You are on page 1of 49

2014-11-17, S. E. & O.

Voice, Direction, Case, and Alignment in Proto-Indo-European


IV. The Proto-Indo-European Case System

ROLAND A. POOTH

This paper is work in progress. It is part (or chapter) IV of a cluster of five papers
with the following subtitles:
I. Introduction and methodological remarks (planned)1
II. The Proto-Indo-European Voice System (cf. Pooth manuscript a)
III. Transitivity Direction in Proto-Indo-European (cf. Pooth manuscript c)
IV. The Proto-Indo-European Case System
V. Proto-Indo-European Alignment: Summary (in preparation)
This part IV provides an internal reconstruction of the Proto-Indo-European case
system.
Keywords: Reconstruction of Proto-Indo-European morphology and syntax; recon-
struction of the PIE case system.

Avant-propos
The above listed cluster of papers is about suggesting a new model for the
voice, transitivity direction,2 case, and alignment system of the PIE proper parent
language. It is explicitly claimed that PIE differed in these respects from all IE
daughter languages, that is, that PIE differed from Hittite, Vedic, Greek, Latin,
Gothic, Tocharian, etc.

1 Introduction
This paper provides a fresh internal reconstruction of the case system of
the Proto-Indo-European parent language. In advance, the following meth-
odological remarks seem to be necessary.
It is crucial for any potentially real and realistic reconstruction of PIE
morphosyntactic categories to recognize that it must not be done via sole
anachronistic backprojection of IE morphosyntactic or grammatical catego-
ries. For instance, claiming that the parent language should have been a
language of the nominative-accusative alignment type, simply “because”
all the given IE languages correspond by displaying this type of core case
distinction is not very reasonable. First, despite any comparative claim,
this claim is not at all a valid and legitimate logical deduction. Second,

1
Cf. Pooth manuscript b: “Ein Problem der Methode der komparativen Rekonstruktion von
Morphemen, Morphemgrenzen und morphosyntaktischen Kategorien”, downloadable
from my profile at www.academia.edu. It can serve as a provisional part I.
2
For the term direction cf. Wolfart & Carroll 1981: 68, DeLancey 1981, Thompson 1989:
21, Klaiman 1992.
ROLAND A. POOTH 2

with regard to the internal evidence that will be given in the following
sections, it cannot be taken for a very plausible inference any longer.
Third, claiming that the reconstruction of morphosyntactic categories
should be based on IE morphosyntactic equations such as Vedic nom. sg.
= Greek nom. sg. = Hittite nom. sg. = Latin nom. sg., etc. is not only a
methodological mistake, it also has yielded an unfortunate methodological
by-product. Within present-day Indo-European studies the use of typologi-
cal implications leading to the reconstruction of typologically different
morphosyntactic categories, different than the ones of the IE languages
seems to be ‘comparatively forbidden’. From a background of diachronic
morphosyntactic typology and studies in grammaticalization, however,
such an entailment simply is, to be clear, dogmatic nonsense. Instead, the
anachonistic backprojection of IE grammatical categories is of little plausi-
bility from the outset. Assuming such a morphosyntactic Graeco-Aryan
‘stability’ is rather implausible, because it contradicts diachronic morpho-
syntactic typological findings. Instead, it is well-known that the opposite is
true. Crosslinguistically, morphosyntactic categories and syntactic con-
structions quickly, easily, and dramatically change, and show an abundant
formal and functional variation.
There is an illustrative example for such a quick change. From the proto-
language Latin to the Romance daughter languages there was a groundbreak-
ing change of morphosyntax, because roughly from around 1000 or 1100 AD
Romance languages show an article + noun phrase, e.g. in the chanson of the
journey of Charles (datable to after 1108 AD) 7 Il la prist par le poin desoz un
olivier, cf. Modern French le ciel, un ciel, Sardic su chelu, etc. However, the
Strasbourg Oaths, which is the earliest extant Romance text (842 AD), does
not contain any example of the article (whereas the teudisca lingua shows one,
cf. in thes cristianes folches = pro cristian poblo and then eid = sagrament).3
... Pro Deo amur et pro cristian poblo et nostro commun saluament, d’isti di en auant, im quant
Deus sauir et podir me dunat, si saluarai eo cist meon fradre Karlo, et in a(d)iudha et in cadhuna
cosa, si cum om per dreit son fradra saluar dift, in o quid il mi altresi fazet, et ab Ludher nul plaid
nunquam prindrai qui, meon uol, cist meon fradre Karle in damno sit.
... In Godes minna ind in thes cristianes folches ind unser bedhero gehaltnissi, fon thesemo
dage frammordes, so fram so mir Got gewizci indi mahd furgibit, so hald ih thesan minan bruodher,
soso man mit rehtu sinan brudher scal, in thiu thaz er mig so sama duo, indi mit Ludheren in
nohheiniu thing ne gegango, the, minan willon, imo ce scadhen uuerdhen.
... Si Lodhuuigs sagrament, que son fradre Karlo iurat, conseruat, et Karlus, meos sendra, de suo
part non los tanit, si io returnar non l’int pois, ne io ne neuls cui eo returnar int pois, in nulla aiudha
contra Lodhuuig nun li iu er.
... Oba Karl then eid, then er sinemo bruodher Ludhuuige gesuor geleistit, indi Ludhuuig, min
herro, then er imo gesuor forbrihchit, ob ih inan es irwenden ne mag, noh ih noh thero nohhein, then
ih es irwenden mag, uuidhar Karle imo ce follusti ne uuirdhit.

It is well-known that there was no article at any stage of Latin, cf. caelum ‘a
sky, the sky, sky’. It was definitively also lacking in spoken Late Latin of the

3
Cf. Tagliavini 1998: 374f., Ledgeway 2012: 96. Note that the use of the definitive article
þe is found in OE in the annals from 1132 onwards, but not before, cf. Traugott 1972:
18-19. An article + noun phrase, therefore, developed parallelly, but relatively inde-
pendently within the West Germanic dialect continuum (cf. the innovated OHG der,
etc.).
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 3

late Roman Empire.4 In the Late Latin spoken variants there were just different
demonstrative constructions, e.g. mediolum illut, de ipso lapide, etc. (cf. Ledge-
way 2012: 91, 93), but there was no grammaticalized article + noun phrase. 5
Therefore, the article + noun phrase obviously emerged roughly between 842
AD and 1000 or 1100 AD as a parallel, but relatively independent innova-
tion of the Romance variants or dialects. An article + noun phrase must not
be backprojected onto Proto-Romance, which should be identified with spoken
Sub-Standard Classical Latin.6
Compared with the Latin to Romance situation, it is of high plausibility,
therefore, that the IE nominative-accusative alignment emerged as a paral-
lel, but relatively independent innovation of all Vulgar Pre-Indo-European
variants or dialects in a period after the PIE protolanguage had broken up
and developed into a Vulgar Pre-IE variant or dialect cluster. Like the Latin
demonstrative + noun phrase (e.g. ipse episcopus, etc.), there must have
been a preceding PIE construction which developed into the nominative +
accusative construction. I have suggested elsewhere that this specific con-
struction was the PIE antipassive construction (cf. Pooth 2004b, 2013a,
for the whole storyline cf. Pooth manuscript a).
Let me be more explicit. If one wants to decide properly what is of high
plausibility and should be reconstructed in the field of proper PIE morpho-
syntax and what is wrongly supposed or less plausible, one must base a
final decision on internal evidence with a background of linguistic typo-
logical implications. ‘Comparative plus internal reconstruction’ bases its
inferences on internal evidence combined with and based on the com-
parative evidence. It is methodologically more advanced than the one that
is based solely on morphosyntactic equations. Staying within the restricted
boundaries of the comparative method does not lead to “safe”7 reconstruc-
tions, but to anachronistic nonsense. Basing morphosyntactic reconstruc-
tion solely on IE morphosyntactic equations and neglecting typological in-
ferences is anachronistic backprojection with little, if any potential truth
value. Therefore, in my view, the PIE case system that has been recon-
structed solely by the comparative method can only have a provisional and
heuristic status. Syntactic patterns such as, e.g., *h2né:(r) *su(H)néu̯ei
*pék̑u(m) *(é-)déh3t ‘man gave domestic animal to son’ do not have any
linguistic status other than being backprojected representations or sym-
bolizations of younger IE and post-PIE syntactic patterns (like the Vedic
ones). Claiming that a pattern like this would represent a potentially
real(istic) syntactic pattern of the PIE proper protolanguage, however, is an

4
Cf. Ledgeway 2012: 91 quoting a “forthcoming” article by Adams.
5
Cf. Weiss 2009: 522, Adams (“forthcoming”) apud Ledgeway 2012.
6
Another illustrating example is the loss of the Latin neuter gender which must not be
dated to the proper Latin parent language, but developed as a parallel but relatively in-
dependent innovation of Vulgar Latin spoken variants or dialects. This is confirmed by
different Romance gender assignments, e.g. Lat. neuter mel, mellis ‘honey’ → French le
miel, fiel, sel ≠ Spanish la miel, hiel, sal = Cat. la mel, fel, sal, Romanian mierea, fierea,
sarea; Lat. neuter lac ‘milk’, sanguis ‘blood’ → Spanish la leche, la sangre, Cat. la llet, la
sang vs. Romanian masculine lapte(-le), sînge(-le), etc.
7
Cf. Yakubovich 2014: 407.
ROLAND A. POOTH 4

idea that is far too naїve―if not simply mistaken anachronistic morphosyn-
tactic reconstruction. The following overview now provides a sketch of
what I think was the case system of the unitary PIE protolanguage.

2 The PIE case forms: Overview


On the basis of the comparative IE evidence and the PIE word-internal
evidence that will be presented more detailed below I draw the inference
that PIE animate nouns displayed a total number of minimally nine or ten
case forms of the singular number―thus even one or two case forms more
than Vedic Sanskrit (which displays a case system with 8 forms, that is,
nom., voc., acc., instr., abl., dat. gen., loc.). Typologically, the PIE case
system is comparable to the case systems of the Pama-Nyungan languages
of Australia which display an average number of 8 to 10 cases, cf. Blake
2001: 158. PIE forms are coloured blue here. The PIE case forms will be
termed and glossed as follows:

FIGURE 4.1. SG
gloss PIE Vulgar Pre-IE
1 absolutive(-vocative) ABS *pχtɛ́r
10 vocative (?) VOC *pχtɛr a *pχ́ter ~ *=pχter
2 ergative-genitive ERG *pχtɛ́rs > *pχtéː(r) (nom.) 8
3 allative-dative ALL *pχtɛ́rm *pχtérm̥ (acc.)
4 locative-allative LOC *pχtɛ́ri *pχtéri
5 sociative-associative SOC/ASS *pχtɛ́rʔ 9 *pχtérʔ (nom.-acc. du.)
6 ablative-genitive ABL *pχtrɛ́s *pχtrés ~ *pχtrós
7 partitive-genitive PAR *pχtrɛ́m b
8 benefactive-purposive BEN *pχtrɛ́i *pχtréi (dat.)
9 comitative-instrumental COM *pχtrɛ́ʔ 10 *pχtréʔ ~ *pχtróʔ

FIGURE 4.2. PL
gloss PIE Vulgar Pre-IE
1 absolutive(-vocative) ABS *pχtɛ́rɛ *pχtére (nom.-acc. du.)
10 vocative (?) VOC *pχtɛrɛ a
2 ergative-genitive ERG *pχtɛ́rɛs *pχtéres (nom.)
3 allative-dative ALL *pχtɛ́rms *pχtérm̥s ~ *pχtérn̥s (acc.)
4 locative-allative LOC *pχtɛ́ris c new loc. pl.e
5 sociative-associative SOC/ASS *pχtɛ́rɛʔ *pχtéreʔ (nom.-acc. du.)
6 ablative-genitive ABL *pχtrɛ́s d *pχtréːs ~ *pχtróːs (gen. du.)
7 partitive-genitive PAR *pχtrɛ́m d *pχtró(ː)m~*pχtróʔom (gen. pl.)
8 benefactive-purposive BEN *pχtrɛ́i d new dat. pl.e
9 comitative-instrumental COM *pχtrɛ́ʔ d new instr. pl.e

8
The Vulgar Pre-IE neo-lengthened grade emerged by a combination of Pedersen’s law
and “Szemerényi’s law”: PIE *_ɛrs → [ɛɹʃ] with realization of */s/ → [ʃ] after */r/ by
Pedersen’s RUKI rule. Then the */r/ of */rs/ [ɹʃ] was dropped and [ʃ] developed to [ɧ]
~ [h] after */ɛ/ in word Auslaut yielding Vulgar Pre-IE *-eː ['ɛː].
9
Probably realized [pχ'tɛrʔɛ̥] with epenthesis of a voiceless echo vowel.
10
Probably realized [pχ'trɛʔɛ̥]; cf. last fn.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 5

Notes to the figures above:


a
The vocative forms possibly had a low tone opposed to a high tone of the ab-
solutive forms. In utterance second or Wackernagel position, the vocative form was
or became enclitic and had no or lost its accent. In utterance initial position the
accent was shifted to the initial syllable. At least the accent shift seems to be a
Vulgar Pre-IE phenomen.
b
The form of case 7 is primarily reconstructed by means of the method of in-
ternal reconstruction via paradigmatic or ‘system-induced backward analogy’. The
inference is drawn as follows. If cases 6, 8, 9 show *_ɛ́_ in the position before *-s,
*-i, *-ʔ, this is retro-analogically transferrable to case 7. Note that the detransitive
nominal stems show *_ɔ́_ here (e.g. partitive-genitive *ʔɗntɔ́m ‘part of tooth, part of
teeth’). This perfectly matches the form that is reconstructable by means of the
comparative method.
d
For traces of *-is cf. Jasanoff 2009: see sections 6-7 below.
c
In PIE, the oblique cases, that is, cases 6, 7, 8, 9 were transnumeral. Thus they
were underspecified as for number distinctions and did not show specific plural
forms. Sporadically and optionally, the vowel could be lengthened to indicate
plural number.
e
These plural forms were Vulgar Pre-IE innovations: see section 7 below.

3 An internal/synchronic PIE morphological analysis


3.1 The PIE ‘case stems’
After separating the vowel melody or transfix (e.g. *_ɛ_) from the skele-
tal consonant frame the way I have argued for elsewhere11 it is easily ob-
servable that on the skeletal consonant frame tier PIE had case marking by
means of a minimum of five semantic role suffixes *-Ø-, *-s-, *-m-, *-i-,
and *-ʔ-. There was thus a minimum of five superordinate consonantal
‘case stems’. Each subordinate PIE case form was based on one of these
underlying stems as its superordinate.
I use the label superordinate ‘case stem suffixes’ or ‘semantic role
suffixes’ to refer to these five markers and suggest that they indicated the
five superordinate semantic roles given in the figure below. I use the cover
symbol *-K- for these five suffixes.

FIGURE 4.3
*-Ø- PLACE, GOAL, THEME, PRODUCT, PATIENT
*-s- AGENT, SOURCE, COMPAREE
*-m- GOAL AS PART OF A BODY OR WHOLE, RECIPIENT/EXPERIENCER
*-i- PLACE, GOAL, EXPERIENCER, BENEFICIARY, PURPOSE
*-ʔ- COMPANION, INSTRUMENT, PATH, MANNER

The superordinate ‘case stems’ can further be labeled as given in the fol-
lowing figure.

11
Cf. Pooth 2004a, 2009b.
ROLAND A. POOTH 6

FIGURE 4.4
superordinate subordinate cases
*pχ-t_r-_Ø zero case stem ABS (VOC), LOCZ, BENZ
*pχ-t_r-_s *-s- case stem ERG, ABL
*pχ-t_r-_m *-m- case stem ALL, PAR
*pχ-t_r-_i *-i- case stem LOCi, BENi
*pχ-t_r-_ʔ *-ʔ- case stem SOC/ASS, COM

These five ‘semantic role suffixes’ were separable and independent mor-
phemes in PIE. It can be inferred, therefore, that superordinate semantic
role marking was separated and independent from the pragmatic-syntactic
distinction between core (and extended core) case marking versus rela-
tional-attributive or “oblique”, that is, say, ‘adjunctive’12 case marking.
The distinction between such ‘nonrelational’13 core (and extended core)
case forms and ‘relational-adjunctive’ ones was provided by the word form
template on the word form template tier.

3.2 The PIE core case vs. “oblique” distinction


Each PIE case form had a particular underlying nominal ‘word form
template’ (WFT) which clearly had the status of a templatic morpheme. It
provided the distinction of the so-called “strong cases” vs. “weak cases” vs.
“weakest cases”. The word form template (WFT) thus ultimately deter-
mined the position of the vowels of the vowel melody or transfix (e.g. *_ɛ_)
combined with the position of the word form accent on one of these vow-
els within the word form. This is illustrated by the following figure.

FIGURE 4.5.
core :: oblique
ABS *χnɛ́r ::
ERG *χnɛ́r-s :: ABL *χnr_ɛ́_s
ALL *χnɛ́r-m :: PAR *χnr_ɛ́_m
LOC *χnɛ́r-i :: BEN *χnr_ɛ́_i
SOC *χnɛ́r-ʔ :: COM *χnr_ɛ́_ʔ
core WFT oblique WFT
*CCɛ́C-K :: *CCC_ɛ́_K

12
I use this term ‘adjunctive’ just to refer to case forms that are like the ones that crosslin-
guistically have the status of adjuncts. Using this term does not imply that PIE made a
distinction between arguments which are bound by verbal valency vs. unbound adjuncts
as known from languages like, e.g., English or German.
13
I use the term ‘nonrelational’ to refer to forms that do not necessarily imply a relation of
the given nominal concept, e.g. [MAN] to another implicit or explicit topical or non-
topical participant in the context. Forms like PIE *χnrɛ́s ‘from the man’ are ‘relational’,
because they imply a relation of the given concepts [MAN] and [SOURCE] to another
implicit or explicit participant involved in the context. On the other hand, forms like PIE
*χnɛ́rs ‘the man (agent)’ do not necessarily imply a relation of the concept [MAN] to
another participant other than the referent (‘man’) itself.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 7

The word form template thus coded the pragmatic-syntactic distinction


between core (and extended core) case marking versus relational-
attributive or “oblique” or ‘adjunctive’ case marking.
But the word form template was a subordinate by itself. It belonged to a
superordinate set of word form templates, that is, the word form template
set or nominal paradigm (e.g., the one of *pχtɛ́r- ‘father’ given above). This
set can be analyzed as a superordinate template to the subordinate word
form templates. Conventionally it has been termed a nominal inflectional
type (nominal IT). Each PIE nominal stem including its nominal finite word
forms thus belonged to one (or more) of the well-established PIE nominal
inflectional types.14 A respective inflectional type either had a basic or un-
derlying lexical status (e.g. *uɔ́ɗr inam. ‘water’ > Hittite waatar n., etc.15)
or conveyed a specific inflectional or derivational meaning (or a meaning
inbetween inflection and derivation, e.g. *uɛ́ɗɔr inam. ‘water(s)’16 > Hittite
witaar, wedaar).
Thus, all PIE inflectional types were characterized by a combination of
different word form templates (WFTs). These templates were related to
each other by what has been labeled “internal derivation” and what I term
‘vowel transposition’.

It is well-established that the PIE parent language intensively made use of


this morphological strategy called “internal derivation”, cf. Widmer 2004: 29
who gives the relevant literature. However, it is important to recognize that
this morphological strategy was not necessarily a derivational strategy in the
sense of lexical derivation, because it did not necessarily derive a lexical entry
from an underlying lexical entry on the lexical level. Instead, it is the major
aim of this paper to establish the idea that PIE word forms (e.g. *χnrɛ́s) could
also be derived from other word forms (e.g. *χnɛ́rs) by the very same morpho-
logical strategy on the inflectional level. The following internal/synchronic
PIE morphological rule can be reconstructed via inductive reasoning, for more
argumentation cf. Pooth 2004a, 2009b:

Nominal stems and nominal forms were derived from other nominal
stems and forms by transferring/transposing the vowel or more than one
vowel to a different vowel slot or to different vowel slots within the
word form on the word form template tier.

This morphological strategy is here referred to by the label ‘vowel trans-


position’ (VTP). It is rather confusing to use the term “morphological me-
tathesis” (cf. Thompson & Thompson 1969). A morphological transposition or
positional change of segments has a fundamentally different motivation. This
strategy may perhaps ultimately go back to prior cases of spontaneous

14
Cf. Tremblay 1999, 2003.
15
The writing of Hittite is simplified here. I use VV (e.g. waatar) instead of the macron
(wātar) to immitate the plene writing and I ommit the diacritics of š, ḫ.
16
This inflectional type was abundantly productive in PIE and could be built from almost
any nominal stem, cf. Widmer 2004: 67: “[...] aus einer beliebigen Flexionsklasse
[wurde] ein unmittelbarer Wechsel in die [*uɛ́ɗɔr-] Klasse vorgenommen.”
ROLAND A. POOTH 8

methathesis. But once such a morphological strategy has been established as


such, the notion of vowel transposition implies an underlying templatic struc-
ture on a word form template tier with various vowel positions or vowel slots
(*_V_) where the respective vowels switch positions not just for spontaneous
phonological “fun”, but for morphological reasons, that is, as an inflectional
and derivational morphological means.
I hereby generously admit that, with regard to the fusional morphotaxis of
the later IE languages, such a phenomenon may look strange at first sight. To
be fair, I admit that this morphological strategy is typologically rare, but it is
definitively not unknown crosslinguistically. A quite similar morphological
vowel transposition can be found in the Oceanic language Rotumam, e.g. piko
(CV1CV2) ‘lazy’ :: piok (CV1V2C) ‘lazy’, rotuma (CV3CV1CV2) ‘Rotuma’ :: rotuam
(CV3CV1V2C) ‘Rotuma’, etc., cf. Besnier 1987: 201-223, Pooth 2004a: 422 fn.,
Thompson & Thompson 1969.
However, of course it is true that we are fully legitimized to reconstruct
linguistic features for PIE that are typologically rare, but are attested as such.
(NB. A phonological distinction between breathy voiced *dɦ, plain voiceless *t,
and (probably) implosive *ɗ (or glottalized) is definitively among the rarities
of the world’s sound systems.)

Nevertheless, it is a profoundly established comparative fact that within


each PIE inflectional type two or three superordinate sets of finite word
forms were distinguished:

FIGURE 4.6.
a. “strong” forms = core (and part of extended core), e.g. *dɦɛ́gɦɔm
b. “weak” forms = extended core: suffixless loc., e.g. *dɦgɦɔ́m
c. “weakest” forms = oblique case forms, e.g. *dɦgɦmɔ́s

FIGURE 4.7.
a. “strong” forms = core and extended core, e.g. *pχtɛ́rs
b. “weak(est)” forms = oblique case forms, e.g. *pχtrɛ́s

From a background of functional linguistic typology it is thus internally


evident that the given distinction between “strong”, that is, core and ex-
tended core case forms vs. “weak” and/or “weakest” oblique case forms
was entirely functionally motivated in PIE. Given that we do now com-
pletely understand the functional distinction that motivated the given for-
mal distinction between “strong” and “weak(est)”, it has to be concluded
that, at least for the period of synchronic PIE, a phonological motivaton is
rather implausible and excluded.
Let me remind the friendly reader that Tremblay 1999, 2003 has bril-
liantly suggested that the distinction between the so-called “proteroki-
netic” and the so-called “hysterokinetic” type was ultimately motivated by
the functional distinction between ‘nonrelational’ forms vs. ‘relational’
including attributive and possessive forms.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 9

Tremblay 1999, 2003: 248f. has cogently demonstrated that this very
functional distinction must have been synchronically productive within
Avestan. Examples are:

(4.1) sāuuaŋhə̄e dat. sg. ‘for the Sāuuaŋhi’ (Y. 1.3): noun
::
sāuuaŋhaēm acc. sg. ‘the sāuuaŋhi-ish’ (Y. 2.3): adjective

(4.2) frādat̰.fšauue dat. sg. ‘for the Frādat̰.fšu’ (Y. 1.4):


noun, cf. acc. sg. pasūm, Gathic gen. sg. pasə̄uš
::
frādat̰.fšaom acc. sg. ‘the frādat̰.fšu-ish’ (Y. 2.4): adjective

Let me add the following illustrative example from Early Vedic:


(4.3) dā́ru ‘wood’: neuter noun
::
dārúm acc. sg. (RV 7.6.1d): adjective → dāruṇám ~ dā́ruṇam

Remark: I can understand that an inflectional and derivational distinction


‘nonrelational’ vs. ‘relational’ is not easy to understand or to accept with a sole
IE background. Tremblay has thus received a criticism that is inappropriate
and unjustified. In the very same volume Pinault 2003: 150 has replied to Xa-
vier as follows: “Il est ruineux de décrire ces oppositions en termes métaphy-
siques (cf. Tremblay, 1996, p. 103 sq.) comme celle entre substance et acci-
dent ou attribut”. But from a general linguistic perspective and with a typo-
logical background it is exactly valuations like this that make Indo-European
studies look so “extravagant” and solipsistic nowadays.
Let me give an illustrative example for a functional distinction of this type
that Xavier had in mind and others may not be aware of:17
Classical Nahuatl ultimately confirms that some languages mark deviations
from the notion of prototypical referentiality18 by means of a marker or by the
lack of a marker (e.g. Classical Nahuatl -in), cf. tōtol-in ‘turkey hen’ vs. -tōtol-
(without -in) in tōtol-tetl ‘turkey egg, turkey-hen-ish egg’ and tē-tōtol ‘someone’s
turkey hen’ parallel to tē-mich ‘someone’s fish’ (cf. Andrews 2003: 109ff.;
280ff.).
Classical Nahuatl thus uses a NONRELATIONAL form tōtol-in ‘(it is a) turkey
hen’ (which is called “absolute state” by Andrews 2003) to indicate its ‘nonre-
lationality’ or, say, its ‘mono-referentiality’, that is, to indicate a lack of rela-
tion of the given nominal concept [TURKEY HEN] to a different implicit or
explicit participant in the context other than the one that is identifyable as
‘turkey hen’ itself. On the other hand, it uses a RELATIONAL (or “possessive
state”) stem -tōtol- to indicate a relation of the given concept [TURKEY HEN]
to another participant involved in the context which is not identifyable as
‘turkey hen’, but as its possessor. This relational stem is further used to code
17
I am endebted to Xavier for our fruitful discussions on this topic. I know that he would
have liked this typological parallel. The useful term RELATIONAL-ATTRIBUTIVE belongs to a
proper terminology of functional linguistics like all other pragmatical terms, e.g. topical
referent, nontopical referent, referential, etc.
18
Prototypical referentiality is identifyable with ‘nonrelational’ reference, see above fn.
12.
ROLAND A. POOTH 10

the attributive and modifying function of the concept [TURKEY HEN] as a first
member of a compound, e.g. tōtol-tetl ‘turkey egg, turkey-hen-ish egg’ (thus
‘egg pertaining to turkey-hen’ or ‘egg possessing turkey-hen-ishness’).
Therefore, Tremblay’s functional distinction “relativisch” (vs. nonrela-
tional) finds a typological parallel and must be taken serious. Unfortunately,
he died far too young to receive an appropriate understanding.
Widmer 2013: 189, fn. 2 has added the following criticism: “Tremblays
Unterscheidung [...] ist nur schwer nachzuvollziehen”. Widmer ibid. p. 30ff.
himself proposes an overall function of coding “possession” to motivate inter-
nal derivation, e.g. *sɛ́ʔmn inan. ‘seed’ → *sɛ́ʔmɔn- ‘possessing seed, having
seed; pertaining to seed; seedy; god of seed’. But it is noteworthy that Widmer
ibid. p. 31 and p. 73 subsumes the ATTRIBUTIVE function under the possessive
notion, e.g. (PIE *nkú- →) *n̥ku̯ó- (> Toch. A oṅk, B eṅkwe ‘man’) “Tod ha-
bend; sterblich” (Widmer 2004: 73; cf. PIE *nVk- ‘perish, die’, Rix et al. 2001,
s.v. *nek̑-). Widmer’s concept of “possession” is based on the following defini-
tion given by Seiler 1983: 4 (emphasis mine):
“Linguistic POSSESSION consists of the representation of a relationship be-
tween a [concept] and another [concept] [...]” (note that Seiler uses “substance”,
where I prefer using “concept”).
Thus, it finally turns out that Widmer’s definition of “possessive” is identi-
cal to Tremblay’s definition of “relativisch” (cf. Tremblay 2003: 253: “mit der
Rettung verbunden”). Tremblay and Widmer only differ with regard to the
terminology that they have chosen to refer to the linguistic encoding of
RELATIONALITY. It is “nur schwer nachzuvollziehen” why Widmer has written
that Tremblay’s functional distinction “ist nur schwer nachzuvollziehen”, be-
cause it is conceptionally identical to the one he proposes. On the other hand, of
course, it is true that Widmer and Tremblay very much differ with regard to
what they think should be viewed as the underlying derivational base, both on
the formal as well as on the functional level. Tremblay 2003 has made the
brilliant claim that PIE had an elaborate system of root-and-pattern morphol-
ogy including an underlying derivational base on the skeletal consonant frame
(e.g. *ɗ_r-u- ‘tree, wood’, *p_k-u- ‘domestic animal, ...’) and multiple inflec-
tional types or binyans of the Semitic morphological type (Tremblay called it
“semitico more”). Pinault and Widmer, however, still follow Schindler’s unsat-
isfactory idea of a “derivational chain”, “chaîne de derivation”, “Deriva-
tionskette”. Such a view on “internal derivation”, however, has always been a
more restricted perspective, simply because in Schindler’s conception the
strategy of “internal derivation” is restricted to the lexical level. Followers of
this restriction may thus probably simply not reckon with the ultimate formal
mechanisms and functional motivation of PIE vowel transposition:
Tremblay’s brilliant idea is transferrable to the very inside of each PIE
nominal paradigm. It is almost self-speaking that the PIE “oblique” case
forms had a relational-attributive (“relativische”) function. They were used
to modify a given nominal concept and thus to ATTRIBUTIVELY relate the
given nominal concept (e.g. *χnrɛ́s [MAN.SOURCE]) to the given topical
referent (TOP) or to a preceding or following second referent (REF 2) (e.g.
*uɔ́ːkw- [SPEECH]). Syntactically, therefore, they had the status of a modi-
fying and attributive adjunct:
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 11

(4.4) *kuɔ́n-s *uɔ́ːkw *klnɛ́ut *χnrɛ́s


dog-ERG speech.ABS man:ABL
TOP REF 2 verb modifying REF 2, adjunct
literarily ‘dog, to (the) speech, it was listening, the one from man’

(4.5) *kuɔ́n-s *uɔ́ːkw *klnɛ́ut *χnrɛ́m


dog-ERG speech.ABS man:PAR
TOP REF 2 verb modifying adjunct
lit. ‘dog, to speech, it was listening, the one part of (the body of) man’

(4.6) *kuɔ́n-s *ɠwɛ́mt *kluɛ́i


dog-ERG came/went listen:BEN
TOP verb adjunct
literarily ‘dog came/went (hither) to listen (to it)’

(4.7) *kuɔ́n-s *páχt *χnrɛ́i


dog-ERG protected man:BEN
TOP verb adjunct
literarily ‘dog protected it (drove away harm) for the benefit of man’

(4.8) *kuɔ́n-s *ʔɛ́it *χnrɛ́ʔ


dog-ERG went/moved man:COM
TOP verb adjunct
literarily ‘dog moved, accompanied by man’

It can be concluded that the PIE “oblique” RELATIONAL case forms were
word-internally inflected for their “oblique” RELATIONAL function on the
word form template tier.
NB. Let me repeat that such distinctions are not unknown crosslinguisti-
cally. Another example is provided by Ngangela, a southwest Bantu language.
According to Maniacky 2002, Ngangela distinguishes between two case
forms by means of two tones. A case form without initial hight tone is used for
“subjects” (thus König 2008: 218) and for extraposed topics, while the other
one, coded by word-initial high tone (glossed H) is used as a citation form, for
“objects”, and nominal predicates, cf. Maniacky 2002: 48, 68. Maniacky’s ex-
amples are quoted from König 2008: 218 as given below. The English transla-
tions have been provided by her.

(4.9) a. káðila
bird.CLASS7.H (citation form ‘bird’)
b. vakoŋgo va-muʃovo wéetú va-lí na-nðíli
hunter GEN-clan our be strength.H
“The hunters of our clan are strong” (König 2008: 281, Maniacky
2002: 48).
c. kaðila wéekupulúla
bird fly
“The bird flies’ (König 2008: 28, Maniacky 2002: 68).
ROLAND A. POOTH 12

d. muuntu alikutala múcí wa-káamá ayoŋgolá kuutééta


person look.at tree.H big will fell
“The person who is looking at the big tree is going to fell it”
(König 2008: 281)

A case system with one case coding “subjects” and extraposed topics and
another one coding “objects” and nominal predicates is ultimately motivated
by pragmatical distinctions. Ngangela exhibits one case category (kaðila ‘bird’)
for TOPICAL-REFERENTIAL (“subject” and extraposed topics) and another (káðila)
for both NONTOPICAL REFERENTIAL (that is, “oblique-like”) PLUS NONREFERENTIAL-
PREDICATIVE (‘it is (a) bird’) including citation. Such a functional distinction is
a bit similar and definitively comparable to the one proposed for PIE here.
Tremblay’s functional motivation for the distinction between “pro-
terokinetic” vs. “hysterokinetic” stems is thus extendable. The very same
functional motivation is additionally capable of providing a plausible ex-
planation for the PIE distinction between “strong” and “weak(est)” forms.
Let me thus reformulate his finding. In honour of Xavier Tremblay I would
like to term this PIE morphological rule ‘Tremblay’s Rule’:

RELATIONAL-ATTRIBUTIVE nominal stems and oblique or ‘adjunctive’ case


forms were both marked alike for their common relational-attributive or
‘adjunctive’ function by a particular underlying word form template on
the word form template tier.

As already mentioned, this rule is the ultimate reason for the distinction
between, e.g. PIE *pɛ́ku- ‘domestic animal (nonrelational)’ vs. *pkɛ́u- ~
*pkú- ‘[DOMESTIC ANIMAL] in relation to a 2nd participant: someone’s
domestic animal, domestic animal belonging to s.o.’ and ‘s.o. having (a)
domestic animal’ and ‘the concept [DOMESTIC ANIMAL] attributively re-
lated to another participant in the context: s.o. domestic-animal-ish’. I sug-
gest that this PIE synchronic polysemy functioned as follows: (α) It was
either the given referent ‘domestic animal’ that was referred to as being
possessed (possessee) by another implicit or explicit possessor ‘X’―or (β) it
was the concept [DOMESTIC ANIMAL] that was more or less metaphori-
cally possessed by ‘someone’ or the given referent ‘X’:

FIGURE 4.8.
polysemeous:
*pkɛ́u ~ pkú [DOMESTIC ANIMAL]as referent POSSESSED BY ‘X’
‘X’ or given referent HAVING [DOMESTIC ANIMAL]
‘X’ or given referent BEING [DOMESTIC ANIMAL]-ISH19

19
Many noun-like adjectives (e.g. *bɦɛbɦrú ‘beaver-ish brown’) thus belonged to the “hys-
terokinetic” inflectional type due to its relational-attributive function.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 13

Tremblay’s rule thus implies that the notion of NONRELATIONAL versus


RELATIONAL INCLUDING NONREFERENTIAL-ATTRIBUTIVE was a productive inflec-
tional and derivational nominal category of the PIE parent language. This
functional distinction can be illustrated by the following figure:

FIGURE 4.9
case stem NONRELATIONAL RELATIONAL RELATIONAL/
SECONDARY OBLIQUE
*pk_u-_Ø *pɛ́ku *pk_ɛ́_u ~ *pkú
*pk_u-_s *pɛ́ku-s *pk_ɛ́_u-s ~ *pkú-s *pku_ɛ́_s
*pk_u-_m *pɛ́ku-m *pk_ɛ́_u-m ~ *pkú-m *pku_ɛ́_m
*pk_u-_i *pɛ́ku-i *pk_ɛ́_u-i ~ *pkú-i *pku_ɛ́_i
*pk_u-_ʔ *pɛ́ku-ʔ *pk_ɛ́_u-ʔ ~ *pkú-ʔ *pku_ɛ́_ʔ

Last but not least, it is an relevant typological implication of Xavier’s find-


ings that PIE must have had a class of RELATIONAL-POSSESSIVE (plus ATTRIBUTIVE-
ADJUNCTIVE) nominal stems which were derived from nonrelational ones by
“internal derivation” (‘vowel transposition’):

FIGURE 4.10.1
relational-possessive stem
*krɗí ‘someone’s heart, jumper’ a
w w
*ʕ k í ‘someone’s eye, seeer’ a
*ɢɦɛbɦχál ‘someone’s head, roundish holder’ (cf. ɢɦVbɦχ- ‘grab, hold’)
ɦ
*ɗng ú ‘someone’s tongue’
*smɔkrú ‘someone’s beard’ (derived from *smɔ́kur)
*tɛnúχ ‘someone’s body-collective’
*pχtɛ́r ‘someone’s father, protector’
*ɗʕwtɛ́r ‘something’s giver to someone (within a specific event)’ b
*pɔtí ~ *pɔtɛ́i ‘s.o.’s master’ (e.g. PIE *ɗɛ́ms *pɔtɛ́i- ‘the family’s master’ c)

FIGURE 4.10.2
nonrelational stem
*kɛ́rɗ ‘heart, jumper’ a
w w
*ʕ ák ‘eye, seeer’
*ɢɦɛ́bɦχl ‘roundish holder’
*ɗɛ́ngɦu ‘tongue, the one whose function (*-u-) is to *ɗVngɦ-’
*smɔ́kur ‘beard-hair-ish mass’
*ɗáʕwtɔr ‘always giving (habitual aspect)’ b
*pɔ́t ~ *pɔ́ti ‘master’: “acrostatic” *pɔ́ti- abl. *pɛ́tiɔs c

Notes to the figures above:


a
I suggest that the relational nominal stem was derived from an underlying i-
stem. An i-stem could be derived from almost any root noun in PIE.
b
For a distinction between “situative” (= relational) vs. habitual nomina agen-
tis cf. Tichy 1995.
c
The original distinction was given up in the new Vulgar Pre-IE core cases:
thus nom. sg. *pótis, acc. sg. *pótim. But Vedic and Avestan preserve a distinction
ROLAND A. POOTH 14

of two inflectional patterns in the oblique forms, cf. Tremblay 2003: 246 who
gives the older literature. Although a distinction between nonpossessive or
NONRELATIONAL nominal stems versus relational (relational-possessive) ones is
rather absent from the later IE languages, it is not an uncommon crosslinguistic
functional distinction, cf. the situation in Classical Nahuatl given above. But let me
return to this nominal distinction more detailed elsewhere.
Summarizing the given observations, it can be concluded that the case
system of the PIE proper parent language included a functionally ‘hidden’
formal distinction between core and extended core cases versus relational-
attributive or ‘adjunctive’ cases. This syntactically-and-pragmatically moti-
vated distinction between ‘core referential’ marking versus ‘non-core
oblique-adjunctive’ marking was morphologically coded separately from
the encoding of semantic roles.
Therefore, it is finally possible to re-label the given PIE case forms with
regard to that distinction as follows. Note that I leave away the vocative:

FIGURE 4.11.
gloss PIE new glosses
1 absolutive(-vocative) ABS *pɛ́ku NONRELATIONAL.UNMARKED
2 ergative-genitive ERG *pɛ́ku-s NONRELATIONAL-SOURCE

3 allative-dative ALL *pɛ́ku-m NONRELATIONAL-ANIMATE.GOAL


4 locative-allative LOC *pɛ́ku-i NONRELATIONAL-PLACE/GOAL
5 sociative-associative SOC/ASS *pɛ́ku-ʔ NONRELATIONAL-COMPANION

6 ablative-genitive ABL *pkɛ́u-s RELATIONAL-SOURCE


7 partitive-genitive PAR *pkɛ́u-m RELATIONAL-GOAL/PART.OF.WHOLE
8 benefactive-purposive BEN *pkɛ́u-i RELATIONAL-GOAL/PURPOSE
9 comitative-instrumental COM *pkɛ́u-ʔ RELATIONAL-COMPANION

It can finally be concluded that the oblique cases 6, 7, 8, and 9 were


“internally derived” from the core and extended core cases 2, 3, 4, and 5.
Furthermore, is obvious that suffixless locative (*dɦgɦɔ́m) was “internally
derived” from the absolutive case form (*dɦɛ́gɦɔm).

4 Excursus: Understanding PIE vowel transposition


The following rules of “PIE internal derivation” are reconstructed here.
These rules are dated here to a synchronic spoken level of the unitary PIE
proper parent language.

4.1 The transposition of *ɛ


It is easily observable that the so-called “weak” and “weakest” case forms
such as the ABLATIVE-GENITIVE form were internally derived from the corre-
sponding strong forms by a transposition of the vowel of the vowel melody to
a different position within the word form:
(a) VTP in nominal forms: *pɛ́kus anim. ‘domestic animal’ → abl. *pkɛ́us,
*ʔsɛ́nts ‘existing, real, true’ → abl. *ʔsntɛ́s, *gwɦɛ́ns (→ Vulgar Pre-IE *gwɦé:(n) by
analogy) ‘slayer’ → abl. *gwɦnɛ́s.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 15

(b) Verbal agentive plural forms, for instance, were derived from singular
forms by VTP: 1st sg. agt. *gwɦɛ́nm → 1st pl. excl. agt. *gwɦnmɛ́(s), 2nd sg. agt.
dir. *gwɦɛ́nt → pl. *gwɦntɛ́(n), 2nd sg. agt. itr. *gwɦɛ́n → pl. *gwɦnɛ́(n).
The vowel *_ɛ́_ was transposed to a different position within the word form,
either to the vowel slot before the stem suffix (*_VS) or to the one before the
case suffix (*_VK) (NB. K = case suffix).
For instance, *pɛ́ku (PIE animate → IE masculine) ‘domestic animal’ con-
tained an underlying vowel melody template *_V_́ and an underlying word
form template *CVC ́ S-, where *C is for any radical (that is, root) consonant, *S
is for the stem suffix. This word form template assigned the position of the
vowel to a specific vowel slot within the word form. Therefore, the internal
derivation has to be seen as a phenomenon that took place on the word form
template tier:

(4.10) a. *CVC ́ u-K → *CCVu ́ -K


*pɛ́ku-s → *pkɛ́u-s
b. *CCVn ́ t-K → *CCntV_́ K
*ʔsɛ́nt-s → *ʔsntɛ́_s

4.2 The transpositions of *ɔ


It can easily be recognized that the same system of vowel transposition was
applicable to *_ɔ_. The PIE vowels *ɛ and *a were functionally underspecified
or ‘bare’ vowels (and were underlying root vowels before VTP). The discon-
tinuous detranstive marker *ɔ was mapped upon the respective ‘bare’ vowel on
the vowel melody tier yielding detransitive forms (e.g. *uɔ́ɗr, *uɛ́ɗɔr, etc.). Due
to the nature of *ɔ as a specific detransitive morpheme, a combination with
*_ɔ_ remained impossible in agentive forms. Thus agentive forms were not al-
lowed to display *_ɔ_ for functional reasons. There were the following types of
vowel transpositions of underlying *_ɔ́_, *_ɔ_, and *_ɛ́_ɔ_.

4.2.1 The total transposition


According to this rule, the accented *ɔ́ was transposed totally, parallel to
the accented *ɛ́ of the agentive forms:

(4.11) *CCVn ́ tK → *CCntVḰ


*ʔdɔ́nts ‘tooth’ → *ʔɗntɔ́s

4.2.2 The ‘broken’ transposition


There was second, more intricate type of vowel transposition which may be
termed ‘broken’ vowel transposition.
Due to the discontinuous nature of the detransitive morpheme *ɔ, it is pos-
sible to analyze a segmental string with accented vowel */Cɔ́C-/ as containing
two vowels in its underlying segmental string structure: one ‘bare’ vowel *ɛ (or
*a) plus the discontinuous one, both being mapped upon the very same under-
lying vowel slot (_V1_):
ROLAND A. POOTH 16

(4.12) */d ɔɛ́ ru/ realized as *ɗɔ́ru ‘wood’



word form template: *CVC ́ u
underlying word form: *ɗáru ~ *ɗɛ́ru ‘tree’

Under this perspective it turns out to be understandable why the ‘bare’


vowel could be transposed to a different position within the word form while
the discontinous marker remained in its underlying position:

(4.13) *CɔVC ́ u- → *CɔCuV_́ K


*ɗɔ́ru → *ɗɔruɛ́s 20
*liɔ́kwr ‘animal liver, roe’ → loc. *liɔkwɛ́n, abl.-gen. *liɔkwɛ́ns

As opposed to forms like, e.g., *liɛ́ːkwr ‘human liver’, *liɛkwɛ́n, *liɛkwnɛ́s (>
Vedic yaknás) the “weak(est)” forms of this type did not belong to the
PLURATIVE morphological category which was coded by an underlying bivo-
calic vowel melody template *_V1_V2_. Instead, they belonged to the
SINGULATIVE morphological category which was normally coded with a mono-
vocalic template *_V_. The *ɔ within the root of the “weak(est)” forms (e.g
*CɔC_V́_) must have had the morphological status of a remnant of the basic
underlying form with an underlying monovocalic vowel melody template *_V_.
This may sound a bit complicated, but I think it can work. For the singulative
vs. plurative distinction see below.

4.2.3 The ‘reversed broken’ transposition


The relation between *_ɔ́_ and *_ɔ_ɛ́_ also showed up in a reversed way with
underlying word forms displaying an accented vowel _ɛ́_ in a position outside
the root vowel slot and an unaccented *_ɔ_ in a vowel slot to its left or the root
vowel slot. Here, the unaccented *_ɔ_ was transposed to the same vowel slot of
*_ɛ́_ and was thus mapped upon it yielding *_ɔ́_.

(4.14) */CɔCCnVu ́ / → *CCCnɔV́ u-


*/sɔuHnɛ́u/ ~ *sɔuHnú → *suHnɔ́us

It is inferrable, then, that the PIE word for ‘s.o.’s son’ had *ɔ throughout its
nominal paradigm. Gothic gen. sg. m. sunaus points to a preceding Vulgar Pre-
IE *sunóus with o-grade of the suffix. PIE *sɔuHnú-, *suHnɔ́us was first changed
to *sounú-, *suHnóus due to the post-PIE sound law called “Saussure’s effect”,
that is, PIE *CɔRH.CV_ > Vulgar Pre-IE *CoR.CV- with loss of *ʔ, *χ, *ʕw in
that position. Subsequently, the loss was paradigmatically transferred to the
oblique forms in some (but not all) Vulgar Pre-IE dialects. After that, the zero
20
Cf. Homeric Greek gen. sg. n. δουρός :: Hitt. gen. sg. n. taaruwas. There may have been
an “acrostatic” allomorph *ɗɛ́ruɔs. Toch. B genetive-adj. oraṣṣe ‘to wood, made from
wood, wooden’ (e.g. 194b1) reflects Proto-Tocharian *orə́sjyæ- (*doru̯ési ̯o-). This secon-
dary o-stem was derived from the old abl.-gen. comparable to δουρός :: taaruwas. The
equation strengthens the view that the abl.-gen. *ɗrɛ́us (> Vedic dróṣ) once did not be-
long to *ɗɔ́ru-, but rather belonged to an animate stem *ɗáru ~ *ɗɛ́ru ‘tree’, whereas the
abl.-gen. *ɗɔruɛ́s (or *ɗɛ́ruɔs) belonged to *ɗɔ́ru (inanimate) ‘wood’.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 17

grade of the root was paradigmatically leveled―either as Vulgar Pre-IE


*suHnú- or *sunú- from either *suHnóus- or *sunóus. This is probably a plausi-
ble solution for the problem that the IE languages reflect both *suHnú- (>
Vedic sūnú-) and *sunú- (> Gothic nom. sg. m. sunus).
As shown by Tremblay 2003: 246 §3.2.1., Vedic and Avestan point to a dis-
tinction between two PIE stems:
(a) *pɔ́ti- ‘master’ NONRELATIONAL, abl. *pɛ́tiɔs (“acrostatic”) vs. (b) *pɔtí-
‘s.o.’s master’ RELATIONAL, abl. *pɛtɔ́is (> Gathic Avestan patōiš, etc.). I suggest
that the latter was of the same inflectional type as PIE *sɔuHnú-, abl. *suHnɔ́us.
I call this inflectional type “mesostatic with mobile ɔ”.

4.2.4 The ‘static broken’ transposition


There was another type of ‘broken’ vowel transposition of *_ɔ́_. Here, the
accented ‘bare’ vowel *_ɛ́_ remained in its underlying position and the discon-
tinuous *ɔ was transposed to another position within the word form. This kind
of vowel transposition may be termed the ‘static broken’ vowel transposition:

(4.13) *Cɔɛ́Cu → *Cɛ́CuɔK


*ɠɔ́nu ‘knot, joint, angle; knee’ → *ɠɛ́nuɔs

These “weak” and “weakest” forms had an underlying vowel melody tem-
plate *_V_V_ and were identical to adjectival nominal stems with a stative
meaning, e.g. *ɠɛ́nuɔ_ ‘knotty’, cf. Ancient Greek nom.-acc. sg. n. γόνυ :: Hittite
gen. sg. n. geenuwas (ge-nu-wa-aš OS, ge-e-nu-wa-aš OH/NS, cf. Kloekhorst
2008: 467f.). PIE *ɠɛ́nu ‘human joint, chin, knee’, *ɠnɛ́us was the correspond-
ing stem without ɔ, PIE *ɠnɛ́u ~ *ɠnú was the relational stem.
NB. Cf. Kloekhorst 2008, s.v.: “[W]ith these three ablaut grades attested in
the IE languages, it is difficult to reconstruct a PIE paradigm”. But this diffi-
culty is only met with within a monoparadigmatic approach. The IE gender
confusion Gk. γόνυ n., etc., but γένῡς f. ‘chin’, and the parallel Hittite gender
confusion highly speaks in favour of a merger of minimally two paradigms, if
not to say more than two (cf. Avestan acc. sg. žnūm (V 8,61-63; 9,22ff.) point-
ing to a “hysterokinetic” stem *ɠnɛ́u ~ *ɠnú, cf. Tremblay 2004: 577). Thus,
Kloekhorst’s view “(n. > c.)” of Hittite acc. sg. c. ge-nu-un can be seen as a
mistake. Instead, this form directly reflects PIE *ɠɛ́num or *ɠnúm (or even both
forms).
I further suggest that the collective-generic *uɛ́ɗɔr type was internally de-
rived via static broken vowel transposition:

(4.14) *Cɔɛ́CS → *Cɛ́CɔS


*uɔ́ɗr ‘water’ → *uɛ́ɗɔr ‘water(s)’

4.2.4 The ‘reversed static broken’ transposition


Again, the relation between *_ɔ́_ and * _ɛ́_ɔ_ also showed up in a reversed
way. With underlying *_ɛ́_ɔ_, the *_ɛ́_ was transposed to the position of *_ɔ_
yielding *_ɔ́_ with accent. This type of vowel transposition is evident from the
“holokinetic” inflectional type, e.g.
ROLAND A. POOTH 18

(4.15) *Cɛ́CɔS → *CCɔ́S


*uɛ́ɗɔr → loc. *uɗɔ́n (>Vedic udán)
*dɦɛ́gɦɔm → loc. *dɦgɦɔ́m (>Hittite tagaan)

NB. In the PIE verb system, there was a transposition of underlying * _ɛ́_ɔ_
which can be termed ‘cross vowel transposition’. Here, the two vowels simply
switched positions:
(a) Subtype with the accent remaining on the underlying position: *uɛ́iɗɔ(i)
‘recognizes; can be recognized; is visible’ (interminative-durative aspect) →
*uɔ́iɗɛ(i) ‘knows’ (stative-habitual aspect);
(b) Subtype with additional accent shift: *uɛ́iɗɔ(i) ‘recognizes; can be rec-
ognized; is visible’ (interminative-durative aspect) → *uɔiɗɛ́(i) ‘makes s.o. be
recognized, seen’ (distributive-or-factitive-or-causative aspect).

4.3 The ‘plurative broken’ transpositions


A completely parallel broken vowel transposition was applicable to the
agentive and detransitive PLURATIVE word forms, that is, the ones with basic or
underlying geminate vowel slot *_VV_ (realized as *_Vː_) in the root. Here, the
vowel geminate *_ɛ́ɛ_ or the vowel sequence *_ɔ́ɛ_ (both realized as long vow-
els) were broken up into combinations of two short vowels.

4.3.1 The one of underlying *_ɛ́ɛ_


The ‘agentive plurative broken’ transposition was found, for instance, in the
PIE paradigm for ‘human liver’ which was “holokinetic” in my view:

(4.16) *Cɛ́ɛCS → *CɛCɛ́S


*liɛ́ːkwr → loc. *liɛkwɛ́n (→ abl.-gen. *liɛkwnɛ́s > Vedic yaknás)

4.3.2 The one of underlying *_ɔ́ɛ_


The ‘detransitive plurative broken’ transposition was found, for instance, in
the PIE paradigm of *uɔ́ːkw ‘speech, formula, words’ which was also “holoki-
netic” in my view:

(4.17) *Cɔ́ɛCS → *CɛCɔ́S


*uɔ́ːkw → abl.-gen. *uɛkwɔ́s (>Young Avestan gen. sg. vaco)

NB. Cf. Avestan nom. (and Young Avestan gen.) m. vāxš, < Proto-Indo-
Iranian nom. sg. m./f. *u̯ā́kš, gen. sg. *u̯ā́kš, gen.-abl. sg. *u̯acás.

4.4 The derivation of pluratives from singulatives


Therefore, the addition of an unaccented second vowel slot *_V2_ to a
monovocalic basic underlying vowel melody *_V_ can be considered as internal
derivation of PLURATIVE nominal stems and word forms from SINGULATIVE ones.
Thus, on the word form template morpheme tier, plurative word forms were
internally derived from singulative word forms by the addition of an unac-
cented vowel slot *_V_ to the underlying vowel melody template: *_V_ →
*_V1V2_ or *_V1_..._V2_. Examples are:
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 19

(a) nominal forms: *pɛ́kus (animate) ‘single domestic animal’ → *pɛ́ːku (in-
animate) ‘cattle’; *uɔ́kw (animate) ‘single word’ → *uɔ́ːkw (inanimate → IE
feminine) ‘speech, formula, group of words’; probably *χɔ́ui (inanimate) ‘sin-
gle egg’ → *χɔ́ːui (inanimate) ‘group of eggs (in a nest)’ (abl.-gen. *χauiɔ́s →
Vulgar Pre-IE *χau̯i̯ó- ~ *χōu̯i̯ó- ‘egg’).
(b) verbal forms: *stɛ́ut (semelfactive-terminative-telic aspect) → *stɛ́ːut (in-
terminative-durative aspect), etc.

4.5 The derivation of detransitive from agentive-unmarked


It is an implication of the given view on PIE morphology that PIE had de-
transitive or “protomiddle” nominal stems. This is to say that PIE nominal
stems were marked for what may be termed “nominal voice” or nominal dis-
tinctions of the domain of prototypical agency (and prototypical transitivity). The
verbal 3rd person and nominal detransitive morpheme *ɔ was a discontinuous
morpheme that was mapped upon the ‘bare’ vowel of the underlying vowel
melody template (see example 4.12 above). Examples are:
(a) nominal forms: *pɛ́ku- (animate) ‘single domestic animal’ → *pɔ́ku- (in-
animate) ‘product of domestic animal, young cattle’ (cf. Myc. po-ku-ta Kn C4
“Freier, der nur Kleinvieh besitzt” (Tremblay 2004: 575)’, Old Armenian as-
‘fleece’, nom.-acc. sg. asr, nom. pl. ask‘ vs. Vedic paśú- m., etc.; the meaning
‘Jungvieh’ is close to ‘product of domestic animal’); *uɛ́kw- (animate) ‘speaker,
act(ion) of speaking’ → *uɔ́kw- (inanimate) ‘the result or product of speaking,
word’; *χáui- (animate) ‘bird’ → *χɔ́ui- (inanimate) ‘product of bird, egg’ (see
below section 5.1.), etc.
A sample of more examples:
(1) *ʔɔ́su, *ʔɛ́suɔs inan. ‘good thing, realium’ vs. *ʔɛ́sus, *ʔsɛ́us ‘existence, life, life
force, strongness, vitality’ [+animate natural force], cf. Hittite aassu n. vs. Vedic ásu-,
ásoṣ* m.;
(2) *mɔ́dhu inan. ‘product of honey, alcoholic honey, mead’ :: *mɛ́dhu inan. ‘honey’,
cf. Toch. B mot “boisson alcoolisée”. The meaning ‘product of honey’ has already been
its PIE meaning (pace Pinault 2004: 183, who thinks that *mɔ́dhu would have to mean
“sweetness”. This mistaken preconception has lead to his opinion that Toch. B mot
“must” be a loan from Middle-Iranian, particularly from Sogdian mwδy, acc. sg. mwδ
< *madu “boisson alcoolique” (which is possible, but not compulsory);
(3) *pɔ́χiu inan. ‘cattle: the protected undergoer’ :: *páχius, *pχiɛ́us anim. ‘herds-
man: the protecting agent’, cf. Greek πῶυ n. ‘cattle’ vs. Vedic abl.-gen. sg. pāyóṣ (RV
3.15.4) ‘herdsman’ m. (cf. Tremblay 2004: 578);
(4) *χɔ́iu, *χáiuɔs inan. ‘time, age, Zeitalter, eternity (as independent from animate
life)’ :: *χáiu, *χiɛ́us ‘[+human] or [+animate] lifespan, age of vital force, age of
animate vigor’, cf. Gothic aiws ‘time, eternity’, Avestan gen. sg. yaoš ‘lifespan (refer-
ring to animate lifetime)’ (cf. Tremblay 2004: 577);
(5) *ɢɦɛ́bɦχl *ɢɦbɦχáls (“proterokinetic”) inan. (→ Vulgar Pre-IE *ɢɦebɦχlā f.) ‘holder,
frame: head [+human] or [+animate]’ (cf. OHG gibilla f. ‘head’, Greek κεφάλη f.
‘head’, Toch. A śpāl ‘head’) vs. *ɢɦɔ́bɦχl, *ɢɦɛ́bɦχlɔs = adj. *ɢɦɛ́bɦχlɔ- (“acrostatic”) ‘in-
animate holder, head of a house: gable’ (cf. OHG gibil m. ‘gable’, Gothic gibla (<
*γeβ(u)lōn), ON gafl ‘Giebelreiter’, NE gable (< *γaβ(u)la- m.)). Both stems are derived
from a root *ɢɦVbɦχ- ‘seize, hold, keep’ (cf. Rix et al. 2001, s.v. “ghebhh2-”);
(b) verbal forms: *gwɦɛ́n(t)(-i) (SEMELFACTIVE-TERMINATIVE-TELIC aspect) →

*g ɔ́n(-i) (TRANSITIONAL aspect), etc. (for the PIE verbal voice marking system
cf. Pooth manuscript a).
ROLAND A. POOTH 20

Parallel to the verbal 1st person sg. and 2nd person and 2nd or 3rd person col-
lective detransitive voice marker *-χ-, there was a regular nominal detransi-
tive suffix *-χ- (which also had a collective meaning). Note that “nominal
voice marking” is rare, but not absent crosslinguistically. For instance, there
are “middle nouns” in Temiar, cf. Benjamin 2011.

5 The PIE case forms


5.1 Ergative-genitive and ablative-genitive
The most prominent core case, that is, the PIE case marked by sg. *-s,
pl. *_ɛ_s is now termed ERGATIVE-GENITIVE case and is functionally inter-
preted as the case for the agent role and possessor role. An ergative-
absolutive alignment was first reconstructed for PIE by Van Wijk 1902.
However, it is has to be recognized that PIE was not a canonical ergative-
absolutive language, but one of the “active-stative” or “fluid-S” type, cf.
Dixon 2010 I: 77. There was a predominant semantic orientation of PIE
case morphosyntax, similar to the semantic orientation of the oblique cases
which is still evident in Early Vedic. The “ergative” must thus be viewed as
a semantic case for the semantic agent role which is volitionally acting,
controlling the action, and responsible for the action.
The emergence of the neo-lengthened ablaut grade (e.g. *χné:(r) <
*χnɛ́rs) by sound law (“Szemerényi’s law”) is datable to the Vulgar Pre-IE
period. In my view, this ablaut grade, which coded nominative and loca-
tive singular forms, has nothing to do with the PIE system of morphology.
It must sharply be distinguished from the PIE long (or geminate) vowels
which coded PLURATIVE nominal stems (e.g. *liɛ́ːkwr, *liɛkwnɛ́s > Vulgar Pre-
IE *ʔiéːkwr, *ʔiekwnés > Vedic yaknás, *pɛ́ːku (inanim.) ‘cattle’, etc.). To
properly reconstruct PIE nominal word forms, it is inevitable to eliminate
the neo-lengthened grade from the whole system of morphology. The sec-
ondary character of *χné:(r) is internally evident.
The ergative-genitive case marker *-s can be identified with the *-s of
the ablative-genitive case marker (cf. Van Wijk 1902).21 These two case
forms are viewed here as two subordinate instances of a superordinate
skeletal sigmatic consonant frame case stem, where different positions of
vowel slots *_V_ further distinguished these two forms:

(4.18) *ku_n-_s
↘ *kuɔ́ns ‘dog (animate agent)’ and ‘dog’s’
↘ *kunɔ́s ‘from dog(s)’

The ergative case is thus interpreted here as a semantic SOURCE OF


ANIMATE AGENTIVE FORCE/POWER case. It can be seen as the case for the
source of the agentive power and agentive causation force. Crosslinguisti-

21
The pronominal ablative, however, was marked by *-ɗ, e.g. *ʔmɛ́ɗ, *tuɛ́ɗ, etc. (> Vedic
mád, tvád, etc.). In Vulgar Pre-IE, *-ed was extracted and transfered to the o-stems, e.g.
*-oʔed (with hiatus) > Vedic -ād, etc.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 21

cally, such an identity of ablative case and ergative case is quite common,
cf. Heine & Kuteva 2002, s.v. “ergative”.
There are a few archaic IE genitive forms that show full grade or even
lengthened grade (sic!) of the root:

FIGURE 4.12.1.
*ɗɛ́m-s ‘family’s’ > Old Avestan də̄ṇg (cf. də̄ṇg paiti-)
*nɛ́ɠwt-s ‘dawn’s’ > Hittite neekuz22
*ɠwɛ́n-s ‘woman’s’ > Old Irish nom. and gen. sg. f. bé
*uɔ́ik-s ‘settlement’s’ > Old Prussian Wais- of Waispattin, etc.
*nɔ́ɠwt-s ‘night’s’ > Latin nox ‘at night’ (adverb, frozen gen.)
*uɔ́ːkw-s ‘speech’s’ > Avestan gen. sg. m. vāxš

Furthermore, there are a few archaic IE genitive forms that show zero
grade of the suffix. Tremblay 2004: 579, fn. 30 also gives Avestan garəbuš
(V. 9.38), maδuš (V. 14.17) ‘of wine’ and some more problematic cases:

FIGURE 4.12.1.
*máː-tr-s > Vedic gen. sg. f. mātúr (patúr is secondary)
*bɦráχ-tr-s > Vedic gen. sg. m. bhrā́tur (dā́tur should be secondary23)

These archaic IE genitive forms can be understood differently than


Schindler 1967a, 1972 once did it. Their functional prehistory is detectable
by help of PIE *ɗiɛ́u- which is a “hysterokinetic” nominal stem with the
meaning ‘bright/shining one: sky, daylight’. Its underlying verbal root is
*ɗVi- ‘shine, be bright’. Originally, it was not a proper root noun, but was
marked by a suffix *-u- (skeletal *ɗi-u-). The corresponding nom. sg. form
obviously received the neo-lengthened grade in Vulgar Pre-IE (*diéus →
*diéːus > Vedic dyáuṣ, etc.). In that period, this stem was reanalyzed as a
neo-root-noun. It is a relevant fact that Vedic dyáuṣ reflects two abl.-gen.
sg. forms dyóṣ and divás. The first one reflects *diéus which is formally
identical to the old s-case stem form *ɗiɛ́u-s (which was its form before
receiving the neo-lengthened grade in Vulgar Pre-IE). It has to be con-
cluded that the Vedic abl.-gen. form dyóṣ is an archaism. It is less plausible
to interpret it as a secondary form, whereas divás looks a bit younger.
NB. There is further internal evidence. The Vedic paradigm (nom. sg. m. véṣ
~ víṣ ‘bird’, acc. sg. m. vím, abl.-gen. véṣ, etc.) does not go back to a root noun
“*h2uóis, *h2uéis”, pace Schindler 1969, cf. the criticism of de Vaan 2008, s.v. It
rather reflects the RELATIONAL “hysterokinetic” paradigm *χuɛ́is ~ *χuís ‘s.o.’s
bird; bird-like, birdish’ (skeletal *χ_u-i-). I think that the Vedic gen.-abl. m.
form véṣ < *χuɛ́is is an archaism that parallels the archaic dyóṣ. In my view,

22
Its meaning ‘dawn’s’ does not make it a good candiate for an abl.-gen. of ‘night’. I sim-
ply conclude that *nɛ́ɠwt ‘dawn, getting bare’ was a nomen actionis, whereas *nɔ́ɠwt
‘night’ was its result ‘the one that is the product/result of dawn, getting bare’.
23
The PIE abl.-gen. of *ɗáʕwtɔr was either *ɗáʕwtrɔs parallel to *bɦɛ́rɔnt, *bɦɛ́rɔnti, *bɦɛ́rntɔs
(> Vedic abl.-gen. sg. bháratas) or *ɗʕwtrɔ́s parallel to *uɛ́ɗɔr, *uɗɔ́n, *uɗnɔ́s.
ROLAND A. POOTH 22

both forms go back to old strong ergative-genitive forms. Latin avis m. ‘bird’,
on the other hand, simply reflects the NONRELATIONAL “proterokinetic” para-
digm *χáuis, *χuɛ́is ‘bird’ (cf. Beekes 1995: 175). The problem of *χu̯V >
Latin avV (avis) can thus be avoided (cf. *χi̯V > *i̯V in iuvenis with loss of *χ
before the glide). The PIE words for ‘bird’ (skeletal *χ_u-i-) can ultimately be
connected with the PIE word for ‘sheep’ (*χɔ́ui) and an underlying root *χVu-
‘dress, be dressed, clothe oneself’, cf. Rix et al. 2001: 275: “h2eu(H)- “(Fußbek-
leidung) anziehen”. The meaning given there, however, is certainly too spe-
cific (n. 1: “Ansatz einer Seṭ-Wurzel [...] ist [...] nicht sicher”). Its etymology is
now easy to understand. Birds are animals that are usually beautifully dressed
in feathers, cf. Modern German Federkleid, etc. On the other hand, the sheep is
the domestic animal that produces wool for garment and dresses. Under this
perspective, it turns out to be easily explicable why these two animal words
are so similar.24
For the above given reason, the archaic IE genitive forms are better not
traced back to forms of the PIE ablative-genitive. Instead, parallel to Vedic
dyóṣ < *ɗiɛ́us these forms should better be traced back to PIE ergative-
genitive case forms in ‘genitive function’ marking the possessor role.
It is thus inferrable that the sigmatic ergative-genitive case could be
used as coding the possessor within a specific genitive or possessive con-
struction. This idea implies that the opponent oblique ablative-genitive
case could only be used in a different adjunct-like construction. It can be
suggested that the ergative-genitive form, e.g. *ɗɛ́m-s (> Old Avestan də̄ṇg,
cf. də̄ṇg paiti-) marked the possessor if it was directly juxtaposed to the left
of the possessee, e.g. *ɗɛ́m-s *pχtɛ́r- ‘the family’s father’: see the examples
given below. It can further be inferred that the possessee role had to be
additionally coded by a RELATIONAL form. Within this specific genitive or
possessive construction both possessor and possessee had to be marked.
Typologically, this is comparable, e.g., to Modern Turkish. Most presuma-
bly, these forms had to be in direct juxtaposition and no other word form
was allowed to come inbetween. One may term this genitive or possessive
construction the “status constructus” of PIE:
24
Another word that can be connected to this root is the PIE word for ‘egg: product of
bird’ (*χɔ́ui, *χáuiɔs). The old idea of a compound “*(H)o-h2u̯i ̯-o-” is rather obscure.
Such an etymology is definitively too ad hoc. Proto-Germanic *VjjV of *ajjaz/iz- points
to a preceding consonant cluster *VjCV (cf. Proto-Germanic *wajju < *uɔ́i ʔu). (1) It may
reflect *ajwaz/iz- with metathesis *ajwV ← *awjV. This *ajwV then differed from *aiwV-,
cf. Gothic aiws ‘time, eternity’ < Proto-Germanic *aiwa-. It may thus reflect *χáuiɔs
(*mɛ́nɔs type) and point to an allomorphic stem *χɔ́uis (cf. Vedic āyuṣ-́ ́
n. besides āyu- n.,
etc.). (2) But it can also reflect *ajj V- < *oːjj V- < *oːwj V- (< *χoːu̯i ̯ó-) and may simply
go back to what is reflected by Latin and Greek (and Albanian ve ‘egg’). These languages
ultimately point to PIE *χɔ́ui. Latin ōvum ‘egg’, Greek ὠιόν (< *χoːu̯i ̯ó-) both point to a
paradigm with long grade *χɔ́ːui, *χauiɔ́s which must have patterned like *uɔ́ːkw, *uɛkwɔ́s
and simply was a collective nominal stem ‘group of egg(s) (in a nest)’. A vr̥ddhi-
derivation from underlying ‘bird’ is ruled out, because the vocalism points to an under-
lying *χɔ́ui, not to *χáui ‘bird’. It can be concluded that it is a much better to recon-
struct a collective stem *χɔ́ːui ‘egg-group’ and an underlying inanimate stem *χɔ́ui ‘egg’.
I reconstruct the following paradigms: (1) *χáui, *χuɛ́is anim. (i-stem, cf. Beekes 1995:
175) ‘bird: the one who dresses, is dressed (in feathers)’; (2) *χuɛ́is ~ *χuís (“hysteroki-
netic”) vs. (3) *χɔ́ui, *χáuiɔs inan. ‘egg: product of bird’; (4) *χɔ́ːui, *χauiɔ́s (→ adj.
*χoːu̯i ̯ó-) ‘group of eggs (in a nest)’; (5) *χɔ́uis, *χáuiɔs animate ‘sheep: the one who is
used for dressing’.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 23

(4.19) a. *ɗɛ́m-s *pɔtɛ́i-m ~*pɔtí-m


family-ERG/GEN master:RELATIONAL-ALL
‘the family’s, to its master’
b. *uɔ́ːkw-s *pɔtɛ́i-m ~*pɔtí-m
speech-ERG/GEN master:RELATIONAL-ALL
‘the speech’s, to/at its master’

A second possessive ablative-genitive construction, on the other hand,


was a simple ablative-genitive adjunct-like apposition. Here, the possessee
was not necessarily coded by a relational stem:

(4.20) a. *pɔ́t(i)-m ... *ɗmɛ́_s


master-ALL family.OBLIQUE_SOURCE
‘to a/the master ... the one of the family’
b. *pɔ́t(i)-m ... *uɛkwɔ́_s
master-ALL speech.OBLIQUE_SOURCE
‘to a/the master ... the one of the speech’

Both constructions and both case forms thus simply merged from PIE to
Vulgar Pre-IE, cf. Avestan gen. sg. m. vāxš besides vacō, Latin nox besides
noctis which are variants like Vedic dyóṣ and divás.
It can be concluded that the few archaic IE genitive forms do not go
back to “weakest” ablative-genitive forms, but reflect the form of the old
“strong” ergative-genitive case.
NB. This entails that the old ad-hoc-reconstruction of a pre-form “**g̑énus”
which was analogically leveled to *g̑énu̯os has to be rejected. The PIE ablative-
genitive form of *ɠɔ́nu must have been *ɠɛ́nuɔs. First, this is what is actually
attested in the IE languages (cf. Hittite geenuwas, s. above). The corresponding
partitive-ablative *ɠɛ́nuɔm is reflected by Greek Hsch. γέυνων < Vulgar Pre-IE
*g̑énu̯oʔom (cf. Tremblay 2004: 577 with fn. 21). Second, this is what also ma-
kes much more sense with regard to the discontinuous detransitive marker *ɔ.
It can be concluded that all forms of the respective paradigm were marked by
this morpheme. It can further be concluded that the reconstruction of the abl.-
gen. form of the “acrostatic” inflectional type must be corrected this way.
Thus, for instance, the Vedic abl.-gen. sg. m. ávyas obviously goes back to
*χáuiɔs, cf. Latin nom. sg. m. ovis :: Vedic nom. sg. m. áviṣ ‘sheep’ (<*χɔ́uis).
There is more comparative evidence: Vedic abl.-gen. sg. vásvas and Homeric
Greek gen. sg. ἐῆος, ἑῆος (< Proto-Greek *ewéhwos) both point to PIE abl.-gen.
sg. *ʔuɛ́suɔs (*ʔuɔ́su ‘good, worthy thing’). There is no comparative evidence
for a ‘Schindlerian’ abl.-gen. “**h1u̯ésus”. *nɛ́ɠwts ‘dawn’s’ never belonged to
the paradigm of *nɔ́ɠwt- ‘night’ (see fn. 21).

5.2 Absolutive and “endingless” locative


It is inferrable via inductive reasoning based on internal evidence com-
bined with typological implications that PIE had a zero- or unmarked
ABSOLUTIVE case (depending on how one prefers to define the notion of
ROLAND A. POOTH 24

‘zero’). This inference is drawn on the basis of the IE unmarked neuter


nominative-accusative case forms. It is inferred that the animate nominals
stems also showed such a case marking before the PIE allative-dative was
reanalyzed to accusative (see below, Pooth 2004b, Pooth manuscript a).
The zero case marking of the IE neuter stems is retrogradely generalized
and transferred to the non-neuter animate nominal stems such as, e.g.
*kuɔ́n- ‘dog’ → abs. sg. *kuɔ́n. Such a generalization is a valid and legiti-
mate step within the progress of internal reconstruction, for the major line
of argumentation cf. Pooth 2004b:

FIGURE 4.13.
INANIMATE casein *-Ø ANIMATE case in *-Ø
*ɗɔ́ru ‘wood’ *ʔɔ́gwɦi ‘snake, worm’
*mɛ́dɦu ‘honey’ *pɛ́ku ‘domestic animal’
*ʔɛ́ku ‘horse’ (with adj. *ʔɛ́kuɔ ‘horse-ish’)
*χnɛ́r ‘man’
*χɔ́ui ‘sheep’ (from *χVu- ‘clothe, clothing’)
*pɔ́ɗ ‘foot’

It is inferrable that all Vulgar Pre-IE variants gave up the zero-marked


animate absolutive case forms parallely but relatively independently, be-
cause all Vulgar Pre-IE variants parallely switched into nominative-
accusative languages with a new animate accusative case form, e.g. Vulgar
Pre-IE *k(u)u̯ónm̥, etc.
But there are some left-overs of the absolutive case:
(a) The PIE abs. sg. form is reflected as Vulgar Pre-IE voc. sg. form, e.g.
*pχ́ter ~ *=pχter (but with a different tone or accent position).
(b) The PIE abs. pl. form *pχtɛ́rɛ merged with the PIE sociative-associative
forms *pχtɛ́rʔ ‘man & co.’, pl. *pχtɛ́rɛʔ ‘man & co.’ or ‘men & co.’. These three
forms developed into Vulgar Pre-IE nom.-acc. dual forms. The Greek nom.-acc.
dual m. πατέρε ‘two fathers’ reflects one of the two first forms, whereas the
third one is reflected by the Vedic nom.-acc. dual form, cf. Vedic pitárā ~
pitárau. In my view, the Vedic -u goes back to a PIE enclitic particle *=u
‘then, and then, and also, in addition’ which fits to the original associative
reading ‘father and then/and also his companion’. Not only for this reason, I
seriously doubt the prehistorical reality of “Jasanoff’s law”, cf. Weiss 2009:
209.
The so-called “endingless” locative-allative case form was also based on
the zero-marked case stem. This case form was marked by a different
vowel melody than the core case forms―it was thus “internally derived”
from the absolutive form:

(4.21) *dɦ_gɦ_m-Ø cf. Hittite


↘ *dɦɛ́gɦɔm > nom.-acc. sg. n. teekan
↘ *dɦgɦɔ́m > loc. sg. n. tagaan
↘ *dɦgɦmɔ́ > tagnaa
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 25

Some nominal stems―mostly stems of the type(s) marked by the discon-


tonuous detransitive marker ɔ―also displayed an allomorphic zero-marked
case form of the benefactive-purposive case (e.g. *dɦgɦmɔ́ ~ *dɦgɦmɔ́i). Par-
allel to the “endingless” locative, this form was also based on the zero-
marked case stem. Like the benefactive-purposive it developed into the
new Vulgar Pre-IE dative case.
I suggest that these zero-marked case forms (e.g. *dɦgɦmɔ́ without *-i)
are still reflected as the Old Hittite so-called “directive” dative forms, e.g.
tagnaa ‘to the earth’, hapaa ‘to the river’. There is no need to reconstruct a
completely separate “directive/allative” case.
I also tentatively suggest that some “endingless” locatives are reflected
by IE vocative sg. forms, e.g. *pkɛ́u ‘for/towards the domesticated animal’
→ ‘o (this one goes out) to/towards the domesticated animal’, whence
Vedic -o, e.g. voc. sg. m. vaso, etc.
NB. In my view, the voc. sg. m. of the thematic o-stems, e.g. *u̯l ̥k ́ we ‘o wolf’
was a parallel Vulgar Pre-IE innovation. It is possible and plausible that these
forms were analogically changed from PIE *ulkwɔ́ ‘dangerous (one)’ to *u̯l ̥́kwe
in Vulgar Pre-IE constructions such as *tu̯é ~ *té *u̯l ́k̥ we ‘(I address to) you, o
wolf’ (going back to prior PIE *tuɛ́ ~ tɛ́ *ulkwɔ́ ‘(to) you, o dangerous one’, cf.
PIE abs. sg. *tɔ́ *ulkwɔ́ ‘this non-topical dangerous one’, *sɔ́ *ulkwɔ́ ‘this topical
dangerous one’). In addition, these forms may have been interpreted as formal
and functional nominal equivalents of the new verbal Vulgar Pre-IE thematic
2nd sg. imperative forms, e.g. *bɦére ‘(you) carry!’ which also had a secondary
*-e (and go back to prior 3rd sg. detransitive intransitive forms like *bɦɛ́rɔ).
I generously admit that there could have been a Vulgar Pre-IE sound law,
changing final unaccented PIE *-ɔ > *-e /_# in absolute auslaut which was
followed by analogical retentions of *-o in that position (in analogy to the
Vulgar Pre-IE nom. sg. m. *-os or a 3rd sg. middle present imperfective ending
*-toi, etc.). Whether there was a sound law or not, I have nevertheless decided
to view forms like *u̯l ̥́kwe ‘o you wolf’ or *bɦére ‘you carry!’ as Vulgar Pre-IE
innovations.

5.3 Allative-dative and partitive-genitive


The PIE allative-dative case form indicated a (PARTITIVE) GOAL AS PART OF
AN ANIMATE BODY OR A PART OF A WHOLE, prototypically a human body, one
that is only partially affected and prototypically an experiencer of an ani-
mate sentience and affection. Since inanimate nominal concepts prototypi-
cally cannot be experiencer of an affection, these were semantically in-
compatible to the allative-dative case marking in PIE. Inanimate nominal
stems thus did not show allative-dative case forms.
I further take the allative-dative case in *-m- for the original recipient
role case of PIE animate nominal stems. Verbs like *ɗVʕw- ‘give; take, re-
ceive, obtain, get’ originally took the PIE allative-dative, that is, the ‘pre-
IE-accusative’ to mark the recipient.
The following PIE sentence includes an agent, a patient and a recipient
semantic role. Depending on pragmatical needs, there also may have been
ROLAND A. POOTH 26

a word order with recipient role before patient role, because PIE had dis-
course-pragmatically triggered word order.

(4.22) *χnɛ́r-s *χɔ́ui *pχtɛ́r-m *ɗáʕw-t


man-ERG sheep.ABS father-ALL give:SEM:AGT:SG:3-DIR
agent patient recipient
lit. ‘man (agent) gave sheep (theme) to/at father (recipient)’

This morphosyntactic pattern differs from what has been reconstructed


by the comparative method within the traditional model. It has already
been outlined above, however, that the ‘old comparative model’ can only
have the provisional status of representing or symbolizing the younger IE
and post-PIE morphosyntactic patterns. With regard to morphology and
syntax, the ‘old comparative model’ has little to do with the actual PIE
proper language (cf. Pooth manuscript b).
It is internally evident via comparison that the genesis of the dative case
in *-ei is not datable to a period prior than Vulgar Pre-IE. It is internally
evident that *-ei was paradigmatically generalized not before the post-PIE
period. The full grade ablaut of the suffix, e.g. Vedic agnáye, matáye, etc.
points towards a prior zero-marked locative *mntɛ́i which was remodelled
by extension of *-ei → Vulgar Pre-IE *mn̥-téi̯-ei, etc.
This extension strengthens the following generalization: Since many of
the ‘comparative IE dative’ forms in *-ei look as if they were of recent ori-
gin, the entire dative case in *-ei looks like a Vulgar Pre-IE innovation.
Therefore, I hereby draw the inference that *-ei spread as a new marker of
the Vulgar Pre-IE dative case originating from the agentive-unmarked
benefactive-purposive case forms, e.g. *χnrɛ́i ‘for (the benefit of) a/the
man, for a/the man, for man, for the men, for men’.
As I have outlined in detail elsewhere (cf. Pooth 2004b, Pooth manu-
script a), I think that the PIE antipassive construction including oblique
allative-dative case marking (in *-m-) of the patient or theme, e.g. *χnɛ́rs
*stɛ́uɔ *kuɔ́nm ‘man was praising at/to dog’ was reanalyzed as the new
transitive construction ‘man praised dog’ in Vulgar Pre-IE (see Pooth
2004b, Pooth manuscript a for the argumentation). As a consequence, the
former allative-dative (indirect) goal case became the new Vulgar Pre-IE
accusative case for the direct goal or theme or patient role. From that very
moment onward, Vulgar Pre-IE more or less urgently needed a new dative
case for the recipient semantic role. To suit this purpose, the suffix *-ei of
the benefactive-purposive was generalized as a new Vulgar Pre-IE dative
case suffix. Note that such a grammaticalization path BENEFACTIVE →
DATIVE is widely attested, cf. Heine & Kuteva 2002: 54.
The former PIE benefactive-purposive case, however, was mainly used
to indicate a beneficiary role and a purpose, that is, an indirect goal:

(4.23) a. *pχtɛ́r-s *dɦɛ́ʔt *χnrɛ́i


father-ERG put:SEM:AGT:SG:3DIR man:BEN
‘father made it for (the benefit of) (a/the) man’
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 27

b. *pχtɛ́r-s *ɠwɛ́mt *uiɗɛ́i


father-ERG come:SEM:AGT:SG:3DIR see:BEN
‘father came here/there to see (him/her/it/someone)’

As a consequence, I think that the Vulgar Pre-IE innovative dative case


has to be eliminated from a proper model of the PIE case system. Thus I
hereby dare to refuse to reconstruct a dative case with suffix(es) *-ei, *-oʔei
→ *-ōi for any stage prior than Vulgar Pre-IE. Instead, the forms of the
Vulgar Pre-IE accusative case (e.g. *pχtérm̥) must now functionally be re-
transformed to the case forms they were before, namely PIE animate alla-
tive-dative case forms. This point may lead to a longlasting controversy,
but I think that it is internally evident.
Another PIE case form that contained the suffix *-m- was the partitive-
genitive case. The partitive-genitive case had the function of attributing
the property BEING PART OF A WHOLE OR BEING AMONG A WHOLE OR THE SUM OR
A LOT (of the given nominal concept) to the particular participant:

(4.24) *pχtɛ́r-s *páʕwt *mɛ́dɦuɔm


father-ERG swallow:SEM:AGT:SG:3DIR alcoholic.honey:PAR
‘father swallowed (part) of the alcoholic honey-drink’

Both the allative-dative case form and the partitive-genitive case form
were derived from an underlying allative-dative plus partitive-genitive case
stem with superordinate case suffix *-m-.

(4.25) *ku_n-_m
↘ *kuɔ́nm ‘at/to dog’
↘ *kunɔ́m ‘(at/to) part of dog(s)’

Thus both the PIE allative-dative case form and the partitive-genitive
case form were derived from an underlying indirect goal case stem. Interest-
ingly, there is some crosslinguistic overlap in this functional domain, cf.
Heine & Kuteva 2002: 333f.: “partitive ← [...] possessive ← (3) dative, (5)
locative”. This overlap may speak in favour for a functional closeness of
the partitive and indirect goal case meaning.
It is relevant for a proper understanding of the relationship between the
animate allative-dative case and the partitive-genitive case that within the
PIE antipassive construction, the animate dative-allative case could also be
used in partitive-like function:

(4.26) *χnɛ́r-s *pʕwɔ́ *ʔɔ́gwɦim


man-ERG swallow:SEM:ITR:3SG:DTR snake-ALL
‘man swallowed (part) of the snake/worm’,
literarily ‘man swallowed at/to snake’

Only later, the PIE partitive-genitive case form developed into the new
Vulgar Pre-IE genitive plural case form. The new portmanteau suffix *-om
ROLAND A. POOTH 28

(which orginated from forms like *kunɔ́m) was then completely generalized
as a Vulgar Pre-IE marker of the genitive plural. Furthermore, the o-stem
suffix *-o- was extended by *-om yielding *-oʔom (with simple hiatus). This
secondary suffix also was generalized to some extent (e.g. *χn(e)róm ~
*χn(e)róʔom > Vedic narā́m).
In fact, the morphological spread of the new gen. pl. ending *-om is
clearly confirmed by the Vulgar Pre-IE demonstrative gen. pl. *tóisom ~
*tóisoʔom which goes back to a prior PIE demonstrative loc. pl. *tɔ́is that
was pleonastically extended by *-om ~ *-oʔom (see below section 6.1.).
The Gothic gen. pl. m. -ê (e.g. dagê, wulfê vs. -ô otherwise) can go back to a
Vulgar Pre-IE gen. pl. ending *-eː(m) with lengthening of *-em (*_ɛm) in para-
digmatic analogy to its productive variant *-oː(m) < *-oʔom. If this is true, the
ending *-eː(m) simply was an allomorph of *-oː(m) and was thus transferrable to
Vulgar Pre-IE o-stems (e.g. *u̯l ̥́kweː(m) ~ *u̯l ̥́kwoː(m) < *u̯l ̥́kweʔem ~ *u̯l ̥́kwoʔom
parallel to *u̯l ̥́kwes(i̯)o ~ *u̯l ̥́kwos(i̯)o, etc.). There is no compelling explanation
for why Gothic shows a gen. pl. in -ê besides -ô among the proposals that have
been suggested so far. All these explanations suffer from being ad-hoc-
assumptions. With regard to all the given IE ablaut variants (e.g. abl.-gen. sg.
*-es ~ *-os) it cannot be implausible―and it is definitively not ad hoc―to infer
that this suffix simply reflects a Vulgar Pre-IE ablaut variant which was com-
pletely eliminated in all the other IE branches. But note that Gothic -ê is not
urgently needed to legitimate the internal reconstruction of PIE forms like
*χnrɛ́m vs. *kunɔ́m.
Note that the Vedic irregular gen. pl. form nŕ̥n ̄ ‘(part) of men, among men’
(e.g. RV 1.121.1a nr̥̄́m̐ḥ ... devayatā́ṃ) is identical to the regular acc. pl. m.
form nr̥̄́n of Vedic nár- ‘man’. This Early Vedic synchronically irritating overlap
now finds an easy diachronic explanation and finally confirms a diachronic
functional closeness of both cases. Vedic nr̥̄́n thus goes back to a Vulgar Pre-IE
allomorpic variant of *χnérn̥s ~ *χnŕ̥ns reflecting the old PIE allative-dative pl.
form *χnɛ́rms, but with Vulgar Pre-IE analogical leveling of the root ablaut
*χnŕ̥- according to the other plural forms (e.g. loc. pl. *χnŕ̥-su, etc.). The fact
that Vedic nr̥̄́n is attested as an allomorphic variant of narā́m (<*χneróʔom)
besides an even younger variant nr̥ṇā́m clearly shows that the genitive plural
category was subject to abundant allomorphy and innovation within Vulgar
Pre-IE.
Note, finally, that Archaic Latin acc. sg. m. seruom = gen. pl. seruom con-
firms a prior formal identity of the allative-dative and the partitive-genitive of
the PIE “thematic” inflectional type (e.g. *nɛ́uɔm, see figure 4.18 below).

5.4 Locative-allative and benefactive-purposive


The locative-allative case form was either based on the case stem with
suffix *-i- or on the zero-marked case stem. In parallel to the locative-
allative case, the PIE benefactive-purposive case had two allomorphic
forms. There was one case form which was based on the case stem with
suffix *-i-. The other form was based on the zero-marked case stem. Both
case forms, that is, locative-allative and benefactive-purposive were thus
derived from a common underlying case stem.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 29

(4.27) a. *χn_r-_i
↘ *χnɛ́ri ‘to/at/in/... man’
↘ *χnrɛ́i ‘for man/men’
b. *ku_n-i
↘ *kuɔ́ni
↘ *kunɔ́i → Vulgar Pre-IE dat. sg. *kunéi
ɦ ɦ
c. *d _g _m-Ø cf. Hittite
↘ *dɦgɦɔ́m > loc. sg. n. tagaan
↘ *dɦgɦmɔ́ > “directive” dative sg. n. tagnaa

With regard to the evident spread of the new dative sg. suffix *-ei within
the Vulgar Pre-IE period (as outlined in detail above) it is very plausible to
infer that the former PIE detransitive benefactive-purposive case forms of
the *kunɔ́i or *dɦgɦmɔ́i type were remodelled to forms with this new and
productive dative singular suffix *-ei, e.g. *dɦgɦmɔ́i → Vulgar Pre-IE
*dɦgɦméi > Hittite taknii, etc. Therefore, I do not regard forms like *dɦgɦméi
(and *dɦgɦém) but *dɦgɦmɔ́i (and *dɦgɦɔ́m) as original PIE forms belonging to
*dɦɛ́gɦɔm. Note that *dɦgɦɔ́m with ɔ́ before m is ultimately confirmed by
Hittite tagaan.

5.5 Sociative-associative and comitative-instrumental


The PIE case stem marked by the suffix *-ʔ, e.g. *ku_n_ʔ was the su-
perordinate case stem for the notion of COMPANION(S), COMPANY,
INSTRUMENT, MANNER, and PATH. Besides a sociative reading (e.g. ‘with a
dog, with the dog’) the “strong” form (e.g. *kuɔ́nʔ) also had an associative
meaning ‘dog & co.’, that is, ‘dog with a companion or in company’. The
“weak” relational form, however, only had a comitative-sociative meaning
(e.g. ‘with, through/by a/the dog(s)’) including an instrumental reading:

(4.28) *ku_n-_ʔ
↘ *kuɔ́nʔ ‘with the dog’ or ‘dog & co.’
↘ *kunɔ́ʔ ‘with dog(s)’

As just mentioned, the sociative-associative so-called “singular” forms,


e.g. *kuɔ́nʔ were thus polysemeous (and, to a certain extent, transnumeral)
and had two major readings: (α) a sociative reading ‘with dog’ and (β) an
associative dual or plural number reading ‘dog with (his) companion or
company’. I suggest that this PIE synchronic polysemy functioned as fol-
lows: (α) It was either a/the given referent, e.g. *χnɛ́r *kuɔ́nʔ ‘man with
dog’ that was referred to as being accompanied by the given concept, e.g.
[DOG]―or (β) it was the concept [DOG] itself that (as the given referent)
was referred to as being accompanied by ‘someone’, that is, an implicit ‘X’.
(Note that this polysemy runs in parallel to the one proposed for the rela-
tional nominal stems, see figure 4.8. above).
ROLAND A. POOTH 30

FIGURE 4.14.
polysemeous:
*kuɔ́nʔ ‘X’ or given referent ACCOMPANIED BY [DOG]
[DOG]as referent ACCOMPANIED BY ‘X’

It is easy inferrable that the forms with associative reading, e.g. *kuɔ́nʔ
‘dog & co.’ developed into nom.-acc. dual forms, e.g. *kuónʔ ~ *kuóneʔ
‘two dogs’ in Vulgar Pre-IE.25

6 Number marking [provisional sketch]


6.1 Number forms of PIE animate nominal stems
To keep this section brief: In my view, PIE animate nominal stems were
canonical “single object nouns”, that is, individuated entitiy nominal stems
(cf. Rijkhoff 2002: 54). Besides forming SINGULAR forms, animate nominal
stems , e.g. *χnɛ́r- ‘man’ displayed forms of the following number catego-
ries:
(a) Animate nominal stems displayed canonical PLURAL (PL) number
forms, e.g. erg. pl. *χnɛ́rɛs ‘men (agent)’, abs. pl. *χnɛ́rɛ ‘men’ referring to a
countable plural number of men.
(b) PIE also had a special ASSOCIATIVE (ASS) number category, e.g. *χnɛ́rʔ
‘man & co.’ referring to one man and either one or several companions.
Such a category is not preserved in the later IE languages. However, I think
it is plausible to reconstruct it. I suggest that associative forms (e.g. ‘dog &
co.’) were functionally narrowed to dual number forms (e.g. ‘two dogs’) in
Vulgar Pre-IE.
(c) There were optional ASSOCIATIVE PLURAL (ASS.PL) forms, e.g. *χnɛ́rɛʔ
‘man & co.’, referring to one man or more than one men with either one or
several companions.
(d) In my view, PIE also had transnumeral-polysemeous COLLECTIVE
(COL) forms, e.g. *χnɛ́rχ ‘man-group (plural number)’ and ‘(single) man
belonging to a group of men’. I suggest that PIE had both an inflectional
collective number and further derivational collective stems. Both were
obviously coded by *-χ- (e.g. *ɠwɛ́nχ-, abl.*ɠwnáχs ‘women-group, woman
belonging to a group’). For the line of argumentation see Eichner 1985. It
has lately been argued by Litscher 2014 that PIE should have had an in-
flectional collective number category; see the recent book by Neri &
Schuhmann 2014.
(e) I further suggest that here were corresponding optional COLLECTIVE
PLURAL (COL.PL) forms, e.g. *χnɛ́raχ ‘man-group, men-groups, men belong-
ing to a group, men belonging to groups’.
Opposed to the core and extended core case forms, the PIE oblique case
forms, that is, ablative-genitive, partitive-genitive, benefactive-purposive

25
Similar Kölligan 2004: 211f. and fn. 42 who refers to Haudry, J. 1994: L’Indo-Européen.
Paris, p. 37.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 31

and comitative-instrumental were transnumeral and thus did not display


specific plural, nor associative, nor collective forms.
A paradigm of a PIE animate nominal stem, for instance *χnɛ́r- ‘man’ is
given in the following figure.

FIGURE 4.15.

SG PL ASS ASS.PL COL COL.PL


ABS *χnɛ́r *χnɛ́rɛ *χnɛ́rʔ *χnɛ́rɛʔ *χnɛ́rχ *χnɛ́raχ
ERG *χnɛ́rs *χnɛ́rɛs
ALL *χnɛ́rm *χnɛ́rms
LOC *χnɛ́ri *χnɛ́ris *χnɛ́riʔ *χnɛ́riχ
SOC *χnɛ́rʔ
ABL *χnrɛ́s
PAR *χnrɛ́m
BEN *χnrɛ́i
COM *χnrɛ́ʔ

Jasanoff 2009: 141 has given IE evidence pointing to Vulgar Pre-IE *-is
(“argueably with instr. pl. value”). However, I suggest that―at least mor-
phologically―this ending obvisouly goes back to locative plural form
marked by *-i-s (structured like *-m-s). I think that the PIE locative plural
forms could cover a broad “oblique” meaning. It is inferrable from the
comparative and internal evidence that PIE once must have had locative-
plural-and-sociative-plural case syncretism. The Vulgar Pre-IE demonstra-
tive plural form *tóːis (≡ *tṓis) very much looks like a proper locative plu-
ral form (*-i-s) (structured like *tóːns (≡ *tṓns) ~ *tóns), but turns out to
function as an instr. plural form in the IE languages (e.g. Vedic táis, etc.).
Therefore, I simply conclude that it goes back to a locative plural form that
also had the function of a sociative plural form. There is possibly no better
explanation for this old morphological riddle.
I suggest that from PIE to Vulgar Pre-IE, the loc. pl. *tɔ́is was morpho-
logically lengthened to *tóːis (> Vedic táis) in analogy to other plural
forms in Vulgar Pre-IE, e.g. acc. pl. *tóːns ~ *tóns, *u̯l ̥́kwoːns ~*u̯l ́k
̥ wons,
where the lengthened stem vowel *-oː- was analogically transferred from
the Vulgar Pre-IE nom. pl. m. *u̯l ̥́kwoːs ~*u̯l ́k
̥ woi. The analysis of Jasanoff
2009: 143 (who parses this form as “*tói + *-is” and analyzes “*tói-s” as a
“collective genitive” form) may not be correct.
Jasanoff 2009: 141 has not used the following internal evidence that
speaks in favour of a simple pleonasm: The suffix *-is is ultimately con-
firmed not only by Old Latin (inscriptional) sokiois > Latin sociīs, but also
by the demonstrative plural form *tɔ́is which was obviously simply pleo-
nastically extended by *-om ~ *-oʔom yielding the Vulgar Pre-IE demon-
strative gen. pl. *tóisom ~ *tóisoʔom (> Vedic téṣām). Jasanoff’s own pars-
ROLAND A. POOTH 32

ing (“*tói-s-oHom”, cf. Jasanoff 2009: 154 with fn. 15) is less plausible and
problematic, because there is no independent evidence, neither for a “col-
lective genitive” form “tói-s”, nor for a collective stem “tói-”.26 Therefore,
we should better reconstruct the loc. pl. form as *tɔ́is with *-i-s. It is more
promising to simply infer that *tóːis (≡ *tṓis) has undergone a secondary
lengthening like other plural forms as given above.
It is an entailment of this view that the original functional scope of PIE
locative *-i-s also covered the genitive plural function, that is, the endod-
ing of a possessor role. This is exactly paralleled by the enclitic pronouns
*=mɔi, *=tɔi, *-sɔi which once indicated the PIE locative-allative case plus
the possessor role (and even more roles).
Thus it looks as if forms in *-i-s could be used to supply all the PIE
transnumeral oblique cases (6, 7, 8, 9) including the sociative case (5) with
a more specific plural form.

6.2 Number marking of PIE animate nominal stems


The PIE proper parent language made use of two morphological strate-
gies to mark plural number forms of nominal stems.
(a) The first morphological means was a transfixal marker. The plural
forms of the two core cases, that is, absolutive and ergative case were
coded by a PLURATIVE vowel *_ɛ_ in the rightmost vowel slot before the case
suffix slot (_K) (but *_a_ before *-χ-):

FIGURE 4.16.1.
SG PL
ABS *χnɛ́r :: *χnɛ́r_ɛ
ERG *χnɛ́r-s :: *χnɛ́r_ɛ_s
ABS.ASS *χnɛ́r-ʔ :: *χnɛ́r_ɛ_ʔ
ABS.COL *χnɛ́r-χ :: *χnɛ́r_a_χ
WFT *CCɛ́C-K :: *CCɛ́C_ɛ_K

The optional collective and associative plural forms *χnɛ́raχ, *χnɛ́rɛʔ


were also formed this way. As just mentioned, I assume that the corre-
sponding so-called “singular” form, e.g. *χnɛ́r-ʔ was transnumeral and
could refer to either, for instance, ‘man & co.’ or ‘accompanied men, men &
co.’.
(b) On the other hand, PIE also made use of agglutinating suffixation.
The plural forms of the two other “strong” cases, that is, the ones of the
extended core cases allative-dative and locative-allative were simply suf-
fixed by a plural number (N) suffix *-s (~ *-n ?) to the right of the respec-
tive case suffix (K):
26
It can seriously be doubted that Jasanoff 2009: 146ff. is correct by reconstructing “*tói-”
as a neuter “collective” stem. The *-i looks more like a locative suffix added to *tɔ́_. It is
thus possible that the Vulgar Pre-IE nom. pl. form *tói simply goes back to a PIE loc. sg.
form which originally had adjunct status and thus did not show case, number, and gen-
der agreement, e.g. *tɔ́i *χnɛ́rɛs ‘at that place men (agent)’ and *tɔ́i *ɗɔ́ru(χ) ‘at that
place wood(s)’ (besides *tɔ́ːs *χnɛ́rɛs and *tɔ́ɗ ~ *táχ *ɗɔ́ru(χ)).
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 33

FIGURE 4.16.2.
SG PL
ALL *χnɛ́r-m :: *χnɛ́r-m-s
LOC *χnɛ́r-i :: *χnɛ́r-i-s (~ *χnɛ́r-i-n ?)
WFT *CCɛ́C-K :: *CCɛ́C-K-N

6.3 Case and number forms of PIE verb-like adjective stems


It can be doubted that Jasanoff 2009: 141 is correct by claiming that
“the shorter form *-os must have been the original PIE dat.-abl. pl. mor-
pheme”. The reason for doubting such an inference is simple. It is a com-
parative fact that Hittite very clearly reflects a former generalization of the
plural case suffixes throughout the various stem formations: cf. nom. -es,
acc. -us, gen. -an ~ -as, dat.-loc. -as. Differently from Vedic, an Anatolian
language like Hittite is not a perfect candidate to preserve a specific plural
ending in its original environment. The nom. pl. c. -es for instance, has
been generalized to the Hittite a-stems beyond its presumable i-stem (or
athematic) origins, cf. Hittite (OH) nom. pl. c. laales ‘tongues’ (la-a-le-eš
OS, nom. sg. la-a-la-aš). Therefore, the Proto-Anatolian dative plural suffix
*-os (cf. Hittite antuhsas (an-tu-uḫ-ša-aš OH/MS) ‘hominibus’, uddanas ‘re-
bus’, etc.) can go back to almost any PIE nominal stem formation. At first
sight, *-os looks more like an ending with a thematic o-stem origin.
I hereby suggest that the Vulgar Pre-IE nominal o-stems (e.g. *u̯l ̥́kwo-
‘wolf’) did not go back to PIE proper nouns like *χnɛ́r- ‘man’, etc. Instead, I
think that this nominal inflectional type originally was an attributive ad-
jective inflectional type. For instance, the Vulgar Pre-IE noun for ‘wolf’
obviously only became a proper noun via retraction of the accent to the
initial syllable. Thus, *u̯l ̥́kwo- obviously reflects a PIE adjective (and not a
noun) stem *ulkwɔ́- ‘dangerous, wolf-ish’. It is a relevant fact that this ad-
jective is still preserved as such by Vedic a-vr̥ká- ‘not harming, un-
dangerous, un-wolf-wish, protecting from harm or enemies.’
It is internally evident that the PIE o-stem inflectional type originally
did not distinguish between “strong” vs. “weak(est)” forms. As a parallel to
*dɦgɦɔ́m, *dɦgɦmɔ́ (both derived from *dɦɛ́gɦɔm) it is reasonable to infer that
this inflectional type also had an “endingless” locative-allative form and an
“endingless” benefactive-purposive form (these case forms were identical
then) besides the regular one(s) in *-i, e.g. *ulkwɔ́ ~ *ulkwɔ́i.
In Vulgar Pre-IE the new nom. pl. forms (*u̯l ̥́kwoːs ~ *u̯l ́k ̥ woi) were mor-
phologically analyzed as containing the suffix *-es (e.g. of *χnér-es). This
was done in parallel to the parsing of the new dat. sg. *u̯l ̥́kwo- + *-ei
(*u̯l ̥́kwoʔei) and a new abl. sg. *u̯l ̥́kwo- + *-ed (*u̯l ̥́kwoʔed).
However, the length of the stem vowel of the later nom. pl. forms did
not originate from a suffixation of *-es. Instead, in PIE the long vowel was
simply the realization of the addition of *_V_ (*_ɛ_) to the vowel melody
template:
ROLAND A. POOTH 34

FIGURE 4.17.1.
SG PL
ABS *ulkwɔ́ :: *ulkwɔ́_ɛ → *ulkwɔ́ː
ERG *ulkwɔ́_s :: *ulk ɔ́_ɛ_s → *ulkwɔ́ːs
w

WFT *CCCɔ́_K :: *CCCɔ́_ɛ_K → later reanalyzed as *-óʔes

Thus parallelly to the PIE proper nouns, the extended core cases were
marked by addition of the plural suffix *-s. In addition, the long vowel of
the core case forms was analogically transferrable.

FIGURE 4.17.2.
SG PL
ALL *ulkwɔ́_m :: *ulkwɔ́_m-s ~ *ulkwɔ́ː_m-s
LOC *ulkwɔ́_i :: *ulkwɔ́_i-s ~ *ulkwɔ́ː_i-s
WFT *CCCɔ́_K :: *CCCɔ́_K-N

As just mentioned, I suggest that this inflectional type included an “end-


ingless” locative-allative and benefactive-purposive form. Its corresponding
plural forms were thus marked by the plural suffix *-s.

FIGURE 4.17.3.
SG PL
LOCZ *ulkwɔ́_Ø :: *ulkwɔ́_Ø-s ~ *ulkwɔ́ː_Ø-s
BENZ *ulkwɔ́_Ø :: *ulkwɔ́_Ø-s ~ *ulkwɔ́ː_Ø-s
WFT *CCCɔ́_K :: *CCCɔ́_K-N

The plural form outcome simply was *ulkwɔ́s ~ ~ *ulkwɔ́ːs yielding a


Vulgar Pre-IE dat. pl. *u̯l ̥́kwos ~ *u̯l ̥́kwoːs (which existed besides the new
allomorphs *u̯l ̥́kwobɦi(s) ~*u̯l ̥́kwoibɦi(s) ~ *u̯l ̥́kwoibɦi(n) or the likes). Note
that the Vulgar Pre-IE allomorphy of *u̯l ̥́kwo- ~*u̯l ́k ̥ woi- before the segment
*-b i(s) clearly points towards an allomorphy *ulk ɔ́ ~ *ulkwɔ́i. Thus, yes, I
ɦ w

infer that the Proto-Anatolian dat. pl. forms in *-os and Hittite -as ulti-
mately go back to such “endingless” locative-allative and benefactive-
purposive forms (glossed LOCZ and BENZ in the paradigm below).
I finally suggest that the PIE adjective inflection originally did not par-
ticipate in all the grammatical distinctions that PIE proper nouns were
coded for. As already mentioned, the adjective lacked the distinction of
“strong” vs. “weak(est)” case forms. It perhaps displayed only optional
number marking in PIE:
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 35

FIGURE 4.18. The PIE verb-like adjective inflection

d
SG PL ASS COL
optional? optional?
w
ABS *ulk ɔ́ *ulkwɔ́ː *ulkwɔ́ʔ *ulkwáχ 27 (cf. *nɛ́uaχ)
ERG *ulkwɔ́s *ulkwɔ́ːs
ALL *ulkwɔ́m *ulkwɔ́(ː)ms
LOC *ulkwɔ́
Z
*ulkwɔ́(ː)s c *ulkwɔ́ʔ *ulkwáχ
LOCi *ulkwɔ́i *ulkwɔ́(ː)is *ulkwɔ́iʔ *ulkwáχi
SOC *ulkwɔ́ʔ *ulkwáχʔ
ABL *ulkwɔ́s a *ulkwáχs e
PAR *ulkwɔ́m b *ulkwáχm e
BENZ *ulkwɔ́ c *ulkwáχ
BENi *ulkwɔ́i *ulkwáχi e
COM *ulkwɔ́ʔ *ulkwáχʔ

Additional remark to this figure:


a
Cf. Hittite gen. sg. -as = nom. sg. c. -as. It is internally evident that the
gen. sg. case form was later remodelled to Vulgar Pre-IE *u̯l̥kwósi̯o ~*u̯l̥kwési̯o
~ *u̯l̥kwóso ~ *u̯l̥kwéso. It is further internally evident that the abl. sg. case
form *u̯l̥kwóʔed was a post-PIE innovation (NB. these are the Vulgar Pre-IE ad-
jective forms of *u̯l̥kwó- ‘dangerous, wolf-ish’ ―not the ones of the noun for
‘wolf’).
b
Cf. Archaic Latin gen. pl. seruom = acc. sg. m. seruom.
c
It must have been the suffixless benefactive-puroposive form that was ex-
tended by *-ei yielding Vulgar Pre-IE *u̯l̥kwóʔei (with hiatus). This strengthens
the reconstruction of a corresponding plural form *u̯l̥kwós ~*u̯l̥kwóːs (> Hittite
-as) (see the discussion above).
d
I think that the proper PIE parent language did not display plural forms
belonging to the *ulkwáχ or the *nɛ́uaχ inflectional type(s). The comparative
evidence clearly leads to the conclusion that the respective IE plural forms are
Vulgar Pre-IE parallel, but relatively independent innovations. I suggest that
PIE *ulkwáχ and *nɛ́uaχ simply were corresponding COLLECTIVE-PLURATIVE
forms of *ulkwɔ́ and *nɛ́uɔ, respectively. These specific COLLECTIVE-PLURATIVE
forms were used to code optional agreement for “nominal aspect” with
collective and plurative (and thus inanimate) head nouns and also abstract
action head nouns (e.g. *ʔɛ́r(i)- (inanim.) ‘coming together’) (and with nouns
like *náχu ‘boat, it is used to swim’ for some other reason28), e.g. *uɔ́ːkw
*ulkwáχ ~ *ulkwɔ́ ‘dangerous language, speech, formula’ (→ Vulgar Pre-IE
feminine *u̯óːkws *u̯l̥kwā ‘id.’) vs. *uɔ́kw *ulkwɔ́ ‘dangerous (single) word’ (→

27
This type was not identical in meaning to the type *ulkwíχ ‘belonging to the group of the
dangerous one’ or *ulkwíʔ ‘belonging to the company of the dangerous one’ which may
point to a prior PIE root noun or i-stem *uɛ́lkw- (*uɛ́lkwi-) ‘dangerous agent (animate)’
and ‘dangerous action (inanimate)’.
28
Maybe because a *náχu ‘boat’ was not owned and used by a single person, but belonged
to the clan and was thus owned and used by a group of people. Perhaps the collective-
plurative adjectives were often used to classify the given head noun like this.
ROLAND A. POOTH 36

Vulgar Pre-IE masculine *u̯ókws *u̯l̥kwós ‘id.’). I will return to the difficult topic
of the emergence of the post-PIE feminine gender elsewhere; see the articles in
Neri & Schuhmann 2014.
e
These forms were subject to extensive remodelling in Vulgar Pre-IE.

Furthermore, there is a more or less relevant typological implication of


the finding that PIE was of the agglutinating (plus transfixing and tem-
platic) morphotactic type. Adjectives sometimes lack case and number
(and gender) agreement with the given head noun in languages of this type
(e.g. in Modern Turkish).29 Therefore, perhaps even the superordinate case
marking, that is, the suffixes *-s, *-m, *-i, *-ʔ were absent from the para-
digm of these adjective stems in an early stage of PIE. Maybe we can only
safely reconstruct two forms *ulkwɔ́, *nɛ́uɔ and *ulkwáχ, *nɛ́uaχ, respec-
tively, for this particular class of adjective stems PIE.30
Last but not least, it is important to recognize that adjective stems of
this class morphologically exactly match 3rd person singular and 3rd person
collective detransitive, that is, “protomiddle” verb forms (e.g. *gwɦnɔ́, *stɛ́uɔ
and *gwɦnáχ, *stɛ́uaχ, etc.). This morphological identity is definitively not
just due to accident.
I have thus decided to speak of a class of PIE verb-like adjectives.31
Verb-like (vs. noun-like) adjectives are attested in many languages around
the world, cf. Dixon 2010 (see the chapter on adjectives).

6.4 Number forms and marking of PIE inanimate nominal stems


It is typologically most plausible to infer from the comparative and in-
ternal evidence that the inanimate nominal stems were “set nouns” and
thus transnumeral (cf. Rijkhoff 2002: 54). These nominal stems could take
an optional collective number marker *-χ and an optional associative
marker *-ʔ.
A paradigm of a PIE inanimate nominal stem, for instance *ɠɔ́nu ‘knot,
knotty joint, angle; knee’ is given in the following figure.

29
Swahili (another strongly agglutinating language) has prefixal class agreement of adjec-
tives with the given head, but adjectives of Arabic origin remain uninflected.
30
Thus also Kortlandt art269e.pdf, p. 4: “The Proto-Indo-European thematic paradigm was
probably uninflected [...]”.
31
PIE also had noun-like adjective stems. These were internally derived from underlying
proper nouns, e.g. *ɗɔrú- ‘wooden’ from *ɗɔ́ru- ‘wood’, etc. PIE also had a class of “delo-
cative” adjective stems, e.g. *ɗɔruí- ‘in wood, wooden’ from loc. *ɗɔ́ru-i ‘in wood’, etc.
But I will return to this interesting topic elsewhere.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 37

FIGURE 4.19.

TRANSNUMERAL ASS COL/PL


optional
ABS *ɠɔ́nu *ɠɔ́nuʔ *ɠɔ́nuχ
ERG GEN *ɠɔ́nus
LOC *ɠɔ́nui *ɠɔ́nuiʔ *ɠɔ́nuiχ
SOC *ɠɔ́nuʔ
ABL *ɠɛ́nuɔs
PAR *ɠɛ́nuɔm
BEN *ɠɛ́nuɔi
COM *ɠɛ́nuɔʔ

In PIE, the sociative-associative form, e.g. *ɠɔ́nuʔ had both a sociative


case reading ‘with joint(s), with knee(s)’ and an absolutive case and asso-
ciative number reading ‘with joint(s), knee(s) & co., accompanied joint(s),
knee(s)’ including ‘joint & co., knee & co.’ and thus also ‘two joints, two
knees’ with a subordinate dual reading (but not a dual meaning).
The corresponding optional locative form *ɠɔ́nuiʔ had a meaning
‘at/to/with joint(s) &co., knee(s) & co.’.
As I have argued for more detailed elsewhere, inanimate locative forms
were used in “O function” (cf. Dixon 1979, 2010) in the PIE antipassive
construction which was later reanalyzed as a new transitive construction
from PIE to Vulgar Pre-IE (cf. Pooth manuscript a). Thus inanimate locative
forms such as *ɠɔ́nuiʔ were reanalyzed as new neuter “object” forms, that
is, neuter accusative forms. Due to the lack of old allative-dative forms and
the use of old PIE absolutive forms in both the accusative and nominative
function, there was a principle homonymy of neuter accusative and nomi-
native forms in Vulgar Pre-IE. These two form were thus additionally used
as neuter nominative forms by that time.
As mentioned above, I suggest that the PIE forms with associative num-
ber reading (e.g. ‘knee & co.’) developed into Vulgar Pre-IE nom.-acc. dual
number forms (→ ‘two knees’). Such a grammaticalization path
ASSOCIATIVE ‘referent & co., referent WITH company’ → DUAL ‘two refer-
ents’ is functionally very plausible and I would bet that it is attested any-
where crosslinguistically.32
It follows that both the PIE locative associative form (e.g. *ɠɔ́nuiʔ) and
the PIE absolutive associative33 form (*ɠɔ́nuʔ) should have developed to

32
It is not listed in Heine & Kuteva 2000, but this does not mean that such a grammatical-
ization path does not exist.
33
There was no ergative form of this number in PIE. The associative and collective num-
ber forms were treated as if they were inanimate stems. The PIE inanimate gender as-
signment was triggered by the features inanimate plus nonindividual. Nonindividual
concepts including animate associations and groups (e.g. *pɛ́ːku ‘group of domestic ani-
mals, cattle’) received inanimate gender assignment (cf. Pooth manuscript a).
ROLAND A. POOTH 38

Vulgar Pre-IE nom.-acc. dual allomorphs (*ɠónuʔ ~ *ɠónu̯iʔ). In the later


Vulgar Pre-IE variant cluster, the second then ousted the first form and *-iʔ
(> *-iː or *-iʔe) became the regular Vulgar Pre-IE neuter nom.-acc. dual
ending, cf. Greek ὄσσε = OCS oči ‘both eyes, two eyes’.
There is perhaps no better explanation for why the IE neuter nom.-acc.
dual forms go back to forms with two agglutinating suffixes *-i-ʔ with the
same structure as *-i-s.

7 New Vulgar Pre-IE oblique forms


Judging from the comparative IE evidence and following part of the ar-
gumentation and reconstruction provided by Jasanoff 2009, it is internally
evident that the new Vulgar Pre-IE dative case, the comitative-instrumental
case and the ablative case―but not including its genitive reading―and
also the locative case were supplied by new allomorphic by-forms not be-
fore a post-PIE period.
Cf. Jasanoff 2009: 139: “The absence of grammaticalized bh-forms in
Anatolian suggest that the creation of bh-cases was a later development
than the separation of Anatolian [...]” (but see below).
These Vulgar Pre-IE allomorphic by-forms and their endings must have
looked like as follows (I drop the accent):
(a) instr. and dat.-abl. *χnr̥-bɦi(ː)(s) ~ *χnr̥-bɦo(s) ~ *χnr̥-bɦi̯o(s) (the
latter is an obvious pleonasm; a possible combination is *χnr̥-bɦi + *-os cf.
Jasanoff 2009) or simply *-bɦi- plus the coda of *-bɦo(s);
(b) allomorphs with a segment *-m-: *χnr̥-mi(ː)(s) ~ *χnr̥-mo(s);
(b) loc. (pl.) *χnr̥-su (maybe already *χnr̥-si).

NB. Excursus on Old Phrygian. This language maybe had both endings -pi
and -po. The inscription from Vezirhan/B-05 4b shows a form patorpi
(<*pχtr̥bɦi) in eśtat patorpi ke ... andati where andati (-dati< *dɦéʔti) is compa-
rable to Boeoitian Greek ἀνεθε̄.
N = Neumann 1997: 18, B = Brixhe 2004: 65 (cf. Gorbachov 2008: 92, Sowa 2008: 83)
(emphasis mine):
4 N: panta vebras aḍunpoṣkey eśtat patorp ike ..... ạṇ dati
B: pạntạ vebrạṣ ạdun p(?)os key estṣt pator.(?)ike[...]ẹ[..]ạndạti [cf. Greek ἀνεθε̄]
mine: estat patorpi ke ..... andati
estat ‘he put it (sc. sint imenan) there’,
patorpi ‘by the fathers’ (?)
ke ‘and’
andati :: ἀνεθε̄, cf. Brixhe 2004, s.v.

Two stancas later Vezirhan 6b shows andopo (+andorpo?) postois which


looks pretty much like a case form of Old Phrygian anar ‘man’ with a case
ending -po (<*χnr̥bɦo). This form is directly followed by what may be an ad-
jective posto- ‘?’ with an ending -ois (cf. Greek -οις). -ois is also attested in
Üyücek/B-04 brạterạis pạtriyiọ(?)is(?)k[e] (cf. Brixhe 2004: 39). andopo is pre-
ceded by what looks like a 3rd sg. aorist injunctive mediopassive form of a verb
‘do, make, produce’, namely ganato ‘is done’.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 39

6 N: nidus .akaliyay karatu panato andop opostois klaniv


B: nidus ad kaliyay karatu panato andopopostois klam(?)iv[..?]
mine: ni dus ad kaliyay karatu ganato andopo postois klaniv
ni = prohibitive or indefinite particle, dus :: Vedic duṣ- ‘bad’? ad = Latin ad, ad
kaliyay ‘to Kallias’ (dat. sg. m.), karatu gen. pl. (?),
ganato < post-PIE *g̑ń̥h1-to ‘is done’ (?);
klaniv :: Vedic śr̥nú ‘listen’ (??) < *k̑l̥néu ~ *k̑l̥nú (??)

The symbols ¾ p and ± g are almost identical in this inscription. A reading


OTANA± ganato with ± is possible. Compare B-05 5b vay niptiyay daker ‘for
their own niptiya (family?) they have made it/it has been made’ and karatu
engnatuś ‘to the ones born among the Karato’ (ÀYTAN±NE, with N± as against
Neumann’s “enpsatuś”). Old Phrygian thus maybe had ganato vs. -gnato- (paral-
lel to Greek). I suggest that andopo postois was a NP in the instrumental or
dative-ablative plural. (Note that the d-epenthesis +andorpo besides anar
would also run parallel to the one in Greek ἀνδρ-). End of excursus.

It is internally evident from Mycenean Greek wi-pi ‘with violence, vio-


lently’ (cf. Jasanoff 2009: 144) and other such relics that these new
oblique by-forms were once transnumeral before developping into narrow
plural forms. While thus being functionally narrowed to specific plural
markers, these new endings were interpreted as extendable by the plural
markers *-s or *-n (cf. Homeric -φι(ν)). The vowel of the new narrow plural
ending sporadically underwent a plural lengthening (cf. Proto-Balto-Slavic
*-miːs (≡ *-mīs) with long vowel).
As mentioned above, the Vulgar Pre-IE allomorphy of *u̯l ̥́kwo- ~*u̯l ́k ̥ woi-
before the segment *-b i(s) points towards a PIE allomorphy *ulk ɔ́ ~
ɦ w

*ulkwɔ́i.
The comparative evidence thus strengthens the inference that these seg-
ments *-bɦi-, *-bɦo-, *-mi-, *-mo- were once attached to ...
(a) ... underlying transnumeral or singular locative-allative forms of o-
stems (cf. *u̯ĺ̥kwo(i)- + *-bɦi(s) > Proto-Indo-Iranian inst. pl. m. *u̯ŕ̥kaibɦiš
~ *u̯ŕ̥kaːiš > Vedic vŕ̥kebhiṣ ~ vŕ̥kaiṣ, etc.),
(b) ... an accent-reduced and vowel-reduced (in this case basically
vowel-less) athematic stem *χnr-, etc.

7.1 PIE compounds with a pronominal second member


Thus, as already suggested in a similar way by Jasanoff 2009: 139ff.
part of these new singular and plural forms presumably were grammatical-
ized from original PIE compounds with a pronominal second member. A
compound with a pronominal second member was perfectly grammatical
in PIE. The first member of that compound must have lost its transfix or
had a transfix that was reduced by one vowel.
ROLAND A. POOTH 40

FIGURE 4.20.
*χnt-bɦí “sidewise”, cf. Jasanoff 2009: 139 (cf. *χánt- ‘front, the (other)
side to the front’), thus literarily ‘side-with, side-by’
*χnr-bɦí literarily ‘he in man-company, he man-with’

NB. Let me add to the proposal of Jasanoff 2009 that *bɦí was not simply a
pronominal adverb in my view, but had full independent pronoun status in
PIE. It was structured like PIE *smí ‘the one, the one associated, the homoge-
neous one, the same one, the identical one’ and ‘1’, further well-known exam-
ples are PIE *ʔí, *sí, *tí, *kí, ..., *kwí ‘who, anyone’, etc. I think that *bɦí had a
meaning close to literarily ‘he in company, he with x’. The first element before
the semi-vowel *-í obviously was either a deictic-demonstrative pronominal
root (*s-, *t-, *k- ‘here’, etc.) or a pronominal root with classifying or classi-
fyer-like function. *bɦɔ́- was simply the attributive form corresponding to *bɦí
(which made up a full NP). This functional distinction of independent pro-
nouns *kwí-, *bɦí-, etc. vs. attributive pronominal forms *kwɔ́-, *bɦɔ́-, etc. was
among the brilliant ideas of W. Cowgill (personal communication with D.
Ringe, cf. Ringe 2006). In Vulgar Pre-IE these forms must have been subject to
abundant functionally reinterpretation, because what may once have been a
PIE classifyer system was completely lexicalized and given up in Vulgar Pre-IE.
*mí presumably was another independent pronoun with a similar meaning
literarily ‘he associated’, compare *smí ‘the one alike/associated, the homoge-
neous one, the same one, the identical one’ and ‘1’. Again, *mɔ́- was just its
attributive form.
Thus, a plausible ultimate source and starting point for a grammaticali-
zation of the given pronominal forms to Vulgar Pre-IE secondary locative-
allative and new dative and also ablative case suffixes are old PIE com-
pounds with a pronominal second member (e.g. *bɦí and the likes). These
pronouns presumably once fully agreed in case marking (but not necessar-
ily in number marking) with the given head referent (gloss: ATTR = at-
tributive):

(4.29) a. *pχtɛ́r-s *ɠwɛ́mt *χnr-bɦí-s


father-ERG came man-he.in.company-ERG
lit. ‘father came, he (agent) in man-company’
b. *pχtɛ́r-s *χnr-bɦɔ́_s *ɠwɛ́mt
father-ERG man-in.company.ATTR_ERG came
lit. ‘father, the man-accompanied one (attributive), came’

c. *pχtɛ́r *gwhnɔ́ *χnr-bɦí


father.ABS was slewn man-he.in.company-ERG
lit. ‘father was slewn, he in man-company’

7.2 A PIE construction with loc. + pronoun


An additional plausible source is a PIE construction including an “end-
ingless” locative or *-i-locative (e.g. *ulkwɔ́(i)) which preceded the referen-
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 41

tial pronoun *bɦí- as its adjunct (modifyer). This construction must have
had a very similar function like the one given above:

(4.30) a. *pχtɛ́r-s *ɠwɛ́mt *ulkwɔ́(i) *bɦí-s


father-ERG came dangerous.LOC he.in.company-ERG
‘father came, to-the-dangerous-one-he-in-company’
→ ‘father came he (was) accompanied to (with) the dangerous one’
b. *pχtɛ́r-s *ulkwɔ́(i) *bɦɔ́-s *ɠwɛ́mt
‘father, to the dangerous one the accompanied one, came’
→ ‘father came accompanied to (with) the dangerous one’
c. *pχtɛ́r *gwhnɔ́ *ulkwɔ́(i) *bɦí
father.ABS was slewn dangerous.LOC he.in.company-ERG
lit. ‘father was slewn, to the dangerous one he in company’
→ ‘father was slewn accompanied to (with) the dangerous one’

After all, I suggest that the new oblique case forms ultimately emerged
via syntactic reanalysis from older PIE constructions including pronominal
forms that agreed in case (but maybe not in number) with the topical ref-
erent. I thus suggest that―in this particular case―the *-s was not necessar-
ily the plural suffix *-s. But it was ‘reanalyzable’ as a plural suffix, because
it was formally identical to the old plural number suffix *-s (which should
have had an allomorph *-n parallel to the verbal plural marking, cf. Pooth
manuscript a, b, c).
Finally, it is possibile and plausible that PIE *bɦí already had lost its full
pronoun status after PIE broke up and before being grammaticalized the
way as it has just been suggested. The original pronoun ‘he in company, he
WITH x’ was presumably reinterpreted as a simple Vulgar-IE adverbial
particle *bɦí ~ *=bɦi ‘in company, with, by, at’. The old PIE pronoun sys-
tem must have undergone abundant reinterpretations.
NB. Thus, the Proto-Anatolian instr. endings *-(i)di ~ *-(i)d(a) (> Hit-
tite instr. -it ~ -et ~ -d(a) ~ -t(a)) presumably originated from two
sources: (a) PIE compounds with pronominal second member*-dɦí-, *-dɦɔ́-
(parallel to *-bɦí-, *-bɦɔ́- and *-mí-, *-mɔ́- above) and (b) a PIE locative con-
struction like the one given above, cf. Hittite instr. kissar-it ~ kissar-ta
‘with hand(s)’. Note that these endings cannot go back to plain *tV, cf.
Hoffner & Melchert 2008: 77f.

FIGURE 4.21.
*X(-i) dɦí literarily ‘X-LOC he.there (?)’ → ‘at/by/with X(s)’
*gɦesr-dɦɔ́ literarily ‘hand-there.ATTR’ → ‘at/by/with hand(s)’

Such a parallel innovation can further strengthen the view that pre-
Proto-Anatolian once also participated in forming some innovated oblique
by-forms, but did not participate in their functional narrowing to plural
number forms. It seems to be impossible to reconstruct an “Inner-IE” iso-
ROLAND A. POOTH 42

gloss. Like all other branches, Proto-Anatolian also separated after partici-
pating in forming new oblique by-forms. The Proto-Anatolian instrumental
in *-(i)di ~ *-(i)d(a) just adds a new Vulgar Pre-IE oblique by-form to the
ones ending in *-bɦi-, *-bɦɔ- or *-mi-.

7.3 A PIE construction with loc. (pl.) + pronominal adverb


However, a new Vulgar Pre-IE locative plural form, e.g. *u̯ĺ̥kwoi(s)su
most probably emerged from a different construction, that is, one with
juxtaposition of the PIE pronominal adverb *sú ‘thus, in this way, at this
way, at this place’ or the like (cf. *tú ‘there, thus’, *kwú ‘where, why, how’,
cf. Vedic kúha ‘where’, etc.).

(4.31) *pχtɛ́r-s *ʔɛ́st *ulkwɔ́is *sú


father-ERG was.there dangerous.LOC at this place
‘father was/sat there, at the dangerous ones, at this/that place’

Hence → *u̯ĺ̥kwois *=su → *u̯ĺ̥kwoi(s)su (> Vedic vŕ̥keṣu). *-su then be-
came a new locative plural suffix and was generalized to other stems, e.g.
Vulgar Pre-IE *χnŕ̥-su. But I admit that there may be other plausible possi-
bilities as well.

7.4 Conclusion
Let me sum up: Whatever be their actual source, I hereby firmly refuse
to backproject any of these new oblique plural forms including their end-
ings to the proper PIE protolanguage. Claiming that these forms belonged
to the proper case system and to each nominal paradigm of the unitary PIE
parent language is an obvious anachronistic mistake.
The ones that once spoke this language thus used similarly looking con-
structions that preceded these new by-forms, but they would not have used
forms like “PIE *u̯ĺ̥kwoibɦis, *u̯ĺ̥kwoisu, ...” in their later functions. They
would have used the respective locative-allative plural forms (e.g. PIE
*χnɛ́ris *ulkwɔ́(ː)(i)s ‘at/to/by/... dangerous men’) instead.

8 Final remark
Above I have provided a fundamentally new reconstruction and an inter-
nal/synchronic morphological analysis of the Proto-Indo-European case sys-
tem.
The traditional description of the case system of the protolanguage as pre-
sented, for instance, by Tichy 12000, 22004, 32009 and by the contemporary
handbooks suffers from its obvious linguistic anachronism. The ‘old’ model
even cannot be termed a coherent reconstruction of the various case systems
(!) of the Vulgar Pre-IE dialects or variants.
Reconstructions like “PIE *u̯ĺ̥kwoibɦis, *u̯ĺ̥kwoisu, ...” are nothing but pseudo-
reconstructions. They can only have the status and value of phonological trans-
ponates or equation formulas, that is, forms that represent and symbolize the
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 43

given IE equations. The naїve comparative practice of backprojecting these


formulas onto an alleged prior chronological layer solely by undoing the given
detectable sound laws, however, has only yielded a strange kind of scientific-
artificial pseudo-protolanguage―a language that has little to do with proper
Proto-Indo-European.
“Proto-Indo-European Proper”,34 however, must have been a synchronically
relatively unitary and less divergent, perhaps even, say, ‘standardizable’ lan-
guage, spoken presumably by a relatively small number of people somewhere
between Eastern Europe (Ukraine, Southern Russia) and Western Central Asia
before breaking up into an incoherent and increasingly less mutually under-
standable and diverging (and ‘non-standardizable’) Vulgar Pre-Indo-European
variant or dialect cluster.
It has to be recognized that it is far more plausible to infer that the given
forms *u̯ĺ̥kwoibɦis, *u̯ĺ̥kwoisu, ... did not exist in this PIE protolanguage.35
Let me conclude as follows. It is most plausible ...
(a) ... that despite the traditional comparative claim the Proto-Indo-
European proper parent language did not display nominative-accusative align-
ment; instead it was of the “active” or “fluid-S” type;
(b) ... that PIE was an “exotic” non-Indo-European-like language and had a
very different (transfixing-templatic and agglutinating) morphotaxis than its
post-PIE fusional daughter languages (cf. Pooth 2004a, 2009b);
(c) ... that PIE exhibited many non-Indo-European-like grammatical catego-
ries, morphosyntactic distinctions and syntactic constructions (which are not
preserved as such in the younger IE languages):
(i) PIE had a class of RELATIONAL nominal stems opposed to NONRELATIONAL
ones; (ii) PIE had “protomiddle nominal stems”, e.g. *nɔ́ɠwt ‘night’ (vs. agen-
tive-unmarked *nɛ́ɠwt ‘dawn’) which are typologically comparable to the ones
of Temiar, cf. Benjamin 2011, etc.

Representation of reconstructed forms


Reconstructed “PIE Proper” word forms are coloured blue and additionally
marked by the conventional preceding asterisk *. The I.P.A. symbols ʔ, χ, ʕw
correspond to the algebraic h1, h2, h3. The vowels are represented as ɛ, ɔ, ɛː, ɔː
≡ e, o, ē, ō. Vowel length is symbolized by ː (I.P.A.). The symbols for i, u, m, n,
r, l remain unchanged (i, u, m, n, r, l), but the redundant diacritics are om-
mited. The consonant phonemes p, ɓ, bɦ, t, ɗ, dɦ, k, ɠ, gɦ, kw, ɠw, gwɦ, q, ʛ, ɢɦ
(besides s) are represented here the way as has lately been suggested by Küm-
mel 2012.
Reconstructed Vulgar Pre-IE dialectal forms are not coloured blue and are
represented by the traditional symbolization, except that I use the I.P.A. sym-
bols ʔ, χ, ʕw instead of the algebraic h1, h2, h3 and k, g, gɦ, q, ɢ, ɢɦ instead of k,̑
g̑, g̑h, k, g, gh―and that vowel length is also symbolized by ː (I.P.A.).

34
For this term cf. Ringe 2006: 5-6.
35
This also holds true, in my view, for the entire verbal “thematic” (*-o/e-) conjugation,
cf. Meillet 1931.
ROLAND A. POOTH 44

References
Aldridge, E. 2011: “Antipassive in Austronesian Alignment Change”, Manuscript,
http://faculty.washington.edu/eca1/pdf/Alignment.pdf
Anthony, D. 2007: The horse, the wheel and language. How bronze-age riders from the
Eurasian Steppes shaped the modern world. Princeton & Oxford.
Beekes, R. S. P. 1995: Comparative Indo-European Linguistics. Amsterdam & Phila-
delphia.
— 2011: Comparative Indo-European Linguistics. Second edition. Revised and cor-
rected by Michiel de Vaan. Amsterdam & Philadelphia.
Benjamin, G. 2011: “Deponent Verbs and Middle-Voice Nouns in Temiar”, Austroa-
siatic Studies: papers from ICAAL4. Mon-Khmer Studies Journal Special no. 2.
Ed. by S. Srichampa & P. Sidwell. Dallas, Salaya & Canberra, 11-37.
Besniers, N. 1987: “An autosegmental approach to metathesis in Rotumam”, Lin-
gua 73, 201-223.
Blake, B. J. 2001: Case. Second edition. Cambridge (1st edition 1997).
Bossong, G. 1999: “Le marquage de l’expérient dans les langues de l’Europe”, Ac-
tance et valence dans les langues de l’Europe, Ed. by J. Feuillet. Berlin & New
York, 259-294.
Brixhe, C. 2004: “Corpus des inscriptions paléo-phrygiennes. Supplement II”,
Kadmos 43, 1-130.
Brugmann, K. 1916: Grundriss der vergleichenden Grammatik der indogermanischen
Sprachen. 2nd ed. Strassburg.
Bybee, J., R. Perkins & W. Pagliuca 1994: The Evolution of Grammar. Tense, Aspect,
and Modality in the Languages of the World. Chicago & London.
Clackson, J. 2007: Indo-European Linguistics: an Introduction. Cambridge.
Delbrück, B. 1867: Ablativ localis instrumentalis. Berlin.
Dixon, R. M. W. 1977: A Grammar of Yidiɲ. Cambridge.
— 1979: Ergativity. Cambridge.
— 2000: “A typology of causative: form, syntax and meaning”, Changing Valency.
Ed. by R. M. W. Dixon & A. Y. Aikhenvald. Cambridge, 30-83.
— 2010, 2012: Basic Linguistc Theory. Vol. I & II 2010, Vol. III 2012. Oxford.
Donohue, M. & S. Wichmann 2008 (eds.): The Typology of Semantic Alignment. Ox-
ford.
Dowty, D. 1991: “Thematic proto-roles and argument selection”, Language 67,
547-619.
Drinka, B. 1999: “Alignment in Early Proto-Indo-European”, Language Change and
Typological Variation. In Honor of Winfred P. Lehmann on the Occasion of his
83rd Birthday. Ed. by C. F. Justus & E. C. Polomé. Washington DC, Vol. II,
464-500.
Eichner, H. 1972: “Die Etymologie von heth. mēḫur“, Münchener Studien zur
Sprachwissenschaft 31, 53-107.
— 1985: “Das Problem des Ansatzes eines urindogermanischen Numerus ‘Kollek-
tiv’ (‘Komprehensiv’)”, Grammatische Kategorien, Funktion und Geschichte: Ak-
ten der VII. Fachtagung der Indogermanischen Gesellschaft, Berlin 20.-25. Feb-
ruar 1983. Ed. by B. Schlerath & V. Rittner. Wiesbaden, 134-169.
Fortson IV, B. W. 2004: Indo-European Language and Culture. Oxford & Malden MA.
Fox, Anthony 1995: Linguistic Reconstruction. Oxford.
Geldner RV = K. F. Geldner 1923ff. Der Rig-Veda. Göttingen. Nachdruck 2003,
Cambridge, Ma. & London.
George, C. H. 2005: Expressions of Agency in Ancient Greek. Cambridge.
Givón, T. 2001: Syntax. Volume I & II. Amsterdam.
Gorbachov, Y. 2005: “Toward the Interpretation of the Old Phrygian Inscription
from Vezirhan”, Harvard Working Papers in Linguistics 11, 41-57.
Hale, K. 1985: “On Nonconfigurational Structures”, Harvard Studies in Korean Lin-
guistics. Proceedings of the 1985 Harvard Workshop on Korean Linguistics (Har-
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 45

vard WOKL-1985), July 12-13, 1985. Ed. by Susumu Kuno, John Whitman,
Ik-Hwan Lee, Young-Se Kang. Seoul, 62-70.
Hale, M. 1987: “Notes on Wackernagel’s Law in the language of the Rigveda”,
Studies in Memory of Warren Cowgill. Papers from the fourth East Coast Indo-
European Conference. Cornell University June 6-9 1985. Ed by C. Watkins. Ber-
lin & New York, 38-50.
Haspelmath, M. & A. D. Slims 22010: Understanding Morphology. Second Edition.
London.
Heine, B. & T. Kuteva 2002: World Lexicon of Grammaticalization. Cambridge.
Hoffner, H. A. Jr. & H. C. Melchert 2008: A Grammar of the Hittite Language. Wi-
nona Lake.
Hopper, P. J. & S. A. Thompson 1980: “Transitivity in Grammar and Discourse”,
Language 56 (2), 251-299.
Jasanoff, J. H. 1979: “The position of the ḫi-conjugation”, Hethitisch und Indoger-
manisch. Hrsg. von Ernst Neu und Wolfgang Meid. Innsbruck, 79-90.
— 1998: “The thematic conjugation revisited”, Mír Curad. Studies in Honor of Cal-
vert Watkins. Ed. by Jay H. Jasanoff, H. Craig Melchert and Lisi Oliver. Inns-
bruck, 301-16.
— 2003: Hittite and the Indo-European Verb. Oxford & New York.
— 2009: “*-bhi, *-bhis, *-ōis: Following the trail of the PIE instrumental plural”,
Internal Reconstruction in Indo-European. Ed. by J. E. Rasmussen & T.
Olander. Copenhagen, 137-149.
Keydana, G. 2005: “Indogermanische Akzenttypen und die Grenzen der Rekon-
struktion”, Historische Sprachwissenschaft 118, 19-47.
— 2009: “Wackernagel in the Language of the Rigveda. A Reassessment”, Manu-
script, www.keydana.de → download_wackernagel_extended.pdf.
— 2011a: “Wackernagel in the Language of the Rigveda. A Reassessment”, His-
torische Sprachforschung 124, 106-33.
— 2011b: “Proterokinetische Stämme, Akzent und Ablaut”, Manuscript,
www.keydana.de → download_Keydana_Proterokinetische_Staemme_Akzent
_Ablaut[1].pdf.
Klaiman, M. H. 1992: “Inverse Languages”, Lingua 88, 227-61.
Klimov, G. A. 1974: “On the character of language of active typology”, Linguistics
131, 11–25.
— 1977: Tipologija jazykov aktivnogo stroja. Moscow.
Kloekhorst, A. 2008: Etymological Dictionary of the Hittite Inherited Lexicon. Leiden.
König, C. 2008: Case in Africa. Oxford.
Kortlandt, F. 1981: “1st sg. middle *-H2”, Indogermanische Forschungen 86, 123-
136.
— 1983: “Proto-Indo-European verbal syntax”, Journal of Indo-European Studies 11,
307-324.
— art269e: “An outline of Proto-Indo-European”, Manuscript, www.kortlandt.nl →
art269e.pdf.
Kümmel, M. J. 1996: Stativ und Passiaorist im Indoiranischen. Göttingen.
— 1998: “Wurzelpräsens neben Wurzelaorist im Indogermanischen”, Historische
Sprachforschung 111, 191-208.
— 2004: “Zur o-Stufe im idg. Verbalsystem”, Indo-European Word Formation.
Proceedings of the Conference held at the University of Copenhagen Oct. 20th-
22nd 2000. Ed. by J. Clackson and B. Olsen. Kopenhagen, 139-158.
Kurzová, H. 1986: “Typologie und die Rekonstruktion der indogermanischen Mor-
phosyntax”, Folia Linguistica 20, 49-86.
Lakoff, G. 1987: Women, Fire, and Dangerous Things: What categories reveal about the
mind. Chicago.
Lakoff, G. & M. Johnson 1980: Metaphors We Live by. Chicago.
Ledgeway, A. 2012: From Latin to Romance. Morphosyntactic Typology and Change.
Oxford.
ROLAND A. POOTH 46

Lehmann, W. P. 1974: Proto-Indo-European Syntax. Austin.


— 1995: Residues of pre-Indo-European Active Structure and their Implications for the
Relationships among the Dialects. Innsbruck.
— 2002: Pre-Indo-European. Washington DC.
Letuchiy, A. 2009: “Towards a typology of labile verbs: Lability vs. derivation”,
New Challenges in Typology. Transcending the Borders and Refining the Distinc-
tions. Ed. by A. Arkhipov & P. Epps. Berlin, 223-244.
Litscher, R. 2014: “Voraussetzungen für ein feminines Genus und Implikationen
für das Kategoriesystem des frühindogermanischen Nomens”, in Neri &
Schuhmann 2014: 137-165.
LIV = Rix et al. 2001.
Lubotsky, A. 1997: A R̥gvedic Word Concordance. 2 Volumes. Cambridge, MA.
Luraghi, S. 1987: “Reconstructing PIE as an Ergative language. A Test”, Journal of
Indo-European Studies 15 (3-4), 359-379.
— 2001a: “Syncretism and classification of semantic roles”, Sprachtypologie und
Universalienforschung 54 (1), 35-51.
— 2001b: “Some remarks on Instrument, Comitative, and Agent in Indo-
European”, Sprachtypologie und Universalienforschung 54 (4), 385-401.
Mallory, J. P. & D. Q. Adams 1997 (eds.): Encyclopedia of Indo-European Culture.
London.
Malzahn, M. 2010: The Tocharian Verbal System. Leiden & Boston.
Maniacky, J. 2002. Tonologie du ngangela. PhD thesis, INALCO, Paris.
Matasović, R. 2000: “Uses and misuses of typology in Indo-European linguistics”,
125 Jahre Indogermanistik in Graz. Ed by M. Ofitsch & C. Zinko. Graz, 327-
339.
— 2011a: “Latin pudet me, piget me, paenitet me and the active clause alignment in
PIE”, paper presented at the International Conference on Historical Linguistics,
Osaka, July 2011.
— 2011b: “Clause alignment in Proto-Indo-European”, Manuscript, http://
mudrac.ffzg.unizg.hr/~rmatasov/ClauseAlignmentInPIE.pdf.
McCarthy, J. 1979: Formal Problems in Semitic Phonology and Morphology. Doctoral
dissertation, MIT. Cambridge, MA.
— 1981: “A Prosodic Theory of Nonconcatenative Morphology”, Linguistic Inquiry
12, 373-418.
Meillet, A. 1931: “Caractère secondaire du type thématique indo-européen”, Bulle-
tin de la Société de Linguistique de Paris 32 (2) (numéro 96), 194-203.
— 1937: Introduction à l’étude comparative des langues indo-européennes. Paris.
Melchert, H. C. 2012 (ed.): The Indo-European Verb. Proceedings of the Conference of
the Society for Indo-European Studies, Los Angeles 13-16 September 2010.
Wiesbaden.
Neri, S. & R. Schuhmann 2014 (eds.): Studies on the Collective and Feminine in Indo-
European from a Diachronic and Typological Perspective. Leiden & Boston.
Neumann, G. 1997: “Die zwei Inschriften auf der Stele von Vezirhan”, Frigi e
Frigio. Atti del 1° Simposio Internazionale Roma, 16-17 ottobre 1995. A cura di
R. Gusmani, M. Salvini, P. Vannicelli. Rom, 13-32.
NIL = Wodtko et al. 2008.
Pinault, G.-J. 2003: “Sur les thèmes indo-européens en *-u- : dérivation et éty-
mologie”, Indogermanisches Nomen. Akten der Arbeitstagung der Indoger-
manischen Gesellschaft, Freiburg, 19. bis 22. Sept. 2001. Ed. by Eva Tichy,
Dagmar S. Wodtko, Britta Irslinger. Bremen, 153-188.
Pooth, R. A. 2000: “Stativ vs. Medium im Vedischen und Avestischen”, Historische
Sprachforschung 113, 88-116.
— 2001: “Studien zur frühurindogermanischen Morphologie I. ‘Stativ’, ‘Medium’
und ‘Perfekt’”, Historische Sprachforschung 114, 220-258.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 47

— 2004a: “Ablaut und autosegmentale Morphologie: Theorie der urindoger-


manischen Wurzelflexion”, Indogermanistik - Germanistik - Linguistik. Ed by.
M. Kozianka, R. Lühr & S. Zeilfelder. Hamburg, 401-471.
— 2004b: “Zur Genese der späturidg. thematischen Konjugation aus frühuridg.
Medialformen”, Indogermanische Forschungen 109, 31-60.
— 2009a: “Der urindogermanische Progressiv”, Protolanguage and Prehistory. Akten
der XII. Fachtagung der Indogermanischen Gesellschaft ... in Krakau. Ed. by R.
Lühr & S. Ziegler. Wiesbaden, 381-406.
— 2009b: “Proto-Indo-European Ablaut and Root Inflection”, Internal Reconstruc-
tion in Indo-European. Ed. by J. E. Rasmussen & T. Olander. Copenhagen,
229-254.
— 2011: “Die 2. und 3. Person Dual und das Medium”, Indogermanistik und Lin-
guistik im Dialog. Akten der XIII. Fachtagung der Indogermanischen Gesellschaft
... in Salzburg. Ed. by T. Krisch & T. Lindner. Wiesbaden, 473-83.
— 2012: “Zum Aufkommen transitiver Verben in frühen Vedischen am Beispiel
1
r̥”, The Indo-European Verb. Proceedings of the Conference of the Society for
Ind-European Studies, Los Angeles 13-16 September 2010. Ed. by H. Craig Mel-
chert. Wiesbaden, 267-84.
Rijkhoff, J. 2002: The Noun Phrase. Oxford.
Ringe, D. 2006: From Proto-Indo-European to Proto-Germanic. Oxford.
Rix et al. 2001 = LIV = Lexikon der indogermanischen Verben. Unter der Leitung
von H. Rix ..., 2. erw. und verbesserte Auflage. Wiesbaden 2001.
Rumsey, A. 1987: “The Chimera of Proto-Indo-European Ergativity”, Lingua 71,
297-318.
Sasse, H.-J. 1978: “Subjekt und Ergativität: Zur pragmatischen Grundlage primärer
grammatischer Relationen”, Folia Linguistica XII/3-4, 219-252.
— 1982: “Subjektprominenz”, Fakten und Theorien. Beiträge zur romanischen und
allgemeinen Sprachwissenschaft. Festschrift für Helmut Stimm zum 65. Geburt-
stag. Ed by S. Heinz & U. Wandruszka. Tübingen, 267-286.
— 1993: “Syntactic Categories and Subcategories”, Handbücher zur Sprach- und
Kommunikationswissenschaft 9.1. Syntax. Ed. by H. Steger & H. E. Wiegand.
Berlin & New York, 646-686.
— 1991a: “Aspect and Aktionsart: a reconciliation”, Belgian Journal of Linguistics 6,
31-45 [= Perspectives on Aspect and Aktionsart. Ed. by C. Vetters & W.
Vandeweghe].
— 1991b (ed.): Aspektsysteme. Arbeitspapier Nr. 14 (Neue Folge). Institut für
Sprachwissenschaft. Universität zu Köln.
Schaffner, S. 2001: Das Vernersche Gesetz und der innerparadigmatische grammatische
Wechsel des Urgermanischen im Nominalbereich. Innsbruck.
― 2003: “Der Beitrag des Germanischen zur Rekonstruktion der urindoger-
manischen Akzent- und Ablautklassen”, Indogermanisches Nomen. Derivation,
Flexion und Ablaut. Akten der Arbeitstagung der Indogermanischen Gesellschaft,
Freiburg, 19. bis 22. Sept. 2001. Ed. by E. Tichy, D. S. Wodtko, B. Irslinger.
Bremen, 203-218.
Schindler, J. 1967a: “Zu hethitisch nekuz”, Zeitschrift für Vergleichende Sprachfo-
schung (KZ) 81, 290-303.
― 1967b: “Das idg. Wort für “Erde” und die dentalen Spiranten”, Die Sprache 13,
191-205.
― 1967c: “Tocharische Miszellen”, Indogermanische Forschungen 72, 239-249.
― 1969: “Vogel und Ei im Indogermanischen”, Die Sprache 15, 144-167.
― 1972: “L’apophonie des noms racines indo-européens”, Bulletin de la Societé de
linguistique de Paris 67, 31-38.
― 1975: “L’apophonie des thèmes indo-européens en r/n”, Bulletin de la Societé de
linguistique de Paris 70, 1-10.
― 1977: “A thorny problem”, Die Sprache 23, 24-35.
ROLAND A. POOTH 48

Schmalstieg, W. 1981: “Ergativity in Indo-European”, Bono Homini Donum. Studies


in Memory of J. A. Kerns. Ed. by Y. L. Arbeitman & A. Bomhard. Amsterdam,
242-258.
Schmidt, K. H. 1979: “Reconstructing Active and Ergative Stages of Pre-Indo-
European”, Ergativity. Towards a theory of grammatical relations. Ed. by F.
Plank. London, 333-345.
— 1980: “Zur Typologie des Vorindogermanischen”, Linguistic Reconstruction and
Indo-European Syntax. Proceedings of the Colloquium of the Indoger-
manische Gesellschaft (Pavia 1979). Ed. by P. Ramat. Amsterdam, 91-112.
Seiler, H. 1983: Possession as an operational dimension of language. Tübingen.
― 1988 [1990]: “The dimension of Participation”, translated & ed. by F. Leal.
Función 7. Universidad de Guadalajara.
Sowa, W. 2008: Studien zum Phrygischen. Göttingen.
Stang, C. S. 1932: “Perfektum und Medium”, Norsk Tidskrift for Språgwedenskap 6,
29-39.
— 1965: “Indo-Européen *GwŌM, *D(I)I̯ĒM”, Symbolae linguisticae in honorem
Georgii Kuryłowicz. Wrocław
— 1942: Das slavische und baltische Verbum. Oslo.
Tagliavini, C. 1998: Einführung in die romanische Philologie. 2., verbesserte Auflage.
Tübingen & Basel.
Taylor, John R. 11989, 32003: Linguistic Categorization. Prototypes in Linguistic The-
ory. 3rd editon 2003. Oxford.
Thompson, L. C. & M. T. Thompson 1969: “Metathesis as a grammatical device”,
International Journal of American Linguistics 35, 213-219.
Tichy, E. 1980: “Zum Kasusgebrauch bei Kausativa transitiver Verben”, Die
Sprache 26, 1-18.
— 1993: “Transponierte Rollen und Ergänzungen beim vedischen Kausativ”, In-
dogermanica et Italica. Festschrift für Helmut Rix zum 65. Geburtstag. Ed. by G.
Meiser. Innsbruck, 436-460.
— 1995: Die Nomina agentis auf -tar- im Vedischen. Heidelberg.
— 12000, 22004, 32009: Indogermanistisches Grundwissen für Studierende sprachwis-
senschaftlicher Disziplinen. 2. und 3. überarb. Aufl. Bremen.
Traugott, E. Closs 1972: The History of English Syntax. New York.
Tremblay, X. 2003: “Interne Derivation: „Illusion de la réconstruction“ oder ver-
breitetes morphologisches Mittel? Am Beispiel des Awestischen”, Indoger-
manisches Nomen. Akten der Arbeitstagung der Indogermanischen Gesellschaft,
Freiburg, 19. bis 22. Sept. 2001. Ed. by Eva Tichy, Dagmar S. Wodtko, Britta
Irslinger. Bremen, 231-259.
— 2010: “Jenseits von Schindler? Die Bedeutungen der drei Wurzelnomina-
Ablauttypen” MSS 64, 2004 [2010], 181-221.
de Vaan, M. 2008: Etymological Dictionary of Latin and the Other Italic Languages.
Leiden & Boston.
Vaillant, A. 1936: “L’ergatif indo-européen”, Bulletin de la Société de Linguistique de
Paris 37, 93-108.
Van Wijk, N. 1902: Der nominale Genitiv Singular im Indogermanischen in seinem
Verhältnis zum Nominativ. Zwolle.
Watkins, C. 1962: Indo-European Origins of the Celtic Verb I. The Sigmatic Aorist.
Dublin.
— 1969: Indogermanische Grammatik III/1. Geschichte der indogermanischen Verbalf-
lexion. Heidelberg.
Weiss, M. 2009: Outline of the Historical and Comparative Grammar of Latin. Ann
Arbor & New York.
Widmer, P. 2004: Das Korn des weiten Feldes. Interne Derivation, Derivationskette und
Flexionsklassenhierarchie: Aspekte der nominalen Wortbildung im Urindoger-
manischen. Innsbruck.
Voice, Direction, Case, and Alignment in PIE: III. The PIE Case System 49

— 2013: “Akzent und Ablaut, externe und interne Derivation in der Nominalkom-
position”, Indo-European Accent and Ablaut. Ed. by G. Keydana, P. Widmer &
Th. Olander. Copenhagen, 187-195.
Wodtko et al. 2008 = NIL = Wodtko, D. S., B. Irslinger, C. Schneider 2008:
Nomina im indogermanischen Lexikon. Heidelberg.
Wolfart, H. C. 1978: “How many obviatives: Sense and reference in a Cree verbal
paradigm”, Linguistic studies of native Canada. Ed. by E.-D. Cook & J. Kaye
Vancouver, 255-272.
Wolfart, H. C. & J. F. Carroll 21981: Meet Cree: A practical guide to the Cree language.
2nd, revised ed. (1st ed. 1973). Edmonton.
Yakubovich, I. 2014: “Reflexes of Indo-European ‘ē-statives’ in Old Indic”, Transac-
tions of the Philological Society 112 (3), 386-408.
Zúñiga, F. 2006: Deixis and Alignment. Inverse Systems in Indigenous Languages of the
Americas. Amsterdam & Philadelphia.

Further references (to my manuscripts)


Pooth, R.A. manuscript a: “Voice, Transitivity Direction, Case, and Alignment in Proto-Indo-
European. II. The Proto-Indo-European Voice System”, downloadable from my pro-
file at www.academia.edu.
— manuscript b: “Ein Problem der Methode der komparativen Rekonstruktion von Morphe-
men, Morphemgrenzen und morphosyntaktischen Kategorien”, downloadable from
my profile at www.academia.edu.
— manuscript c: “Voice, Transitivity Direction, Case, and Alignment in Proto-Indo-
European. III. Transitivity Direction in Proto-Indo-European”, downloadable from
my profile at www.academia.edu.

Roland A. Pooth
Merheimer Str. 117
50733 Köln-Nippes
roland.pooth@gmx.de

You might also like