Hostname: page-component-89b8bd64d-4ws75 Total loading time: 0 Render date: 2026-05-06T19:57:31.063Z Has data issue: false hasContentIssue false
  • English
  • Français

Problems of Tocharian Phonology

Published online by Cambridge University Press:  22 April 2026

George S. Lane
George S. Lane*
Affiliation:
University of North Carolina
Get access Rights & Permissions [Opens in a new window]

Extract

The forty odd years which have elapsed since the first publication of a Tocharian text leave much for the comparativist to do. As clear as is the affinity of both dialects A and B to Indo-European, even so inexplicable are still many of the countless special developments that have led to their remarkable appearance alongside the other languages of the family. It is, in fact, the problem of the relationship of Tocharian to the other Indo-European dialects which has most interested scholars from the very beginning, while very little time has been devoted to the comparative study of the two dialects themselves. Indeed, their interrelationship is as yet quite imperfectly understood. As remarked already by M. Sylvain Levi, in speaking of the vocabulary, the more one observes their fundamental identity, the more one is struck by their divergences. The remark may apply no less appropriately to the phonology.

Information

Type
Research Article
Information
Language , Volume 14 , Issue 1-Part1: January-March , March 1938 , pp. 20 - 38
Copyright
Copyright © 1938 by the Linguistic Society of America

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Article purchase

Temporarily unavailable

References

1 Hoernle, JASB 62.39 f. (1893).

2 I use here the name ‘Tocharian’ in the established (American) sense for both dialects A and B, without regard to the appropriateness of the name for both or even one of the dialects. Lévi's ‘langue Ārśi’ (= A) and ‘langue de Koutcha’ or ‘Koutchéen’ (= B) may be preferable on several counts, but the former is yet quite unfamiliar in English. Should we translate perhaps as ‘Arshian'? ‘Kuchean’ is of course familiar in England through the works of Sir M. A. Stein and others. Cf. Lévi, Le ‘Tokharien B’, langue de Koutcha, Journal Asiatique 1913 (2).311 ff., 1933 (222).1 ff. (The latter also in Fragments de Textes Koutchéens, Cahiers de la Société Asiatique 1re Série, II, Paris 1933).

3 Fragments de Textes Koutchéens 31 ff.

4 Indogermanen und Germanen 115.

5 Geschichte der indogermanischen Sprachwissenschaft, zweiter Teil, 5, 2 (Tocharisch).35.

6 Archiv Orientální 2.320 ff. (1930).

7 Journal Asiatique 1912 (19).339 ff.

8 Above ftn. 2. Henceforth abbr. Fragm.

9 Göttingen, 1931. Henceforth abbr. Toch. Gramm.

10 Cf. also Feist, Indogermanen und Germanen3 115 ftn. 1; Fraenkel, IF 50.7.

11 Le Groupement des Dialectes Indo-européens 19 f. So also now Benveniste, ‘Tokharien et Indo-Européen’, in Germanen u. Indo-germanen, Festschrift für H.Hirt 1.235.

12 Groupement 31.

13 Cf. Meillet, MSL 17.286.

14 Meillet, Indogerm. Jahrbuch 1.18; Schrader, Reallexikon2 2.353; Schulze, Kleine Schriften 260, etc.

15 Schulze, Kl. Schr. 240 ftn. 3, 260; Fraenkel, IF 50.97.

16 JAs. 1911 (18).633. Smith's (Videnskabsselskabets Skrifter 2. Hist.-Filos Kl., 1910, no. 5) early derivation from IE *leiqw-:Grk. λείπω, etc. is certainly to be discarded (cf. Meillet, l.c.), but is still quoted by Schwentner (Tocharisch 35), Lévi (Fragm. 32) and Reuter, Journal de la Soc. finno-ougrienne 47.4.13.

17 Lidén, Aufsätze Kuhn 142 f.

18 Falk-Torp, Etym. Wtb. 744; Walde-Pokorny 2.242.

19 Meillet, JAs. 1912 (19).116; MSL 17.286; Idg. Jahrb. 1.14; Fraenkel, IF 50.8.

20 Cf. Fraenkel, IF 50.19; Hermann, KZ 50.307.

21 Cf. Schulze, Kl. Schr. 252 ftn. 4; Fraenkel, IF 50.7.

Professor Sapir (Language 12.263) assumes loss of post-vocalic IE d in Tocharian citing this group and A kri ‘will’, käryā- ‘bedenken’, which he connects with the IE words for ‘heart’, Grk. καρδία, Lat. cor, cordis, etc. To arrive at the proper vocalism in the latter case however he starts from an (otherwise unknown) *kred- reduced form of *k̂red- from dissyllabic *kered-. The comparison of Skt. çrad-dadhāti ‘believes’, Lat. crēdō, however, is no support for the base, since the first member of these verbal compounds is probably not a word for ‘heart’, but rather an Indo-Iranian-Italo-Celtic religious term indicative of the magical properties of an object, cf. MIr. cretair, W. creir, crair ‘relics (of the Saints), holy thing’ (Walde-Pokorny 1.423; Walde-Hofmann 287; and especially Vendryes, Rev. celt. 44.90 ff.). Against this loss of post-vocalic d, I would cite B preściye ‘mire, filth’: Lith. brendù, brìsti, Russ.-ChSl. bredu, bresti ‘wade’, etc. (Lidén, Stud. z. toch. Sprachgesch. 7 f.; but we might have here IE dh of course); A kät-k- ‘überschreiten, vorübergehen’ (Toch. Gramm. 427), B kät-k- ‘tomber, passer, arriver à‘: Lat. cado ‘fall', Skt. çad- ‘fall off’ (so Meillet, in Lévi, MSS Remains 378 f.); A āti, B ‘grass’: Lat. ador ‘spelt’, Goth. atisk ‘field of grain’.

22 Fraenkel, IF 50.230. But A oko ‘fruit’ = B oko, hence more probably (after Lidén, Stud. z. toch. Sprachgesch. 34) to Lith. úoga ‘berry’, Goth. akran ‘fruit (of the field)‘, etc. (Walde-Pokorny 1.173), rejected by Fraenkel, l.c. Cf. below §4, 1.

23 Fraenkel, IF 50.222 ftn. 2. For Schrader's A ko-, B kan-, Poucha's A ko-, B ken-, cf. above.

24 Schrader, Reallex.2 2.255.

25 Meillet, JAs. 1912 (19).113; MSL 15.327 ff.; Schrader, Reallex. 1.635; etc.

26 Fraenkel, IF 50.16 f.

27 Meillet, Idg. Jahrb. 1.16; Fragm. 37 (quoted by Lévi). Connection with A, B śwā-, śu- ‘eat’ (as e.g. Fraenkel, IF 50.7) seems highly improbable.

28 So Fraenkel IF 50.8 (but eventually with different IE root connection, cf. ftn. 27). The older comparison with Lat. homō, Goth, guma, old Lith. žmuõ, etc. ‘man’ is certainly to be discarded.

29 So Schrader, Reallex. 1.246 and cited also by Poucha, Arch. Or. 2.323. Toch. Gramm. gives only yoṁ ‘Spur’.

30 Schulze, Kl. Schr. 261.

31 Fraenkel, IF 50.227 ftn.

32 Meillet, Idg. Jahrb. 1.19 suggests Turkish origin without closer identification, but apparently with reference to the group of Osmanli gün.

33 Connection with any one of several roots in au-, or eu- is of course possible, cf. e.g. *au- ‘weave’ (Walde-Pokorny 1.16), possibly orig. ‘begin weaving’, cf. Lat. ordīrī in the wider sense ‘begin’; or perhaps *eu- ‘put on’, of clothing, etc. (op. cit. 1.109f.), had a more general sense originally. But such connections cannot of course be demonstrated.

34 I fail to find evidence to support Meillet's statement (Introduction7 99) that Tocharian keeps IE o and a distinct. For the cases of retained o in both dialects cf. §4, 1.

35 Meillet, JAs. 1912 (19).113.

36 Reuter, Jour. de la Soc. Finno-ougr. 47.4.9.

37 Meillet, JAs. 1911(18).147, Idg. Jahrb. 1.19; Kretschmer, Glotta 20.66 f.; Benveniste, Hirt-Festschrift 2.235. But no one seems to have envisaged the difficulty that Toch. ṁ does not equal final m but n. Greek and Hittite show simply the extension of n from m, phonetically correct in final position, to the oblique cases (Hitt. gen. taknas), cf. Sturtevant, Hitt. Gramm. 136. The root has m, cf. Grk. χαμαί, Lat. humus, etc. (Walde-Pokorny 1.662 ff.).

38 Cf. also Meillet, MSL 17.284.

39 Groupement 28.

40 Contrary to Meillet's assumption (MSL 18.18) based on the palatilized ṣṣ from sk.

41 Cf. Meillet, MSL 17.284.

42 Reuter (Jour. Soc. Finno-ougr. 47.4.13) revives again the ghost of labial development from labiovelars by deriving the group from *wer-gw- an otherwise unattested formation (cf. Walde-Pokorny 1.272).

43 Groupement 20 f.

44 Poucha, Archiv Orientální 2.324.

45 Meillet, MSL 19.159.

46 Schulze, Kl. Schr. 257.

47 Meillet, MSL 19.159.

48 JAs. 1911 (17).456.

49 MSL 17.284 f.

50 Schrader, Reallex.2 2.70, etc.

51 JAs. 1912 (19).112; Walde-Pokorny 2.267.

52 Cf. Brugmann, Grundr.2 2.543.

53 Meillet, JAs. 1912 (19).115 f.; Boisacq, Dict. étym. 999.

54 Pedersen's comparison (Groupement 32) with MHG un-bil ‘ungemäss’ is surely erroneous because of the A vocalism.

55 Holthausen's comparison (IF 39.66) with NHG Frist, etc. wrecks on the guttural of the B forms, and also on A tāpärk ‘now’, which is probably related For the Gmc. group cf. Walde-Pokorny 2.34.

56 Meillet, JAs. 1911 (17).451; MSL 17.284.

57 Ibid.

58 Lidén, Studien zur toch. Sprachgesch. 1.17 ff., but without B peret.

59 Meillet, MSL 18.28; Fraenkel, IF 50.227 ftn. 1.

60 Meillet, JAs. 1911 (18).148; Toch. Gramm. 2.

61 Benveniste, Hirt-Festschrift 2.236.

62 Stud. z. toch. Sprachgesch. 34, but probably not from ā as he assumes.

63 Sapir (Language 12.179 ftn. 15) invokes Brugmann's å, that is o not of the e/o-series (and for which he substitutes ọ), to explain the retention of A okät, and in opäśśi, opśi ‘dexterous, skilful’, which he connects with Lat. opus, etc., as opposed to IE o in gradation with e which gives A a. This explanation comes to grief on at least two points. First å (ọ) also gives A a in PIE *oqw- (Walde-Pokorny 1.169 ff., Brugmann, Grundr.2 1.153 ff.), cf. A ak (B ek, cf. above 3, 1) ‘eye’, and in PIE *potis (Walde-Pokorny 2.77 f., Brugmann, l.c.), cf. A pats ‘husband’ (no B equivalent). Secondly IE o (alternating with e) gives also A o; cf. ṣom sg. obl., ṣome pl. nom. of sas, fem. säṁ ‘one’ (above §3, 2 with reference), and also Grk. ὁρΦνό㰡, A orkäm, B orkaṁñe probably stand in ablaut relationship to Grk. , Goth. riqis, etc. cf. Walde-Pokorny 2.367; Benveniste, Hirt-Festschrift 2.236.

64 IF 50.229.

65 Schulze, Kl. Schr. 239; Benveniste, Hirt-Festschrift 229.

66 Pedersen, Groupement 39.

67 Begriff der Totalität 23, 35, 53, 60.

68 Toch. Gramm. 306 f.; Meillet, MSL 18.416. Holthausen (IF 39.65) takes the k as radical and connects with Grk. πυκνός ‘close, thick’, but does not envisage the forms in -nt-.

69 Sturtevant, Hitt. Gr. 91; Benveniste, Hirt-Festschrift 2.235.

70 Schrader, Reallex.2 2.645; Walde-Hofmann 283; Benveniste, op. cit. 234.

71 IF 50.6. But on the interrelationship of the forms cited, cf. Walde-Pokorny 1.148, 289; Walde-Hofmann 64 f.

72 Cf., for example, Meillet, Idg. Jahrb. 1.18; Schulze, Kl. Schr. 255 ftn. 1; Fraenkel, IF 50.7.

73 In his criticism of my paper on Tocharian vocalism (LSA meeting Chicago, December, 1936), and later by personal letter, Professor Sapir suggests the presence of a sort of u-epenthesis in Tocharian. That is, PIE *bhāĝhu-s > Pre-Toch. *pāku- > PToch. *pāuku- > *pauk- whence A pok-e, B pauk-e. This view is exceedingly fascinating for these particular words, in as much as it affords an explanation also for B obl. pokai, which may then represent the vocalism inherited from another IE case, e.g. dat. *bhāĝhewai, loc. *bhāĝhēu, *bhāĝhewi, etc.—provided, of course, that IE ā may give Toch. o.

74 Schrader, Reallex.2 1.443.

75 Meillet, MSL 18.386.

76 Cf. Holthausen, IF 39.65 without the B forms.

77 Meillet, JAs. 1911 (18).150; Schulze, Kl. Schr. 255 ftn. 4; Fraenkel, IF 50.7.

78 Manuscript Remains of Buddhist Literature Found in Eastern Turkestan, ed. by A. F. R. Hoernle, Oxford 1916.

79 Cf. Pedersen, Groupement 32 ftn., but for B ekalymi ‘subject’ (to which corresponds prob. A akälyme ‘zugewendet‘) cf. above, §3, 1 end.

80 Meillet, Idg. Jahrb. 1.14; MSL 18.20.

81 Meillet, JAs. 1911 (17). 456, MSL 18.20.

82 But so apparently Petersen, Lang. 11.197 with ftn. where he derives A from *ṇ-gn∂tos.

83 So Meillet in Lévi, MSS Remains 377.

84 Cf. Pedersen, Vgl. Gramm. d. kelt. Sprachen 1.46 ff., 2.7 f.; Thurneysen, Hdb. d. Altir. 493 f.