Brugmann’s view on Aryan gender has not been unchallenged. The weakest points in his arguments are, of course, that there are so few old naturally feminine words in -a and -i to take as starting-points for such a thoroughgoing modification of the grammatical system, and that Brugmann was unable to give any striking explanation of the concord of adjectives and pronouns with words that had not these endings, but which were nevertheless treated as masculines and feminines respectively. It would lead us too far here to give any minute account of the discussion which arose on these points;[99] one of the most valuable contributions seems to me Jacobi’s suggestion (Compositum u. Nebensatz, 1897, 115 ff.) that the origin of grammatical gender is not to be sought in the noun, but in the pronoun (he finds a parallel in the Dravidian languages)—but even he does not find a fully satisfactory explanation, and the Aryan gender distinction reaches back to so remote an antiquity, thousands of years before any literary tradition, that we shall most probably never be able to fathom all its mysteries. Of late years less attention has been given to the problem of the feminine, which presented itself to Brugmann, than to the distinction between two classes, one of which was characterized by the use of a nominative in -s, which is now looked upon as a ‘transitive-active’ case, and the other by no ending or by an ending -m, which is the same as was used as the accusative in the first class (an ‘intransitive-passive’ case), and an attempt has been made to see in the distinction something analogous to the division found in Algonkin languages between a class of ‘living’ and another of ‘lifeless’ things—though these two terms are not to be taken in the strictly scientific sense, for primitive men do not reason in the same way as we do, but ascribe or deny ‘life’ to things according to criteria which we have great difficulty in apprehending. This would mean a twofold division into one class comprising the historical masculines and feminines, and another comprising the neuters.
As to the feminine, we saw two old endings characterizing that gender, a and i. With regard to the latter, I venture to throw out the suggestion that it is connected with diminutive suffixes containing that vowel in various languages: on the whole, the sound [i] has a natural affinity with the notion of small, slight, insignificant and weak (see Ch. XX § 8). In some African languages we find two classes, one comprising men and big things, and the other women and small things (Meinhof, Die Sprachen der Hamiten 23), and there is nothing unnatural in the supposition that similar views may have obtained with our ancestors. This would naturally account for Skr. vṛk-ī ‘she-wolf’ (orig. little wolf, ‘wolfy’) from Skr. vṛkas, napt-ī, Lat. neptis, G. nichte, Skr. dēv-ī, ‘goddess,’ etc. But the feminine -a is to me just as enigmatic as, say, the d of the old ablative.
XIX.—§ 19. Final Considerations.
The ending -a serves to denote not only female beings, but also abstracts, and if in later usage it is also applied to males, as in Latin nauta ‘sailor,’ auriga ‘charioteer,’ this is only a derived use of the abstracts denoting an activity, sailoring, driving, etc., just as G. die wache, besides the activity of watching, comes to mean the man on guard, or as justice (Sp. el justicia) comes to mean ‘judge.’ The original sense of Antonius collega fuit Ciceronis was ‘A. was the co-election of C.’ (Osthoff, Verbum in d. Nominal-compos., 1878, 263 ff., Delbrück, Synt. Forsch. 4. 6).
The same -a is finally used as the plural ending of most neuters, but, as is now universally admitted (see especially Johannes Schmidt, Die Pluralbildungen der indogerm. Neutra, 1889), the ending here was originally neither neuter nor plural, but, on the contrary, feminine and singular. The forms in -a are properly collective formations like those found, for instance, in Lat. opera, gen. operæ, ‘work,’ comp. opus ‘(a piece of) work’; Lat. terra ‘earth,’ comp. Oscan terum ‘plot of ground’; pugna ‘boxing, fight,’ comp. pugnus ‘fist.’ This explains among other things the peculiar syntactic phenomenon, which is found regularly in Greek and sporadically in Sanskrit and other languages, that a neuter plural subject takes the verb in the singular. Greek toxa is often used in speaking of a single bow; and the Latin poetic use of guttura, colla, ora, where only one person’s throat, neck or face is meant, points similarly to a period of the past when these words did not denote the plural. We can now see the reason of this -a being in some cases also the plural sign of masculine substantives: Lat. loca from locus, joca from jocus, etc.; Gr. sita from sitos. Joh. Schmidt refers to similar plural formations in Arabic; and as we have seen (Ch. XIX § 9), the Bantu plural prefixes had probably a similar origin. And we are thus constantly reminded that languages must often make the most curious détours to arrive at a grammatical expression for things which appear to us so self-evident as the difference between he and she, or that between one and more than one. Expressive simplicity in linguistic structure is not a primitive, but a derived quality.