The Tocharian verb

The Tocharian verb Miguel Carrasquer Vidal, November 2014 Unlike the Tocharian nominal system, which has innovated and eroded to the point of being only of minor importance for the reconstruction of the PIE nominal system, the Tocharian verb seems to have maintained many inherited features of the PIE verbal system, and is highly relevant to its reconstruction. Like that of Hittite and the other Anatolian languages, the Tocharian verbal system in some respects is difficult to reconcile with the so-called I do-G eek e al syste that had ee postulated for PIE in the classical reconstructions of Brugmann et al. The state-of-the-art reference on the Tocharian verbal system is Malzahn 2010. Present classes I: Athematic presents. Most forms show the expected e-grade (but no ablaut). A few verbs show original *o-grade, pointing to olō-verbs. II: Thematic presents. Besides e-grade, a number of verbs show ē-grade, pointing to thematized Narten-presents. III, IV: Middles with invariable thematic suffix *-o-. The origin of this category, and the details on how it split into classes III, with short root vocalism (ä, i, u) and suffix *æ, and IV, with full root vocalism and suffix *å, are quite controversial. Based on the evidence from Hittite and Gothic thematic middles, and on the Tocharian Subjunctive III, I would opt for deriving the suffix from a stressed thematic vowel with invariant shape *-ó-, possibly in combination with the suffix *-ā. V: Athematic presents with suffix *-ā-. These a e athe ati p ese ts ade f o oots ith a- ha a te p esu a ly, at so e point, seṭ-roots, i.e. roots ending in a laryngeal). VI: Athematic presents with *- ā-. Cf. LIV category 1m (*neH-presents). VII: Athematic presents with nasal infix. Malzahn does not link this class directly with LIV 1k (n-infix presents), but explains them as resulting from an inner-Tocharian metathesis -Ck- ā > -C kā. I would add that in Hittite, as well, the metathesis is only triggered by roots ending in -k(k) (and -h(h)). VIII: Thematic presents with suffix -se. This category has merged with the next one in ToA, and, like present Class IX, the sigmatic suffix tends to increase the valency of the verbal root. However, an origin separate from that of the sk̂é-presents still seems the most likely. A relation with the s-aorist seems plausible, but because of the zero grade of the root, it cannot be an old s-aorist subjunctive. It can be a thematic s-aorist (cf. preterite class VI for the asigmatic case), turned presentic by application of Malzah s tezzi-principle1. IX: Thematic presents with suffix -sk̂éClass IXa has zero grade of the root and corresponds exactly with LIV category 1p. Class IXb, which contains the vast majority of Tocharian present causatives, has e-grade of the root, a d i itial a e t i ToB, poi ti g to lost edupli atio . I stead of Malzah s ela o ate hai of events (pp. 457-9), I would suggest that class IXb indeed reflects original causatives without *-ské-, but that the whole class simply reflects the present tense of Preterite class II. 1 The formation of presentic forms out of preterite ones by the simple procedure of adding primary endings. The model is Hittite tezzi states, says , out of the oot ao ist ste *dheh1-. As to the i egula ities ega di g a- ha a te i p ese t lass IXa (IXb is irrelevant, as the *-sk̂e is a secondary addition there), it is worth pointing out that Lithuanian ste-verbs more often than not show an acute root, pointing to a pre-form *-h2sk̂é-. This I have previously analyzed as evidence for an origin of the (h2)sk̂é-verbs in the incorporated verb *h2aisk̂- to seek, ask, de a d . Verbs in final -tk derive from roots ending in a dental followed by *-sk̂e. X: Thematic presents with suffix *- äsk/ āskApparently a combination of classes VI and IX. XI: Thematic presents with suffix *-säskApparently a combination of classes VIII and IX. XII: Thematic presents with suffix *-(ä)ññ-. Most likely a class made up out of denominatives from n-stems, and deverbatives from *neHstems. Subjunctive classes Like the present classes, except: I, V: Athematic subjunctives Either no ablaut or root ablaut o ~ ø. Usually initial accent in ToB. Class V additionally shows the suffix *-ā-. II: Thematic subjunctives From former thematic presents: it is unlikely that any Tocharian thematic subjunctive continues an inherited PIE thematic subjunctive. III: Middles with invariable thematic suffix *-o-. Intransitive (middle only) counterparts to class I subjunctives. IV: Thematic subjunctives with suffix -i. Probably from PIE *je-presents. VII: Thematic subjunctives with suffix -ñ. Ultimately from LIV 1l (nu-verbs), with generalization of the analogical thematized suffix *-ne- > *-ñ(ä)-, instead of expected *-nä- (< *-nu). Preterite classes 0: Middle root preterites. Middle-only, attested only in ToA, with root grade *o and *ē. In ToB, these verbs are in preterite class III, with an s-ending. I: Preterites with suffix -ā. With respect to ablaut, there are 5 subclasses:  e-grade in the singular, o-grade in the plural (!), and zero grade in the middle.  zero grade in the singular and middle, o-grade in the plural (!).  ā throughout.  ä (*e) throughout.  ā throughout. II: Causative preterites. Reduplication in ToA (patterns *Cæ-CäC-ā and *C æ-C äC-ā), long vowel (pattern *C āC-ā and, in the ptc., *C æ-C äC-) in ToB, with initial accent. Makes preterites to causative verbs with short vocalism (ä, i, u). III: s-Preterites. The main characteristic of this class is the desinence -s ā in the 3sg. and in the whole of the middle (except for the ToA verbs in preterite class 0). The root, if it ablauts at all, has *æ (*o) or * æ (*ē) vocalism in the active, zero-grade (ä) in the middle. IV: Causative preterites with suffix -ṣṣā. Makes preterites to causative verbs with full vocalism. Evidently created by adding preterite -ā to the presents in *-(h2 sk̂e. Initial accent in ToB as in class II and present class IXb. V: Preterites with suffix -ñ ñ ā. Preterites to class XII subjunctives. VI: Thematic aorists. From the PIE thematic root aorist with endings *-óm, *-és, *-ét, etc., in part retained in To ha ia he eas all othe p ete ites ha e adopted the pe fe t e di gs . The PIE optative gives the Tocharian optative and imperfect, and there are also seven classes of imperatives. From the comparative point of view, the classes that most interest me at the moment are subjunctive classes I, V, present/subjunctive class IXb, and preterite classes I, II and III/0. We will take a closer look at them (Malzahn 2010: 111-214, 274-316 and 433-466). Preterite III/0 Preterite III, as we saw above, consists of two main subgroups: a. *o grade in the whole active, zero grade in the middle. There are signs of reduplication, e.g. otkasa-me sepa ated < *wäwætk-, orsa-c a a do ed < *h2eh2or(H)-, 3pl. arar-c < *h2eh2rH-. b. *ē grade in the whole active, zero grade in the middle. There are no signs of reduplication. Preterite III (fundamentally, subgroup a) is linked to Subjunctive class 1, with *o ~ zero ablaut, and initial accent in ToB. Preterite I Preterite I likewise can be divided into two subgroups: a. *e-grade in the active singular, *o grade in the active plural, and zero grade in the middle. No signs of reduplication. b. zero grade in the active singular, *o grade in the active plural, and zero grade in the middle. No signs of reduplication. Subgroup 1 can be linked to Subjunctive class 5 with ablaut *o ~ zero, and initial accent in ToB. Subgroup 2 can be linked to Subjunctive class 5 forms without ablaut, and normal accent in ToB. Preterite II Preterite II can also be divided into two groups: a. *e-grade in the active, reduplicates with *ē in ToA. b. zero-grade in the active, reduplicates with *o in ToA. In ToB, like in Vedic *sa-sd-úr > sedúr, Germanic *ge-gbum > gē u , *he-ht- > *hē2t-, or Lithuanian *le-lk- > lėk̃ -ė, the reduplication has been replaced by a long root vowel - ā- (*C æC ä C - > *C āC -). The corresponding present/subjunctive is class IXb, with root vocalism *e, initial accent in ToB (pointing to *-i- or *-e- reduplication), and secondarily suffixed -(ä)sk-. Inherited categories of the h2e-conjugation As I a gued i y a ti le Problems in PIE verbal morphology , the p e-PIE perfective *h2e-conjugation spawned the following six verbal paradigms: pret. sg. 1/2 pl. 3pl. pres. sg. 1/2 pl. 3pl. intr. Ci-CéRCCi-CRCCí-CRCintr. Ce-CóRCCe-CRCĆ-CRC- tr.sg. CérCCéRCCRCtr.sg. CórCCóRCCéRC- tr.pl. Cí-CRCCi-CRCCí-CRCtr.pl. Ć-CRC-, ĆCCe-CRC- , CeCĆ-CRC-, ĆC- We have, respectively, the old pe fe ti e atego ies: reduplicated present, the root aorist and the edupli ated ausati e ao ist, a d the eo-i pe fe ti e atego ies: the edupli ated pe fe t, the olō-presents and the Narten-presents. Note that, for semantic reasons, the intransitive *h2econjugation present (i.e. the perfect) eventually develops into a preterite category, while the o espo di g i t a siti e p ete ite the edupli ated p ese t , y the tezzi-principle, acquires presentic function. I would argue that the core subclasses of the Tocharian preterite classes I, II and III directly continue the PIE intransitive, transitive and causative/iterative paradigms, with the reflexes of the perfective and neo-imperfective paradigms blended as follows: Preterite I, Subclass a: blend of root aorists and olō-imperfects, mainly transitive. The active singular has *e from the root aorist, the active plural *o from the olō 1/2pl., analogically extended to the 3pl. as well. Preterite II: blend of the reduplicated aorist and reduplicated Na te -imperfects, mainly iterativecausative. The reduplication syllable is *ē from the Narten verbs, and therefore survives into ToA, and the root has zero grade or *e-grade, as in the Vedic reduplicated aorist. Preterite III, Subclass a: the reduplicated present had ceased to be a preterital category, so this subclass would appear to continue only the now preterital PIE perfect, with *o-grade and reduplication, mainly intransitive. The *o-grade in the active must originally have been confined to the singular only. Subsequently, two further categories were incorporated into Preterites I and III: 1. intransitive thematic *h2e-conjugation (middle?) aorists, which I will call *ã-verbs , which joined Preterite class I; 2. s-ao ists eo-pe fe ti es , hi h joi ed P ete ite lass III. The s-aorist in Tocharian The s-aorist in PIE is a verbal category devised to supply aorists to imperfective (*mi-conjugation) verbal roots. It is the prime example of what I, in i itatio of Malzah s tezzi-principle, would like to call the dais-principle: presentic roots could be made aoristic/preterital by adding the corresponding endings. Since the transitive 3sg. preterite ending of the *h2e-conjugation was *-s (also 3pl. *-ér-s), the main characteristic of the eo-pe fe ti es as *-s in the 3sg. This *-s, in roots ending in a single consonant, would have triggered Szemerényi lengthening, which is why the s-aorist paradigm, besides generalized *-s-, also shows generalized *ē-grade throughout (except perhaps in the 1/2pl.). At first glance, it looks as if in Tocharian the *-s was confined to its birth-place, the 3rd person sg. (the extension to the whole of the middle is a later development), whereas the *ē-grade, on the other hand, as ge e alized to all pe so s of the a ti e. This does t st ike e as a e y plausi le s e a io. We should not overlook the fact that the 2sg. and pl. preterite endings in Tocharian (*-stā and *-sä) already had an *-s-, etymologically derived from stems ending in a dental (*-ts-tx and *-ts-dhxw), with the *-s- reinterpreted as part of the ending. Additionally, PIE *-s- is regularly lost in Tocharian before nasals in word-internal position, which would explain why *-s- is absent from the first persons (assuming *-mh2a > -wā and *-me > -m(ä)). If so, the Tocharian s-aorist is in origin no different from the s-aorist elsewhere in IE. The presence of *-s- in the perfect 2nd person endings may indeed also have played a role in the merger of the original Preterite III (= PIE perfect) and the s-aorist. Another factor must have been the merger of the root vocalism *o (> *æ) and *ē > * æ). After the merger of the two paradigms, the old perfects, analogically after the s-aorists, extended the uniform ablaut grade *æ to the active plural (even in such middle forms as are efle ted i the ToA P ete ite . Reduplication, except for the traces noted above, was also lost, as the Preterite III does not show the usual signs of former reduplication (i.e., initial accent) in ToB. The *ã-ver s and To harian a- hara ter Preterite I, as stated above, consisted of original root aorists and original molō-preterites (by the dais-principle), blended in such a way as to produce the, from an IE point of view, iza e i e ted ablaut pattern *e ~ *o. Some of the roots involved were sẹ roots, and ended in PT *ā. This eventually must have attracted the category of *ã-verbs into the sphere of preterite class I. As I hypothesized i Problems in PIE verbal morphology , the e di gs *-ã and *-̃, with persistent zero grade of the root, were originally the regular outcomes of a thematic *h2e-conjugation 1st person *-a-h2a and 3rd person *-e-e, reflected as such in the Latin subjunctive/future 1st person -am, 3rd person -et. In Baltic, the endings make the -o (largely intransitive) and -ė (largely transitive) preterites. The Baltic data clearly prove the circumflex character of the desinences, but we may assume that the distinction was lost in Pre-Tocharian. The intransitive category of *-ã verbs was preserved into Pre-Tocharian, until they merged with Pret. I. After the merger of the *ã-verbs with the original Preterite I, they acquired analogical *o-grade in the active plural (at least in ToA), but retained the zero grade of the root in the active singular (PT *ä). The ending *-ā was analogically transferred to aniṭ-verbs of preterite class I, which in this way a ui ed se o da y a- ha a te , and eventually to all Tocharian preterite classes except VI. The corresponding subjunctives The three preterite classes I, II and III regularly correspond to subjunctive classes V, IXb and I. The causative preterites of class II had *ē-reduplication, and either zero-grade or *e-grade of the root. The corresponding presents (> subjunctives) likely had *i- or *e-reduplication and *e-grade of the root (and the suffix *-sk̂e added secondarily). The *e-grade is not entirely as expected, but is also found in the Vedic causative aorists when the root ends in -eT or -eR. Preterite class Ia (old olō-verbs and root aorists) corresponds to the class V subjunctive with *o ~ zero ablaut, original reduplication, and suffix -ā. Class Ib (old ã-verbs) corresponds in ToB with middle-only class V subjunctives without ablaut or reduplication, and suffix -ā. Preterite class III (old perfects, s-aorists) corresponds to Subjunctive class I, with *o ~ zero ablaut, original reduplication, and absence of the suffix -ā. The fact that subjunctive classes I and V are identical, except for the presence or absence of the suffix -ā, suggests that levelling has taken place, apparently in favour of the perfect paradigm, with reduplication and *o ~ zero ablaut, and against the olō-paradigm where we would expect *o ~ *o and no reduplication. The Tocharian root aorist The root aorist (and the related reduplicated present and aorist) originally had h2e-conjugation endings. In previous papers, I have argued that there are a number of facts supporting this view:    The Greek root aorists ἔδωκα ga e , ἧκα sent and ἔθηκα put . If the e di gs of the oot aorist were originally those of the hi-conjugation, the -κ- in Greek can be explained in the same way as the -κ- in the first perfect, as the hardening of *-HH- in the 1st person sg. (*-h2e) of verbs ending in a laryngeal. Parallel forms can be seen in Latin (faciō ~ fē ̄ do , ia iō ~ iē ̄ th o , the othe Itali la guages U . face; Osc. afakeit, fefacid, Venet. whagsto do , as well as Phrygian (addaket, addaketor affect ). The 3pl. ending -úr (< *-́r < *-ér-s) in Vedic. The Ablaut of the root aorist, with full *e-g ade i the . a d . pl. Malzah s 5:1 Ablaut), as a direct result of the fact that the hi-conjugation endings were originally, before the suffixation of *-e, syllabic only in the 3pl. (*-ár). At the sa e ti e, I stated that [t]he t a sfe of the oot ao ist to the mi-conjugation happened before the split-off of the Anatolian languages, where we have e.g. the mi-verb (< root aorist) tē i (*dhéh1-mi I state, I say , s. the hi- e eo-impe fe ti e tehhi (*dhó(i)h1-h2ai I put . One may wonder whether, in the case of Tocharian, it is really necessary to assume that the rootaorist first assumed mi-conjugation endings, only to give them up again in favour of the Tocharian preterite endings, which are clearly based on the endings of the h2e-conjugation. The relative chronology of the split-offs of Anatolian and Tocharian would suggest that, yes, it is necessary to assume such a sequence of events. The existence of preterite class VI (thematic aorist, partially with *mi-endings), is also suggestive... In any event, we can add at least one Tocharian form to the Greek, Italic and Phrygian k-aorists given above: ToB tākā-, the suppletive preterite (I) and subjunctive (V) to the verb nes- to e . To uote Malzah p. : Mu h as o e ould like to take tākā- as a reflex of a PIE active perfect in view of the Greek perfect ἕστηκα [...] , a pe fe t should ha e esulted i Pret. III and Subj. I. The form tākā- is therefore a root aorist *stéh2-h2a, completely parallel to ἔδωκα, ἧκα and ἔθηκα. Summary tr. Pret. I root aor. olō Subj. V ã-verbs intr. caus. Pret. II red. aor. Narten Subj. IX intr. Pret. III perfect Subj. I s-aor. tr. References Carrasquer Vidal, Miguel, 2014, Problems in Indo-European verbal morphology. Malzahn, Melanie, 2010, The Tocharian verbal system, Leiden-Boston.