudhero– to designate those who belong to the same ethnic stock and enjoy the status of “free men.”
It thus appears that the notion of “liberty” was constituted from a socialized notion of “growth,” the growth of a social category, the development of a community. All those who issued from that “stock” are endowed with the quality of * (e)leudheros.
We can now return to liber and recognize the connection between the several different notions it designated. The god Liber and the adjective liber may coexist without the name of the god being an application of the adjective. Liber, like the Venetic Louzera, is the god of growth of vegetation, later specialized in the domain of viniculture.
Eleútheros, liber: the pair of words now illuminates the origin of the notion of “liberty.” In Latin, as in Greek, all the ideas which we connect with the word “free” appear from the earliest texts on: the word is used with reference to the free man in the city, and the man who is free of illness, of suffering (with the genitive). In Homer, eleútheron ē̂mar (ἐλεύθερον ἦμαρ) ‘the free day’ designates the day which is that of the free man, the state of being free, and it is opposed to doúlion ē̂mar (δούλιον ἦμαρ) ‘the day of slavery’.
We grasp the social origins of the concept of “free.” The first sense is not, as one would be tempted to imagine, “to be free of, rid of something”; it is that of belonging to an ethnic stock designated by a metaphor taken from vegetable growth. Such membership confers a privilege which a stranger and a slave will never possess.
Let us consider finally the term liberi ‘children’. It shows a double peculiarity: first, it is only used in the plural; further, and this is particularly important, it designates the children by age only, not by social condition. Nevertheless, liberi ‘children’ is nothing more than the plural of the adjective liber. It is explained by a very ancient formula which accompanied the celebration of marriage and which we find in legal texts and in Plautus. It describes the purpose of marriage. The man who gave his daughter in marriage addressed the future husband with the words liber(or)um quaesundum causa (or gratia) ‘to obtain legitimate children’. This formula recurs in Greece, where it is well established through the allusions of Attic orators, by a quotation of Menander, and various legal texts. The pronouncement is literally the same: epì paídōn gnēsíōn sporâ̄i (ἐπὶ παίδων γνησίων σπορᾷ) ‘to generate legitimate children’. If we keep to the proper sense of liber, we can translate the Latin formula literally as “to obtain free (beings)”; the aim of marriage is precisely to give to those who will be born the status of free persons by legalizing their birth. It is in this phrase, and only by implication, as an object of quaerere ‘to obtain’, that liberi has taken on the sense of “children”; by itself, the plural liberi is equivalent to paîdes gnḗsioi (παῖδες γνήσιοι) of the Greek formula. It was in the legal language that this development of meaning originated. There are numerous legal terms which passed into the common vocabulary of Latin. Thus liber, which corresponds to gnḗsios ‘of free birth’, ended up by forming an independent term, liberi ‘children’. Such is the formation of the notion of “liberty,” which we have been able to reconstruct by combining facts which at first sight seemed irreconcilable and by resuscitating a deep-lying conceptual image, that of “the stock.” 
The history of this term throws light on the formation of the concept of the “free man” in those languages where it is expressed by a derivative of *leudh-, such as the Gr. eleútheros, by showing the primary notion from which the concept evolved.
But the genesis of the corresponding term was different in other parts of the Indo-European world, where different terms have prevailed and remain in use today. What especially deserves our attention is the Germanic frei (German frei, English free). Thanks to favorable conditions for comparison, here, too, we can describe the genesis of a word which has become synonymous with Gr. eleútheros but which evolved, along quite different lines, notions relating to the individual and not to the society.
The dialect distribution of the forms in the present case seems complementary to that of *(e)leudheros in the sense that neither Greek nor Latin possess the etymological correspondent. Conversely, the languages which share with Germanic the word frei did not use derivatives of *leudh– to express the notion of “free.” In this way a lexical distribution came about between the dialects which permits us to compare two distinct processes which started from different points and finally converged.
The evolution which has produced frei ‘free’ in Germanic starts not from a verbal root, but from an Indo-European adjective which can be reconstructed as *priyos. This alone is worth noting: everything has evolved, from common Indo-European times onwards, from a nominal form, from an adjective, attested as such in Indo-Iranian, Slavic, Germanic and Celtic, which has remained productive. The second fact worth noting is the sense of *priyos. This term indicates a notion of an emotional character which appears clearly in Indo-Iranian, where Sanskrit priya, Av. frya– means “dear.” The adjective is in fact charged with the sentimental overtones which we attach to the word “dear,” i.e. it qualifies those for whom we feel affection. But in certain idiomatic usages it refers to personal possessions and even to parts of the body. It can be shown that this was the original sense: *priyos is the adjective for personal belongings, implying not a legal but an emotional connection with the “self” and always prone to take on a sentimental coloring. The result is that, according to the context, it can be translated sometimes by “his own” and sometimes by “dear, beloved.” This aspect of the notion is the one most apparent and which becomes most frequent: thus priya– in Vedic qualifies the beings most closely associated with the person and which are “close” to him in affection: the feminine priyā ‘dear’ was substantivized and became the name for “wife.” The personal sphere also occasionally comprises the relations between man and the gods, thus expressing a sort of “mutual belonging.” Vedic priya-, Av. frya– thus enter into the religious terminology.
On the basis of this ancient adjective, Slavic has coined a present denominative prijajǫ (Russ. prijaju) ‘to show oneself favorable, to show affection’, from which comes the agent noun prijateľ ‘friend’, known in all Slavic languages.
In Germanic, too, the sentimental value is apparent from Gothic onwards in the verb frijon ‘to love’ (translating Gr. agapân, phileîn) and in the abstract noun friaþwa ‘love’. The participle frijonds ‘friend’, OHG friunt ‘friend’ survives in this sense to our own days (German Freund, English friend).
But Gothic also possesses the adjective freis ‘free, eleútheros’ with the abstract noun frijei ‘liberty, eleuthería’, that is to say, the literal correspondent of the ancient *priyos, but with quite a different sense, that of “free.” It shares this function with Welsh rhydd ‘free’, which also goes back to *priyos. There is thus in Gothic a division between frijon ‘to love’ and freis ‘free’. This peculiar lexical situation suggests that the passage of freis to the sense of “free” was due in Gothic to Celtic influences, where *priyos signifies only “free.” Perhaps it is even a direct borrowing in Gothic from Celtic. This specialization is not attested anywhere except Celtic and Germanic.
The evolution from the Indo-European sense of “personal, dear” to that of “free” which appears in Celtic and Gothic may be explained by the exclusiveness of a social class. What was a personal qualification of a sentimental kind became a sign of mutual recognition which was exchanged between members of the class of the “well-born.” It is a tendency of closed sections of society to develop among those who belong to it the sense of closely belonging to the same group, and to evolve a distinctive vocabulary. The term which in its first form expressed an affectionate relationship between persons, *priyos, took on an institutional sense when it became the name for the members of a kind of class “friendly society” and later the denomination for a social status, that of “free” men. 
Finally, a last word for “free” is the ancient Iranian āzāta– (Persian āzād). It properly signifies “born of the stock,” the preverb ā– marking the descent towards and up to a present moment. It is always birth in a succession of generations which guarantees the condition of a “free man.”
The history of these terms imposes the conclusion that words for individual social status and class status are often connected with individual notions such as that of “birth,” or with terms for friendship, like those which are applied to each other by members of closed groups. These names mark them off from strangers, slaves and, in general, from those who are not “well-born.”
We must draw attention to a fact which is rarely commented upon: how closely connected with certain forms of society some of the terms are which define the individual in his personal status.
A whole group of words with different interconnections will serve to illustrate these relations, some of them directly, others in a more distant way. We shall first consider the Gr. adjective ídios (ἴδιος), which is connected with the notion of “private, what belongs to somebody,” as opposed to what is public or common to all. The origin of the term has been much discussed. It could not be solved until an Argive inscription was found (on Dorian territory) with the word whediestas (whεδιεστας), which was recognized as the local form of the classical term idiṓtēs (ἰδιώτης). This form whediestas is of great interest because of its orthography with wh– (going back to an original initial *sw-), as well as the vocalism e of the first syllable. It shows that the initial i– of ídios is an ancient e– that has been assimilated to the following –i-. In addition to this, the formation of whediestas does not accord exactly with that of idiṓtes. The Argive word belongs to a category of social terms in –estās, Ionic-Attic –estēs, like Gr. penéstēs ‘mercenary, domestic’ (in Thessaly). But the root is identical in the Argive whediestas and in Gr. idiótēs, and this is now reconstructed as *swed-. In two slightly different forms, we have here the Greek designation for “the individual, the private citizen,” as opposed to the public personage, the one who holds power or fulfils a public function. As so often, each of the Indo-European languages has used in its own way an inherited root and each has made its own specific derivatives. This is the case with the Greek term in question, for which no other languages offer a correspondent.
However, there exists a related form in the Latin adjective sodālis, a derivative in –ālis from a stem sod-, which can be traced back to *swed-. Between sodālis ‘companion, colleague’, especially “member of a religious college,” and the Greek whediestas, in spite of the difference in institution, there appears a common trait, that of a closed circle around the “private person,” or a closed professional group. This trait specifies it and separates it from the rest of society by conferring a special status. The characterization remains a social one; it takes its place among the words for classes and functions, as is shown respectively by the Greek formation in –estās and the Latin one in –ālis.
Let us now consider the radical element itself, *swed-, an enlarged form of the basic term *swe. This *swe, which is attested in a long series of different words, is a very important term of the Indo-European vocabulary. Its intrinsic meaning can be seen in isolation in a definite morphological category (see below on the pronoun). Its final e is fixed, constant, without vocalic alternation; it is therefore not the ending of an inflected term. We have here a vestige of an archaic state: *swe remains fixed also in compounds or derivatives.
Its final –e is found in a small group of other words which likewise testify to a very ancient linguistic stage and which survived as such in various parts of the vocabulary: e.g. *kwe, an enclitic meaning “and,” Gr. te (τε), Lat. que, Skt. –ca; the root, with another vocalic grade, is found in the stem of the relative interrogative *kwo-, Gr. po– (πο- : πότερος, πόσος) and in *kwi, Gr. ti, tis (τι, τις). But *kwe, with the fixed final ending e, has the form and function of a particle, and it is not susceptible to inflexion or alternation.
Another word which presents this final –e is the numeral *penkwe ‘five’, Gr. pénte (πέντε), Lat. quinque, Skt. panca, the endings of which, –te, –que, –ca, exactly reproduce the forms of the connective particle: Gr. –te, Lat. –que, Skt. –ca.
This word *swe has given rise to an adjective indicating “personal belonging”: Skt. sva-, Lat. suus, Gr. *swós (*σwός). We must note that *swos is not in Indo-European the pronoun of the third person singular, as might be supposed by the relation of Lat. suus to meus and tuus. We instinctively make suus the third term of the series. Just as we put I, you, he in the verbal inflexion, it seems normal to us to have the pronominal series my, your, his. The relationship of these forms was quite different in Indo-European: *swos is the reflexive and possessive pronoun equally applicable to all persons.
This is what we still see today in the Slavic languages: Russian has svoj for “(my, your, his, our, your, their) own.” Similarly, Gothic swes ‘own, personal’ was used with reference to any person whatsoever. Again, in Sanskrit sva– was used without distinction where, with us, the insertion of mine or yours would be necessary. This neutrality as regards the person reveals the fundamental sense of the word. 
It has already been noted above (Book Two, Chapter One) that *swe appears in the ancient compound *swe–sor ‘sister’ as well as in *swekrū– ‘mother-in-law’, *swekuro– ‘father-in-law’.
In this connection we may note a peculiar feature of the terms for kinship formed from *swe in Slavic, Baltic, and particularly in Germanic; in this group the terms derived from *swe refer to kinship by alliance and not to consanguineous kinship. This is a common feature of a whole group of terms: Russian svat ‘suitor’ and also “related by marriage” (for instance for the relationship between the husband’s father and the wife’s father); svojak (a derivative from svoj ‘own’) ‘brother-in-law’, svest’ (feminine) ‘wife’s sister’; Lithuanian sváinis, ‘wife’s brother, sister’s husband’, fem. svainé ‘wife’s sister, brother’s wife’; Old High German swîo, geswîo ‘brother-in-law, sister’s husband’. If we have in these derivatives survivals of an ancient lexical state of affairs, we can see how interesting they are for the interpretation of those fundamental words common to all Indo-European languages which seem to be composed with *swe, that is, “sister” (*swesor-) and “parents-in-law” (*swekrū-, etc.). It would mean that these terms connect those so designated with the other exogamic “moiety.” In fact, the sister belongs there potentially, and the mother-in-law does so in fact. Theoreticians, who might be prompted by the present study to reconsider the analysis of kinship in Indo-European societies, will be better able to assess the significance which is to be attached to this observation.
This *swe is likewise the stem of the Gr. word étēs (ἔτης) ‘kinsman, relation’ and hetaîros (ἑταῖρος) ‘companion’. These two words, which are used together in Homer and in competition with one another, are closely related in sense, although they differ in their suffixation. It would be necessary to study the passages in which the two words occur if we wanted to make an exact distinction between them. It seems, however, that hetaîros has a more precise signification: “companion,” “friend” in the exercise of some activity, in battle, but it is not properly speaking a term of kinship, while étēs designates “kinsmen” in general.
In étēs ‘kinsman’ and also, dialectally, “fellow citizen,” “private person,” the root *swe points to a connection with whediestas ‘private person’. In the two words the same fundamental notion is evident, a notion which we also detect in another semantic family in Greek, the perfect eíōtha (εἴωθα) ‘to be accustomed to’ and the noun éthos (ἔθος) ‘habit’. The verbal form and the nominal form particularize the notion of “habit” as a distinctive mark and manner of being individual.
We may thus identify *swe in several groups of Greek forms where it is specialized by distinctive affixes:
These few examples illuminate the relationships which connect the concept signified by the root *swe with a group of derivatives, all implying a bond of a social character of kinship or sentiment, such as companionship, alliance (by marriage) and friendship.
If we now take a comprehensive view of all the derivations based on the stem *swe, we observe that they divide along two conceptual lines. On the one hand *swe implies the membership of a group of “own people”; on the other it specializes the “self” in its individuality. The interest of such a notion is evident, both for general linguistics and for philosophy. The notion of “self,” of the reflexive, is crystallized here. It is this expression which a person uses to delimit himself as an individual and to refer to “himself.” But at the same time the subjectivity is expressed as a “belonging.” The notion of *swe is not limited to the person itself; from the beginning it implies a tight and closed group which encompasses the “self.”
All that is ascribable to *swe becomes *swos, Lat. suus ‘his’ (in the sense indicated above), and ownership proper is defined only within the group included within the limits of *swe. Thus, to return to the Greek terms, *swe explains at the same time ídios ‘peculiar to oneself’ and hetaîros, which implies a bond with an age group or a profession. The situation which has been reconstituted by this connection reproduces the proper sense of Indo-European *swe, which implies both distinctiveness from all else, the isolation of the “self,” the effort to separate oneself from everything which is not *swe, and also, within the exclusive circle thus marked off, the close relationship with those who form part of it. From this comes the double heritage, both idiṓtēs, the isolated member of society, and also the sodālis, the member of a closed fraternity.
This duality survived, as is revealed by the etymology, in the two forms se of Latin, which have become independent; the reflexive se, indicating “self,” and the separative se-, sed ‘but’, marking distinction and opposition.
We see here again (as in the case of liberty) that it is society and social institutions which furnish concepts which are apparently the most personal. In this great lexical complex made up of numerous subdivisions which has evolved from the term *swe, institutional values consort with those of personal self-reference, and these prepare the way, at a higher degree of abstraction, for the grammatical categories of “person.”
This double relationship is apparent in the historical facts; Sanskrit sva– signifies “his,” but in a technical sense which goes beyond mere personal possession. Sva– is applied to the person who forms part of the same tight group; this term plays an important role in legal provisions affecting property, inheritance or the succession to titles and honors. The corresponding term exists with the same technical meaning in Latin. In the Law of the Twelve Tables, there is a clause relating to inheritance: “if a man dies intestate, heres suus nec escit (= non sit), and if he has no heir who is a suus.” The expression heres suus is also an archaism, for suus, if it had only a possessive sense, would not be necessary. A heres who is a suus, this is what the provision intends: there is no transmission of property outside the sui, that is to say the closed group of immediate descendants; it remains within the group of collaterals.
We observe all kinds of developments which start from these connections. Gradually legal kinship, and the consciousness of self, the connections of confraternity and individuality, are constituted as autonomous concepts and develop groups of new terms. But the comparison and analysis of these lexical families reveal their initial unity and lay bare the social foundations of the “self” and the “inter-self.”
[ back ] 1. Cf. our article “Liber et liberi,” Revue des Etudes Latines, XIV, 1936, pp. 51-58.
[ back ] 2. A recent bibliography of the problem is given, with a different interpretation, by F. Metzger, Zeitschrift für vergleichende Sprachforschung, 79 (1965), p. 36ff.
[ back ] 3. We have no occasion here to study the formal relationship between the alternating stems *swe and *se. For a reconstruction of an older state of affairs we refer to Bulletin de la Société de Linguistique de Paris 50, 1954, p. 36ff. The stem *sw– is also relevant to the formations of the derivatives that figure in the present study.