Notes on Verbal Governing Compounds in Tocharian (2018)

Proceedings of the 28th Annual UCLA Indo-European Conference Los Angeles November 11th and 12th, 2016 Edited by David M. Goldstein Stephanie W. Jamison Brent Vine with the assistance of Angelo Mercado HEMPEN VErLAG BrEMEN 2018 Cover illustrations: Wheeled vehicles depicted on Bronze Age vessels and petroglyphs, from Kuzmina, E. E. (2007) The Origin of the Indo-Iranians, Leiden, Brill; Fig. 34. reproduced with the kind permission of the author. Bibliografische Information Der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http://dnb.d-nb.de abrufbar. ISBN: 978-3-944312-56-9 www.hempen-verlag.de © 2018 Hempen Verlag, Bremen Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Umschlaggestaltung: Detemple Design, Igel b. Trier Gedruckt auf alterungsbeständigem Papier Printed in Germany Notes on Verbal Governing Compounds in Tocharian* HANNES A. FELLNER Leiden University 0. Tocharian nominal morphology is still understudied, especially from the perspective of historical comparative linguistics. Thus one of the desiderata of Tocharian nominal morphology is the treatment of compounding. Bernhard’s (1958) unpublished dissertation is still the standard monograph on this topic from a synchronic point of view. Since Tocharian texts are written without word boundaries, compounds need to be identified by other means. Bernhard (1958:21 and 75; see also Marggraf 1970:64–81) established two criteria for the identification of compounds: stress in Tocharian B and the morphological shape of the first compound member. Stress distribution in Tocharian B can be determined from vowel alternations affecting the underlying central vowels /ə/ and /a/, which show different realizations depending on the placement of stress. Underlying /ə/ is realized as [ʌ] 〈a〉 when stressed and as [ə] 〈ä〉 when unstressed; underlying /a/ is realized as [a] 〈ā〉 when stressed and [ʌ] 〈a〉 when unstressed. These patterns are presented in Table 1. Table 1. Tocharian B central vowels a Underlying vowel /ə/ /a/ a Stressed allophone 〈a〉 [ʌ] 〈ā〉 [a] Unstressed allophone 〈ä〉 [ə] 〈a〉 [ʌ] In this and the following two tables the underlying forms are given in slashes in IPA only, whereas the surface representations are given in standard Tocharian transcription with IPA in square brackets. The vowel 〈ä〉 [ə] is always unstressed (at least in Classical Tocharian B), the vowel 〈ā〉 [a] is always stressed. The stress of the vowel 〈a〉 [ʌ] can be inferred * I would like to thank Laura Grestenberger, Jay Jasanoff, Bernhard Koller, Melanie Malzahn, Alan Nussbaum, and Georges-Jean Pinault for discussion of various topics of this paper and the editors, especially David Goldstein and Brent Vine, for comments and suggestions. The usual disclaimer applies. David M. Goldstein, Stephanie W. Jamison, and Brent Vine (eds.). 2018. Proceedings of the 28th Annual UCLA Indo-European Conference. Bremen: Hempen. 53–69. 54 Hannes A. Fellner from alternations it displays across morphologically related forms (both innerand interparadigmatically). The basic descriptive generalization for Tocharian B stress is that disyllabic words bear the accent on the initial syllable, whereas polysyllabic words usually bear the accent on the second syllable (Krause 1952:10; cf. Krause and Thomas 1960:43; see also Marggraf 1970), as illustrated in Table 2. Table 2. Tocharian B stress pattern Simplex Compound TB surface form tarya [ˈtʌrja] täryā-yäkne [tərjˈajəkne] Underlying representation /tərya/ /tərya/ + /yəkne/ Gloss NOM.SG.F ‘three’ ‘threefold’ The basic stress pattern leads to the surfacing of the underlying vowel not realized in simplex words that pertains to the stem of the first compound member. This is illustrated in the following examples. Table 3. Vowels in simplex words vs. first compound members Surface form TB ost [ˈost] TB yasa [ˈjʌsʌ] TB osta-ṣmeñca [ostˈʌʂmeɲʈ͡ʂa] TB ysā-yok [jsˈajok] TA pältsäk a TA wäs TA pälska-pāṣe TA wsā-yok a Underlying representation /ostə/ /yəsa/ /ostə/ + /ṣmeñca/ /yəsa/ + /yok/ /pəlskə/ /wəsə/ /pəlskə/ + /paṣe/ /wəsa/ + /yok/ Gloss ‘house’ ‘gold’ ‘householder’ ‘gold-colored’ ‘thought’ ‘gold’ ‘protecting thought’ ‘gold-colored’ The t in pältsäk is an emergent stop like the d in Greek GEN.SG.M ἀνδρός from underlying /anros/ to NOM.SG.M ἀνήρ (for the concept of emergent stop, see Ohala 1997). If stress cannot be determined due to lack of vowel alternation in Tocharian B or, generally, in Tocharian A, and stem vowels are inconclusive or absent, it is only possible to identify compounds as such if stems are involved that already have been classified as parts of existing compounds based on the aforementioned criteria, e.g., TA rī-pāṣe ‘protecting the city’ based on TA pälska-pāṣe ‘protecting thought’. Not all compounds in Tocharian have been identified so far. However, Notes on Verbal Governing Compounds in Tocharian 55 one subset of compounds is fairly well established, namely the verbal governing compounds (VGC) with governing second member (SM).1 1. The VGCs were first recognized and described by Bernhard (1958:134–7) and systematically treated by Malzahn (2012). Malzahn identifies two main types of VGCs, (1) a type that ends in TB -a TA -∅ (TB -rita-type; TB yolo-rita ‘seeking evil’: TB ritā- ‘seek’; TA ṣotre-lyāk ‘seeing signs’: TA läkā- ‘see’) and (2) a type that ends in TB -i TA -e (TB -yāmi-type; TB yolo-yāmi ‘evil doing’: TB yām‘make, do’; TA rī-pāṣe ‘protecting the city’: TA pās- ‘protect’). According to Malzahn, the B-rita-type is synchronically built to verbal roots with a-character and the B-yāmi-type to verbal roots without a-character. The acharacter of Tocharian roots is comparable to the seṭ and aniṭ distinction of Sanskrit roots. Synchronically, whenever a root forms a class V or VI subjunctive, the root has a-character (see Malzahn 2010:24 with literature). Compare TB ritā-: subjunctive V 3SG.MID rita-tär vs. TB yām-: subjunctive I+II 3SG.MID yām-tär. Similar to the i in Sanskrit seṭ-roots, the a of roots with a-character in Tocharian ultimately goes back to a vocalized laryngeal. However, the picture was blurred in Tocharian by several changes in the verbal system, so that roots with acharacter do not necessarily go back to roots which ended in a laryngeal (see Jasanoff 2013 with earlier references). Malzahn argues that the B-rita-type is diachronically descended from inherited compounds with SM in *eh2-abstracts or root compounds with second members that ended in a laryngeal, whereas the B-yāmi-type is an inner-Tocharian innovation based on PIE root compounds. While Malzahn’s work is an important contribution to the understanding of the VGCs in Tocharian, it is now possible to add previously overlooked synchronic and novel diachronic evidence for a fresh look at these compounds. Building on Malzahn’s work, the goal of this paper is to shed light on the synchronic patterning and diachronic affiliation of both VGC types. 2. In this section, I present VGCs in their context to clarify their meaning and syntactic properties, which were not treated by Malzahn (2012) in depth. Type 1 VGCs end in TB -a TA -∅: 1 There are no traces of verbal governing compounds with governing first member as of yet. 56 Hannes A. Fellner TB -pälk (: pälkā- ‘see’) (1) TA keṃ-pälk (: keṃ ‘false’) CEToM, A 62b6 keṃ-pälkāsaṃ mar cmimträ false-see.LOC.PL PROH born.1PL.OPT.MID May we not be reborn among heretics! TB -rita (: ritā- ‘seek, long for’) (2a) TB kärtse-rita (: kartse ‘good, favor, salvation’) CEToM, PK NS 29b3f. lantuñeṣṣe ṣecake kärtse-rita säsweṃtse royal.NOM.SG lion.NOM.SG good-seek.NOM.SG lord.GEN.SG The royal lion is the one who seeks welfare of the lord/is the welfare seeker of the lord. Stress clearly indicates that kärtse-rita is a compound since the simplex of the first member is kartse. If the genitive säsweṃtse really depends on the first member of the compound—which is not entirely clear from the context of this sentence—kärtse-rita säsweṃtse would be similar to cases like ModE. snow removal of streets or NHG Absturzverursacher des Computers [crash-perpetrator.NOM.SG the.GEN.SG computer.GEN.SG] ‘perpetrator of the computer’s crash’.2 (2b) TB yikne-rita (: yakne ‘way, manner’) CEToM, B 31a3 kuse yikne-ritañ sosoyoṣ […] who way-seek.NOM.PL satisfied.NOM.PL Those who are longing for the (right) way are satisfied … (2c) TB yolo-rita (: yolo ‘evil’) CEToM, IOL Toch 8b2 mār-ñäktesa warñai cwī yolo-ritañ ///3 god-Māra.PERL.SG et cetera this.GEN.SG evil-seek.NOM.PL To God Māra et cetera those seeking evil for this one … 2 3 For the problem surrounding arguments of non-heads see Härtl 2013 with literature. /// indicates a damaged edge of a manuscript fragment. Notes on Verbal Governing Compounds in Tocharian 57 TB -lyaka TA -lyāk (: TB TA läkā- ‘see’) (3a) TB läkle-lyaka (: lakle ‘pain, suffering’) CEToM, PK AS 7Ea6 […] wnolmi maiwaññesāk warñai entwek läkle-lyakāñ being.NOM.PL youth.PERL.SG with then pain-see.NOM.PL mäskentär be.3PL.PRS.MID … the beings are then the ones who see suffering starting from youth. (3b) TA ṣotre-lyāk (: ṣotre ‘sign, mark, manifestation’), CEToM, A 310b5 ṣomaṃ nu ṣotre-lyākāñ ṣeñc some but sign-see.NOM.PL be.3PL.PRS.ACT But some were seeing signs. TB -tsaika TA -tsek (TB TA tsikā- ‘form, shape’) (4a) TB lwaksā-tsaika (: lwāke ‘pot, vessel’, pl. lwāksa) CEToM, B 3a2 kos tsaikaṃ lwaksā-tsaik=4 aścemaṣṣeṃ how many shape.3SG.SUBJ.ACT pot-form.NOM.SG earthen.OBL.PL bhājanta kautalñ= āke po toṃ vessel.OBL.PL breaking.OBL.SG end.OBL.SG all this.NOM.PL However many earthen vessels the potter will shape these are all breaking in the end …5 Even though the first member of lwaksā-tsaika is inflected, stress again indicates that it is a compound. Inflection in first members of compounds, especially for plural, is not uncommon, cf. ModE. lice-infested, arms trade etc. or NHG Rinderwahn [cow.GEN.PL-craze.NOM.SG] ‘mad cow desease, i.e., bovine spongiform encephalopathy’, Waffenschmuggel [arm.ACC.PL-smuggling.NOM.SG] 4 5 = indicates sandhi. Cf. the corresponding passage from the Sanskrit Udānavarga—an early Buddhist topically organized collection of the utterances of the Buddha and his disciples—Uv. 1.12: yathāpi kumbhakāreṇa mṛttikā bhājanaṃ kṛtam | sarvaṃ bhedana paryantam evaṃ martyasya jīvitam “It is with the life of mortal man as with the shining vases made from clay by the potter—they all finish by being destroyed.” 58 Hannes A. Fellner ‘weapons smuggling’ etc. A similar formation is also found in Tocharian A in the next example. (4b) TA kuntis-tsek (: kunti ‘bowl, pot’, pl. kuntis) CEToM, A 19b5f. piṣṣ ānand ptāñäkte rakeyo bṛhadyuti go.2SG.IMP.ACT Ānanda Buddha-god word.INST.SG Bṛhadyuti kuntis-tsekānac peṃ pot.OBL.PL-form.ALL.SG speak.2SG.IMP.ACT Ānanda, go and speak with the word of the Buddha god to Bṛhadyuti, the potter. Type 2 VGCs end in TB -i TA -e: TB -ākṣi (: āks- ‘announce, proclaim’) (5) TB pelaikn-ākṣi (: pelaikne ‘law’) CEToM, B 19a2 pelaikn-ākṣiṃ karttse palkas law-proclaim.OBL.SG good look.2SG.IMP.ACT Look good at the one proclaiming the law! TB -aiyśi (: aik- ‘know, recognize’) (6) TB po-aiyśi / poyśi (: po ‘all’) CEToM, B 22b2 yāmṣane poyśi vyākarīto ce͡ u do.3SG.PRT.ACT all-know.NOM.SG prophecy.OBL.SG this.OBL.SG The all-knowing one made this prophecy … TA -kämṣe (: kän- ‘come about, occur’) (7) TA ākāl-kämṣe (: ākāl ‘wish’) CEToM, A 23a5 ākāl-käṃṣeśśi ākāl eṣant wish-fulfill.GEN.PL wish.OBL.SG giving.NOM.SG (You) who grant the wish of the ones fulfilling the wishes … TB -näkṣi (: näk- ‘destroy’) (8) TB läkle-näkṣi (: lakle ‘pain, suffering’) CEToM, B 229b4 läkle-näkṣi säkw-aiṣṣeñcai käṣṣi […] pain-destroy.VOC.SG fortune-giving.VOC.SG teacher.VOC.SG O teacher who destroys pain and gives fortune … Notes on Verbal Governing Compounds in Tocharian 59 TA -pāṣe (: pās- ‘protect’) (9a) TA pälska-pāṣe (: pältsäk [TB palsko] ‘thought’) CEToM, A 318/9A6 /// lāntsac lyutār potaträ k͡ uyalte queen.ALL.SG more honor.3SG.PRS.MID because pälska-pāṣe wrasom /// thought-protect.NOM.SG being.NOM.SG … honors the queen because a thought-protecting being … (9b) TA rī-pāṣe ‘city-protecting’ (: ri ‘city’) CEToM, A 253a8 täm pälkoräṣ rī-pāṣe śodhane this.OBL.SG seeing.ABL.SG city-protect.NOM.SG Śodhana.NOM.SG yakäṣ upaśodhanenac träṅkäṣ Yakṣa.NOM.SG Upaśodhana.ALL.SG say.3SG.PRS.ACT After having seen this, the city-protecting Yakṣa Śodhana says to Upaśodhana … TB -pilṣi (: pil- ‘listen, prick up one’s ears’) (10) TB klausa-pilṣi ‘ear-pricking’ (: klautso ‘ear’, du. klau(t)sa-) CEToM, IOL Toch 246a4 se ṣamāne śilnāntäṃ wetāntäṃ which.NOM.SG monk.NOM.SG quarreling.GEN.PL fighting.GEN.PL ṣamāneṃts klausa-pilṣi kalträ monk.GEN.PL ear-pricking.NOM.SG stand.3SG.PRS.MID pāyti pātayantika-sin.OBL.SG Whichever monk stands next to quarreling and fighting monks pricking his ears (commits) a pātayantika-sin. Furthermore, there is TB käryor-pläṅṣi (: karyor ‘buying; trade, commerce’; pläṅk(ā)- ‘be for sale’) without context in the fragment CEToM, IOL Toch 129b5 6 Cf. also Malzahn and Fellner (2015) for the edition of these fragments. 60 Hannes A. Fellner and TB yolo-yāmi (: yolo ‘evil’; yām ‘do’) without context in the fragment CEToM, B 265.2. As nominals derived from verbs the Tocharian VGCs are similar to participles and show some of the syntactic properties and functions associated with this word class, a feature typical for verbal governing compounds across IE languages.7 Type 1 in TB -a TA -∅ is used attributively in (2a) and (2c), predicatively in (3a, b), and substantivally in (1), (2b), (4a), and (4b). The latter examples show that the substantival use of TB -tsaika TA -tsek was lexicalized as the name of a profession, a phenomenon also common among VGCs in other IE languages.8 Type 2 in TB -i TA -e is equally used substantivally (5, 6, 7) and attributively (8, 9a, 9b, 10). It is important to note that type 1 in TB -a TA -∅ resembles the other agentive formations in TB -a TA -∅ in syntax, function, and inflection. These agentives consist of TB aknātsa TA āknats subst. ‘fool’, adj. ‘ignorant’, formations in TB -ntsa (e.g., TB wapāntsa ‘weaver’: wāpā- ‘weave’), formations in TA -nt (e.g., TA pekant ‘painter’: pikā- ‘paint, write’, formations in TB -nta TA -nt (e.g., TB kauṣenta TA koṣant ‘killing’: TB kau- TA ko- ‘kill’), the nt-participle in TB -ñca TA -nt (e.g., TB preñca TA prant ‘bearing, carrying, taking’: TAB pär‘bear, carry, take’), and formations in TB -uca (e.g., TB kärstauca ‘cutting’: kärstā- ‘cut’). All of these inflect in the same way and can have masculine and feminine referents: nom. sg. TB -a TA -∅, obl. sg. TB -ai TA -āṃ, nom. pl. TB -añ9 TA -āñ, obl. pl. TB -aṃ TA -ās. 3. Turning to the affiliation of the VGCs with the verbal system, the earlier proposal by Malzahn (2012) must be modified. Malzahn argues that the B-ritatype is built to verbal roots with a-character and the B-yāmi-type to verbal roots without a-character. However, a closer look shows that the two types of VGCs are conditioned by verbal stem formation. Table 4 presents a summary of the da- 7 8 9 Cf. Ved. -bhārá- ‘carrying’, attributive in bhūribhāra- [much-carrying] ‘carrying plenty’ RV 1.164.13, substantive, in muṣkábhāra- [testicle-carrying] ‘bearing testicles’ RV 10.102.4; Gk. βουληφόρος [counsel-carrying] ‘counsel bearing’ attributively in Il. 1.144 and 12.414, substantivally, in Il. 5.180 and Il. 7.126; Lat. lūcifer [light-carrying] ‘bringing light’ attributively, e.g., in Lucr. 5.726, substantivally (as the planet Venus), e.g., in Varr. R. 3.5.17. Cf. Skt. ayaskāra- [iron-making] ‘smith’; Gk. δρυτόμος [wood-cutting] ‘woodsman’; Lat. agricola [field-tending] ‘peasant’ (cf. also Gk. βουκόλος [cow-tending] ‘herdsman’). A specifically feminine nom. pl. -ana is so far only attested once and is probably analogically based on the productive feminine inflection in nom. sg. -a, obl. sg. -ai, nom./obl. pl. -ana. Notes on Verbal Governing Compounds in Tocharian 61 ta. Type 1 in TB -a TA -∅ is generally associated with athematic verbal formations. There is a straightforward connection of the compound second member and the present formation, except in the case of TB -lyaka TA -lyāk (cf. pret. I stem TB lyak-), for which see below (§5, pp.64–6). Type 2 in TB -i TA -e is not just built to roots without a-character, but systematically connected to thematic verbal formations. TB -aiyśi, TA -pāṣe, TB -yāmi are associated with simple thematic formations, the rest of the second compound members of this class with complex thematic verb stems (class VIII < *-se/o-, IX < *-sk̑ e/o-, and XI with some innerTocharian remodeling ultimately also < *-sk̑ e/o-). Table 4. Formation of verbal stems Type 1 Type 2 SM TA -pälk TB -rita TB -lyaka TA -lyāk TB -tsaika TA -tsek TB -ākṣi TB -aiyśi TA -kämṣe TB -näkṣi TA -pāṣe TB -pilṣi TB -pläṅṣi TB -yāmi Prs. n/a VI V n/a XI II IX VIII II n/a VIII n/a Averbo Subj. V ritanaṃ mid. ritanatär V mid. lkātär V lkāṣ mid. lkātär V 3sg. aksaṣṣäṃ aiśtär mid. (3pl.) käṃsantär nakṣäṃ mid. nakṣtär mid. pāṣtär plaṅ(k)ṣäṃ II II III I + II II I/II II I + II 3sg. mid. pälkātär mid. rītatär lakaṃ mid. lkātär TB tsaikaṃ ākṣäṃ opt. aiśitär mid. knatär nakäm mid. pāṣtär mid. piltär plyañcän yāmäṃ 4. Malzahn (2012) traces VGC type 1 in TB -a TA -∅ back to inherited SM abstracts in *-eh2 (e.g., according to Malzahn, Gk. Ὀλυμπιονίκης [Olympic gameswinning] ‘conquerer in the Olympic games’) and/or root compounds with second members that ended in a laryngeal (e.g., according to Malzahn, Lat. agricola [field-tending] ‘peasant’ and indigena [within-born] ‘native’) under the assumption that both final *-eh2 and *-H gave PT *-ā- > TB -a TA -∅. Malzahn treats type 2 in TB -i TA -e as an inner-Tocharian innovation based on PIE root compounds that were remodeled as hysterokinetic *n-stems. Both proposals have problems that will be addressed below. Malzahn’s diachronic derivation for type 1 has phonological and morphological problems. As Malzahn herself notes there are no traces of root SMs ending in 62 Hannes A. Fellner a laryngeal attested in the Tocharian B-rita-type. However, there is the question of what root structures of SMs with final laryngeal would have yielded PT *-ā-. In the SMs of roots in *-EH and *-RH only the full-grades *-aH, *-RH and the zerogrades -H, -R̥ H could lead to PT *-ā-, where the result of the vocalized laryngeal was analyzed as a stem formant or morpheme.10 However, besides the fact that ablaut in root compounds is rare even in a conservative branch such as IndoIranian and only occurs in a small number of root SMs,11 there seem to have been morphophonological constraints in Indo-European that prevented final laryngeals in root SMs to surface. Fellner and Grestenberger (2016:142) argue that in root SMs of the structure *-CVRH there are not many contexts in which laryngeal vocalization is predicted and that even if the nominative singular *-CVRH-s exerted influence on the rest of the paradigm, it is doubtful that this would have prevented the “νεογνός-rule”.12 Root SMs of *CeR-roots are extended with a t-formant in Indo-Iranian and Greek, e.g., Ved. deva-stú-t- [god-praising] ‘praising the gods’, Ved. pari-kṣí-t- [around-dwelling] ‘dwelling around’, Gk. περι-κτί-τ- [arounddwelling] ‘neighbor’, πυλ-άρ-τ- [gate-fastening] ‘gate-fastener’. At least in Greek13 the t-formant is also found in root SMs from roots in -RH (in the zerograde), e.g., προ-βλή-τ- [forth-throwing/thrown] ‘thrown forward, jutting out’ (: *gʷelh1 ‘throw’), ἀ-δμῆ-τ- ‘unbroken’ (: *demh2 ‘tame’), ἀ-γνώ-τ-14 ‘unknown’ (: *g̑ neh3 ‘know’).15 10 There are virtually no traces of o-grade SM in root compounds in Indo-European and fullgrade *-eh1 would have given palatalizing *-jä, which would have been lost in both Tocharian languages, whereas *-eh3 would have given PT *-æ > TB -e TA -∅; for *-eh2 see §4, p.64. 11 See Scarlata 1999:732–4. SMs of roots in *-EH usually show the full grade, e.g., -dhā́- ‘putting’ (: *dʰeh1 ‘put’), -pā́- ‘protecting’ (: *peh2 ‘protect’), -dā́- ‘giving’ (: *deh3 ‘give’). Examples of SM with, e.g., -dhá- , -pá-, -dá- go back to the structure -CH-ó-, cf. also n.15. 12 Cf. the formulation of Weiss (2009:113): -CRHV- > -CRV- “in non-initial syllables of ‘long’ words (including compounds and reduplicated forms)”: *neu̯ o-g̑ n̥ h1-ó-s > *neu̯ o-g̑ no-s > Gk. νεογνός ‘new-born’; see also Kuiper 1961, Mayrhofer 1986:65. 13 For the Latin examples sometimes cited in this context, e.g., Lat. sacer-dō-t- [sacrum-giving] ‘priest’, prae-gnā-t- [before-giving birth] ‘pregnant’, especially, locu-plē-t- [abundancereplete] ‘rich, wealthy, opulent’ see Nussbaum 2016. 14 Gk. ἀ-γνώ-τ- probably had e-grade since (-)CR̥ HC- sequences with accent gave disyllabic *(-)CəREC- in Greek. 15 SMs from roots in -RH (in the zero grade) appear also with a suffix, e.g., Ved. tuvi-gr-á‘much devouring’ (: *gʷerh3 ‘devour’), Gk. νεο-γν-ός ‘new-born’ (*g̑ enh1 ‘beget’), Lat. benign-us [good-natured] ‘kind’ (: *g̑ enh1 ‘beget’). Notes on Verbal Governing Compounds in Tocharian 63 From a comparative perspective, the evidence for root SMs with intact rootfinal laryngeal is therefore not very compelling.16 Malzahn’s comparanda of Lat. agricola ‘peasant’ and indigena ‘native’—which are usually cited as prime examples for root SMs with final laryngeal—are more economically explained as *eh2-derivatives from *o-stem SMs (cf., e.g., Gk. βου-κόλος ‘cowherd’, Lat. -genus ‘-born’) à la Fellner and Grestenberger 2016, see below (§6). The other morphological problem concerns the assumption that PT *-ā- could go back to abstract SMs in *-eh2. As already noted by Debrunner (1954:249): “Gemäß einer wohl grundsprachlichen Regel sind Abstrakta auf -ā- nicht üblich hinter Nominalstämmen.” This rule basically works as a morphological substitution rule that is synchronically opaque in the IE languages17 and applies regardless of whether the continued *eh2-stem was an abstract in the daughter language: Ved. śraddhā́- ‘belief’ → a-śraddhá- ‘unbelieving’, Ved. jihvā́- ‘tongue’ → mádhu-jihva- ‘having a sweet tongue’, Ved. jyā́- ‘bowstring’ → r̥ tá-jya- ‘having a good bowstring’; Gk. ἄγρη ‘hunt, chasing’ → πάν-αγρος ‘chasing all’; Gk. νεφέλη ‘cloud, Gewölk’ → ἀ-νέφελος ‘unclouded, cloudless’, Gk. τυπή ‘blow, wound’ → χαλκό-τυπος ‘having a wound made by a bronze weapon, inflicted with arms of bronze’. There is also replacement with *-i-, e.g., Gk. ἀλκή ‘strength’ → ἄν-αλκις, -ιδος ‘without strength, defenseless, weak’, Lat. barba ‘beard’ → imberbis ‘without a beard, beardless’, Lat. lingua ‘tongue’ → trilinguis ‘having three tongues’.18 From a comparative perspective, there are generally only a few examples of original SM in *-eh2—i.e., *eh2-stems that are not themselves derived from other SMs like *o-stems or root nouns. These are usually (endocentric) determinative compounds and almost all look like einzelsprachlich formations, cf., e.g., Ved. māṃsa-bhikṣā́- f. ‘request for meat’, dur-háṇā- f. ‘evil rage’, Gk. οἰνο-χόη [wine-pouring out] ‘wine vessel’. For the origin of the VGCs in *-eh2, see Fellner and Grestenberger 2016. 16 For apparent counterexamples found in Vedic, e.g., tuvi-ṣváni- (beside tuvi-ṣván- ‘roaring loudly’: *su̯ enh2 ‘sounding’) see Grestenberger (2017), who argues that these cases are better analyzed as i-substantivizations. 17 Cf. the “substituting” *-i- in the first compound member of bahuvrīhi compounds of roots belonging to the Caland system (Ved. r̥ jrá- ‘shining’ → r̥ jiśvan- ‘having shining dogs’, Gk. ἀργός ‘shining’ → ἀργικέραυνος ‘having bright lightning’). 18 Diachronically, this *i-substitution presumably started out as a substantivization of *oadjectives; see Grestenberger 2017. 64 Hannes A. Fellner The main reason why SMs in *-eh2 cannot have given type 1 VGCs in PT *-ā- > TB -a TA -∅ is that PIE *eh2 gives PT *å > TB o TA a in all environments, as shown in Fellner 2014a. Therefore, it is phonologically impossible to derive the TB -rita-type from *-eh2 SMs of any origin. Malzahn’s account of type 2 VGCs is phonologically possible. The proposed *-ēn of a hysterokinetic *n-stem would indeed account for the inflection of this type. However, the exact morphological and/or semantic mechanisms by which these compounds were created are left unexplained. 5. Based on the new evidence presented in the previous sections it is possible to offer a new solution for the origin of the VGCs in Tocharian. Since it is reasonable to suppose that type 1 belongs to the other agentive formations in Tocharian in TB -a TA -∅ (§2, p.60) it is possible to take its inflection as the continuation of the inherited so-called individualizing *(o)n-suffix that was associated mainly with thematic adjectives (Gk. Στράβων ← στραβός ‘squinting’, Lat. Catō ← catus ‘sharp’, Go. weihs ‘holy’ ← Go. weiha ‘priest’)19 and prone to readjectivization (*h1róudʰo- ‘red’ → *h1róudʰōn- ‘red (one)’: Gmc. strong. adj. *rauda-, Lith. raũdas etc. ‘red’ → Gmc. weak adj. *raudan- ‘red’; see Nussbaum 2014). As shown in Fellner 2014b, the inflection of the nt-participle in TB -ñca TA -nt (TB preñca TA prant ‘carrying’: TAB pär- ‘bear, wear, carry, take’) with nom. sg. TB -a TA -∅, obl. sg. TB -ai TA -āṃ, nom. pl. TB -añ TA -āñ, obl. pl. TB -aṃ TA -ās goes straightforwardly back to *-ōn-. After the suffix allomorph *-ō- was extended throughout the paradigm (which occurred independently in other Indo-European branches as well, e.g., Latin), the whole paradigm of ProtoTocharian and subsequently the individual Tocharian branches is derivable almost entirely by sound law (see Table 5). Concerning the nt-participles, Tocharian thus parallels Germanic in employing the formant *-ōn- to “extend” or recharacterize the basic stem. In Gothic and Old Norse, present participles inflect according to the Germanic weak declension, cf. the Go. nom. sg. m. -nds,20 acc. sg. m. -ndan, nom./acc. pl. m. -ndans. 19 For athematic base, cf., e.g., Ved. gurú- ‘heavy’ → grā́ van- ‘(grind) stone’ (also OIr. bráu ‘millstone’), Gk. Πᾶν < *pāu̯ ōn- < *pau̯ sōn- < *peh2-us-ōn- (cf. Ved. Pūṣan- < *puh2-s-n- < *ph2u-s-n-): *peh2 ‘protect’. 20 The nominative singular masculine preserves the old “strong” inflection, but the “weak” nom. sg. m. -nda is occasionally attested. 65 Notes on Verbal Governing Compounds in Tocharian Table 5. Development of the Tocharian paradigm nom. sg. obl. sg. nom. pl. obl. pl. a b c Toch. TB -a TA -∅ TB -ai TA -āṃ TB -añ TA -āñ TB -aṃ TA -ās PT ← < ←c < < < < < a < Pre-PT *-ō < PIE *-ō *-ānä < *-ōn-m̥ ← *-on-m̥ *-āñä < *-ōn-es ← *-on-es *-āns < *-ānäs < *-ōn-n̥ s ← *-on-n̥ s *-å b The paradigmatic leveling of the allomorph *-ā- on the way to Tocharian B is not too problematic and similar to the leveling of the stem allomorph -r- in Lat. honor, honōris < OLat. honōs, honōris. Following Jasanoff forthcoming, final PIE *-ō > PT *-å > TB -o, TA -∅. The expected obl. sg. †-aṃ in Tocharian B (which would have been synchronically conspicuous) was perhaps already replaced by -ai in analogy to nouns of class VI of the type yoko, obl. -ai m. ‘thirst’ (which continues amphikinetic i-stems) at a stage when the nominative singular of both noun classes still had Pre-B *-å. Thus, the TB -rita-type can then be viewed as reflecting SMs in *-ōn-. The starting points of these individualizing *ōn-stems were either original SMs of the R(o)-o-type (the corresponding simplex nouns are well attested in Tocharian; see Malzahn 2013). or root SMs (at a time when there was no difference between secondary *-ōn- of thematic or athematic stems). The former could be reflected by TB -tsaika TA -tsek if the original o-grade does not come from the associated verbal formation; the latter seems to be represented by TB -lyaka TA -lyāk < *-lēg̑ - (cf. Lat. lēx ‘law’, il-lēx, -ēgis ‘without law’), which is synchronically not associated with a present or subjunctive formation, but with the preterite TB 3sg. lyāka and the imperfect TA 3sg. lyak ‘saw’ (cf. Lat. lēgī ‘gathered’, Alb. mblodhi ‘collected’; see further Jasanoff 2012), and these are synchronically conspicuous themselves. In the light of the latter, it is very likely that PIE root compounds were the basis of the creation of type 1 VGCs in Tocharian. In this way, there is another independent parallel between Tocharian and Germanic besides the nt-participles: VGCs in Germanic going back to PIE root compounds were also characterized by *-ōn-, e.g., OS folk-togo ‘leader of the people’ (: tiohan ‘lead’), OE yrfe-numa [inheritance-taker] ‘heir’ (: niman ‘take’), OHG ārunt-poro ‘news carrier’ (: beran ‘carry’). 66 Hannes A. Fellner Since type 2 is associated with thematic verb stems, there is also a scenario to account for type 2 in TB -i TA -e < *-ēn-. Nussbaum (2016:289) argues for a derivational process *o-stem adjective → individualizing *ē̆t-stem individualization (which like their *ōn-stem equivalents are prone to readjectivization), e.g., Gk. ἀργός ‘shining’ → ἀργής, ἀργέτ- ‘bright’, Pre-Lat. *dīu̯ -o- ‘brilliant’ (Lat. dīum ‘daylit sky’) → *dīu̯ et- (Lat. dīves, dīvit- ‘rich’), PIE *steip-o- ‘rigid’ (Gmc. *steifa- ‘stiff’ in OE stīf) → *steipe-t- (Lat. stīpes, stīpit- ‘stick, stake’). For this derivational scheme he also adduces examples from compounds, e.g., PIE *-sth2-o- (Ved. prati-ṣṭhá- [against-standing] ‘resistant’) → -sth2ē̆t- (Av. haŋᵛharə-stāt- ‘staying in hiding’, Lat. anti-stes, -stit- [in front-standing] ‘priest’), PIE *-kʷol(h1)-o- (Gk. βου-κόλος ‘cowherd’, Lat. an-culus [around-tending] ‘servant’, Av. fra-cara- ‘going forth’) → *-kʷol(h1)-ē̆t- (Av. fra-carāt- ‘going forth’), Pre-Lat. *-poro- (Lat. per-perus [across-traversing] ‘perverse’) → *-p(o)r-ē̆t- (Lat. interpres, -pret- [between-carrying] ‘messenger’). The basis of the *ē̆t- derivatives according to Nussbaum (p.c.) was the e-grade of the thematic vowel to which an amphikinetic *-(o)t- was added, with *-e-ot- giving *-ēt-. If this kind of vowel contraction is accepted, the same prehistory could in principle be behind Malzahn’s proposed ending *-ēn- of type 2 VGCs. The core of these compounds would then go back to R(o)-o- SMs to which *-(o)n- was added with the same development of *-e-on- > *-ēn- as with *-e-ot- > *-ēt-.21 That thematic quasi-participial formations become associated with thematic verbal formations is trivial. 6. Nussbaum (2014) notes that *(o)n-stems behave like *eh2-stems concerning individualization/substantivization, with a tendency to become readjectivized: *h1róudʰo- ‘red’ → *h1róudʰōn- ‘red (one)’: Lat. (via Sab.) rūfus ‘red’ → Rūfō ‘red(-headed) guy’, Gmc. strong. adj. *rauda- → weak adj. *raudan- ‘red’; cf. Gmc. *weiha- in Go. weihs ‘holy’ → Gmc. *weihan, Go. weiha ‘priest’. Fellner and Grestenberger (2016) deal with the VGCs of the Gk. βουζύγης ‘yoking oxen’ and the Lat. agricola type, e.g., βαθυ-δίνης ‘deep-eddying’: δινάω/έω ‘whirl’, Ὀλυμπιο-νίκᾱς ‘Olympic victor’: νικάω ‘prevail’; agri-cola ‘farmer’: colō, -ere ‘cultivate’, indi-gena ‘native’: gignō/genō, -ere ‘beget’. Fellner and Grestenberger (2016), following Nussbaum (2014) and Melchert (2014), argue that Gk. βουζύγης- and Lat. agricola-type VGCs constitute an archaic compound type in 21 Assuming that type 1 and type 2 both go back to R(o)-o- SMs it is possible to surmise that they each generalized different stems, *-ōn- < *-o-on- in the case of type 1 and *-ēn- < *-e-on- in the case of type 2. For this possibility in simplex nouns, see Höfler 2015:232–6. Notes on Verbal Governing Compounds in Tocharian 67 which the *eh2-suffix is used as masculine individualization/substantivization based on the following derivational pattern: *X-i̯ ug- / *X-i̯ ug-o- ‘xyoked/yoking’ (Gk. -ζυξ, Ved. -yuj- / -ζυγος) → *X-i̯ ug-eh2- -ζύγης ‘one who is x-yoked/yoking’. Thus Malzahn’s claim (2012) that the type 1 VGCs in TB -a TA -∅ can be compared with the VGC of the Gk. βουζύγης and Lat. agricola type is mutatis mutandis not incorrect. There are even quasi word equations of VGCs with -ō̆n and *-eh2: OHG -poro TB -lyaka TA -lyāk with *-ō̆n with *-ō̆n Gk. -φόρᾱς22 Lat. -lēga23 with *-eh2 with *-eh2 7. In conclusion, type 1 VGCs in TB -a TA -∅ pattern with athematic present stem classes, whereas type 2 VGCs in TB -i TA -e pattern with thematic present stem classes. Type 1 belongs together with other agentive formations in TB -a TA -∅. Similar to the nt-participles of this class, the VGCs go back to a formation with individualizing *-(o)n- and are therefore directly comparable with the Germanic VGCs. It is also possible that individualizing *-(o)n- formations were at the heart of type 2. The *(o)n-stem VGCs of Tocharian and Germanic (which evolved independently) are functionally comparable to Gk. βουζύγης- and Lat. agricola-type VGCs, thus providing more evidence for the similarity of the function of *-(o)n- and *-eh2 à la Nussbaum (2014). References Debrunner, Albert. 1954. Altindische Grammatik II/2: Die Nominalsuffixe. Göttingen: Vandenhoeck & Ruprecht. Bernhard, Franz. 1958. Die Nominalkomposition im Tocharischen. Ph.D. diss., Göttingen University. CEToM. A Comprehensive Edition of Tocharian Manuscripts. http://www.univie.ac.at/ tocharian/. Accessed April 30, 2017. Cooper, Adam I., Jeremy Rau, and Michael Weiss (eds.). 2013. Multi Nominis Grammaticus: Studies in Classical and Indo-European Linguistics in Honor of Alan J. Nussbaum on the Occasion of his Sixty-fifth Birthday. Ann Arbor: Beech Stave. Fellner, Hannes A. 2014a. PIE Feminine *-eh2 in Tocharian. In Neri and Schuhmann 2014, 7–21. Leiden: Brill. ———. 2014b. Tocharian Special Agents: The nt-Participle. Tocharian and Indo-European Studies 15.53–67. 22 E.g., Gk. σαμ-φόρας [σάν-carrying] ‘horse branded with the letter σάν’. 23 E.g., Lat. col-lēga [together-gathering] ‘colleague’. 68 Hannes A. Fellner Fellner, Hannes A. and Laura Grestenberger. 2016. The Greek and Latin Verbal Governing Compounds in *-ā- and their Prehistory. In Bjarne Simmelkjaer Sandgaard Hansen, Benedicte Nielsen Whitehead, Thomas Olander, and Birgit Anette Olsen (eds.), Proceedings of the 14th Fachtagung der Indogermansichen Gesellschaft, 17–22 September 2012, Copenhagen, 135–50. Wiesbaden: Reichert. Grestenberger, Laura. 2017. On “i-Substantivizations” in Vedic Compounds. In Bjarne Simmelkjær Sandgaard Hansen, Adam Hyllested, Anders Richardt Jørgensen, Guus Kroonen, Jenny Helena Larsson, Benedicte Nielsen Whitehead, Thomas Olander, and Tobias Mosbæk Søborg (eds.), Usque ad Radices: Indo-European Studies in Honour of Birgit Anette Olsen, 193–206. Copenhagen: Museum Tusculanum. Härtl, Holden. 2013. Arguments of Non-heads. In Holden Härtl (ed.), Interfaces of Morphology. Berlin: Akademie, 163–177. Höfler, Stefan. 2015. Denominale Sekundärderivation im Indogermanischen: Eine Ochsentour. Münchener Studien zur Sprachwissenschaft 69.219–43. Jasanoff, Jay H. 2012. Long-Vowel Preterites in Indo-European. In H. Craig Melchert (ed.), The Indo-European Verb: Proceedings of the Conference of the Society for Indo-European Studies, Los Angeles 13–15 September 2010, 127–35. Wiesbaden: Reichert. ———. 2013. The Tocharian Subjunctive and Preterite in *-a-. In Cooper, Rau, and Weiss 2013, 105–20. ———. Forthcoming. The Phonology of Tocharian B okso ‘ox’. To appear in a Festschrift. Krause, Wolfgang. 1952. Westtocharische Grammatik I: Das Verbum. Heidelberg: Winter. Krause, Wolfgang, and Werner Thomas. 1960. Tocharisches Elementarbuch I: Grammatik. Heidelberg: Winter. Kuiper, Franciscus B. J. 1961. Zur kompositionellen Kürzung im Sanskrit. Die Sprache 7.14–31. Malzahn, Melanie. 2010. The Tocharian Verbal System. Leiden and Boston: Brill. ———. 2012. Verbale Rektionskomposita im Tocharischen. In Velizar Sadovski and David Stifter (eds.), Iranistische und indogermanistische Beiträge in memoriam Jochem Schindler (1944–1994), 113–35. Vienna: Österreichische Akademie der Wissenschaften. ———. 2013. Cutting around “temós”: Evidence from Tocharian. In Cooper, Rau, and Weiss 2013, 165–74. Malzahn, Melanie, and Hannes A. Fellner. 2015. Lifting up the Light: tläś and lkäś in Tocharian A. Tocharian and Indo-European Studies 16.61–79. Marggraf, Wolf-Jürgen. 1970. Untersuchungen zum Akzent in Tocharisch B. Ph.D. diss., University of Kiel. Mayrhofer, Manfred. 1986. Indogermanische Grammatik I/2: Lautlehre [Segmentale Phonologie des Indogermanischen]. Heidelberg: Winter. Notes on Verbal Governing Compounds in Tocharian 69 Melchert, H. Craig. 2014. PIE *-eh2 as an ‘Individualizing’ Suffix and the Feminine Gender. In Neri and Schuhmann 2014, 257–71. Neri, Sergio, and Roldand Schuhmann (eds.). 2014. Studies on the Collective and Feminine in Indo-European from a Diachronic and Typological Perspective. Leiden: Brill. Nussbaum, Alan J. 2014. Feminine, Abstract, Collective, Plural: Disconnected Remarks on Each. In Neri and Schuhmann 2014, 273–306. ———. 2016. Replacing locus ‘place’ in Latin locuplēs. In Dieter Gunkel, Joshua T. Katz, Brent Vine, and Michael Weiss (eds.), Sahasram Ati Srajas: Indo-Iranian and Indo-European Studies in Honor of Stephanie W. Jamison, 276–95. Ann Arbor: Beech Stave. Ohala, John J. 1997. Emergent stops. Proceedings of the 4th Seoul International Conference on Linguistics (SICOL).84–91. Scarlata, Salvatore. 1999. Die Wurzelkomposita im Ṛg-Veda. Wiesbaden: Reichert. Weiss, Michael. 2009. Outline of the Historical and Comparative Grammar of Latin. Ann Arbor: Beech Stave.