Sex and Agreement: (Mis)matching Natural and Grammatical Gender in Greek1

1 Research for this paper was done while the author was an Associate of Harvard’s Center for Hellenic Studies in 2019. A preliminary version was presented at the Round Table on “Greek Language and Grammatical Gender” at Cankarjev dom in Ljubljana (January 14, 2020). The author wishes to thank the organizers, Jerneja Kavčič and Christina Manouilidou, for their invitation and hospitality. 2 Cf. Corbeil (2008: 80); Wartelle (1982: 66) translates ἄρρην as ‘mâle’ in reference to humans, i.e. children (Rhet. 1361a6), but as ‘masculin’ in reference to noun classes (Rhet. 1407b6-8), θηλύς as


INTRODUCTION
The Greek word γένος may refer to 'sex' as well as 'gender' . The concept of grammatical gender is obviously connected with the idea of biological sex, as emerges from the use of the adjectives ἄρρην 'male' and θηλύς 'female' to distinguish masculine and feminine nouns. According to Aristotle, it was Protagoras who introduced the concept of grammatical gender: (1) Πρωταγόρας τὰ γένη τῶν ὀνομάτων διῄρει, ἄρρενα καὶ θηλέα καὶ σκεύη.
Protagoras distinguished the classes of nouns, males and females and things.
(Arist., Rhet. 1407b) I prefer to translate ἄρρενα καὶ θηλέα here as 'male and female' , i.e. male and female beings, rather than 'masculine and feminine' (sc. noun classes), because of their juxtaposition with σκεύη 'things' . 2 The choice of terminology 1 Research for this paper was done while the author was an Associate of Harvard's Center for Hellenic Studies in 2019. A preliminary version was presented at the Round Table on "Greek Language and Grammatical Gender" at Cankarjev dom in Ljubljana (January 14, 2020). The author wishes to thank the organizers, Jerneja Kavčič and Christina Manouilidou, for their invitation and hospitality. 2 Cf. Corbeil (2008: 80); Wartelle (1982: 66) translates ἄρρην as 'mâle' in reference to humans, i.e. children (Rhet. 1361a6), but as 'masculin' in reference to noun classes (Rhet. 1407b6-8), θηλύς as suggests a division between animate beings, subdivided into male and female, on the one hand, and inanimate objects on the other. 3 Aristotle himself seems to prefer the term τὰ μεταξύ 'the in-between' (Poet. 1458a). 4 Dionysius Thrax is the first grammarian we know of to have used the terminology which has become accepted in the Greek and Roman grammatical tradition: (2) γένη μὲν οὖν εἰσι τρία· ἀρσενικόν, θηλυκόν, οὐδέτερον There are in fact three genders: masculine, feminine and neuter. (GG 1.1.24) Dionysius adds that others distinguish two additional genders: κοινόν τε καὶ ἐπίκοινον 'common and epicene' (GG 1.125). 5 Both can be used to refer to male as well as female beings, but whereas common nouns distinguish grammatical gender by agreement, epicene nouns do not. Examples of common nouns given by Dionysius include ὁ ~ ἡ ἵππος 'horse ~ mare' and ὁ ~ ἡ κύων 'dog ~ bitch'; examples of epicene nouns are restricted to animals and include ἡ χελιδών 'swallow' [m/f] and ὁ ἀετός 'eagle' [m/f] (GG 1.125).

EPICENE NOU NS
Aesop's fables unsurprisingly abound with such epicene nouns. The fable of the eagle and the fox, for instance, seems to be about two female animals and their young, but ὁ ἀετός being an epicine masculine noun (and one of the examples cited by Dionysius Thrax) as opposed to ἡ ἀλώπηξ, which is an epicine feminine noun, both trigger obligatory grammatical agreement patterns on pronouns and participles which have no relation with their biological sex: (3) ἀετὸς καὶ ἀλώπηξ φιλίαν πρὸς ἀλλήλους ποιησάμενοι πλησίον ἑαυτῶν οἰκεῖν διέγνωσαν … καὶ δὴ ὁ μὲν ἀναβὰς ἐπί τι περίμηκες δένδρον ἐνεοττοποιήσατο· ἡ δὲ εἰσελθοῦσα εἰς τὸν ὑποκείμενον θάμνον ἔτεκεν.  'féminin, de sexe ou de genre féminin ' (1982: 193), in reference to the same passages, and σκεῦος as 'mot (nom, adjectif, pronom) neutre ' (1982: 388). 3 Cf. Schmidhauser (2010: 501), Novokhatko (2020: 107). 4 Singular τὸ μεταξύ (Arist., Poet. 166b; Soph. el. 173b). 5 Dionysius' wording ἔνιοι δὲ προστιθέασι τούτοις ἄλλα δύο 'but some add to these two others' (GG 1.1.24) indicates that he was not the inventor of the traditional terminology. The sex of the eagle is undetermined in both versions, ἀετός being an epicine masculine noun (and one of the examples cited by Dionysius Thrax) and seemingly irrelevant for the purpose of the fable. The two words for 'tortoise' , ἡ χελώνη and ἡ χελύς, are both epicine feminine nouns and both are used alternately in the Homeric hymn to Mercurius to refer to the same mountain tortoise: χέλυς ὄρεσι ζώουσα 'a tortoise [f] ' (h.Merc. 44). The sex of the tortoise in the first version of the fable (4a) is therefore undetermined and, again, seemingly irrelevant. The agreement of the participle θεασαμένη and the pronoun αὐτή with χελώνη is, in other words, obligatory and purely grammatical. In the second version, however, the turtle is overtly marked as male by the agreement of the participle θεασάμενος and the pronoun αὐτός with χέλυς, which would have been ungrammatical, had it not been for the added adjective ἄρρην. One can only guess at the reason(s) why the author of this version thought it necessary to explicitly present the tortoise as a male-because he wants to "fly like an eagle" out of male vanity, male arrogance, male hybris or perhaps all of the above?

NATUR AL GENDER AND DECLENSION
In a well-known scene from Aristophanes' Clouds, Socrates is presented as having even more original, albeit quite radical solutions to the problem of common nouns in his education of Strepsiades on the topic of gender assignment and gender marking (Nub. 658-93). 6 Socrates is playing on the ambiguity on the ambiguity of the adjective ἄρρην, when he asks Strepsiades which four-legged animals are properly male / masculine (τῶν τετραπόδων ἅττ' 6 As for the source for the scene, Wackernagel (1928: 4), Corbeil (2008 and Willi (2003: 99) acknowledge Protagoras, Sommerstein (1982: 196) and Henderson (1998: 97 59 ) Prodicus. Dover mentions Protagoras in connection with "the genders of nouns", but refers to Prodicus in connection with the use of ὀρθῶς at Nub. 659 (1968: 182 To resolve the referential or, if you like, sexual ambiguity of the word, Socrates offers a radical solution to the problem (of which only he is apparently aware) and on the spot creates the feminine ἀλεκτρύαινα 'hen' , which he contrasts with the poetic masculine ἀλέκτωρ 'cock' (Nub. 666) to avoid the epicene ἀλεκτρυών. The otherwise unattested neologism ἀλεκτρύαινα is obviously formed on the analogy of other pairs referring to opposite sexes in the animal kingdom such as λέων 'lion' ~ λέαινα 'lioness' , δράκων 'snake' ~ δράκαινα 'shesnake' , λύκος 'wolf ' ~ λύκαινα 'she-wolf ' , σκύλαξ 'dog' ~ σκυλάκαινα 'bitch' . 8 By doing so, the Aristophanic Socrates presents himself as a proponent of the principle that nouns referring to animate beings belonging to different sexes ought to be differentiated by different endings. Aristophanes, to be sure, used ἀλεκτρυών as a "properly epicene" noun according to Athenaeus (9.374c), who quotes him to illustrate the fact that in fifth-century Attic this was common usage: 9 (6a) ᾠὸν μέγιστον τέτοκεν, ὡς ἀλεκτρυών She's laid a huge egg, like a cock. (Ar., fr. 193) (6b) πολλαὶ τῶν ἀλεκτρυόνων βίᾳ ὑπηνέμια τίκτουσιν ᾠὰ πολλάκις It happens that many [f] cocks [m/f] by necessity lay wind-eggs. (Ar.,fr. 194) 7 Ignoring the fact that fowls are not quadrupeds, as Wackernagel wittingly remarks (1928: 1). 8 On the productivity and extension of the suffix see Chantraine: "le suffixe -αινα a pris en grec un développement nouveau, il a servi à désigner des animaux, surtout des animaux méprisés" (1933: 107). The oldest examples of the formation include δέσποινα 'mistress' ~ δεσπότης 'master' (etymologically of a 'house') and θέαινα in the formulaic verse κέκλυτέ μοι πάντες τε θεοὶ πᾶσαί τε θέαιναι 'hear me, all gods and all goddesses ' (Il. 19.101,Od. 8.5) and variations thereupon (Il. 8.20,Od. 8.341). 9 τὸν δ᾿ ἀλεκτρυόνα … οἱ ἀρχαῖοι καὶ θηλυκῶς εἰρήκασι 'the ancients used the word ἀλεκτρυών also to refer to the hen' (Athen. 9.373e).
The translation of (6a) and (6b) is Henderson's, who undoubtedly intended to emphasize Socrates' ἀπορία with the common noun ἀλεκτρύων, but the agreement of πολλαί in (6b) leaves no doubt about the sex of the fowl (as if laying eggs was not enough to convince anyone). 10 The principle of correspondence between sex and gender is even more hilariously illustrated with Socrates' second rebuke of Strepsiades' lack of gender awareness. When the latter (correctly) uses the feminine article with a seconddeclension noun, i.c. τὴν κάρδοπον 'the trough' (Nub. 669), the former retorts that by doing so he is 'turning a feminine into a masculine noun' (ἄρρενα καλεῖς θήλειαν οὖσαν, Nub. 671). When Strepsiades asks him how on earth he managed to do that, Socrates replies: ὥσπερ γε καὶ Κλεώνυμον 'well, obviously, just like Cleonymus' (Nub. 673a), adding: ταὐτὸν δύναταί σοι κάρδοπος Κλεωνύμῳ 'clearly, κάρδοπος can be the same to you as Κλεώνυμος' (Nub. 674). This provokes an obscene wordplay on the part of Strepsiades (Janse forthcoming a), who asks how he should say the word correctly. Socrates' answer is again mind-boggling: This is a remarkable innovation: instead of replacing the feminine article with its masculine equivalent (τὸν κάρδοπον), Socrates moves the noun to the first declension (τὴν καρδόπην) to align the grammatical gender of the noun, indicated by the agreement of the article, with its dedicated inflectional class. Strepsiades is again unable to distinguish biological sex from grammatical gender and thus fails to understand why a trough should be 'female' (τὴν καρδόπην θήλειαν;Nub. 679a The point of Socrates' digression is that nouns belonging to the second declension should be masculine and those belonging to the first declension 10 Strepsiades, to be sure, learned his lesson well when he enlightens Phidippides not to use the epicene noun ἀλεκτρυών to refer to both sexes, but to call the masculine fowl ἀλέκτωρ and the feminine ἀλεκτρύαινα (850-1). 11 Strepsiades later uses his newly acquired knowledge to put off his first creditor: οὐκ ἂν ἀποδοίην οὐδ' ἂν ὀβολὸν οὐδενί | ὅστις καλέσειε κάρδοπον τὴν καρδόπην 'I wouldn't repay not even an obol to anyone | who calls the trough κάρδοπος' (Nub. 1250-1).
feminine-whether naturally (φύσει), conventionally (θέσει), οr both. 12 Socrates clearly treats Σωστράτη as a feminine noun referring to a female person, 13 but Strepsiades apparently understands Σωστράτη as a feminine noun referring to an effeminate male, hence his reassignment of Κλεώνυμος to the first declension. 14 Apart from male-female doublets in personal names belonging to the second and first declension respectively, there are of course many doublets in nouns, e.g. κόρος 'boy' ~ κόρη 'girl' , δοῦλος 'slave' [m] ~ δούλη 'slave' [f], θεός 'god' ~ θεά 'goddess' , etc.-not to mention the very common first and second-declension adjectives like καλός ~ καλή. It seems therefore quite reasonable for Socrates to fix, so to speak, the oddity of second-declension nouns triggering grammatical agremeent patterns on articles and adjectives usually reserved for first-declension nouns. As a matter of fact, many grammatically feminine second-declension nouns have been "repaired" in the course of time, either by imposing masculine agreement patterns on them or by moving them to the first declension (Jannaris 1897: 111-2). A well-known example, discussed by Wackernagel (1928: 3) in terms of analogy and more recently by Coker (2009: 40-2) in terms of category formation, is ἡ ἄσβολος 'soot' [f], which appears as ἡ ἀσβόλη in Semonides (fr. 7.61 West) but as ὁ ἄσβολος in Hipponax (fr. 138 West) according to Phrynichus (Praep. soph. 28.1 Borries), 15 both variants condemned by Photius. 16
(9) κἀγώ, παρθένος γὰρ ἔτ' ἦ κοὐκ ἐξῆν πώ μοι τεκεῖν | ἐξέθηκα, παῖς δ' ἑτέρα τις λαβοῦσ' ἀνείλετο and I, being still an unmarried maiden and not allowed to give birth, exposed [the child], and some other girl took it up and adopted it. (Ar., It is clear that the male (sic) chorus leader "speaks of himself metaphorically as an unmarried girl who had a baby and (in accordance with a common Greek custom) left it to die in the open country", in the words of Dover, who astutely adds that παρθένος is here "not a biological term, 'virgin' , but a social term, 'unmarried'" (1968: 167). 19 The original meaning is borne out by the juxtaposition of παρθένος and παῖς δ' ἑτέρα τις 'some other girl' (Nub. 531). The fact that the word can be combined with other nouns seems to indicate that it was originally an adjective, e.g. γυναῖκα | παρθένον (Hes., Theog. 513-4), θυγάτηρ παρθένος (Xen., Cyr. 4.6.9). 20 The meaning 'maiden' also underlies the use of παρθένος in connection with ἠίθεος in Homer The clearly archaic and poetic word ἠίθεος can be reconstructed as *ἠϝίθεϝος, which is presumably related to Proto-Indo-European *h 1 u̯ id h eu̯ -'unmarried' . It is thus cognate with Sanskrit vidhávā, Old Church Slavonic въдова vŭdova, Latin uidua, Old Irish fedh, Welsh gweddw, Gothic widuwō and Old English widuwe, all meaning 'widow' . Chantraine questions the traditional etymologie: "il est difficile de tirer le nom du jeune homme non marié de celui de la veuve" (1968-80: 408), but Beekes connects the meanings 'widowed ' and 'unmarried' (2010: 512) and concludes that it was originally an adjective (1992: 178). 22 It may be noted that Latin uidua is not only used to refer to a widow, 23 but also to an unmarried woman, notably in Tullia's urge to her husband Tarquinius Superbus, Rome's last king: se rectius uiduam et illum caelibem futurum fuisse contendere 'that it would have been juster for her to be unmarried and for 19 Cf. Sommerstein (1982: 187), Henderson (1998: 83), pace Sissa (1990: 86). 20 If Klingenschmitt's (1974) etymology is correct, παρθένος is originally a compound adjective, which would explain the fact that it is a second-declension adjective of two endings. 21 Cf. Hdt. 3.49.15-6. 22 A more detailed explanation is given in Beekes (1992). 23 As in Palinurus' warning to Phaedromus: dum abstineas nupta, uidua, uirgine … ama quidlubet 'as long as you stay away from a married woman, a widow, a virgin … love whatever you like' (Plaut., Curcullio 1.1.37).
him to be single' (Liv. 1.46.7). The juxtaposition of uidua with caelebs is very instructive, as the latter is also used to refer to a person who is single "through being unmarried, widowed, or divorced" (OLD, s.v.). Perhaps even more instructive is the following line from Propertius' tirade against Isis, where uidua is combined with puella: quidue tibi prodest uiduas dormire puellas? 'or what's in it for you that girls should sleep without men?' (Prop. 2.33.17). Finally, it should be noted that the adjective uiduus is also used to refer to men without women, e.g. iuuit uiduos rapta Sabina uiros 'the rape of the Sabine women aided the wifeless men' (Ov., Ars 1.102). Its Greek equivalent is also occasionally used in combination with feminine nouns referring to female persons, e.g. κόρη ᾔθεος 'unmarried girl' (Eup., fr. 362 Kassel-Austin = 332 Kock). 24 The Etymologicum Magnum has an interesting comment on Eupolis' use of ᾔθεος: (11) ᾔθεος· ὁ ἄπειρος γάμου νέος. σπανίως δὲ ἐπὶ παρθένου, ὡς παρ' Εὔπολι ᾔθεος: a youth inexperienced in sex; rarely in reference to a παρθένος, as in Eupolis.  This brings us back to παρθένος 'maiden' as a social term in the sense of 'unmarried girl' (cf. supra). The use of the phrase οὐκ ἐξῆν πώ μοι τεκεῖν by the chorus leader in (9) indicates that a respectable παρθένος should not have children, but if she did, she could still be called a παρθένος. The interpretation of παρθένος as 'virgin' constitutes therefore a secondary semantic narrowing, based on the premise that "the categories of virgins and unmarried women were ideally identical" (Ogden 1996: 107 140 ). For this reason it was assumed to be part of the αἰδώς of a παρθένος not to engage in sexual relations before marriage. This emerges clearly from the epic formula παρθένος αἰδοίη 'respectable maiden' in reference to Astyoche, who was still an unmarried girl when she was impregnated by Ares in her father's house (Il. 2.514). The same formula is used in reference to newly created Pandora by Hesiod (Theog. 571, Op. 70). In Sophocles' Trachiniae, Deianeira "contrasts her own anxieties as a married woman with the peace and freedom of a young girl before marriage" (Easterling 1982: 93), until she is called 'a wife instead of a maiden' (ἀντὶ παρθένου γυνὴ, Tr. 148). The latter is nevertheless described as living 'a carefree life in the midst of pleasures' (ἡδοναῖς ἄμοχθον βίον, Tr. 147). Such "pleasures" could include sex with a married man, because Heracles refers to Iole as 'the unmarried daughter of Eurytus' (τὴν Εὐρυτείαν παρθένον, Tr. 1220), who he has nevertheless slept with him (τοῖς ἐμοῖς πλευροῖς ὁμοῦ κλιθεῖσαν, Tr. 1225-6). 25 The idea that a maiden should ideally remain a virgin until she becomes a wedded wife (γυνή) gave rise to the semantic narrowing of παρθένος. 26 Compare, for instance, the definition of γυνή and παρθένος by Ptolemy of Ascalon: (12) γυνὴ παρθένου διαφέρει· γυνὴ μὲν γὰρ καλεῖται κυρίως ἡ ἤδη ἀνδρὸς πεῖραν εἰληφυῖα, παρθένος δὲ ἡ μήπω μυηθεῖσά ποτε ἀνδρός γυνή is different from παρθένος; γυνή is generally the word for a woman who has had sexual experience with a man, παρθένος for a woman who has not yet been initiated by a man. (Ptol. 61 Palmieri) Pollux' definition of the verbs διακορεύω and διαπαρθενεύω, both meaning 'deflower' , implies the idea of virginity as well: (13) τὸ δὲ τῆς παρθένου παρθενίαν ἀφελέσθαι To take away a maiden's virginity. (Poll.,Onom. 3.42 Bethe) In the Judeo-Christian context, it is of course the virgin birth of Jesus that gave rise to the generalization of the sense 'virgin' . According to the Gospel of Luke, Mary is described as παρθένον ἐμνηστευμένην ἀνδρί 'a maiden / virgin engaged to a man' (Lc. 1.27). When the angel Gabriel announces that she will get pregnant, she asks how this could possibly be, since she does not 'know a man' , i.e. carnally (ἄνδρα οὐ γινώσκω, Lc. 1.34). 27 Mary's fiancé Joseph is of course, technically speaking, a man, but in Matthew's version of the story it is made clear that 'he took her as his wife and did not get to know her [carnally] until she had borne a son' (παρέλαβεν τὴν γυναῖκα αὑτοῦ καὶ οὐ ἐγίνωσκεν αὐτήν ἕως οὗ ἔτεκεν υἱόν, Mt. 1.25). John Chrysostom is therefore justified to ask the question that must have been on many people's lips: (14a) πῶς τίκτει ἡ Παρθένος καὶ μένει παρθένος; How is it possible that the Virgin gives birth and remains a virgin? (Hom. in Mt. 4.6 Field) He could and should perhaps also have asked: 28 26 For a very thorough discussion of the Greek concept of "virginity" see Sissa (1990). 27 Compare the description of Isaac's future wife Rebecca: παρθένος ἦν, ἀνὴρ οὐκ ἔγνω αὐτήν (Gen. 24.16), where παρθένος translates the Hebrew bətūlāh. 28 Clement of Alexandria gave of course the only possible answer: μία δὲ μόνη γίνεται μήτηρ παρθένος 'only one woman becomes a virgin mother' (Paed. 1.6.42.1). A longer discussion is given by Gregory of Nyssa (Or. dom. = PG 1136.15 Migne).
(14b) πῶς γαμεῖται ἡ Παρθένος καὶ μένει παρθένος; How is it possible that the Virgin gets married and remains a virgin?
Even though the mystery surrounding Mary's virginity remained, there was no doubt about her sex nor about her parental or, indeed, her marital status. It is therefore surprising that παρθένος remained a second-declension noun in the vast majority of the early Christian writers. Coker invokes "its religious significance" (2009: 51) to explain the overwhelming frequency of the second-declension noun (2009: 49, tab. 6) as opposed to its meagerly attested first-declension alternative. Coker found nine dated examples of παρθένα instead of παρθένος in the TLG, six plural and three singular. The plural examples obviously do not refer to the Virgin Mary, a rather important fact which has escaped Coker's attention, but the (two, not three) singular examples do and this is of course noteworthy. The first example is taken from the Catena on the Epistle to the Hebrews and is very remarkable, as both the secondand the first-declension noun are used in the same text, which is dated to the fifth (!) century: (15a) γέγονεν υἱὸς Δαυίδ, σῶμα λαβὼν ἐκ τῆς ἁγίας παρθένου He was born a son of David, receiving his body from the Holy Virgin. The second example is found in the Late Byzantine Etymologicum Gudianum, where the legal status of children is discussed and παρθένιος is one of the terms to refer to illegitimate children: (16) παρθένιος δὲ ὁ ἐκ τῆς παρθένας ἔτι νομιζομένης γεννώμενος παρθένιος refers to the son born from a woman who is considered to be a virgin (?) (EG 410.34 Sturz) In Modern Greek, παρθένος has become a masculine second-declension noun used to refer to male virgins, 29 as opposed to the feminine noun παρθένα used to refer to a female virgin, including the Virgin Mary, e.g. in the invocation Παναγία μου Παρθένα or more colloquially, with a hypocristic term of endearment, Παναγίτσα μου Παρθένα-but the old epicine form continues to 29 The masculine παρθένος was already used in the New Testament book of Revelation to refer to men 'who were not defiled [sic] by women' (οἳ μετὰ γυναικῶν οὐκ ἐμολύνθησαν, Apoc. 14.4).
be used as well, though not in combination with a hypocoristic: *Παναγίτσα / Παναγία μου Παρθένε.
Some of the words listed by Ptolemy have feminine doublets which are derived from the same stem: παιδίσκος ~ παιδίσκη, νεανίσκος ~ νεανίσκη, μειρακίσκος ~ μειρακίσκη, 31 νεανίας ~ νεᾶνις, πρεσβύτης ~ πρεσβῦτις. 32 The word ἔφηβος, originally a second-declension adjective of two endings referring to the age class of ἥβη 'adolescence' , 33 hence theoretically applicable to adolescent boys and girls alike, 34 came to be used in fourth-century Athens as a legal term for boys who entered a two-year period of military training in their eigteenth year (Arist., Ath. 42). 35 In reference to adolescent girls the now common noun ἔφηβος is found from the sixth century onwards, and again in 31 On the positive and negative connotations of diminutive nouns in -ίσκος / -ίσκη referring to persons see Chantraine (1933: 408-9). 32 The details of the relationship between the obvious cognates γέρων and γραῦς / γραῖα are disputed, cf. Chantraine (1968-80: 235), Beekes (2010: 285). 33 Compare the phrase ἐφ' ἥβης (Ar., Eq. 524). 34 Compare the expression ἐς ἥβην ἦλθεν ὡραίαν γάμων 'she came to the marriagable age' (Eur., Hel. 12). 35 For a recent assessment of the Athenian ephebeia in the fourth century see Friend (2019). a legal context. In his paraphrase of the Justinian Code, Theophilus Antecessor, for instance, mentions οἱ ἄρρενες ἔτι δὲ καὶ αἱ θήλειαι ἔφηβοι 'the male and also the female adolescents' who are under the guardianship of a curator (κουρατωρεύονται) until they are old enough (at the age of twenty-five) to manage their property (Par. inst. 1.23.7-10). In the ninth-century successor to the Justinian Code, the so-called Basilika, ἔφηβος is used in combination with παρθένος (Bas. 2.2.12), θυγάτηρ (39.1.41) and κόραι (Scholia in Bas. I-XI 60.37.78.3). In Modern Greek, έφηβος is still being used as a common noun in high-register scientific jargon, but colloquially ο έφηβος now has a feminine counterpart: η έφηβη.
In many cases, however, παῖς is lexically opposed to its female counterpart, as in Eumaeus' account of the fate of Odysseus' parents  It is clear that παίδων is here used generically in reference to both Odysseus and Ctimene, 37 the latter being identified as θυγατέρι 'daughter' (15.364) as opposed to Odysseus, who is twice referred to as παιδός 'son ' (355, 358).
(Hom., Οd. 10.529-30) The second example from Homer's Iliad is a variant of the first: 37 It may be noted that the superlative ὁπλοτάτην instead of the metrically equivalent 'binary' comparative ὁπλοτέρην suggests that Laertes and Anticlea had more than two children. 38 Cf. Chantraine (1968-80: 473), Beekes (2010: 606 Liddell and Scott's remark that "Hom. uses ἰφθίμη of women; but ἴφθιμοι ψυχαί, κεφαλαί, speaking of men" (LSJ, s.v. ἴφθιμος) is echoed by Montanari: "Hom. -η referring to women; -ος with fem. nouns speaking of men" (2015: 995). It is tempting to accept this explanation for an apparent mismatch in natural and grammatical gender agreement, but one is left wondering why πολλάς [f] should be left out of the game, when πολλούς [f] would have been a viable and metrically equivalent alternative. Alternatively, it has been argued that ἴφθιμος is a second-declension adjective of two endings, except "bei Frauen" (Schwyzer 1950: 32).
Returning to the use of παῖς in reference to sons, it is clear that the plural may be used to refer to male and not to female children, as when Hector is met with "the womenfolk at large" (Kirk 1990: 155) at the Scaean gates: Here, as in the case of (19), the daughters are referred to by θύγατρας, the sons by παῖδας, but the identification of the latter can only be deduced from the context: the men return from the battlefield and the women are anxious to know if they are still alive. Shortly thereafter the sleeping quarters of Priam's children in his palace are described: Again the identification of παῖδες as 'sons' is made possible by their conjunction with their wives and the mention of Priam's daughters in the following line (κουράων δέ, Il. 6.247). Herodotus relates how the Hyperborean maidens (referred to as κόρας at 4.33 and παρθένοισι at 4.34) who had come to Delos to bring offering but had died there, were honoured by the Delians: κείρονται καὶ αἱ κόραι καὶ οἱ παῖδες οἱ Δηλίων 'both the girls and the boys cut their hair' , sc. in honour of the maidens (4.34).
Finally, there is of course the possibility of signalling the sex of the child by making articles or pronouns agree with the noun, as in Menander's Epitrepontes, where one of the girls (κόραις, Epit. 477) Habrotonon was invited to play for at the Tauropolia is later referred to as τὴν παῖδα (Epit. 480), ἐλευθέρα[ς | παιδός 'of a freeborn mother' (Epit. 495-6). Smicrines' daughter is called παῖδ' ἐπίγαμον 'marriagable girl' (Epit. 1115) and referred back to by the demonstrative pronoun ταύτην (Epit. 1119). 40
The diminutives of παῖς are either male (παιδίσκος) or female (παιδίσκη), but the most frequently used are neuter: τὸ παιδίον / τὸ παιδάριον. Looking at the respective positions of the neuter diminutives παιδίον and παιδάριον and the masculine nouns παιδίσκος and παῖς in Ptolemy of Ascalon's division of age classes (18), one might be inclined to look for a correlation between grammatical and natural gender, but a παῖς is generally not deemed old enough to be able to engage in sex-as opposed to a μειράκιον, who is considered to be young enough to still go to school according to  and old enough to have sexual relationships . Although the sex of a παιδίον does not seem to matter a lot, it is sometimes explicitly identified, e.g. θηλύ παιδίον (Plut., Pomp. 53.4) versus ἄρρην παιδίον (Ar., Lys. 748b). 41 There are many cases in which παῖς and παιδίον are used interchangeably to refer to the same child, e.g. τῷ ἂν οἴκῃ τῶν ἀνδρῶν τὸ παιδίον, τούτου παῖς νομίζεται 'to whom of the men the παιδίον resembles, the παῖς is adjudged to be his ' (Hdt. 4.180). 42 Aesop's fable about the boy who went hunting for grasshoppers begins with παιδίον and ends with ὁ παῖς (9b Hausrath-Hunger). Socrates discusses Protagoras' principle τὸ πάντων μέτρων 'the measure of all things' in reference to a παιδίον who is immediately thereafter referred to as τοῦ παιδός (Plat., Theaet. 168d).
There is, however, a very interesting and remarkable case of a mismatch between the grammatical and the natural gender of a baby in Menander's Epitrepontes. The usual words to refer to the baby are παιδάριον (Epit. 245,464,473,646,986) and παιδίον (Epit. 266, 268, 269, 295, 302, 311, 354, 355, 403, 448, 533, 539, 569, 864, 896, 956, 1131). 43 Once the baby is addressed as ὦ φίλτατον τέκνον (Epit. 856). On three occasions, however, it is referred to as παῖς and identified as a boy. When Syrus reveals to Smicrines that the 41 In reference to the latter, Sommerstein suspects that "there may be a play on sklēros 'hard' which, in later Greek at any rate, could also mean 'tough, virile'" (1982: 196 The baby is anaphorically referred to by the demonstrative pronouns οὑτοσί and οὗτος. 44 The use of the masculine οὗτoς instead of the neuter τοῦτο presents the infant as a young man who has the authority to claim the jewelry for himself. In other words, Syrus lets the baby speak on his own behalf, even though he identifies himself as its legal guardian (κύριος, Epit. 306). He then asks whether the gold trinkets should be kept τῷ παιδίῳ … ἕως ἂν ἐκτραφῇ 'for the child … until he is grown up' (Epit. 311), confirming its status as an infant. The demonstrative pronoun now used to refer back to the baby is not the masculine οὗτος, but the neuter τοῦτο (Epit. 314). Then, however, Syrus says the following: By using the masculine ο[ὑτο]σὶ ὁ πα]ῖς, Syrus is again presenting the baby as a young adult freeman who has the right to self-determination. Finally, παῖς is used in the phrase χρήματ' … ὀρφανοῦ | παι]δός 'the possessions … of an orphan boy' (Epit. 397-8), where the masculine noun is also used to emphasize the legal rights of the boy once he is an adult. 45 I would like to conclude with a brief discussion of the use of μειράκιον. In (Pseudo) Hippocrates' division of age classes (17), μειράκιον is used to refer to an adolescent boy between fourteen and twenty-one years, i.e. between παῖς and νεανίσκος, the latter being a full-grown, but still young, man. Ascalon, however, distinguishes μειράκιον from μεῖραξ in his division (18), which is remarkable as the same Ptolemy elsewhere distinguishes the two in the following way: (27) μειράκιον καὶ μεῖραξ διαφέρει· μειράκιον μὲν λέγεται ὁ ἄρσην, μεῖραξ δὲ ἡ θήλεια Τhere is a difference between μειράκιον and μεῖραξ: the male is called μειράκιον, the female μεῖραξ. (Ptol. 94 Palmieri) Moeris gives the following specification: Given the obvious relationship between the two words, it seems surprising that the diminutive should be used to refer to male youths, whereas the base form from which it is derived is used to refer to female youths. Etymologically, μεῖραξ is related to Sanskrit márya-'young man, lover' and maryaká-'small man' . 46 The latter is a formation independent of μεῖραξ, but the former suggest that μεῖραξ itself was derived from an unattested *μεῖρος, which would go back to Proto-Indo-European *mer-i̯ o-'young (girl or man)' (Beekes 2010: 921). Chantraine (1933: 379) suggests that nouns in -αξ may have been originally adjectives, e.g. μύλος 'mill' → μύλαξ 'millstone' , λίθος 'stone' → λίθαξ 'stony' as in λίθακι ποτὶ πέτρῃ 'against the stony rock' (Hom., Od. 5.415). Herodian says that μεῖραξ, -ᾰκος is feminine by analogy with other words in -αξ with a short suffix vowel such as ἡ κλῖμαξ, -ᾰκος 'ladder' , ἡ πῖδαξ, -ᾰκος 'spring' as opposed to masculine nouns with a long suffix vowel such as ὁ Φαίαξ, -ᾱκος 'Phaeacean' , ὁ θώραξ, -ᾱκος 'breast' (Hdn. GG 3.2.631). However, animate nouns in -αξ are often common nouns, e.g. σκύλαξ 'puppy' , δέλφαξ 'swine' , σπάλαξ 'mole rat' , so it is not inconceivable that μεῖραξ was originally a common noun as well. This would imply that the masculine use of μεῖραξ in "later writers" (LSJ) is not necessarily an innovation or an extension. 47 The use of μεῖραξ to refer to a male youth is found in the story of the seven Maccabean martyrs who were one by one tortured and killed by Antiochus 46 Other cognates have been suggested, but rejected by Chantraine (1968-80: 678) and Beekes (2010: 921-2 IV Epiphanes. 48 In the first version of the story, the third oldest is referred to as νεανίσκος (2 Macc. 7.12), the seventh and youngest as νεανίας (7.25, 7.30) and μειράκιον (7.25). In the second version, they are collectively called μειρακίσκοι (4 Macc. 8.1), μειράκια (8.14, 14.4), νεανίαι (8.5, 8.27, 14.9), νεανίσκοι (14.12) and even ἄνδρες (14.11), but also μείρακες (14.8) and οἱ ἱεροὶ μείρακες (14.6). 49 It is worthy of note that the Greek of 2 and 4 Maccabees is considered "literary and Atticistic" by Thackeray (1909: 13). 50 As a matter of fact, the distinction between μειράκιον / μειρακίσκος on the one hand and νεανίσκος / νεανίας on the other is as spurious as in other cases quoted earlier in reference to Ptolemy's life cycle (18). Leaving aside μειρακίσκος and νεανίας, it is interesting to observe that both μειράκιον and νεανίσκος can be used to refer to "the junior partner in homosexual eros" (Dover 1989: 85). In Plato's Charmides, Socrates says of the eponymous youth: (29) οὐ γάρ τι φαῦλος οὐδὲ τότε ἦν ἔτι παῖς ὤν, νῦν δ᾿ οἶμαί που εὖ μάλα ἂν ἤδη μειράκιον εἴη When Charmides enters the room, followed by a host of other lovers (πολλοὶ δὲ δὴ ἄλλοι ἐρασταί, 154c), Socrates consistently refers to him with masculine pronouns (ἐκεῖνος, 154b; αὐτόν, 154d), wherupon Chaerephon asks him: (31) τί σοι φαίνεται ὁ νεανίσκος; What do you think of the νεανίσκος [m]? (Plat., Charm. 154d) It appears that a sexually active μειράκιον can not only trigger male attention but masculine agreement patterns as well, despite the neuter gender of the noun. Nεανίσκος thus fits the natural gender better than μειράκιον.
48 Antiochus IV was the first of the Seleucids to persecute Jews, which resulted in the Maccabean revolt (167-160 BC The masculine gender of the demonstrative pronoun οὗτος may have been triggered by that of ἐκεῖνος, which refers back to Κριτοβούλου, which is of course a masculine proper name, but it may equally well have been triggerd by the fact that Cleinias is portrayed as being ahead of his age. He is nevertheless still refered to as τὸ μειράκιον by Socrates in his description of the same seating plan in which Cleinias was first identified by Crito (273b). Socrates agrees with Crito that Cleinias is well developed for his age (ὃν σὺ φῂς πολὺ ἐπιδεδωκέναι, 273a) and goes on to say that he was followed by a host of lovers (ἐρασταὶ πάνυ πολλοί, 273a), just as Charmides was described in his eponymous dialogue. In other words, the context is again erotically charged.
In the first eristic scene (272d-277c), Cleinias is first referred to as τουτονὶ τὸν νεανίσκον and immediately thereafter as τῷ μειρακίῳ τούτῳ (275a). The context is no longer erotically charged, as Socrates' purpose is to have Euthydemus and Dionysiodorus persuade Cleinias 'to ensue wisdom and practise virtue' (ὡς χρὴ φιλοσοφεῖν καὶ ἀρετῆς ἐπιμελεῖσθαι, 275b). He is twice characterized by Socrates as a νέος who is by his very nature susceptible to corruption (οἷον εἰκὸς περὶ νέῳ, 275b). He urges the two sophists to make trial τοῦ μειρακίου (275b) and they agree provided ὁ νεανίσκος (275c) is willing to answer their questions. Socrates continues his account as follows: (33)  Though a neuter noun, μειράκιον triggers masculine agreement patterns on the participle ἀπορήσας and the pronoun αὐτόν, which in its turns triggers masculine agreement on the particple τεθορυβημένον. It seems as if the youth is considered to be a (young) man of reason who is able to refute the sophists despite his ἀπορία, as is clear from Socrates' reassurement: (34) θάρρει … καὶ ἀπόκριναι ἀνδρείως, ὁπότερά σοι φαίνεται Do not worry … and answer like a man, whatever you think it is. (Plat., It is tempting to explain to masculine agreement pattern in (33) by the subsequent use of ἀνδρείως in (34), which Socrates apparently uses to convince Cleinias that he is a (young) man of independent thought. Dionysiodorus, however, is convinced that τὸ μειράκιον (275e) will be confuted, no matter what his answer will be, and Socrates knows he is unable to advise τῷ μειρακίῳ (276a), who continues to be referred to as τὸ μειράκιον in the ensuing interrogation (276b-d ter; 277b).
At the beginning of the first protreptic scene (227d-282e), as Euthydemus is about to press τὸν νεανίσκον (277d) for the third fall (πάλαισμα, as in a wrestling game), Socrates continues his account as follows: (35) καὶ ἐγὼ γνοὺς βαπτιζόμενον τὸ μειράκιον, βουλόμενος ἀναπαῦσαι αὐτό … παραμυθούμενος εἶπον And I, perceiving that the μειράκιον [n] was going under and wanting to give it [n] some breathing space … encouraged him with these words.
(Plat., Euthyd. 277d) All of a sudden, Cleinias is presented as a helpless little boy who is "getting into deep water" (LSJ) and this time τὸ μειράκιον triggers neuter agreement patterns on the participle βαπτιζόμενον, here of course indistinguishable from its masculine equivalent, and the anaphoric pronoun αὐτό, as opposed to αὐτόν at 275d (33). 52 The idea that Cleinias is too young to be able to tackle questions of such magnitude is later explicitly stated by Socrates, when he explains to the bewildered Cleinias that good fortune is not the greatest of all good things (τὸ μέγιστον τῶν ἀγαθῶν, 279c): (36) ἡ σοφία δήπου … εὐτυχία ἐστί· τοῦτο δὲ κἂν παῖς γνοίη Wisdom surely … is good fortune; even a child would see that.
(Plat., Euth. 279d) The particle δήπου combines "the certainty of δή" with "the doubtfulness of που", but "often the doubt is only assumed μετ' εἰρωνίας" (Dover 1954: 267). 53 That this is certainly the case here appears from Socrates' subsequent comment: (37) καὶ ὃς ἐθαύμασεν· οὕτως ἔτι νέος τε καὶ εὐήθης ἐστί And he wondered at this; he is still so young and ignorant. (Plat.,Euth. 279d) At the end of the first protreptic scene, Socrates urges Euthydemus and Dionysiodorus again to show Cleinias how "to ensue wisdom and practise virtue": (38)  In (38), τὸ μειράκιον triggers masculine agreement patterns on the anaphoric pronoun αὐτόν, as opposed to αὐτό at 277d (35), and on the participle λαβόντα, which suggests that Socrates is now treating Cleinias again as being compos mentis in that he assumes him to be capable of acquiring ἐπιστήμη to become a 'good man' (ἀγαθὸν ἄνδρα). It seems as if the use of τόνδε τὸν νεανίσκον in the second part of Socrates' statement is intended to suggest that he is actually a boy on the brink of manhood.
Crito is equally impressed upon hearing Socrates' account of this: Clearly, Crito could not believe that a μειράκιον would be able to speak in such a clear and sensible way. The masculine agreement on the anaphoric pronoun αὐτόν again indicates that Crito considers Cleinias to have grown out of the age class of μειράκιον and to be no longer in need of education.
An even more remarkable shift in grammatical gender agreement appears in Plato's Protagoras, when Agathon is introduced as follows: (41)  In this passage, Agathon is presented as a relatively young μειράκιον. 55 The noun triggers neuter agreement on the adjectives καλόν τε κἀγαθόν, which refer to his "birth and breeding" (Lamb 1924: 115), but masculine agreement on the next adjective καλός, which refers to his current appearence. It is again tempting to see in this grammatical gender mismatch an attempt at 54 The word ἄνθρωπος is used in this very passage (285b). 55 For speculations about Agathon's age see Denyer (2008: 84).
connecting the μειράκιον both with its past (τὴν φύσιν) and with its present (τὴν ἰδέαν).
In his current state, Agathon is obviously sexually active, as is made clear by Socrates' suspicion that he is Pausanias' παιδικά. About the latter word Dover says: "The Greeks often used the word paidika in the sense of 'eromenos' . It is the neuter plural of an adjective paidikos, 'having to do with paides' , but constantly treated as if it were a masculine singular " (1989: 16). In the passage just quoted (41), παιδικά is used as the predicative complement of ὤν, which shows masculine agreement, even though it refers back to μειράκιον.
There are many more cases of this kind of (mis)match between grammatical and natural gender. I conclude with some examples in which a neuter diminutive is used to refer to a female referent. The first one comes from Aristophanes' Wasps: (42)  In (42), τὸ γύναιον triggers neuter agreement on the first participle ὑποθωπεῦσαν, but the second participle καθεζομένη is feminine, which agrees with the natural, not the grammatical gender of τὸ γύναιον. A very similar example comes from the Septuagint: (43) καὶ αὐτοὶ εὑρίσκουσιν τὰ κοράσια ἐξεληλυθότα ὑδρεύσασθαι ὕδωρ καὶ λέγουσιν αὐταῖς· εἰ ἔστιν ἐνταῦθα ὁ βλέπων; καὶ ἀπεκρίθη τὰ κοράσια … In (43), τὰ κοράσια triggers neuter agreement on the participle ἐξεληλυθότα, but the anaphoric pronoun αὐταῖς is feminine, the gender of which is again determined naturally, not grammatically. The following clause is therefore all the more remarkable, as the verb ἀπεκρίθη is singular, because the subject τὰ κοράσια is neuter. This is of course the normal agreement pattern for neuter plural subjects (van Emde Boas et al. 2019: 322), but in this particular case it indicates that the grammatical and not the natural gender prevails again.
The final example is taken from the story of Jesus' healing of the daughter of Jairus, one of the rulers of a Galilean synagogue. It is transmitted in three versions in the synoptic gospels. Mark's version begins as follows: Here again the feminine pronoun αὐτῇ does not agree with the neuter diminutive τὸ θυγάτριον. 56 Matthew (9.18) and Luke (8.42) read θυγάτηρ instead of θυγάτριον, which explains the feminine agreement in ἐπ' αὐτήν in the version of the former (ibid.). Jesus' intervention is interrupted by a hemorrhaged woman and in the meantime Jairus' daughter has died. Jesus immediately goes to his house and says the following to the grieving crowd according to Mark: (45) τὸ παιδίον οὐκ ἀπέθανεν ἀλλὰ καθεύδει … καὶ κρατήσας τῆς χειρὸς τοῦ παιδίου λέγει αὐτῇ· ταλιθα κουμ, ὃ ἐστιν μεθερμηνευόμενον· τὸ κοράσιον, σοὶ λέγω, ἔγειρε· καὶ εὐθὺς ἀνέστη τὸ κοράσιον καὶ περιπάτει· ἦν γὰρ ἐτῶν δώδεκα … καὶ εἶπεν δοθῆναι αὐτῇ φαγεῖν Again feminine pronouns are used to refer to neuter diminutives: the second αὐτῇ (5.43) refers back to τὸ κοράσιον (5.43), but even more remarkable is the first αὐτῇ (5.41), which refers back to τὸ παιδίον (5.39) and τοῦ παιδίου (5.41). In Matthew's version, Jesus uses the neuter diminutive τὸ κοράσιον (9.24), which is again referred back to by a feminine pronoun in the phrase ἐκράτησεν τῆς χειρὸς αὐτῆς (9.25). Luke uses the feminine noun ἡ θυγάτηρ with female agreement patterns throughout his version of the story, with one exception: he uses the common noun ἡ παῖς [f] instead of the neuter diminutive τὸ κοράσιον to translate ταλιθα (Aramaic ). Judging from (43), (44) and (45) it seems safe to conclude that Greek girls behave exactly like German girls. The use of feminine pronouns to refer to the German neuter diminutive Mädchen has become a textbook example of a clash between semantics and grammar. Braun and Haig conclude that the choice depends both on the "semantics of age" (2010: 70) and on the "semantics of femaleness" (2010: 82), which is perfectly applicable to the examples just discussed, except that the definition of "femaleness" in terms of "age" differ in the case of Greek girls. The same holds, mutatis mutandis, for the use of masculine pronouns to refer to the neuter diminutives παιδίον and μειράκιον, which is equally dependent on the semantics of age and maleness.

CONCLUSION
In this paper I have discussed selected mismatches between natural and grammatical gender and the ways in which grammatical agreement is sometimes used to repair such mismatches. Epicene nouns ( §2) are sometimes overtly marked to reveal the natural gender of their referents, such as the male tortoise in (4b). The natural gender of common nouns ( §3) can be overtly marked by agreement on articles, pronouns, adjectives and participles, as in the case of the cock in (6). Masculine second declension nouns such as θεός are prototypically associated with male referents, as opposed to feminine first declension nouns such as θεά which are prototypically associated with female referents. Apparent mismatches of natural and grammatical gender often result in the reassignment of a noun to the other declension, such as the feminine seconddeclension noun παρθένος, which eventually became a first declension noun, i.c. παρθένα ( §4).
Nouns referring to human beings of the same sex sometimes differ in grammatical gender ( §5). In the division of the life cycle of male human beings according to (Pseudo) Hippocrates (17) and Ptolemy of Ascalon (18), the neuter τὸ παιδίον is younger than the masculine ὁ παῖς, who in turn is younger than the neuter τὸ μειράκιον, who in turn is younger than the masculine ὁ νεανίσκος. There seems no logical or, indeed, natural reason to shift gender twice in the coming of age of boys. The case of the common noun παῖς reveals that if the natural gender is not explicitly marked by agreement or, indeed, by the addition of the gendered adjectives ἄρσην / ἄρρην and θηλύς, it is either ambiguous, especially in the plural (παῖδες = 'children' , whether male or female) or, quite often, exclusively male (παῖδες = 'sons'). In the latter case, the opposition between male and female children is often expressed by antonyms, e.g. παῖδες ~ θύγατρες (22).
Diminutive nouns offer the most exciting insights in the way natural and grammatical gender interact and, indeed, clash. Neuter diminutives normally trigger neuter agreement patterns, but sometimes the semantics of age and "maleness" / "femaleness" have an impact on the choices speakers and writers make. Grammatically neuter nouns such as παιδίον, μειράκιον, γύναιον, κοράσιον and θυγάτριον are sometimes referred to by masculine and feminine pronouns, and in some cases even trigger 'gendered' agreement on adjectives or participles, as in the case of μειράκιον in (33) and (41). Braun and Haig conclude their study of the use of feminine pronouns to refer to German Mädchen that "people perceive biological gender as more relevant for adults