Proto-Indo-European Nominal Morphology. Part 1: The Noun

Language Arts 1 | 2015 | VERSION 2015-12-23

OPEN

ACCESS Freely available online

Proto-Indo-European Nominal Morphology. Part 1. The Noun Roland A. Pooth* FIU Cologne, University of Cologne ‡ , Leiden University‡ , Max Planck Institute for the Science of Human History◊ Abstract: This article provides an overview of Proto-Indo-European noun morphology. It gives an introduction into its templatic morphotactic type and provides a fundamentally revised reconstruction of Proto-Indo-European grammar. Keywords: Reconstruction of morphology, Proto-Indo-European nominal morphology, Proto-Indo-European grammar **Citation: Pooth, R. A. (2015): “Proto-Indo-European Nominal Morphology. Part 1. The Noun”, Language Arts 1, issue version 2015-12-23, author manuscript version 2015-12-23, 38 pp. plus comments and reviews Editor: Dr. Roland A. Pooth, Merheimer Str. 117, D 50733 Cologne (Nippes), Western Germany Written: Summer and autumn 2015; published: 2015-11-11 (preliminary version for https://leidenuniv.academia.edu/RolandPooth), 2015-12-23 Copyright: © 2015 R. A. Pooth. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. *E-mail: [email protected] Language Arts is an open-access freesheet for linguistic arts, pre-publication, DIY publication, and post-publication amendment. It is edited by the FIU Cologne (cf. J. Beuys 1978: “Aufruf zur Alternative”, Frankfurter Rundschau, Dec. 23; J. Beuys & H. Böll 1973: "Manifesto on the foundation of a 'Free International School for Creativity and Interdisciplinary Research'", in: C.M. Joachimides & N. Rosenthal (eds.) 1974: Art into Society, Society into Art: Seven German Artists [...]. Institute of Contemporary Arts, London, pp. 49ff., reprinted in: J. Beuys 1993 (C. Kuoni, ed.): Joseph Beuys in America: Energy Plan for the Western Man. Four Walls Eight Windows, pp. 149ff.) The FIU logo is cited here. Language Arts is published under the Creative Commons Attribution 4.0 Licence (CC BY 4.0) explicitly for post-publication amendment and supplementary correction of author manuscripts. Language Arts issues are constantly corrected. A fresh PDF version will be generated in due course. Language Arts reviewing is accomplished by means of a continuous correction process: Everybody is invited to send their comments and reviews to the editor’s e-mail address. They will be attached to the fresh PDF version. Language Arts articles can be cited with reference to the issue version and the version of the author mansucript as suggested above.** Language Arts has no responsibility for the persistence or accuracy of URLs for external or third-party Internet websites referred to in this publication. The sole responsibility for the article’s form and contents remains with the author. The reader should take care to download the latest Language Arts issue version from https://leidenuniv.academia.edu/RolandPooth. Language Arts is also available by e-mail.*

1 Introduction The present linguistic reconstruction of Proto-IndoEuropean nominal word forms, word formation patterns and morphosyntactic rules is initially based on the most archaic Indo-European word forms and word formation patterns. It is predominantly grounded on the athematic nominal stems, first of all on the patterning of the archaic IE1 root nouns such as Vedic yúj- m. ‘allied, associate’, f. ‘alliance, association’ and the archaic so-called athematic2 stems with overt suffix, e.g. Vedic paśú- m. ‘domestic animal, sheep’, páśu- n. ‘livestock, cattle’ (-u-), etc. Secondly, it is based on isolated and completely irregular but obviously archaic word forms such as Hittite gen. sg. ne-ku-uz /nēkuz/ ‘dusk’s, evening’s’. Such isolated archaisms are phonologically mapped onto PIE by undoing the respective sound laws. A form generated in this way is called a phonological transponatum (TM), e.g. ™nékwts (™négwts), cf. ‡ Alumnus: Dr. (2014) in Comparative Linguistics, Leiden University; M.A. in Historical-Comparative Linguistics, General Linguistics, Philosophy, University of Cologne. ◊ IE Lexical Cognacy Database (IELex) 2, Data Entry Group (supervised by P. Heggarty), Nov./Dec. 2015. 1 Abbreviations: adj. = adjective; adv. = adverbial; an. = animate; f. = feminine; Gk. = Greek; Hitt. = Hittite; IE = Indo-European; in. = inanimate; Lat. = Latin; m. = masculine; OCS = Old Church Slavonic; OHG = Old High German; PG = Proto-Germanic; PIE = Proto-IndoEuropean; plt. = plurative; redup. = reduplicated or reduplication; rel. or rl. = relational; Ved. = Vedic.; vlad. = verb-like adjective. 2 Note that the term “thematic” is no longer used to refer to the stems with an overt suffix other than Greek -ω/ο/ε- and -α/ᾱ/η- and its equivalents.

Hittite ne-ku-uz /nēkuz/ ‘of the evening’. The PIE form can be reconstructed as *nɛ́ɠwts. The stem *nɛ́ɠwt- was derived from the root *n_ɠw- ‘to be/become/make naked, bare, colourless’. As a methodological obligation, the linguistic reconstruction of PIE noun morphology must ultimately be based on comparative IE equation formulas. A comparative equation formula is indicated by ₠ here, e.g. ₠u̯óg̑ɦo- m. ‘waggon’. Based on a prime axiom of comparative philology, comparative equation formulas like this are normally interpreted as PIE word forms. However, the traditional evaluations were often done much too rashly and automatically. Many interpretations of equation formulas as PIE word forms suffer from the lack of diachronic linguistic plausibility. The reason for being more sceptical about the traditional interpretations is simple. Many comparative equation formulas such as ₠u̯óg̑ɦo-, ™u̯óg̑ɦah2- are solely grounded on the IE PRODUCTIVE thematic word formation patterns, cf. Greek (ϝ)όχος :: OCS vozŭ m. ‘waggon’, Mycenean Greek wo-ka, etc.3 Nouns belonging to IE productive formation patterns, however, are always and principally prone to be innovative in a particular respect, which is to say that they are prone to be result of a formal or functional morphosyntactic innovation. Equation formulas like ₠u̯óg̑ɦo- and transponata like ™u̯óg̑ɦah2- are thus prone to represent common IE secondary equations, such as result from a parallel but relatively independent formal This holds for most of the stems reconstructed by the NIL and for all thematic stems reconstructed by the LIV, cf. Meillet 1931; Pooth 2014b: chapters 7-8.

3

Roland A. Pooth

2

and functional emergence, presumably in close areal dialectal contact. 4 These equation formulas are no longer automatically interpreted as belonging to the PIE noun inflection here. It is explicitly claimed that, before being acceptable as a potentially real and realistic PIE word form, the given equation formula must undergo an additional methodological procedure which is based on the well-established method of INTERNAL RECONSTRUCTION.5 The equation formula must first be diachronically analyzed, interpreted, and identified as matching the most archaic PIE word formation patterns from the perspective of internally reconstructable PIE to IE morphosyntactic rules and tendencies. If necessary, an equation formula must thus be modified and formally retransformed to match the older and most archaic pattern, if the respective PIE pattern was evidently different and if this older pattern is evidently reconstructable. It is reasonable to be more cautious and refrain from any automatic backprojection of such secondary equation formulas to avoid a diachronic misinterpretation of productive IE word formation patterns. It is neccesary at this point of reasoning within a proper linguistic reconstruction to make the most plausible decisions as a prehistorical linguist and to sharply distinguish between primary and more reliable equation formulas (such as ₠ pék̑u- ‘domestic animal; sheep’) and secondary, less reliable or unreliable equation formulas. Whereas the former are formally close to linear results of a common inherited word form (e.g. PIE *pɛ́ku- an. ‘domestic animal, sheep’), the latter are much less reliable and cannot automatically be interpreted as such linear continuations. Nevertheless, if possible, secondary equations can also be used to reconstruct PIE word forms. But these word forms need to be formally or functionally modified and morphosyntactically reanalyzed according to a more general diachronic morphosyntactic analysis of PIE. Formally and functionally, PIE word forms do not necessarily have to exactly match secondary equation formulas. They may as well be proxies. An illustrative example for what is implied in this section can be provided by the reconstruction of a PIE inanimate singulative detransitive6 root noun *uɔ́g̈- with the reconstructable meaning ‘moving thing, moved thing; waggon’. This PIE root noun is reconstructed here by means of the equation formula ₠u̯óg̑ɦo- m. ‘waggon’ and the transponatum ™u̯óg̑ɦah2- by retransforming and remodeling both forms into a PIE root noun. This is done here, because IE root nouns are less productive than o-stems and ah2-stems and it is inferential and even well-established that root nouns preceded IE thematic nouns in many similar cases. The corresponding IE equations, their formulas and the transponatum are presented in Figure 1. Note that my representation of PIE word forms, e.g. *uɔ́g̈ in. ‘waggon’, differs from the traditional one; see section 2 below. It is thus explicitly claimed here that a form ₠u̯óg̑ɦo- m. ‘waggon’ and its function can only have the value of a 4 Comparanda are the Germanic have-perfect constructions, which parallely emerged in areal contact and yielded secondary equations. PG only had possessive constructions with *haβjanan ‘to hold, keep’ and *aiɣanan (*aih) ‘to own’. 5 Cf. Fox 1995. 6 The term DETRANSITIVE is used here for “proto-middle” or “premiddle”. The term singulative refers to a form without morphological plurative marking. I will return to these notions below.

secondary and less reliable equation formula and is not necessarily a potentially real and realistic PIE proper noun, despite all previous claims of comparative philology. With regard to the traditional methodological view on how to reconstruct morphosyntactic categories, such as nouns, or gender, or case forms and functions, comparative philology is much too confined to the set of categories found in the IE languages. It is morphosyntactically too dogmatic and naїve.7 In short, equation formulas like ₠u̯óg̑ɦo- m. ‘waggon’ need not be analyzed as PIE proper nouns. label IE form IE equation formula phonological transponatum provisional form reconstructed PIE form

examples Greek (ϝ)όχος :: OCS vozŭ m. ‘waggon’, Mycenean Gk. wo-ka ₠ u̯óg̑ɦo™u̯og̑ɦah2*u̯óg̑ɦo-, *u̯og̑ɦah2inanimate root noun *uɔ́g̈, verb-like adjective *uɔ́g̈ɔ ~ *uɔg̈ɔ́, COL/PL forms *uɔ́g̈aχ ~ *uɔg̈áχ

Figure 1. IE forms, provisional forms, and reconstructed forms

For structural reasons, it is inferential that all stems ending in /-ɔ-/ as well as all stems ending in /-aχ-/, e.g. PIE *ʔɛ́kuɔ-, *uɔ́g̈ɔ- ~ *uɔg̈ɔ́- once belonged to a different morphosyntactic word class, namely the class of VERB-LIKE 8 9 ADJECTIVES. As outlined elsewhere, PIE verb-like adjectives were either idential to the detransitive oblique stem, e.g. *ulkwɔ́-, *ɠɛ́nuɔ-, or were derived from an underlying root noun or a classified noun by suffixation or suprasegmental mapping of /ɔ/, and more regularly by additional accent shift, yielding stem final -ɔ́-. It is inferential that many of these adjectives were simply nominalized parallelly and independently only within a Vulgar Post-PIE10 period in areal contact yielding parallel but independent IE secondary equations. An illustrative example for an evident secondary nominalization is provided by the comparative IE equation formula for ‘wolf’ which is ₠u̯ĺ̥kwo- m., cf. Gothic wulfs (< PG *wulfa-) :: Vedic vŕ̥ka-, cf. Greek λύκος, Latin lupus. It is evident that the stem ₠u̯ĺ̥kwo- was the outcome of a secondary nominalization encoded by accent shift to the initial syllable. It is further inferential that ₠u̯ĺ̥kwo- was derived from a previous stem *ulkwɔ́- ‘dangerous, harmful, dreadful, wolf-ish’, as continued by Vedic vr̥ká- adj. ‘harmful, not safe’, attested as privative avr̥ká‘safe’, which is still an adjective with accent on the suffix.11 In the present (and following) analysis, the PIE stem *ulkwɔ́- was both used as a verb-like adjective stem and as oblique stem (e.g. abl.-gen. *ulkwɔ́s) of a root noun and adjective *uɔ́lkw- an. ‘habitual dreadful one or dangermaker’ as well as ‘dangerous (inanimate head noun)’, abl.gen. *ulkwɔ́s, opposed to *uɛ́lkw- in. ‘danger’ and ‘dangerous (animate head noun)’. 12 So even if many IE languages show a younger nominalized form (₠u̯ĺ̥kwo-), it is far from Cf. Pooth 2014a. Cf. Dixon 2013 for a typological overview. 9 Cf. Pooth 2015. 10 For this term and a definition cf. Pooth 2015. 11 Cf. Kloekhorst 2008: 950 on Hitt. walkuwa- n. meaning something negative, maybe ‘harm, danger’ as suggested by Lehrman 1987. 12 Cf. Pooth 2015. 7 8

Proto-Indo-European Nominal Morphology. Part 1. The Noun

assured that this secondary nominalization already occurred in the proper parent language, simply because accent shift is a common old IE morphological strategy to derive thematic nouns from adjectives and adjectives from thematic nouns. As for their form and function, provisional equation formulas like these are less reliable and cannot that easily or automatically be considered PIE proper nouns any longer. As just mentioned, they will be analyzed as verb-like adjectives in the sequel and part two of this article. After all, it is necessary to reconstruct not just PIE inflected word forms or nominal stems, but also all PIE morphotactic and morphological rules and tendencies. After the thorough application of the comparative method to all IE word forms, as undertaken by comparative philology, it is about time to apply an advanced methodological procedure to all these primary, reliable or secondary, less reliable equation formulas. It is necessary to introduce an advancement of the method of linguistic reconstruction at this point of the reconstruction of the PIE parent language. This additional procedure consists of the application of a proper, adequate, and up-to-date linguistic morphological analysis to all the equations formulas and transponata.13 For the time being, it is urgently necessary to bid farewell to the traditional linear segmentation into reduplication (prefix), root, suffixes, and inflectional endings with regard to PIE nominal (and verbal) morphology. It is necessary to start the morphological analysis completely anew. The sole backprojection of IE morpheme boundaries onto the parent language, resulting in the parsing of ₠u̯óg̑ɦo- into root (including o-grade ablaut) and suffix -o-, has turned out to be a quite problematic transfer of a younger morphological segmentation to a very different language. However, to be clear, such a transfer of morpheme boundaries from IE to PIE, which was a different language, is illegitimate within a proper linguistic morphological analysis. The application of the comparative method to morpheme boundaries can finally be regarded as circular reasoning. This methodological mistake has become more and more evident within the course of the last decades due to the increasing theoretical separation of general diachronic and historical linguistics and general linguistics from comparative philology. However, even otherwise outstanding comparative philologists just did not recognize this problematic deduction, simply because it was almost never considered a problem, 14 although the mistaken deduction is logically evident. 15 In short, this linguistic reconstruction of PIE nominal morphology is not based on the comparative IE segmentation any longer. It is founded on the average methodological and typological knowledge of general synchronic, diachronic, and historical linguistics. It is analytic, functionally probabilistic, and polymethodological. It is based on the application of multiple analytical methods of linguistic reconstruction to the entire IE material. It is based on the introduction of a sharp distinction of equation formulas from actually reconstructable PIE word forms (e.g. *uɔ́g̈-, *ʔɛ́ku-). It is based on the sharp distinction of more reliable primary equation formulas and less 13 14 15

Cf. Pooth 2014d, 2015. There were famous exceptions, e.g. Lehmann 1974, 1975, 2002. Cf. Pooth 2014d.

3

reliable secondary equation formulas (e.g. ₠u̯óg̑ɦo- or ₠ h1ék̑u̯o-), as illustrated above. Note that equation formulas and transponata are represented here by the traditional notation (e.g. ₠u̯óg̑ɦo- m. ‘waggon’). Like traditional comparative philology, this reconstruction is also based on the application of the comparative method to the sum of IndoEuropean word forms. But in a sense it is even more cautious than comparative philology, because it tries to avoid anachronistic misinterpretations of younger word forms by not automatically considering a secondary equation formula a potentially real and realistic PIE word form. It is based on the insight that the use of the comparative method can only be an initial and provisional step within the process of reconstruction of morphology and syntax and that the present results of comparison in these fields should not be overestimated. It differs from traditional comparative philology by claiming that a mere comparison of morphology and syntax is not enough. It is thus initially based on an up-to-date linguistic analyis of the reconstructable primary equation formulas and transponata, and a subsequent analytical and interpretative adjustment of the secondary equation formulas and transponata to the results of the application of this proper linguistic morphological analysis. The overall morphological analysis is thus somewhat more independent from comparison than it was before, and it conforms to the principles of general linguistics. After application of these multiple methods of linguistic reconstruction to the entire Indo-European material, the common PIE parent language will now turn out to be of a very different linguistic type, typologically quite different than all its younger IE daughter languages.

2 Phoneme inventory and representation This article is not about PIE phonological issues, but it must offer an overview of the phoneme inventory and the representation of the phonemes. The traditional representation is not maintained here for reasons that have already been outlined elsewhere.16 Instead, I principally make use of the symbols of the International Phonetic Association (IPA). The PIE segmental phoneme inventory comprised a minimum of 31 phonemes including twenty-three consonant phonemes, two semi-vowel phonemes /i u/ and six full vowel phonemes /a aː ɛ ɛː ɔ ɔː/:

/p b̤ ɓ t d̤ ɗ k g̈ ɠ kw g̈w ɠw q ɢ̤ ʛ ʔ s χ ʕ m n r l u i a aː ɛ ɛː ɔ ɔː/ The traditional mediae 〈b d g̑ g gw〉 are reconstructed as voiced implosive stops here, generally following the glottalic theory, in particular the suggestion of Kümmel 2012. The traditional mediae aspiratae 〈bh dh g̑h gh gwh〉 are reconstructed as PIE breathy or murmured voiced plosives here. 17 The three “laryngeals” are reconstructed here as

Cf. Pooth 2015. Kümmel 2012 and p.c. has suggested that the breathy voiced stops emerged by the time when the implosives became plain voiced plosives. However, the progressive assmilation a.k.a BARTHOLOMAE’S LAW, e.g. */b̤ud̤tɔ́/ → [b̤uʣ'd̤ɔ] (> Vedic buddhá-) is much easier to understand in terms of a breathy voice laryngeal feature, since plain voiced stops more typically undergo a lag assimilation to a following voiceless stop, whereas a rightward spread of the breathy voice feature is quite unproblematic. I would expect a resultant †[b̤uʦ'tɔ] if PIE had plain 16 17

Roland A. Pooth

4

two fricatives /χ ʕ = h2 h3/ and a glottal stop /ʔ = h1/. The vowel phonemes, including the semi-vowels, are given in Figure 2. PIE consonant phonemes, also including the two semi-vowels, are given in Figure 3. The marginal but not absent PIE phoneme /ɓ/ is enclosed in parentheses. Key to the columns: 1. plain voiceless stops; 2. breathy or murmured voiced stops; 3. voiced implosives; 4. voiceless fricatives (including the sibilant); 5. voiced epiglottal/pharyngeal fricative or approximant; 6. nasal stops; 7. lateral approximant; 8. alveolar trill or approximant; 9. semi-vowels (vocalic glides or approximants); 10. glottal stop. closed mid open

front i

mid ɛ ɛː

different morphological levels within the autosegmental morphological analysis are conveniently termed morphological tiers or morpheme tiers. The vowel melody is analyzed as an independent segmental string, morphotactically independent from the agglutinating affixation on the skeletal consonant frame tier. It is mapped onto the underlying vowel melody template, abbreviated VMT, which consists of underspecified vowel slots, symbolized by _V1_ and _V1V2_ or _V1_+_V2_, etc. The different tiers can be illustrated by Figures 4 and 5. vowel melody tier detransitive vowel tier

back and rounded u ɔ ɔː

a aː

bilabial alveolar velar labiovelar postvelar epiglottal glottal

2 b̤ d̤ g̈ g̈w ɢ̤

3 (ɓ) ɗ ɠ ɠw ʛ

4

5

s χ

6 m n

‘bare’ vowel tier

ɛ

vowel phoneme moraic tier

µ

7 l

8 r

9

ʕ

ʔ

consonant frame tier root and inflectable base tier classifying suffix tier case suffix tier

Figure 3. PIE consonant phonemes including semi-vowels

µ ⎜

µ

⎜

⎜

V1

V2

⎜ ⎜

⎜ ⎜ ⎜ ⎜ ⎜

´ ⎜

tonal tier word form CV template tier

⎜

µ

word form template tier stress accent tier

i u

ɛ

⎜

vowel melody template tier

10

⎜

⎜

vowel melody moraic tier

Figure 2. PIE vowel phonemes inlcuding semi-vowels 1 p t k kw q

ɔ

H C ⎜ ⎜

ɠ

⎜

V

C

⎜ ⎜

n

C

⎜ ⎜ ⎜

u

⎜

V

C

⎜ ⎜ ⎜ ⎜

s

Figure 4. Segmental and templatic analysis of PIE *ɠɛ́nuɔs

3 Nominal morphotaxis As argued in detail elsewhere,18 it is inferential that PIE morphology was of the root and pattern morphology type. According to the convenient definitions, languages of this type are made up from a skeletal consonant frame (abbreviated CF) and a transfix or vowel melody (abbreviated VM):

[...] In these languages, the root in a number of common binyanim[19] or paradigms may be analysed as being made solely of consonants, while the pattern of vowels which are found around the consonants and the particular vowels filling up the pattern provide morphological information comparable to that often given by affixation. This analysis leads to discontinuous roots and discontinuous morphs interacting with the roots, [...]. A transfix is a particular type of affix, one which is completely interwoven with its base. Typically, it is a series of vowels which surround and interact with a base which in turn can be analysed as a series of consonants. For example, Arabic katab ‘he wrote’, kitaab ‘book’, kaatib ‘clerk’ (where the root is *ktb, indicating ‘writing’) illustrate the transfixes _a_a_, _i_aa_ and _aa_i_. Such transfixes [...] are discontinuous affixes attached to discontinuous bases, [...]. (Bauer 2004: 93, 102)

3.1 Morphological tiers Following the terminological tradition of the autosegmental approach, as launched by McCarthy 1981, the term vowel melody is used here as a synonym of transfix. The

voiced plosives. It is thus inferential that PIE had breathy voiced plosives in addition to implosive stops. 18 Cf. Pooth 2000: last fn., 2001, 2004a, 2009b, 2014acd, 2015; Tremblay 1999, 2003. 19 The term binyānî́m (pl.) is borrowed from Classical Hebrew grammar.

The PIE word form analyzed in Figure 4 is *ɠɛ́nuɔs. It is the PIE oblique ablative-genitive form of *ɠɔ́nu in. ‘knot, joint, angle’. Oblique forms of the o/e-graded acrostatic type, e.g. *ɠɔ́nu, *ɠɛ́nuɔs, are reconstructed here with a disyllabic trochaic metrical structure (σ́ σ) and bivocalic underlying vowel melody template _V_V_. The oblique forms had the same vowel melody as verb-like adjectives, e.g. *ɠɛ́nuɔ- ‘knotty’, except that the first had a trochaic metrical structure with the accent on the initial syllable (σ́ σ), whereas the latter more regularly had an iambic structure with accent on the final syllable (σ σ́). The ablativegenitive form *ɠɛ́nuɔs is what is actually confirmed by the IE material, cf. Greek nom.-acc. sg. n. γόνυ :: Hittite gen. sg. n. gēnuwaš (OS ge-nu-wa-aš, OH/NS ge-e-nu-wa-aš 20 ). Likewise, the Vedic abl.-gen. sg. m. ávyas goes back to *χáuiɔs, corresponding to *χɔ́uis, cf. Lat. nom. sg. m. ovis :: Vedic nom. sg. m. ávis ‘sheep’. Vedic abl.-gen. sg. vásvas and Homeric Greek gen. sg. ἐῆος, ἑῆος (< Proto-Greek *eu̯éhu̯os) both point to PIE *ʔuɛ́suɔs, abl.-gen. of *ʔuɔ́su in. ‘worthy thing’. The corresponding PIE partitive-genitive case form *ɠɛ́nuɔm is continued by Gk. Hsch. γέυνων 21 (™g̑énu̯oHom). As outlined in more detail below, the younger Vulgar Post-PIE genitive plural case forms ending in *-ɔm and *-ɔʔɔm (™-oHom) go back to PIE transnumeral partitive-genitive case forms, e.g. PIE *ʔuɔ́su *ʔuɛ́suɔm ‘a worthy thing among (and part of) the worthy thing(s)’. Most of the IE gen. pl. forms reflect a secondary suffixation of the reanalyzed suffix *-ɔm to the thematic stem, e.g. 20 21

Cf. Kloekhorst 2008: 467f. Cf. Tremblay 2003: 577 with fn. 21.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

segment/ template _ɛ_ _V_ CV́C-uσ́ σ 23 p_k-u-s ― *pɛ́kus

5

3.2 Consonant frame gloss22

In addition to transfixation on the vowel melody tier, PIE made use of the morphotactic strategy of agglutination on the consonant frame tier. Nominal classification as for shape and nominal aspect including deicticmarker-like suffixes for the referential function, and inflectional case and number marking were provided by agglutinating suffixation on the consonant frame tier. After a Figure 5. Segmental and templatic analysis of PIE *pɛ́kus. proper subtraction of all the discontinuous full vowel phonemes /ɛ ɛː ɔ ɔː/ from the C-chain structure gloss example translation segmental string, the agglutinating skeletal w w u_k ro_otspe_ak*uɔ́k ‘word’ C-chain, that is, reduplication prefix, inflectn_ɠw-_tro_ot-_Cna_ked-_CL1 *nɔ́ɠwt ‘night (first half)’ able base, including root, and suffix(es) ɠw_n-_χro_ot-_Cwo_man-_CL3 *ɠwɛ́nχ ‘group-woman’ su_p-_rba_se-_Cfall.a_sleep-_CL4 *suɔ́pr ‘dream’ remains. Recall that the inflectable base can p_k-t_nro_ot-C_Cco_mb-CL1_CL4*pktɛ́n ‘s.o.’s comb(er)’ be split up into the proper root and an opd̤_g̈-_mro_ot-_Cgrou_nd-CL6*d̤ɛ́g̈ɔm ‘ground, earth’ tional derivational suffix, that is, the sod̤ɛ-d̤_g̈-_u- redup-ro_ot-_C- redup-grou_nd-CL8*d̤ɛd̤g̈ú‘groundling, fish’ called enlargement. It is further inferential Figure 6. A few PIE nominal consonants frames that the so-called mobile s- was a derivational or inflectional prefix. I use the underline _ to represent a morpheme boundary lexical and derivational inflectional between the consonant frame and the vowel red. infl. base ref. shape “modal” “aspect” ref. case number melody. Furthermore, the underline also -1 0 1 2 3 4 5 6 7 indicates the position of the vowel slot Cɛ CV1V2C V3C V4C V5C V6C V7C V8C V9C Cɛ t l i ʔ t Ø s within the inflectable base, e.g. *d̤_g̈-_mn u χ s s s ‘ground, classified as large, extended, flat’. A n m r n m few PIE nominal consonant frames or consoɗ n ɗ i nantal skeletons are given in Figure 6. The d̤ K d̤ ʔ (b̤) order of affixes in nominal stems is given in CL1A CL5A CL7 CL2 CL1A ABS/VOC/LOC PL Figure 7. In line 3, in between the grey lines, CL1B CL5B CL8 CL3 CL1B ERG/GEN the consonantal morpheme slots are numCL1C CL6 CL4A CL1C ALL/PAR bered -1 to 7 with 0 being the inflectable CL1D CL4B CL1D LOC/BEN CL1E CL10 CL1E SOC/COM base. The vowel slots of the vowel melody (CL9) are included in line 4. Lines 5 and 6 of Figure 7 contain a list and a numbering of the Figure 7. PIE nominal morphotaxis respective classifying suffixes, as it is used here. The letters a, b, c, d, e after the num*ɠɛ́nuɔ (₠g̑énu̯o-), yielding a secondary morphological labers indicate allomorphic suffixes of the respective class. ʔ ryngeal hiatus in between the two vowels *-ɔ ɔm. This Let me add a brief overview of the functions of the classisegmental sequence was further fused and again reanafying suffixes. I will return to them below. I suggest anaʔ ₠ lyzed as a new and longer gen. pl. suffix *-ɔ ɔm ( -oHom). lyzing the five alveolar suffixes -t-, -s-, -n-, -ɗ-, -d̤- as referThere is no need to reconstruct a phonemic laryngeal in ential suffixes in complementary distribution, encoding this position. Recall that there is no comparative evidence prototypical reference (with a nominalizing function), e.g. for gen.-abl. forms of the shape **h1u̯ésus or **g̑énus of this *nɔ́ɠw-t- in. ‘night’, *mɛ́ʔ-n-s- ‘moon: measurer’ besides inflectional type. Schindler’s 1967a reconstruction of the *mɛ́ʔ-n_ɔ_t- ‘moon-ish, moon(s) (generic); month(s)’. I use oblique forms of the o/e-graded acrostatic inflectional type the provisional term “modal” classifying suffixes for the must be corrected this way. two semi-vocalic suffixes -i- (CL7) and -u- (CL8). It is obviEach PIE inflected nominal word form consisted of a ous that the suffix -u- was a possessive marker, indicating a minimum of two obligatory overt morphemes. The first obligatory overt morpheme was the skeletal discontinuous 22 I use the general glosses redup- = reduplication, ro_ot- = discontiinflectable base, abbreviated IB, including the proper root nuous root, ba_se- = discontinuous inflectable base, because PIE (R). The second overt morpheme was the discontinuous regular roots and discontinuous inflectable bases were interrupted by a inflectional vowel melody, abbreviated VM, which was vowel slot _V_, e.g. *p_k- ‘to fix wool, fix hair, comb’, *ʔu_s- ‘worthy’, completely interwoven with the discontinuous inflectable etc. Furthermore, I principally follow the Leipzig Glossing Rules. Most of the glosses are the convenient ones, e.g. ABS = absolutive; ADJ = base, including the root. The other morphemes or moradjective, ALL = allative-dative; AN = animate, ASS = associative; BEN pheme slots could remain non-overtly zero-marked or = benefactive-finalis; CL = class(ifier); COL = collective; DEM = deunmarked, depending on how one prefers defining the monstrative; DTR = detransitive-middle; ERG = ergative-genitive; IN = notion of zero in languages of this type. inanimate; OBL = oblique (stem); PAR = partitive-genitive, PL = plural I tier vowel melody (or transfix) vowel melody template word form CV template word form metrical structure consonant frame case suffix number marking PIE word form

or underspecified SINGULATIVE or underspecified nonoblique, nonrelational nonoblique, nonrelational ro_ot-CL8 ERG or agent/possessor case SINGULAR or underspecified ‘single domestic animal, sheep’ ACTIVE

further use the glosses PLT = plurative, RL = relational, melody or bare vowel of the vowel melody. 23 Symbols: σ́ σ = trochaic foot, σ σ́ = iambic foot.

VM

= vowel

Roland A. Pooth

6

natural function, quality, or property, specified by the respective base, e.g. *ʔɛ́k-u- an. ‘horse: its natural property is swiftness, being swift’. This suffix -u- was also part of the extended possessive suffix -u_nt- (~ -m_nt- after -u-), e.g. *pku-mɛ́nt- ‘having a lot of domestic animals or sheep’. The class 7 suffix -i- was used in a more general and underspecified derivational function. It was a kind of nominalizer that was mainly used to derive classified nominal stems from root nominal stems, e.g. *pɔ́t- (root noun) ‘being able on one’s own, master’ → *pɔ́t-i- (class 7 noun) ‘id.’ → *pɔtɛ́i- ~*pɔtí- (relational stem) ‘s.o.’s master’. PIE further had two shape classes: A pointed, round(ish), small shape was coded by the suffixes -l- (CL5A) and its oblique counterpart -n- (CL5B), e.g. *sáχu-l-, abl.-gen. *suχnɛ́s n. ‘sun (round)’, *χáɓ-l- n. ‘apple (round)’. A large, extended, flat, big, augmentative shape was marked by the suffix -m(CL6); examples will be given below. The notion of nominal aspect, subsuming the classifiers -ʔ- (CL2), -χ- (CL3), and -r- (CL4A) ~ n- (CL4B) refers to the internal structure of the given concept.24 -K- is used as a cover symbol for the DIMINUTIVE or SIMILATIVE suffixes -k-, -g̈-, -ɠ-, -kw-, -g̈w-, -ɠw-, -q-, -ɢ̤-, -ʛ-. Further abbreviations in Figure 7 are: infl. = inflectable, red. = reduplication, ref. = referential.

3.3 Nominal voice marking As just outlined, PIE had transfixation on the vowel melody tier, which was a superordinate tier of the detransitive vowel tier, the bare vowel tier, the vowel phoneme moraic tier, the vowel melody moraic tier, and the vowel melody template tier; see Figures 5 and 8. As a typological peculiarity, PIE regularly exhibited “middle nouns”, that is, detransitive nominal stems. Nominal voice marking is typologically rare, but it is definitively not absent.25 Underlying unmarked agentive-active or underspecified nominal stems were distinguished from detransitive (DTR) nominal stems, e.g. *ɗáru- (perhaps besides *ɗɛ́ru-) an. ‘tree’, cf. Proto-Celtic *daru- f. ‘tree’, Old Irish daur ‘id.’ vs. *ɗɔ́ru- in. ‘wood’, cf. Vedic dā́ru ‘wood’, etc, both derived from the PIE root *ɗ_r- ~*ɗar-.26 Such detransitive “middle nouns” were marked by the discontinuous autosegmental or suprasegmental detransitive morpheme ɔ /ɔ/. This detransitive vowel was suprasegmentally mapped onto a bare vowel /ɛ/ or /a/, occupying a vowel slot of the underlying vowel melody template, as illustrated by the PIE example *ɠɔ́nu- in. ‘inanimate joint, angle, node’ in Figure 8. The underlying and basic nominal stem is *ɠɛ́nu an. ‘animate or human joint, angle, node: knee, chin’. the introduction of a vowel phoneme moraic tier and a vowel melody moraic tier can help to represent the mapping of two underlying vocalic segments /ɔ/ and /ɛ/ and the respective two moraic units of these segments onto one and the same monovocalic vowel melody template _V1_ which was monomoraic. This analysis has the advantage of keeping the discontinuous vowel as a moraic unit in the representation of the analysis. This vowel turns up again, reCf. Rijkhoff 2002. 25 Cf. Benjamin 2011. 26 Some PIE roots of the structure */C_r-/ had allomorphic a-roots, e.g. *ʔ_r- ~ *ʔar-, *mar- ~ *m_r-, cf. Latin mare n. ‘sea, ocean’ < PIE abl.gen. *máriɔs, abs. *mɔ́ri in. ‘habitually retracting (tidally)’, derived from *m_r- ‘to vanish, fade, wither, retract, go away, die’. 24

maining in its position in the vowel slot of the inflectable base, when the other moraic unit is transposed to another position within the word form, e.g. gen. *ɗɔ́rus → abl.-gen. *ɗɔruɛ́s. Therefore, this analysis describes the suprasegmental mapping quite adequately. Note that the position of the discontinuous vowel within the analyis, that is, whether it is posited to the left or to the right of the ‘bare’ vowel is irrelevant. One could perhaps also graphically represent it before or behind the ‘bare’ vowel. It occupies a position left to the ‘bare’ vowel /ɛ/ in the scheme given in Figure 8 just due to my arbitrary choice. vowel melody tier detransitive vowel tier

ɔ

⎜ ⎜ ɛ ⎜ ⎜

‘bare’ vowel tier vowel phoneme moraic tier

µµ

/

⎜

vowel melody moraic tier

µ ⎜

vowel melody template tier

V ⎜ ⎜

word form template tier stress accent tier

´ ⎜

tonal tier word form CV template tier consonant frame tier root and inflectable base tier classifying suffix tier case suffix tier

H C ⎜ ⎜

⎜

V

ɠ

C

⎜ ⎜

n

C

⎜ ⎜ ⎜

u

C

⎜ ⎜ ⎜ ⎜

Ø

Figure 8. Suprasegmental mapping of /ɔ/ onto /ɛ/

The corresponding agentive-active or unmarked forms only contained the unmarked or ‘bare’ vowel phonemes /ɛ/ or /a/, e.g. *ɠɛ́nu an. ‘animate or human joint, angle, node: knee, chin’. Thus, agentive-active nominal stems were unmarked in this morphological sense. The detransitive nominal stems, on the other hand, were marked by the discontinuous detransitive voice marker ɔ or by its continuous suffixal counterpart -χ-, or by both markers, e.g., in the verb-like adjective collective-plural form *g̈ɔʔráχ ‘left-over-ish, remaining’,27 cf. Greek χώρᾱ f. ‘empty area, land’, χήρᾱ f. ‘widow’, see examples (1) and (2). It is obvious that nominal and verbal detransitive voice was marked by the same discontinuous detransitive marker ɔ and by the same continuous detransitive suffix -χ-. (1) *sá-χ ɠwɛ́n-χ g̈_ɛ_ʔ-r_á_χ DEM-CL3\ABS woman-CL3\ABS ro_VM_ot-CL4A_PLT_CL3\ABS (α) ‘this woman (belonging to a group) left-over-ish’ → ‘this left-over woman’; (β) ‘this group of left-over women’

(2) *sáχ28 nɔ́ɠwt g̈_ɔ_ʔ-r_á_χ dito night.ABS ro_DTR_ot-CL4A_PLT_CL3\ABS ‘this left-over (empty) night (= naked/bare sky, conceptualized as a collective of left-over lights after sunset)’

27 This is derivative of the heteroclitic *g̈ɔ́ʔr, abl.-gen. *g̈ɛ́ʔnɔs ~ *g̈ɛ́ʔrɔs in. ‘non-human remaining(s)’, derived from *g̈_ʔ- ‘to go away, duck; leave s.o., be left’ (> Vedic hā ‘id.’). The agentive-active stem *g̈ɛ́ʔr had a meaning ‘[+human] remaining(s)’. Note that *g̈ɛʔrɛ́ɗ- ‘s.o.’s humanleft-over-ish one’ (> Latin hērēs, hērēdis m. ‘heir’) points to a plurative base *g̈ɛː́ʔr, see section 3.4. 28 On optional plurative semantic congruency see section 7.2.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

It is crucial for any understanding of PIE templatic morphology recognizing that all derivational and inflected word forms coded by the discontinuous vowel phoneme /ɔ/ and the continuous detransitive suffix /-χ-/ belonged to the detransitive voice category, regardless from their respective position within the given word form. The suprasegmental mapping of the discontinuous detransitive vowel morpheme ɔ upon the ‘bare’ vowel ɛ or else the vowel a of a-roots seems to be one of the most fascinating properties of PIE morphology. PIE had a distinction between active-agentive or unmarked, more underspecified nominals (e.g. *pɛ́ku- an. ‘single domestic animal, sheep’) and detransitive or detransitive-deagentive-inactive nominals (e.g. *pɔ́ku- in. ‘product of domestic animal’). Further examples are (cf. also footnote 25): (3)

(4) (5) (6)

(7)

(8) (9)

*pɛ́ku- an. ‘domestic animal, sheep’ → *pɔ́ku- in. ‘product of domestic animal: fleece, young cattle’, cf. Mycenean Greek (Kn C4) po-ku-ta /pokutās/ “Freier, der nur Kleinvieh besitzt”,29 OArm. as- ‘fleece’, nom. acc sg. asr, nom. pl. ask‘ :: Vedic paśú- m., etc. *uɛ́kw- an. ‘speaker’, in. ‘volitional act of speaking’ → *uɔ́kw- in. ‘result or product of speaking: word’ (→ *uɔ́ːkwin. ‘group of words, formula, speech’)

*χáui- an. ‘bird’ → *χɔ́ui- in. ‘product of bird, egg’ (→ *χɔ́ːui- in. ‘group of eggs (in a nest)’, abl. *χauiɔ́s → *χɔːuiɔ́- = ₠h2ōu̯i ̯ó-)

*ʔɛ́sus, abl. *ʔsɛ́us an. ‘existence, life, life-force, strongness, vitality’ [+animate natural force] → *ʔɔ́su, abl. *ʔɛ́suɔs in. ‘good thing, realium’ , cf. Vedic ásu- ‘lifeforce’ m. vs. Hittite āššu ‘thing’ n.

*mɛ́d̤u in. ‘sweetness, honey’ → *mɔ́d̤u in. ‘product of honey, alcoholic honey, mead’, cf. Toch. B mot “boisson alcoolisée”;30 there is no positive evidence for a status as a loan from Middle-Iranian, cf. Sogdian mwδy, acc. sg. mwδ “boisson alcoolique”31 *páχiu-, abl. *pχiɛ́us an. ‘herdsman: protecting agent’ → *pɔ́χiu in. ‘cattle: protected undergoer’, Vedic abl.-gen. sg. pāyós m. ‘herdsman’, but Greek πῶυ n. ‘cattle’32

*χáiu-, abl.-gen. *χiɛ́us ‘animate lifespan, age of vital force, age of vigor’ → *χɔ́iu, *χáiuɔs in. ‘time-span, age, eternity’, cf. Avestan gen. sg. yaoš ‘lifespan (referring to animate lifetime), but PG *aiwaz > Gothic aiws ‘time, eternity’ (referring to a time and eternity in general)33

(10) *ɢ̤ɛ́b̤χl ‘(human) roundish seizer/holder: head’, abl.-gen. *ɢ̤b̤χáls an. ‘holder, frame: head [+human] or [+animate]’ (→ Vulgar Post-PIE *ɢɦebɦχlā f.) → *ɢ̤ɔ́b̤χl, abl. *ɢ̤ɛ́b̤χlɔs :: verb-like adj. *ɢ̤ɛ́b̤χlɔ- ‘inanimate holder, head of a house, gable’, cf. OHG gibilla f. ‘head’, Greek κεφάλη f. ‘head’, Toch. A śpāl ‘head’, but OHG gibil m. ‘gable’, Gothic gibla (< PG *γeβ(u)lōn), ON gafl ‘Giebelreiter’, NE gable (< *γaβ(u)la- m.); both stems were obviously derived from the root *ɢ̤_b̤χ- ‘to seize, hold, keep, grab’.34 Cf. Tremblay 2003: 575. Cf. Pinault 2003: 183. 31 Pace Pinault ibid. 32 Cf. Tremblay 2003: 578. 33 Cf. Tremblay 2003: 577. 34 Cf. LIV, s.v. “ghebhh2-”.

7

In the system of PIE noun inflection voice marking was rather derivational. It was used to code detransitive nominal stems which were evidently derived from underlying unmarked ones in the majority of cases. But in the system of root nouns and root adjectives, detransitive voice marking was derivational and inflectional. In the verb system, however, voice marking was completely inflectional. Detransitive root nouns, mainly denoting habitual agents as well as results, products, artefacts, were derived from underlying unmarked root nouns by detransitive suprasegmental vowel mapping: (11) *g̈wɛ́n an. ‘slayer; beating, slaying one’ (agent noun), in. ‘beating, slaying’ (action noun, verbal abstractive) → *g̈wɔ́n in. ‘result of beating: wound, ban’ (result noun)

Inanimate root adjective forms were internally derived from animate ones and were coded by detransitive suprasegmental vowel mapping: (12) *ʔrɛ́ud̤ an. ‘red one, animate head’ → *ʔrɔ́ud̤ in. ‘red one, inanimate head’

(13) *sɔ́

ʔrɛ́ud̤ red.AN.ABS.SG

(14) *tɔ́ɗ sáχul DEM.IN.ABS sun.ABS ‘this sun, this red one’

ʔrɔ́ud̤ red.IN.ABS

χnɛ́r man.ABS.SG ‘this man, this red one’

DEM.AN.ABS.SG

3.4 Plurative marking PIE distinguished a basic and underived singulative or underspecified (polysemous) monovocalic vowel melody template (_V1_) from a specifically bivocalic PLURATIVE 35 vowel melody template (_V1V2_), glossed PLT. Whereas the single (and short) vowel melody template indicated verbal and nominal singularity (or aspectual underspecification), the geminated (or long) vowel was a specific marker of both verbal and nominal plurality or plurativity, including verbal duration or a state. The plurative meaning was coded by gemination of the vowel slot on the vowel melody tier of the word form template tier (_V1_ → V1V2_), as illustrated, e.g., by PIE *pɛ́ːku in. ‘livestock, cattle, collective group of domestic animals’ in Figure 9. Such a templatic morpheme is sometimes called a “chroneme”, because the vowel gemination/length conveys a specific meaning and has morpheme status. This mechanism of “internal derivation” took place on the vowel melody template tier and (parallelly) on the word form template tier. The distinction of basic nominal stem and its derivative is still partially preserved by the masculine stem ‘domestic animal’ vs. the neuter stem ‘cattle’ in Early Vedic (RV):36 (15) *pɛ́ku an. ‘domestic animal’, abl.-gen. *pkɛ́us, loc. *pkɛ́ui (> Ved. nom. sg. masculine paśús, abl.-gen. paśós, dat. paśáve) → *pɛ́ːku in. ‘livestock, cattle’, abl.-gen. *pɛkuɛ́s (> Ved. nom. sg. neuter páśu, abl.-gen. sg. paśvás, dat. páśve)

29 30

35 I use this term as a functional superordinate for PLURAL, COLLECTIVE, NONINDIVIDUATED, ABSTRACTIVE, GENERIC, AUGMENTATIVE. Augmentatives

usually denote the concept [MORE]. Cf. Tremblay 1999.

36

Roland A. Pooth

8

(16) *χ_kw- (root) ‘animate water’ → *χɛ́ːkwi in. ‘big water, ocean’, verb-like adj. *χɛːkwiɔ́- > ON ǽgir

vowel melody tier detransitive vowel tier ‘bare’ vowel tier

ɛ

vowel phoneme moraic tier

µµ

The PIE transfixal plurative marker (illustrated by Figures 9 and 10) is formally similar to the later IE “Vr̥ddhi” ablaut grade of the root. Several instances of forms showing the “Vr̥ddhi” ablaut grade of the root are simply continuations of forms with a PIE plurative marker.37 PG *hōnaz- n., for instance, clearly denoted a bigger fowl, not just a hen. 38 Its meaning was not simply “possesive”. Originally, it was not a possessive “Vr̥ddhi” derivative,39 but a continuation of the original augmentative plurative reading [MORE]. It can finally confirm the analysis provided here; see (17).

\

⎜

⎜ ⎜

vowel melody moraic tier

µµ ⎜ ⎜

vowel melody template tier

VV ⎜ ⎜ ⎜ ⎜

word form template tier stress accent tier

´ ː

⎜ ⎜

tonal tier

Hː

word form CV template tier

C ⎜ ⎜

consonant frame tier root and inflectable base tier classifying suffix tier case suffix tier

⎜ ⎜

Vː

C

C

⎜ ⎜

p

C

⎜ ⎜ ⎜

k

⎜ ⎜ ⎜ ⎜

u

(17) *kán 40 an. ‘singer, bird’, *kɔ́n an. ‘habitual singer’ → *káːn, *kɔ́ːn in. ‘big(ger) singer, group of singers’ → *kɔːnɔ́- (or *kaːnɔ́-?) group-of-singer-ish, big singer-ish’ (→ PG *hōnaz/ez- n. ‘bigger bird, fowl, hen’ pointing to ™Kōn-os/es-); PIE also had a nominal stem *kánɔn ‘singing’ (> PG *hanō(n) ~ *han-in- m. ‘cock’)

Ø

Figure 9. PIE vowel slot gemination

vowel melody tier detransitive vowel tier

⎜ ⎜ ɛ ⎜ ⎜

‘bare’ vowel tier

\

vowel phoneme moraic tier

/

⎜ ⎜

µµ ⎜ ⎜

vowel melody template tier

VV ⎜ ⎜ ⎜ ⎜

word form template tier stress accent tier

´ ː

⎜ ⎜

tonal tier

consonant frame tier root and inflectable base tier classifying suffix tier case suffix tier

Vowel slot gemination as a plurative marker also occurred with additional suprasegmental vowel mapping; see example (18) and Figure 10. Again, the position of ɔ within the autosegmental analyis, that is, whether it is posited to the left or to the right or whether it is mapped onto the first vowel slot _V1_ or second vowel slot _V2_ of _V1V2_ → _Vː_, is irrelevant to the overall analysis. Both /ɔɛ/ or /ɛɔ/ were realized as a long vowel [ɔː]. The templatic plurative marker also occurred as an allomorphic inflectional plural marker of animate nouns in the core cases absolutive plural, ergative plural, and absolutive associative-plural and absolutive collective-plural, as illustrated by Figures 11 and 12. It was inserted into the vowel slot _V8_ directly before the respective case suffix. As illustrated by Figures 12 and 13, the transfixal plurative plural marking only occurred in the paradigm of animate nominal stems. As illustrated by Figure 13, the transfixal plurative plural marker was not used with inanimate nominal stems, simply because inanimate “singular” forms were transnumeral and did not need plural marking. Collective forms (e.g. *ɗɔ́ruχ) could be used as suppletive plural forms of inanimate nominal stems, but this was only a facultative possibility, not an obligatory rule.41

µµµ

vowel melody moraic tier

word form CV template tier

(18) *χɔ́ui in. ‘product of bird: egg’ → *χɔ́ːui in. ‘group of eggs (in a nest)’ → χɔːuiɔ́_ ‘it is eggs-in-a-nest-ish’ (I will return to the PIE word for ‘egg’ below)

ɔ

Hː C ⎜ ⎜

⎜ ⎜

Vː

C

C

⎜ ⎜

χ

C

⎜ ⎜

⎜ ⎜

⎜

u

⎜ ⎜

i

Ø

Figure 10. PIE vowel slot gemination and suprasegmental mapping

vowel melody tier detransitive vowel tier ‘bare’ vowel tier

ɛ

ɛ

vowel phoneme moraic tier

µ

µ

⎜ ⎜

vowel melody moraic tier

µ

⎜

⎜

V1

V2

⎜

´ ⎜

tonal tier

H C ⎜ ⎜

p

Figure 11. PIE transfixal plural marking

⎜

V1

C

⎜ ⎜

k

C

⎜ ⎜ ⎜

u

⎜ ⎜ ⎜ ⎜ ⎜

V2

In my view, PIE stems like *ɗɛiuɔ́- ‘sky-ish’ were verb-like adjectives. The pertensive-possessive reading was coded by the detransitive marker /ɔ/, whereas the bi-vocalic VMT _V_+_V_ indicated the stative adjectival meaning. *ɗɛiuɔ́- was not necessarily originally derived from *ɗiɛ́u‘shiny one, sky-god’. PIE also had a nonrelational stem *ɗɛ́iu-. The formal relation of *ɗɛiuɔ́- :: *ɗ_i_u- (*ɗiɛ́u-) represented a templatic rule to derive adjectives of this type from relational stems, e.g. (*χákmn‘(big) stony one’ →) *χkmɛ́n- → *χkɛmnɔ́- (> PG *hemna- ‘heaven’). But PIE also had *χkmɛnɔ́ (→ PG *hemina- ‘id.’). The derivation of verblike adjectives will be treated in more detail in the sequel (part 2) of this article. For the time being cf. Pooth 2015. 38 Cf. Darms 1978: 122-133. 39 Pace Kroonen 2013: 240, Darms ibid. 40 I reconstruct *kan- ‘to sing’ (*kan- ~ *k_n-) with a velar *k because I assume that it is also the root of the enlarged inflectable base *k_n-s- ‘to announce, honour in public, praise, recite’, cf. LIV, p. 326. Otherwise change *k to *K = *k or *q. 41 This is a common typological pattern, cf. Rijkhoff 2002. 37

⎜

⎜

word form template tier stress accent tier

consonant frame tier root and inflectable base tier classifying suffix tier case suffix tier

⎜

µ

vowel melody template tier

word form CV template tier

⎜

C

⎜ ⎜ ⎜ ⎜

s

Proto-Indo-European Nominal Morphology. Part 1. The Noun

form *χnɛ́r *χnɛ́r_ɛ *χnɛ́r_ɛ_s *χnɛ́r-ʔ *χnɛ́r-ʔ *χnɛ́r_ɛ_ʔ *χnɛ́r-χ *χnɛ́r_a_χ

gloss

ABS.SG ABS.PL

ERG.PL

ABS.ASS SOC

ABS.ASS.PL ABS.COL

ABS.COL.PL

translation

‘single man (non-agent)’ ‘countable men (non-agent)’ ‘countable men (agent)’ ‘man & companion; man in company’ ‘accompanied by man/men’ ‘man & companion(s)’ or ‘men & co.’ ‘group of men’ or ‘man in a group’ ‘group(s) of men’ or ‘men in a group’

Figure 12. Number and nominal aspect marking of animate stems

9

overall lexical paradigm. The overall PIE verbal paradigm consisted of a set of multiple binyans. On the other hand, each nominal stem had an underlying, basic inflectional type, whereas other inflectional types had a more derivational status. However, each overall nominal paradigm also consisted of a set of more than one inflectional type (minimally two). PIE verbal “binyanim” were inflectional aspectual templates,42 whereas nominal inflectional types were less inflectional, but more derivational.

4.1 Stem gradation form *ɗɔ́ru *ɗɔ́ru-ʔ *ɗɔ́ru-ʔ *ɗɔ́ru-χ *ɗɔ́ːru

gloss ABS

ABS.ASS SOC

ABS.COL/ PL ABS.PLT

translation ‘single piece of wood; woods’ ‘wood(s) & co.’ ‘wood(s)-accompanied’ ‘group(s) of wood(s); wood(s) in a group’ ‘id.’ or ‘bigger wood’

Figure 13. Transnumerality of inanimate stems

stem *kɛ́rɗ*krɛ́ɗ-

gloss heart.BASEheart.LOC/RL-

*krɗɛ́_

heart.OBL-

translation ‘heart’ ‘in/at/by heart’/ ‘s.o.’s heart’/ ‘heart-ish’

Figure 14. PIE nominal stem gradation I

stem *pɛ́ku*pkɛ́u-

gloss sheep.BASEsheep.LOC/OBL/RL-

*pkú-

sheep.RL

*pkuɛ́_

sheep.RL.OBL-

translation ‘domestic animal, sheep’ ‘in/at/by domestic animal’/ ‘s.o.’s domestic animal’/ ‘domestic-animal-ish’ ‘s.o.’s domestic animal’/ ‘domestic animal-ish’

Figure 15. PIE nominal stem gradation II stem *χnɛ́r*χnrɛ́_

gloss man(kind).RLman(kind).RL.OBL_

translation ‘man’

Figure 16. PIE stem gradation III

4 Internal modification The PIE word form template, abbreviated WFT, had the status of a templatic morpheme. The word form template determined the position of the vowel(s) of the vowel melody combined with the position of the word form accent and tone on one vowel within the word form. The word form template was thus meaningful, and a particular position of the vowels within the word form determined a particular meaning. The word form template itself belonged to a superordinate paradigmatic set of word form templates. In the nominal inflection, this superordinate set of templates is conveniently termed the “inflectional type”. In the verbal inflection, however, I make use of the term “binyan” which is borrowed from Classical Hebrew grammar. It has to be recognized that both a PIE inflectional type as well as a PIE “binyan” are not identical to the

Each PIE noun had minimally two, maximally three nominal stems (or four including one allomorphic stem: see Figure 15) displaying different vowel melodies which paradigmatically belonged together. This stem gradation made up a system of nominal internal modification running in parallel with the system of verb gradation;43 see Figures 14, 15, 16. I suggest to use the term BROKEN locative instead of “endingless” locative for a suffixless PIE locative, e.g. *krɛ́ɗ (> Vedic śrad- of śraddhā-), whereas I suggest to use the term SOUND locative for the form marked by the suffix -i, e.g. *χnɛ́ri. These two terms are borrowed from Classical Arabic grammar.44 The stem of the broken locative and the oblique stem were internally derived from the underlying basic stem (base), as illustrated in Figures 14 and 15. The nominal gradation presented in these Figures is well-established and common sense, except that the traditional concept of the oblique stem *krɗɛ́_ is different (namely ₠k̑r̥d-). But the traditional analysis failed to notice PIE transfixation. This should now be regarded as an analytical mistake. The stem of the PIE relational nouns,45 that is, the relational nominal stem, was identical to the stem of the broken locative. Relational nouns, as defined, usually convey an inherent semantic possessee role, that is, they bear the meaning of being possessed by a contextual (external) possessor. Crosslinguistically, such a class of nouns is quite common. Although all IE languages lack such a class of nouns, it is very reasonable to reconstruct them for PIE.46 Claiming that the absence of such a class of noun in all IE languages would constitute a comparative proof that PIE must also have lacked it, is just an illegitimate logical deduction and an unfortunate dogmatic misuse of the comparative method. I will return to the polysemy of the PIE relational nouns in a more detailed manner below. Relational nouns had only two nominal stems, namely the relational nominal stem and an additional oblique stem, as illustrated by PIE *χnɛ́r ‘man’47 in Figure 16. Within the Cf. Pooth 2015, see the sections on aspect. Cf. Pooth 2015. 44 Recall that the Arabic broken plural is a plural form which lacks plural suffixes and is solely marked by a particular transfix. 45 PIE relational nouns, e.g. *pɔtɛ́i- ~ *pɔtí- ‘s.o.’s master’, were first identified as such by Tremblay 2003. Tichy 1995 recognized the distinction of relational agent noun, e.g. *ɗʕtɛ́r- ‘someone’s/something’s giver, giver of s.th.’, from habitual nominals *ɗɛ́ʕtɔr- ‘giving’ and noticed the relational semantics of the former, albeit her label “situativ” is too specific. 46 Cf. Tremblay 2003, Pooth 2015. 47 Perhaps derived from a heteroclitic *χánr in. ‘mass of faces, persons: mankind’ with the same root as *χán-t- (CL1A) an. ‘face, front’ and maybe *χán- an. ‘face, front, person’, cf. maybe *χiu-χán- an. ‘life-ofvigour-person, young person’, cf. Vedic ánu- m. ‘person’, if not derived from *χ_n-ʔ- ‘to breathe’, but from *χ_n-. 42 43

Roland A. Pooth

10

broken locative/ traditional comparative model of PIE noun morbase/ relational stem/ oblique stem/ pholoy, the respective stems have been termed abs. sg. primary oblique stem secondary oblique stem IE reflexes “strong” stem, “endingless locative” stem, “weak” or 48 *ɠɔ́nu (*ɠɛ́nɔu) *ɠɛ́nuɔ_ cf. Hitt. gēnuwaš “weakest stem” stem. However, the use of the tradi*ɗɔ́ru *ɗɔrɛ́u *ɗɔruɛ́_ cf. Gk. δουρός tional terms is not maintained here. Instead, I make *pɛ́ku *pkɛ́u *pkuɛ́_ *liɛ́ːkwr *liɛkwɛ́n *liɛkwnɛ́_ cf. Vedic yaknás use of the terms BASIC STEM (BASE), BROKEN LOCATIVE *pχtɛ́r *pχtrɛ́_ or RELATIONAL STEM, and OBLIQUE STEM (primary and *ʕrɛ́ːɠ *ʕrɛ́ɠɛ_ cf. Lat. rēx, rēgis secondary oblique stem). Note that the broken loca*uɔ́ːkw *uɛkwɔ́_ cf. YAv. vāxš, vaco tive stem was used as primary oblique stem of nonFigure 17. PIE nominal stem gradation IV relational nouns in the case of the “proterkinetic” or “proterodynamic” inflectional type, e.g. *pɛ́ku- → *pkɛ́u-, whereas the secondary oblique stem *pkuɛ́_ consonant frame. I have decided to keep on claiming that belonged to the relational stem *pkɛ́u- ~ *pkú-. To distinthis morphological strategy must be reconstructed and, guish both oblique stems, I use the term primary oblique indeed, is quite realistic. However, it is rather improper to stem vs. secondary oblique stem of relational stems use the term “metathesis” for PIE vowel transposition. A (glossed RL.OBL), as illustrated by Figures 16 and 17. morphological transposition or positional change of segments has a different motivation. This strategy may per4.2 Internal inflection and derivation haps ultimately go back to prior cases of spontaneous methathesis. But once such a morphological strategy has It is a remarkable typological peculiarity of PIE that deribeen established as such, the notion of vowel transposition vational nominal stems as well as inflected word forms implies an underlying templatic structure on a word form were thus derived from underlying nominal stems or intemplate tier with various vowel positions or vowel slots flected word forms by transferring or transposing the (*_V_) where the respective vowels switch positions not just vowel or vowels to different vowel slots within the word for the phonological fun of it, but for morphological reaform on the word form template tier. This morphological sons, that is, as an inflectional and derivational morphostrategy is here referred to by the label VOWEL logical means. TRANSPOSITION. An example is: PIE vowel transposition, at least synchronically, was of(19) *χnɛ́r- → oblique stem *χnrɛ́_ (see Figure 16) with underten, but not always and not necessarily accompanied by an ́ ́ lying word form templates *CCVC- → *CCCV_ additional internal change of the accent position within Comparative evidence strengthens the inference that PIE the word form (by accent shift); see the following example: made abundant use of this morphological strategy. I admit (21) *pɛ́ːku- in. ‘livestock, cattle’ → *pɛkɛ́u- → *pɛkuɛ́_ with that such a morphological strategy is typologically rare underlying word form templates C_V1́ V2_Cu- → and poorly attested, if it is attested in this specific way at ́ ́ C_V 2 _C_V1_u- → C_V2_Cu_V1 _ all. However, a quite similar and comparable strategy, The PIE word form accent was part of the word form namely morphological metathesis is well-attested among template. The PIE accent was thus somehow “free”, which the languages of the world. A very similar morphological is to say that it was unpredictable from syllable structure vowel transposition is found, for instance, in the Oceanic or word form structure. In word forms with more than one language Rotuman:49 full vowel or semi-vowel, one of these two full vowels or a (20) Rotuman (Oceanic, Austronesian): piko (CV1.CV2) ‘lazy’ semi-vowel had to bear a contrasting high tone (high → piok (CV1V2C) ‘lazy’, rotuma (CV3.CV1.CV2) ‘Rotuma’ pitch) stress (intensity) accent, opposed to an unstressed 50 → rotuam (CV3.CV1V2C) ‘Rotuma’ (non-intense) lower tone of the other vowel or vowels in The Rotuman switch of segments is much closer to phothe word form. The word form accent, therefore, was not a nological metathesis, but it also has a morphological motiproperty of any morphological unit other than the word vation and is thus comparable to what is described as form template morpheme, and it provided grammatical vowel transposition for PIE here. There are a several other distinctions. The accent position within the word form, languages, e.g. Amarasi (Austronesian), that make use of combined with other morphological means, was definitory morphological metathesis.51 In any case, claiming that the for the identification of a given form as belonging to a absence of such a morphological strategy in IE languages particular paradigmatic combination of vowel melodies, would constitute a comparative proof that PIE must also that is, to a particular paradigm. have lacked it, is just an illegitimate logical deduction and For instance, verb-like adjectives were derived from a anachronistic circular reasoning. Within a transfixing and respective underlying 3sg durative-stative detransitive otherwise agglutinating language, such a morphological intransitive verb form (which was the derivational base) strategy makes perfect sense, because the vowels are indeby accent shift to the word final vowel: pendent from the consonant frame and any vocalic posi(22) *lɛ́uqɔ- ‘s.o. is, was bright, shining’ (verb form of the sotional switch does not at all affect the grammatical meancalled “stative Narten middle” type) → *lɛuqɔ́- verb-like ing which is coded by the agglutinating morphology on the

adjective ‘(it is) bright, shining’

This form is identical to the plurative (collective and generic) nominal stem. 49 Cf. Besnier 1987. 50 Cf. Besnier 1987: 201-233. 51 Cf. Edwards 2014 on Amarasi; Thompson & Thompson 1969 48

PIE relational adjectives were either identical to the relational stem and the ‘broken’ locative stem or were derived from the sound locative form by accent shift to the word final vowel or by additional ‘broken’ vowel transposition, a

Proto-Indo-European Nominal Morphology. Part 1. The Noun

morphological strategy that will be treated in more detail below.52

(23) *ɗɔ́ru ‘wood’ → *ɗɔrɛ́u ‘broken’ loc. form ‘in wood, at wood’ = relational adjective stem ‘wooden, wood-ish, pertaining/associated to wood, possessing wood’

(24) *ɗɔ́ru ‘wood’ → *ɗɔrú- relational adjective ‘wooden, wood-ish, pertaining/possessing wood, associated to wood’; cf. adjective dārúm acc. sg. (RV 7.6.1d) (25) *ɗɔ́ru ‘wood’ → *ɗɔ́rui loc. ‘in/at wood’ → *ɗɔruí- ~ *ɗɔruɛ́i- relational adjective ‘wooden, wood-ish, pertaining to wood, associated to wood, possessing wood’

But recall that in the system of verbal root and pattern morphology inflection it is quite impossible to identify a particular derivational base other than the lexical discontinuous inflectional base (inlcuding the root), e.g. *g̈w_n- ‘to slay, kill, hunt, beat’. In the nominal inflection, on the other hand, the lexical item included the vowels of the vowel melody (e.g. *ɗɔ́ru-) and the lexical entry was thus different from the inflectional base or root. The following rules of PIE VOWEL TRANSPOSITION are reconstructed here. These rules are dated here to a synchronic spoken level of the unitary PIE proper parent language, that is, the stage with the highest level of PIE lingustic unity and homogenity, mutual understanding, and a kind of ‘PIE tribal identity’. The term ‘complete transposition’ or ‘unbroken transposition’ refers to a complete transferal and full transposition of a vowel from an underlying position within the word form to another vowel slot of the word form template. This morphological strategy can be illustrated by the following examples: (26) *pɛ́ku- an. ‘domestic animal, sheep’ → *pkɛ́u- with underlying word form templates C_V́1_Cu- → CC_V́1_u-

(27) erg.-gen. sg. *ʔsɛ́nts ‘existing, real, true’ → abl.-gen. *ʔsntɛ́s (28) abs. sg. *pɛ́rtu in. ‘its function is to go through: firth’ → ‘broken’ loc. *prtɛ́u → abl.-gen. *prtuɛ́s (29) erg.-gen. sg. *g̈wɛ́ns (→ Vulgar Post-IE *gwɦé:(n) by analogy) ‘slayer’ → abl.-gen. *g̈wnɛ́s (> Vedic °ghnás)

In complete or ‘unbroken’ vowel transposition the vowel /ɛ/ was completely transferred and transposed to a different vowel slot within the word form. In the following figures 18ff., the segmental transfer is indicated by arrows and brackets. The analytical unit (V) that is transferred and transposed to another position within the word form is given in parentheses. The arrows ( →) represent the segmental transfer and transposition. Example (26), as analyzed in Figure 18, had a monovocalic underlying vowel melody template (*_V_) and an underlying word ́ form template *C_V_C-_S-, where C_ is used as a cover symbol for any consonant of the inflectable base and _S- is used as a cover symbol for any consonantal classifying suffix. The position of the vowel was assigned to a specific vowel slot within the word form on the word form template tier. Therefore, the morphological strategy of internal inflection and internal derivation can be seen as a phenomenon that operated on the word form template tier; see Figure 18. In Figures 18ff., the word form CV template tier is marked by the grey shading. 52

Cf. Pooth 2015.

11

vowel melody tier detransitive vowel tier ‘bare’ vowel tier

ɛ

⎜

vowel phoneme moraic tier

µ

⎜

vowel melody moraic tier

µ

⎜

vowel melody template tier

(V)

→ V

word form template tier stress accent tier

(´)

→ ´

tonal tier

(H)

→ H́

(V)

→ V

word form CV template tier consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

⎜ ⎜

⎜ ⎜

⎜

⎜

⎜

C ⎜ ⎜ ⎜

⎜

C

C

⎜ ⎜ ⎜ ⎜

⎜ ⎜

p

k

u

C

⎜ ⎜ ⎜ ⎜ ⎜

s

Figure 18. Complete transposition of underlying act. sgt. stems vowel melody tier detransitive vowel tier

ɔ

⎜ ⎜ ɛ ⎜ ⎜

‘bare’ vowel tier vowel phoneme moraic tier

µµ

/

⎜

vowel melody moraic tier

µ ⎜

vowel melody template tier

(V)

word form template tier stress accent tier

⎜

(´)

→ ´

tonal tier

(H)

→ H

(V)

→ V C C

word form CV template tier consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

→ V

⎜

⎜ ⎜

⎜

C ⎜ ⎜ ⎜

ʔ

C

⎜ ⎜ ⎜

s

⎜

⎜

⎜

⎜ ⎜ ⎜

⎜ ⎜ ⎜

n t

C

⎜ ⎜ ⎜ ⎜ ⎜

s

Figure 19. Complete transposition of underlying DTR SGT stems

The same system of complete vowel transposition was applicable to the accented full vowel /ɔ́/. According to what may be termed an ‘unbroken’ transposition rule, the accented full vowel was transposed in its entirety, in parallel with the accented /ɛ́/ of the agentive-active forms. This type of transposition can be illustrated by the following Figure 19. This rule is evident from external comparative evidence; see example (30) below.

(30) *ʔɗɔ́nt- detransitive relational nominal stem ‘always eating, biting; being eaten, bitten’ (with loc. *ʔɗɔ́nti) → abl.-gen. *ʔɗntɔ́s (> Vedic nom. sg. m. dán, abl.-gen. datás, etc.)

Ad (30a): PIE also had another detransitive relational stem *ʔɔɗɛ́nt- ‘someone’s tooth’, abl.-gen. *ʔɗɔ́nts ~*ʔɗntɔ́s of the inflectional type described below, whence Greek ὀδών, but cf. Aolic ἔδοντες; PIE further had a corresponding agentive-active stem *ʔɗɛ́nt- > Lat. dēns ‘tooth’. It is inferential that all these stems merged in Vulgar Post-PIE. The following example (31) can also illustrate a complete transposition of accented /ɔ́/:

Roland A. Pooth

12

(31) *ʔsɔ́nt- detransitive relational nominal stem ‘always being there’ and ‘having been there; resultant from being there’, as confirmed by Latin sōns ‘guilty’, derived from ‘having been there’ (besides a habitual agentive reading detransitive participles also had a resultative nonagentive reading) → abl.-gen. *ʔsntɔ́s

The same system of vowel transposition was applicable to the unaccented full vowel /ɔ/ in the acrostatic inflectional type with underlying bi-vocalic vowel melody template; see examples (32) and (33):

whereas the discontinuous vowel /ɔ/ remained in its original position. This transposition was combined with additional accent shift to the new position of the ‘bare’ vowel /ɛ́/. Thus, in a sense, in this type of tranposition the detransitive vowel was a remnant or left-over of the underyling derivational base (e.g. *ɗɔ́ru, gen. *ɗɔ́ru-s). This type of tranposition is illustrated by Figure 20. vowel melody tier detransitive vowel tier

(32) *nɛ́pɔt- an. ‘decendant(s), offspring’ with ‘sound’ loc. *nɛ́pɔti → abl.-gen. *nɛ́ptɔs (→ continued by Vedic abl.gen. sg. náptur53);

With regular acrostatic -ur in parallel with abl.-gen. sg. mā́tur, bhrā́tur, pátyur. 54 For an overview of the former problems that can be solved by this inference cf. Fellner & Grestenberger 2015; for the function of the Hittite participles in -ant- cf. Frotscher 2013. 53

⎜ ⎜ (ɛ) ⎜ ⎜

‘bare’ vowel tier vowel phoneme moraic tier

(33) *b̤ɛ́rɔnt-, loc. *b̤ɛ́rɔnti, abl.-gen. *b̤ɛ́rntɔs (> Vedic bhárantwith loc. sg. bháranti, abl.-gen. bháratas)

Thus, PIE had agentive-active participles such as, e.g., *ʔsɛ́nt- ‘existing, being there, real, true’ and detransitive counterparts, e.g. *ʔsɔ́nt- ‘having been there, having the property of having been there’ (> Latin sōns ‘guilty’). The latter had a labile detransitive, including anticausative and factitive or indirect causative meaning. They are continued by the Hittite voice-neutral participles in -ant-.54 The morphological strategy of vowel transposition further included the internal change of an underlying form with discontinuous detransitive marker /ɔ/ mapped onto the accented ‘bare’ vowel /ɛ́/, that is, with stressed /ɔ́/, to a derived form with two vowels in the word form including unaccented /ɔ/ and an accented vowel /ɛ́/ in two different vowel slots. This internal modification is labeled ‘broken’ vowel transposition here. The ‘breaking’ of /ɔ́/ to /ɛ́/ and /ɔ/ elsewhere ran in parallel with the internal change of a plurative form with underlying double (geminate) vowel slots (_VV_ → _Vː_) to a form with two vowels in different vowel slots (_V_ and _V_), which is labeled ‘geminate breaking’ here. Thus, ‘broken’ vowel transposition is a more intricate type of vowel transposition. Due to the discontinuous and autosegmental status of the detransitive vowel /ɔ/, an underlying segmental string with accented vowel */Cɔ́C-/, e.g. */ɗɔ́ru-/ is analyzable as containing two input vowels in its underlying segmental string structure, namely the ‘bare’ vowel /ɛ́/ and the discontinuous vowel /ɔ/, both being mapped onto the very same vowel slot _V_ of the word form template; see Figure 8 above. Under this perspective, that is, with regard to the two vowels on the vowel input tier, it turns out to be understandable that one of these two vowels could be transposed to a different position within the word form, while the other remained in the position of the underlying derivational base. This type of vowel transposition is termed ‘broken’ vowel transposition, because the underlying segmental string */Cɔ́C-/, which thus consisted of two underlying vowels, mapped onto each other, was ‘broken’ up again into its two different vowels, so-to-speak. There was a first subtype of this ‘broken’ transposition, where the ‘bare’ vowel was transposed to a different position,

ɔ

ɛ ⎜

µɔ(µ)

→ µ

/

⎜

vowel melody moraic tier

⎜

µɔ

µ

⎜

⎜

vowel melody template tier

Vɔ(V)

→ V

word form template tier stress accent tier

(´)

→ ´

(H)

→ H

Vɔ(V)

→ V

⎜ ⎜ ⎜

tonal tier word form CV template tier consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

⎜ ⎜

C ⎜

⎜

⎜

⎜ ⎜

ɗ

⎜

C ⎜ ⎜

r

C

⎜ ⎜ ⎜

u

C ⎜ ⎜ ⎜ ⎜ ⎜

s

Figure 20. Broken transposition _ɔ́_Ø_ → _ɔ_ɛ́_

As illustrated by Figure 20, the ‘bare’ vowel /ɛ/ (including its moraic unit, both given in epentheses) was transposed to another position within the word form, leaving the other moraic unit µɔ of /ɔ/ in its underlying position, that is, in the root vowel slot. More examples are: (34) *liɔ́kwr in. ‘animal liver, roe’ → loc. *liɔkwɛ́n, abl.-gen. *liɔkwɛ́ns, cf. Lat. iocur, iocineris, Lat. na. sg. n. iocur (Plin. Nat. 32.50), gen. iocineris (besides iecur, iecore < PIE *liɛ́ːkwr, loc. *liɛkwɛ́n, abl.-gen. *liɛkwnɛ́s55) (35) *uɔ́ɗr in. ‘water’ → loc. *uɔɗɛ́n, abl.-gen. *uɔɗɛ́ns (as allomorph of acrostatic *uɛ́ɗnɔs), cf. Goth. gen. sg. n. watins (nom. wato); the allomorphic use of some inflectional types must have yielded some variation

There was another subtype of ‘broken’ vowel transposition, at least synchronically. Here, the accented ‘bare’ vowel */ɛ́/ remained in its underlying position and the discontinuous ɔ was transposed to another position within the word form. This kind of vowel transposition can be termed the ‘static broken’ vowel transposition. It is illustrated by Figure 21 (and by Figure 8). I reconstruct an animate nominal stem PIE *ɠɛ́nu- an. ‘human joint, angle, knot: chin, knee’, abl.-gen. *ɠnɛ́us as the corresponding underlying active-agentive and unmarked nominal stem. The traditional difficulty of reconstructing a single PIE paradigm56 for all attested IE forms for ‘chin, knee’ is only met with within a monoparadigmatic approach. The IE Vedic yaknás still continues the original accent position, pace Viti 2015: 102, fn. 3, cf. Pooth 2014d; cf. further Weiss 2009: 240, Rix 1965. 56 Cf. Kloekhorst 2008: 467f.: “[W]ith these three ablaut grades attested in the IE languages, it is difficult to reconstruct a PIE paradigm.” 55

Proto-Indo-European Nominal Morphology. Part 1. The Noun

13

gender confusion of Gk. γόνυ n., etc., but γένῡς f. ‘chin’, and the parallel Hittite gender confusion speaks in favour of a merger of the given two paradigms. Young Avestan acc. sg. žnūm (V 8,61-63, V 9,22ff.) further points to a corresponding “hysterodynamic” (HD) stem *ɠnɛ́u ~ *ɠnú an. 57 Kloekhorst’s suggestion “(n. > c.)” to explain the common gender of Hitt. acc. sg. c. ge-nu-un is of lower plausibility. Instead, this form directly continues PIE *ɠɛ́num (all.-dat. sg.) or *ɠnɛ́um ~ *ɠnúm, or may also continue a merger of both forms. The PIE stem *ɠnɛ́u- ~ *ɠnú- an. was thus the relational (possessed) nominal stem with a meaning ‘someone’s human joint: knee, chin’. As already outlined above (cf. Figure 8), the oblique stem of the corresponding inanimate stem *ɠɔ́nu in. ‘joint, angle, knot’, namely *ɠɛ́nuɔ- had an underlying vowel melody template *_V_V_ and was identical (or similar) to verb-like adjective stems with a stative meaning, e.g. *ɠɛ́nuɔ- ‘jointish, angle-ish, knotty’ (besides allomorphic *ɠɛnuɔ́- ‘id.’). vowel melody tier detransitive vowel tier

ɔ

ɔ

vowel phoneme moraic tier

µ (µ)

→ µ

vowel melody moraic tier

µ

‘bare’ vowel tier

ɛ

⎜

/

⎜

(µ)

⎜

⎜

(Vɔ)V

→ Vɔ

⎜ ⎜

word form template tier stress accent tier

⎜ ⎜

´

⎜

⎜

tonal tier

consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

⎜ ⎜ ⎜

⎜

vowel melody template tier

word form CV template tier

⎜ ⎜ ⎜

⎜

H

C ⎜ ⎜

However, the arrows from left to right, that is, from non-oblique stem to derivational oblique stem (as given in Figure 21) can only be represented in such a direction within a synchronic inner-PIE analysis. From a diachronic pre-PIE to PIE perspective it is more plausible to infer that the actual derivational direction took the opposite direction. In this context it is relevant to recognize that, as a rule, the PIE basic and underlying non-relational nominal stems had a trochaic metrical structure (σ́ σ) and were subject to a high-ranking morphological constraint which can be termed the ‘minimum-of-syllables constraint’ (MINSYL). The rule was: ‘Make the basic and underlying non-relational stem a monosyllable, if possible.’ Therefore, it is inferential that the origin of the suprasegmental mapping of the detransitive vowel /ɔ/ was probably simply a former continuous suffix **-ɔ or the like and a disyllabic form of the shape **Cɛ́Cɔ or the like. It can now be suggested that this form underwent a morphological metathesis which was very similar, if not almost identical, to what is found in Rotuman (Oceanic, Austronesian) (see example (19) above). This metathesis was ultimately triggered by the strong regularity of the MINSYL constraint and further by analogy to the other stems, perhaps pre-PIE58**wɛ́kwɔ (CV1.CV2) ‘speaking-detransitive’ (or similar) → **uɛ́ɔkw (CV1V2C) ‘id.’ > *uɔ́kw :: *ukwɔ́ in parallel to **wɛ́kwV (CV1.CV) ‘speaking-agentive’ > PIE *uɛ́kw (CVC.) ‘id.’, but **wVkwɛ́ (with an iambic metrical structure σ σ́) > PIE *ukwɛ́_. Such a scenario is quite similar to the one found in Rotuman (Oceanic, Austronesian): piko (CV1.CV2) ‘lazy’ → piok (CV1V2C) ‘lazy’, rotuma (CV3.CV1.CV2) ‘Rotuma’ → rotuam (CV3.CV1V2C) ‘Rotuma’. 59 In my view, such an emergence of the PIE templatic system of vowel transposition from a prior more linear morphological system is quite plausible. But it cannot be further investigated here, since it clearly surpasses the possibilities of a potentially reliable linguistic reconstruction. Nevertheless, within a synchronic account of PIE morphology, the direction must in any case be from non-oblique to oblique stems. Thus at least synchronically the detransitive plurative *uɛ́ɗɔr type was internally derived from an underlying base with /ɔ́/ in the root or inflectable base vowel slot by ‘static broken’ vowel transposition. Derivatives of this kind may thus be termed ‘BROKEN’ PLURATIVES, in parallel with the label “broken plural” in Classical Arabic; for example:

⎜

⎜

⎜

(Vɔ)V

C

C

⎜

⎜

⎜

ɠ

→ Vɔ

C ⎜ ⎜

⎜

⎜ ⎜

n

⎜ ⎜ ⎜

u

s

Figure 21. Acrostatic broken transposition _ɔ́_Ø_ → _ɛ́_ɔ_ vowel melody tier detransitive vowel tier

(ɔ)

→

vowel phoneme moraic tier

(µ)

→ µµ

vowel melody moraic tier

⎜ ⎜

⎜ ⎜ ⎜

‘bare’ vowel tier

tonal tier

consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

⎜

µ

C ⎜ ⎜ ⎜

s

→ V

⎜ ⎜ ⎜ ⎜ ⎜ ⎜

(Vɔ)

(36) *uɔ́ɗr in. ‘water’ → *uɛ́ɗɔr in. ‘water(s)’

⎜

(Vɔ)

word form template tier stress accent tier

word form CV template tier

/

⎜

vowel melody template tier

ɔ

⎜ ⎜ ɛ ⎜ ⎜

⎜ ⎜

´

⎜

H

⎜

C

⎜ ⎜

u

C

⎜ ⎜

χ

C

⎜ ⎜ ⎜

n

→ V

C

C

⎜ ⎜ ⎜ ⎜

⎜ ⎜ ⎜ ⎜ ⎜

u

s

Figure 22. ‘Reversed broken’ transposition _ɔ_ɛ́_ → _Ø_ɔ́_

These pre-PIE forms are fictitious and cannot be reconstructed. They are given here to illustrate the possibiliy of a Rotuman-like emergence of PIE nominal templatic morphology. 59 Cf. Besnier 1987: 201-233. 58

57

Cf. Tremblay 2003: 577.

The transpositional relation between the above given stems also showed up in a reversed way modifying underlying “hysterodynamic” relational stems which displayed a stressed ‘bare’ vowel */ɛ́/ in a position outside the root or inflectable base vowel slot and an unaccented */ɔ/ in a vowel slot to its left slot; see Figure 22. In this ‘reversed broken’ type of vowel transposition, the unaccented /ɔ/ was transferred to the vowel slot of the accented vowel */ɛ́/ and was thus once more

Roland A. Pooth

14

suprasegmentally mapped onto it, yielding stressed */ɔ́/, but in this case the mapping took place outside the root vowel slot. As correctly seen by Tremblay,60 Vedic and Avestan reflect a distinction between two PIE stems for ‘master’, namely *pɔ́ti- nonrelational ‘master’, abl.-gen. *pɛ́tiɔs (acrostatic) vs. *pɔtí- relational ‘someone’s master, master of s.o./s.th.’, abl.-gen. */ptɔ́is/ (> Proto-Indo-Iranian *patáiʃ > Gathic Avestan patōiš, etc.). I suggest that the latter was of the same inflectional type as PIE *sɔuχnú- ‘someone’s son, lit. somone’s the-one-who-has-to-be-born’, abl.-gen. *suχnɔ́us. In my view, this PIE relational nominal stem was derived from the inflectable base *s_u-χ- ‘to be made pressed out, be born’ with a factitive-detransitive enlargement suffix *-χ-, ultimately derived from the root *s_u- ‘to press’. Note that the fricative *-χ- can be based on Toch. B suwō ‘sow’ pointing to a stem ™suu̯áh2-. As already argued elsewhere,61 it is inferential that the PIE word for ‘s.o.’s son’ had the marker ɔ throughout its nominal paradigm. The Gothic gen. sg. m. sunaus including its ending -aus points to ™sunóus with o-grade of the suffix. In my view, PIE *sɔuHnú-, *suHnɔ́us was first changed to *sɔunú-, *suχnɔ́us due to SAUSSURE’S EFFECT, that is, PIE *CɔRH.CV_ > Vulgar PostPIE *CɔR.CV- with loss of *H (= ʔ χ ʕ) in that position. Subsequently, the loss was paradigmatically transferred to the oblique forms in some (but not all) Vulgar Post-PIE dialects. After that, the zero grade of the root was paradigmatically leveled, either as Vulgar Post-PIE *suχnú- or *sunú- from either *suχnɔ́us- or *sunɔ́us. In my view, this is the most plausible solution for the problem that the IE languages reflect both an equation formula ₠suHnú- (cf. Vedic sūnú-, etc.) and ₠sunú- (cf. Gothic nom. sg. m. sunus, etc.). Recall that it is probably a bit too naїve to reproject one of these two variants directly to the PIE parent language (see section 1). Excursus: I suggest that Dybo’s law 62 was a different sound law and should not be confused with Saussure’s effect and its subsequent analogical paradigmatic levellings. Since OHG dūmo, but OSwed. þumi ‘thumb’ is a body part, this variants presumably arose from a relational paradigm *tɔuHmɛ́n ‘s.o’s thumb’, abl.-gen. *tuHmɔ́ns ~ *tuHmnɔ́s exhibiting the same inflectional type as the precursor of PG *sunu-. Thus these variants simply continue the same variation due to the same parallel levelings. This example may thus not be a case of Dybo’s law at all. Furthermore, PG *ɣlana- n. ‘shine’63 may simply belong to a root *ɢ̈l_n-, of which both *ɢ̈l_n-d̤- and *ɢ̈l_n-ɗ- are enlarged derivatives;64 thus PG *ɣlana- may not be a case of this law, too. It is perhaps possible to suggest that Dybo’s law was restricted to a loss of the glottal stop before R = m n l r in pretonic position immediately before the syllable boundary (*CVh1.RV́ > *CV.RV́) which is evident form OE delu ‘tit’, OHG tila ‘id.’ < PG *delō, but Gk. θηλή f. ‘breast’ pointing to ₠dɦeh1lah2-.65 Latin fūmus, Cf. Tremblay 2003: 246 §3.2.1. Cf. Pooth 2015. 62 Cf. Kroonen 2013: xx-xxi for Germanic, Weiss 2009: 99 for Latin, Zair 2006, all three with further references. 63 Cf. Kroonen 2013: xx-xxi. 64 Cf. LIV, p. 200. 65 Cf. Neri 2005, Zair 2006 (restriction to *h1 and *h3). 60 61

vowel melody tier detransitive vowel tier

ɔ

‘bare’ vowel tier

(ɛ)

→

vowel phoneme moraic tier

(µ)

→

vowel melody moraic tier

⎜ ⎜

⎜ ⎜

vowel melody template tier

⎜ ⎜ ɛ ⎜ ⎜

µµ

/

⎜

µ ⎜

(V)

→

word form template tier stree accent tier

(´)

→

tonal tier

(H)

→

H

(V)

→

V

word form CV template tier consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

⎜ ⎜ ⎜

C ⎜ ⎜ ⎜

d̤

⎜

C

⎜ ⎜

g̈

V

⎜ ⎜

´

⎜

⎜

C

⎜ ⎜ ⎜ ⎜

m

C

⎜ ⎜ ⎜ ⎜ ⎜

Ø

Figure 23. ‘Reversed broken’ transposition _ɛ́_ɔ_→ _Ø_ɔ́_

however, seems to suggest that the fricative *-χ- was not affected by this law.66 Perhaps the voiced fricative was also not affected, because the short vowel of PG *kwiwa- (> Gothic qius) may have arosen simply by analogy to its variant PG *kwikwa-, which may go back to a reduplicated form *ɠwiɠwʕ(u)ɔ́ with regular loss of the fricative. This may also hold true for the corresponding forms, since the analogy could have happened very early. Note by the way that the reduplicated verb-like adjective resembled the PIE reduplicated verb form *ɠwiɠwʕ(u)ɔ́. Dybo’s law seems to be a quite restricted phenomenon but may have given rise to sporadic analogical pretonic shortenings in the Vulgar Post-PIE dialect or variant bundle. End of excursus. Furthermore, the relation between */ɔ́/ and */ɛ́/ + */ɔ/ of the ‘static broken’ vowel transposition also showed up in a reversed way. Thus, 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. It is illustrated by Figure 23. Note that the corresponding animate stems were acrostatic, not holokinetic; see examples (37) and (38): (37) *uɛ́ɗɔr in. ‘water(s) (generic, collective-plural) → loc. *uɗɔ́n (> Ved. udán) (→ abl.-gen. *uɗnɔ́s :: verb-like adj. *uɗnɔ́_ ~ *uɗrɔ́_ ‘it is water-ish, in/at/by water’

(38) *d̤ɛ́g̈ɔm in. ‘ground, earth, classified as extended, flat’ (Hitt. na. sg. n. tēkan ‘earth’) → loc. *d̤g̈ɔ́m (> Hitt. loc. tagān) (→ abl.-gen. *d̤g̈mɔ́s > Hitt. gen. taknāš)

Two parallel types of ‘broken’ vowel transposition were applicable to the agentive and detransitive plurative word forms, that is, the ones with basic or underlying geminate vowel melody template *_VV_ (realized as *_Vː_) in the vowel slot of the inflectable base. Here the vowel geminate *_ɛ́ɛ_ or the vowel sequence *_ɔ́ɛ_ (which were both realized as long vowels) were ‘broken’ up into combinations of two short vowels (*_V_ + _V_). This is illustrated by Figure 24 and Figure 25.

66

Cf. Weiss 2009: 99 for Latin with references.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

15

vowel melody tier detransitive vowel tier ‘bare’ vowel tier

ɛ

vowel phoneme moraic tier

µ(µ)

→ µ

µ(µ)

→ µ

V(V)

→ V

⎜

⎜

vowel melody template tier

⎜

⎜

⎜

⎜

word form template tier stress accent tier

⎜ ⎜ ⎜ ⎜ ⎜

´

⎜

tonal tier

consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

⎜

⎜

vowel melody moraic tier

word form CV template tier

ɛ

\

⎜

H

C ⎜ ⎜ ⎜

l

⎜

C

V(V)

⎜ ⎜ ⎜

→ V

C

c

⎜ ⎜ ⎜ ⎜

⎜ ⎜

i

kw

N

Figure 24. Broken transposition of underlying active plurative stems; key: -N- is used as a cover symbol for -r- ~ -n-

vowel melody tier detransitive vowel tier

(ɔ)

‘bare’ vowel tier vowel phoneme moraic tier vowel melody moraic tier vowel melody template tier

⎜

⎜

⎜

⎜

⎜

\

→ µɔ

(µ)µ

→ µ

(V)V

→ V

/

⎜ ⎜

⎜ ⎜ ⎜ ⎜

tonal tier

consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

⎜ ⎜ ⎜

ɛ

(µ)µµ

word form template tier stress accent tier

word form CV template tier

→ ɔ

⎜ ⎜ ⎜

C ⎜ ⎜ ⎜

⎜ ⎜

´

⎜

H

⎜ ⎜

⎜

(Vɔ)V

→ Vɔ

C

C

⎜ ⎜ ⎜

⎜ ⎜

u

⎜ ⎜

kw

s

Figure 25. Broken transposition of detransitive plurative stems I vowel melody tier detransitive vowel tier

ɔ

ɔ

⎜

‘bare’ vowel tier

⎜

ɛ

ɛ

\

⎜

⎜

vowel phoneme moraic tier

µ(µ)

→ µ

vowel melody moraic tier

µ(µ)

→ µ

vowel melody template tier

V(V)

→ V

word form template tier stress accent tier

´

tonal tier

H

word form CV template tier consonant frame tier root and inflectable base tier classifying suffix(es) tier case suffix tier

⎜

⎜

⎜

⎜

⎜ ⎜

⎜ ⎜ ⎜ ⎜ ⎜ ⎜

⎜

C ⎜ ⎜ ⎜

b̤

⎜

V(V)

C

→ V

⎜ ⎜

r

Figure 26. Broken transposition of detransitive plurative stems II

C

⎜ ⎜ ⎜ ⎜ ⎜

s

Ad Figure 24: The plurative broken transposition *liɛ́ːkwr → loc. *liɛkwɛ́n → abl.-gen. *liɛkwnɛ́s is confirmed by Vedic abl.-gen. yaknás, etc.67 The plurative broken transposition of underlying */ɔ́ɛ/ → ['ɔː] was found, for instance, in the PIE paradigm of *uɔ́ːkw in. ‘speech, formula, words’ which was also “holokinetic”, 68 cf. Avestan nom. (and Young Avestan gen.) m. vāxš, < Proto-Indo-Iranian nom. sg. m. f. *u̯ā́kʃ, gen. sg. *u̯ā́kʃ, gen.-abl. sg. *u̯acás, as illustrated by Figure 25. There was a last type of detransitive ‘broken’ vowel transposition which ran completely in parallel to the nondetransitive one (see Figure 24). Here the detransitive vowel /ɔ/ was mapped onto both vowels. The long vowel [ɔː] was thus ‘broken up’ into two detransitive vowels /ɔ/ + /ɔ/ and the resultant oblique forms were thus somehow double-marked for detransitive voice. This type of transposition is illustrated by Figure 26. Epilogue to this section: I admit that these PIE vowel transposition rules may look very strange and unfamiliar from a sole comparative IE perspective. Some scholars not used to morphotactic types other than the fusionalinflectional synthetical type may perhaps object that these rules are not well-founded on comparison and will perhaps try to discredit the given analysis, because it is so different from the traditional model. However, such an objection is completely unjustified. The above reconstructed morphological derivational and inflectional rules are perfectly grounded on the external IE comparative evidence and thus on the most plausible potentially real and realistic PIE word forms. These rules must cogently be reconstructed as such for PIE within a linguistically proper morphological analysis of the reconstructable PIE word forms. The main difference between the traditional model, its analysis, and its reasoning and the templatic analysis and its reasoning lies in a different kind of scientific caution. Traditional comparative philologists almost anxiously avoid any diachronic scenario that fundamentally deviates from traditional Indo-European morphosyntax. Doing so, however, they are necessarily and hopelessly stuck to the IndoEuropean-ish set of morphosyntactic categories. Their linguistic reconstruction is necessarily condemned to be just anachronistic backprojection of Indo-European-ish categories. The templatic analysis, on the other hand, tries to be more cautious in this respect by avoiding any such anachronistic backprojections and misinterpretations, and it also does not take over the traditional comparative circular reasoning in the fields of morphology and syntax with regard to constructions, functions, and categories. Any sceptisism about the templatic analysis that operates with the argument that PIE “must” look “exactly” like an IE language with regard to its set of morphosyntactic categories is just based on anachronistic circular reasoning. Hoffmann’s premature model for the linguistic reconstruction of morphology and syntax and morphosyntactic categories 69 cannot be taken serious by any historical and prehistorical linguist any longer. It may perhaps be hard to accept from a more traditional philological viewpoint, but PIE was definitively not one of the Indo-European lan67 68 69

Cf. Pooth 2014d. Cf. Ibid. Cf. Hoffmann 1970.

Roland A. Pooth

16

(unmarked) locative form of nonrelational nouns. This identity can be illustrated by the “three” stems, which are actually ONE stem (*pkɛ́u-), given in the last three lines of Fig*pkúures 27 and 28. Many noun-like adjectives *pkuɛ́_ thus belonged to the hysterodynamic inflecFigure 27. PIE basic stem, primary oblique and relational stem, and secondary oblique stem tional type due to fact that it also had a relational-attributive function and thus included relational adjectives, e.g. *b̤ɛb̤rú- ‘beaver-ish form gloss translation brown’. I suggest that the PIE synchronic *pɛ́ku sheep.ABS dito (*pkɛ́u) sheep.LOC (α) ‘x as place/goal-related to domestic animal; polysemy of relational stems ran as follows: x = referent, domestic animal = place/goal’ (α) It was either the referent, e.g. ‘domestic *pkɛ́u ~ *pkú sheep.RL.ABS (β) ‘domestic animal as related to/possessed by x; animal’, that was referred to as being place or domestic animal = referent, x = possessor’ goal of (and thus spacially related to) another *pkɛ́u ~ *pkú sheep.RL.ADJ.ABS (γ) ‘x related to [DOMESTIC ANIMAL]; x = referent, [DOMESTIC ANIMAL] = attribute contextual referent, or it was (β) the referent ‘domestic animal’ that was referred to as a Figure 28. Polysemy of the PIE relational stems possessee referent, possessed by another implicit or explicit possessor x, or (γ) it was the supra-lexical concept [DOMESTIC guages and it was not at all a morphosyntactic look-alike ANIMAL] 72 as attribute, function, quality, property that of Ancient Greek, Vedic or Hittite. From a taxonomic and was more or less metaphorically possessed by ‘someone’ or typological perspective, PIE was definitively not Indothe given referent x. In functional terms, this underspecifiEuropean. Any such claim70 must immediately be rejected, cation-polysemy can be interpreted as a kind of ‘nominal because it is based on a false preconception and misleading lability’ as for the semantic roles place (location) or goal reasoning. and possessor or possessee in addition to underspecification with regard to reference and attribution. PIE relational nominal stems were not categorically specified for 4.3 Functional motivations any of these semantic roles or pragmatic functions. The PIE made use of the two or three (or four) nominal stems given contextual referent ‘x’ or ‘someone’ was either reto formally distinguish between three superordinate types ferred to as metaphorical possessor of the concept of case forms. The PIE case system will be treated in more [DOMESTIC ANIMAL], as in (γ), or as the possessed referdetail below. The PIE five core case forms, including exent ‘domestic animal’ itself, as in (β), or else as ‘domestic tended core case forms, i.e. absolutive, ergative-genitive, animal’ and place (location) or goal of a given external allative-dative, sociative-associative, and the ‘sound’ locareferent ‘x’ or ‘someone’, as in (α). In short, the relational tive with suffix -i were derived from the (most) basic stem stem *pkɛ́u- solely indicated a possessive and spacial (loca(glossed BASE). The ‘broken’ locative form, i.e. the locative tional) relation of the concept [DOMESTIC ANIMAL] to form without the suffix -i had no further case suffix and another contextual participant, and it was both referential was identical to the ‘broken’ locative stem or relational (nouny) and attributive (modifying, adjectival). Tremblay73 stem. Finally, the four oblique case forms, that is, ablativehas cogently demonstrated that the internal derivation of genitive, partitive-genitive, benefactive-purposive, comita“hysterodynamic” nominal stems from underlying nouns tive-instrumental were derived from the oblique stem, and was still productive synchronically within Old and Young in the case of the proterokinetic inflectional type from the Avestan. Illustrative examples are (39) and (40). I have ‘broken’ locative or relational stem, e.g. ‘sound’ loc. *pɛ́kui, added an archaism from Early Vedic: see (41). relational stem *pkɛ́u-, abl.-gen. *pkɛ́us, etc.; see Figures 27 stem *pɛ́ku*pkɛ́u-

gloss sheep.BASE sheep.OBL-/ sheep.RLsheep.RL sheep.rl.obl-

translation ‘domestic animal, sheep’ primary oblique stem/ relational stem ‘s.o.’s domestic animal, sheep; domestic animal-ish’ secondary oblique stem

and 28. As just mentioned, RELATIONAL71 (RL) nominal stems, viz. stems indicating an additional possessee role and a contextual (external) possessor were formally identical to the ‘broken’ locative stem. A relational nominal stem was thus marked for its relationality or for taking the role of possessee by the very same vowel melody that also coded the

70 Somone told me that Prof. Jürgen Untermann once made a claim similar to “das Urindogermanische war ja eine indogermanische Sprache”. I am not sure if he really did. 71 I make use of the term RELATIONAL to indicate the relation of possessee to possessor and modifying concept to referent. Note that “relational” is also used with a quite different meaning in the typological literature. In Mayan, for instance, “relational nouns” are a class of lexical items that tend to denote spatial relations, concepts such as ‘above’, ‘below’, etc. which are expressed by prepositions in languages like English. These “relational nouns” are obligatorily marked for possession, the grammatical possessor being the nominal that corresponds to the complement of a preposition in English.

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

(40) noun/compound frādat̰.fšauue dat. sg. ‘for the Frādat̰.fšu’ (Y. 1.4): (cf. acc. sg. pasūm, Gathic gen. sg. pasə̄uš) :: frādat̰.fšaom acc. sg. ‘the frādat̰.fšu-ish’ (Y. 2.4): adjective

(41) noun dā́ru n. ‘wood’ :: adjective dārúm acc. sg. (RV 7.6.1d) → more regular dāruṇám ~ dā́ruṇam

Tremblay’s functional distinction “relativisch” vs. nonrelational finds a noteworthy typological comparandum. I suggest that the proposed underspecification is comparable to the polysemy of the so-called “possessive state” of Classical Nahuatl nouns. Classical Nahuatl seems to encode deviations from prototypical (and non-relational) reference by means of the lack of the so-called “absolutive suffixes”. Supra-lexical nominal (and verbal) concepts are given in capitals and square brackets. Cf. Tremblay 1999, 2003: 248f.

72 73

Proto-Indo-European Nominal Morphology. Part 1. The Noun

Examples from Classical Nahuatl (Uto-Aztecan, Central Mexico) are:

(42) tōtol-in ‘turkey hen’ :: -tōtol- (without -in) as the modifying member in the compound tōtol-tetl ‘turkey egg, turkey-hen-ish egg’ and in possesssive state forms tē-tōtol ‘someone’s turkey hen’74 (43) mich-in ‘(it is a) fish’ :: tē-mich ‘someone’s fish’ (with stem -mich-)75

Classical Nahuatl thus uses a ‘nonrelational’ form tōtol-in ‘(it is a) turkey hen’, usually termed “absolute state”, to indicate a lack of relation 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, Classical Nahuatl uses a ‘relational’ stem -tōtol-, usually termed “possessive state”, 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. It is a relevant typological observation that the very same “possessive state” stem (lacking the “absolutive suffix” -in-, etc.) is further used to code 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 more literally ‘egg pertaining to or related to turkey-hen’ or ‘egg possessing turkey-hen-ish-ness’. This double function of the Classical Nahuatl “possessive state” is thus comparable to the underspecification-polysemy I hereby reconstruct for the PIE relational nominal stems. It is thus evident that Tremblay’s distinction “relativisch” vs. nonrelational can be encoded by particular markers.76 Widmer has added the following criticism: “Tremblays Unterscheidung [...] ist nur schwer nachzuvollziehen”.77 Widmer himself78 proposes an overall function of coding “possession” to motivate internal derivation, e.g. *sɛ́ʔmn in. ‘seed, sowing’ → *sɛ́ʔmɔn- ‘possessing seed, having seed; pertaining to seed; seedy; god of seed’. But it is noteworthy that Widmer79 also subsumes the attributive function under the possessive notion, e.g. PIE *nkú- → ™n̥ku̯ó- (> Toch. A oṅk, B eṅkwe ‘man’) “Tod habend; sterblich”,80 derived from PIE *n_k- ‘to perish, die’ (cf. LIV, s.v. *nek̑-). Widmer’s concept of “possession” is based on the following definition given by Seiler (note that Seiler uses “substance”, where I prefer using “concept”): Linguistic POSSESSION consists of the representation of a relationship between a [concept] and another [concept] [...]. (Seiler 1983: 4, italics mine)

It turns out that Widmer’s definition of “possessive” is functionally identical to Tremblay’s definition of “relaCf. Andrews 2003: 109ff., 280ff. Ibid. 76 Cf. Tremblay 1999, 2003. Recall that Pinault 2003: 159 has objected to this brilliant discovery as follows: “Il est ruineux de décrire ces oppositions en termes métaphysiques [...] comme celle entre substance et accident ou attribut”. Such an objection, however, is misleading from a typological perspective, simply because such formal distinctions are attested among the languages of the world. It is deplorable that Xavier died too young before receiving an appropriate understanding. 77 Widmer 2013: 189, fn. 2. 78 Ibid. p. 30. 79 Ibid. p. 30, p. 73. 80 Ibid. p. 73. 74 75

17

tivisch”. Widmer failed to recognize the underlying conceptional identity. Tremblay and Widmer only differ with regard to the terminology that they have chosen to refer to the linguistic encoding of relationality, that is, the encoding of a relationship between a [CONCEPT]1 and another [CONCEPT]2. It is true, however, that Widmer and Tremblay differ with regard to what they think should be taken for the underlying derivational base, both on the formal as well as on the functional level. Tremblay has made the brilliant claim that PIE had an elaborate system of root and pattern morphology including an underlying derivational base on the skeletal consonant frame and multiple nominal inflectional types and verbal binyans of the Semitic morphological type. He called it “semitico more”.81 His view on PIE nominal morphology thus conforms to the one proposed here. However, others still follow Schindler’s extremely unsatisfactory idea of a “derivational chain”, “chaîne de derivation”, “Derivationskette”. Such a view on internal derivation has always been a more restricted perspective, simply because in Schindler’s conception the strategy of internal derivation was restricted to the lexical and derivational level, whereas in Xavier’s and mine it is also seen as an inflectional strategy. Followers of Schindler’s preconception are thus simply unable to reckon with the mechanism of wordinternal transfixation and ultimate functional motivations of PIE noun inflection, simply because they must cling to the traditional segmentation and the parsing of the PIE word forms into root with ablaut, suffix(es) with ablaut, and inflectional ending. Applying the IE segmentation to PIE word forms, however, has turned out to be based on circular reasoning82 and can be regarded as an inner-PIE analytical mistake. It can be concluded that Tremblay’s brilliant idea “relativisch” vs. nonrelational is transferrable to the very inside of each PIE nominal paradigm. It is almost self-evident that the PIE oblique case forms (e.g. *pkɛ́us) had a relational attributive (Tremblay’s “relativische”) function and thus encoded a relationship between a referential [CONCEPT]1 and another, modifying and attributive [CONCEPT]2. The oblique case forms were thus used to attributively relate the given nominal [CONCEPT]2 to a given referential [CONCEPT]1. Syntactically, therefore, they had the status of a modifying and attributive adjunct and were realized as right-attachment. (I have dropped the glossing of the verb form in the following examples): (44) *kuɔ́n-s uɔ́ːkw klnɛ́ut χnrɛ́s dog-ERG speech.ABS man:ABL:ADJUNCT literally ‘as for dog, to speech it was listening, the one from man’ (45) *kuɔ́n-s ɠwɛ́mt kluɛ́i dog-ERG listen:BEN:ADJUNCT lit. ‘dog came to listen’

(46) *kuɔ́n-s pɛ́ːku páχt χnrɛ́i dog-ERG cattle.ABS protected man:BEN:ADJUNCT lit. ‘dog protected cattle for the benefit of man’ 81 82

Cf. Tremblay 2003. Cf. Pooth 2014d.

Roland A. Pooth

18

(47) *kuɔ́n-s ʔɛ́it χnrɛ́ʔ dog-ERG man:COM:ADJUNCT lit. ‘dog moved, accompanied by man’

Therefore, it is inferential that the PIE relational (possessed) nominal stems were word-internally inflected for their relational function on the word form template tier much the same way as attributive oblique forms were word-internally inflected for their syntactic adjunct(ive) status. The following inference is thus dedicated to Xavier. I would like to make it memorizable by the label TREMBLAY’S DERIVATIONAL RULE, because he was the first to correctly identify the functional distinction of *pɔ́tis, abl.gen. *pɛ́tiɔs from its relational derivative *pɔtɛ́is ~ *pɔtís, abl.-gen. *ptɔ́is, although he did not correctly identify the original inflectional type of the latter (and did not analyze the o-grade the way I do now). Tremblay’s derivational rule thus runs as follows: PIE relational (“relativische”) nominal stems were inflected according to the “hysterodynamic” or “hysterokinetic” inflectional type. (Let me add to this that there were additional “mesostatic” inflectional types, e.g. the one just mentioned) These inflectional types were derived from underlying nonrelational classified nouns by internal derivation, that is, by vowel transposition. Examples are: (48) *ɢ̤ɛ́b̤χl- ‘round holder’ → *ɢ̤b̤χál- an. ‘someone’s head’; cf. Greek κεφάλη f. ‘head’, Toch. A śpāl ‘head’, OHG gibilla f. ‘gable’; derived from the root *ɢ̤_b̤χ- ‘to seize, hold, keep, grab’

(49) **ɗɛ́ng̈u- ‘whose function is to **ɗ_ng̈-’ → *ɗng̈ɛ́u- ~ *ɗng̈ú- ‘someone’s tongue’, derived form the base *ɗ_ng̈(perhaps related to *ɗ_n-s- ‘to perceive, make perceive, learn, teach’, perhaps ultimately derived from **ɗ_n- ‘to perceive, taste’)

(50) *ɗɛ́i-u- an. ‘whose functional nature is to be bright, shiny’ → *ɗiɛ́u- an. ‘someone’s bright sky, our daylight-god’, adj. ‘bright, shiny’

(51) *smɔ́kur- ‘beard-hair’ → *smɔkrɛ́u- ~ *smɔkrúone’s beard-hair, someone‘s beard’

‘some-

Excursus: It is obvious that the regular methathesis *-ur→ *-rɛ́u- ~ rú- went hand in hand with this derivation. It is inferential that it was ultimately due to the onset maximization principle (OM) combined with PIE syllabification and metrical templatic rules. The derivational base */smɔ́kur/ had a trochaic metrical structure (σ́ σ). It was initially syllabified as *['smɔ]σ[kur]σ, but PIE trochaic structures tended to be syllabified with a closed (heavy) initial syllable, whereas iambic structures (σ σ́) tended to be syllabified with an open (light) initial syllable. The second syllable was thus also syllabified as *['smɔk]σ[wr]σ with /u/ in the onset of the second syllable and /r/ in its nucleus. The derivative, however, had an iambic structure, and according to the OM of stressed syllables, the stressed syllable *['kur]σ underwent metathesis yielding a structure /'kr_u/ with an open initial syllable *[smɔ]σ and a second syllable with maximized onset *[smɔ]σ['krɛu]σ ~ *[smɔ]σ['kru]σ. In addition, this was heavily influenced by analogy to other relational nominal stems with a second syllable ['rɛu] or ['ru] such as, e.g., */b̤ɛb̤rɛ́u-/, etc. End of excursus. More relational nominal stems are:

(52) **b̤áːg̈u- → *b̤a(ː)g̈ɛ́u- ~ *b̤a(ː)g̈ú- ‘someone’s forearm’, cf. Vedic bāhú- ‘forearm’; recall that the root *b̤aɠ- had a meaning ‘to get, take, give, distribute a portion (in the mouth, to eat)’; the ‘forearm’ was derived from a root *b̤ag̈- which perhaps had a similar meaning; note that aroots sporadically exhibit such variants with different laryngeal features (53) *tɛnúχ- ‘someone’s body’, lit. ‘s.o.’s stretched collective/group (of body-parts)’, derived from the root *t_n‘to stretch, be stretched’, cf. Vedic tanū́-

(54) *pχtɛ́r- an. ‘someone’s father’, *sɔuHnú- an. ‘someone’s son’; several other kinship terms belong here, except for *máː an. ‘mother’ and *suɛ́sɔr ‘self-woman, sister’

(55) *ɗáʕtr- ‘giving’, erg.-gen. sg. *ɗáʕtrs (besides *ɗáʕtɔr‘always giving’; these later merged yielding Vedic dā́tar-, gen.-abl. sg. dā́tur) → *ɗʕtɛ́r- ‘s.o.’s giver, its giver’ that is, either ‘someone’s (possessor) giver’ or ‘something’s giver’, that is, ‘giver of a something (theme)’, continued as the Vedic dātár- type

(56) *ʔaltɛ́r- ~ *ʔantɛ́r -‘the other one (in relation to someone contextually presupposed)’, derived from a pronominal derivational base *ʔál- ~ *ʔán- (~ *ʔ_n-) ‘the other one, the one beyond’; cf. *ʔalí- ~ *ʔaní- ‘other-ish’; note that Hittite annal(l)a/i- adj. ‘former’ can simply contain a reduplicated stem (*ʔan-ʔal- + -í-, -lí- or -lɔ́_ or the like); annaz adv. ‘formerly’, annišan adv. ‘formerly once’ can thus also be related;83 if they all do, this speaks in favour of *ʔa-n- (probably the root was simply *ʔa-)

Furthermore, root nouns and class 1 nouns were turned into relational stems by suffixation of the class 7 suffix -i-, additional internal modification, and accent shift: (57) *ʕákw- ‘seer’, in. ‘eye’ → *ʕkw-í- ‘someone’s eye’

(58) *pɔ́t- an. ‘master, powerful maker, self, self-maker, selfruler’ → *pɔt-í- an. ‘someone’s master’ (see above) (59) *kɛ́r-ɗ- in. ‘jumper, swinger: 84 heart’ (class 1b) → *kr-ɗ-í‘someone’s heart’ ~ *krɗɛ́i-, etc., cf. Lithuanian širdìs :: Vedic hŕ̥daya- (with h for *ś-, cf. śrád-°)

5 Overview of PIE inflectional types Given the above reconstructed PIE internal inflectional and derivational rules, it is quite easy to generate all potentially possible PIE inflectional types. We are now in a perfect position to fundamentally understand the underlying morphosyntactic and functional derivational motivations and the overall “raison d’ être” of the PIE inflectional types. The various inflectional types were reconstructed most predominantly by Eichner, Schindler, Rix, grounded on work done by Kuiper and Pedersen.85 These scholars deserve the status of being the first scholars to envision a ‘templatic-like’ morphology. But we all know that they never used the terms template and transfixation and clinged to the traditional segmentation into root, suffixes, endings. The templatic model can thus be seen as a more advanced and revised version of the model of the “Wiener 83 It is quite unproblematic to derive the meaning ‘the former’ from ‘the other one’, for other or former views cf. Kloekhorst 2008: 173f. 84 Ultimately derived from the root *k_r- ‘to swing, jump, dance’, cf. LIV, p. 556 with fn. 3. 85 Cf. Eichner 1973, Schindler 1972, 1975, Rix 1965.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

RN no. 1 2a 2b 3a(~b?) 4a(~b?) 4c

abs. sg. *ɗɛ́m *pɔ́ɗ *ɗɔ́m *ʕrɛ́ːɠ *b̤ɔ́ːr *uɔ́ːkw

erg.-gen./gen. *ɗɛ́ms *pɔ́ɗs *ɗɔ́ms *ʕrɛ́ːɠs *b̤ɔ́ːrs *uɔ́ːkws

abl.-gen. *ɗmɛ́s *pɔɗɛ́s *ɗmɔ́s *ʕrɛ́ɠɛs (~ *ʕrɛɠɛ́s?) *b̤ɔ́rɔs (~ *b̤ɔrɔ́s?) *uɛkwɔ́s

19

the English gerund, e.g. hittting by being underspecifiedpolysemous as for individuated reference. But note that the individuated referent was not specified by the numeral ‘one’ in PIE. The PIE root nominals had flexible gender depending on the given contextual referent. This flexible gender is evident from Vedic yújmasculine ‘allied’, but yúj- feminine ‘alliance, connecting, connection’ (verbal abstract noun), which goes back to a PIE root nominal *iɛ́uʛ- ~ *iúʛ- with flexible gender depending on the contextual referent. Root nouns were thus used either with individual reference to a particular participant role involved in the event indicated by the verbal root or with reference to the verbal event itself without any individuated participant involved in the event. An illustrative example is *ʔɛ́r-; see (60):

gloss ‘household, family’ ‘habitual down-goer: foot’ ‘house’ ‘enduring ruler’ ‘habitual away-carrier: thief’ ‘group of words, speech, voice’

Figure 29. Inflectional types of root nouns

Schule” and Kuiper’s and Pedersen’s ideas. Formally, the model of the “Wiener Schule” and mine are more or less identical, but they fundamentally differ in the overall underlying analysis and in the overall functional assignments and derivational conceptions. The templatic model also reckons with accent types, but accent shift is not viewed as the most prominent morphological strategy anymore. The model is not inspired by Lithuanian accent types any longer. In addition, the templatic model is capable of assigning particular functions to the various types. It is also capable of assigning a more specific function to the types who turn out as types with o-grade. Recall that the ograde is not understandable from the perspective of the model of the “Wiener Schule”. This is a clear advantage in favour of the templatic model.

5.1 Root nouns The reconstructable inflectional types of root nouns or root nominal stems are given in Figure 29.86 These types were generated by the above reconstructed morphological rules of vowel transposition. The reconstruction of a single stem with root ablaut *uɔ́ik- :: *uɛ́iks was an anachronistic mistake. The distinction of the active-agentive stems, e.g. *uɛ́ik- ‘clan (human beings)’, from corresponding detransitive stems e.g. *uɔ́ik- ‘settlement of the clan, village’ was not recognized in the traditional model. The erg.-gen. sg. an. *ɗɛ́ms once belonged to the animate word for ‘family, household’, but not to its inanimate derivative *ɗɔ́m- ‘house, building’;87 see the section on the formal identity of the erg. and the gen. below. Its root also had the meaning ‘to make s.o./s.th. belong to the household/family; be, become, make tame’ besides ‘to build, to make a house’. There were two detransitive plurative types, cf. Gk. φώρm. ‘thief’ = Lat. Plaut. Peon. 1384 ms. B for, Lat. fūr m. ‘id.’.88 Young Avestan gen. sg. m. vaco and nom. sg. and gen. sg. m. vāxš < Proto-Indo-Iranian abl.-gen. sg. *uacás :: nom. sg. and gen. sg. *uā́kʃ = */uā́ks/. The plurative type 4ab thus indicated habitual agents, whereas the plurative 4c type encoded non-agentive nominal plurality, a group of products or the like. Perhaps the same was true for the distinction of the types 2a and 2b. Many PIE root nouns were verbal adjectives (participles) and verbal nouns. More precisely they were root nominals, not solely root nouns. They were typologically similar to 86 I use the abbreviations RN = root nouns and adjectives, CN = classified nominal stems, HD = “hysterodynamic” relational nominal stems. Note that these abbreviations differ from the provisional classification given in Pooth 2015, where I used A = RN, B = CN, C = HD. 87 Pace Schindler 1972 & co. 88 Cf. Weiss 2009: 143 with fn. 46.

(60) *ʔɛ́r- (~ *ʔár-) animate ‘hitting one: ram’ (reference to an individual which is capable of being agent, triggering animate gender), cf. Umbrian acc. sg. erietu ‘ram’, Latin ariēs m. ‘ram’, etc.89 = *ʔɛ́r- inanimate ‘hitting, contacting, meeting’ (abstract nouns have inanimate gender); cf. Gk. ἔρις f. ‘quarrel’

The underspecification of the stem was not disambiguated by constructions like English the hitting vs. the one hitting, the hitting one. But it was syntactically disambiguated by the respective gender agreement of a modifying adjective. (61) a. *ʔɛ́r ram.ABS.SG ‘big ram’

mɛ́ɠ-χ_ɔ_nt big.CL3_DTR_CL13.ABS\ANIMATE.HEAD

b. *ʔɛ́r mɛ́ɠχ hitting.ABS big.CL3.ABS\INANIMATE.HEAD ‘big hitting, meeting, contacting’

PIE root nominal stems, like verb forms, were inflected for voice, that is, they were marked for distinctions of the domain of agency-agentivity and semantic transitivity. The unmarked agentive-active root nouns regularly denoted an agent or natural force role or an agentive action or powerful natural action or process, whereas detransitive root nouns indicated a stative-habitual agent or a non-agentive semantic role (product, theme, patient). Furthermore, detransitive nouns indicated a stative-habitual action or process or a nonagentive process or state. Thus, habitual agency-agentivity by an animate being was also coded by detransitive voice principally conforming to the inactual and stative or ‘stativized’ semantic nature of habitual actions. Like verb forms, root nominals were also inflected for verbal and nominal aspect by the same “chroneme”. Two superordinate aspect categories, namely singulativesinglefactive vs. plurative-plurifactive were distinguished on the vowel melody template tier, as outlined above. Example (62) can illustrate the marking of voice and nominal aspect of root nouns:

(62) a. *uɛ́kw-

89

(α) an. ‘speaking one’ → ‘speaker’ (β) ‘speaking natural force’

Cf. NIL s.v.; EIC s.v.

Roland A. Pooth

20

b. *uɔ́kw-

(γ) in. ‘action of speaking’ (abstract noun)

(α) an. ‘habitually speaking one’, adj. ‘having the habit of speaking’ (β) in. ‘habit of speaking’ (abstract noun) (γ) in. ‘product/theme/patient of speaking’ → ‘spoken one: word’ (δ) in. ‘process/result/state of speaking’ (nonagentive verbal abstract noun)

c. *uɛ́ːkw- (α) an. ‘big/more/plurally speaking one’ (β) in. ‘plurality of speakers’ (γ) in. ‘plurifactive action of speaking’

d. *uɔ́ːkw- (α) an. ‘big/more habitually speaking one’ (β) in. ‘having the abstract(ive) habit/quality of speaking’ → ‘voice’ (γ) in. ‘plurality of products/themes of speaking’ → ‘group of words, speech, formula’ (δ) in. ‘big/more/plurative-plurifactive process/result/state of speaking’

The plurative-plurifactive aspect was underspecifiedpolysemous for individuation. It indicated an individuated referent involved in a plurifactive event, or referred to the plurifactive event itself, or indicated a plurality of referents involved in the event. PIE root nouns, unlike verb forms, were further inflected for case, number and oblique vs. nonoblique. PIE also had reduplicated nominal stems (RD) with the same underspecification-polysemy, e.g. *b̤ɛ́b̤ɛr-, *mɛ́mɔlχ‘plurative product of grinding: flour, meal, dust’. They also belonged to the suffixless nominal stems, because they lacked a classifying suffix; see Figure 30: RD no. 1 2

abs. sg. *b̤ɛ́b̤ɛr *mɛ́mɔlχ

‘broken’ loc./ ‘sound’ loc. *b̤ɛb̤ɛ́r *mɛmɔ́lχ

3a 3b

*d̤ɛ́d̤ʔi *d̤ɛ́d̤ʔi

*d̤ɛd̤ʔɛ́n *d̤ɛ́d̤ʔn-i

abl.-gen. gloss *b̤ɛb̤rɛ́s an. ‘brown one’ *mɛmlχɔ́s an. ‘habitual grinder’, in. ‘grinded mass’ *d̤ɛd̤ʔnɛ́s in. ‘whey, sour milk’ *d̤ɛd̤ʔnɛ́s an. ‘the one putting’

Figure 30. Inflectional types of reduplicated nominal stems

Additional notes to Figure 30: There also was a reduplicated verb-like adj. stem *b̤ɛb̤r_ɔ́_ and a classified stem *b̤ɛ́b̤r-u- ‘brown animal; beaver’ (class 8), cf. Vedic babhrú‘brown’, etc. For PIE *mɛ́mɔlχ- cf. Hittite nom.-acc. sg. n. me-e-ma-al (OS), me-ma-al (OS), me-em-ma-al (NH), gen. sg. me-ma-al-la-aš (MH?/NS), me-ma-la-aš (OH/NS), me-em-mala-aš.90 PIE further had a few irregular reduplicated heteroclitic classified nouns or root nouns in -i- ~ -n- (class 7 ~ class 1). There was a subtype which was suffixed by the classifying suffix -i- (class 7) as, e.g., in the case of *d̤ɛ́-d̤ʔ-i- an. ‘the one putting s.o./s.th. there (perhaps distributively here and there)’, cf. Vedic nom. sg. m. dádhis ‘granting, putting’ (RV 10.46.1c). Another subtype was a root noun and contained an enlarged inflectable base in -i- as, e.g. in the case of *d̤ɛ́-d̤ʔ-i- in. ‘sour milk, whey’. It is inferential that in both cases the -i- was substituted by -n- in the loc. and oblique stem, cf. Vedic dádhi n. ‘sour milk, whey’, abl.-

gen. dadhnás. The obviously quite archaic class of PIE reduplicated nouns seems to need a more detailed comparative philological investigation.

5.2 Classified nouns The following inflectional types, which are reconstructed within the templatic approach, are formally identical or almost identical to the traditional ones.91 The only difference lies in the analysis.

CN no. 1a 1b 2a 2b 2c 2d 3a 3b 4a 4b 4c 4d 4e 5a 5b 5c 5d

Cf. Kloekhorst 2008: 575 who also reconstructs “*mé-molh2-”.

*ʔɔ́ːku *d̤ɛ́g̈ɔm *mɛ́nɔs *kuɛ́ːnɔs *sɔ́kwχɔis

sound loc. sg./ broken loc. *pɛ́kui *krɛ́ɗ *liɔkwɛ́n *ɗɔ́rui ~ *ɗɔrɛ́u *nɔ́ɠwti (~ *nɠwɔ́t?) *ɠɔ́nui (~ *ɠɛ́nɔu?) *ɠɛ́rχas /ɠɛ́rχɛs/ *liɛkwɛ́n *χauɔ́i? /χɛuɔ́i/ ~ *χɔuɔ́i ~ *χɔːuɛ́i? ~ *χɔːuɔ́i? *ʔɔ́ːkui ~ *ʔɔːkɛ́u *d̤g̈ɔ́m *mɛ́nɔs *kuɛnɔ́s *sɔkwχɔ́i

abl.-gen. *pkɛ́us *krɗɛ́s *liɔkwɛ́ns *ɗɔruɛ́s *nɠwtɔ́s *ɠɛ́nuɔs *ɠɛ́rχsɛs *liɛkwnɛ́s *χauiɔ́s? /χɛuiɔ́s/ ~ *χɔuiɔ́s ~ *χɔːuiɛ́s? ~ *χɔːuiɔ́s? *ʔɔːkɛ́us *d̤g̈mɔ́s *mɛ́nsɔs *kuɛnsɔ́s *sɔkwχiɔ́s

Figure 31. Inflectional types of classified nouns [Key: orange: proterodynamic, yellow: acrostatic, grey/blue/white: holo-/amphikinetic]

Additional notes to Figure 31: Cf. Lat. gen. sg. n. iocineris, na. sg. n. iocur ~ iecur ‘liver’;92 cf. Hom. Gk. gen. sg. n. δουρός = Hitt. gen. sg. n. GIŠtar!-wa-aš (₠doru̯ós); cf. Vedic gen.-abl. sg. n. yaknás, Latin na. sg. n. iecur, abl. iecore (< *ieceri), Greek na. sg. n. ἧπαρ, etc.; cf. f. PG *ajja-93/*ajjazn. ‘egg’ (™h2au̯i ̯os-, ™h2ou̯i ̯os- or ₠h2ōu̯i ̯ó- ‘egg’; I will return to the PIE word for ‘egg’ below). The type 4 maybe had a few allomorphic subtypes 4abcd and 4e. Since the pattern *χɔuiɔ́s matches the oblique stem of the *sɔkwχiɔ́ type it was perhaps the most regular oblique stem of type 4. The abl.-gen. *ʔɔːkɛ́us was later remodelled to *ʔɔːkɛ́u̯ɔs (™h1ōk̑éu̯os), which is directly continued by the ablaut pattern of its Greek continuation. The ‘broken’ loc. of types 5abc was remodelled by levelling to the more frequent e-grade in analogy to the ‘broken’ locative forms of the types 1b, 2ab, 3ab, 4bc, e.g. *mɛ́nɔs → *mɛ́nɛsi (matching the equation formula ₠ ménesi). PIE *kuɛ́nɔs- in. ‘divine force’ is continued by cf. Young Avestan nom.-acc. sg. n. spānō ‘divine force’, inst. spanaŋha (Y 9.27).94 For type 5d cf. Vedic sákhi m. f. ‘friend, fellow’, acc. sg. sákhāyam < PIE all.-dat. sg. *sɔ́kwχɔi-m ‘to the friend’. Recall that I have interpreted this type 5d as a noun-like adjective type elsewhere (a type in between noun and noun-like adjective). Its suffix -i- was the class 7 suffix. It seems to be related to the oblique stem of type 4b and thus perhaps presupposes an underlying abstract noun *sɔ́ːkwχ-iCf. Widmer 2004, Viti 2015. Cf. Pooth 2014d, Rix 1965. Cf. Kroonen 2013. 94 Cf. Widmer 2004: 50. 91 92 93

90

erg. sg. an./ abs. sg. in. *pɛ́kus *kɛ́rɗ *liɔ́kwr *ɗɔ́ru *nɔ́ɠwt *ɠɔ́nu *ɠɛ́ːrχs *liɛ́ːkwr *χɔ́ːui

Proto-Indo-European Nominal Morphology. Part 1. The Noun

in. ‘fellowship, friendship’, an abstract noun derived from *sɔ́(ː)kw-χ- in. ‘group of followers’.95

5.3 Relational nouns Since relational nominal stems were derived from corresponding non-relational stems by vowel transposition, various different types of such stems were possible. These inflectional types are traditionally termed “hysterodynamic”. To give some credit to the “Leiden School” I will subsume the relational stems under the label “hysterodynamic” inflection (abbreviated HD). Notes: The type HD 1 is quite confirmed. For the type HD 2 cf. *pɔχimɛ́n- ‘the *pɔ́χimn-ish = protection-ish one’; its abl.-gen. *pɔχimnɛ́s was remodelled to Proto-Greek gen. sg. m. *poiménos. Some types are just added to this list by generalizing the established derivational rule and presupposing that it also was true for the other stems. base CN no. 1a

derived HD no. 1

1b 2a

1 2

2b 2c

2 3

2d 3a

4 5

3b 4a 4b 4c 4d 4e 5a 5b 5c 5d

5 4 6 7a 7b 7a 3 4 4 6

relational stem *pkɛ́u- (~ *pkú-) (*ɗiɛ́u-) *krɛ́ɗ*liɔkwɛ́n(*pɔχimɛ́n-96) *ɗɔrɛ́u (~ ɗɔrú-) *nɠwɔ́t(*uɗɔ́r-) *ɠɛnɔ́u*ɠɛrχás(*ʔsu-mɛnɛ́s-) *liɛkwɛ́n*χauɔ́i~ *χɔuɔ́i~ χɔːuɛ́i-? ~ χɔːuɔ́i-? *ʔɔːkɛ́u*uɗɔ́n*mɛnɔ́s*kuɛnɔ́s*sɔkwχɔ́i-

secondary oblique stem, e.g. abl.-gen. *pkuɛ́s (*ɗiuɛ́s) *krɗɛ́s *liɔkwnɛ́s *ɗɔruɛ́s *nɠwtɔ́s (*uɗnɔ́s ~ *uɗrɔ́s) *ɠɛnuɔ́s *ɠɛrχsɛ́s (*ʔsu-mɛnsɛ́s) *liɛkwnɛ́s *χauiɔ́s ~ *χɔuiɔ́s ~ χɔːuiɛ́s? ~ χɔːuiɔ́s? *ʔɔːkuɛ́s *uɗnɔ́s*mɛnsɔ́s *kuɛnsɔ́s *sɔkwχiɔ́s

Figure 32. HD inflectional types of relational nominal stems

base CN no. 2ab

derivative MS no. 1

6abcde

2

relational stem *sɔuχnɛ́u(~*sɔuχnú-) *CɔːCɛ́u- ?

secondary oblique stem, e.g. abl.-gen. *suχnɔ́us *CɛCɔ́us ?

Figure 33. ‘Mesostatic’ inflectional types of relational nominal stems

According to the present analysis, PIE also had minimally one “mesostatic” inflectional type of relational nominal stems (abbreviated MS). Here belonged the PIE word for ‘(someone’s) son’. Such stems perhaps had a more ‘non-derivational’ status; see Figure 33.

5.4 Generic detransitive plurative nouns The *uɛ́ɗɔr type, usually termed “amphikinetic”97 or “holokinetic”, 98 the *mɛ́nɔs type, and the *kuɛ́ːnɔs type (see 95 96 97

Cf. Pooth 2015. Cf. Gk. ποιμήν :: piemuõ ―the PIE inflectable base was *p_χ-i-. Cf. Steer 2015 who uses the term “amphikinesis”.

21

Figure 31, no. CN 5abc) are interpreted as detransitive plurative inflectional types. Recall that I use this term plurative as a superordinate for plural, collective, generic, augmentative. These inflectional types were less basic inflectional types. As a rule, they were secondarily derived from any underling nominal stem by templatic mapping of their vowel melodies upon the respective underlying consonant frame. Examples are: (63) *g̈ɛ́i-m- ‘winter, frost (classified by class 6 as an extended season)’, loc. *g̈iɛ́m (= relational stem/adj. *g̈iɛ́m-99) → *g̈ɛ́imɔn- in. ‘winter(s)’, adj. ‘winter-ish’

(64) *uɔ́ɗr- in ‘water’ → *uɛ́ɗɔr- in. ‘water(s)’, adj. ‘water-ish, of the water-type’ (65) *sɛ́Ɂmn- in. ‘seed’ → *sɛ́Ɂmɔn- adj. ‘seed, seed-ish’

(66) *pχtɛ́r- an. ‘s.o.’s father’ → *páχtɔr- adj. ‘father-ish’

Thus, almost any classified nominal stem, even relational nouns like *pχtɛ́r- an. ‘father’, could be turned into such a derived stem by superimposing the vowel melodies of the inflectional types CN 5abc (and 5d) upon the respective underlying consonant frame. These plurative stems were underspecified as for the distinction of reference from attribution (modification), in parallel with the (verbal) root nouns. Their nouny generic reading (simply glossed GNR here) can be illustrated by the examples below: (67) a.

b. c.

*uɔ́ɗr páʕ-t water.ABS swallow.NDUR.AGT.SG-DIR ‘he swallowed a portion of water’ *uɛ́ɗɔr nɛ́ páʕ-t water.GNR.ABS NEG dito ‘he did not swallow water (any water)’ *uɛ́ɗnɔ_m nɛ́ páʕ-t water.OBL_PAR NEG dito ‘he did not swallow water (partitive)’

These inflectional types also served to encode noun-like adjectives (and verbal participles) and thus served to code what may be termed associative-similative nominal stems, e.g. *sɛ́ʔmɔn- ‘associated to seed, seed-ish, see-like, seedtype-like’, as given in example (65) above. The inflectional type CN 5a was thus also used to code equativecomparative nominal (noun-like adjective) stems suffixed by -i_s- (class 7 + class 1). (68) *ʔuɛ́siɔs- ‘(a/the) worthy-type like; the worthy one as parameter of comparison’

These equative-comparative nominal stems could be used both with absolute reading ‘worthy in a special way compared to standard world knowledge’ or else in equative-comparative constructions ‘worthy compared to s.o. who is worthy’ → ‘as worthy as’ and ‘worthy compared to s.o. who is not’ → ‘worthier than’; cf. the excursus below. Finally, the inflectional types 5abc also served to code simultaneous-imperfective verbal participles, e.g. *uɛ́kɔnt‘willing, desiring’. Therefore, it is perhaps possible to speak of a common class of plurative-generic-similative and associative nominal stems. The various readings of all these nominal stems included a kind of similative or asso98 99

Cf. Schindler 1975a. Cf. Latin hiems m. ‘id.’ continuing the relational nominal stem.

Roland A. Pooth

22

ciative concept [OF THE TYPE], [LIKE] or [AS] and can be translated by help of their English euqivalents. The polysemy (or polyfunctionality) of this inflectional type can be illustrated by the Figure below. abs. sg. *uɛ́ɗɔr *ʔuɛ́siɔs *uɛ́kɔnt

meaning/translation (α) generic noun ‘water as type, not as a specific token’ (β) adj. ‘like water’, ‘as water’ (α) adj. ‘worthy in a special way’ (β) adj. (equative) ‘as worthy as’ (γ) adj. (comparative) ‘worthier than’ participle ‘as willing (*u_k-) one ...’

Figure 34. The PIE polysemy of the PIE generic DTR PLT types

Excursus: As just mentioned, equative-comparative nounlike adjectives, that is, nominalized root adjectives in -i_swith class 7 (-i-) + class 1b (-s-) marker, e.g. *ʔuɛ́siɔs-, derived from the root *ʔu_s- ‘worthy, worthiness’ were also coded by this inflectional type. As in many languages of the world, the standard-of-comparison participant was coded in the ablative case. The PIE comparative construction is illustrated by example (69a). The PIE equative construction, on the other hand, had to be coded by the associative-similative-equative enclitic particle *=nɛ ‘like, likewise’ or *=uɛ ‘or, or likewise, likewise’, as illustrated by examples (69b). The plurative detransitive nominal stem *ʔuɛ́siɔs- was not obligatory in the equative construction; other types of adjectives could be used alike, as illustrated by example (69c). (69) a.

*ʔɛ́ku ʔuɛ́siɔs χáuiɔs horse.ABS.SG worthy.CL7.DTR.CL1B:PLT.ABS sheep.ABL literally ‘horse, worthier than sheep’

b. *ʔɛ́ku ʔuɛ́siɔs ɠwáʕu=nɛ dito dito cow.ABS=particle literally ‘horse, as worthy as cow’ c.

*ʔɛ́ku ʔuɛs-is_ɔ́ χáuiɔs dito worthy-CL7.CL1_DTR\VLAD.ABS dito lit. ‘horse is worthier than sheep’

b. *ʔɛ́ku ʔuɛs-is_ɔ́ ɠwáʕu=nɛ dito dito cow.ABS=particle lit. ‘horse is as worthy as cow’ c.

6 Metrical issues Recall that PIE syllabification was driven by wellestablished syllabification constraints, which are universal guidelines. These can be formulated for PIE as follows:100 According to the HNUC constraint, the nucleus was assigned to the segment with the highest sonority. The PIE full vowels /ɛ ɛː a aː ɔ ɔː/ always formed the syllable nucleus. When there was more than one segment destined to be the nucleus of a syllable, the nucleus was assigned to the one higher in sonority, except when a word-edge sonorant preceded another sonorant and thus conflicted with the onset maximalization principle. The constraints ONSET, MAXO, ALIGNNUC, ALIGN have been formulated universally as follows: Syllable should have onsets; a segment to the left of a syllable nucleus constitutes its onset (ONSET). Syllabify as many consonants as possible within the onset, that is, maximize onsets (OM). Align the right edge of a syllable nucleus with the right edge of a syllable (ALIGNNUC), that is, minimize codas. Keep in mind that, if possible, for every morpheme boundary there should be a syllable boundary (ALIGN). Finally, recall the Sonority Sequencing Principle (SSP): “Between any member of a syllable and the syllable peak, only sounds of higher sonority rank are permitted.” 101 This principle, however, should be toned down for PIE as follows: If possible, adjust the sequence of segments to the sonority scale with increasing sonority from the right word-edge to the syllable nucleus (peak) and decreasing sonority from the nucleus to the left word-edge and try to syllabify the phonological word as a monosyllable; cf. Figure 35.

*ʔɛ́ku ʔuɛ́s-u ɠwáʕu=nɛ dito worthy-CL8.ABS.SG cow.ABS=particle literally ‘horse, worthy as cow’

As suggested above and elsewhere, PIE exhibited a distinction between non-predicative (noun-like) adjectives and predicative (verb-like) adjectives. The corresponding predicative function thus had to be coded by a verb-like adjective, which is glossed VLAD in the examples given below. (70) a.

extendable to derive verb-like adjectives again. I will return to the PIE class of verb-like adjectives in more detail in the sequel of this article. End of excursus.

*ʔɛ́ku ʔuɛs-u_ɔ́ ɠwáʕu=nɛ dito worthy-CL8_DTR\VLAD.ABS dito lit. ‘horse is worthy like cow’

Such verb-like adjectives could be nominalized again by suffixation of the class 1c suffix -n-, e.g. *ʔuɛsisɔ́n- ‘the worthier one’ or *suaχɗisɔ́n- ‘the sweeter one’. These nominalized stems are the source of the Proto-Germanic comparative stems in *-izō, *-izan-, *-izin-. However, such class 1c nominalized verb-like adjectives were not further

ω ⎜

σ

o ⎜ ⎜ ⎜

ρ ν

⎜

['pkt ɛ

κ ⎜

n]σ

Figure 35. MINSYL metrical template e.g. *pktɛ́n rel. ‘s.o.’s comber’ (abs. sg.)

On the level of the phonological word ω, PIE had a highranking mononsyllabicity constraint of morphologically underlying forms, namely the MINIMUM-OF-SYLLABLES CONSTRAINT (MINSYL). The rule was: If possible, syllabify ω as a monosyllable or syllabify ω with the lowest possible number of syllables. This high-ranking PIE syllabification principle, taken together with the other syllabification constraints, yielded various metrical syllabification templates. Figure 35 illustrates the syllabification of PIE abs. sg. *pktɛ́n (rel. nominal stem) ‘comber’ as a monosyllable. Key to Figures 36ff.: F = metrical foot, σ = syllable, o= syllable onset, ρ = syllable rhyme (or rime), ν = syllable 100 101

Cf. Kobayashi 2004: 22; Byrd 2010, 2015; Cooper 2015. Clements 1990: 284ff.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

nucleus, κ = syllable coda; R = any sonorant (m n r l i u), P = plosive or implosive stop, C = any consonant, $ or . or ]σ = syllable boundary. PIE *pktɛ́n- ‘hair-fixer, comber: comb’ is continued by Lat. pecten m. ‘comb’ and Gk. κτɛίς, κτενός m. ‘id.’. It was another relational nominal stem. Its original relational meaning was ‘someone’s comber’. It was structured in parallel with the relational agent nominal stem *pk-tɛ́r- an. ‘hairfixer, comber’. Both were derivatives of the root *p_k- ‘to fix hair, comb’, cf. Greek πέκω, Lithuanian pešù ‘to pluck, pull at the hair’. It can be reconstructed as beginning with a heavy onset cluster consisting of three stops *pkt_ (PPP_). This tripartite onset shape is thus reconstructable for PIE, 102 simply because *pktɛ́n- is presupposed both by Greek κτɛίς, which shows deletion of the initial *p, and by Latin pecten (< *pectēn), which exhibits a secondary post-PIE schwa-epenthesis (inserted to the initial syllables of oblique forms, see Figure 37). Usually, morphologically underlying forms were syllabified as a sequence of a stressed syllable and an unstressed syllable (σ́ σ) (with σ́ = σ̍). This is usually termed a trochaic foot in metrical phonology.103 A sequence of an initial stressed light syllable, e.g. ['pɛ]σ and a subsequent unstressed syllable ['kus]σ is thus also termed a trochaic foot; see the following Figure: ω ⎜

F σ́1 ⎜ ⎜

σ2 ⎜ ⎜

['pɛ]σ [kus]σ Figure 36. Trochaic template of underlying word forms

PIE morphologically underlying (and non-reduplicated) forms were predominantly syllabified either as a closed or heavy monosyllable, that is, as a so-called “moraic trochee”,104 as a trochaic foot (σ́ σ), or as a dactylic foot, which consists of a trochee, extended by a second unstressed syllable (σ́ σ σ). The syllabification of morphologically underlying forms was metrically left-headed and had a trochaic metrical/rhythmical disposition. On the other hand, it has now increasingly become clearer that morphologically derived (and non-reduplicated) forms were either syllabified as a closed or open syllable with the nucleus more towards the right word-edge (and I will term this a “moraic iamb” in parallel with Hayes’ term “moraic trochee”) or they were syllabified as an iambic foot (σ σ́), extended iambic foot (σ σ́ σ) or an anapaest (σ σ σ́), which consists of an unstressed initial syllable and a subsequent iambic foot. The syllabification template of the non-acrostatic nonbasic and derived forms such as the abl. *pktnɛ́s from *pktɛ́n was thus metrically right-headed and had an iambic metrical disposition. The derivational syllabification rule was thus: If possible, syllabify a morphologically nonacrostatic non-basic and derived word ω´ as a monosyl102 103 104

Pace Byrd 2010. Cf. Hayes 1995: 71. Ibid.

23

labic “moraic iamb” and put the stress more towards the right-edge, or syllabify it as an iambic or extended iambic foot (or as an anapaest), if possible with unstressed light and open syllable (σ̆1) and stressed heavy and closed syllable (σ́2). Such a foot is illustrated by Figure 37 below. ω ⎜

F σ́2

σ1 ⎜ ⎜

⎜ ⎜

[pk]σ [tnɛs]σ Figure 37. Iambic template of derived word forms

Outside the verb inflection and the verb-like adjective inflection, there was a tendency to ommit initial heavy syllables in the iambic foot (σ̆1 σ́2). Initial syllables of iambs, extended iambs, and anapaests, who had their word accent on the non-initial syllable, were often light and open syllables. This tendency conforms to the principle of NOCODA of unstressed syllables.105 However, this tendency did not hold for the PIE acrostatic inflectional patterning. Non-basic and derived forms belonging to the acrostatic patterning had a different syllabification template. The acrostatic derivational syllabification had a trochaic disposition. The rule was: If possible, syllabify a morphologically derived (acrostatic) word ω´´ as a trochee or dactyl with an initial heavy, that is, long or closed syllable (σ̄1) and do not change the position of the stress. Such a trochaic foot can be illustrated by the abl. forms of *pɛ́ːku in. ‘cattle, group of domestic animals’ and *ɠɔ́nu in. ‘joint, knot, angle’. ω ⎜

F σ́1 ⎜ ⎜

σ2 ⎜ ⎜

['pɛk]σ [uɛs]σ ['ɠɛn]σ [uɔs] σ Figure 38. Trochaic template of acrostatic derived word forms, e.g. *pɛ́kuɛs (abl.-gen. of *pɛ́ːku) or *ɠɛ́nuɔs (abl.-gen. of *ɠɔ́nu)

Oblique forms of classified nominal stems of the acrostatic inflectional types usually had such a metrical structure. Their initial closed syllable aligned with the rightedge of the root or inflectable base. PIE acrostatic stems such as, e.g., *pɛ́ːku in. ‘cattle’ thus lacked ablative-genitive forms of the structure †pɛ́kɛus, whereas such a form *pkɛ́us when derived from an underlying “proterokinetic” stem *pɛ́ku- an. ‘single dometic animal, sheep’ was well-formed. Therefore, it can be concluded that the morphological strategy of vowel transposition was in a sense sensitive to acrostatic stress assignment. The transposition of an unstressed vowel /ɛ/ to a position outside the root or inflectable base yielding an initial stressed light and open syllable, e.g. †pɛ́kɛus †['pɛ]σ[kɛus]σ was disfavoured and ommitted in the acrostatic inflectional types. Therefore, the 105

Cf. e.g. Hall 2011: 332 with references.

Roland A. Pooth

24

acrostatic inflectional types perhaps also lacked a broken locative form of the shape †pɛ́kɛu †['pɛ]σ[kɛw]σ, but had ‘sound’ locatives instead, e.g. *pɛ́ːkui → ['pɛː]σ[kwi]σ ~ ['pɛːk]σ[wi]σ (?), whereas non-acrostatic inflectional types had a ‘broken’ locative form, e.g. *liɛkwɛ́n. Furthermore, metrical templatic structures can serve as a reasonable motive for why PIE exhibited so-called “schwebeablaut”, e.g. of abs. *liɛ́ːkwr, loc. *liɛkwɛ́n, abl. *liɛkwnɛ́s (> Vedic yaknás). The rule once maybe ran as follows: An open initial syllable with NOCODA of unstressed syllables, e.g. *['liɛ]σ instead of *['lɛi]σ was favoured in forms with an iambic metrical structure.

(71) *lɛ́ikwr ~ *lɛ́ipr in. ‘remaining(s)’ → loc. *likwɛ́n ~ *lipɛ́n → abl.-gen. *likwnɛ́s ~ *lipnɛ́s, cf. Toch. B (°)l(y)ipär, lyīpär n. ‘remainder, residue, remnant’, cf. Adams 2013 II: 602: “(as if) from PIE *léipr̥” (72) *lɛ́ikwɔr (~ *lɛ́ipɔr) → loc. *likwɔ́n → abl.-gen. *likwnɔ́s

(73) but *liɛ́ːkwr ‘human sticky mass: liver’ → loc. *liɛkwɛ́n → abl.-gen. *liɛkwnɛ́s

“Schwebeablaut” was perhaps favoured in forms such as

*liɛ́ːkwr to avoid a tri-moraic super-heavy initial syllable *['lɛːi]σ. It can be concluded that PIE templatic morphology

was intimately intertwined with PIE metrical structural templatic rules and PIE syllabification rules or tendencies.

7 Gender, individuation, and number 7.1 Gender and individuation Despite all previous claims, PIE lacked a tripartite gender system, as continued in Vedic, Greek, Latin, and still preserved, e.g., in Modern German. Instead, PIE had what may be termed an inferential and conceptually induced animate and inanimate gender distinction. This type of gender assignment system was triggered not so much as a lexical feature of a given nominal stem, but by the immediate referential semantics with regard to individuation and agency, agentivity and animacy, more precisely by AGENTIVE CAUSATION ABILITY intertwined with NOMINAL ASPECT in the given reference act. individuated/ singulative nominal aspect AGENTIVE POWER/ NO AGENTIVE POWER/ is able to be agent: is unable to be agent: animate being inanimate thing *pɛ́ku*pɔ́ku‘domestic animal’ ‘fleece’ ANIMATE

nonindividuated/ plurative nominal aspect *pɛ́ːku‘livestock, cattle’

INANIMATE

Figure 39. PIE gender assignment; orange: animate

As illustrated by Figure 39 and as confirmed, e.g., by Vedic paśú- m. ‘individuated domestic animal, sheep’ but páśu- n. ‘livestock, cattle’ as well as Old Armenian asr ‘fleece’ (™pók̑ur 106 ), the animate (agentive) gender was assigned to PIE nominal stems denoting an individuated human being or animal or another individuated entity capable of being a conscious, controlling and willing agent. Inanimate gender, on the other hand, was assigned 106

Cf. Tremblay 2003.

to nouns denoting individuated entities incapable of being an agent, e.g., to natural products and human artefacts and inanimate things. Inanimate gender was automatically assigned to a noun denoting a semantically plurative entity, that is, a group, an uncountable mass or substance or another nonindividuated entity, including an abstract noun or verbal noun, regardless from whether the reference was to a group of human beings or animals or things; see Figure 39. Speakers of PIE thus had a cultural knowledge of natural agency/agentivity. They classified natural entities by knowing which entities are capable of doing something consciously, with a motive, willingly controlling the beginning and end of the action, and which are not. But there were cultural peculiarities. The PIE word for ‘word’ *uɔ́kw-, for instance, had animate gender even though it is a human product and an outcome of speaking. It had animate gender, because a word usually causes an effect or affection in another participant (e.g. by magic speech). Natural products (e.g. *χáɓl- in. ‘apple’107) were always inanimate, because they were seen as a products of natural causers, whereas the corresponding ‘tree’ (e.g. *ɗáru- an. ‘tree’) was animate, because it is was seen as the natural producer (causer) of fruit, bark, wood, leaves, etc. The PIE word for ‘heart’ *kɛ́rɗ- had inanimate gender. It was thus conceived of as having little conscious control or will―as a muscle which is controlled by a superior magic natural force called ‘life’. The ‘vivid existence, life force, power of life, life energy, vigour’ *ʔɛ́su- itself was exceptionally animate, even though this concept is an abstract concept. There were some other incoherencies or discrepancies which were due to the given cultural ideology and world view.

7.2 Plurative semantic congruency Differently from the classical IE languages Vedic, Greek, Latin, etc., the proper PIE parent language had no feminine gender, even if it is reconstructable for a pre-ProtoAnatolian period (vide infra). Instead, PIE exhibited what may be termed semantic congruency with regard to nominal aspect marking. Semantic congruency is defined here as formal agreement of nominal aspectual (singulative, singular; plurative, plural) marking of the modifyer within the NP which is solely motivated by the nominal aspectual semantics of the given head, that is, by individuation vs. Its relational derivative *χaɓɛ́l- also had a meaning ‘someone’s pancreas, gland’ which was animate because of its nature as a producer of a body liquid, cf. Modern Icelandic epill m. ‘id.’ < PG *apila- (cf. Kroonen 2013: 32). The root *χaɓ- was perhaps the PIE root for ‘juice, fruit liquid, liquid of an animate producer’, thus *χáɓl- ‘the roundish juicy one’. Recall that the PIE word for ‘(lifely) water, river’ exhibited several variants. Hitt. ḫapa- c. ‘river’, OIr. aub, gen. abae, MWe. afon ‘river’ point to *χab̤- or *χaɓ- ‘lifely/flowing water, river’ (Lat. amnis ‘stream, river’ is ambiguous). But Vedic ā́p- f. ‘flowing water’ and other cognates clearly point to a root with voiceless stop *χ_p-. This may be related to *χ_kw- of Lat. aqua ‘water’, Goth. aƕa ‘flowing water, river’, since PIE also had a few roots exhibiting variants with either *p or *kw. Note that PG *sapi-, *safan-/sappan- n. ‘juice’ (cf. OHG sapf, saf, ON safi, OE sæp n.) also exhibits a variation of *p and *f and may thus be a derivative of prefixed variants *s-χaɓ- ~ *s-χap-, cf. Lat. sapa f. ‘must, new wine boiled thick’ (which is rather not related to sapiō from *s-ʔ_p‘to perceive, taste’ (cf. LIV, p. 519), which may be a derivative of *ʔ_p‘to take’) 107

Proto-Indo-European Nominal Morphology. Part 1. The Noun

nonindividuation of the head of the NP. PIE plurative collective nouns such as, e.g., *g̈ɛ́ːsr- in. ‘hand, classified as a collective (as a group of fingers)’, as well as nonindividuated abstract nouns, e.g. *mɛ́nti- in. ‘thinking’, which were always inanimate, non-obligatorily triggered the use of plurative (abstractive, collective, generic, augmentative; glossed COL here) or plural (PL) number forms of verb-like adjectives, as illustrated by the examples below. (74) a.

*mɛ́nti ulkwáχ thinking.ABS dangerous.COL.DTR\VLAD.ABS lit. ‘thinking, they (collective) are dangerous’

25

within the NP was non-obligatory. The corresponding singular-singulative forms could be used instead. However, in the later post-PIE period the respective constructions were reanalyzed as constructions with feminine gender agreement and many former inanimate abstract nouns (e.g. *mɛ́nti ‘thinking’) were thus lexicalized to post-PIE feminine nouns. (77) a.

b.

b. *mɛ́nti ulkw-r_ɔ́ dito dangerous-PL_DTR\VLAD.ABS lit. ‘thinking(s), they are dangerous’

c.

(75) a.

*mɛ́nti ulkw_ɔ́ dito dangerous_DTR\VLAD.ABS lit. ‘thinking, (it is) dangerous’ uɛ́ɗɔr tɛ́pa_χ water.GNR.ABS hot.DUR_COL.DTR\VLAD.ABS ‘water(s), they (colllective) are hot’

b. uɛ́ɗɔr tɛ́p_r_ɔ (~ tɛ́pɔr) dito hot.DUR_PL_DTR\VLAD.ABS ‘water(s), they are hot’

c.

uɛ́ɗɔr tɛ́p_ɔ dito hot.DUR._DTR\VLAD.ABS ‘water(s), it is hot’

Furthermore, PIE nonindividuated collective nouns and abstract nouns and some other nouns (e.g. *nɔ́χu- in. ‘boat’) triggered a non-obligatory agreement of the modifyer within the NP with regard to the CL3 (collective, group-like) classifying suffix, as illustrated by (a) below: (76) a.

*nɔ́χu ʔɔːkú-χ ~ʔɔːkɛ́u-χ boat.ABS swift/swiftness.RELATIONAL-CL3 lit. ‘boat, swiftness-ish one’

b. nɔ́χu boat.ABS ‘id.’

ʔɔːkú ~ ʔɔːkɛ́u swift/swiftness.RELATIONAL

Thus nouns like *nɔ́χu- in. ‘boat’ also triggered this formal agreement. In my view, the PIE word for ‘boat’ was inanimate and acrostatic (abs. *nɔ́χu, abl. *náχuɔs). It was remodeled to a feminine noun in Vulgar Post-PIE and was paradigmatically leveled to *nɔ́χus ~ *náχus, *náχu̯os (whence Vedic nā́uṣ f. and generalized nā́v-). It was not inflected like *ɠwáʕu- an. ‘cow’. The PIE collective number agreement (e.g. PIE *nɔ́χu ɗɔrúχ ~ ɗɔruíχ ‘a woodencollective boat’) perhaps classified its possessor as a group or as belonging to a group: a boat was very presumably not owned and controlled by a single person at that time, but by a group of people. A social sharing of particular human artefacts within a group thus maybe triggered collective number agreement of a modifier of such a head noun.108 As just mentioned, in PIE this formal agreement A typological comparandum is ‘o possession in Tongan (Austronesian, Tonga), e.g. ko e huo ‘o Sione PRS DEF.ART hoe O.POSS John ‘this is John’s hoe’, but ko e hele ‘a Sione PRS DEF.ART knife A.POSS John ‘this is John’s knife’; cf. Bennardo 2000: 269: “In the past, a hele [‘knife’] as made of kafe (bamboo) or ngeesi fingota (shell), and a huo [‘hoe’] was made of langa (stone), [...] not everybody had access to a huo; many people shared one”. 108

PIE non-obligatorily plurative agreement mɛ́nti ulkwáχ thinking.ABS dangerous.COL.DTR\VLAD.ABS ‘dangerous thinking’

Post-PIE feminine gender agreement *mɛ́nti-s u̯lkwáχ [wl.'kwah] ~[wl.'kwaː] thinking-NOM.SG.F dangerous-NOM.SG.F ‘dangerous thinking’

This scenario entails that nominative and accusative case forms of feminine stems, e.g. nom. sg. f. *méntis, acc. sg. f. *g̑ɦesérm̥ were created not before a post-PIE period. Several acc. sg. feminine forms were derived from previous broken locative forms, e.g. *g̈ɛsɛ́r → *g̑ɦesérm̥ (+ *-m), simply because the broken locative was used in O109 or P position within the PIE antipassive construction and was later reanalyzed as accusative form. The fact that acc. sg. f. forms resemble such broken loc. forms thus strengthens the reconstruction of a PIE antipassive construction with locative forms of inanimate stems in O or P position. It is important to recognized that the Hittite acc. sg. common gender form kiššaran ‘hand’ strongly speaks in favour of a loss of this new feminine gender from Vulgar Post-PIE to Proto-Anatolian. In my view, pre-ProtoAnatolian must have participated in the previous PIE to Vulgar Post-PIE lexicalization of a feminine gender. Otherwise the existence of acc. forms like Hittite acc. sg. kiššaran of common gender nouns corresponding to feminine nouns of other IE languages is very hard to understand within the overall (and very plausible) scenario that the feminine agreement emerged from previous inanimate collective forms like the one given above. Therefore, the loss of the feminine gender is datable to a period after the creation of new Vulgar Post-PIE feminine nominative and accusative forms of original PIE inanimate stems. (such as, e.g., acc. sg. *g̑ɦesér- m̥). It is inferential that the creation of a feminine gender was a kind of ‘by-product’, so-to-speak, of the loss of the old system of nominal aspect marking and the overall loss of the system of nominal classification, as well as the alignment change to a nominative/accusative type. Therefore, most of the traditionally reconstructed feminine nom. sg. forms in -s (e.g. ₠méntis) as well as acc. forms like ₠g̑ɦesér-m̥ can be regarded as postPIE innovations and the respective correspondences can be interpreted as secondary equation formulas. In my view, it is better to delete feminine nominative and accusative forms like ₠ménti-s and ₠ménti-m from the inventory of reconstructable PIE proper forms.

7.3 Deponency of collective number As outlined elsewhere, PIE exhibited grammatical deponency (besides verbal lexical deponency). PIE verbal col109

Cf. Dixon 1997, 2010, 2012.

Roland A. Pooth

26

lective number marking was a deponent number marking, because of its semantic nature as deviation from the transitivity prototype. Participants within a joint actional event were conceived of as a non-discrete cluster participant with a low degree of individuation deviating from the feature [+indiviuated A vs. O] of the transitivity prototype. 2nd and 3rd person collective number forms were formally identical and marked by the detransitive suffix -χin addition to transfixation.110 Examples: (78) a. *χnɛ́r_ɛ_s χɔ́ui g̈wn-t_á_χ man_PL_ERG sheep.ABS slay-DIR_NDUR_2/3COL:DTR lit. ‘men (countable) slew sheep together’

b. *χɔ́ui χnɛ́r_ɛ_s g̈wn-s_á_χ sheep.ABS man_PL_ERG slay-INV_DNUR_2/3COL:DTR lit. ‘as for sheep: men (countable) slew them together’ c. *χnɛ́r_ʔ χɔ́ui-m g̈wn_á_χ sheep-ALL slay-ITR:NDUR_2/3COL:DTR man_ASS lit. ‘man & company together slew sheep’

Excursus: The Vedic 3rd dual active ending -tām, pointing to ₠-tah2(m), is still evidently also found as a middle secondary (aorist) ending in Early Vedic,111 cf. 3rd du. aor. ind. middle adhītām, which synchronically must belong to the paradigm of the aorist middle adhithās, adhita, adhīmahi, ...

transitivity prototype and had to be used within the PIE antipassive construction. (80) a. *χnɛ́r-ʔ χɔ́ui-m-s g̈wnáχ man-ASS sheep-ALL-PL slay:ITR:NDUR:2/3COL:DTR ‘man & company together slew (at) sheep’

b. *χnɛ́r_ɛ_ʔ χɔ́ui-m-s g̈wnáχ man_PL_ASS sheep-ALL-PL slay:ITR:SEM:2or3COL:DTR ‘man/men & company together slew (at) sheep’

c. *χnɛ́r_a_χ χɔ́ui-m-s g̈wnáχ man_PL_COL sheep-ALL-PL slay:ITR:SEM:2or3COL:DT ‘men (as collective group) together slew (at) sheep’

Alongside the intransitive antipassive construction there was a transitive construction including ergative plural forms for countable plural agents and collective-sociative inverse-transitive and direct-transitive forms, marked by the inverse suffix -s- and the direct suffix -t-, see examples (78a) and (78b) above. Thus the use of the antipassive construction triggered by associative and collective forms in agent or causer role ran in parallel to the use of this construction with inanimate stems in causer role: (81) a.

(79) RV 10.4.6 tanūtyajeva táskarā vanargū́ raśanābhir daśábhir abhí adhītām ‘... ... ... the two were fixed.’ “So wie zwei den Leib hingebende, im Wald umhergehende Räuber, mit den zehn Zügeln (raśanābhir daśábhir, sc. mit den Fingern) wurden die beiden (sc. Reibhölzer) [...] festgemacht” (Pooth 2011).

A middle origin is also evident in the case of the 2nd and 3rd person dual forms jániṣṭām (RV), jániṣṭam (AV) of the iṣaorist stem (á-)jániṣ-ta, because this stem is otherwise middle tantum. Another otherwise middle tantum paradigm includes the “exceptionally active” dual forms AVP 1.92.1 vr̥ṇītām and RV 1.180.4b avr̥ṇītam. It has to be concluded that these endings clearly belonged to the PIE detransitive voice category.112 In Vulgar Post-PIE, therefore, additional new middle endings had to be created. End of excursus. Following Eichner, COLLECTIVE number was a morphosyntactic nominal number category in PIE. 113 Nominal collective number forms, as well as associative number forms had to agree with the above mentioned collective verb forms. Only later, these collective verb forms developed into IE 2nd and 3rd dual forms.114 As outlined above, PIE also had associative number forms. The first two examples given below can illustrate its use. Associative and collective number forms of the ergative case and the allative-dative case, however, did not exist in PIE, because an associative or collective agent or causer deviates from the 110 Compare the so-called “D-effect” of the Athabascan 1st plural forms, cf. Thompson 1996: 361, which has been identified with middle voice marking. Thompson’s example from Mattole is ʔi-š-aŋ IND-1s-eat ‘I eat’ vs. ʔi-di-d-aŋ IND-1p-MID-eat ‘we eat’. The “D-effect” provides crosslinguistic evidence that readings like these can be marked for detransitive voice. 111 Cf Pooth 2011. 112 Ibid. 113 Cf. Eichner 1985, Litscher 2014. 114 Cf Pooth 2011.

b.

c.

antipassive construction with inanimate “A” *ɗɔ́ru b̤nɔ́ χnɛ́r-m wood.ABS injure:NDUR:DTR:ITR:3SG man-ALL ‘lance injured (at) man’

ungrammatical construction with inanimate “A” †ɗɔ́ru b̤ɛ́n-t χnɛ́r wood.ABS injure:NDUR:AGT:SG-DIR\3 man.ABS ‘lance injured man’

noncanonical antipassive with animate “A” *χnɛ́r-s ʔrɔ́ pχtɛ́r-m man-ERG dito father-ALL ‘man accidently/unwillingly hit (my) father’

A topical inanimate causer or cause in the absolutive case could not be cross-referenced by an agentive-active transitive verb form of the predicate. It also triggered the use of the antipassive construction, as illustrated by example (81a) and (81b). It can be concluded that associative and collective forms of both animate and inanimate nouns were treated as if they were (quasi-) inanimate nominal forms. This perfectly fits to the idea that inanimate gender assignment was automatically triggered by nonindividuated concepts including plurative nominal stems, as outlined above; see Figure 39.

7.4 Gender and number As already mentioned, PIE nominal number distinctions differed depending on the given gender assignment. PIE animate nouns (almost always) were countable nouns115 and had distinct singular vs. plural vs. associative vs. collective forms. In addition, they had associative plural and collective plural forms, as just mentioned. Inanimate nouns, even when their meaning implied reference to an individuated entity like ‘a piece of wood’ were transnumeral set nouns.116 They displayed associative forms and an optional collective (and suppletive plural) forms (e.g. *ɗɔ́ru, *ɗɔ́ruʔ, *ɗɔ́ruχ); see Figure 42. 115 116

Cf. Rijkhoff 2002. Ibid.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

27

possessor role, but not the “A” case of a canonical ergative language; 2. absolutive for patient, theme, direct goal. Extended ABS core cases were: 3. allative-dative for ALL-DAT indirect goal, less affected patient and, LOC-ALL recipient; 4. locative-allative for goal, SOC-ASS place, experiencer; 5. sociativeFigure 40. Internal derivation of oblique case forms from core and extended core case forms associative for companion and path. Oblique cases were: 6. ablative-genitive for source, comparee; 7. partitiveSG PL ASS ASS.PL COL COL.PL genitive; 8. benefactive-purposive for 1. ABS *χnɛ́r *χnɛ́rɛ *χnɛ́rʔ *χnɛ́rɛʔ *χnɛ́rχ *χnɛ́raχ beneficiary, abstract goal; 9. comitative2. ERG-GEN *χnɛ́rs *χnɛ́rɛs instrumental for companion, instrument. 3. ALL *χnɛ́rm *χnɛ́rms 4. LOC *χnɛ́ri *χnɛ́ris *χnɛ́riʔ *χnɛ́riχ Animate nouns also had vocative forms 5. SOC *χnɛ́rʔ (10.), see Figure 41. 6. ABL *χnrɛ́s Inanimate nouns generally lacked the 7. PAR *χnrɛ́m 8. BEN *χnrɛ́i allative-dative, that is, the recipient role 9. COM *χnrɛ́ʔ case; see Figure 42. The ergative-genitive 10. VOC *χnɛ̀r *χnɛ̀rɛ *χnɛ̀rʔ *χnɛ̀rɛʔ *χnɛ̀rχ *χnɛ̀raχ case form of inanimate nouns could not Figure 41. Case forms of animate nouns [Key: red: core and extended core case forms; yellow: be used in agent or causer or ergative oblique case forms (transnumeral); pink: loc. pl. forms (could be used as general oblique pl. forms); grey: vocative forms] function of this form, but it was used in genitive (possessor) function, e.g. *uɔ́iks ‘settlement’s’, e.g. *uɔ́iks *pɔtí ‘the settlement’s master’; I will return to this in the TRANSNUMERAL ASS COL following sections. The vocative case form of inanimate 1. ABS *ɠɔ́nu *ɠɔ́nuʔ *ɠɔ́nuχ nouns was not derived from the absolutive form, but from 2. GEN *ɠɔ́nus 3. ALL the broken locative form by tonal change. The paradigm 4. LOC *ɠɔ́nui *ɠɔ́nuʔ *ɠɔ́nuiχ given in Figure 42 is the one of PIE *ɠɔ́nu- in. ‘joint, angle, 5. SOC *ɠɔ́nuʔ hook, limb, link’. 6. ABL *ɠɛ́nuɔs syntactic status of base core core extended core extended core extended core

7. 8. 9. 10.

PAR BEN COM VOC

basic case ERG-GEN

underlying case form *pɛ́ku-s *pɛ́ku *pɛ́ku-m *pɛ́ku-i *pɛ́ku-ʔ

derived case form *pkɛ́u-s *pkɛ́u *pkɛ́u-m *pkɛ́u-i *pkɛ́u-ʔ

derived oblique case [derived nominal stem] ABL-GEN [relational stem] PARTITIVE-GEN BENEFACTIVE-PURPOSIVE COMITATIVE-INSTRUMENTAL

*ɠɛ́nuɔm *ɠɛ́nuɔi *ɠɛ́nuɔʔ *ɠɛ̀nɔu

Figure 42. Case forms of inanimate nouns [Key: blue: core and extended core case forms; grey: oblique forms, voc.; light green: loc. pl. forms]

8 The PIE case system PIE had an elaborate system of 10 cases, typologically comparable to the case system of many Pama-Nyungan languages of Australia.117 PIE case forms were marked by means of both suffixation and transfixation. Case marking was initially provided by suffixation of the four semantic role suffixes -s, -m, -i, -ʔ, and zero combined with internal modification, which was internal inflection. It is thus another typological peculiarity of PIE that the oblique case forms were derived from the corresponding nonoblique or core case forms by internal modification, more exactly by vowel transposition, as outlined in detail in the preceding sections; see Figure 40. The PIE case system thus included a formal distinction between core and extended core cases versus relational or attributive or oblique or adjunctive cases. It can be concluded that a syntactically and pragmatically motivated distinction between core and referential marking versus non-core oblique or adjunctive marking was morphologically coded separately from the encoding of semantic roles. PIE animate nouns, e.g. *χnɛ́r- ‘man’ (see Figure 41), formed the following 10 case forms. Core case form were: 1. ergative-genitive, which was the case for the agent or 117

Cf. Blake 2001; Pooth 2014c.

8.1 Alignment In terms of the convenient S, A, O (or P) typology 118 PIE was a “fluid-S ergative language”, that is, a language with semantic alignment of the active type, as illustrated by the following examples. Note that the overall meaning of the PIE root *p_ɗ- was close to German fallen ‘to fall’, fehltreten ‘to slip’, treten ‘to kick’, auftreten ‘to appear’, eintreten ‘to enter’, etc. Translations are as literal as possible: (82) a.

b. c.

*χnɛ́r pɔ́ɗ man.ABS fall/go.down:TRANSITIONAL :DTR:ITR:3SG ‘man fell, slipped’, literally ‘it fell man’ *χnɛ́r pɛ́ɗ-t man.ABS fall/.../kick:NONDURATIVE:AGT:SG-DIR\3 ‘he (topical/proximative) kicked the man’ *χnɛ́r-s pɛ́ɗ-t man-ERG dito ‘man (willingly) (topical/proximative) put his foot down, tread/trod, appeared (on the scene)’

PIE also had a canonical and a non-canonical antipassive construction. The latter was reanalyzed as post-PIE transitive nominative-accusative construction; see (83c). (83) a.

b.

118

*g̈wɛ́r pɛ́ɗɔ wild.ABS fall/...:DURATIVE/IMPERFECTIVE:SG-ITR\3 ‘wild (animal) (unwillingly) is falling/will/shall fall’ *χnɛ́r g̈nɔ́ pɛ́ku-m man.ABS slay:NDUR:SG-ITR\3 domestic.animal-ALL ‘man (unwillingly and by accident) slew a domestic animal’

Cf. Dixon 1997, 2010, 2012.

Roland A. Pooth

28

c.

*χnɛ́r-s g̈ɛ́nɔ g̈wɛ́r-m-s man-ERG slay:DUR/IPFV:SG-ITR\3 wild-ALL-PL ‘man is (willingly) hunting at wild animals’

Excursus: It has recently been suggested by Jasanoff 2015 that PIE had “split-S” alignment, motivated by two conjugation classes, formally very similar to the two Hittite conjugation classes. In his view, PIE had accusative alignment of the conjugation of the *g̈wɛ́nt(i) type, whereas it had a canonical ergative alignment of the conjugation of the *kɔ́nqɛ(i) type (cf. PIE *k_nq- ‘to hang, to feel unsafe’). In Jasanoff’s scenario, the accusative alignment was “mechanically extended” and yielded the “protomiddle”. But the major problem of Jasanoff’s proposal is of course that PIE verbs cannot be classified in terms of two entirely lexical conjugations the Hittite way. Even if PIE *kɔ́nqɛ and thus *k_nq- was a deponent detransitive verb, this did not hold for verbs like *ɗáʕt ‘s.o. gives, gave s.th. to s.o.’ vs. *ɗɔ́ʕɛ(i). Both were formed from *ɗ_ʕ- ‘to give, spend; take, receive’. The lexicalization and semantic specification of Hittite daḫḫe, dāi ‘takes’ as a verb of the ḫi-conjugation was a post-PIE innovation, and the comparative evidence rather speaks in favour of a verbal root-and-pattern morphological type including multiple binyans, much more than just two.119 Reconstructable verb forms like *kɔ́nqɛ(i) have to be interpreted as detransitive forms of underlying deponent verbs within a reconstructed proto-synchronic description of PIE. As a rule, detransitive forms were labile verb forms, which still holds for many middle forms in Early Vedic.120 These labile verb forms could be used semantically transitively (with factitive-causative meaning) within the PIE intransitive antipassive construction, a construction that preceded both the thematic conjugation and the pre-ḫi-conjugation. 121 In the scenario proposed here, the accusative alignment was entirely post-PIE and emerged as a parallel reanalysis of the PIE antipassive construction with allative-dative case of animate nouns (marked by *-m and *-m-s). The PIE absolutive case is continued by the IE neuter nominative-accusative forms, because these were not allowed to occur as transitive agents in the ergative case and also lacked a recipient allative-dative case. Examples are: (84) a.

*tɛ́uɛ

ɗɔ́ru

lɔ́ɢ̤ɛ(i)

2sg.GEN wood.ABS lay/lie/bend:STAT:DTR:ITR:3SG(PROG)

‘your wood (lance) always lies/is always lying there’

Ditransitive constructions including agentive transitive verb forms normally had indirective (patient identical to theme) alignment. 122 Animate recipients either had allative-dative or locative case marking (Key: P = patient, T = theme): (86) a.

b.

PIE direct transitive construction *χnɛ́r-s knq-t_ɔ́ ɗɔ́ru man-ERG hang-DIR_NDUR:DTR:3SG wood.ABS ‘man hang up a lance’

b. PIE antipassive construction *χnɛ́r-s knq_ɔ́ ʔɔ́g̈wi-m man-ERG hang_NDUR:DTR:ITR:3SG worm/snake-ALL ‘man made (at) the snake hung’ Cf. Pooth 2015 (see the sections on verb morphology). 120 Cf. Pooth 2014 on Early Vedic middle syntax. 121 Cf. Pooth 2009a. 119

mɛ́d̤u páʕ-t honey.ABS swallow:NDUR:AGT:SG-DIR P ‘man swallowed (a portion of) honey’

*χnɛ́rs mɛ́d̤u ɠwɛ́n-m ɗáʕ-t dito dito woman-ALL give/take/get.NDUR:AGT:SG-DIR T R ‘man gave (a portion of) honey to (a/the) woman’

When R was definite or an anaphoric third person pronoun the neutral alignment was perhaps preferred, that is, both R and T were in the absolutive case.123

(87) a.

b.

*χnɛ́rs ɠwɛ́n dito ABS R ‘man gave the

mɛ́d̤u

ABS

ɗáʕ-t

give/take/get.NDUR:AGT:SG-DIR

T woman (a portion of) honey’

*χnɛ́rs sí mɛ́d̤u ɗáʕ-s ABS ABS give/take/get.NDUR:AGT:SG-INV dito R T ‘man gave topic (a portion of) honey’

Note that non-oblique case forms of personal pronouns (first or second person) did not encode semantic roles, but had pragmatic alignment including a topical or proximative form (glossed TOP) opposed to a corresponding antitopical or obviative form (glossed OBV): (88) a.

b.

*χnɛ́rs ʔmɛ́ dito 1SG.OBV R ‘man gave me (a

mɛ́d̤u ɗáʕ-s

ABS

give/take/get.NDUR:AGT:SG-INV

T portion of) honey’

*ʔɛ́ tuɛ́ mɛ́d̤u ɗáʕ-m 1SG.TOP 2SG.OBV dito give/take/get.NDUR:AGT:SG-1 R T ‘I gave you (a portion of) honey’

However, PIE had indirective (P = T) alignment of personal pronominal clitics, whereas R (and the possessor role) was coded in the locative-genitive form:124

(89) a.

b. PIE antipassive construction *χnɛ́r-s lɔ́ɢ̤ɛ(i) ʔɔ́g̈wi-m man-ERG dito worm/snake-ALL ‘man (habitually) makes the snake lie down; is always making/bending the snake’

(85) a.

*χnɛ́r-s man-ERG

b.

*tú mɔi mɛ́d̤u ɗáʕ-s 2SG.TOP 1SG.LOC dito give/take/get.NDUR:AGT:SG-INV R T ‘You gave me (a portion of) honey’ *ʔɛ́ tɛ χnɛ́r(m) ɗáʕ-m 2SG.TOP 2SG.ACC ABS(ALL) give/take/get.NDUR:AGT:SG-1 T R ‘I gave you to the (a/the) man’

However, the syntactically intransitive antipassive construction exhibited neutral 125 alignment with allativedative case marking of animate P = T = R vs. locative case marking of inanimate P = T = R. Recall that inanimate nominal stems lacked allative-dative case forms and Cf. Haspelmath 2014-2015: 22; Dryer 1986. Cf. Haspelmath 2014-2015: 27ff. Cf. Pooth 2015. 125 Cf. Haspelmath ibid. 122 123 124

Proto-Indo-European Nominal Morphology. Part 1. The Noun

were coded by the broken or sound locative in the antipassive construction. (90) a.

*χnɛ́rs ɠwɛ́nm ERG

ALL

kuɔ́nm ɗʕ_ɔ́

dog-ALL give/take/get_NDUR:DTR:ITR:3SG

R T lit. ‘man made (at) woman take/receive (at) dog’

b. *χnɛ́rs ɠwɛ́nm

ɗɔ́rui ɗʕ_ɔ́ wood-LOC dito R T lit. ‘man made (at) woman take/receive (at) lance’ ERG

c.

ABS

*χnɛ́rs pɛ́ku-i cattle-LOC R lit. ‘man made (at) ERG

g̈rasmɛ́n ɗʕ_ɔ gras.LOC dito T cattle take/receive (at) gras’

The neutral alignment of the antipassive construction was due to the fact that the antipassive construction was syntactically intransitive and could not exhibit two core case forms, but was limited to only one of them.

8.2 Ergative, genitive, and ablative It was first recognized by the Leiden scolar van Wijk126 that the later IE nominative case suffix -s should be identified with the -s of the ablative-genitive. A formal identity of ablative case and ergative case is quite common among the languages of the world.127 There are a few archaic IE genitive forms that show full grade or even lengthened grade of the root, e.g. YAv. gen. sg. m. vāxš ‘speech’s’; see Figure 43. Furthermore, there are a few archaic IE genitive forms that show zero grade of the suffix. Tremblay has also added Avestan garəbuš (V. 9.38), maδuš (V. 14.17) ‘of wine’ and some more problematic cases;128 see Figure 43. GEN

*ɗɛ́m-s *nɛ́ɠwt-s *ɠwɛ́n-s *uɔ́ik-s *nɔ́ɠwt-s *uɔ́ːkw-s *máː-tr-s *b̤ráχ-tr-s

meaning ‘family’s’ ‘dusk’s ‘family’s’ ‘settlement’s’ ‘night’s’ ‘speech’s’ ‘mother’s’ ‘brother’s’

IE continuation OAv. də̄ṇg (də̄ṇg paiti-) Hitt. nekuz OIr. nom. and gen. sg. f. bé OPrussian Wais- of Waispattin, etc. Lat. nox ‘at night’ (adverbial gen.) YAv. gen. sg. m. vāxš Vedic gen. sg. f. mātúr Vedic gen. sg. m. bhrā́tur

Figure 43. PIE ergative-genitive or genitive forms

These archaic IE genitive forms can be entirely understood now.129 The meaning ‘dusk’s’ of Hittite nekuz does not make it a perfect candidate for being the original abl. of the PIE word for ‘night’. It is clear now that *nɛ́ɠwt‘dusk, getting bare’ was an action noun, whereas *nɔ́ɠwtin. ‘night’ was the corresponding derivative indicating its result or product: ‘night: product/result of dusk, getting bare’, marked by the discontinuous detransitive marker ɔ. The functional prehistory of forms like *nɛ́ɠwts is detectable by help of PIE *ɗiɛ́u- which is a relational “hysterodynamic” nominal stem with the meaning ‘s.o.’s bright one, shining one: sky, daylight’. The underlying PIE root was Cf. van Wijk 1902. Cf. Heine & Kuteva 2002, s.v. “ergative”. Cf. Tremblay 2003: 579, fn. 30. 129 Pace Schindler 1967a, 1972 126 127 128

29

*ɗ_i- ‘shine, be bright’. It was not a root noun, but was marked by the class 8 suffix -u- (skeletal *ɗ_i-_u-). The corresponding nom. sg. form obviously received the ‘neolengthened’ grade in Vulgar Post-PIE (*diéus → *diéːus > Vedic dyáuṣ, etc.130 In that period this stem was reanalyzed as a ‘neo-root-noun’. It is a relevant fact that Vedic dyáuṣ has two abl.-gen. sg. forms dyóṣ and divás. The first one (™di ̯ éus) is formally identical to the s-case form *ɗiɛ́u-s. It is inferential that this was the PIE ergative-genitive form before it received the ‘neo-lengthened’ grade as nominative singular marker in Vulgar Post-PIE. It has to be concluded that the Vedic abl.-gen. form dyóṣ constitutes an archaism. It is less plausible to interpret it as secondary. The second form divás shows the more regular gen.-abl. suffix of root nouns. There is further internal evidence: The Vedic word for ‘bird’ (nom. sg. m. véṣ ~ víṣ ‘bird’, acc. sg. m. vím, abl.-gen. véṣ, etc.) does not go back to a root noun “*h2uóis, *h2uéis”.131 It rather continues the PIE relational “hysterodynamic” *χuɛ́is ~ *χuís ‘s.o.’s bird; bird-like, birdish’ (skeletal *χ_u-i-). It is inferential that the Vedic gen.-abl. m. form véṣ < *χuɛ́is is thus another archaism that runs in parallel with dyóṣ. It can be concluded that both forms go back to PIE ergative-genitive forms. Latin avis m. ‘bird’, Umbrian avif, OArm. haw on the other hand, simply go back to the nonrelational paradigm *χáuis, *χuɛ́is ‘bird’.132 (NB. The problem of *χu̯V > Latin avV (avis) can thus be avoided; cf. *χi ̯V > *i ̯V in iuvenis with loss of *χ before the glide). The PIE words for ‘bird’ (skeletal *χ_u-i-) is etymologically related to the PIE word for ‘sheep’ (*χɔ́ui) and an underlying root *χVu- ‘dress, be dressed, clothe oneself’.133 The meaning given in the LIV, p. 275, however, is too specific (“Anm. 1: Ansatz einer Seṭ-Wurzel [...] ist [...] nicht sicher”). The etymology is now clearer: Many birds are beautifully dressed in feathers, cf. German Federkleid, etc.; and the sheep is the domestic animal that produces wool for garment and dresses. Under this perspective, it turns out to be easy to understand why these two PIE animal words look so similar. Another word that is ultimately derived form this root is the PIE word for ‘egg: product of bird’ (PIE *χɔ́ui). The idea of a compound “*(H)o-h2u̯i ̯-o-” is obscure and definitively too ad hoc. 134 Proto-Germanic *VjjV of *ajjaz/izpoints to a preceding consonant cluster (cf. PG *wajju < *uɔ́i ʔu): *wj > *jj. 1st possibility: It can continue the original PIE oblique stem of this inflectional type, e.g. abl.-gen. *χauiɔ́s ~ *χɔuiɔ́s, in parallel with *uɔ́ːkw, *uɛkwɔ́s (see Figure 31), syllabified as *[χa]σ['wjɔs]σ due to its iambic metrical structure, with *wj > PG *jj and resyllabification to *aj.jaz-. 2nd possibility: It can go back to *ajj V- < *ɔːjj V< *(χ)ɔːwj V-, syllabified as *[χɔː]σ['wjɔ]σ due to its iambic metrical structure with *wj > PG *jj, resyllabification plus shortening of the vowel, corresponding to what is continued by Latin and Greek, and Albanian ve ‘egg’, etc. 135 Latin ōvum ‘egg’, Greek ὠιόν (< ₠h2ōu̯i ̯ó-) both point to a para130 For the spread of the neo-lengthened grade as an innovative marker of nominative sg. forms cf. Pooth 2014d. 131 Pace Schindler 1969, see de Vaan 2008, s.v. 132 Cf. Beekes 1995: 175, pace Zair 2011: 304. 133 Cf. LIV, p. 275: “h2eu(H)- “(Fußbekleidung) anziehen”. 134 Cf. Zair 2011. 135 Cf. Zair 2011: 292.

Roland A. Pooth

30

digm with long vowel *χɔ́ːui, *χauiɔ́s (or *χɔuiɔ́s). This should have patterned like *uɔ́ːkw, *uɛkwɔ́s and can now be interpreted as a collective nominal stem ‘group of egg(s) (in a nest)’. A “Vr̥ddhi” derivation from an underlying ‘bird’ is ruled out, because we would expect †h2ēu̯i ̯ó-.136 There is no evidence for †h2óu̯̯i- ‘bird’, and the wrong vocalism is decisive.137 It is a much better to reconstruct a PIE plurative collective stem *χɔ́ːui ‘group of eggs (in a nest)’ as derived from an underlying inanimate stem *χɔ́ui ‘egg’. I reconstruct the following PIE paradigms: (91) a. *χáui, *χuɛ́is an. (following Beekes 1995: 175) ‘bird: the one who dresses/is dressed (in feathers)’ (“proterokinetic” underlying base) b. *χuɛ́is ~ *χuís ‘someone’s bird; bird-ish’ (HD) c. *χɔ́ui, *χáuiɔs in. ‘egg: product of bird’ d. *χɔ́ːui, *χauiɔ́s (or *χɔuiɔ́s) ‘group of eggs (in a nest)’ (→ verb-like adj. *χɔːuiɔ́-)138 e. *χɔ́uis, *χáuiɔs an. ‘sheep: the one who makes dressing/garment’, cf. Toch. B āuw, nom. pl. f. awi ‘ewe’

For the reasons given above, the archaic IE genitive forms given in Figure 43 should not be traced back to PIE oblique ablative-genitive case forms. In parallel with Vedic dyóṣ < *ɗiɛ́us these forms more plausibly go back to forms of the PIE ergative-genitive case in possessive function marking the possessor role. The two case forms ergativegenitive and ablative-genitive can now be analyzed as two subordinate instances of a superordinate skeletal sigmatic consonant frame case stem. Therefore, only different positions of the vowels of the vowel melody further distinguished these two forms, e.g. *ku_n-_s (+ ɔ́) → *kuɔ́ns ‘dog (animate agent)’ and ‘dog’s’ vs. *kunɔ́s ‘from dog(s)’. The oblique ablative-genitive case form was thus derived from the ergative-genitive case form by vowel transposition, as illustrated by Figure 40 and outlined in detail in the sections above. Therefore, the ergative case is interpreted here as a source core case, indicating the source of the animate agentive causation force.

8.3 The PIE genitive construction It is further inferential that the ergative-genitive case was used to encode the possessor role within a specific genitive (or possessive) construction. This inference implies that the oblique ablative-genitive case could only be used as an adjunct. It can be suggested that the ergative-genitive form, e.g. *ɗɛ́m-s (> OAv. 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. This inference further implies that the possessee role had to be additionally coded by the relational stem. Within this specific PIE genitive construction both possessor and possessee were marked (cf. Modern Turkish). Most presumably, these forms had to be in direct juxtaposition and no other word form was allowed to stand in between. One may term this genitive or possessive construction the “status constructus” of PIE: Cf. Zair 2011: 304. Pace Zair 2011. 138 PIE also had corresponding class 1b stems *χɔ́ui-s-, *χɔ́ːui-s-. A ‘broken’ locative = relational stem *χɔːuiɛ́s- ‘s.o.’s group of eggs’ is continued by Doric ὤεον, Argive pl. ὤβεα. 136 137

(92) a. b.

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

The possessive ablative-genitive construction, on the other hand, was a simple adjunct in apposition and rightattachment. Here, the possessee was not necessarily coded by a relational stem: (93) a.

b.

*pɔ́t(i)-m ... ɗmɛ́s master-all family:OBL:ABL/GEN ‘to a/the master ... the one of (from) the family’ *pɔ́t(i)-m ... uɛkwɔ́s master-all speech:OBL:ABL/GEN ‘to a/the master ... the one of (from) the speech’

Both constructions and both case forms simply merged from PIE to Vulgar Post-PIE, cf. Avestan gen. sg. m. vāxš besides vacō, Latin nox besides noctis. These are variants like Vedic dyóṣ and divás. It can be concluded that the few archaic IE genitive forms do not go back to ablativegenitive forms, but are relics of the ergative-genitive.

8.4 Absolutive, broken locative, and vocative PIE must have had a zero-marked or unmarked absolutive case. This inference is drawn on the basis of the IE unmarked or zero-marked neuter nominative-accusative case forms. It is inferential that the animate nominals stems also showed such a case marking before the PIE allativedative was reanalyzed to accusative.139 The zero case marking of the IE neuter stems is thus retrogradely generalized chronologically and transferred to the non-neuter animate nominal stems such as, e.g. *kuɔ́n- an. ‘someone’s dog’ → abs. sg. *kuɔ́n. Such a generalization is a perfectly valid and legitimate step within the process of internal reconstruction, against all comparative claims. It is inferential that all Vulgar Post-PIE variants gave up animate absolutive case forms due to the alignment change from the previous active “fluid-S” to nominative-accusative alignment, which, according to my view, happened parallely but relatively independently within the Vulgar Post-PIE variant or dialect cluster. According to the present analysis all post-PIE variants parallely switched into nominativeaccusative languages with a new animate accusative case form, e.g. Vulgar Post-PIE *k(u)u̯ónm̥, as well as new feminine nominative forms, e.g. Vulgar Post-PIE *méntis, as outlined above. But there are some left-overs of the absolutive case. First of all, the PIE abs. sg. form is reflected as Vulgar Post-PIE voc. sg. form, e.g. PIE voc. sg. *pχtɛ̀r > Vulgar Post-PIE *pχ́ter ~ *=pχter (> Gk. πάτερ, etc.). Secondly, the PIE abs. pl. form *pχtɛ́rɛ ‘(countable) fathers’ merged with the PIE sociative-associative forms *pχtɛ́rʔ ‘father & co.’, pl. *pχtɛ́rɛʔ ‘id.’ or ‘fathers & co.’. These three forms developed into Vulgar Post-PIE nom.-acc. dual forms. The Greek nom.-acc. dual m. πατέρε ‘two fathers’ continues one of the two first forms or both, whereas the third one is continued by the Vedic nom.-acc. dual form, cf. Vedic pitárā ~ pitárau. Note that it is impossible to 139

Cf. Pooth 2004a.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

generate an equation formula (Gk. πατέρε ≠ Ved. pitárā ~ pitárau). Thus, something different happened here. The simplest solution is to suggest that the Vedic -u goes back to the PIE enclitic particle *=u ‘then, and then, and also, in addition’ which perfectly fits to the original associative reading:140 (94) a.

PIE *pχtɛ́r_ɛ_ʔ =u father_PL_ASS =and.also ‘father and also his companion(s)’

b. > Vedic pitár-au father-NOM/ACC.DUAL.MASCULINE ‘two fathers’

The broken locative 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 internally derived from the absolutive singular form, as outlined above. Some nominal stems, mostly stems of the types marked by the discontinuous detransitive marker /ɔ/, also displayed an allomorphic zero-marked case form of the benefactive-purposive case, e.g. *d̤g̈mɔ́i ~ *d̤g̈mɔ́ > Hitt. loc. sg. n. tagnā (viz. the alleged “allative” or “directive”). In parallel with the broken locative, this form was also based on the zero-marked case stem. In parallel with the benefactive-purposive, it also developed into the new Vulgar Post-PIE dative case. These zero-marked case forms are still reflected as the Old Hittite “directive” dative forms, e.g. tagnā ‘to the earth’, ḫapā ‘to the river’; and there is no need to reconstruct a completely separate “directive” case for PIE. I finally suggest that some broken locative forms are reflected by neuter IE vocative sg. forms, e.g. *ʔuɔsɛ́u ‘at/to worthy thing’ → ‘(this one goes out) to the worthy thing, o worthy thing’. I suggest that such broken locative forms are the source, e.g., of the Vedic voc. sg. ending -o of stems like vasu-, e.g. vaso, etc. Excursus: The vocative sg. m. forms of the later thematic o-stems, e.g. ₠u̯l ̥́kwe ‘o wolf’ may go back to forms similar to 2pl verb-like adjective forms, used to modify a speech-actparticipant (SAP = 1st or 2nd person), with a vowel melody and VMT similar to *(ʔ)mɛ́nɛ ~ *(ʔ)mɛ́mɛ ‘(it is) mine’, *tɛ́uɛ ‘(it is) yours’, etc. PIE verb-like adjective SAP forms thus perhaps ended in *_ɛ_ instead of *_ɔ_ (3rd person), e.g. 2sg?/pl *lɛ́uqɛ ~ *lɛuqɛ́ ‘you are shiny, bright’, but 3sg *lɛ́uqɔ ~ *lɛuqɔ́ ‘is shiny, bright’, in parallel with 2pl verb forms, e.g. *stɛ́uɛ ‘you (pl.) are praising s.o.’, and in parallel with the formal opposition between *tɛ́ ‘you’ (2sg personal pronoun) and *tɔ́- ‘this’ (pronominal adjective). Later, these forms were perhaps interpreted as nominal equivalents of the new verbal Vulgar Post-PIE thematic 2nd sg. imperative forms, e.g. *bɦére ‘(you) carry!’ which also had *-e (PIE 3rd sg. detransitive intransitive *b̤ɛ́rɔ → Vulgar PIE *bɦére). The evidence for a sound law is scant. But I admit that more arguments are needed. End of excursus.

8.5 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 af140 “Jasanoff’s law”, as quoted, e.g., by Weiss 2009: 209 with references, was perhaps restricted to cases like PIE *ɗɛɗɔ́ʕɛ > Vedic dadáu.

31

fected and prototypically an experiencer of an animate sentience and affection. Since inanimate nominal concepts prototypically cannot be experiencers of an affection, these were semantically incompatible with the allative-dative case marking in PIE. Inanimate nominal stems thus did not show allative-dative case forms. Therefore, I take the allative-dative case in -m- for being the original recipient role case of PIE animate nominal stems. Verbs like *ɗ_ʕ- ‘to give, take, receive, obtain, get’ originally took the PIE allative-dative, that is, the ‘pre-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 a word order with recipient role before patient role, since PIE must have had discourse-pragmatically triggered word order. (95) *χnɛ́r-s χɔ́ui pχtɛ́r-m ɗáʕ-t man-ERG sheep.ABS father-ALL give:NDUR:AGT:SG-DIR\3 ‘man (agent) gave sheep to father’

(96) *χnɛ́r-s χɔ́ui pχtɛ́r-m ɗʕ-t_ɔ́ man-ERG sheep.ABS father-ALL give:NDUR-DIR_DTR3\SG ‘man (agent) gave/took sheep top father for his own benefit’

This morphosyntactic pattern differs from what is usually reconstructed by the comparative method within the traditional model. It has already been outlined above, however, that the traditional comparative morphosyntactic 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, however, the traditional comparative model is, to be clear, ‘pseudoPIE’ syntax. It is just similar to the actual PIE syntactic patterns. Grounded on comparison, it is inferential that the genesis of the dative case in *-ei is not datable to a period prior than Vulgar Post-PIE. It is internally evident that *-ei was paradigmatically generalized in a period with fixed vocalism of inflectional endings and loss of ablaut in these morphemes. 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 Post-PIE *mn̥-téi ̯-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 origin, the entire dative case in *-ei looks like a Vulgar Post-PIE innovation. I hereby draw the inference that *-ei spread as a new marker of the Vulgar Post-PIE 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 already mentioned, PIE had an antipassive construction including oblique allative-dative case marking (m-, pl. -m-s) of the patient or theme, e.g. *χnɛ́rs *stɛ́uɔ *kuɔ́nm ‘man was praising at/to dog’, which was reanalyzed as the new transitive construction ‘man praised dog’ in Vulgar Post-PIE. As a consequence, the former allativedative indirect goal and recipient (R) case became the new Vulgar Post-PIE accusative case for the direct goal or theme or patient role. From that very moment onward, Vulgar Post-PIE had neutral aligment (R = T = P) and more or less urgently needed a new dative case for the recipient (R) role. To suit this purpose, the suffix *-ei of the

Roland A. Pooth

32

benefactive-purposive was generalized as a new Vulgar Post-PIE dative case suffix, again yielding Vulgar Post-PIE indirective (T = P vs. R) alignment. Note that such a grammaticalization path benefactive → dative is widely attested. 141 The original PIE benefactive-purposive case, however, was mainly used to indicate a beneficiary role and a purpose, that is, an indirect goal: (97) *pχtɛ́r-s d̤ɛ́ʔt father-ERG put:NDUR:AGT:SG:3DIR ‘father made it for man’

χnrɛ́i man:BEN

(98) *pχtɛ́r-s ɠwɛ́mt uiɗɛ́i come:NDUR:AGT:SG:3DIR see:BEN father-ERG ‘father came to see ...’

For this reason the Vulgar Post-PIE innovative dative case has to be eliminated from a proper model of the PIE case system. I hereby dare to make the claim that the dative case with suffix(es) *-ei, *-oʔei → *-ōi should better not be reconstructed for any stage prior than Vulgar PostPIE. Instead, the forms of the Vulgar Post-PIE accusative case (e.g. *pχtérm̥) must now be functionally retransformed to the case forms they were before, namely PIE ANIMATE ALLATIVE-DATIVE case forms. This change is internally evident and inevitable, in my view. 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: (99) *pχtɛ́r-s páʕ-t mɛ́d̤uɔm father-ERG swallow:NDUR:AGT:SG-DIR alcoholic.honey:PAR ‘father swallowed (part) of the alcoholic honey-drink’

Thus, 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-, *ku_n-_m → *kuɔ́nm ‘at/to dog’ → *kunɔ́m ‘at/to part of dog(s)’. In other words, both the PIE allative-dative case form and the partitive-genitive case form were derived from an underlying INDIRECT, NOT TOTALLY AFFECTING OR TRANSFORMING GOAL case stem. Recall that there is some crosslinguistic overlap in this functional domain, cf. Heine & Kuteva 2002: 333f.: “partitive ← [...] possessive ← (3) dative, (5) locative”. This overlap may point to 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, used within the PIE antipassive construction, the animate dative-allative case could also be used in partitive-like function: (100) *χnɛ́r-s *pʕɔ́ *ʔɔ́gẅ im man-ERG swallow:NDUR:ITR:3SG:DTR snake-ALL ‘man swallowed at (part of) snake/worm’

Only later, the PIE partitive-genitive case form developed into the new Vulgar Post-PIE genitive plural case form. The new portmanteau suffix *-om, which orginated from forms like *kunɔ́m, was then completely generalized as a Vulgar Post-PIE marker of the genitive plural. As already mentioned, the o-stem suffix *-o- was extended by *-om yielding *-oʔom (with simple hiatus, see above). This sec141

Cf. Heine & Kuteva 2002: 54.

ondary suffix also was generalized to some extent (e.g. *χn(e)róm ~ *χn(e)róʔom > Vedic narā́m). Actually, the morphological spread of the new gen. pl. ending *-om is clearly confirmed by the Vulgar Post-PIE demonstrative gen. pl. *tóisom ~ *tóisoʔom, which goes back to a prior PIE demonstrative loc. pl. *tɔ́is that was pleonastically extended by *-om ~ *-oʔom. Of course it is still possible that the Gothic gen. pl. m. -e (e.g. dage, wulfe vs. -o otherwise) goes back to a Vulgar Post-PIE gen. pl. ending *-eː(m) with lengthening of *-em (*_ɛm) in paradigmatic analogy to its more productive variant *-oː(m) < *-oʔom. If this was true, the ending *-eː(m) would have to be interpreted as an allomorph of *-oː(m) which was transferrable to Vulgar Post-PIE o-stems (e.g. *u̯l ̥́kweː(m) ~ *u̯l ̥́kwoː(m) < *u̯l ̥́kweʔem ~ *u̯l ̥́kwoʔom parallel to *u̯l ̥́kwes(i ̯)o ~ *u̯l ̥́kwos(i )̯ o, etc.). There is no convincing explanation for why Gothic shows a gen. pl. in -e besides -o among the proposals that have been suggested so far. Most of these explanations suffer from being ad hoc and unconvincing. With regard to all the given IE ablaut variants (e.g. abl.-gen. sg. *-es ~ *-os) it is definitively not ad hoc to infer that this suffix simply reflects a Vulgar Post-PIE ablaut variant which was completely eliminated in all the other IE branches. But note that Gothic -e is not urgently needed to legitimate the internal reconstruction of PIE forms like *χnrɛ́m vs. *kunɔ́m. There is better evidence. The Vedic irregular gen. pl. form nr̥̄́n ‘(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 otherwise irritating formal overlap can now find a very easy diachronic explanation and can finally confirm a diachronic functional closeness of both cases (acc. pl. and gen. pl.). It is now inferential that Vedic nr̥̄́n simply goes back to a Vulgar Post-PIE allomorpic variant of *χnérn̥s ~ *χnŕ̥ns continuing the old PIE allative-dative pl. form *χnɛ́rms, but with Vulgar Post-PIE 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 (< VPIE *χneróʔom) besides an even younger variant nr̥ṇā́m clearly confirms that the genitive plural category was subject to abundant allomorphy and innovation within Vulgar PostPIE. It also confirms that the ‘pre-accusative’ was used as illustrated by example (100). Note that Archaic Latin acc. sg. m. = gen. pl. seruom confirms a prior formal identity of the allative-dative and the partitive-genitive of the PIE verb-like adjective inflection. After all, the PIE formal connection of both case forms is evident.

8.6 Locative and benefactive-purposive The locative (place and goal) case form was either based on the case stem with suffix *-i or on the zero-marked case stem. In parallel with the locative case, the PIE benefactive-purposive case had two allomorphic forms. One was based on the case stem with suffix *-i. The other one 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. With regard to the evident spread of the new dative sg. suffix *-ei within the Vulgar Post-PIE period, as just outlined, it is

Proto-Indo-European Nominal Morphology. Part 1. The Noun

inferential that the PIE detransitive benefactive-purposive case forms (*kunɔ́i, *d̤g̈mɔ́i, etc.) were remodelled to forms with the new dative singular suffix *-ei, e.g. *d̤g̈mɔ́i → Vulgar Post-PIE *dɦgɦméi > Hittite taknī, etc. Thus I do not regard forms like *dɦgɦméi (and *dɦgɦém) but *d̤g̈mɔ́i (and *d̤g̈ɔ́m) as original PIE forms belonging to *d̤ɛ́g̈ɔm. Note that a broken locative *d̤g̈ɔ́m with ɔ́ before m is clearly confirmed by Hittite tagān, whereas Vedic kṣámi is not decisive, since it continues a secondarily attached -i.

8.7 Company, instrument, manner, path The underlying skeletal case stem marked by the glottal stop suffix *-ʔ was the superordinate case stem for the notion of companion(s), company, instrument, manner, and path e.g. *ku_n_ʔ. This company, instrument, manner, and path suffix *-ʔ perhaps had a variant *-ɗ,142 since PIE exhibited some variation with regard to these two phonemes */ʔ/ and */ɗ/, e.g. *uɔ́ɗr in. ‘water’ ~ *uɔ́ʔr in. ‘id.’ (→ Ved. vā́ri- ‘id.’). The variant *-ɗ was perhaps restricted to the pronominal inflection. A Vulgar Post-PIE spread of the suffix *-ed to the nominal inflection is internally evident, cf. *u̯l ̥́kwoʔed (> Proto-Baltic gen. sg. m. n. *-ād, Vedic vŕ̥kād ‘from the wolf’) with morphological hiatus due to the suffixation of *-ed to the stem (in parallel with the new endings *-oʔei and *-oʔom). I suggest that the ending *ed was simply extracted from *ʔmɛ́ɗ (→*ʔm-éd). In any case, the suffix *-ɗ once encoded PIE pronominal oblique ablative case forms, e.g. *ʔmɛ́ɗ ‘from me’, *tuɛ́ɗ ‘from you’ (> Vedic mád, tvád, etc.). The suffix *-ɗ was perhaps a prior polyfunctional ablative and associative-sociative and comitative-instrumental suffix or else a prior associativesociative and comitative-instrumental suffix that came to be used in ablative function in the pronominal system. Besides a SOCIATIVE reading (e.g. ‘with dog, accompanied by dog’) the extended core case form (e.g. *kuɔ́nʔ) also were ABSOLUTIVE ASSOCIATIVE (glossed ASS) forms (e.g. ‘dog & co.’, that is, ‘dog with a companion or in company’). But the oblique stem form only had a comitative-instrumental meaning (e.g. ‘with, through, by dog’). The extended core case form was thus polysemeous and transnumeral. It had two readings: (α) ‘with dog, accompanied by dog’ (glossed SOC); (β) ‘dog with (his) companion or company’, as illustrated by Figure 44. In other words, this polysemy functioned as follows. It was (α) either the given referent, e.g. *χnɛ́r kuɔ́nʔ ‘man with dog’, that was referred to as being accompanied by the given concept [DOG]. Or it was (β) the concept [DOG] itself that was the referent and was referred to as being accompanied by ‘someone’, that is, by an implicit underspecified companion. Such a polysemy obviously was recurrent in PIE, since it ran in parallel with the polysemy of the PIE relational nominal stems; see Figure 28 above. Such a polysemy belongs to the recurrent polysemy of PIE nominals which are both nouns and adjectives and can be used both referentially as a canonical noun form (‘dog & co.) and as an adjective-like modifyer of another referent (‘accompanied by dog’).143 Finally, it is quite plausible to infer that the forms with associative

33

reading, e.g. *kuɔ́nʔ ‘dog & co.’ developed into nom.-acc. dual forms, e.g. *kuónʔ ~ *kuóneʔ ‘two dogs’ from PIE to Vulgar Post-PIE.144 form pɛ́ku-ʔ

gloss sheep-SOC

pɛ́ku-ʔ

sheep-ASS

Figure 44. Polysemy of associative-sociative case forms

9 Classifying suffixes PIE exhibited nominal classifying suffixes or classifiers. These were used to encode the PIE word classes termed classified nouns and classified nominals. The classifiers are equivalent to the traditional nominal stem suffixes. As an implication of my analysis, it is inferential that these classifiers fused with part of the vowel melody from PIE to post-PIE.

9.1 Basic classifiers PIE had a mixed and multifunctional classifier system with a group of (alveloar) suffixes for prototypical reference (class145 1a -t-, 1b -s-, 1c -n-, 1d -ɗ-), three nominal aspect suffixes (classes 2-4), two canonical shape classifiers (classes 5 and 6) and a class 9 (-b̤-) for colors and animals. There were two less canonical classifiers of class 7 (-i-) and class 8 (-u-), which I term “modal” classifiers. The classes are glossed CL1, CL2, etc. The cover symbol -K- is used for any velar and post-velar stop suffix (-k- -g̈- -ɠ- -kw- -g̈w- -ɠw-q- -ɢ̤- -ʛ-). I am not sure whether these suffixes are further separable into different classes. class no. 1

2 3 4 5 6 7 8 9 10

suffix(es) 1a -t1b -s1c -n1d -ɗ1e -d̤-ʔ-χ4a -n4b -r5a -l5b -n-m-i-u-b̤-K-

classificatory meaning prototypical reference, nominalizing (NMZ) pair, part of a pair group, part of a group

shapeless, mass, abstract roundish, pointed, small, diminutive

extended, big, augmentative underspecified derivational suffix, nominalizing (NMZ) function, necessity, quality, property animal (9a), colour (9b) diminutive, singulative, similative

Figure 45. PIE basic classifiers

Prototypical reference, as it is defined here, is reference to a singulative and single concept, e.g. English man referCf. Kölligan 2004: 211f and fn. 42; Haudry 1994: 37. I use the term class for a category marked by a particular classifier. Its is not implied that a class was a canonical noun class of the Bantu type; for a typology cf. Seifart 2010. 144

Cf. Koekhorst 2012. 143 Recall that “nominals” are nominal stems used both referentially and as noun-like adjectives (in modifying function). 142

meaning (α) ‘x as accompanied by domestic animal’ x is the given referent and domestic animal is its companion (β) ‘domestic animal as accompanied by x’ x is a companion and domestic animal is the referent

145

Roland A. Pooth

34

ring to a single concept [MAN]. Modifiers, e.g. strong in English the strong bow imply the conception of a participant ‘x’, e.g. [BOW], modified by a second concept, e.g. [STRONG]. They are used for ‘bi-conceptual reference’ which is reference to two concepts, that is, referent and modifier. Compounds, e.g. English strongbow also have ‘biconceptual reference’ [STRONG] + [BOW] and thus deviate from prototypical reference. Class 1: PIE class 1 suffixes were not allowed to code the modifying member of a compound. Class 1 nouns, therefore, did not occur as the modifying member of a compound. Let me thus suggest that the function of the many (alveolar) class 1 suffixes is typologically comparable to the Classical Nahuatl “absolutive” markers (e.g. -in-, -tl-). Cass 2: This suffix was used to classify a single part of a pair or the pair itself. It was identical to the sociativeassociative case suffix. The class 3 suffix was used to encode groups and suchlike, whereas the class 4 suffix(es) classified a shapeless mass. (Perhaps it had a variant -ɗ-.) Classes 5 and 6: PIE had two canonical shape suffixes: class 5a -l-, 5b -n- for roundish, pointed, small, diminutive and class 6 -m- for extended, long, big, augmentative. Examples of class 6 stems are: (101) *ʔárʔm an. ‘arm’, abl.-gen. *ʔrʔmɛ́s versus *ʔɔ́rʔm ~ *ʔrɔ́ʔm in. ‘oar, paddle’, abl.-gen. *ʔrɛ́ʔmɔs, both derived from an inflected base *ʔarʔ- ~ *ʔ_rʔ- ~*ʔr_ʔ- ‘arm, oar, paddle, row’, perhaps ultimately derived from an underlying root *ʔar-~ *ʔ_r ‘join, fit, ...’ (102) *kwɛ́rm an. ‘worm, classified as long’, abl.-gen. *kwrmɛ́s, derived from the root *kw_r- ‘to carve, cut’, literally ‘the long cut one; insect’ → Vulgar Post-PIE *kwŕ̥mi-, with -iextension

(103) *d̤ɛ́g̈ɔm in. ‘earth’ = ‘big (extended) ground’ > Hitt. tēkan n. ‘earth’, etc.

Class 7: I suggest that the class 7 suffix (-i-) was semantically underspecified and was predominantly used to derive a classified nominal stem from an underyling root noun or root adjective or from another classified nominal stem. Thus, this suffix -i- mainly changed the word class and perhaps had no specific function other than being a kind of nominalizer.

(104) *ʔár- ~ ʔɛ́r- → *ʔár-i- ~ *ʔɛ́r-i- all ‘ram, the hitting one’ (animate) (cf. Greek ἔριφος m. f., cf. NIL, p. 233ff.)

(105) *ʔɛ́r- → *ʔɛ́r-i- both ‘hitting, meeting, coming in contact’ (inanimate) (cf. Vedic r̥cchá-ti, cf. Pooth 2012, 2014: chapter 6; cf. Greek ἔρις f. ‘quarrel’) (106) *ʕákw- → *ʕákw-i- both ‘seeing; seer: eye’ (animate)

(107) *kɛ́r-ɗ- → *kɛ́r-ɗ-i- (thus derived from a class 1d (-ɗ-) stem) both ‘jumper: heart’ (inanimate)

Some class 7 (-i-) nominals were heteroclitics. Their oblique forms showed an allomorph -n- instead of the -i-. This -n- was thus a more general oblique classifier of classes 1, 4, 5, 7 and was easily transferrable to other classes in this general oblique function. Class 8: The class 8 suffix (-u-), on the other hand, was semantically more specified. It was used to derive a classified nominal stem from an underlying root noun or adjective indicating its (natural) function, necessity, its (natural)

quality, or a possession or a (natural or functional) property.

(108) *su_χɗ- → *suáχɗ-u- ‘sweetness, sweet one’

(109) *m_d̤- → *mɛ́d̤-u- ‘sweetness, sweet one’

(110) *ʔ_k- → *ʔɛ́k-u- ‘swiftness, swift one’

(111) *ʔu_s- → *ʔuɛ́s-u- ‘worth, worthy (animate head)’ vs. *ʔuɔ́s-u- ‘worthy thing (inanimate head)’

(112) *ʔ_s- → *ʔɛ́s-u- ‘vigour, lifeliness; good (animate)’ (cf. Vedic ásu- m.) vs. *ʔɔ́s-u ‘good thing (inanimate)’ (cf. Hitt. āššu n. ‘id’)

(113) *(s-)n_χ- ‘to swim’ → *nɔ́χ-u- ‘the thing’s function is to swim: boat’ (114) *d̤_g̈- ‘ground’→ *d̤ɛd̤g̈-ɛ́u- ~*d̤ɛd̤g̈-ú- (a form of the *b̤ɛb̤r-ú- type) ‘groundling, fish’

Ad (114): The meaning ‘groundling’ seems to be the best etymology of Greek ἰχθῦς m. ‘fish’, etc. The ἰ should be a remnant of a PIE reduplication syllable: thus PIE *d̤ɛ-d̤g̈-ú> *dɦəg̑ɦdɦú- (metathesis of the root in syllable onset) > *dəg̑ɦdɦú- (Grassmann of the former reduplication syllable) > *ʔəg̑ɦdɦú(ː)- (with sporadic d > ʔ) (which turned into > *ʔjə-khthúː- > *hi-khthúː- 146 ) > Proto-Greek *ikhthúː-. The meaning of the PIE root *d̤_g̈- was thus ‘ground’, cf. *d̤ɛ́g̈ɔm- ‘big ground: earth’ > Hitt. tēkan n. ‘earth’, etc.

9.2 Derived combinatory stem suffixes In addition to the above given ten basic classes, PIE had secondary combinations of two or three primary (monophonemic) classifiers. These are termed combinatory classifiers or combinatory stem suffixes here. These suffixes were mainly used for derivational purposes, viz. to derive verbal agent or instrument nouns, participles, nouns with a meaning of a nominalized participle, and verbal abstract nouns from underlying verbs. I suggest that the many suffixes, e.g. *-nt- ~ *-ns-, etc. were just allomorphic variants in complementary distribution, encoding either core case stem or oblique stem(s). I see no reason or means for a further functional separation. Class 1a-e served as the derivational base for the combinatory classes 11, 12, 14, 15, 16, 17, 18, and 29. There must have been heteroclitics (e.g. 11ab -tl- ~ -tr-, etc.). The classes 11, 12, and 13 (-nt-, -ns-) were mainly used to derive verbal agent or instrument nouns and participles from the underlying root or lexical base: An example of a class 11 stem is PIE *pɔ́ʕtl in. ‘drinking instrument’, abl.-gen. *páʕtrɔs (→ *póh3tros → VPIE ™póh3tlo- ~ ™póh3tro- > Vedic pā́tram :: Latin pōculum ‘id.’; the latter confirming the suffix -tl-).147 Several instrument nouns of this type should go back to this class. It must have been a functional variant of the class 12 suffixes, cf. *pktɛ́n- an. ‘s.o.’s comber’. An example of a class 12 stem is perhaps PIE *lɔ́uqsl, *lɔuqsɛ́n- in. ‘small roundish illuminating thing’ (with verblike adj. *lɔuqsnáχ-148 > Latin lūna- f. ‘moon’, Avestan adj. raoxšna- ‘gleaming’, Old Prussian lauxnos (pl.) ‘stars’) and Cf. Bozzone 2023: 11. Rather acrostatic, not proterokinetic, pace Pooth 2015. 148 Stefan Höfler p.c. has reminded me that this verb-like adj. may also go back to a later addition of -nó-, -náh2-. This would presuppose PIE *lɔ́uqs- in. ‘id.’, a class 1b stem. 146 147

Proto-Indo-European Nominal Morphology. Part 1. The Noun

PIE *χáksl, *χksɛ́n- an. ‘little pointed (shoulder) joint (movable), axle’ besides *χáks an. ‘shoulder joint, joint, axle’, derived from the PIE root *χ_k- ‘edge, point, top, pointed, sharp’.149 In my view, class 13 was derived from class 4 by addition of the prototypically referential (and thus individualizing) class 1 suffixes to a verbal noun, e.g. *ʔɛ́ɗr- ~*ʔɗɛ́n‘biting, eating (abstractive)’ → *ʔɗɛ́nt- ‘biting, eating one’. The class 13a suffix -nt- could also be used to derive nounlike adjective stems from transnumeral inanimate stems, e.g. *ɗɔ́ru- in. ‘wood(s)’ → class 13a *ɗɔ́runt- ‘the woodone, wood-ish one, wooden’. The same derivation was applicable to the *uɛ́ɗɔr- type, e.g. *uɛ́ɗɔr- in. ‘water(s)’ → *uɛ́ɗɔn-t- in. ‘the water-one, water-ish one’, matching the participles of the *uɛ́kɔnt- type. These noun-like adjective stems, with or without a suffix -i (locative or class 7?), could be used in parallel with their base as causer or cause within the antipassive contstruction:150 (115) a. *ɗɔ́ru-nt(-i) χnɛ́r-m ʔrɔ́ lance-CL13A(-CL7?) man-ALL hit:NDUR:DTR:ITR:3SG literally ‘lance at man hit’ χnɛ́r-m ʔrɔ́ b. *ɗɔ́ru lance.ABS man-ALL hit:NDUR:DTR:ITR:3SG literally ‘lance at man hit’

Classes 14 and 15 were mainly used to derive verbal abstract nouns, e.g. *mɛ́n-ti- ‘thinking’, or nouns with a meaning of a nominalized participle, e.g. *sɔuχnú- ‘s.o.’s son: the one of s.o. that has to be born’. (Note that PIE had a class 15f variant *sɔuχiú- ‘id.’ and thus allomorphic suffixes -iu- ~ -nu-). Classes 16 and 17 were also used to derive verbal abstract nouns, e.g. *kwɔ́i-nχ- ‘payment, punishment’ with verb-like adjective *kwɔi-náχ-, derived from *kwɔ́i-r- (class 4) in. ‘paying’. Class 18, 19, 20: The class 4 shapeless, mass, and abstract(ive) classifying suffix could be attached to the primary classifiers of class 1, 6, and 8 yielding combinatory suffixes which were also used to derive verbal abstract nouns, e.g. *mɛ́ːχur- in. ‘time’.151 A well-known example is *sɛ́ʔmn- ‘sowing, seed’, etc. These original verbal abstract nouns were also used with a more concrete, individualized reading, e.g. *sɛ́ʔmn- ‘seed’ (a comparandum is German die Bedienung f. ‘waiter, waitress’). Classes 7 and 8 served as the base for the combinatory classes 21, 22, 23, 24, 25 and 15f. Classes 21 and 22 are treated in section 9.3 below. Classes 23 and 24 mainly consisted of pertensive-possessive nominal stems, e.g. *ulkwíʔ- ‘belonging to a pair of dangerous ones; s.o.’s belonging as part of a pair to the dangerous one’ 152 or *ulkwíχ- ‘belonging to a group of dangerous ones, s.o.’s belonging to a group of dangerous ones’. (Nominal stems of these classes are continued by stems of the Vedic vr̥kī ́ḥ Cf. NIL, pp. 259ff., 287ff. Such parallel stems with and without -nt-, further extended by -s or combined as -i-s constitute the origin of the Hittite ergative sg. n. suffix -anza, -anzi=ya (Hitt. pl. -antes, Luwian sg. -antiš, pl. -antinzi). 151 Cf. Widmer 2004. 152 Recall that PIE relational nominals were underspecified for reference vs. attribution. They had both a referential relational-possessed noun reading ‘someone’s [CONCEPT]’ and a pertensive-possessive adjectivelike attributive/modifying reading ‘belonging (as a possesee) to [CONCEPT]’ or ‘having the property [CONCEPT]’. 149 150

35

type). For class 25 cf. *tɛnúχ- ‘s.o.’s body, classified as a collective of body parts’, internally derived from its nonrelational base *tɛ́ːnuχ- ‘collection of body parts’. PIE even had more derivational suffixes. An example of a class 26 nominal stem is perhaps *χiu-Hɛ́n- ‘life-of-vigourperson, young male’. But this may have been a compound with a root noun *-χ_n- ‘face, person’ as its second member. I am not sure. Class 27 was mainly used to derive participles of the type *ɠusmʔnɔ́ ~ *ɠusmnɔ́ ‘having taken pleasure in, having been pleased by s.th.’. Besides this nondurative stem there was a durative-interminative stem, continued as OArm. marmin ‘body, corpse’ < PIE *mr-m(ʔ)ɛ́nɔ ‘mortal, having to die’ :: PIE *mr-mʔnɔ́ ‘having died, dead’. The shape of the suffix has been reconstructed as *-mʔnɔ́-, but *-mnɔ́also existed (presumably as its allomorph).

class no. 11

12

13 14 15a-e 15f 16a-e 16fg 17 18

19 20 21 22 23 24 25 26 27 28 29

combined suffixes 1+5/4?

class no. 11a 11b 11c 11d 1+5 12a 12b 12c 12d 4+1 13a 13b 1+7 14a 14b 14c 1+8 15a 15b 15c 7+8 15f 1+3 16a 16b 16c 4+3 16f 16g 1+7+3 17 1+4 18a 18b 18c 18d 6+4 19a 19b 8+4 20a 20b 7+1 21a 21b 21c 8+1 22a 22b 22c 7+2 23 7+3 24 8+3 25 26 6+2 27a 6+2+1 27b 6+1 27c 8+4/1+1 28a 28b 1+6 29a 1+6+1 29b

class further suffix(es) no. variants -tl11e -ɗl-tr11f -ɗr-sl11g -d̤l -sr11h -d̤r-tl12e -ɗl-tn12f -ɗn-sl12g -d̤l-sn12h -d̤n-nt13c -nɗ-ns13d -nd̤-ti14d -ɗi-si14e -d̤i-ni-tu15d -ɗu-su15e -d̤u-nu-iu-tχ16d -ɗχ-sχ16e -d̤χ-nχ-rχ-nχ-d̤iχ- (used as infinitive) -tr18e -ɗr-tn18f -ɗn-sr18g -d̤r-sn18h -d̤n-mr-mn-ur-un-it 21d -iɗ-is21e -id̤-in -ut22d -uɗ -us22e -ud̤-un -iʔ-iχ-uχ-Hn- (a root noun -χan-?) -mʔn-mʔnt-mn-unt-mnt- (-mnt- after -u-) -tm-tmt-

Figure 46. PIE derived stem suffixes

Roland A. Pooth

36

Class 28 served to derive associative-possessive adjectives from any underlying nominal stem by suffixation of the suffix -uɛ́nt-/untɛ́_ to the oblique stem and subtraction of one vowel of the underlying vowel melody, e.g. PIE *pku-mɛ́nt- ‘being associated by domestic animals, having a lot of domestic animals’, derived from *pɛ́ku- minus its vowel. The derivational base was thus treated as if it was a first part of a compound. Likewise, the class 29 suffix -tmɔ́- was used to derive elative-superlative verb-like adjectives from any underlying nominal stem by suffixation of -tmɔ́- to the oblique stem and subtraction of one vowel of the underlying vowel melody, e.g. *χnrtmɔ́- ~ *χnrtmtɔ́- ‘best-man-ish’, derived from *χnɛ́r- ‘man’ minus its vowel. The derivational base was thus also treated as if it was a first part of a compound. The adjectives willl be treated in more detail in the sequel (part two) of this article.

9.3 More on classified nominals Classified nouns were often derived from an underlying root noun or root adjective (e.g. *mɛ́l-i-t-, cf. examples (118abcd)). But they could also be derived directly from the discontinuous root or inflectable base or from the discontinuous consonant frame of another classified nominal stem by suffixation on the consonant frame tier. An example is: (116) *b̤_r- → *b̤_r-u- → *b̤r-u-ʔ- → relational nominal stem *b̤r-ú-ʔ- in. ‘eye-brow(s), classified as a pair’ (maybe there was no underlying base (**b̤ɛ́r-u-) other than the skeletal consonant frame in such cases) (117) *s_uχ- → *s_uχ-n_u- → relational nominal stem *sɔuχnúan. ‘s.o. son’ (was there an underlying nonrelational base **sɔ́uχnu-?)

PIE nonrelational classified stems of class 1, 2, 3, 4, 5, 6 and class 9a animal nouns (e.g. *ʔɛ́r-b̤- an. ‘ram’) differed from PIE root nouns in being more canonical noun stems, whereas the classified nominals of class 7 and 8 and class 9b colour adjectives (e.g. *ʔál-b̤- ‘white’) were more underspecified and served both referentially as nouns and modifying adjectives. Therefore, PIE classified nouns of class 1, 2, 3, 4, 5, 6, 9a, 10 were not as underspecified as root nominals. To yield a classified modifying adjective they had to be turned into classified relational nominals by vowel transposition including accent shift, as outlined in detail above. The class of relational nominal stems (e.g. *ɗʕtɛ́r- ‘s.o.’s giver, giver of sth.’, etc.), as a subordinate, thus belonged to the class of classified nominal stems. Thus, PIE had a more flexible word class of classified nominals. The class 7 suffix (-i-) and the class 8 suffix (-u-) were used to code classified nominals exhibiting both a nouny referential function (e.g. like English sweetness) and a modifying function (like adjectives, e.g. sweet). PIE classified nominals in modifying function can thus be interpreted as NOUN-LIKE ADJECTIVES. They were inflected like classified nouns and showed agreement for case and number with the given head noun. In my view, further agreement for class was optional. Differently from root adjec-

tives, classified nominals did not obligatorily show voice marking triggered by gender; see the following examples. (118) a. *mɛ́d̤u mɛ́li honey.ABS.SG softness/soft.CL7.ABS ‘portion of honey, softness-ish one’

b. *mɛ́d̤u mɛ́li-t honey.ABS softness/soft.CL7-CL4A\ABS ‘(portion of honey) honey, the soft one’

c. *uláχ-r mɛ́li(-n) wool-CL4A softness/soft.CL7(-CL4B) ‘wool, softness/soft (mass)’

d. *mɛ́li-t softness/soft.CL7-CL1A ‘the softness/soft-one’ (in some variants this became the new word for ‘honey’)

Ad (118): The PIE word for ‘wool’ should have been *uláχr. It was classified as a mass or substance by the class 4ab suffix -r- ~ -n- and was derived from the PIE inflectable base *ul_χ- ‘to make (us) covered’, derived from the root *u_l- ‘to cover’. PIE also had a verb-like adj. *ulχnɔ́-, *ulχnáχ-. This adjective *ulχnɔ́- was nominalized to a Vulgar Post-PIE noun by nominalizing accent shift to the word initial syllable. Lithuanian vìlna f. ‘wool’ :: Vedic ū́rṇā- ‘wool’ both point to *u̯ĺ̥Hnā- f. Doric Greek λᾶνος- n. ‘wool’ can continue Proto-Greek *u̯lā́nos- (™u̯láh2no-) or *u̯lH ̥ nós- (™u̯lHnó-). Hittite ḫulana- cannot be equated to the Doric Greek root λᾶν-. The latter cannot go back to ? h2u̯ĺ̥h1neh2-, because this should turn out as Greek †αϝέλενᾱ, but not as λᾶνος-. A derivation of Hittite ḫulana- from ? h2u̯ĺ̥h1neh2- causes further problems, because it should have yielded †ḫull(a)na, but not ḫulana-. Can the Hittite form go back to *u̯ĺ̥χnā-? After all, it is perhaps better to take Hittite ḫulana- as a loan, cf. Kloekhorst 2008: 358. Note that the alleged ?h2u̯ĺ̥h1neh2- has no etymology, whereas the etymology proposed here is highly plausible. Last but not least, some classified nominals behaved more like noun-like adjectives and thus probably had more obligatory voice marking, e.g. *ʔɛ́s-u- ‘(animate) goodness, good (animate)’ vs. *ʔɔ́s-u- ‘(inanimate) goodness, good thing, good (inanimate)’. (119) a. *χnɛ́r ʔsú man.ABS.SG someone’s.goodness/good.CL7.RL.ABS ‘a good man’

b. *ʔɛ́ːku ʔɔːkú s.o.’s.swiftness/swift.PLT.CL7.RL.ABS horse.PLT.ABS ‘a swift group of horses’

However, such issues will be treated in more detail in the sequel and part two of this article.

Acknowledgements I particularly wish to thank Robert Tegethoff and all those who participated in the academia session on the aforegoing version of this article. In memory of Xavier Tremblay and Hans-Jürgen Sasse.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

37

References 1. 2. 3. 4. 5. 6. 7.

8. 9.

10. 11. 12. 13.

14. 15. 16. 17. 18. 29. 20. 21. 22. 23. 24. 25. 26. 27.

28.

Adams, D.Q. 2013: A Dictionary of Tocharian B. Revised and Greatly Enlarged. 2 Vol. Leiden & New York. Andrews, J.R. 1975: Introduction to Classical Nahuatl. Austin. Andrews, J.R. 2003: Introduction to Classical Nahuatl. Revised edition. Norman. Bauer, L. 2004: A Glossary of Morphology. Edinburgh. Beekes, R.S.P. 1995: Comparative Indo-European Linguistics. Amsterdam & Philadelphia. Beekes, R.S.P. 2011: Comparative Indo-European Linguistics. Second edition. Revised and corrected by Michiel de Vaan. Amsterdam & Philadelphia. Behrens, L. 2002: “Structuring of word meanings II: Aspects of Polysemy”, Lexicology: An international handbook on the nature and structure of words and vocabularies. Handbücher zur Sprachund Kommunikationswissenschaft (HSK) 21.1. Ed. by A. D. Cruse, F. Hundsnurscher, M. Job, P. R. Lutzeier. Berlin & New York, 319-337. Behrens, L. & H.-J. Sasse 1997: Lexical Typology: A Programmatic Sketch. Arbeitspapier Nr. 30 (Neue Folge). Institut für Sprachwissenschaft. Universität zu Köln. Benjamin, G. 2011: “Deponent Verbs and Middle-Voice Nouns in Temiar”, Austroasiatic Studies: papers from ICAAL4. Mon-Khmer Studies Journal Special no. 2. Ed. by S. Srichampa & P. Sidwell. Dallas, Salaya & Canberra, 11-37. Bennardo, G. 2000: “Possessive Markers in Tongan: A Conceptual Approach”, Sprachtypologie und Universalienforschung 53 (3/4), 269-280. Besnier, N. 1987: “An autosegmental approach to metathesis in Rotuman”, Lingua 73, 201-223. Blake, B.J. 2001: Case. Second edition. Cambridge. Bozzone, C. 2013: “Initial “Yod” in Greek and the Etymology of Gk. ἵππος ‘horse’”, Proceedings of the 24th Annual UCLA IndoEuropean Conference. Ed. by S. W. Jamison, H. C. Melchert & B. Vine. Bremen, 1-26. Brugmann, K. 1916: Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. 2nd ed. Strassburg. Byrd, A.M. 2010: Reconstructing Proto-Indo-European Syllabification. Ph.D. dissertation, University of California, Los Angeles. Byrd, A.M. 2015: Reconstructing Proto-Indo-European Syllabification. Leiden & Boston. Clackson, J. 2007: Indo-European Linguistics: an Introduction. Cambridge. Cooper, A. I. 2015: Reconciling Indo-European Syllabification. Leiden & Boston. Darms, G. 1978: Schwäher und Schwager, Hahn und Huhn. Die Vr̥ddhi-Ableitungen im Germanischen. München. Dixon, R.M.W. 1979: Ergativity. Cambridge. Dixon R.M.W. 2010, 2012: Basic Linguistic Theory. Vol. I & II 2010, Vol. III 2012. Oxford. Donohue, M. & S. Wichmann 2008 (eds.): The Typology of Semantic Alignment. Oxford. Dryer, M.S. 1986: “Primary Objects, Secondary Objects, and Antidative”, Language 62 (4), 808-845. Edwards, O. 2014 (presentation): “The structure of metathesis in Amarasi”, talk given on 2014-12-18 at the University of Cologne, Institut für Linguistik, Allgemeine Sprachwissenschaft. EIC = Mallory, J.P. & D.Q. Adams (eds.) 1997: Encyclopedia of Indo-European Culture. London & Chicago. Eichner, H. 1973: “Die Etymologie von heth. mehur”, Münchener Studien zur Sprachwissenschaft 31, 53-107. Eichner, H. 1985: “Das Problem des Ansatzes eines urindogermanischen Numerus ‘Kollektiv’ (‘Komprehensiv’)”, in: Grammatische Kategorien, Funktion und Geschichte. Akten der VII. Fachtagung der Indogermanischen Gesellschaft, Berlin 20.-25. Februar 1983. Ed. by B. Schlerath & V. Rittner. Wiesbaden, 134-169. Fellner, H. & L. Grestenberger 2015 (handout): “Gemeinsame Unterschiede und unterschiedliche Gemeinsamkeiten der hethitischen und tocharischen Nominalmorphologie: Die Reflexe der *-nt- und *-mh1no-Partizipien”, talk given at 100 Jahre Entzifferung des Hethitischen - Morphosyntaktische Kategorien in Sprachgeschichte und Forschung. Arbeitstagung der Indogermanischen Gesellschaft, Philipps-Universität Marburg, 21. bis 23. September 2015.

29. 30. 31. 32. 33.

34. 35. 35. 37. 38. 39. 40.

41. 42. 43. 44. 45. 46. 47. 48.

49. 50. 51. 52. 53. 54.

55. 56. 57.

Fortson, B.W. IV 2009: Indo-European Language and Culture: An Introduction. 2nd edition. Malden, Oxford & Carlton. Fox, A. 1995: Linguistic Reconstruction. An Introduction to Theory and Method. Oxford. Frotscher, M. 2013: Das hethitische -ant-Partizip und seine indogermanischen Grundlagen: Semantik, Morphologie, Syntax. Doctoral dissertation. Università degli studi di Verona. Hall, T.A. 2011. Phonologie. Eine Einführung. 2., überabeitete Auflage. Berlin & New York. Haspelmath, M. 2014-2015: Ditransitive Constructions. Annual Review of Linguistics 2015 (1), 19-41. Review in Advance, September 17, 2014, doi: 10.1146/annurev-linguist-030514-125204, www.linguistics.annualreview.org Haspelmath, M. & A. D. Slims 2010: Understanding Morphology. Second Edition. London. Haudry, J. 1994: L’Indo-Européen. Paris. Hayes, B. 1995. Metrical Stress Theory. Chicago. Heine, B. & T. Kuteva 2002: World Lexicon of Grammaticalization. Cambridge. Hoffmann, K. 1970: “Das Kategoriensystem des indogermanischen Verbums”, Münchener Studien zur Sprachwissenschaft 28, 19-41. Hoffner, H. A. Jr. & H. C. Melchert 2008: A Grammar of the Hittite Language. Winona Lake. Jasanoff, J.H. 2003: Hittite and the Indo-European verb. Oxford. Jasanoff, J.H. 2015 (handout): “Hittite and the IE verb 100 years after Hrozný; or, What happened to the perfect in Hittite?”, talk given on 22 September, 2015 at 100 Jahre Entzifferung des Hethitischen - Morphosyntaktische Kategorien in Sprachgeschichte und Forschung. Arbeitstagung der Indogermanischen Gesellschaft, Philipps-Universität Marburg, 21. bis 23. September 2015. Kobayashi, M. 2004. Historical Phonology of Old Indo-Aryan Consonants. Tokyo. Keydana, G. 2005: “Indogermanische Akzenttypen und die Grenzen der Rekonstruktion”, Historische Sprachwissenschaft 118, 1947. Keydana, G. 2013: “Proterokinetische Stämme, Akzent und Ablaut”, in: Keydana, Widmer & Olander 2013: 31-62. Keydana, G., P. Widmer & T. Olander (eds.) 2013: Indo-European Accent and Ablaut. Copenhagen. Klimov, G.A. 1974: “On the character of language of active typology”, Linguistics 131, 11–25. Kloekhorst, A. 2008: Etymological Dictionary of the Hittite Inherited Lexicon. Leiden. Kloekhorst, A. 2012: Hittite ḫūmantet and the PIE instrumenal ending. Power point presentation, pdf version 2012-09-06, https://www.leidenuniv.academia.edu/AlwinKloekhorst Kroonen, G. 2013: Etymological Dictionary of Proto-Germanic. Leiden & Boston. Kümmel, M.J. 2012: “Typology and reconstruction. The consonants and vowels of Proto-Indo-European”, The Sound of IndoEuropean. Phonetics, Phonemics, and Morphophonemics. Ed. by B. Nielsen Whitehead, T. Olander, B. A. Olsen & J. E. Rasmussen. Kopenhagen, 291-329. Lehmann, W.P. 1974: Proto-Indo-European Syntax. Austin. Lehmann, W.P. 1995: Residues of pre-Indo-European Active Structure and their Implications for the Relationships among the Dialects. Innsbruck. Lehmann, W.P. 2002: Pre-Indo-European. Washington. Lehrman, A. 1987: “Anatolian Cognates of the Proto-IndoEuropean word for ‘Wolf’, Die Sprache 33, 13-18. Litscher, R. 2014: “Voraussetzungen für ein feminines Genus und Implikationen für das Kategoriesystem des frühindogermanischen Nomens”, in Neri & Schuhmann 2014: 137-165. LIV = Kümmel, M.J. & H. Rix et. al. 2001: Lexikon der indogermanischen Verben. Unter der Leitung von H. Rix ..., 2. erw. und verbesserte Auflage bearb. von M. Kümmel und H. Rix. Wiesbaden. Mayrhofer, M. 1986: Indogermanische Grammatik I/2: Lautlehre. Heidelberg. McCarthy, J. 1979: Formal Problems in Semitic Phonology and Morphology. Doctoral dissertation, MIT. Cambridge, MA. McCarthy 1981: “A Prosodic Theory of Nonconcatenative Morphology”, Linguistic Inquiry 12, 373-418.

Roland A. Pooth

38

58. 59. 60.

61. 62. 63.

64. 65. 66.

67. 68.

69. 70.

71.

72.

73.

74.

75. 76. 77. 78. 79. 80. 81.

Meillet, A. 1931: “Caractère secondaire du type thématique indoeuropéen”, Bulletin de la Société de Linguistique de Paris 32 (2) (numéro 96), 194-203. Meillet, A. 1937: Introduction à l’étude comparative des langues indo-européennes. Paris. Neri, S. 2003: I sostantivi in -u del gotico: Morfologia e preistoria. Innsbruck. Neri, S. 2005: “Riflessioni sull’apofonia radicale di protogermanico *namōn ‘nome’”, Historische Sprachforschung 118, 201-50. Neri, S. & R. Schuhmann 2014 (eds.): Studies on the Collective and Feminine in Indo-European from a Diachronic and Typological Perspective. Leiden & Boston. NIL = Wodtko, D. S., B. Irslinger & C. Schneider 2008: Nomina im indogermanischen Lexikon. Heidelberg. Pinault, G.-J. 2003: “Sur les thèmes indo-européens en *-u- : dérivation et étymologie”, Indogermanisches Nomen. Akten der Arbeitstagung der Indogermanischen 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. Pooth, R.A. 2001: “Studien zur frühurindogermanischen Morphologie I. ‘Stativ’, ‘Medium’ und ‘Perfekt’”, Historische Sprachforschung 114, 220-258. Pooth, R.A. 2004a: “Ablaut und autosegmentale Morphologie: Theorie der urindogermanischen Wurzelflexion”, Indogermanistik - Germanistik - Linguistik. Ed by. M. Kozianka, R. Lühr & S. Zeilfelder. Hamburg, 401-471. Pooth, R.A. 2004b: “Zur Genese der späturidg. thematischen Konjugation aus frühuridg. Medialformen”, Indogermanische Forschungen 109, 31-60. Pooth, R.A. 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. Pooth, R.A. 2009b: “Proto-Indo-European Ablaut and Root Inflection”, Internal Reconstruction in Indo-European. Ed. by J. E. Rasmussen & T. Olander. Copenhagen, 229-254. Pooth, R.A. 2011: “Die 2. und 3. Person Dual und das Medium”, Indogermanistik und Linguistik im Dialog. Akten der XIII. Fachtagung der Indogermanischen Gesellschaft ... in Salzburg. Ed. by T. Krisch & T. Lindner. Wiesbaden, 473-83. Pooth, R.A. 2012: “Zum Aufkommen transitiver Verben in frühen Vedischen am Beispiel 1r̥”, 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 Melchert. Wiesbaden, 267-84. Pooth, R.A. 2014a (ms.): “Ein Problem der Methode der komparativen Rekonstruktion von Morphemen, Morphemgrenzen und morphosyntaktischen Kategorien”, manuscript version 2014-0807, https://leidenuniv.academia.edu/RolandPooth Pooth, R.A. 2014b: Die Diathesen Aktiv vs. Medium und die Verbsemantik im Vedischen der R̥gveda-Saṃhitā. Proefschrift. Defended 2014-10-23. Universiteit Leiden, Leiden University Repository, https://openaccess.leidenuniv.nl. Pooth, R.A. 2014c (manuscript): “Voice, Transitivity Direction, Case, and Alignment in Proto-Indo-European. IV. The Proto-IndoEuropean Case System”, mansucript version 2014-11-17, https://leidenuniv.academia.edu/RolandPooth Pooth, R.A. 2014d (ms.): “More evidence for Proto-Indo-European transfixes: Two types of “lengthened grades””, ms. version 201412-05, https://leidenuniv.academia.edu/RolandPooth Pooth, R.A. 2015 (manuscript): “A typological overview of ProtoIndo-European”, manuscript version 2015-07-23 https://leidenuniv.academia.edu/RolandPooth Rijkhoff, J. 2002: The Noun Phrase. Oxford. Ringe, D. 2006: From Proto-Indo-European to Proto-Germanic. Oxford. Rix, H. 1965: “Lat. iecur, iocineris”, Münchener Studien zur Sprachwissenschaft 18, 79-92. Schaffner, S. 2001. Das Vernersche Gesetz und der innerparadigmatische grammatische Wechsel des Urgermanischen im Nominalbereich. Innsbruck. Schaffner, S. 2003: “Der Beitrag des Germanischen zur Rekonstruktion der urindogermanischen Akzent- und Ablautklassen”, in: Tichy, Wodtko & Irslinger 2003: 203-218.

82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93.

94. 95.

96. 97. 98. 99. 100. 101.

102. 103. 104. 105.

106. 107. 108. 109.

110. 111.

Schindler, J. 1967a: “Zu hethitisch nekuz”, Zeitschrift für Vergleichende Sprachfoschung 81, 290-303. Schindler, J. 1967b: “Das idg. Wort für “Erde” und die dentalen Spiranten”, Die Sprache 13, 191-205. Schindler, J. 1967c: “Tocharische Miszellen”, Indogermanische Forschungen 72, 239-249. Schindler, J. 1969: “Vogel und Ei im Indogermanischen”, Die Sprache 15, 144-167. Schindler, J. 1972: “L’apophonie des noms racines indoeuropéens”, Bulletin de la Societé de linguistique de Paris 67, 3138. Schindler, J. 1975a: “L’apophonie des thèmes indo-européens en r/n”, Bulletin de la Societé de linguistique de Paris 70, 1-10. Schindler, J. 1975b: “Armenisch erkn, griechisch ὀδύνη, irisch idu”, Zeitschrift für Vergleichende Sprachforschung 89, 53-65. Schindler, J. 1977: “A thorny problem”, Die Sprache 23, 24-35. Seifart, F. 2010: “Nominal classification”, Language & Linguistics Compass 4, 719-736. Seiler, H. 1983: Possession as an operational dimension of language. Tübingen. Stassen, L. 1997: Intransitive Predication. Oxford. Steer, T. 2015: Amphikinese und Amphigenese. Morphologische und phonologische Untersuchungen zur Genese amphikinetischer Sekundärbildungen und zur internen Derivation im Indogermanischen. Wiesbaden. Tichy, E. 1995: Die Nomina agentis auf -tar- im Vedischen. Heidelberg. Tichy, E., D.S. Wodtko & B. Irslinger. 2001. Indogermanisches Nomen. Derivation, Flexion und Ablaut. Akten der Arbeitstagung der Indogermanischen Gesellschaft, Freiburg, 19. bis 22. Sept. 2001. Bremen. Thompson, C.L. 1996: “The Na-Dene middle voice: An impersonal source of the D-Element”, International Journal of American Linguistics 62 (4), 351-378. Thompson, L.C. & M.T. Thompson 1969: “Metathesis as a grammatical device”, International Journal of American Linguistics 35, 213-219. Tremblay, X. 1999: Études sur les noms suffixaux athématique de l’Avesta. Thèse doctorat. É.P.H.É. IVe section. Paris. Tremblay, X. 2003: “Interne Derivation: „Illusion de la reconstruction“ oder verbreitetes morphologisches Mittel? Am Beispiel des Awestischen”, in Tichy, Wodtko & Irslinger 2003: 231-259. Tremblay, X. 2010: “Jenseits von Schindler? Die Bedeutungen der drei Wurzelnomina-Ablauttypen”, MSS 64, 2004 [2010], 181221. Tremblay, X. 2013: “Zwei? Vier? Sechs? Zur Anzahl der Nominalablauttypen und ihrem Grundwesen (Apophonica VII)”, IndoEuropean Accent and Ablaut. Ed. by G. Keydana, P. Widmer & T. Olander. Copenhagen. de Vaan, M. 2008: Etymological Dictionary of Latin and the Other Italic Languages. Leiden & Boston. Viti, C. 2015: “Reconstructing Proto-Indo-European Accent Paradigms”, Journal of Indo-European Studies 43 (1/2), 100-139. Weiss, M. 2009: Outline of the Historical and Comparative Grammar of Latin. Ann Arbor & New York. Weiss, M. 2012: “Intersting i-stems in Irish”, Multi Nominis Grammaticus. Studies in Classical and Indo-European linguistics in honor of Alan J. Nussbaum on the occasion of his sixty-fifth birthday. Ed. by A.I. Cooper, J. Rau & M. Weiss. Ann Arbor & New York, 340-56. Widmer, P. 2004: Das Korn des weiten Feldes. Interne Derivation, Derivationskette und Flexionsklassenhierarchie. Aspekte der nominalen Wortbildung im Urindogermanischen. Innsbruck. Widmer, P. 2013: “Akzent und Ablaut, externe und interne Derivation in der Nominalkomposition”, in: Keydana, Widmer & Olander 2013: 187-195. van Wijk, N. 1902: Der nominale Genitiv Singular im Indogermanischen in seinem Verhältnis zum Nominativ. Zwolle. Woodhouse, R. 2015: “Some observations on the putative dual reflexes of PIE *CRHC in Greek and Armenian, Francis’ law and Greek αὐχήν”, Studia Etymologica Cracoviensia 20, 257-72, doi: 10.4467/20843836SE.15.017.2803, www.ejournals.eu/SEC Zair, N.A. 2006: “Dybo’s law: Evidence from Old Irish”, Oxford University Working Paper in Linguistics, Philology, and Phonetics 7, 215-26. Zair, N. 2011: “PIE ‘bird’ and ‘egg’ after Schindler”, Münchner Studien zur Sprachwissenschaft 65, 287-310.

Proto-Indo-European Nominal Morphology. Part 1. The Noun

39

Comments and reviews

Author’s response

Everybody is invited to send their comments and reviews to the editor’s e-mail address. They will be attached to the latest PDF version.

The author is invited to send a response to the editor’s e-mail address. It will also be attached to the latest PDF version.

Comments can be formal or informal.

The response can be formal or informal.

Example comment 1 “A monumental achievement. Virtually without precedent in the field of Indo-European linguistics. Destined to be a postmodern classic.”

;-)

Quang Phúc Đông, South Hanoi Institute of Technology

Example comment 2 “I found this paper an extremely muddled paper with a large number of deficits.” Reviewer A

Example comment 3 “It is written in a clear style and would be understood by any reader.” Reviewer B

[Comments 2 and 3 are quoted from p. 180 of Smith, R. 2006: “Peer review: a flawed process at the heart of science and journals”, Journal of the Royal Society of Medicine 99, 178182.]

Language Arts 1 | 2015 | VERSION 2015-12-23

Author manuscript version 2015-12-23