Peyrot 2016: Language contact in Central Asia: On the etymology of Tocharian B yolo 'bad' (original) (raw)

Language contact in Central Asia: On the etymology of Tocharian B yolo ‘bad’

Michaël Peyrot
Berlin-Brandenburg Academy of Sciences and Humanities

Abstract

Tocharian B yolo ‘bad, evil’ is often connected with Hittite idālu- ‘id.’, after a suggestion of Čop and Rasmussen. The similarities between the Tocharian word on the one hand and Khotanese yola- ‘falsehood’ and Old Turkish yavlak ‘bad, evil’ on the other would be due to chance or to borrowing from Tocharian. Here it is argued that we should rather return to a suggestion of Hansen, who assumed that both the Tocharian and the Khotanese word are borrowed from Turkish. More precisely, it seems that the Old Turkish word was borrowed into Khotanese and from there into Tocharian.

In this paper, the etymology of the Tocharian B word yolo ‘bad, evil’ is discussed. The main part will be devoted to synchronic points, and, based on Hansen (1940: 162), to possible borrowing relationships of this word, as it shows important similarities in form and meaning to Khotanese yola- ‘falsehood’ and Old Turkish yavlak ‘bad, evil’, both neighbouring languages. However, I will first treat a possible Indo-European etymology of yolo.

1. Indo-European connections

Jens Elmegård Rasmussen, whom we sadly honour with this volume, proposed an Indo-European etymology for Tocharian B yolo that is widely accepted today (1999: 147-148, originally 1984). Rasmussen was inspired by an etymological proposal of Čop (1975: 204-205), who connected yolo to, among others, Ved. ādhrá- ‘poor, needy’ and reconstructed *ēdhlo-. Also related, according to Čop, was Hitt. idālu- ‘evil, bad’, which he derived from ∗edhu{ }^{*} e d^{h} u. A crucial element of Čop’s derivation was the assumption of a problematic sound change ∗−dhl{ }^{*}-d^{h} l - to ∗−ll{ }^{*}-l l for Tocharian, which he promised to discuss in a future study.

Rasmussen’s etymology is much more convincing. It is based also on a correspondence between Tocharian B yolo and Hitt. idālu- ‘bad, evil’, but instead of Ved. ādhrá-, he connected it with Gr. óóovn ‘pain’, Arm. erkn ‘labour pains’ and OIr. idu ‘pain’, which are cognate according to Schindler (1978). The group of Gr. óóovn of course requires unaspirated ∗d{ }^{*} d rather than the aspirated ∗dh{ }^{*} d^{h} in Cop’s variant, and this works much better for Tocharian. While there are no parallels for Cop’s idea that aspirated ∗dh{ }^{*} d^{h} is lost before ∗l{ }^{*} l in Tocharian, it is clear that unaspirated ∗d{ }^{*} d is lost at least before ∗r{ }^{*} r and ∗u{ }^{*} u, and the latter context applies in this word. As a direct source for yolo, Rasmussen assumed a sákhā-type * h1h_{1} edyoló (i)(i), of which the nominative singular was generalised throughout the paradigm. This became first *edyoló, where the ∗d{ }^{*} d stood before ∗u{ }^{*} u and was lost. Then, through the Tocharian vowel changes, the immediate result * ewoló became * yowelō, and this form in turn underwent “umlaut and schwa deletion” to become *ywolo. Finally, the ww was lost before oo to yield the actually attested yolo.

Although the derivation of yolo from *edyoló takes many steps, it works quite well nevertheless. That is to say, there is virtually no way to reconstruct a Proto-Indo-European *edyoló on the basis of only Tocharian B yolo, which allows many more different sources. However, if the assumed preform is taken as the starting point, yolo is an acceptable outcome. The only difficulty that I see is the umlaut that took place in the intermediate form *yowelō: in view of

the wide variety of types of umlaut, the question is what kind of umlaut this was. Presumably, Rasmussen meant oo-umlaut of ∗e{ }^{*} e to yield oo, which is what Čop - more explicitly - assumed for his original version of the etymology with Hitt. idālu-.

The problem is that oo-umlaut does occur, and oo may also be the result of an umlauted ee, but there is no oo-umlaut of ee. Rather, oo-umlaut affects a preceding aa to become oo; or ee may be affected by a following uu or ww to become oo. The first change is found in the oo-presents of class 4, e.g. TB plontotrā ‘rejoices’ < *plantotr. The second type is found e.g. in TB or ‘wood’ from *eru, in turn from *doru. As far as I can see, neither of the two sound laws could have applied here: the root vowel was not ∗a{ }^{*} a, and the suffix contained no ∗u{ }^{*} u.

There are, however, more sound changes that yield Tocharian B o. Especially the ∗w{ }^{*} w in the sequence * we { }^{\text {we }} may have caused an ultimate oo. One possibility is that after aa-syncope had taken place in * yawelo, the combination * we in the resulting *ywelo turned to oo. A parallel for this sound change is poñ ‘say!’ from *pweñ (Peyrot 2010: 71-72). The question is, however, if the conditions of this sound law are met in yolo; possibly, it operated only in closed syllables (as suggested l.c.), while we evidently have an open syllable here. A second option is a contraction accross ww. A parallel for this change may be the oo in the final syllable of the preterite participles of the type tärkau, tärkos ‘let go’, from *tarkawes. In my view, the resulting oo derives from a contraction of ∗ae{ }^{*} a e (op. cit. 71-73), but other scholars would claim that the ww causes the final outcome to be rounded. Although it is not completely parallel, owe might also have yielded oo through contraction (cf. in particular Pinault 2008: 533). A third option is oo-umlaut of ∗u{ }^{*} u. This umlaut seems to be needed to explain e.g. TB klyomo ‘noble’ < *kleu-mōn and okso ‘ox’ < *uksōn.

Although none of the sound laws just referred to are exactly parallel, it is in my opinion difficult to exclude that any of them operated in yolo. The problems with Tocharian o-vocalism are complex and many of the sound laws have not yet been worked out in detail. It is therefore not warranted to discard this etymology because of the oo-vocalism in the first syllable.

Rasmussen’s etymology is widely accepted today. It is found, for instance, in Adams’ dictionary (1999: 513) and in NIL (p. 209; cf. further e.g. Winter 2011: 232-233). However, the approval is not universal. Notably, Kortlandt does not accept the connection of Greek óóv νη\nu \eta etc. with ∗h1ed{ }^{*} h_{1} e d-, the root for ‘eat’. He makes a strict distinction between the root ∗h1ed{ }^{*} h_{1} e d - and another root, ∗h3ed{ }^{*} h_{3} e d-, which means ‘bite’ (e.g. 2003: 7, 94-95). Within his framework, Tocharian B yolo cannot be connected with Greek óóv νη\nu \eta.

However, even if one does not accept Schindler’s reconstruction of the word for ‘pain’, Hitt. idālu- and Tocharian B yolo could belong together, and this part of Čop’s and Rasmussen’s etymology would still hold. Nevertheless, the evidence is certainly weakened, even more so since CLuw. has a cognate of Hitt. idālu- without −l-l - (next to one with −l−-l- ), which shows that the original formation to be reconstructed for Proto-Anatolian is *?eduo- ‘evil’ < * h1ed(k)uo−h_{1} e d^{(k)} u o-, according to Kloekhorst (2008: 420-422).

It seems worthwhile, therefore, to reconsider the possibility that Tocharian B yolo is not inherited from Proto-Indo-European, but perhaps borrowed from either Khotanese or from Old Turkish, as proposed by Hansen (1940: 162). However, first a closer look at the synchronic behaviour of yolo is necessary.

2. Synchronic points

Synchronically, yolo is a difficult word because it can be used as an adjective as well as a noun: Adams (1999: 512) distinguishes a noun yolo from a homophonous adjective. Also the relationship between the noun and the adjective is evaluated in different ways. For instance, Pinault

(2008: 318) argues that the noun is derived from the adjective, so that we would reconstruct an original adjective, while Hilmarsson (1987:36) argues for the exact opposite and derives the adjective from the noun. If we take a look at the forms in actual usage, we see that yolo is mostly used as an adjective. However, even though the noun is less frequent, the nominal use cannot be explained away and the existence of a homophonous noun is absolutely certain.

In most cases, yolo is used as an adjective: not only does it show concord for number and gender, it is also frequently used to modify nouns and only rarely independently. Especially frequent is the combination with yāmor ‘act, deed’, followed by palsko ‘thought, mind’. Compare:

The paradigm of this adjective runs as follows:

MASCULINE FEMININE
nom.sg. yolo yolyiya
obl. yolaim yolyiyai
gen. yolopi -
nom.pl. yolaiñ yolaina
obl. yolaim yolaina
gen. yolaināṃts yolainamts

Since the feminine of the adjectives never has a genitive singular (Peyrot 2012: 197-200), the paradigm is complete, an absolute rarity for Tocharian: all forms are attested.

Apart from its use as a regular adjective, yolo is strikingly frequent in compounds, where it always has the shape yolo. In these compounds, it is mostly used to render Sanskrit terms prefixed with duss−d u s_{s^{-}}‘bad’, e.g.

(compare B308b6 kārtse-were ‘of good smell’ Skt. sugandhini; Bernhard 1958: 188)

So far there has been no need to assume a homophonous noun. The noun is needed for occurrences of yolo as a direct object, when it should be in the oblique case in Tocharian. The most frequent combination with yolo in the role of an object is yolo yam- ‘do wrong, commit evil’, for instance:

In view of the frequent occurrence in Tocharian of fixed combinations of uninflected words with verbs, very similar to the English phrasal verbs, one could in the case of yolo yam- argue that this is also a fixed combination and possibly not the best evidence for productive use of a noun yolo. Parallel to the phrasal verbs lāre yam- ‘love’, apākārtse yam- ‘make manifest’ and

[1]

  1. 1maˉ1 m \bar{a} is added against the manuscript by Sieg (1938: 13); differently, CETOM (14 June 2013). ↩︎

erkatte yam- ‘treat badly’ discussed by Meunier (2013: 139-142), yolo could here be taken as the nom.sg.m. of an otherwise inflected adjective, incorporated into the verb. Also the agent noun combination yolo-yāmi ‘wrongdoer’ may blur the picture, as in this case the form yolo can be explained from the fact that it is a compound, parallel to the compounds mentioned above, where yolo is not inflected either.

Yet there is more evidence that yolo is a noun, purely morphological and therefore very solid: the form yolo also combines with secondary case suffixes. Normally, these secondary case suffixes are attached to the oblique case, which would be either yolaim or yolyiyai, but never yolo - except when yolo is a noun. Most frequent is the ablative yolomeṃ ‘from evil’ (5x); of the locative yolone and the perlative yolosa ‘through evil’ I found each one example.

It is impossible to discuss here in general the phenomenon that the nom.sg.m. of an adjective is used as a noun in Tocharian B, but it is highly unusual. Nevertheless, a good parallel can be adduced. The semantic opposite of yolo, the adjective kartse ‘good’, often found together with yolo to express ‘good and bad’ or ‘good or bad’, shows very much the same behaviour. Also kartse has a very distinct oblique, namely obl.sg.m. krent, obl.sg.f. kartsai, and it is likewise used as a noun, as can be seen from the combination of the nom.sg.m. kartse with secondary case suffixes. For kartse we also find the locative, once perhaps the causal, and further rather frequently the allative kartseśc ‘towards good’ (and its variants kärtsescc, kartseś, etc.; a quick count yields 27 examples). The existence of a noun kartse next to the homophonous adjective is further confirmed by the derived adjective kärtseṣse, which presupposes a nominal base. 2{ }^{2}

In passing, it may be pointed out that the statistics of the inflexional morphology of the words ‘good’ and ‘evil’ give a nice insight into the moral standards of the speakers of Tocharian B language: while kartse is frequently attested in the allative, yolo is rather found in the ablative.

Thus, yolo is a regular and very well attested adjective, whose nom.sg.m. is also used as a noun, and the same situation is found with its semantic opposite kartse ‘good’. Probably, the usage of these two words is somehow correlated, since they are so often used side by side: one could argue that the use of the nom.sg.m. of the one is analogical after the other. In the case of kartse, it is obvious that its use as a noun with an oblique singular that is identical to the nominative must be secondary: the behaviour of its ntn t-suffix in the inflected forms of the adjective is very well known and there is absolutely no reason to assume that kartse is something other than the nom.sg.m. However, yolo has forms that are difficult to interpret, but seem to suggest that the form yolo was originally used also outside the nom.sg.m. This in turn may be taken as an indication that the strange behaviour of kartse ‘good’ and yolo ‘bad’ started out with yolo and was then taken over by kartse.

The forms that suggest that yolo was in origin not restricted to the nom.sg.m. are the gen. sg.m. yolopi and the obl.sg.m. variants yolon and yoloynä. The gen.sg.m. yolopi is an unambiguous adjective form because the ending - pi is only found with adjectives; however, it is formed from the nominative yolo, not from the oblique yolaim, which would be regular. Compare for instance nom.sg.m. kartse, obl.sg.m. krent, gen.sg.m. kreñcepi, with the genitive formed from the stem of the oblique. The obl.sg.m. variants are both archaic hapax legomena:

2 For the sake of completeness, it may be noted that both have regularly derived abstract nouns: yolaiññe ‘evil, evil things’, and kärtsauñe ‘goodness, good work’.

Since these are only two forms, and they are not identical, it is conceivable that both are mistakes. However, both are found in clearly archaic and otherwise well-written and wellpreserved texts, so that one would first have to assume that they are genuine forms of the language. In my view, both indicate that the original obl.sg.m. of yolo was yolo, not yolaim. The first, yolon, must have the frequent oblique marker −n(−m)-n(-m) added; the second, yoloynä, seems to be a blend of the more archaic yolon and the classical form yolaim. 4{ }^{4}

The paradigm of yolo also presents problems on a different morphological level: no such adjective class exists. In тEB, the paradigm is given a class number, but yolo is the only member ( 231231\231 231, class II.2). On the other hand, a large number of nouns follows exactly the pattern that yolo follows in the masculine. This is precisely the argument that was put forward by Hilmarsson (1987:36) in order to prove that yolo was originally a noun, and that its use as an adjective is secondary.

3. Khotanese yola- and Old Turkish yavlak

Although Hilmarsson admits that his assumption that yolo was originally a noun might seem ad hoc, he points out (l.c.) that Khotanese yola- ‘falsehood’ is also a noun and not an adjective. His comparison with Khotanese can only be meaningful if it is assumed that both words are related in some fashion. Indeed, a connection between the two words seems likely, because they are obviously close in form as well as in meaning.

The meaning of Khot. yola- is not easy to gather from the texts, because it mostly occurs in lists of evil acts where it is very clear that it is something negative, but its exact meaning is in many passages difficult to establish. Thus, Bailey first (1967: 286; also Skjærvø 2004: II, 329a) translated it as ‘evil act’, but later (1979: 343b) as ‘falsehood’. It is attested as the translation of Skt. mrṣā ‘untrue, wrong’ and śāthya ‘wickedness, deceit’.

An inner-Iranian etymology for the word was provided by Bailey (1979:343b5). According to him, an original yaula- 6{ }^{6} reflects *iyaßla-, and this in turn *vidab(a)la-. A verb *vi-dab’deceive’ is indeed attested in Middle Iranian with MPth. wdyfs- ‘be deceived’, MPrs. wyyb-. However, he needs to assume that the prefix is a dialectal variant ∗i{ }^{*} i - for ∗vi{ }^{*} v i-; and he thinks that the original root shape * yab- is proved by OTr. yavlak, borrowed, according to Bailey, from Khotanese.

However, it is not likely that Old Turkish yavlak ‘bad, evil’ is a borrowing from Khot. yola-. First of all, it is unclear where the final −k-k in Turkish should have come from. Second, and more importantly, yavlak has an excellent inner-Turkish etymology. Although the underived root yav- is not attested as such, a number of non-trivial cognates are found (for this argument, see already Hansen 1940: 162), and all of these are well attested from early stages:

[1]

  1. 3 In the edition, this form is given as yoloytā, but it really has to be read yoloynā, i.e. <yoloyną> (Peyrot 2013: 317), which is clear from high-resolution photographs that have recently been made available.
    4 The spelling of yoloynä and yolaim conceals that the relevant difference is in fact only the oo vs. aa of the second syllable: yoloynä can be analysed as yolo-ynä, where ynä corresponds to the final of yolaim.
    5 And already (1946: 203). For further references to earlier literature see also Van Windekens (1971: 450-451), who offers an Indo-European etymology of yolo that fails to convince.
    6 The word is both spelled with oo and aua u in the first syllable (in Late Khot. even with a¨\ddot{a} ), but this in itself does not prove older * aua u, because oo and aua u are mostly used indiscriminately. ↩︎
yaviz 'bad'
    \(\rightarrow\) denominative yavizla- etc.
yavri- 'become weak'
    \(\rightarrow\) causative yavrit- etc.
yavgan 'coarse, unsympathetic'

Of these formations, at least yaviz (attested already in the Orkhon inscriptions) and yavri- are likely to be old. As concerns the first word, Erdal (1991: 324) notes that the suffix " −(X)z-(X) z appears to have been rather obsolescent in Old Turkic: A number of lexemes evidently of this formation have no living bases, or have lost semantic contact with them." Furthermore, Tekin (1969: 58,65 ) points out that the correspondence between the verb yavrı- and the adjective yaviz proves an archaic etymological relationship for which he reconstructs * yavır. This * yavır was the direct source of yaviz, while derivation with the denominal suffix −1−-1- yielded yavrı-. A further derivative yaman ‘bad’ < *yav-man is adduced by Tezcan (1981: 72), but this word is attested relatively late (from Kāšgarī onwards), while good and early examples of the suffix -man are difficult to find (Erdal 1991: 387). These formations all point to a verb * yav-, possibly meaning ‘be bad, be weak’ (Clauson 1972: 871a; and see the relevant derivatives). For yavlak itself, Clauson (op. cit. 876b) assumes a homophonous noun * yav besides, since yavlak seems to be an adjective (and also a noun, see Erdal 2004: 143) in −k-k derived from a denominal verb in -la: noun * yav →\rightarrow verb * yavla- →\rightarrow adjective yavlak.

In view of all these cognates, there is no reason to assume that Old Turkish yavlak is a borrowing; 7{ }^{7} all inner-Turkish evidence rather clearly points to an inherited word. To this must be added, obviously, that the inner-Iranian etymology of Khotanese yola- offered by Bailey is not without problems. 8{ }^{8}

Thus, with Hansen (1940: 162), it seems much more attractive to assume that Old Turkish yavlak was not borrowed from, but into Khotanese. On the formal side, this is unproblematic. As concerns the exact shape of the Old Turkish word, the pronunciation of yavlak was probably approximately [yawla:q]: the vv in the first syllable may have been pronounced as a semivowel [w]; the vowel of the second syllable must have been long, as shown by Oguz/Kıpčak yavlāk (Clauson 1972: 876b-877a); and the final −k-k was pronounced as a uvular [q]. On the basis of this phonetic shape, Khotanese yola- can be explained without difficulty. Since the vv in the first syllable of yavlak was probably pronounced as a semi-vowel, Khotanese may have borrowed the first syllable with a diphthong aua u. In this case, the synchronic result was oo at any rate, because this monophthongisation took place very early in Khotanese. 9{ }^{9} However, even if the vv was borrowed as a consonant, the final outcome would have been oo : the groups ∗−aβl{ }^{*}-a \beta l - or *-afl- would also have become -aul- and then ol- (Emmerick 1989: 212). Also the long vowel of the second syllable may be reflected in the Khotanese word. Although it is everywhere given

[1]

  1. 7 According to Erdal (1991: 388), “There are indications that the root came from Iranian in one form or another or, indeed, through more than one channel.” However, he does not specify which indications he is referring to. Tezcan (1981: 72) does cite the idea of Clauson (1972: 871b) that a Turkish word yava ‘foolishness’ is borrowed from Persian yāva ‘absurd, foolish’, but in the end he rather prefers the Persian word to be originally Turkic, precisely because of the group of obviously related words cited above. The only other indication of Iranian origin seems to be Bailey’s explanation of Khot. yola-. And also Erdal admits, “The lexemes found in Turkic can, however, very well be related among themselves by folk-etymology if not genuine inner-Turkic derivation.” (l.c.).
    8 In a supplement Bailey writes on yola-, “For Turkish, beside yablaq ‘bad’, note also yabīz, yawuz, yawrī-” (1979: 511a-b). This could perhaps be interpreted as a critical note to his own etymological suggestion, but in any case he does not admit it explicitly. Skjærvø (2004: 11, 329a) gives the etymology of yaula- as unknown.
    9 That is, of secondary aua u that has arisen in the prehistory of Khotanese. Old Iranian *au develops into Khotanese uˉ\bar{u} (Emmerick 1989: 212). ↩︎

as yola-, the Old Khotanese nom.pl. yole (the most frequent form) could as well be from a stem yolà-. 10{ }^{10} However, this point must not be overvalued. First, the stem class remains uncertain, and second, the long stem vowel aˉ\bar{a} of the aˉ\bar{a}-stems does not surface as aˉ\bar{a} anywhere in the paradigm even in the oldest material. Finally, the final −k[4]-k[4] may simply have been lost in the borrowing process, as this was certainly a final that was not found in Khotanese. But if the word was borrowed with the final, the kk may have been lost afterwards, as it was generally lost in this position (after a long vowel and in a final syllable) in Pre-Khotanese (Emmerick 1989: 214).

By contrast, Tocharian B yolo is not so easily derived from Old Turkish yavlak, despite the obvious similarity: there is no reason to suppose that the −k-k would be lost, and it is questionable whether the initial syllable yav- would turn up in Tocharian B as yo-. It seems, therefore, much more attractive to assume that yolo was borrowed from Khotanese yola-. In that case, it is possible that aua u was already monophthongised in Khotanese before the word was borrowed, or at least it sounded so different from Tocharian au that it was borrowed as oo. The final of the Khotanese word was apparently also the closest to Tocharian B -o, itself a frequent final in that language. 11{ }^{11} Probably, Khotanese yola- was at first borrowed into Tocharian B as a noun yolo. Later this noun was inflected, and it received, for instance, a pl. yolaiñ, while archaic traces of the uninflected yolo are preserved synchronically. Finally the word became an adjective when also a fem. yolyiya with the corresponding paradigm was created.

4. The Proto-Tocharian word for ‘bad, evil’

As a closing note, we may ask ourselves whether the Proto-Tocharians had any knowledge of evil at all, if the Tocharian B word for it was borrowed. As it happens, Tocharian A provides us with two words, which both seem to go back to Proto-Tocharian.

The first and most frequent is Tocharian A omäskem ‘bad, evil’ (adj./noun). Hilmarsson (1991: 131) derives this word from mäsk- ‘be’. The initial o- would be a reflex of the negative prefix ∗en{ }^{*} e n-, while the final eme m would be an abstract suffix: ∗{ }^{*} en-moske-ñne. As a semantic parallel and a possible model for the formation, Hilmarsson adduces Skt. a-sat ‘evil, evil person’.

The second Tocharian A word is more difficult: umpar ‘evil’ (noun; see Carling 2009: 68a). Since all words beginning with uu - are very clear loan words, one would think that umpar is a loan word, too. However, as a source for borrowing does not present itself, another possibility is to see umu m - as an irregular development of original om-. That umpar is not borrowed but inherited is suggested in particular by the comparison with kem-par ‘wrongly’: the common element in um-par and kem-par suggests that both are inherited words. This in turn requires the assumption of a special development to explain the initial uu - of umpar, precisely since all other instances of initial uu - are found in loan words. A plausible connection for the element um- is indeed found in TB aume ‘misery’ (Van Windekens 1976: 539).

Thus, omäskem is built with genuine Tocharian elements, but not necessarily of ProtoTocharian age, while umpar seems to go back to Proto-Tocharian. Therefore, the speakers of Proto-Tocharian will have had knowledge of evil.

[1]

  1. 10 The Middle Khotanese (according to Skjærvø 2002: lxxiv) nom.sg. yaula IOL Khot 165/1b21 and the Late Khotanese acc.sg. yäla and ins.-abl. yäla jsa are not diagnostic.
    11 Theoretically a long −aˉ-\bar{a} might have been particularly close to Tocharian B oo, as long ∗aˉ{ }^{*} \bar{a} has developed into TB oo (Peyrot 2010: 71-73). However, this can hardly be the correct explanation here, since the stem-final −aˉ-\bar{a} does not appear in the paradigm, as remarked above. ↩︎

References

Adams, Douglas Q. 1999. A dictionary of Tocharian B. Amsterdam & Atlanta: Rodopi.
Bailey, Harold W. 1946. Supplementary note to Asica. Transactions of the Philological Society 45. 202-206.

Bailey, Harold W. 1967. Khotanese texts VI: Prolexis to the Book of Zambasta. Cambridge: Cambridge University.
Bailey, Harold W. 1979. Dictionary of Khotan Saka. Cambridge: Cambridge University.
Bernhard, Franz. 1958. Die Nominalkomposition im Tocharischen. Diss. Göttingen.
Carling, Gerd. 2009. Dictionary and thesaurus of Tocharian A. Part 1: A-J. In collaboration with Georges-Jean Pinault and Werner Winter. Wiesbaden: Harrassowitz.
CETOM = A comprehensive edition of Tocharian manuscripts. www.univie.ac.at/tocharian/.
Clauson, Gerard. 1972. An etymological dictionary of pre-thirteenth-century Turkish. Oxford: Clarendon.
Čop, Bojan. 1975. Studien im tocharischen Auslaut I. Ljubljana: Univerza v Ljubljani.
Emmerick, Ronald E. 1989. Khotanese and Tumshuqese. In Rüdiger Schmitt (ed.), Compendium linguarum Iranicarum, 204-229. Wiesbaden: Reichert.
Erdal, Marcel. 1991. Old Turkic word formation. A functional approach to the lexicon. Wiesbaden: Harrassowitz.
Erdal, Marcel. 2004. A grammar of Old Turkic. Leiden & Boston: Brill.
Hansen, Olaf. 1940. Tocharisch-iranische Beziehungen. Zeitschrift der Deutschen Morgenländischen Gesellschaft 94. 139-164.
Hilmarsson, Jörundur G. 1987. The element -ai(-) in the Tocharian nominal flexion. Die Sprache 33.34-55.
Hilmarsson, Jörundur G. 1991. The nasal prefixes in Tocharian. A study in word formation. Reykjavík: Málvísindastofnun Háskóla Íslands.
Kloekhorst, Alwin. 2008. Etymological dictionary of the Hittite inherited lexicon. Leiden & Boston: Brill.
Kortlandt, Frederik H. H. 2003. Armeniaca: Comparative notes. Ann Arbor.
Meunier, Fanny. 2013. Typologie des locutions en yām- du tokharien. Tocharian and IndoEuropean Studies 14. 123-185.
NIL=N I L= Dagmar S. Wodtko, Britta Irslinger & Carolin Schneider. 2008. Nomina im Indogermanischen Lexikon. Heidelberg: Winter.
Peyrot, Michaël. 2010. On the formation of the Tocharian preterite participle. Historische Sprachforschung 121 (2008). 69-83.
Peyrot, Michaël. 2012. The Tocharian A match of the Tocharian B obl.sg. ai. Tocharian and Indo-European Studies 13. 181-220.
Peyrot, Michaël. 2013. The Tocharian subjunctive. A study in syntax and verbal stem formation. Leiden & Boston: Brill.
Pinault, Georges-Jean. 2008. Chrestomathie tokharienne. Textes et grammaire. Leuven & Paris: Peeters.
Rasmussen, Jens Elmegård. 1984 Miscellaneous problems in IE languages III. Arbejdspapirer udsendt af Institut for Lingvistik, Københavns Universitet 4. 135-149.
Rasmussen, Jens Elmegård. 1999. Selected papers on Indo-European Linguistics. Copenhagen: Museum Tusculanum.
Schindler, Jochem. 1978. Armenisch erkn, griechisch óóóv η\eta, irisch idu. Zeitschrift für Vergleichende Sprachforschung 89. 53-65.

Sieg, Emil. 1938. Die Kutschischen Karmavibhañga-Texte der Bibliothèque Nationale in Paris (zu Prof. Sylvain Lévi’s Ausgabe und Übersetzung). Zeitschrift für Vergleichende Sprachforschung 65. 1-54.
Skjærvø, Prods O. 2002. Khotanese manuscripts from Chinese Turkestan in the British Library. London: The British Library. [Reprinted with corrections 2003.]
Skjærvø, Prods O. 2004. This most excellent shine of gold, king of kings of sutras: The Khotanese Suvarnabhāsottamasūtra. Cambridge (MA): Harvard University.
TEB = Wolfgang Krause &\& Werner Thomas. 1960. Tocharisches Elementarbuch, I. Grammatik. Heidelberg: Winter.
Tekin, Talât. 1969. Zetacism and sigmatism in Proto-Turkic. Acta Orientalia Hungarica 22. 51−8051-80.
Tezcan, Semih. 1981. Kutadgu Bilig dizini üzerine. Belleten 45. 23-78.
Thomas, Werner. 1958. Zum Ausdruck der Komparation beim tocharischen Adjektiv. Zeitschrift für Vergleichende Sprachforschung 75. 129-169.
Van Windekens, Albert J. 1971. Études de phonétique tokharienne XVI: Nouvelles notes sur le préfixe intensif A aˉ−\bar{a}- B aa-, etc. Orbis 20. 447-453.
Van Windekens, Albert J. 1976. Le tokharien confronté avec les autres langues indo-européennes, vol. I. Louvain: Peeters.
Winter, Werner. 2011. Vowel lengthening before distinctively voiced consonants in Tocharian. Tocharian and Indo-European Studies 12. 221-238.

Etymology and the European Lexicon

Proceedings of the 14th 14^{\text {th }} Fachtagung
der Indogermanischen Gesellschaft, 17-22 September 2012, Copenhagen

Edited by
Bjarne Simmelkjær Sandgaard Hansen, Benedicte Nielsen Whitehead, Thomas Olander and Birgit Anette Olsen