Published online by Cambridge University Press: 22 April 2026
The historical interpretation of the inflectional forms of Tocharian nouns has lagged behind for an obvious reason. The endings of the verbs show their IE affinities at once, and such changes as have occurred present no more obstacles to their explanation than those in the other IE languages. On the other hand the case endings of the nouns, most of all those of the nominative plural, e.g., -a
m, -ñ, -ant, -u, do not recall directly any IE case formatives, and therefore at first seem to defy explanation. Nevertheless it is not possible that the Tocharian declensions have no IE antecedents at all, for such a contrast between the origin of nouns and verbs is inconceivable. Explanations of Tocharian case endings must take as their starting point IE conditions. This is the only line of attack which holds forth promise of attaining any results, and even in this way it cannot be hoped to settle once for all the history of every variety of declensional pattern. While the writer believes that he can make some suggestions which will win assent, he also recognizes that other details will necessarily be subject to change after maturer consideration.
1 The genitive plural is classified as a primary case by SSS (i.e., Schulze, Sieg, und Siegling, Tocharische Grammatik) because it is not always based upon the oblique, but it is nevertheless omitted from consideration here because sometimes its form is plainly not of IE origin, and in other instances it borrows the ending of the singular, which distinctly puts it on the border line between primary and secondary cases.
2 The neuter plurals in -ā from feminine collectives in -ā would be a parallel if any evidence for such an origin of any Tocharian plural could be found otherwise.
3 In Hittite all IE quantitative vowel distinctions have disappeared, so that, e.g., u and ū are treated alike.
4 The representation of IE e by y in these two words is puzzling, but the question of its origin does not affect the morphological aspects discussed here as long as there can be no doubt of the ultimate identity of the words compared.
5 The reason for clothing this and other statements in IE terms is merely the desirability of avoiding commitment on vexed chronological problems which do not affect the questions discussed. That the process was actually performed in IE times is of course out of the question.
6 It is true that IE knew denominative adjectives in -went- and -ment- (Skt. -vant- and -mant-), but Tocharian shows no signs of the previous existence of this category.
7 It probably would be an error to connect the apparently passive meaning of these participial formations with the real passive use of Hitt. -ant- (see Sturtevant, HG 158). We must assume that the latter is a secondary development, since the Hittite herein stands altogether by itself.
8 After consonants an ä necessarily appears before the -nt-. Historically it may either be a svarabhakti vowel, or -änt- is IE -
nt-.
9 So, e.g., vēžamas, OBlg. vezom
b from ve
bú, vezą ‘veho’.
10 Often with assimilation of y to a preceding sound, e.g., aśśāñ beside aśyãñ.
11 This statement of equivalence ignores the long ē of Lat. -ēs, which, of course, is known to be secondary.
12 This result implies that the first steps in the palatalization of the preceding t must have been completed before the loss of the following e.
13 Since the m-participles and m-adjectives mostly were originally o-stems, the ending -ä
o here represents a spreading of the s-stem formation beyond its original sphere.
14 That is, IE -us-ī, weak grade of -wes- with -ī, nominative of the feminine suffix -yā-.
15 The fact that Toch. -i and -e are only nominative, but not oblique endings, prevents their identification with the IE -oi of the neuter dual of o-stems, e.g., Skt. yug
e.
16 Cf. Hrozný, Congr. Ling. 1935 xix.
17 With syllabic
n (-
ns) after consonants.
18 See Part 4 for examples.
19 Except for the difference in the quantity of the a. Its shortness in Tocharian was no doubt due to some analogy.
20 Oblique forms in -
m also occur: masc. ca-
m and neut. nom. obl. ta-
m. That the preceding vowel is here a instead of ä may well be due to a secondary development.
21 The process is well illustrated by the resulting modern forms, although it really took place in a prehistoric period.
22 For trit ‘third’ and the like see Part 6.
23 In monosyllables, as shown by the pronouns kus = Lat. quis and the demonstrative fem. nom. pl. to
s = Skt. tã
h, IE ∗tās, final -s was retained. To this conclusion ko ‘cow’ presents no obstacles, since it occurs only as an oblique (SSS 110), where it came from -w
m, and was thus a form like Lat. bovem. The nominative may well have been ∗kos, or it became ko analogically.
24 The connection of -e with feminines also prevents taking it as derived from accented forms in IE -oi, which would correspond to the -i from enclitics in -oi. Of course the IE ∗moi, etc. were enclitic only, but such forms became accented secondarily in some languages, and in Tocharian the genitive ke (all three genders) from kus ‘qui, quis’ is certainly an analogical form of this type. Yet connection with it of the genitive ending -e would leave unexplained the pronounced feminine associations of the latter.
25 Note, however, the ending -y, which occurs in [ñä]kteññāy, gen. of
äkteññā ‘goddess’ and in a number of proper names taken over from Sanskrit (SSS 86). If, as seems probable, this is the same as -i, it went from the nouns of relationship to the word for goddess, and there took the form -y after the stem vowel ā.
26 The initinal ñ- for the m- of ∗moi came from the plural IE ∗nos. See Lang. 11.200 f.
27 This etymology presupposes the origin of the m from earlier pm. For the y cf. note 4.
28 The striking analogy -ōs (nom. pl. o-stems) : -ō (dual) = -es (nom. pl. consonant stems) : -e (dual) need not be discarded with the realization of the IE age of the -e. The process simply took place in IE or pre-IE times.
28 That -tos and -tis became -ts was mentioned in Part 4.