¹ The endings of the act. ipf. in the other OArm. conjugations, e.g., sg. 1st layi (: pres, lam), Inowi (: pres. lnowm), and in the post-classical sbj. ipf., e.g., bericei (or ēi), are of course analogical; so also is the use of the endings -inn, -th(e)a, etc. in the OIr. pret, future (‘conditional‘) and sbj. pt.

² A. Meillet, Esquisse d'une grammaire comparée de l'arménien classique² 127(1936).—This work is hereafter cited as Esq.²—Cf. also Kieckers, IF 35. 108–14 (1915).

³ The earlier literature is reviewed by H. Zeller in Brugmann-Thumb(-Debrunner-Sommer), Gdr. d. idg. Sprach- und Altertumskunde 2.4.2. 93–4 (1927). This k' Meillet came to regard (Esq.² 56) as a special development of IE final s. However, as k' is known to derive elsewhere from IE ∗t and ∗s (e.g., in k'oyr 'sister') it seems reasonable to derive its occurrence in personal endings from the addition of enclitic personal pronouns, i.e., 1st pl. ∗-mes + …(cf. Skt. vayám, Goth, weis 'we'), 2d pl. ∗-te + s, …(cf. Welsh chwi 'ye', Goth. [dat. and acc.!] i-zivis); on the 2d pl. cf. the similar incorporation of the pronoun into the ending in Welsh 2d pl. cerwch; other but less immediate parallels are common enough. In the 1st pl. ending the syncope of e between m and k' must have been fairly late, as shown by the fact that assimilation of the labial nasal has not yet taken place in OArm. (but cf. NArm. berenk'); before syncope occurred the monosyllabic ending ∗-mek' was abstracted from the verb and used as an (independently accented) 1st pl. personal pronoun, nom.; ultimately the old oblique cases were reformed so as to begin with m- also; cf. the abstraction and use as a personal pronoun of the NIr. ending 1st pl. -maid or -muid in certain northerly dialects (cf. Pedersen, Vgl. Kelt. Gr. 2.310 (1913); the Christian Brothers' Irish Grammar §270 (n.d.), though sharply condemning the usage, gives as examples chonnaic muid é ‘we saw him’, chonnaic sé muid 'he saw us'). One is tempted to suppose that Lith. mēs 'we', with which Arm. mek‘ is usually compared, has a similar origin; if so, it was abstracted at a time when a primary ending was still possible in the 1st pl. of Lithuanian verbs. The ultimate exclusive use of the secondary ending here (e.g., dìrbame) would have been due to avoidance of ambiguity between primary ∗-mes and the periphrastic reflexive 1st pl. -m-s, the use of the lengthened grade of the secondary ending (cf. Ved. -mā) having evidently proved an insufficient means of distinction. Brugmann, Gdr.² 2.2. §384.1 anm. 1, merely ascribes the initial m- of such pronouns to the influence of the corresponding personal endings, an explanation probably adequate for OCS 1st pl. my, the oblique cases of which have initial n-. Somewhat later in pre-Armenian the pl. nom. of the second personal pronoun (sg. dow) was analogically reformed as dowk’. Here, however, the oblique cases remained unaffected. Somewhat imperfectly on the analogy of dowk' arose the verbal form pass. aor. 2d pl. berarowk (: 2d sg. berar) discussed in §10 below. The element k' having been thus schematized as a sign of plurality, it came to be attached to the sg. nom. of most nouns to form the pl. nom. (ban : bank', but hayr : hark‘), and eventually to the sg. instr. to form the pl. instr. (baniw : baniwk’, but oskerb : oskerawk'); on the tendency to schematization thus revealed; cf. the far more thorough-going use of -(n)er as a sign of plurality in NArm. noun declension, e.g., sg. nom. cer, gen. cerow · · · : pl. nom. cerer, gen. cererow · · ·

⁴ Esq.² 126–7.

⁵ Esq.² 127; this was undoubtedly Meillet's final view, as shown by his language: 'Mais une forme telle que berēr est certainement ancienne, en effet berēr repose sur ∗bheretor et coïncide avec phr. aβ-β∈ρερoρ'. Earlier in the same paragraph, however, he strangely retains from the first edition (1903) what seems to be an incompatible explanation: 'la 3^me personne ∗-y-r aurait un aspect particulier parce qu'elle reposerait sur une ancienne form monosyllabique d'imparfait ∗ēst, cf. Skt. ah, Gr. .¹ In the same paragraph Meillet derives the post-classical beriwr (a by-form to berēr) from a form ∗bheretro 'avec métathese'. As to this last derivation and its implications I prefer to reserve judgment.

⁶ J. A. Kerns and B. Schwartz : Structural Types of the IE Medio-Passive Endings, Lang. 13. 263–78 (1937).

⁷ Esq.² 125.

⁸ Cf. however Ose. ipf. 3d pl. fufans; It. ipf. 1st pl. eravamo, 2d pl. eravate as illustrating a somewhat similar tendency.

⁹ As far as the i-vocalism is concerned, we might as well start out from ∗esēm, but it is doubtful whether interior -s- would disappear without a trace, cf. sg. gen. -oy < ∗-os .

¹⁰ Brugmann, Gdr.² 1. §236 (1897); Meillet, Esq.² 19, 55.

¹¹ Sg. acc. am supposedly from ∗samām or the like may be the nom., or may be analogical to the sg. acc. of stems ending in a short vowel.

¹² Cf. §9 below.

¹³ Meillet, Esq.² 127.

¹⁴ Goodwin, Syntax of the Moods and Tenses of the Gk. Vb.^3–4 §462 (1897?).

¹⁵ J. Morris Jones, A Welsh Grammar Historical and Comparative §171.22.; §180.4.3; §181.6 (1913).

¹⁶ Vgl. Kelt. Gr. 2. §605.2 (350).

¹⁷ This Morris Jones illustrates (WGr. §171.2.2) among other examples by rendering Vedic kāmáyēta rā samr bhávitum ‘a king would like to be a supreme ruler’ into Welsh as carai brenin fod yn benadur.

¹⁸ The forms are taken from Schulze, Sieg, u. Siegling, Toch. Gr. §461 (1931); cf. the authors' language at the close of this paragraph: ‘Die B-Formen stehen innerhalb ihres Dialektes nicht so vereinzelt da wie yes and ses in A, denn sie gehoren genetisch vielleicht zusammen mit der fur B allgemein giltigen Imperfektbildung, die mit dem Optativ …zwar das stammbildende i (bez. oy) und die Personalendungen teilt, aber im Gegensatz zum Optativ die Stammform des Prasenssystems zeigt.‘

¹⁹ Of course if this explanation were adopted, all attempts to connect these endings with the equally puzzling Indo-Iranic athematic middle primary 2d du. ∗-āthai, 3d du. ∗-ātai and secondary 2d du. -āthām, 3d du. -ātām would have to be given up; for the conventional view cf. Bartholomae, KZ 29.283 ff. (1888); Brugmann, Gdr.² 2.3.601 (1916); Thumb (-Hirt), Hdb. d. Skt.² §432 with Hirt's note in the Nachtrag (1930); Reichelt, Awest. Elb. §256 (1909).

²⁰ Is such an Avestic form as pres. 3d du. -myāsaⁱte (Yasna 33.1) an archaism whose origin antedates the spread of an originally optatival vocalism? Cf., however, Bartholomae, in Geiger-Kuhn, Gdr. d. iran. Phil. 1 §120.1; §300.4.

²¹ Cf. note 6 above.

²² Esq.² 119.

²³ E.g., ind. 3d sg. esa-r-i; ipv. 3d sg. esa-r-u.

²⁴ Brugmann, Gdr.² 2.3 §125; Meillet, Esq.² 107.

²⁵ For the possibility of redistribution of functionally overlapping forms, cf. the retention in OIr. of the substantival pl. nom. ending of masc. o-stems (IE ∗-ōs) as a pl. voc. (firu, i.e., [a]fhiru), as against its complete replacement in the pl. nom. proper by the pronominal ending (IE ∗-oi, whence OIr. fir).

²⁶ Bartholomae, Studien zur idg. Sprachgesch. 2.75 (1891); apparently accepted in Hirt, Idg. Gr. 2. §92§2 (1921); Lat. erās, etc., are accounted of late and individual origin in Sommer, Hdb. d. lat. L. u. F.^2,3 §345.4 (1914); Brugmann, Gdr.² 2.3 §108 end; Stolz-Schmalz, Lat. Gr.⁵ (rev. Leumann-Hoffmann) §238 (1928); Kieckers, Hist. Lat. Gr. 2.316 (1931).

²⁷ To postulate the use of a primary ending in pre-Armenian aorists is legitimate enough, in view of the use of Dor. Gk. -μ∈, Lat. -mus, in secondary tenses; cf. also §§11, 12 below.

²⁸ Somewhat different is Meillet's explanation of the loss of m in the 1st pl. (Esq.² 122).

²⁹ Meillet, Revue des études armeniennes 10.183 (1930); Esq.² 124.

³⁰ Many points affecting the development of IE t in Armenian are obscure (cf. Brugmann, Gdr.² 1. §577); Meillet, Esq.² 33). If the suggestion here made is pertinent, then the derivation of sg. abl. -oy < ∗-o-tos (Brugmann loc. cit.) would have to be given up, and the form regarded as an analogically used genitive; on the other hand, sg. instr. -aw could be derived from IE ∗-ā-to(s), cf. Hit. sg. abl. anaz (Sturtevant, Hit. Gr. §§196–7 [1933]), rather than from ∗-ā-bhi, as is generally done.

³¹ So Hirt, Idg. Gr. 4 §46 (1928).

³² Further details as to the use of -t), rather than acceptance of Hirt's view of its origin, are found in Vondrak(-Grümenthal), Vgl. Slav. Gr.² 2.126 (1928); N. v. Wijk, Gesch. d. aksl. Spr. 1.213–4 (1931); P. Diels, Aksl. Gr. §108 and anmm. 2, 3 (1932).

³³ Vgl. Kelt. Gr. 2 §605 end.

³⁴ IF 34.408–10 (1915).

³⁵ Hdb. d. altirischen 1. §579 (1909); in §87, however, he speaks of bered as being ‘vielleicht aus ∗bhereto’.

³⁶ Vgl. Kelt. Gr. 2 §605; I do not, however, understand Pedersen's insistence that the enclitic element ‘muss mit einem Vokal …angelautet haben’.

³⁷ Cf. Thurneysen, Hdb. d. altir. 1 §97, §98.1; §152; §154; §87.1.

³⁸ The analogical origin here suggested for the simplex pres. 2d pl. -the does not, to be sure, entirely account for its final vowel being e rather than i.

³⁹ The development of analogical 3d pl. -tis rhyming with 1st pl. -mis was doubtless favored by the parallel pair 3d pl. -tar, 1st pl. -mar (prosoposemics) in the conjunct deponent inflections; probably only when -mis, -tis, though still felt as primarily medio-passive, began being used actively did -mar, -tar begin to appear in the act. preterite. Later still, at the beginning of MIr. times, the tendency toward rhyme between the 1st pl. and 3d pl. endings in any one tense led to the use in the present (simplex) of 1st pl. -mit (> NIr. -maid, etc.), although its t was really a suffixal object pronoun, to match 3d pl. -it < ∗-onti, etc.

⁴⁰ Thurneysen, Hdb. d. altir 1 §179; Pedersen, Vgl. Kelt. Gr. 1 §334 (1909).

⁴¹ Cf. bronn < ∗brusnos, etc. (Thurneysen, Hdb. d. altir. 1 §148a). MIr. 1st sg. -ind is of course purely orthographical, basing on a confusion due to the sound-change by which the historic group nd, though still often so written in MIr., had become nn; cf. Dottin, Manuel d'irlandais moyen 1 §11 (1913) : the NIr. orthography is everywhere nn.

⁴² Cf. note 1 above.

⁴³ Cf. note 6 above.