¹ Beiträge zur geschichte des geschlechtswechsels der englischen substantiva, p. 1.

² Das grammatische genus in Laamons Brut, p. 69; cf. also pp. 66-68.

³ Das grammatische geschlecht in der Ancren Riwle, pp. 32 f.

⁴ Grammatisches und psychologisches geschlecht im englischen, pp. 6-9. This paper was first printed in Nachrichten von der königlichen gesellschaft der wissenschaften zu Göttingen, Geschäftliche mitteilungen, 1912, Heft 2, pp. 102-121, but was reprinted in Berlin with copious notes in 1913 (Weidmannsche Buchhandlung). My quotations are made from the later edition.

⁵ Zur geschichte des grammatischen geschlechts, p. 5. Lindelöf shows, in his Beiträge zur kenntnis des altnorthumbrischen (Mémoires de la Société Neo-Philologique à Helsingfors, i, 1893), that in the language of the Lindisfarne Gospels there is great confusion and inconsistency in the use of noun genders, but does not attempt to explain the causes that brought about this condition; Carpenter, in Die Deklination in der northumbrischen evangelienübersetzung der lindisfarner handschrift, pp. 66-68, adds nothing to Lindelöf's material and nothing of importance to his discussion of the material. Breier, in his Eule und Nachtigal, pp. 101-107, states the facts in regard to the retention of the Old English grammatical genders in the Owl and the Nightingale but contributes nothing towards explaining the loss of gender in later Middle English; his statement that “der untergang des gram, genus nimmt . . . einen ähnlichen verlauf wie der verfall der flexion des substantivs, in der ja das gram, geschlecht zum ausdruck kam” (p. 101) expresses a view that is discussed below in note 8. Philippsen, in Die deklination in den Vices and Virtues, pp. 72-83, deals with the retention and confusion of genders in the text he studies but contributes nothing to an explanation of the loss of gender except the statement that “ein früher verfall des grammatischen geschlechts setzt stets auch einen frühen verfall flexionsverhältnisse voraus” (p. 72).

⁶ “Mit dem aufgeben der geschlichtigen formunterschiede des artikels, des adjektivs und des attributiven fürwortes und bei der völligen angleichung der deklination der substantive aller geschleehter musste die erinnerung an das frühere grammatische geschlecht sich fast gänzlich verlieren,” Englische grammatik, 3 ed., 1880, i, 263.

⁷ Emerson, History of the English language, pp. 289 f., 303 ; Sweet, New English grammar, i, 305 f.; Einenkel, Grundriss der germanischen philologie, 1 ed., i, 910 f.; Bradley, Making of English, pp. 48 f.; Keluza, Historische grammatik der englischen sprache, 2 ed., ii, 139 f.

⁸ Altho in Old English, as in Latin and other languages, nouns of different genders are characterised to a great extent by certain types of inflection, noun gender does not consist in or depend upon noun inflection. No mere change in the system of noun inflection, however extensive, can cause the loss of grammatical gender. Changes of noun inflection, by establishing new associations, are likely to cause confusion and shifting of genders, and it is probable that the confusion of genders which occurred in early Middle English was to a great extent the result of changes of noun inflection which had taken place thru the operation of sound change and analogy. But the loss of grammatical gender cannot be explained by this cause alone.

⁹ The Old English gender-distinctive forms of the weak adjective inflection (-a in the masculine nominative singular and -e in the neuter accusative singular) were not gender-distinctive in Middle English.

¹⁰ Modern Language Review, xiv, 97-102.

¹¹ Classen, p. 98.

¹² Hoffmann, p. 68.

¹³ Glahn, p. 21; cf. p. 9. Cf. also Morsbach's expression, in passage quoted above: “mit ausschluss der personalia, die bei diesem prozess keine rolle spielen konnten.”

¹⁴ Hoffmann, p. 68; see also p. 57.

¹⁵ Landwehr, p. 35.

¹⁶ Morsbach, pp. 10 f.

¹⁷ The statement made in this sentence is erroneous. As will be shown later in this paper, the gender-distinctive forms of the demonstrative and definitive article, even when they modified nouns like wife, maiden, and woman, were as a rule in concord with grammatical gender. The usage Körner illustrates was only occasional.

¹⁸ Körner, p. 27. See also the last sentence of the passage quoted from Körner at the beginning of this paper.

¹⁹ In using the words “evade” and “uncertainties” I do not intend to imply on the part of speakers of Middle English any consciousness of evasion or uncertainty.

²⁰ Die syntax in den werken Alfreds des grossen, i, section 238. For another Early West-Saxon example see Dialogs of Gregory the Great, ed. Hecht, p. 17, 17 ff., MS. C. For other examples in poetical texts see Genesis 2242 f., 2380 f. See also note 26 below.

²¹ I have printed Skeat's text as prose and modified the punctuation.

²² For examples of maiden, wif, and wimman used with neuter or masculine forms of the article or strong adjective but not followed by a personal pronoun see Laamon 3202, 9067, 9543 f., 14349, 14364, 16018, 22227, 23619, 25959 ff., 26132, 26137.

²³ Old English Homilies, I, 223, 18 f.; I have modified the punctuation. Note also es meidenes 241, 12.

²⁴ Landwehr's comment upon these passages is as follows: “Dass ein und dasselbe wort an verschiedenen stellen verschiedene genera (das des etymons und geschlechtslosigkeit z. b.) zeigen kann, wird kaum auffallen (vergl. p. 41 u. ff.). Wenn aber, wie in folgenden beispielen, in ein und demselben satze das gleiche wort zweigeschlechtig behandelt wird, und zwar so, dass das direkt dabei stehende pronomen (hier der artikel) erhaltung des genus des etymons zeigt, das entfernter stehende pronomen aber, das das wort wieder aufnimmt, bereits neutrum ist, so können wir hieraus folgern, dass in diesen fällen das gefühl für das gr. genus nicht mehr stark war. In dem gebrauch des geschlechtigen artikels können wir hier nur eine spontane weiterführung eines sprachgebrauchs erblicken, der in einer früheren periode noch lebendig und lebensfähig war” (p. 51). If Landwehr had recognized that grammatical gender which was in conflict with natural gender was expressed in Old English and Middle English primarily by the gender-distinctive forms of the strong adjective declension and definite article and demonstrative, and only occasionally by the personal pronoun, he would not have made this comment. The personal pronoun is conclusive evidence of grammatical gender only when it is used in conflict with natural gender.

²⁵ I followed the same principle in collecting the data given earlier in this paper in regard to the conflicts with natural gender in the gender-distinctive forms of the strong adjective declension, sē, sēo, þæt, and other pronominal words in Beowulf. That is, I disregarded the masculine and neuter genitive and dative singular and the masculine-feminine forms of ***hwã. I also disregarded the occasional plural gender-distinctive forms of the strong adjective declension.

²⁶ Inasmuch as his and him are indifferently masculine and neuter, they can be in conflict with natural gender only when they refer to masculine or feminine nouns that denote females (like wĩfman and wīf). But such nouns are almost always referred to by a feminine pronoun. The cases in which his and him refer to masculine nouns that denote lifeless objects (e. g. Beowulf 1528 f., ā wæs forma sī / dēorum mādme, þæt his dōm ālæg) are ambiguous, even apart from the tendency to use a neuter pronoun for referring to masculine and feminine nouns denoting lifeless objects, as in—sīo ecg gewāc / brūn on bāne, bāt unswīor, / þonne his īod-cyning þearfe hæfde (Beowulf 2577 ff.).

²⁷ I have treated the masculine nominative he (heo) and feminine nominative heo (he) as distinct in spite of occasional ambiguities of spelling.

²⁸ I offer the data contained in these tables as close approximations. I have gone thru all of the texts twice to make my collection of conflicts as complete as possible, but doubtless a few have escaped my notice. The totals of the gender-distinctive forms in the various texts are (except in two or three instances) based on one count. The actual totals would prove, I believe, to be slightly greater than those given in the tables, for one is more likely to miss forms in counting than to count forms that ought not to be counted, and in the cases in which I did make a recount I found my second total to be greater than my first. The probable error in the determination of the totals therefore tends to neutralize the probable error in the determination of the number of the conflicts.

²⁹ The greater percentage of conflicts in the Middle English texts is due partly to causes stated above on page 96, but still more, I think, to the abstract nature of the Middle English material as compared with the Old English. We should naturally expect a higher percentage of conflicts in homilies and similar material than in narrative. The surprising fact is that in material so abstract as the Bodley and Lambeth Homilies the percentage is not greater. Of course in limited portions of text the percentage may be much greater than any of those given in the table. For example, the first of Ælfric's Lives of Saints (ed. Skeat), which is largely a discourse on the nature of the soul, has a very high percentage of conflicts. But discourses of this kind are exceptional even in literature, and they are generally of limited extent. I might add that the unexpectedly high percentages of conflicts in Orosius and the Brut are due to the use of feminine pronouns referring to Romebyrig, Carthage, burh, etc.

³⁰ Modern Language Review, xiv, 100. On the following page Mr. Classen says: “it is clear from the evidences of Old English that there existed a strong sense of sex. . . . Such a sense of sex might readily develop a sense of sexlessness, or a sense of the distinction between living and lifeless things. How far this sense may have been strengthened by the possibility [italics mine] that the personal pronouns are most commonly used in the masculine and feminine to refer to man and woman, we do not know. But it is a possibility which ought not to be lost sight of.” These sentences might very well have suggested to me the mode of investigation used in the present paper if I had not already formulated the thesis here presented and accumulated material sufficient for its establishment a year or two before Mr. Classens' article appeared. I am glad, however, to acknowledge that I have been considerably aided in the presentation of my thesis by being able to use as a point of departure that portion of Mr. Classen's article quoted earlier in this paper. To discuss the article as a whole would take more space than can be given to it here. Mr. Classen does not appear to be acquainted with the investigations of Körner, Hoffmann, Landwehr, and Morsbach (Glahn's investigation was probably not available at the time he wrote). If he had been acquainted with them he would probably have distinguished more clearly than he does between the two distinct problems of the loss of grammatical gender and the establishment of natural gender in Middle English. As I have tried to show in the body of this paper, the loss of grammatical gender might have occurred without the establishment of natural gender, and the establishment of natural gender (expressed by the personal pronoun) might have occurred without the loss of grammatical gender.

³¹ Indeed, natural gender exists together with grammatical gender in Latin, Greek, Germanic, and the Indo-European languages generally. When a Roman remarked to his companion of a girl whom both had noticed but of whom neither had yet spoken, “Pulchra est” or “Illa pulchra,” the use of the feminine was determined by natural gender. So also in “Ignarus sum,” “Curiosa es,” etc. And a clause (which of course has no grammatical gender) is always neuter.

³² I followed in collecting the data on which this table is based the same system that I followed in collecting the data from Old and Middle English texts on which the previous tables are based. Of the Gothic pronoun is, si, ita, etc., the Old High German pronoun (h)er, siu, iz, etc., and the Old Saxon pronoun hē, siu, it, etc., I included only the forms that are positively gender-distinctive. The same forms are positively gender-distinctive in all four dialects except that in Old High German the neuter genitive singular es is positively gender-distinctive whereas the corresponding forms in the other dialects are only negatively gender-distinctive. I have gone thru the Gothic, Old High German, and Old Saxon texts only once, but I believe the percentages I have arrived at are a sufficiently close approximation for the purpose for which I offer them.

³³ Old English examples of this usage would be superfluous here, and examples from the other dialects are scarcely less so. Among those that occur in the texts examined for the purpose of this paper are: thaz kind . . . inan, Tatian, iv, 11; uúib sie, Tatian xxviii, 1; themo magatine . . . siu, Tatian, lxxix, 9; that kind . . . he, Heliand 382 ; that frī . . . ira, Heliand 435 f.; them uuībe . . . siu, Heliand 445 f.; barn . . . ina Heliand 798; &thornata barn . . . izai, St. Mark, 5: 41. In these texts I have not found any neuter noun denoting a female referred to by a neuter pronoun; kind and barn are referred to either by a neuter pronoun or by a pronoun corresponding to the actual sex of the child.

³⁴ I have noted the following clear examples in Old High German: Otfrid, iv, 16, 33; Tatian, lxxii, 5. In Heliand 1509 f. himile (mas.) and eru (fem.) are referred to by neuter that. To the Old English examples already given of hit referring to a masculine or feminine noun I may add: Beowulf 779, 1234, 2248, and possibly 3161; Orosius, p. 174, 4-8; Ælfric, Homilies, i, 88, line 3 from bottom.

³⁵ As a matter of fact this usage is no more peculiar to Germanic than to Old English. It is impossible of course to give for Latin and Greek comparative statistics in precisely the form that was used for the Germanic dialects, because Latin and Greek have no third personal pronoun (that is no gender-distinctive demonstrative used exclusively as a substantive) and because the pronominal words that perform in Latin and Greek the function of the Germanic third personal pronoun are so diverse morphologically. But if we examine the substantive use of these pronominal words in Latin and Greek we find that they are seldom in conflict with natural gender. The Latin Vulgate text of St. Luke (ed. Nestle) contains about 759 forms of ille, is, hic, ipse, and iste used substantively, of which 25 are in conflict with natural gender, giving a percentage of .0329. The Greek text of St. Mark (ed. Westcott and Hort) contains about 593 forms of , and demonstrative δ and s used substantively, of which 21 are in conflict with natural gender, giving a percentage of .0354.