¹ 'Sur l'emploi de l'auxiliaire ‘avoir’ pour marquer le passé', Mel. van Ginneken 85 ff.

My thanks are due to Hans Vogt and John Lyons for their helpful comments on a preliminary draft of this paper.

² ‘La construction passive du parfait transitif’, BSL 48.52 ff. (1952).

³ Discussions of this type often distinguish between a grammatical and a semantic subject. The former is formally definable and usually refers to the noun or pronoun which is in a particular (e.g. ‘nominative’) case and/or in concord with the verb or with a particular verbal element. The latter (also variously described as ‘real’ or ‘psychological’) generally remains undefined and intuitive, since situational or logical correlates such as ‘actor’ or ‘topic’ break down in a number of instances.

This double use of the term can lead to confusion. And since a grammatical category set up for one language cannot be identified as such with a grammatical category in another, the grammatical criterion is not valid for a cross-linguistic comparison. For such purposes it is the semantic criterion that is relevant, and ‘subject’ in the present study will therefore always imply semantic subject.

This admittedly poses theoretical problems of definition. But in the absence of a satisfactory solution there exists a formal correlate, such that the relative order of subject and object exponents in any one of the languages is preserved in translation into any other; and as Greenberg's investigations have shown, there is a virtual universal whereby (in sentences of neutral style and emphasis) a noun, pronoun, or noun phrase corresponding to the intuited semantic subject precedes any noun etc. corresponding to the semantic object (Universals of language 61; cf. also Jakobson, ibid. 212). By invoking this factor of order, we are of course introducing a form of grammatical criterion to fill out inadequacies of semantic definition—but one that has been shown empirically to have cross-linguistic relevance.

⁴ For further discussion see Benveniste, BSL 54. 57 ff. (1959), 55. 127 (1960).

⁵ Cf. Schwyzer-Debrunner, Griechische Grammatik 2.150.

⁶ RR ii.2.5 ‘antiqua fere formula utuntur, cum emptor dixit: Tanti sunt mi emptae (sc. oues)? et ille respondit: Sunt’.

⁷ Voegelin, Linguistic structures of native America 143 (1946).

⁸ Robins, The Yurok language 25, 50.

⁹ LSNA 202, 209.

¹⁰ For details cf. Trager, IJAL 20.173 ff. (1954).

¹¹ Lg. 28.368 ff., 376 (1952).

¹² LSNA 246.

¹³ Thus Swadesh, ‘South Greenlandic (Eskimo)‘, LSNA 30 ff.; Schulz-Lorentzen, A grammar of the West Greenland language.

¹⁴ Kleinschmidt, Grammatik der grönländischen Sprache.

¹⁵ Bourquin, Grammatik der Eskimo-Sprache … an der Labradorküste.

¹⁶ Cf. Thalbitzer, Handbook of American Indian languages 1032,1057 (BAE Bull. 40, 1911).

¹⁷ LSNA 44; cf. Bourquin, op.cit. 14.

¹⁸ Op.cit. 18; cf. Kleinschmidt, op.cit. 14, 69.

¹⁹ The relevant details apply equally to Abkhaz; cf. Deeters, ‘Der abchasische Sprachbau’, Nachr. von der Ges. der Wiss. zu Göttingen, Phil.-Hist. Kl. 1931, 289 ff.

²⁰ Cf. Allen, TPS 1956. 153.

²¹ a if it is the second element, na if it is later in the verbal complex.

²² r is the basic form, with d as a morphophonemic variant.

²³ It is true that in this as in many examples the ‘infix’ might be regarded as a second prefix. But since members of this system are frequently embedded between the elements of a compound verb (e.g. y-š ‘t'∂-l-xd ‘she lifted it’, from the verb š't'∂-x), the term ‘infix’ is appropriate.

²⁴ Cf. Genko, Abazinskij jazyk 97 ff.

²⁵ Cf. Deeters, op.cit. 297 f.

²⁶ Four cases are commonly attributed to Kabardian; but as Trubetzkoy notes (Mél. Bally 81), two of these are really postpositional formations. Cf. Jakovlev, Grammatika literaturnogo kabardino-čerkesskogo jazyka 112 f.

²⁷ Cf. Grammatika lit. kab.-čerk. jazyka 48 f. (Kabardino-balkarskij naučno-issledovatel'skij institut).

²⁸ Cf. Dumézil, Études oubykhs 14; Vogt, Dictionnaire de la langue oubykh 151.

²⁹ This number is agreed on (though with some difference as to their identity) by Dirr (Einf. in das Studium der kaukasischen Sprachen 238 f.) and Hjelmslev (La catégorie des cas 162). There are in addition a large number of ‘local cases’, of which the counts vary; Hjelmslev sets the total number of cases at 48.

³⁰ Cf. Žirkov, Lakskij jazyk: Fonetika i morfologia 138.

³¹ ‘Un aspect du problème actif-passif dans le verbe’, Grammaire et psychologie 128 ff. (numéro spécial du Journal de psychologie, 1950); cf. Deeters, Handbuch der Orientalistik 1.vii, 55, 59.

³² Esquisse d'une grammaire du géorgien moderne 116 = NTS x.6 (1938).

³³ ‘Sur la notion du sujet en géorgien’, Mél. van Ginneken 183.

³⁴ Einführung in die georgische Sprache 498.

³⁵ Vogt, Esquisse 48 = NTS ix.48 (1938) : ‘l'agent au passif s'exprime par l'instrumental s'il est un nom de chose, par postposition si c'est une personne.‘

³⁶ The ‘versions’ of the Georgian verb express particular nuances of semantic relationship between subject and object(s).

^36a A more obvious connexion, reminiscent of Old Persian, is seen in participial constructions of the type čemi dacerilia es cigni, lit. ‘my written-is this book’, i.e. ‘this book has been written by me’ (Vogt, Esquisse 254 f.).

³⁷ On these cf. Chanidzé, ‘Le sujet grammatical de quelques verbs intransitifs en géorgien’, BSL 58. 1 ff, (1963); Deeters, Handbuch der Orientalistik l.vii, 61.

³⁸ This applies even where the subject is not actually expressed by a nominal or pronominal form, but is implied by the verbal person marker; e.g. mimiceria cerili mis-t'vis ‘I have written him a letter’.

³⁹ E.g. Kellogg, Grammar of the Hindi language §412, §785.

⁴⁰ Cf. Allen, Acta ling. vi. 71 ff. (1950–1).

⁴¹ Cf. Allen, Indian linguistics 21.10 (1960).

⁴² ‘A problem of morpheme identity in a Rajasthani dialect’, (unpublished; cf. LSA Bulletin 27 .4 (1954).