2014-12-26 1 , S. E. & O.
The Proto-Indo-European aspect system
ROLAND A. POOTH
“Was wir als Anfänge [d.h. als Urindogermanisch] glauben nachweisen zu können, sind ohnehin schon ganz späte [d.h. nachurindogermanische] Stadien.” à la Jakob Burkhardt Weltgeschichtliche Betrachtungen quoted from Parzinger 2014: 12
The reconstructable Proto-Indo-European language exhibited an elaborate inflectio-nal aspect system including over twenty specific aspect categories. From PIE to the Vulgar Pre-Indo-European dialect or variant cluster, many of these aspect categories were semantically broadened and merged, and their number decreased. The aspect system was “fused” with tense distinctions and was thus remodelled to a tense and aspect system including, among others, a prominent PRESENT IMPERFECTIVE tense and aspect category.
Keywords: Proto-Indo-European verb morphology, aspect, Aktionsart, tense.
1 PIE verb morphotaxis
It has recently been discovered that “Proto-Indo-European Proper” 2 verb morpholo-gy was of the root and pattern morphology (RPM) type. 3 According to the definition given by Bauer 2004: 93, languages of this type have the following property:
“[...] In these languages, the root in a number of common binyanim [4] or paradigms may be analysed as being made solely of consonants, while the pattern of vowels which are found around the consonants and the particular vowels filling up the pat-tern provide morphological information comparable to that often given by affixation. This analysis leads to discontinuous roots and discontinuous morphs interacting with the roots, [...].”
1 This paper has been published at www.academia.edu on 2014-12-26 as provisional grey literature and work in progress and will undergo further revisions. Merry christmas!
2 This term has been coined by Ringe 2006: 5. I avoid terms like “Early PIE” or “Late PIE”, because labels like this presuppose the notion of two chronological layers of PIE which I find a bit problematic. The term Vulgar Pre-Indo-European does not refer to a more or less homogeneous and standardizable languages, but to a divergent post-PIE variant or dialect cluster (comparable to what will be spoken in many parts of the world someday after the decline of the English standard language).
3 Cf. Pooth 2004a, 2009b; Tremblay diss. 1999, 2003.
4 Cf. Classical Hebrew binyānî ́ m.
ROLAND A. POOTH 2
Besides the skeletal consonant frame each PIE verbal finite word form contained a transfix, cf. Bauer 2004: 102:
“A transfix is a particular type of affix, one which is completely interwoven with its base. Typically,it is a series of vowels which surround and interact with a base which in turn can be analysed as a series of consonants. For example, Arabic katab ‘he wrote’, kitaab ‘book’, kaatib ‘clerk’ (where the root is *ktb, indicating ‘writing’) illust-rate the transfixes _a_a_, _i_aa_ and _aa_i_. Such transfixes [...] are discontinuous affixes attached to discontinuous bases, [...].”
I use the term vowel melody for transfix, however,because I follow the terminological tradition of the autosegmental approach (cf.McCarthy 1981).The different morpheme levels within the autosegmental morphological analysis are termed “tiers”. 5
The vowel melody (VM) on the vowel melody template tier (e.g. _V_) was morpho-tactically independent from the agglutinating affixation on the skeletal consonant frame (CF) tier (e.g.*d ɦ i-d ɦ _ʔ-t-i) An example of the different tiers is given below:
The separation of the different morphological tiers can also be illustrated by the fol-lowing figure; the entire word form is PIE *stɛ́utoi ‘topical referent is praising s.o. for topical referents’s own benefit’:
5. Cf. McCarthy 1981.
The Proto-Indo-European aspect system 3
In addition to transfixation on the vowel melody tier and the word form template tier, PIE morphology made use of the morphotactic strategy of agglutination on the con-sonant frame tier.After subtraction of the discontinuous full vowels (*ɛ, *ɛː, *ɔ, *ɔː) an agglutinating C-chain (reduplication, root, infix, suffix) remains (ro_ot = discontinuous root, interrupted by a vowel slot _V_):
The reduplication templatic prefix, the two aspect suffixes (i.e., *-nV 4 u-, *-sk-), and the aspect infix were in complementary distribution. IE reduplicated present stems like Greek διδάσκω ‘I teach’ are obvious secondary innovations showing affix pleo-nasm (PIE *ɗiɗnsɔ́: PIE *ɗnsskɔ́→) *didn̥ssk ̑ó/é- > *didn̥sk ̑ó/é-, cf. Old Avestan didąs ‘teaches’, dīdaiŋhē ‘I experience, get to know’, etc. PIE had a vocalic prefix (*ɛ́-) always attracting the accent (e.g. *ɛ́-ʔɛst ‘was there, existed, sat (down) there). Although being vocalic, it did not belong to the vowel melody. It was used optionally to specify past tense reference.It is used as past tense prefix in Greek, Phrygian and Indo-Iranian (and is partially preserved in Armenian). The order of affixes in PIE verb forms is given in the following figure.
6 In principle, I follow the Leipzig Glossing Rules. Glosses: DIR = DIRECT-TRANSITIVE direction, INV = INVERSE-TRANSITIVE direction, PROG = PROGRESSIVE aspect, DEB = DEBITIVE mood, EXCL = 1PL.EXCLUSIVE, ITR = INTRANSITIVE-UNDERSPECIFIED direction (cf. Pooth manuscript e), etc.
7 The H is used for ‘voice suffix slot’, because V is used for ‘vowel slot’.
ROLAND A. POOTH 4
Cover symbols:
T narrative past tense prefix (*ɛ́-)
RE reduplication templatic prefix
R “root” or lexical base
A aspect infix or suffix (*~n~, *-nV 4 u-, ...) 8
M modal suffix (*-iV 5 ʔ-, *-ʔs-)
P person-and-direction 9 suffix (1EXCL *-m-, 1INCL *-u-, 2/3DIR *-t-, 2/3INV *-s-, 2/3ITR *-Ø-)
H detransitive voice suffix (*-χ-)
N number suffix (1PL *-s-, 2PL *-n-, 3PL *-r- ~ *-n-, 2/3COL *-χ-)
D direction-and-deixis suffix (2/3COL.PROX *-m, DIR *-t-, INV *-s-)
F final progressive aspect suffix (*-i-) or debitive mood suffix (*-u-)
V vowel slot of the vowel melody
Each PIE verbal finite word form thus obligatorily consisted of a minimum of two overt morphemes:
(a) the skeletal “root”, that is, the discontinuous lexical base (LB) 10 and
(b) the discontinuous inflectional vowel melody (VM).
The other morphemes or morpheme slots could remain non-overtly zero-marked or “unmarked”, depending on how one prefers to define the status or notion of zero (Ø). The following figure can illustrate the internal morphotaxis of a PIE 2 nd person PLURAL DETRANSITIVE DIRECT-TRANSITIVE PROGRESSIVE form *ɗiɗnstχáni ‘you (pl.) (TOP) are experiencing /getting to know REF now & then ...’: 11
Cf. Old Avestan dīdaiŋhē ‘I experience, get to know’ from the root *ɗ_ns-, cf. LIV, p. 118f. NB. From PIE to Vulgar Pre-IE, *-tχán (without *-i) was reanalyzed as a new 2nd pl. active or “neoactive” portmanteau “ending”. It is reflected as Proto-Indo-Iranian 2nd pl. present imperfective (so-called “primary”) active ending *-thanā ̆ > Vedic -thanā ̆ , cf. Pooth 2011.
8 A o = onset of the aspect suffix syllable template (e.g. *-i_ of *-i_ʔ-), A c = coda of the aspect suffix; likewise M o and M c .
9 For the term direction cf. Wolfart & Carroll 1981: 68; DeLancey 1981; Thompson 1989: 21; Klaiman 1992; for “Transitivity Direction in Proto-Indo-European” cf. Pooth manuscript e.
10 A lexical base (LB) can further be separated into the proper root and its derivational or quasi-derivational “enlargement” (ENL).
11 Abbreviations: TOP = topical referent, REF = non-topical referent.
The Proto-Indo-European aspect system 5
To illustrate the PIE to Vulgar Pre-IE great morphotactic fusion, the consonant frame can be separated into two major parts, namely
(a) the “aspect and mood stem” and
(b) the “ending”:
The “aspect and mood stem” was later fused to the IE aspect and mood stem (e.g. *dʰi-dʰ_ɛ́_ʔ- → *dʰidʰéh-ti , etc.).
The respective “word form ending”, on the other hand, was fused to the respective IE portmanteau suffix, that is, the so-called “inflectional ending” for tense, aspect, mood, person, number, and voice (e.g. *-t-χ_á_n → *-th2an, etc.).
The distinction of “aspect and mood stem” and “ending” is solely motivated to illust-rate the post-PIE morphotactic fusion. It is not implied that these stems were “PIE Proper” synchronic morphological units. The emergence of post-PIE and IE fusional aspect and mood stems (e.g.*didéns- ~ *didn̥s-´) was thus triggered by the process of morphotactic fusion. (It was not triggered by suffixation of clitics to a fusional word form.)
Remark: However, the younger IE so-called “thematic” stems and the IE “sigmatic” stems developed from a later resegmentation:
(a) From PIE to Vulgar Pre-IE, many 3 rd person SINGULAR DETRANSITIVE INTRANSITIVE forms 12 were pleonastically extended by the new productive middle “endings” (*-to(i) ~ *-tor(i)). In parallel, the former PIE 3 rd person SINGULAR INVERSE forms were pleonastically extended by the 3 rd singular so-called “secondary” “ending” (*-t).
(b) Within the Vulgar Pre-IE dialect or variant cluster, the respective first segment of *-oto(i) ~ *-otor(i), *-eti, and the one of *-st were reanalyzed as a stem-final suffix *-o- ~ *-e- and *-s-. This resegmentation mainly occured in the so-called “Inner IE” part of the Vulgar Pre-IE dialect or variant cluster. This “morphotactic internalization” has brilliantly been described by Watkins 1962, 1969. In honour of Watkins it is called “Watkins’ law”, cf. Collinge 1985: 239. This resegmentation is transferrable to the IE *-sk ̑ o/e-stems (see figure 7):
12 These have been termed “stative”, but this label is inappropriate. Instead, they were 3 rd sg. detransitive forms used in a PIE intransitive construction (including the antipassive construction), cf. Pooth 2000 and manuscript a. They had labile semantics.
13 This form is reflected as Homeric Gk. 3 rd sg. pres. ind. mid. εἴδεται, 1 st sg. εἴδομαι.
ROLAND A. POOTH 6
A parallel suffix pleonasm happened to the corresponding 3 rd person singular de-transitive intransitive forms of the progressive aspect which were marked by the suf-fix *-i in the final morpheme slot (F). These PIE specifically PROGRESSIVE aspect forms were semantically broadened to Vulgar Pre-IE present imperfective tense and aspect forms. They were extended, then, by the productive, specifically present im-perfective middle “endings” *-toi ~ *-tori,etc. or by the productive active (including the “neoactive”) ones (*-ti ~ *-ei ~ *-eti, etc.). The subsequent resegmentation of the re-spective first part of the pleonastic “ending” to stem-final suffixes (middle *-oi- toi , *-i- toi , *-i̯o- toi , *-ei- toi and “neoac- tive” *-i/i̯- ei , *-i̯e- ti , *-ei̯e- ti ) 14 is now datable to the Vulgar Pre-IE period. The “internalization” must have happened slightly before or by the time when Proto- Anatolian split up from the dialect or variant cluster:
NB. 3 rd person sg. forms of the *ɠɔ́nʔi type are reflected as the Vedic so-called “passive aorist” (Vedic jáni, etc.). Cf. Proto-Germanic *đuγaiþ(i), Vedic duhyate :: Gaulish dugiiontiio, etc. Cf. Latin 3 rd sg. pres. ind. mid. oritur < *ʕwóritor(i) ~ *ʕwóritoi (← PIE *ʕɔ́ri) :: Hittite 3 rd sg.pres ind. act. araai < *ʕwró́i (PIE *ʕwrɔ́i), pl. ariyanzi < *ʕwr̥i̯ónti ~ *ʕwrónti. The 3 rd person pl. forms *ʕwr̥i̯ónti ~ *ʕwr̥i̯óntoi were created via paradigmatic levelling of *-i- in Vulgar Pre-IE. 15
Systematic suffix pleonasms were a general post-PIE tendency. These “prolonged” word forms were triggered by paradigmatic analogical leveling or regularization, that is, the analogical introduction of the new productive portmanteau “endings” (i.e., 3 rd *-toi, *-ti ~ *-ei, etc.). It is thus possible to reconstruct the following two PIE forms of the 3 rd sg. detransitive intransitive progressive via internal reconstruction:
14 A second, but minor source of IE yod-present stems were derivational “root enlarge- ments” in *-i-, e.g. *ɗ_χ-_i- ‘separate, detach, devide, distribute, share’ from underlying *ɗ_χ- ‘id.’ (cf. LIV, s.v. “deh 2 (i ̯)-”).
15 This offers a plausible explanation for why the partciple of Hittite araai does not show any *-i ̯-, cf. araant-, cf. Kloekhorst 2008: 200.
16 I am sorry, but I do not “assume, as a descriptive fact, ‘acrostatic’ iterative-causatives of essentially the traditional sort at the level of PIE” (Vine 2012: fn. 11).
The Proto-Indo-European aspect system 7
NB. Vedic śrā ́ myate ‘gets tired, slack’ is a reflex of Vulgar Pre-IE *k ̑ róːm(H)i̯etoi going further back to PIE *krɔ́ːm(H)i ‘is getting/being slack’. It is evident that the Vulgar Pre-IE suffix of the *su̯óːpi̯eti type was *-i̯o/e-, cf. the discussion by Vine 2012: 548-555.
A brief remark on suffix pleonasms: There are plenty of well-known cases of suffix pleonasms crosslinguistically, cf. the discussion and the examples of Haspelmath 1993: 297 with references. Cases such as Afro-American Vernacular English child-Ø :: children-s, Modern Dutch kind-Ø :: kinder-en or Vulgar Latin esse-re (cf. the Latin irregular infinitive esse) show that the first part of pleonastically chained suffixes can be reanalyzed as belonging to a new stem allomorph (esse → essere ). It is thus evident that suffix pleonasms lead to additional allomorphy (cf. Haspelmath 1993: 299). The fact that Vedic stáve ~ stavate, śáye ~ śáyate, juṣat ~ juṣáta, etc. are variants without any functional difference thus only strengthens the case for “Watkins’ law”.
Note that the given diachronic scenario has the implication that the *-i̯o/e- (and *-Hi̯o/e-) thematic stems and the ones in *-sk ̑ o/e- and *-éi/éi̯o/e- came into being a bit earlier than the simple “thematic” ones (*-o/e-). Such a chronology is supported by the Proto-Anatolian situation.
2 The six basic transfixal aspects
It is among the most intricate and fascinating facets of the “PIE Proper” system of transfixing and templatic morphology that the vowel melodies were integrated into a reciprocal system of verbal internal inflection, that is, a system of verbal base modifications by means of vowel transposition (VTP). I will return to a definition of this phenomenon below.
PIE had an internal inflectional system of minimally six transfixal verbal “grades” 17 (counted by Roman numbers I, II, III, IV, V, VI). These “grades” were altering vowel patterns which were mapped upon the underlying vowel melody template (*_V_, *_VV_, or *_V_V_). Each grade thus consisted of a particular vowel melody (e.g. *_ɛ_) or a combination of derivationally related vowel melodies (e.g. *_ɛː_ → *_ɛ_ɛ_).
Thus in “PIE Proper” it was not the root or the suffix that contained a particular “ablaut grade” like in Vulgar Pre-IE or in the IE languages. Instead, the term “grade” is now transferred to identify a property of the entire word form. The essence of the grade system is the classification of verbal finite word forms on the basis of their respective underlying word form template. Each PIE verbal finite word form belonged to a specific grade. Each grade, as so defined, had semantic correlates. The six PIE grades signalled six basic (inflectional) aspectual meanings. They could further be combined with reduplication, infixation, and suffixation yielding more specified (derivational) aspectual meanings.
17 The term “grade” is borrowed from the term for tonal patterns of Hausa grammar, e.g. Hausa jeefaà (grade I) ‘throw sth.’ :: jèefaa (grade II) ‘throw at s.o.’ :: jeefoo (grade VI) ‘throw (in this direction)’, cf. Newman 1973: 298. This term is also used to describe the aspect system of Creek (Muskogee, spoken by the Seminole tribe, Oklahoma and Florida, USA), cf. Martin 2011: 43ff., 241ff., chapters 8 and 28.
ROLAND A. POOTH 8
The PIE grades and their vowel melodies are given in the figures below.Grade III, IV, V, VI were deponent grades. Only detransitive forms belonged to these grade; there were no corresponding agentive-active forms. A deponent grade is indicated by d here:
These six transfixal grades including their combined vowel melodies were motivated by so-called “internal derivation”. This derivational strategy implied the following morphological mechanisms:
(a) Vowel transposition:
Templatic vowel transposition was an important morphological mechanism in PIE. By definition, it implied an internal change of the position of a vowel or vowels in the respective vowel slot(s) within the word form template:
18 The PIE NONDURATIVE aspect was either semelfactive-deliminative, that is, it indicated a single event within its two boundaries occurring once, or it was terminative (including a termina-tion) or telic (including a goal).It was more underspecified as for duration than a perfective aspect and was not incompatible to the PROGRESSIVE aspect suffix *-i which derived a progressive durative meaning from otherwise nondurative or underspecified polyactional roots. The gloss NONDUR used here is equivalent to the gloss SEM used in earlier manuscripts (Pooth manuscrips a-e).
19 The vowel melodies given in brackets were the vowel melodies without the superimposed discontinuous marker *ɔ. These unmarked vowel melodies coded detransitive 1 st person singu-lar and 2 nd person forms in combination with the continuous detransitive suffix *-χ-.These vowel melodies were identical to the unmarked agentive-active vowel melodies of grades I and II, respectively.
20 The detransitive forms of this type also had a “future-prospective” or potential reading, e.g. PIE *mɛ́rɔ ‘is mortal, can die, will die, shall die’. Forms with this reading developed to subjunctive stems *méro/e- within Vulgar Pre-IE.
21 The vowel melodies are coloured red here.
The Proto-Indo-European aspect system 9
NB. A quite similar morphological strategy is found in the Oceanic language Rotu-mam: e.g., piko (CV 1 CV 2 ) ‘lazy’ :: piok (CV 1 V 2 C) ‘lazy’, rotuma (CV 3 CV 1 CV 2 ) ‘Rotuma’ :: rotuam (CV 3 CV 1 V 2 C) ‘Rotuma’, 22 etc., cf. Besnier 1987: 201-223, Pooth 2004a: 422, fn.
The PIE 1st person exclusive, 1st person inclusive, and 2nd person plural agentive forms were internally derived from the corresponding singular forms by means of vowel transposition, see the following figure:
The detransitive forms of grade I were internally derived from the detransitive forms of grade IV by vowel transposition. 23
22 Cf. Besnier 1987 who speaks of “vowel metathesis”, but the term “metathesis” should better be restricted to a non-morphological switch of segments.
23 Cf. Jasanoff 1978, footnote 29: “The possibility that the stems under discussion owe their zero-grade to a process of internal derivation is not unattractive [...]”.
24 This form is reconstructed by the method of internal reconstruction, that is, by internal subtraction of *-i. The corresponding progressive form *b ɦ ɔ́ud ɦ i is reflected by Vedic bódhi.
ROLAND A. POOTH 10
The detransitive 2nd /3rd person collective-plural forms were internally derived from the 2nd person singular forms:
The respective vowel melody of the 3rd person plural detransitive forms was internally derived from the one of the corresponding singular forms by vowel transposition:
NB. These 3rd person plural forms were marked for plural number by the 3rd person plural number suffix *-r- (or *-n- before *-t-). Verbal and nominal “internal derivation” further implied the following morphological means:
(b) Accent shift:
A second PIE morphological means was simple accent shift: e.g., *lɛ́uqɔ ‘is, was shining’ → *lɛuqɔ́- (adjective) ‘bright, shining’. 25
But simple accent shift was more important in the system of nominal derivation than in verbal morphology: e.g.,*ɗɔ́ru inanimate ‘wood’→ *ɗɔrú- ‘wooden’ (adjective), loc. *ɗɔ́rui ‘in/at wood’ → *ɗɔruí- ‘in wood (ATTRIBUTIVE), wooden’ (cf.Modern German ein Stuhl in Holz ‘a wooden chair’), etc.
(c) Vowel slot gemination:
Another salient PIE morphological means was gemination of the vowel slot on the vowel melody tier of the word form template:
The agentive forms of grade II were internally derived from the ones of grade I by means of vowel slot gemination:
25 Cf. Greek λευκό- ‘white’; the corresponding verb form is reflected by Vedic rócate ‘is shining’.
The Proto-Indo-European aspect system 11
NB. From PIE to Vulgar Pre-IE, the 3 rd person inverse-transitive forms were pleonastically extended (e.g. *ɗɛ́ːks→ *dɛ́ːk ̑ st).
(d) Suprasegmental vowel mapping:
A tremendously important and even more intricate morphological means of PIE was the suprasegmental mapping of the discontinuous DETRANSITIVE marker *ɔ (or else the phono- logical feature [+round] 26 ) upon one of the underlying vowel slots of the vowel melody template. It is illustrated by the following figure:
It is crucial for any deeper understanding of PIE morphology to recognize that all word forms coded by *ɔ, no matter in which position within the word form, and also all word forms coded by its continuous counterpart *-χ- belonged to the DETRANSITIVE voice category. The following detransitive forms were internally derived from the corresponding agentive word forms by suprasegmental vowel mapping:
26 This depends on how one prefers to analyze this phenomenon. Note that the PIE “bare” vowel phoneme */ɛ/ was realized as [ɑ] or [a] before and after *χ, but as [ɒ] before and after *ʕw. Both realizations are written <a> here. Otherwise it was realized as [æ] or [ɛ] (written <ɛ> here). I think that the realization [ɑ] or [a] before and after *χ was older than [æ] or [ɛ] otherwise. PIE also had *a and *aː in onomatopoietics and “small word forms”, e.g. *máː ‘mummy’ vs. *mɛ́ː PROHIBITIVE. Some roots had variants, e.g. *ʔVr- ~ *ʔar-. I think that, e.g., *ʔarɔ́ was older than its regular variant *ʔrɔ́ (1st binyan).
ROLAND A. POOTH 12
NB. This system offers a functional explanation for the later different ablaut grades of *-me(s) and *-mo(s). The 1 st person plural middle ending *-mor of Proto-Celtic and Proto-Italic cannot be a complete innovation: *-mo (without *-r) should be seen as archaic. The plural middle endings *-mo(s)d ɦ χ, *-u̯osd ɦ χ, *-(s)d ɦ u̯o/e, and *-onto, on the other hand, should be taken for Vulgar Pre-IE innovations (cf. Pooth 2011); see below.
It can be concluded that the grade I agentive singular forms (*dʰɛ́ʔt > Vedic 3rd sg. aor. inj. act. dhā´t, etc.) were the most “basic” forms. Many detransitive forms (e.g. *dʰɔ́ʔ, *dʰɔ́ʔi > Vedic 3rd sg. aor. inj. mid. dhā ́yi) were internally derived from the underlying agentive forms (e.g. *dʰɛ́ʔ) by mapping *ɔ upon a vowel slot of the vowel melody template. In terms of markedness, therefore, the detransitive voice forms contained more morphological material than the underlying agentive voice forms. Note that this situation is quite different from the one found in the most archaic IE languages, where active and middle forms show the same number of morphemes, e.g. Vedic 3rd sg. pres. ind. act. bhára-ti vs. mid. bhára-te.
The grade I detransitive forms were internally derived from the grade IV forms, e.g. *bʰɔ́udʰ ‘wakes up, woke up; gets/got attentive’ → *bʰudʰɔ́ ( 27 ) ‘recognizes/recognized sth./s.o.; is/was/gets/got attentive towards s.o./sth.’. The forms of grade III (*uɔ́iɗɛ) were internally derived from the grade II detransitive forms, e.g. *uɛ́iɗɔ(i) ‘s.o. can see; s.o./sth. can be seen, is visible; is apparent’ → *uɔ́iɗɛ(i) ‘knows sth./s.o.’ (> Vedic véda, etc.).
It can thus finally be concluded that a monovocalic underlying vowel melody template *_V_ had a nondurative or a (nondurative) transitional aspectual meaning (e.g. *uiɗɔ́ ‘s.o. found sth./s.o.’ 28 ), whereas a bivocalic (or “stereovocalic”) underlying vowel melody template *_VV_ or *_V_V_ had a durative, a (durative or “plurative”) stative-habitual or a (“plurative”) distributive-iterative aspectual meaning. In other terms, a monovocalic vowel melody had a SINGLEFACTIVE-SINGULATIVE (BOUND, PUNCTUAL) or else an UNDERSPECIFIED aspectual meaning, whereas a bivocalic vowel melody conveyed a specific PLURIFACTIVE-PLURATIVE (EXTENDED, INTERNALLY MULTIPLIED, EXTERNALLY MULTIPLIED or UNBOUND) aspectual meaning:
One may use the term superordinate vowel melody template aspects to refer to these two templatic aspect distinctions. The terms “imperfective” vs. “perfective” are rather inappropriate here, because the singlefactive aspect was compatible to the progressive aspect and thus conveyed a somewhat different aspectual meaning (which was less specified than the perfective aspect).
27 This form is reflected as the Greek thematic aorist ἐπυθόμην.
28 This form is reflected as the IE thematic aorist *u̯idó/é-, e.g. Vedic ávidat.
The Proto-Indo-European aspect system 13
As an exception, the 3rd pl. forms of the so-called “Narten type”, namely *stɛ́ur, *stɛ́unt, *stɛ́urs and the 3rd pl. of grade V (*suɔ́pr) showed a monovocalic-singlefactive (underspecified) vowel melody template (*_V_), but belonged to a pattern of word forms which exhibited a plurifactive vowel melody template otherwise. 29
3 The PIE progressive aspect
The word final morpheme slot -F (see figure 4) provided a slot for the PROGRESSIVE aspect suffix *-i or the debitive mood suffix *-u, but the latter will not be dealt with here. In PIE, the progressive aspect marker *-i could be attached to almost any finite verb form except for verb forms whose grammatical meaning was completely incompatible to the progressive durative aspectual meaning (cf. Pooth 2009a); see the figure below:
Remark: As already mentioned above, this suffix *-i was fused with the other suffixes of the “ending” from PIE to Vulgar Pre-IE and became part of the so-called “primary (portmanteau) endings” marking the Vulgar Pre-IE present imperfective tense + aspect category. The most productive Vulgar Pre-IE present imperfective aspect and tense “endings” are given in the following figure. Note that there was more variation; the 1st and 2nd person pl. forms, for instance, could optionally lack the *-i.
29 Note that these forms were marked for plural number by *-r- and *-n- (before *-t-). These markers thus may have coded plurality of the given event concept.
30 The new 1 st pl. and du. incl. middle endings obviously developed by attachment of a former clitic *=dʰχ (or *=dʰa ~ *=dʰi?) whose origin remains obscure. This clitic may be related to the deictic clitic *=dʰi that was attached to 2nd sg. imperative forms, cf. PIE *ʔɛ́s=dʰí ‘be there!’ (> Vulgar Pre-IE *ʔesdʰí ~ *ʔsdʰí). It may also be a form *dʰʔχ́ ‘one does/did it (for one’s own benefit, ...); it was done (by a group of people)’ (formed like a 1st sg. middle *g̑nʔχ́ ) which as a relic contained the old collective meaning of the marker *-χ- found both in 1st person sg. and 2nd person detransitive forms.
ROLAND A. POOTH 14
Additionally, there were new Vulgar Pre-IE middle “endings” with a new suffix *-r(i). I follow the “old” view that this extension ultimately originated from PIE 3 rd person plural detransitive intransitive forms (e.g. *stɛ́uɔr ~ *stɛ́urɔ ‘some people praised s.o.’) which were coded by the PIE 3 rd person plural marker *-r- (in the number slot -N-). I think that the PIE 3 rd person plural detransitive intransitive forms were reanalyzed as new 3 rd person singular middle forms within the new Vulgar Pre-IE passive construction. This new passive construction emerged by addition of an oblique causer or agent to the original 3 rd person plural intransitive construction:
(1) a. PIE (oblique agent ungrammatical) *χnɛ́r-Ø *stɛ́uɔ_r man-ABS:SG praise:DUR:DTR _3PL\ITR ‘as for the man, (some) people praised him’
b. Vulgar PIE (oblique agent grammatical) *χnéː(r) *stéu̯-or *pχtr-és ~ -ós man:NOM:SG praise-3SG.IPFV.IND.MID father-ABL/GEN.SG ‘the man was praised by the father’
Note that there is a second source for the new Vulgar Pre-IE passive construction. It also emerged by addition of an oblique causer or agent to the original 3rd person singular intransitive construction:
(2) a. PIE (oblique agent ungrammatical) *χnɛ́r-Ø *stɛ́uɔ man-ABS:SG praise:DUR:DTR:ITR_3SG ‘as for the man, someone praised him’
b. Vulgar PIE (oblique agent grammatical) *χnéː(r) *stéu̯-o *pχtr-és ~ -ós man:NOM:SG praise-3SG.IPFV.IND.MID father-ABL/GEN.SG ‘the man was praised by the father’
Crosslinguistically, both grammaticalization paths are well-known, cf. Haspelmath 1990: 49-50. A conflation of the passive and the middle function is confirmed. Thus *-r(i) was soon extended to be used as a general middle marker. The new Vulgar Pre-IE pres. imperfective middle “ending” variants are given in the following figure:
31 For the 2 nd person and 3rd person dual “endings” cf. Pooth 2011.
32 The origin of this ending is obscure. It may go back to a voc. sg. form, e.g. *bʰudsdʰu̯é of a verbal adjective, e.g. *gʷʰntuɔ́ , *bʰudʰtuɔ́ (>Vulgar Pre-IE *bʰudsdʰu̯ó-). It may also go back to a verb + auxiliary compound *gʷʰn-dʰʔ-u_ɔ́ ‘slaying-do-1PL.INCL_DTR’ ‘we, you and me, do/did slaying (for our own benefit)’ which was later reanalyzed as a 2nd pl. detransitive forms ‘you ... (dito)’.
The Proto-Indo-European aspect system 15
As mentioned above, there were additional pleonastic Vulgar Pre-IE 3rd person singular and plural present imperfective middle and neoactive “endings”:
As also mentioned above, there were additional pleonastic Vulgar Pre-IE 3 rd person singular and plural middle or neoactive non-present “endings”:
ROLAND A. POOTH 16
4 PIE verbal binyans
As already mentioned above, the underlying word form template (WFT) had the status of a templatic morpheme in PIE. By means of the underlying word form template, the position of the vowels of the vowel melody combined with the position of the word form accent on one of these vowels within the word form was determined. The word form template belonged to a superordinate set of word form templates. I have decided to term this superordinate template bundle the word form template set. It can also be termed more conveniently the “inflectional type”. For its brevity, however, I make use of the term binyan which is borrowed from Classical Hebrew grammar. Finally, all binyanim (binyans) were subordinate template sets to the superordinate verbal paradigm. There was nothing in PIE like a verbal lexical “conjugation class” (as found, for instance, in Latin, where the verb laudat, laudāre belongs to the first conjugation, whereas uidet, uidēre be- longs to the second one, etc.). The different “PIE Proper” binyans were fully grammatical. Instead of belonging to a lexical conjugation class, every verb was principally inflectable for each binyan. But as in many languages, there were many defective verbal paradigms. In PIE, this defectiveness was mainly due to a semantic incongruency of a given lexical meaning and the respective grammatical meaning. I will return to this matter elsewhere.
As just mentioned, each PIE verbal finite verb form had an underlying word form template (WFT). This word form template conveyed a specific inflectional meaning (remember: ro_ot is the gloss for the discontinous root or lexical base):
The word form template thus obviously had full morpheme status, because it coded number distinctions and belonged to a binyan with a specific grammatical, that is, aspectual and modal meaning.
The PIE word form accent was part of this word form template (WFT) morpheme. Its position within the word form was definitory for the identi- fication of a given word form as belonging to a specific aspect grade:
The Proto-Indo-European aspect system 17
The PIE accent was “free”, that is, it was unpredictable from syllable structure or phonological word form structure. In word forms with more than one full vowel, one of these two vowels had to bear a contrasting high tone accent, opposed to a lower tone of the other vowel or vowels. The word form accent, therefore, was not a property of any morphological segment other than the word form template (WFT) morpheme. It provided grammatical distinctions.
Since the verbal word form accent was an intonational suprasegment belonging to the verbal word form template (WFT) morpheme and was fully grammatical, PIE did not show different verbal lexical accent types. In the verbal system there was no lexically predetermined accent. The evolu- tion of lexically predetermined accent of verb stems should be seen as a Vulgar Pre-IE phenomenon which must have occured after the great morphotactic fusion. Note that PIE also had no lexical “Narten character” of roots, that is, roots with lexical long vowels.
PIE binyans can thus be defined as a combination of different word form templates. Remember that these word form templates were related to each other by so-called “internal derivation”. Within each PIE binyan, three types of finite word forms were distinguished:
The word form template morpheme is thus separable into two subordi- nate morphemes:
(a) The word form template (morpheme WFT) provided number and aspect distinctions. It was classified as belonging to a specific grade and vowel melody template (morpheme VMT) and it belonged to a superordinate word form template set or binyan which conveyed a specific aspectual (and also modal) meaning. To now provide the reader with an impression of how the PIE binyans looked like and were morphologically structured, the PIE radical binyans are given in the figures below. The first one is given in the following figure. It is the aorist-like NONDURATIVE or basic aspect. I have decided to term this inflectional pat- tern the PIE first binyan. To save space, I leave away the asterisk (*) marking reconstructed word forms in the figures/tables below. The vowel melody and the accent are coloured red:
ROLAND A. POOTH 18
Remark: This first binyan is reflected by two Vedic and Greek verbal stems belonging to two different aspect categories:
(a) the imperfective “root present” stem, and
(b) the perfective “root aorist” stem. I propose the following developments: The progressive forms of this first binyan (e.g. *gʷʰɛ́nt-i, etc.) once were predominantly used with ongoing present time reference and thus developed into Vulgar Pre-IE present imperfective tense and aspect portmanteau forms. Subsequently, the corresponding nonprogressive forms either developed into corresponding non-present imperfective forms (e.g. Vedic han :: áhan, etc.) or were further narrowed to “root aorists” (e.g. Vedic gán, ágan, etc.). The drift can be illustrated by the following figure:
A class of totally terminative or totally telic roots, e.g. *gʷ_m- ‘come hither, go there’ perhaps generally lacked progressive forms in PIE (†gʷɛ́mt-i). Many different stems (e.g. the one preceding Vedic gáccha-ti ‘go’ etc.) could be used as “stem-suppletive” present imperfective stems in Vulgar Pre-IE. Only later, slightly before Proto-Indo-Iranian and Proto-Greek, but definitively after Proto-Anatolian had left the dialect or variant bundle, the non-imperfective stems were specified to perfective “root aorist” stems and the well-known (neutral-) imperfective vs. anterior-imperfective vs. perfective aspect system was established. By that moment, former PIE progressive forms (e.g. *dʰɛ́ʔti ‘is/was saying, doing’) corresponding to non-progressive forms with a (gradually) terminative or telic meaning (e.g. *dʰɛ́ʔt ‘put, did, said’), as a rule, had to be given up. The reason is simple: By that moment, the Vulgar Pre-IE present imperfective ending *-ti became incompatible to these specific non-imperfective “aorist” stems. One may refer to this rule as the “*dʰɛ́ʔti (> teezzi) dropping rule”:
The Proto-Indo-European aspect system 19
Thus, I would rather not follow the idea that Hittite teezzi ‘says’ was an “innovative backformation”, derived from a former “PIE root aorist” stem. This idea has been labeled the “teezzi principle”. In my view, it can be taken for a Paradebeispiel of anachronistic reprojection of Graeco-Aryan morphosyntactic categories. Instead, I even think that forms like *gʷʰɛ́nt ‘topical referent slew non-topical referent’ once were terminative or telic (and not at all “imperfective-like”) ―,they were only reinterpreted as neutral-imperfective forms, be- cause the corresponding forms in *-ti were used as present neutral- imperfective forms in Vulgar Pre-IE so frequently. 33
Therefore, Hittite teezzi and the corresponding Anatolian forms are clear archaisms and ultimately go back to a PIE 3rd sg. progressive form *dʰɛ́ʔti ‘is/was putting, doing, saying’. The presence of such forms in Hittite and Proto-Anatolian perfectly parallels the abscence of the aorist category in this branch. This ultimately speaks in favour of an innovative nature of the entire aorist category outside Proto-Anatolian. The other PIE five “root formations” or radical binyans are given in the subsequent figures. All forms of the following PIE second binyan (or “acrostatic” “Narten type”) had the word form accent on the vowel in the root vowel slot:
33 Vedic áhim ahan ‘he slew the dragon’, e.g., shows an evident terminative or “telic” meaning (‘... until its death’). The idea that PIE *gʷʰ_n- “must” have had an iterative-like or durative-like lexical aspectual meaning and “must” have meant “wiederholt schlagen” (thus García Ramón 1998), just because this root shows a root present and not a root aorist in Vedic or Greek, is based on the mistaken inference that the IE root presents would reflect an original imperfective-like lexical aspectual meaning, that is, the so-called “Verbalcharakter” of the respective PIE verbal root. However, this is too much a backprojection of Greek and Vedic inflectional aspectual distinctions to the PIE verbal lexicon. Inferring that the imperfective vs. perfective distinction would be “lexically underlying” is, in my view, a severe mistake. Inflectional categories cannot be simply matched 1 to 1 onto a lexicon. Instead, many terminative or telic roots were compatible to the progressive aspect suffix *-i in PIE. Attaching this suffix simply yielded a (derivational-like) durative meaning (like in colloquial Ruhr-German hömma, der is ihn am totschlagen, hilf dem ma bitte ‘listen, he’s beating him to death, please help him’ vs. er hat ihn tot geschlagen, dem kannze nich mehr helfen ‘he has slewn him, you can’t help him anymore’). Therefore, the existence of a root present in IE languages can only tell us that the respective PIE verbal root was compatible to the PIE progressive aspect ― but this does not entail that the root had an imperfective-like meaning.
ROLAND A. POOTH
