Jump to ContentJump to Main Navigation
Hittite and the Indo-European Verb$

Jay H. Jasanoff

Print publication date: 2003

Print ISBN-13: 9780199249053

Published to Oxford Scholarship Online: September 2007

DOI: 10.1093/acprof:oso/9780199249053.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 25 February 2017

(p.228) Appendix 2: The Perfect *u̯óid-e

(p.228) Appendix 2: The Perfect *u̯óid-e

Hittite and the Indo-European Verb
Oxford University Press

Throughout this study we have emphasized the reduplicated character of the PIE perfect. In Ch. 1, the fact that the overwhelming majority of ji-verbs are unreduplicated was used as an argument against the ‘perfect theory’ of the ḫi-conjugation; in later chapters the possibility of taking the perfect as the source of the Tocharian subjunctives of classes I and V and the Tocharian preterites of the type A prakwā, B prekwa was rejected on the same grounds (cf. §§93, 102 f., 117); and in §97 reduplication was argued to have been the morphological feature that distinguished the perfect from its unreduplicated derivational base. On the whole, the comparative evidence strongly confirms the view that the perfect was a reduplicated category in the parent language. Yet it is also clear that one lexical item, the semantically deviant perfect of the root *u̯eid- ‘see, catch sight of, recognize’, was an exception. Reflexes of unreduplicated *u̯óid- / *u̯id-ɛ ‘know’ are widely attested, not only in languages where reduplication might in principle have been lost (cf. Go. 1 sg. wait, OCS vědě, OPr. waidimai (1 pl.), Arm. gitem), but also in languages where it was otherwise regularly maintained (cf. Ved. 1 sg. véda, GAv. vaēdā, Gk.(F)οἶδα‎,OIr. ‧fetar). Opinions differ as to whether the lack of reduplication in *u̯óid- / *u̯id-′ was a secondary, inner-IE development or an original feature.1 The suggestive and more commonly accepted latter view, with its implication that reduplication in the perfect was in the last analysis facultative or even ‘stylistic’, partly explains the willingness of otherwise cautious scholars to countenance ‘perfects’ like *sókH-e ‘knows’ and *pró-e ‘has asked’ as preforms in Hittite and Tocharian. Seductive as it undeniably is, however, to compare Hitt. pakki ‘knows’ with its PIE synonym *ṷóid-e, the comparison is meaningless unless we can estabish the validity of its unspoken premise—that *u̯óid-e was an archaism within PIE itself.

Within the framework presented here, *u̯óid-e is even more anomalous than under more conventional views of the PIE verbal system. As schematically illustrated in §97, the perfect originated as a reduplicated derivative of the stative-intransitive aorist:

ónh1-e ‘was born’   ⇒ *ĝeĝónh1-e ‘is / has been born’

*u̯ā́ĝ-e ’broke (intr.)‘ ⇒ *u̯eu̯ā́ĝ-e ‘is broken’

In keeping with this pattern, we might have expected that the perfect of the root *u̯eid- would be built to the stative-intransitive aorist attested in Ved. avedi (pass. aor.) ‘came to light, was found’:

*u̯óid-e ‘came to light’ ⇒ *u̯eóid-e ‘is visible, is found as’

(p.229) But no such intransitive *u̯eóid-e is actually found. Instead, a new and puzzling fact emerges: *u̯óid-e ‘knows’ was homophonous with *u̯óid-e ‘came to light’, the corresponding stative-intransitive aorist. The unexpected identity of the two forms immediately casts doubt on the putative status of *u̯óid-e ‘knows’ as an age-old inheritance from the remote prehistory of the parent language. Qua perfect, *u̯óid-e cannot be original as it stands: either it was once reduplicated, as some scholars have supposed, or it was created within PIE at a time when the formal overlap with the aorist *u̯óid-e was no longer a relevant issue. To decide between these two possibilities we will first have to determine what would have been the normal treatment, in later PIE and beyond, of the theoretically expected but unattested *u̯eóid-e ‘is visible / recognizable, is found (as)’.

Light is shed on the overall morphological behaviour of the root *u̯eid- by the semantically parallel roots *derk̂ ‘see, glimpse’ and *k̂leu- ‘hear’. Like *u̯eid-, *derk̂- and *k̂leu- formed stative-intransitive aorists, which survive in Ved. ádarśi ‘appeared’ < *dór-e and śrā́vi (GAv. srāuuī) ‘was heard, became famed as’ < *k̂-e, respectively. In principle, the aorists *dór-e and *k̂-e, like *u̯óid-e, ought to have given rise to derived perfects:

*dór-e ‘appeared’

*dedór-e ‘is visible’

*k̂-e ‘became audible’

*k̂e-e ‘is audible, is famed as’

The forms *dedór-e and *k̂elóṷ-e are in fact attested, but not with the expected intransitive meaning. The comparative evidence shows, rather, that the roots *derk̂- and *k̂leu- had transitive active and intransitive ‘passive’ subsystems in late PIE, and that the perfects *dedór-e and *k̂e-e, despite their derivational history as intransitives, were uniformly treated as actives. Compare the actual forms:




dárśam ‘saw’

dadársa, YAv. dadarQsa ‘has seen’, Gk. dådoqje ‘sees’2


ádarsi ‘appeared’

dádse ‘is visible’


ásrot ‘heard’3

susrava, OIr. ro·cúalae ‘has heard’, Gk. jåjktse ‘hear!’


srdvi ‘was famed as’

susruve ‘is famed as’

For these two roots, then, the original function of the perfect was taken over by the newly created perfect middle, while the old perfect was reinterpreted as a transitive perfect active. From the fact that the transitivization of the perfects *dedór-e and *k̂elóṷ-e was a development common to Indo-Iranian and Greek, it can be concluded that the perfect middles *dedr̥k̂-ór ‘is visible’ and *k̂elu-ór ‘is audible, is famed as’ were already part of the verbal system of late PIE (cf. §30, with n. 38).

Apart from the irrelevant detail that the active root aorist *u̯éid-8, *u̯éid-s, etc. was replaced within the parent language by the thematic aorist *u̯id-ó-m, *-é-s, etc.,4 the (p.230) root *u̯eid- patterns in exactly the same way as *derk̂- and *k̂leu-. We thus find




ávidat ‘found’, Gk. ‎(F)ιδε‎ ‘saw’

vivēda (= YAv. vīuuaēδ‎ a) ‘has found’


avedi ‘was found (as)’

vividé ‘is seen, is found (as)’,

pointing, as it might seem, to a PIE system




*u̯id-é-t ‘saw‘ found’

*u̯eóid-e ‘sees, has seen / recognized’


*ṷóid-e ‘appeared, came to light’

*ṷeṷid-ór ‘is visible, is recognized’

Yet despite the overall parallelism of *u̯eid-, *derk̂- and *k̂leu-, and despite the generally assumed PIE antiquity of *dedór-e and *k̂e-e,5 Indo-Europeanists have traditionally been highly resistant to setting up a perfect *u̯eóid-e, and still less a perfect middle *u̯eid-ór, for PIE. There are three main reasons for this reluctance, none cogent:
  1. (1) Because the ‘real’ perfect of *u̯eid- is supposed to have been *u̯óid-e ‘knows’. This, however, is merely conventional wisdom, not established fact; the peculiarities of *u̯óid-e make any analysis, however ‘standard’, provisional.

  2. (2) Because the resultative past meaning of vivéda / vīuuaēδ‎ a in Indo-Iranian is a typologically ‘late’ feature. dadárśa / dādarәsa and śuśrāva, however, have this meaning as well, as do almost all contrastively active perfects in Vedic and Younger Avestan.

  3. (3) Because *u̯eóid-e / vivéda / vīuuaēδ‎a lacks cognates and seems to be a purely Indo-Iranian innovation. But in fact, the assumption of a perfect *u̯eoid- / *u̯eid- provides the only satisfactory basis for explaining Lat. uīdī ‘I saw, have seen’ < *wiwidai, with *-iwi- giving *-ī- as in uīcī ‘I (have) conquered’ (= OIr. do·fích ‘fought’) < *wiwikai.6

The communis opinio notwithstanding, therefore, it would seem legitimate to assume that the original PIE perfect of *u̯eid- was (3 sg.) *u̯eóid-e ‘is visible / recognizable, is found (as)’, parallel to *dedór-e ‘is visible’ and *k̂e-e ‘is audible, is famed as’. Just as the latter two forms split within PIE into a perfect active (*dedór-e ‘sees’, *k̂e-e ‘hears’) and a perfect middle (*dedr̥k̂-ór ‘is visible’, *k̂elu-ór ‘is audible, is famed as’), so *u̯eóid-e gave late PIE *u̯eóid-e ‘sees, is in the condition of having found / recognized’ (> Ved. vivéda, YAv. vīuuaēδ‎ a, Lat. uīdī) and *u̯eid-ór ‘is visible / recognizable, is found (as)’ (> Ved. vividé). It follows that the familiar *u̯óid-e ‘knows’ is neither the (p.231) relic of an earlier period in which perfects were unreduplicated, nor a phonologically evolved form of *u̯eóid-e.

What, then, was *u̯óid-e? A hint is provided by the fact that alongside the well-attested 3 sg. véda ‘knows’, the Rigveda furnishes eleven examples of a middle vidé ‘is known’, mainly but not exclusively in the formulaic phrase yathā́ vidé ‘as is known’ (cf. Kümmel 102 ff.). From a descriptive point of view, vidé is the perfect middle corresponding to véda, and as such could probably have been created at any time between late PIE and the period of our texts. But there is another possibility, more attractive in view of the problem at hand. The root *u̯eid-, in its intransitive reading, was associated with a stative-intransitive system, of which we have already met the stative-intransitive aorist *u̯oid- / *u̯(e)id- (> Ved. avedi) and the perfect *u̯eoid- / *u̯eid-ˊ (> Ved. vivéda). Another member of this derivational family was the root stative-intransitive present *u̯id-ór ‘is / becomes visible / recognizable’, standing in the same relation to the aorist *u̯óid-e ‘came to light’ as e.g. *bhudh-ór ‘is / becomes awake’ (cf. OCS -bъditъ) to *bhóudh-e ‘awoke’, *u̯áĝ-ór ‘is / becomes broken’ (cf. Toch. B wokotär) to *u̯ā́ĝ-e ‘broke (intr.)’, and—especially interesting in view of the parallelism between *u̯eid- and *k̂leu- —*k̂lu-ór ‘is / becomes famed as’ (cf. Latv. sluv, -êt ‘become known’, cited by Kümmel 14, 154) to *k̂-e ‘became audible, became famed as’. PIE *u̯id-ór ‘is / becomes visible / recognizable’ would have given a Vedic stative-intransitive present *vidé, which would presumably have meant ‘is seen, is recognized, is found (as)’. It is natural to see a connection between this expected present *vidé, which is not attested in our texts, and the perfect middle vidé ‘is known’, which is.

The possibility that vidé ‘is known’ was a transformed root stative-intransitive present rather than an ordinary perfect middle is considered by Kümmel, who remarks (104), ‘es wäre auch denkbar, daß in vidé “ist bekannt” ein alter Stativ der Voraussetzung “ist erkennbar” vorläge und dieser erst sekundär aus formalen Gründen zum patientiven Medium / Stativ von véda umgedeutet worden wäre.‘ Kümmel, in other words, contemplates a scenario in which the root stative-intransitive present vidé was ‘captured’ by the perfect véda and pressed into service as its middle. The semantic shift from ‘is / becomes visible’ to ’is known‘ would have followed from the morphological association of the two forms: since véda meant ‘knows’, vidé, by attraction, came to mean ‘is known’.

Kümmel's hypothetical account is doubtless a step in the right direction. More can be explained, however, and the formal identity of the two vidé‘s can be better exploited, by assuming that the meaning ’is known‘ developed from ‘is / becomes visible / recognizable’ within PIE itself, prior to the creation of *u̯óid-e-‘knows’. According to the view that we shall take here, the crucial event in the history of Ved. vidé—the crucial event, in fact, in the history of the whole complex of forms meaning ‘know’ and ‘be known’—was a spontaneous semantic change in the PIE stative-intransitive present *u̯id-ór:

stage I

*u̯id-ór ‘is / becomes visible / recognizable’

stage II

*u̯id-ór (A) ‘is / becomes visible / recognizable’; (B) ‘is known’

(p.232) The extension of the meaning of *u̯id-ór from ‘is / becomes visible / recognizable’ to ‘is known’ hardly calls for comment; it is simply the passive-voice analogue of the commonly assumed change of a PIE perfect meaning ‘sees, is in the condition of having found / recognized’ to ‘knows’. The latter widely assumed event, while making excellent ‘sense’ and often taken for granted, in fact never took place: the PIE perfect of *u̯eid- was *u̯eóid-e, and *u̯eóid-e gave Ved. vivéda and Lat. uīdī.

The establishment of *u̯id-ór in its new meaning would have cleared the way for a key further development. *u̯id-ór ‘is known’ contrasted minimally with another PIE form, the perfect middle *u̯eid-ór ‘is visible, is recognized’. To a learner of PIE at this stage, it would doubtless have appeared as if *u̯id-ór was simply an unreduplicated variant of *u̯eid-ór, and that the absence of reduplication was responsible for the difference in meaning between the two forms (‘is known’ vs. ‘is visible, is recognized’). *u̯id-ór, in short, would have looked like, and tended to pattern as, an unreduplicated or dereduplicated perfect middle:

reduplicated perfect

unreduplicated perfect


*u̯eóid-e ‘sees, has seen / recognized’


*u̯eid-ór ‘is visible, is recognized’

*u̯id-ór ‘is known’

From here it would have been a natural step to fill in the missing active term. The proportion was

mid. *u̯eid-ór ‘is visible / recognized‘ : act. *u̯eóid-e ‘sees, has seen / recognized’ : :

mid. *u̯id-ór ‘is known’ : X,

where X was solved as *u̯óid-e ‘knows’.

*u̯óid- / *u̯id-ɛ ‘know’, by far the best known and most frequently cited perfect in the PIE verbal system, thus emerges not as an archaism but as an inner-IE neologism—a back-formation from its own middle, which in origin was not a perfect at all but a root stative-intransitive present.7 It goes without saying that this analysis clashes with many traditionally held views. It is widely held that the perfect middle was too ‘late’ to play any significant role in the evolution of the PIE verbal system; that Ved. vivéda and YAv. viuuaēδ‎ a were simply inner-Indo-Iranian regularizations of unreduplicated véda and *vaēδ‎ a; and, above all, that *u̯óid-e, with its specialized meaning and lack of visible (p.233) reduplication, was a form of immense antiquity. The fact that these views have become traditional, however, does not make them true. The alleged chronological priority of *u̯oid- / *u̯id- over *u̯eoid- / *u̯eid- is not supported by any actual evidence, and is incompatible with what we now know about the relationship of the perfect to the stative-intransitive root aorist. The perfect *u̯eoid- / *u̯eid- is solidly grounded in the comparative record, being directly attested in Italic as well as in Indo-Iranian and structurally supported by the perfects *dedork̂- / *dedr̥k̂- and *k̂elou̯- / *k̂elu-. And the perfect middle, while indeed a more recent creation than the perfect active within the inferable prehistory of the IE family, was nonetheless a genuine PIE category, fully capable of mediating the entry of *u̯óid-e into the PIE verbal system by the back door.


(1) A full survey of the literature on *u?óid- / *u?id-' is given by di Giovine (1996: 127 ff.), whose own view favours an inner-IE loss of reduplication.

(2) To which can probably be added OIr. ad?condairc ‘saw’, although the possibility of an underlying root aorist cannot be altogether excluded.

(3) Cf. also Gk. ?????‎, ???te‎ ‘hear!‘

(4) The thematic aorist is reflected in Ved. ávidat, Gk. ?‎(F)?de‎, and Arm. egit ‘found’, which together constitute one of the best morphological equations in IE comparative grammar (cf. Ch. 8 n. 5). LIV, which anomalously denies PIE status to the thematic aorist, sets up a root aorist *u?eid- / *u?id-, from which both the thematic forms and Lat. uidi ‘I saw’ are said to be derived (606 f.). But Lat. uidi is better explained otherwise (see below), and it is scarcely credible that the Indo-Iranian, Greek, and Armenian forms represent independent thematizations.

(5) LIV classifies the perfect *dedórk?-e as ‘sicher’ (105); *k?ek?lóu?-e receives a question mark (297), implying merely ‘eine gewisse Wahrscheinlichkeit’ that it goes back to PIE.

(6) Cf. Thurneysen 1946: 435, where other ‘T-preterites’ are also discussed.

(7) There is good reason to believe, in fact, that the shift from *u?id-ór ‘is / becomes visible’ to *u?id-ór ‘is known’ was not absolute; the older (root stative-intransitive) and newer (perfect middle) readings co-existed, perhaps as separate lexical items, in late PIE. In Germanic and Balto-Slavic the presential reading of *u?id-ór (‘is / becomes visible’) led to the back-formation of a present active *u?éid-mi, *-si, *-ti ‘see’ (NB: not *u?e´id-mi, pace Jasanoff 1978a: 108), which survives in OLith. veizdmi ‘I see’, OCS impv. vižd? ‘see!‘ (< *-d-i?es), and possibly OE witan ‘guard, look after’. In relation to the new active, the stative-intransitive paradigm acquired the subsidiary value of a ‘true’ middle (‘sees with reference to oneself’, etc.), and in this capacity gave rise to Go. witaip ‘observes’, Lith. pa-v?di (inf. -e´ti) ‘envies’, and probably OCS vidit? (inf. -eti) ‘sees’. It is also possible that an athematic active *u?éid-mi, etc. existed in the prehistory of Greek, where (F)e?"deta?‎ ‘appears’ can perhaps be seen as the old root stative-intransitive *u?id-ór, with analogical e-grade from the (lost) active *(F)e?"d?‎ ‘see’.