Use the following persistent identifier: http://nrs.harvard.edu/urn-3:hul.ebook:CHS_Benveniste.Indo-European_Language_and_Society.1973.
Chapter 5. Kinship Resulting from Marriage
Except for the husband and wife, for whom no specific terms seem to have existed in Indo-European, the words in this field have a constant form and precise sense—but they are not amenable to analysis. They always designate the tie of kinship through a man—the husband’s mother and father, the husband’s brother, the husband’s sister, the brother’s wife and the husband’s brother’s wife. There is no linguistic fact which would permit us to affirm that *swekuros, the husband’s father, ever designated in a parallel way the wife’s father, that is to say, by the rule of exogamy, the maternal uncle.
In Indo-European, nomenclature of kinship resulting from marriage is opposed to that for consanguineous kinship. This is a distinction which can be verified in modern languages as well as in ancient ones. This kinship as a result of marriage is determined by the position of the wife in the bosom of the family into which she enters: all the same, the terms designating these new ties are subject to variations. At least some modern languages employ the same basic terms as for consanguineous kinship, but they are differentiated by lexical devices. Thus, in French, beau is used as a classifier of kinship by alliance: on the one hand we have père, mère, frère, soeur, fille, fils, and on the other beau–père, belle–mère, beau–frère, belle–soeur, belle–fille, beau–fils. The nouns are identical in both series. In English, too, the same terms serve in both cases, but are differentiated by the addition of in-law, e.g. father-in-law. Each of these two devices has its historical justification. In Old French beau– is often a courtesy term equivalent to gentil ‘kind’; beau–père is thus a polite designation which assimilates the father of the spouse to the father proper. The English father-in-law is more “legalistic”: the “father” is defined according to the “law,” that is to say, canon law. If the same terms are used, it is not because of a sentimental assimilation of the two kinships, but for reason of lexical economy and symmetry: the kinship by alliance employs the same nomenclature as the natural kinship does for connections of filiation (father, mother/son, daughter) and of fraternity (brother/sister). It is a specific classificatory kinship, which serves to define the respective connections of those who find themselves allied by the marriage of their own kin.
But these are modern developments. In ancient Indo-European, on the other hand, the two types of kinship are distinct. No less than consanguineous kinship, kinship by alliance has its own terminology.
To begin with, we find the words for “husband” and for “wife,” which we will consider in their Latin expressions, marītus and uxor.
Marītus is peculiar to Latin: as a matter of fact, there is no Indo-European word signifying “husband.” Sometimes the expression “master” was used, e.g. Skt. pati, Greek pósis (πόσις), without any special indication of the tie of conjugality; sometimes we find “the man,” Lat. vir, Gr. anḗr (ἀνήρ), whereas marītus designated the husband in his legal aspect.
The etymological analysis of marītus raises two distinct problems: that of the formation of the derivative, and that of the sense of the word.
If we consider it only as a Latin derivative, marītus can be interpreted without difficulty. It belongs to a class of well-established derivatives in –ītus parallel to those in –ātus, –ūtus, that is, to secondary formations in which the suffix –to– is added to a root ending in –ī-, –ā-, –ū-, etc.: armātus, cornūtus, aurītus, etc. In virtue of this formation, marītus should signify “provided with the possession of marī-.”
It remains to determine the sense of the root. This has been brought into connection with a group of terms, which from an early date were applied, with some formal variations, sometimes to the young woman, sometimes to the young man: notably Gr. meîrax (μεῖραξ) ‘(young) woman’, secondarily “boy,” meirákion (μειράκιον). Soon, from language to language, one or the other sense predominates. In Latin *mari– seems to have designated a girl of nubile age, and marītus thus signified “in possession of the young woman.”
The Indo-Iranian correspondent marya designates the young man, but with a special status: especially in his amorous relationships, as a suitor, as a gallant (Indra); in brief, a boy of nubile age. This is the usual sense in Indic. In Iranian, marya has taken on an unfavorable meaning: it is a young man who is too audacious, a young fiery warrior, a destroyer, and even a brigand. In fact, this sense is limited to the Avestan texts. Other texts show in Iranian itself the persistence of the ancient sense. Especially clear is the Pehlevi mērak, which signifies “young husband”; mērak, with the corresponding term for the young wife ziyānak, are familiar, affectionate terms. The evidence suggests that in the distant past an institutional value was attached to this term, that of the class of young warriors. That this was a very old word, is shown by the fact that maryanni, designating the warrior class, figures among Indo-Iranian terms which we encounter in the fifteenth century B.C. in the Mitanni texts, where the names of important gods like Indra, Mitra, and the Nāsatya also figure.
Latin and Greek, on the contrary, specialized the term in the sense of the “young (nubile) woman.” This is what made possible the creation of marītus in Latin, literally “provided with *marī-,” a term without known parallels.
To marītus corresponds uxor ‘spouse’, an ancient word of constant sense and limited to Latin. The etymology of uxor is far from clear: the proposal has been made to analyze it as *uk–sor, the second component being the name for the “feminine being,” which appears in *swe–sor “sister.” It would be tempting to assign a classificatory value to this term *sor, which would be identified in the word for “spouse” as well as in that for “sister.” As for the first term *uk-, this analysis links it with the root *euk– ‘to learn, to become used to’, represented by Skt. uc-, Slavic ukŭ ‘teaching’, and in particular by the Armenian verb usanim ‘I learn, I accustom myself’. Now, this verb usanim has been linked with the Armenian term amusin ‘husband, wife’, which, with the prefix am– ‘together’ would then literally mean “the partner with whom common life takes place.” The formation amusin would then explain the sense of *uk– in uxor. It follows from this that uxor, analyzed as *uk–sor, is “the habitual woman, the female being to which one is used.” It must be admitted that such a designation for the wife is far from natural. Besides, no derivative from this root *euk– indicates a relation between individuals or a social relation. What *euk– signifies is of an intellectual order: “to acquire by repeated use,” which leads on to “to learn,” and to “lesson, doctrine”; thus the Gothic bi–ūhts ‘who has the habit’, Slavic vyknǫti ‘learn’, and also Armenian usanim ‘learn’. It is, therefore, uncertain whether we can relate amusin ‘spouse’ (husband and wife) to usanim ‘learn’; the –us– ‘marriage tie’ which seems contained in amusin may be of a different origin. If we have to dissociate these two forms within Armenian itself, the parallelism with uxor disappears.
Another etymological interpretation of uxor leaves it within the vocabulary of kinship by comparing it to a term which, in Baltic, refers specially to the wife: Lith. uošvis ‘wife’s father’ (cf. Lith. uošve ‘mother-in-law’, a secondary feminine form), Lett. uôsvis. This Baltic form is a derivative in –vyas of the type of Skt. bhrātr̥–vya ‘son of the father’s brother’ or Latin patruus, Gr. patrōós (πατρωός); the suffix in question was, therefore, used to form terms of kinship. The prototype of the Lith. uošvis is *ouk(s)-vya-. It would be natural for this term to be applied to the “wife’s father,” if the root *ouk(s)- was already at a predialectal stage a word for the wife. The Latin form *uksor would then only comprise a suffix –or, the sense of “wife” already being given in the root *uks-. This explanation also remains hypothetical in as much as there is as yet no confirmation from a third language. The Ossetic ūs– ‘woman, wife’ cannot be adduced, though this has been proposed, because the dialect form vosae with its initial *w– shows that it had a different origin.
We must thus affirm the specific character of the Latin word uxor, the interpretation of which remains uncertain. It will already be clear that the words denoting kinship resulting from marriage have a double peculiarity in being on the one hand constant in form and precise in sense, but on the other hand, by reason of their very antiquity, difficult to analyze.
The father and mother of the husband are designated respectively by *swekuros and *swekrūs (feminine). The masculine *swekuros is represented by Skt. śváśura, Iranian xvasura, Arm. skesr–ayr, Lat. socer, Gr. hekurós (ἑκυρός), Gothic swaihra, Old Slavic svekrŭ, and, slightly altered, Lithuanian sesuras, Welsh chwegrwn; the feminine *swekrū is represented by Skt. śvaśrū, Arm. skesur, Lat. socrus, Gr. hekurá (ἑκυρά), Got. svaihro, Old Slav. swekry. These correspondences are perfect apart from some slight alterations. In Sanskrit, one finds irregularly śvaś– instead of *svaś-, due to a secondary assimilation, the initial sibilant being guaranteed by Iranian xva– (< *swe). Similar is Lithuanian šeš– for *seš-. The Armenian skesrayr ‘husband’s father’ is a compound (skesr–ayr), which designates the man (-ayr = Gr. anḗr), that is, the husband, of the mother-in-law; skesur ‘mother-in-law’ is the primary term. On the contrary, in Greek, the terms are symmetrical: the feminine has been refashioned from the masculine. In Gothic, too, there has been remodeling: the two terms swaihra (masc.) and swaihro (fem.) have been adapted to one another. By contrast, Latin has preserved the ancient connection between masculine and feminine: socer/socrus < *swekuros/*swekrūs, like the Sanskrit śvaśura-/śvaśrū-.
In the light of this picture, in which all the principal languages are represented, we must conclude that a masculine *swekuros was coupled with a feminine *swekrūs. This is a morphological oddity of which we have no other example. We know of no other opposition masculine/feminine which takes the form of an alternation *-kuro-/-krū-, with its double anomaly. There are no feminine forms in –ū– which could be constituted from a masculine one in –o-; normally we expect a feminine in –ā– or in –ī-. Moreover, the difference of gender does not involve and cannot explain the syllabic variation between *-kuro– and *-krū-.
But let us consider this feminine form by itself: *swekrūs would be anomalous if it was formed from the masculine, but it could be admitted as an autonomous form because there is a type of feminine in ū. It is seen, for instance, in Vedic vadhū– ‘newly married woman’. This raises the possibility that the primary term was the feminine *swekrū-, the masculine *swekuros being secondary. This hypothesis would explain the alterations which were produced in a number of languages. We postulate that *swekrū– is the inherited form, first because it is attested by the agreement of Indo-Iranian, Latin, Slavic and Armenian, and also because it could not have been formed from the masculine, since no similar example exists elsewhere. On the contrary, a number of indications suggest that the word for “father-in-law” has suffered refashioning. This is the case, as we have seen, in Armenian, where the “father-in-law” (of the wife) is called skesr–ayr ‘husband of the mother-in-law’. In Slavic, the masculine svekrŭ ‘father-in-law’ is a secondary form based on the feminine. The Gothic form swaihra ‘father-in-law’ also may have been constituted from an ancient *swekr-, that is, from the feminine stem, not from *swekur-.
But if we now believe that we can approximate better to historical truth by posing as the primary form the feminine *swekrū– ‘husband’s mother’, this still does not give us an explanation of the term. We are even further from it than if we had to proceed from the masculine *swekuro-. In effect, taken on its own, *swekuros has the appearance of a compound: the first term could be *swe-, the same as in the word for “sister.” The second term might be regarded as related to Gr. kúrios (κύριος), Skt. śūra ‘master, he who has authority’. The father-in-law would thus be considered and called master of the family. The flaw in this hypothesis is that a feminine *-krū– is inexplicable, the only justifiable feminine form being the –kura of Greek, but this is secondary. This reason alone would make the analysis improbable. Doubt is increased if we must consider *swekrū– as original. This primacy of the term for “mother-in-law” is, as a matter of fact, quite comprehensible: the husband’s mother is for the young wife more important than the husband’s father: the mother-in-law is the central personage of the household. But this does not explain the interconnection of the terms. The formal relations between *swekuro-and *swekrū– must therefore remain obscure.
The Indo-European word for the “brother-in-law” (the husband’s brother) is to be reconstructed as *daiwer on the basis of the following terms: Skt. devar-, Arm. taygr, Gr. dāé̄r (δαήρ), Lat. lēvir (with an l-, perhaps dialectal, for d-), Old Slav. děverŭ, Lith. dieverìs, Old High German zeihhur. The antiquity of the term is evident, but the true sense escapes us. No analysis of the form *daiwer– is possible; we cannot see an Indo-European root to which it could be related, although it shows a formation in –r-, which is close to other kinship terms.
The correlative term “sister-in-law” (husband’s sister) is less well represented: Gr. galóōs (γαλόως), Lat. glōs, Old Slav. zŭlŭva, Phrygian gélaros (γέλαρος)—to be read gélawos (γέλαwος)—glossed: ἀδελφοῦ γυνή ‘brother’s wife’. According to this last testimony, this would be a reciprocal term denoting both the husband’s sister and the brother’s wife. We must doubtless list here the Armenian word tal ‘husband’s sister’, where t– replaces an ancient c– (ts-) under the influence of taygr ‘husband’s brother’. Here Indo-Iranian is not represented; in spite of this there is a remarkable correspondence between Greek, Latin, Phrygian, and perhaps Armenian.
The last term defines the relationship between “brother’s wives”: it is the name given by the wife to the wives of her husband’s brothers, who live together according to patriarchal rule. This term is everywhere a survival: Skt. yātr̥-; a corresponding form yāθr– may be restored in Iranian on the basis of the Pashto yōr; Phrygian ianater– (ιανατερ-), Gk. einatéres (εἰνατέρες), Lat. ianitrīces, Old Slav. jętry, Lith. intè.
Consequently we can reconstruct *yen°ter, *yṇter-, where the formation in –ter is again evident. But we have no means of interpreting this root.
Everywhere we encounter firm designations with regular correspondences, but the etymological sense escapes us. Several of these terms were replaced at an early date by analytical ones, which were more transparent: “husband’s brother,” “wife’s sister,” etc. A curious situation is revealed if we compare these terms and the notions they express with those we have considered up to now.
If we take into account the fact of classificatory kinship it should theoretically follow that one and the same connection requires a double expression. If a man marries the daughter of his mother’s brother, his maternal uncle becomes his father-in-law. Is this situation attested in the terminology? It does not appear to be the case; we have no proof that *swekuros ever meant anything else than “father-in-law,” that is to say “the husband’s father,” and probably also the wife’s father in certain languages, like Sanskrit and Latin. But Greek has pentherós which, with a different suffix, corresponds to Skt. bandhu– ‘relation’; Armenian has aner ‘wife’s father’ and zokcanč “wife’s mother,” both terms without etymology; in short, Indo-European had no term for the relations of the wife. In fact we must remind ourselves that we have no Indo-European term which specifically designated the maternal uncle. As we have seen, he is called in Latin by a derivative of the word for grandfather; elsewhere the forms are different.
We can envisage two possible explanations. Either we reason with full theoretical rigor and suppose that *swekuros in prehistoric times did designate the maternal uncle, the mother’s brother, and that *swekrū– was the father’s sister, so that the historical sense was the result of a shift. This reconstruction is completely conjectural, and has no linguistic confirmation. Or else we decide that these terms never signified anything else than what they actually denote; they were always strictly applied to the relations established by the wife on her becoming a member of her new family. We should then have to assume that the patriarchal system triumphed at an early date and eliminated, in this series of terms, all trace of the double position which all allied persons occupied in the classificatory type of kinship.
Of the two hypotheses, preference should be given to the second. In any case, there are enough proofs of this patrilineal filiation in the terminology of consanguineous kinship to make it certain that the principle of interpretation itself will not be called into question by subsequent evidence.