Derivational Properties of "Adjectival Roots"

Appeared in Zurück zur Wurzel Struktur, Funktion und Semantik der Wurzel im Indogermanischen Akten der 15. Fachtagung der Indogermanischen Gesellschaft vom 13. bis 16. September 2016 in Wien Herausgegeben von Melanie Malzahn, Hannes A. Fellner und Theresa-Susanna Illés Pages 205-224 Wiesbaden 2022 Reichert Verlag Derivational Properties of “Adjectival Roots” (Expanded Handout) Alan J. Nussbaum Cornell University 1. A point to make first, of course, in discussing the meanings of roots, which is obviously a matter inherent in the question of “adjectival roots,” is that roots are not available for examination as free forms. Their meanings are divined from the lexical family — large or small, verb-heavy or nominal-heavy — to which they give rise. Root semantics are thus constructs and products of analysis, not data to be collected. 1.1 The vast majority of roots are analyzed as having a verbal meaning. These include a large number of especially familiar examples like: *deh3 ‘give’, *g̑heu̯ ‘pour’, *h1em ‘take’, *i̯ eug ‘join’, *k̑leu̯ ‘hear’, *leikw ‘leave’, *men ‘think’ etc. etc. 1.2 It is sometimes possible and credible, however, to attribute a noun-like meaning to a given root. Reasonable candidates for membership in such a class of roots are of two types: 1.2.1 On the one hand, there are root nouns denoting a substantive, the root of which is not found in verbs: *dʰu̯or-/dʰur- ‘door’, *k̑ērd-/*k̑r̥d- ‘heart’, *gwou̯-/*gweu̯- ‘cow’, *sal- ‘salt’ etc. 1.2.2 Another set of cases in which it is reasonable to see a root with the meaning of a noun are those in which a plausibly primary R(oot) + S(uffix) stem denotes a substantive, but where the root once again fails to surface in any verbal forms: *dh(e)g̑h-(o)m- ‘earth’, *(hx)i̯ ē̆kw-r̥/n- ‘liver’ etc. 1.2.3 Neither of these nominal root types is absolutely secure, of course. For it can never be excluded that such forms are ultimately derivatives in one suffix (whether zero or something overt) of roots that are obsolete as bases of verbs or simply fail to have been preserved as such in the evidence that happens to have come down to us. 1.3 If both verbal and substantival roots are sure or possible, the further question — and one which virtually raises itself — is that of the necessity or feasibility of recognizing a class of roots with the meanings of a third type of “content word” — namely adjectives. 2. The question of the advisability of setting up such a category immediately becomes that of determining whether there are one or more objective criteria for analyzing a root as “adjectival” in meaning. 2.1 As something of an aside, it may be useful, though for less immediate purposes, to point out that one thing that is pretty clearly not such a criterion is whether a root’s derivatives form a “Caland system” (CS). 2.1.1 It is, to be sure, true that such systems do often center on an adnominal, and more particularly on a “property concept” (PC) adjective.1 These have been said to be of seven types:2 DIMENSION: large, small, long, short etc. hard, soft, heavy, light, rough, smooth, hot, cold, sweet, sour etc. PHYSICAL PROPERTY: 1 2 Rau 2009: 78-109 with references to Balles 2003, Balles 2006, Balles 2009. Dixon 1982 COLOR: red, green, blue etc. clever, stupid, happy, sad, kind, cruel, proud, humble, envious, wicked etc. AGE: new, young, old VALUE: good (and synonyms), bad (and synonyms), proper, perfect, pure etc. SPEED: fast, quick, slow etc. HUMAN PROPENSITY: 2.1.2 From the central adjectival are derived (a) filling out a semantic derivational “matrix” and (b) via various morphologies (characteristically featuring the synchronically “substitutive”):3 Alternative stem(s) for the positive adjective: -o- / -ro-: *h1roudh-o- (Li. raũdas etc.) / *h1rudh-ró- (L. ruber etc.) ‘red’ -o-/ -i-: *h2erg̑-ó- (ἀργός) / *h2o/erg̑-i- (Hitt. ḫarki-) ‘bright, flashing’ -ro- / -mo-: *kruh2-ro- (Ved. krūrá-) / *kruh2-mo- (YAv. xrūma-) ‘gory’ -ro-/ -i-: *h2r̥g̑-ro- (Ved. ṛjrá-) / *h2(o/e)rg̑-i- (Hitt. ḫarki-) ‘bright’ -ro- / -u-: *tep-lo- (ORuss. teplŭ) / *tep-u- (Ved. tápu-) ‘hot’ -ro- / -o/ent-: *bhr̥g̑h-ro- (TA pärkär, TB pärkare ‘long’) / *bhr̥g̑h-ont- (Ved. br̥hánt- ‘high’) -u- / -o/ent-: *bhr̥g̑h-u- (Hitt. parku-) / *bhr̥g̑h-ont- (Av. bǝrǝzaṇt-) ‘high’ etc. Comparative/ superlative adjective -ro- / -i̯ os-: YAv. tiγra- ‘sharp’ / Ved. téjīyas-, téjiṣṭha-u- / -i̯ os-: Gk. κρατύς ‘powerful’ / κρέσσων, κράτιστος etc. Adjectival abstract (simplex) -o-, -ro- / root noun: *h1roudh-o-, *h1rudh-ró- / *h1rudh- (OIr. rú, rod ‘reddening’) -o-, -mo- / s-stem: *krouh2-o- (Gmc. *hrawa- ‘raw’), *kruh2-mo- / *kreuh2-s- (Gk. κρέας ‘flesh’) Adjectival abstract (1st/ 2nd member of bahuvrihi) -ro-, -o/ent- / -i-: *bhr̥g̑h-ro-, *bhr̥g̑h-ont- / *bhr̥g̑h-i- (YAv. bǝrǝzi-caxra- ‘with high wheels’) -u- / -es- : *dʰr̥s-ú- ‘bold’ (Gk. θρασύς) / *-dʰr̥s-es- (Gk. εὐθαρσής ‘of good courage’) Essive/fientive verb form(s) *h1rudh-ró- ‘red’ (L. ruber, Gk. ἐρυθρός) : e.g. *h1rudh-eh1- (L. rubē[sce]re ‘be[come] red’ etc.) Factitive verb form(s) *leuk-es- ‘light, brightness’ (YAv. raocah-) : *louk-ei̯ e/o-4 ‘illumine’ (Hitt. lukke/azzi ‘set alight’, L. lūcēre ‘shine’ [partly trans.], Ved. rocáyati ‘makes shine’)5 *hxish1-ró- ‘powerful’ (Ved. iṣirá-, Gk. ἱερός) : *hxis-n(e)-h1- (Ved. iṣṇā́ ti ‘speed on’)6 2.1.3 But it is not true that a CS guarantees an adjectival root and vice versa. 2.1.3.1 First, there are plausible adjectival roots, in that they yield PC adjectives and nothing else, which make no CS to speak of: *bhes : *bhos-ó- (Gmc. *baza- > OIc. berr, OE bær ‘bare, naked’, Lith. bãsas [4], OCS bosъ ‘barefoot’). *g̑hreh1 : *g̑hr̥h1-u̯o- (L. rāvus ‘gray) and *g̑hreh1-u̯o- (Gmc. *grēwa- > OIc. grár, OHG grāo etc. ‘gray’). 2.1.3.2 More expectedly, a verbal root may be the basis of a PC adjective to which no real CS is found: Rau 2009: 111-26 This is not the only thinkable morphological reconstruction of this type. But this issue cannot be discussed in the space available here. The point is the existence of a factitive type by substitutive morphological means. 5 Rau 2009: 142, 143. 6 Rau 2009: 145. 3 4 *su̯er (Lith. sveriù svert̃ i ‘weigh’) : *su̯oró- ‘heavy’ (Lith. svarùs), *su̯ēro- ‘id.’ (Gmc. *swēra- > ON svárr, OE swǣr, OHG swār). *g̑neh3 ‘find out’ (Ved. jānā́ ti, Gk. γιγνώσκω etc.) : *g̑n̥h3-ro- ‘knowing’ (L. gnārus) 2.1.3.3 In addition, a PC adjective may be derived from something that is not exactly a “root” in the first place. A CS may, in theory, or may not then develop: *nu ‘now’ (Hitt. nu, Gk. νυ etc.) : *néu̯o- ‘new, young’ (Ved. náva-, L. novus etc.) 2.1.3.4 Perhaps most importantly for present purposes, a CS often developed around a PC adjective derived from a verbal root: *h1legwh (*h1l̥ -n(e)-gwh- > OIr. lingid ‘leaps’) : *h1ln̥gwh-ú- (RV raghú- ‘quick’, Gk. ἐλαχύς ‘slight’); *h1ln̥gwh-ró- (Gk. ἐλαφρός ‘light’, Gmc. *lungra- > OHG lungar ‘eager’, OE lungor ‘quick’; *h1lengwh-ii̯ os-/ *h1ln̥gwh-is- (AV lághīyas- ‘quicker, lighter’, Gk. ἐλάσσων ‘slighter, less’), *h1lengwh-isto- (YAv. rǝṇǰišta- ‘quickest’, Gk. ἐλάχιστος ‘slightest’) *h2eug (*h2u̯o/eg-s- ‘grow’ > Gmc. *wahsa- [OHG wahsan etc.]/ Gk. ἀέξω]) : *h2ug-ró- (Ved. ugrá- ‘great, mighty’, GAv. ugrō ‘id.’; *h2eug-ii̯ os- (Ved. ójīyas- ‘greater, mightier’, Av. aoǰiiah- ‘id.’). This is a cardinal point both in the matter of CSs in general and for the question of “adjectival roots.” It seems that as soon as a PC adjective is generated, from no matter what semantic direction, a CS becomes at least an open possibility (but no more than that: §§2.1.3.1 - 2.1.3.3). 2.1.3.5 In at least one conspicuous case, a fairly robust CS appears to have formed around an item that is not even a PC adjective, descriptively speaking: *pelh1-u-/*pl̥ h1-u- ‘much, many’(Go. filu etc., Ved. purú-) beside *pleh1-ro- (L. plērīque, Gk. πλήρης) *pleh1-i̯ os-/*pleh1-is- ‘more’ (Av. frāiiah-) and *pleh1-isto- ‘most’ (Gk. πλεῖστος) *polh1-u- ‘muchness, great quantity’ (cf. Gk. πολύ). These forms make up a clear CS, even if, like most, it is incomplete. But as it stands, there is neither a PC adjective nor a plausibly adjectival root involved:7 a. This CS is centered on a QUANTIFIER = DETERMINER, not a PC adjective. b. The root is *pleh1- ‘fill (up)’, which is hardly adjectival, which obviously means that not every adnominal that is central to a CS can be or needs to be analyzed as derived from an adjectival root. 2.2 A promising approach to a test for adjectival roots is membership in Watkins’ class of “adjectiveverbs.”8 These show a specified morphology with specific semantics in primary verbs to particular roots. More specifically, these may be characterized as primary (usually R+zero) verbal formations meaning ‘be(come) x’, where ‘x’ coincides semantically with a PC adjective. 2.2.1 Examples introduced by Watkins were: One may perhaps speculate that this anomaly was produced by a secondary development of the QUANTIFIER value of *p(E)lh1-(E)u̯-. Starting from *pleh1- ‘fill (up)’, the idea would be an inner-PIE development: (1) Synchronically deverbative (pre-)PIE *pólh1-u- ‘fullness’ and *pélh1-u-/*pl̥ h1-éu̯- ‘full’. (2) ‘Full’ and therefore ‘plentiful, copious, ample’ counted as PC adjective. (3) A CS was thus authorized: *pleh1-ro-, *pleh1-i̯ os-/*pleh1-is-, *pleh1-isto-. (4) *pélh1-u-/*pl̥ h1-éu̯- ‘ample, abundant’ etc. became the QUANTIFIER ‘much, many’. (5) The deverbative ‘full’ (= ‘filled up’) slot was then taken over by *pl̥ h1-nó- (Ved. pūrṇá-, OIr. lán etc.), and that developed no very extensive CS (cf. §2.1.3.2). 8 Watkins 1971: 68, Watkins 1973: 189-90. 7 a. Hitt. act. 3 sg. ḫāti ‘gets dry’ (pret. 3 sg. ḫāzta, pl. ḫāter; imv. 3 sg. ḫādu; midd. 3 pl. ḫādantari)9 ← *h2ed- ‘dry’ b. Hitt. midd. 3 sg. lukta ‘gets bright’ (pret. 3 sg. luktat);10 Ved. aor. “pass.” aroci;11 Toch. A pret. (III) midd. lyokät ‘got lit up’ and B pres. (III) lyuketär ‘shines’12 ← *leuk ‘bright, light’ c. Gk. perf.-pres. ἔρριγε13 ‘is cold’ (*sesrihxg-) ← *srihxg ‘cold’ 2.2.2 To these may be added other R+zero) verbal formations with comparable semantics:14 a. Hitt. midd. 3 sg. imv. parktaru ‘is to rise’ (cf. Toch. A pres. (III) pärkatär ‘rises’15) ← *bherg̑h ‘high’ b. Hitt. act. 3 sg. nekuzzi ‘gets dark’ (midd. 3 sg. nekutta, pret. nekuttat) ← *negw ‘colorless, dark’ c. Hitt. midd. 3 sg. tukkāri ‘is seen, is important’ (pl. dukkandari; pret. dukkat[i]) ← *tu̯ek ‘visible’ d. CLuv. midd. 3 sg. waššāri16 ‘is good’ ← *h1u̯es ‘valuable, important, dear’ e. Ved. aor. “pass.” áśoci17 ← *k̑euk ‘bright, shining’ f. Ved. pres. cité18 ‘is conspicuous’, aor. “pass.” áceti19 ‘appeared’ ← *kweit ‘visible’ g. Ved. aor. aśvitan20 ‘they brightened’ ← *k̑u̯eit ‘bright, white’ h. Ved. pres. tavīti ‘is strong’ ← *teuhx ‘strong’ 2.2.3 This “adjective-verb” concept comes close to recognizing roots with adjectival meanings. Watkins 1973: 188-90 Watkins 1971: 68-9, Watkins 1973: 190. 11 Jasanoff 2003: 153, 159-60, 168, 208 on the Ved. aor. pass. as a reflex of a h2e-conjugation root aorist and the analysis of this form as an example. 12 Jasanoff 2003:50, 159-60, 180 n.9 on the relevance of these Toch. forms to the reconstruction of a h2e-conjugation root aorist here. 13 Watkins 1973: 190 14 Bozzone 2016: passim promotes the position not so much that there were adjectival roots that made root-stem verbal formations (which is the idea here), but rather (1) that Pre-PIE had a category of “verb-like adjectives” for the assignment to nominals of PC attributes, (2) that these made or survive only as root aorists and (3) that this set of things formed the basis of CSs altogether. It is difficult to say, however, whether all of the verb stems at issue here (§§2.2.1 - 2.2.2) really represent root aorists rather than root presents (or even a primary perfect-present in one case) — and cf. in particular Ved. pres. cité ‘is conspicuous’ beside aor. “pass.” áceti ‘appeared’ (§2.2.2.f below). In addition, although the scenario presented in Bozzone 2016 is called an internal reconstruction of Pre-PIE, it depends more on the support of analogous setups in languages typologically very dissimilar to IE than upon the further processing of what can actually be reconstructed for PIE itself — and where an exercise of the latter kind could easily produce a rather different picture that could rather consequentially include, e.g., a set of primary nominal derivatives from the roots in question at any stage, no matter how remote, where primary verbal derivatives of those roots are envisioned. Discussion of the rest of the proposals made in Bozzone 2016 cannot be accommodated here, particularly as they mostly have to do with CSs per se, which are not the central focus of the present discussion. 15 Jasanoff 2003: 91-2, 94,118. 16 The wašš- of this verb as it stands is probably most immediately transferred from the verbal abstract waššar (n.)‘favor’, which would regularly reflect a *h1u̯és-ro- (type *déh3-ro- > δῶρον etc. ‘gift’): Nussbaum 2014: 250-1. 17 Jasanoff 2003: 155, 164, 208 for the analysis of this as a reflex of a h2e-conjugation root aorist. 18 Jasanoff 2003: 50, 170 for the analysis of this as a root present with stative-intransitive value. 19 Jasanoff 2003: 169, 170 for the analysis of this as a reflex of a h2e-conjugation root aorist. 20 See Bozzone 2016: 32-3. The evidence supplied by the apparent root (aor.) ptcpl. śvitāná- (as of most such R-āna- and R-ant- participles) is not necessarily conclusive, though this point cannot be treated here. 9 10 2.2.4 Necessarily omitted for now is the question of whether there other verbal stem types that are also decidably primary derivatives of adjectival roots.21 At issue would be, for example, R(e)-e/o- presents like: *tep ‘hot’ → *tep-e/o- > Ved. tápati ‘be(comes) hot’ etc. *k̑eubh ‘beautiful’ → *k̑eubh-e/o- > Ved. śóbhate ‘appears beautiful’ etc. 2.3 A second possible test for adjectival roots is provided by Schindler, who proposed a hypothesis tying certain ablaut patterns and specified semantics in root nouns to roots with particular inherent semantics. There are in this well known analysis two classes of agent root nouns: 2.3.1 Agents in R(o/e)- to “normal” verbal roots22 *klop- (Gk. κλώψ) ‘thief’ *klep- (L. cleps: ‘fur’ CGL V 349.51): — agent to *klep ‘steal’. 2.3.2 Agents in R(e/zero)- to “stative roots”23 *h2ner- (ἀνήρ, Ved. acc. náram, Osc. niir etc.) ‘man’ *h2nr- (ἀνδρός, Ved. nŕ̥bhiḥ etc.): — agent to *h2ner ‘be strong’ (vel sim.) that makes no verb forms that are not denominative. *g̑her- (χήρ, L. [h]ēr) ‘hedgehog’ (which shows the strong stem of an e/zero root noun; the weak stem of an o/e root noun to a root in -ER would have introduced zero grade -R and -R̥ already in the proto-language). — agent to *g̑her ‘be prickly’ (cf. *g̑her-s- > L. horrēre ‘bristle’ etc.). 2.3.3 For present purposes the question is whether there is any reason not to rewrite “e/z agents to stative roots” as “e/z substantivizations of adjectival roots.” For cases already mentioned above, this amounts to: *h2ner- ‘strong’ → *h2ner-/ *h2nr- ‘strong (one)’24; *g̑her- ‘prickly’ → *g̑her-/ *g̑hr- ‘prickly (one)’ — to which may be added: *bherg̑h- ‘high’ → *bherg̑h-/ *bhr̥g̑h- ‘high (thing)’ > YAv. bǝrǝz- ‘mountain’. 2.3.4 As an aside, this approach to both the verbal (§2.2) and the nominal (§2.3) material just introduced prompts consideration of “re-glossing” familiar “stative” roots as adjectival: *gel ‘cold’, *h1reudh ‘red’, *k̑reihx ‘outstanding’, *leubh ‘dear’, *magh ‘able’, *nebh ‘moist’, *pleth2 ‘broad’, *tep ‘warm’, *u̯eg̑ ‘lively’ etc. 2.5 The consequence is that there would emerge semantic plus morphological criteria for acknowledging the derivational patterns: PC adjectival root → essive/ fientive primary root aor. and/or root pres. (= “adjective verbs”) PC adjectival root → e/z root noun substantivization of the PC adjective (= “agents to stative roots”) And one and the same root occasionally makes root derivatives of both types: See Rau 2009: 136-84 for a discussion of Caland verbal formations in general and 146-160 for this type in particular. 22 Schindler 1972: 36. PIE o/e root nouns were also “patient”/ result nouns or nomina rei actae: *dom-/ *dem‘house’ (: *dem ‘build’) etc. In both functions these normally became R(o/zero)- in roots of the shapes -ER(T). 23 Schindler 1972: 38. PIE e/zero root nouns were also nomina actionis to “normal” verbal roots: *ker- (κήρ Hom., Pi., trag. lyr.)/ *kr̥r- (Alc. dat. κᾶρι, Alcm. acc. κᾶρα, κάρ˙ θάνατος) ‘doom’ (nomen actionis ‘termination’) to *ker ‘cut (off)’. 24 Root identification of this with Li. nóras (1) ‘wish’ (: norė́ ti ‘want’) and/or OCS nravŭ ‘custom, manner’ is barred both semantically and phonologically (Derksen 2015: s.v. norėti, Derksen 2008: s.v. nȏrvъ; nȏrvь). 21 *bherg̑h-/ *bhr̥g̑h- ‘get high, rise’ > Hitt. parktaru, Toch. A pärkatär *bherg̑h- ‘high’ *bherg̑h-/ *bhr̥g̑h- ‘high (thing), mountain’ > YAv. bǝrǝz*tu̯ek-/ *tuk- ‘be visible’ > Hitt. tukkāri *tu̯ek- ‘visible’ *tu̯ek-/ *tuk- ‘appearance’ > Hitt. tuekk(a)-/ tukk(a)- ‘body’, Ved. tvác- ‘skin’ 3. The two values of the root noun *h2ner-. 3.1 A root noun *h2ner- (ἀνήρ etc.)/ *h2n̥r- (ἀνδρός etc.) ‘man’ is well established. 3.2 But the semantics of ἀνήρ as a 2nd compound member are more complicated. So already in Homer. 3.2.1 -ήνωρ ‘man, men’ 3.2.1.1 This is unremarkable in form and meaning in two compounds of the “τερψίμβροτος type”: a. ῥηξ-ήνωρ ‘breaking (ranks of) men’, an epithet of Achilles (Il., Od.) in inflected formulae. b. φθεισ-ήνωρ ‘destroying men’, an epithet of πόλεμος (Il.) and also formulaic. 3.2.1.2 There is also a more noteworthy -ήνωρ: a. In the “factitive bahuvrihi”25 εὐήνωρ: εὐήνορα οἶνον (δ 622) “wine whose (= whereby) men get a good thing” = ‘making men happy’. b. As a special case of factitive εὐήνωρ: εὐήνορα χαλκόν (ν 19) “bronze(ware) wherein men have valuable property, wherein a man’s wealth consists” = ‘that makes men rich’.26 3.2.2 But there is also a 2nd member -ήνωρ that has the abstract meaning ‘strength, manliness, courage, mettle, high sprits’. 3.2.2.1 Such is the unforced reading of four compounds: a. 1ἀγήνωρ ‘having great manliness, courage’ < ἀγ(α)-ήνωρ ‘having great *aner-’: Hom. formulaic θυμὸς ἀγήνωρ # (frequent Il., Od.) Miscellaneous other such: Τρῶας ἀγήνορας (K 299) PN Ἀγήνωρ (Il.) ἀγηνορίη (Il.) ‘courage’ attributed to heroes. b. 2ἀγήνωρ ‘arrogant’ (2x Il., frequently of the suitors in Od.) and ἀγηνόριαι ‘arrogant behavior’ (1x Il.), simply analyzable as ἀγ(α)-ήνωρ ‘having excessive (cf. A-I ἄγᾱν ‘too much’ < *ága-u̯en[t]) high spirits’. c. the antonym ἀνήνωρ ‘cowardly’ (Od. 2=1x, Hes.). d. ὑπερήνωρ ‘overweening’ (Hes. Th.; ὑπερηνoρέοντ- already Il., Od.), easily taken as an augmentative of pejorative ἀγήνωρ ‘arrogant’ (vs. PN Ὑπερήνωρ, which would be an augmentative of the positive reading surely presupposed by the PN Ἀγήνωρ). But -ήνωρ ‘high spirits’ etc. rather than -ήνωρ ‘man’ is presumable in ὑπερήνωρ in any case. 3.2.2.2 Other accounts of ἀγήνωρ ‘strong, courageous’ are on the books. But they are complicated and not cogent.27 Term devised by J. Schindler. See Nussbaum 2014: 230-1. 27 Specifically, it has been proposed (see DELG2: 10 s.v. ἀγήνωρ) (1) that ἀγήνωρ actually reflects an *ag(e)-ānor- ‘who leads men’ (cf. PNs Ἀγέλαος [Il.+], Ἀγέμαχος [Erythrae 2c.], and Hom. epithets ἀρχέκακος, ἐχέθυμος, μενεπτόλεμος etc.); 25 26 3.2.3 The consequence is that some Gk. compounds have a 2nd member -ήνωρ ‘man’ or ‘men’while there are others in which -ήνωρ means ‘strength, courage’. 3.3 The question that immediately arises, of course, is that of whether there is comparative evidence for a root noun abstract *h2ner- ‘vigor, spirit’ (: Gk. -ήνωρ ‘strength, courage’). The answer seems to be that there are possibilities, but no decisive confirmation as yet. Potentially relevant items (some more suggestive than others), are these: 3.3.1 Ved. sūnára- ‘high-spirited, vigorous’ (of yúvā RV 8.29.1a, e.g.) would support *h2ner- ‘vigor’ if this is a bahuvrihi from *h1su-h2ner-o- ‘well endowed with *h2ner’. But *h1su-h2n̥r-o- would probably be expected. Another possibility would be a determinative sū-nára- ‘very vigorous’, with *h2nr̥- ‘strength’ → *h2nero- ‘strong’ as the pre-form not only of Ved. (*)nara-, but also of a pre-Lat. *nero- that is in any case the default assumption for the basis of PN Nerō, -ōn- and simultaneously that of the divine name Neria/ Neriō. 3.3.2 This abstract *h2ner- could hypothetically figure as well in Ved. sūnṛ́ta- ‘beneficent, magnanimous’ (voc. sū́ nṛta of Indra [8.46.20b], sūnṛ́tā of uṣā́ s [8.9.17b] etc.) on the theory of a *nṛta- ‘high-spirited’ (quasi *h2nr̥-to- : *h2ner- ‘high spirits’ [cf. ἀγ-ήνωρ §3.2.2]). a. More specifically, sūnṛ́ta- could on the one hand be taken to be a determinative like sú-vipra- ‘very inspired’ or sú-pūrṇa- ‘nice and full’. Alternatively there is the possibility of a bahuvrihi with a substantive *nr̥ta- (presumably neut. *nr̥tam) ‘spiritedness’ as its 2nd member to consider. A *h2nr̥-to‘spirited’ would be called for under either theory. (2) that ‘who leads men’ split semantically into ‘courageous’ and ‘arrogant’; and (3) that it was later reanalyzed as ag(a)-ānor- (= ἀγα+ήνωρ) ‘having great courage/ arrogance’ (whence antonymic ἀνήνωρ ‘uncourageous’ and augmentative ὑπερήνωρ ‘overweening’). There are a number of reasons not to accept this account: a. No trace of the putatively original meaning of ἀγήνωρ (‘who leads men’) is found. Nor does it seem possible to point to an actual passage giving the context in which the meaning is likely to have pivoted from ‘who leads men’ to ‘courageous’ or ‘arrogant’. b. In fact, ἀγήνωρ ‘courageous’ is best embedded as an epithet of θυμός and not that of one or more kings or heroes at all. This is especially inconvenient to the scenario. For it demands that ‘who leads men’ should have been revalued as ‘brave’ and then transferred from heroes to the θυμός — at which point not only the expressions in which it had simply and straightforwardly meant ‘who leads men’ but also those in which it meant ‘courageous’ in application to men were all simply eliminated from the Homeric repertoire. It seems fair to say that this postulated sequence of events is decidedly awkward. c. If ἀγήνωρ, which does not occur outside the epics, is a Homeric creation, as it would seem to be, and if it was intended to mean ‘leading men’, it is not at all clear why an epithet that would have as perspicuous, apt, and useful a meaning in epic as ‘who leads men’ should be liable to mis- or reinterpretations at all, and still less why it should have undergone not one, but two semantic revaluations — one positive (‘courageous’) and one negative (‘arrogant’). d. The side of the proposed semantic split on which ‘who leads men’ is supposed to develop to ‘arrogant’, although it doesn’t appear impossible, isn’t so natural-seeming as to be trivial either. e. While the *ag(e)-ānor- analysis requires a semantic history in which ‘who leads men’ partly develops to ‘arrogant’ but partly becomes ‘courageous’, both meanings can be directly derived from an *ag(a)-ānor- in which the first member aga- (< *m̥g̑h2- ‘great’) simply has the independently verifiable meanings that it shows in ἀγᾱν‘too (much)’ on the one hand and ἀγα-κλεής ‘having great fame’ etc. on the other — but only if the second member -ānor- represents a word for ‘courage, spirit’. b. A PIE *h2nr̥tom could be invoked not only for Ved. (*)nṛta- ‘spiritedness’ (as just above), but at the same time for OIr. nert ‘strength’ as well, if the scenario in mind is *h2nr̥tó- ‘strong’ → subst. *h2nértom (‘the strong’ >) ‘strength’ > *nertan > OIr. nert. c. In addition, the Ved. sūnṛ́ta- that occurs three dozen times in the RV is almost always found as fem. sūnṛ́tā-. This makes it thinkable that sūnṛ́tā- ‘generous (one)’ originated in a re-analysis of an abstract sūnṛ́tā- that would in the first instance have been a -tā- abstract28 of adjectival *sūn(a)r- (= Gk. εὐήνωρ) ‘generous’, and where, on the semantic side, ‘generosity’ developed to ‘generosity in person’, whence ‘generous one (fem.)’. The rare non-feminine sūnṛ́ta- would, in this scenario, be entirely back-formed to this re-analyzed fem. sūnṛ́tā- ‘generous (one)’. 3.3.3 Ved. nṛ-vánt- not only can mean ‘full of men’ (e.g. 6.25.6c of kṣáya- ‘settlement’), but also ‘strong’ (e.g. 4.55.4c of śárman- ‘protection’). Here the unanswered question would be, however, that of whether the derivational history (*)nṛ- ‘strength’ → nṛ-vánt- ‘strong’ is to be adopted in preference to nṛ- ‘man’ → nṛ-vánt- ‘manly” and therefore ‘strong’.29 3.3.4 Another trace of *h2ner-/ *h2n̥r- ‘strength’ could theoretically lurk in Hitt. innara- ‘forceful’ (: adv. innarā ‘willfully, on purpose’; factitive innaraḫḫ- ‘strengthen’ etc.) and CLuv. ānnara/i- ‘forceful’. The idea would be an inherited “ἔνθεος” bahuvrihi with compositional -o- of the form *en+h2nr̥r-o- ‘having strength (inside one), strong’. But against this is that the expected syllabification in such a pre-form would presumably be *en+h2n̥r-o- (as in *h2n̥r- > ἀνδρ-). A less debatable pre-form for Hitt. innara- and CLuv. ānnara/i- will be offered below (§7.3.3). 4. The formal identity of a given masc. substantivized adjective (SA) — in this case *h2ner-/*h2n̥r- ‘man’ — with its own adjectival abstract (AA) — in this case *h2ner-/*h2n̥r- ‘strength’ — is not isolated in IE: 4.1 Various types of secondary derivatives made from stem-form adnominals have the same property: Adnominal → Substantivized adnominal (SA) *h2erg̑ó- (ἀργός) ‘bright, shining’ *h2é́ k̑ro- (ἄκρος) ‘high, peaked’ *h2ο/erg̑i- (Hitt. ḫarki- ‘bright’) ‘bright, shining (one)’ *h2ο/ék̑ri- (L. ocris m.) ‘high (one)’ *díi̯ eu̯o- (L. dīus) *díi̯ eu̯e-t- (L. dīves, dīvit-) ‘brilliant, radiant’ *niu̯o- ‘low’ 28 29 30 Abstract (AA) *h2r̥g̑i- (ἀργι-, Ved. ṛji-30) ‘brightness’ *h2ο/ék̑rι- (ὄκρις, ἄκρις f.) ‘point, height’ ‘rich (one)’ *niu̯e-t- (Ved. nivát- f.) ‘depth’ (→ *neiu̯o- > νειός) *u̯eiko- (Gmc. *weiha- > Go. weihs) ‘holy’ *koupēd- (L. cuppēs, cuppēd-) ‘glutton(ous)’ = *u̯eikō̆n- (Gmc *weihan- > Go. weiha m. ‘priest’) *koupēd-(ō̆)n- (L. cuppēdō f.) ‘gourmanderie’ Though note (Debrunner 1954: 617): -tā- abstracts are far less common in RV than those in -tva-. Cf. Debrunner 1954: 876. Hom. ἀργί-ποδας (κύνας), Ved. ṛjí-śvan- (PN). 4.2 It is important to add with regard to Hitt. ḫarki- ‘bright’ and L. dīves, dīvit- ‘rich’ that a masc. SA is liable to become an adjective all over again in what may be called the “weak adjective” (WA) syndrome. 4.2.1 A familiar example is the development of *-(ō̆)n- de-adjectival derivatives from substantivizations to weak adjective in Germanic: *h1roudhó- ‘red’ → SA *h1roudhō̆n- ‘red (one)’ (L. Rūfōn- ‘red(-headed) guy’ = Germanic WA *raudan- ‘red’) 4.2.2 But there are plentiful, even if less systematic, parallels in other branches and with other morphological types of substantivizing derivatives: L. adj. mānus ‘good’ → SA mānēs ‘the goodly ones’, becoming a WA in dī mānēs ‘the goodly divinities’. 4.3 The three derivational types exemplified in §4.1 all produce derivatives of adnominals with the value of (1) SAs or (2) AAs or both: ‘the X1’ (masc. SA) Adnominal ‘X’ ‘the X2’ (fem. AA) — where the meanings of the basis and derivatives in this scheme can be exemplified by something like: (Eric) ‘the Red (one)’ = der, die, das Rote. ‘red’ ‘the red’ (of autumn leaves, of the morning sun etc.) = das Rot. 4.4 This, in turn, suggests: (1) that a root like *h2ner was actually adnominal or even adjectival; and (2) that the derivational process “adjectival root” (*h2ner) → root nominal (*h2ner-/*h2n̥r-) was a substantivization of an adnominal, i.e. a process semantically analogous to adjective → -i- nominal/ -t-nominal/ -n- nominal above. 4.5 That is to say: Secondary derivation of the kinds seen above — e.g. the i-stem type: Adnominal → *h2erg̑ó-‘bright, shining’ *h2é́ k̑ro- ‘high, peaked’ Substantival adnominal (SA) *h2ο/erg̑i- ‘bright, shining (one)’ m. *h2ο/ék̑ri- ‘high (one)’ m. Abstract (AA) *h2(ο/e)rg̑i- ‘brightness’ f. (?) *h2ο/ék̑ri- ‘height’ f. = is semantically paralleled on the level of primary root-noun derivation from “adjectival” roots: Adnominal → *h2ner ‘vigorous, spirited’ Substantival adnominal (SA) *h2ner-ø-/ *h2n̥r-ø‘vigorous, spirited (one); man’ m. = Abstract (AA) *h2ner-ø-/ *h2n̥r-ø- ‘vigor, spirit’ f. (?) — of which other examples would be (cf. §§2.2.1, 2.5): *bherg̑h ‘high’ *leuk ‘bright’ *bherg̑h-ø-/*bhr̥g̑h-ø- ‘high (one)’ m. > YAv. bǝrǝz- ‘high’ *bherg̑h-ø-/*bhr̥g̑h-ø- ‘height’ f. > YAv. bǝrǝz-, OIr. brí ‘hill’ *leuk-ø-/*luk-ø-/* ‘light’ f. > L. lūx, lūc-, Ved. rúc- 5. More inferences are possible. The next one might be this. (1) If “adjectival roots” derive root nouns which have the semantic value of SAs (type *h2ner- ‘man’) that are liable to become WAs (type *bhr̥g̑h- ‘high’), (2) a further possibility is that any given root nominal with adjectival value is potentially from an “adjectival root.” 5.1 And so, for one of the cases just mentioned, the adjectival meaning of YAv. bǝrǝz-31 in a passage like bǝrǝzǝm … vācim (acc. bǝrǝzǝm beside gen. bǝrǝzō) “a high (= loud) … voice” (Yt. 17.61) would lead us to the hypothesis of an adjectival root *bherg̑h ‘high’ (cf. §2.2.2) that formed a SA *bhr̥g̑h-. 5.2 In a similar case, we would suppose that an adjectival root *mā̆k̑ (*mah2k̑) ‘long, tall’ was the basis of a root SA *ma(h2)k̑- ‘long, tall (one)’ that followed the well worn path from SA to WA value, reflected by YAv. mas- ‘big’:32 masō xšaiiete xvarǝnaŋhō “Elle règne sur un grand xvarǝnah”33 (Yt. 5.96) 6. There are, in this connection, other relevant and interesting sets of forms as well. For the usefulness of assuming adjectival semantics for some roots goes well beyond the domain of root adjectives (whence root SAs and root WAs). Suffixes used in derivational processes of the type adjectival stem + suffix → SA and/or AA are also found in derivatives of the type adjectival root + suffix → SA and/or AA. 6.1 Substantivizing and “abstractifying” *-h2- may serve as the first example. a. With a suffixed, stem-form adjective as the basis, we find the following derivational semantics: SA (1) *u̯ēre-h2 ‘true thing’ > OCS věra f. ‘belief’ SA (2) *u̯ēre-h2 ‘true (one)’ >> WA >> feminine: L. vēra ‘true’ (f.) adj. *u̯ēro- ‘true’ > L. vērus AA*u̯ēre-h2 ‘trueness’ > OHG wāra f. ‘loyalty’ b. But the same simultaneous creation of a SA and an AA is accomplished by the addition of the same suffix to the root *meg̑, which it is therefore reasonable to analyze as having the inherently adjectival meaning ‘big’. SA *meg̑-h2 ‘the big’ >> WA ‘big’ > Gk. m., n. μέγα(ς); Hitt. c., n. mekkadj. *meg̑ ‘big’ AA*meg̑-h2 ‘bigness’ (*m̥g̑h2- > Gk. ἀγα-) → external deriv.*m̥g̑h2-ó- ‘big’ (>> Ved. mahá-) — whence SA *mo/eg̑h2-i>> WA (Hitt. mekki-); → external deriv. *m̥g̑h2-u̯n̥t ‘great’ >> PGk. *ágau̯en > A-I ἄγᾱν ‘too much’; → internal deriv. *meg̑-oh2/ *m̥g̑-h2- ‘big’ (Ved. mahā́ -/ mah-). Kellens 1974: 353-355. Kellens 1974: 356. 33 Translation from Kellens 1974: 356. 31 32 6.2 As a second example, we may consider the substantivizing and “abstractifying” denominal derivational formant *-(E)ta. Its behavior with suffix-bearing adjectives as its basis has already been illustrated (§4.1): *díi̯ eu̯o- ‘resplendent’ Hitt. ḫappina- ‘wealthy’ → → SA *díi̯ eu̯e-(o/e)t- ‘splendid (one)’ >> WA L. dīves, dīvit- ‘wealthy’ AA ḫappinatt- ‘wealth’ b. If the AA-forming function of this suffix is also to be found when it makes a root derivative, it is at least thinkable that once again an adjectival basis has had an abstract derived from it — the notable thing being in this case that the element acting as adjectival basis was a root, which would then be analyzed as meaning something adjectival: *negw ‘dark’ (§2.2.1) → AA (fem.) *nógw-t-/ *négw-t- ‘the dark, darkness’ (Hitt. gen. nekuz, Ved. nákt-, Gk. νύξ, νυκτ-, L. nox, noct- etc.) 6.3 Third example: substantivizing/ “abstractifying” -(ō̆)na. The derivational semantics of this formant have also already been illustrated (§4.1): *h1roudhó- ‘red’ (Go. rauþs, Li. raũdas etc.) *koupēd- (L. cuppēs, cuppēd-) ‘glutton(ous)’ → SA *h1roudhō̆n- (L. Rūfōn- ‘redhead’ = Gmc. WA *raudan- ‘red’) → ΑΑ *koupēd-(ō̆)n- (L. cuppēdō f.) ‘gourmanderie’ b. If the AA-forming function of this suffix is also observable when it makes a root derivative, it is entirely thinkable that once again a root from which such an abstract was directly derived was itself adjectival. Potential examples of just that situation are not impossible to find. A PIE root from which are derived words for ‘gray’ and the characteristically gray ‘hare’ may be set up as *k̑as: *k̑as-ó- ‘gray’ (substantivized to Ved. śaśá- m. ‘hare’), *k̑as-no- > L. cānus ‘gray-haired’, *k̑as-ko- > L. cascus ‘grayhaired, old’. Directly from this root, to all appearances, are made both a SA and an AA (even if there is only indirect evidence for the latter):34 SA *k̑ás-on-/*k̑as-n- ‘gray (one)’ (OHG haso etc. ‘hare’/ W. ceinach ‘hare’) * k̑as ‘gray’ (Ved. śaśá-, L. cascus) AA *k̑ás-on-/ *k̑as-n- ‘grayness’ → *k̑asn-o- > L. cānus ‘grayhaired’ 6.4 Fourth example: substantivizing/ “abstractifying” *-ka. Adjectival stems as basis: *séno- ‘old’ (Ved. sána- , Li. sẽnas etc.) → SA *sene-k- ‘old (one)’ > L. senex ‘old man’ b. Adjectival root (?) as basis: 34 See Höfler 2017: 181-2, where, however, a different morphological and derivational analysis is offered. S/WA *mleh2-k-/ *ml̥ h2-k- ‘soft (in the head)’ > βλᾱ́ξ ‘stupid’ *mleh2 ‘soft’ (RV mlāta- ‘softened’, YAv. amrāta- ‘unsoftened’) AA *mleh2-k-/ *ml̥ h2-k- ‘softness’ → *ml̥ h2k-ó- > μαλακός ‘soft’ 7. R(o)-o- derivatives of adjectival roots. 7.1 Earlier in the discussion it was hypothesized that the PIE items *h2nér-ø-/ *h2n̥r-ø-, *meg̑-h2 and *nógw-t-/ *négw-t-, as root-derived, were primary formations. The special proposal, however, was that they were primary derivatives of “adjectival roots,” and so had the semantics (AA and SA >> WA) of certain secondary derivatives to adjectival stems. 7.2 This raises the question of whether there are other primary types of nominals that work as AAs and/or SAs >> WAs when made from plausibly adjectival roots. We may look into this by way of the primaryseeming R(ó)-o- (“τόμος”) type of nominal derivative: a. It is well established that R(ó)-o- nominals have the value of abstracts to verbal roots (RV dábha‘harm’, Gk. γόνος ‘birth’ etc. etc.). b. If, now, e/zero root nouns make verbal abstracts to verbal roots35, but AAs and SAs (>> WAs) to adjectival roots, it is worth investigating whether R(ó)-o- makes AAs and SAs (>> WAs) to adjectival roots beside its verbal abstracts to verbal roots. 7.3 The answer to that question is yes and no. R(ó)-o- AAs do seem to have been a type, but R(ó)-o- SAs >> WAs are insecure for PIE itself. 7.3.1 A random plausible case of a R(ó)-o- AA is *g̑hólh3-o- ‘the green’ > Gk. χόλος (m.) ‘bile’ (> ‘anger’ etc.). This item is better taken as a primary AA than as the substantivization of some derivationally prior adjective for at least two reasons: a. Adjectives substantivized to give abstracts are almost always neut. or fem., not masc. b. This abstract makes a marked and plausibly archaic and inherited type of adjectival derivative of its own: *g̑hólh3-o- ‘greenness, yellowness’ ( χόλος) → *g̑helh3-ó- ‘yellow’ (OIr. gel ‘shining’). This is of the type represented also, and in another plausibly adjectival root (§2.2.1), by *lóuko- ‘the light; bright place’ (Gmc. *lauha- > OE lēah ‘open land’ etc., Li. laũkas ‘open air’ [2 Žem.]) → *leukó- (λευκός ‘white’). And, though neither accent can be confirmed, the same kind of R(e)-ó- adjectival derivative may theoretically be seen in a case like *dʰóubo- ‘depth, deep place’ (Toch. B taupe, A top ‘mine’ [?]36) → *dʰeubó- (Gmc. *deupa- ‘deep’). 7.3.2 Other potential cases of R(ó)-o- as AAs to adjectival roots to consider are: *k̑ók-o- > Ved. śā́ ka- ‘might’ *sór-o- *‘redness, red stuff’ > Ved. sā́ ra- ‘heartwood, solidity’ (cf. ἔρευθος ‘redness’ : L. rōbur, -ōr- ‘oak, sturdiness’) *sróm-o- > Ved. srā́ ma- ‘lameness’ 35 See note 22 above. OLith. daũbas (4) ‘valley’ is not directly useful, of course, for supporting a barytone *dʰóubo-. But it does help suggest an old R(o)-o- stem to this root. See §7.4 below. 36 7.3.3 An especially interesting potential example of a R(ó)-o- AA to an adjectival root should also be mentioned. 7.3.3.1 Hitt. innara- ‘forceful’ (: adv. innarā ‘willfully, on purpose’, factitive innaraḫḫ- ‘strengthen’ etc.) together with CLuv. ānnara/i- ‘forceful’has already (§3.3.4) been brought up. 7.3.3.2 At that earlier point in the discussion the analysis of this Anatolian word as an “ἔνθεος” bahuvrihi with zero-grade 2nd member and compositional -o- was dispreferred because it would require the syllabification *en+h2nr̥r-o-, against which there is evidence. As to other possibilities, an *en+h2nor-o- is also to be disfavored. The 2nd member of a PIE bahuvrihi before compositional -o- should be “zeroed out” (e.g. *du̯i-h2p-o- > Ved. dvīpá- ‘river island’ [: *h2ap- ‘water’] etc.). 7.3.3.3 But an “ἔνθεος” bahuvrihi is still possible with a different account of the 2nd member: Adjectival root *h2ner ‘strong’ → R(ó)-o- AA *h2nór-o- ‘strength’ → *en-h2noró- a. The oxytone accent of the compound in Hitt. is reminiscent of that of bahuvrihis with o-stem 2nd members elsewhere: RV bṛhad-rathá- ‘with a high wagon’ beside RV bṛhád-ratha- ‘id.’ b. The initial-syllable accent (as implied by Čop’s law) of CLuv. ānnara/i- may theoretically reflect an alternative *én-h2noro-. But it is also possible that the actual CLuv. form is ānnari-, and that it continues an inherited SA and/or SA >> WA in -i-. That kind of SA derivative of an adjectival is ultimately a PIE acrostatic (of the “ὄκρις type”), in which the root accent is sometimes preserved: AV ūná- ‘lacking, short of the right amount’ (*h1uh2-nó-) : Gk. εὖνις ‘bereft’ (WA as if < *h1úh2-ni-). 7.3.3.4 Reconstructing an AA *h2nór-o- ‘strength’ is perhaps somewhat arbitrary on the basis of what has been seen so far. But it gets additional support from its utility for clarifying another form: the Ηesychius gloss νωρεῖ· ἐνεργεῖ ‘is effective’. a. There is, of course, a prima facie semantic likelihood to the idea that νωρεῖ (· ἐνεργεῖ) ultimately belongs somehow with *h2ner ‘strong’. But the formal details remain to be filled in. b. The derivational situation, however, is easily elucidated as involving hypothetical: *h2nōró- ‘strong’ ← *h2nóro- ‘strength’ with the vrddhi semantics of, e.g., *u̯ēdo- ‘wet’ (ON vátr , OE wǣt) ←*u̯ed- or *u̯edo -‘water’37 RV mā́ dhvi- ‘sweetened (one)’ (nom.-voc. masc. du. mā́ dhvī) ← mádhu ‘sweet stuff’ YAv. vārǝθraγna/i- ‘victorious’ ← vǝrǝθraγna- ‘victory’ and the derivational history: νωρεῖ < *nōréi̯ e/o- ‘be strong, efficacious’ ← *(h2)nōró- ‘strong’ ← *(h2)nóro- ‘strength’. 7.4 The class of R(o)-ó- stems of the τομός type is also very familiar. When made to verbal roots, these nominals often function as semi-participial adjectives, both active and passive. Gk. τομός ‘cutting, sharp’ is itself a clear example of the former, as is λοιπός ‘left (over), remaining’ of the latter. These semantic values, in turn, can be handled within an analysis of these R(o)-ó- stems as “possessive” derivatives of the R(ó)-o- τόμoς-type nouns that were mentioned above — i.e. *tómh1o- ‘(act of) cutting’ (> τόμoς ‘a cutting, a piece’) → *tomh1ó- ‘having (= performing) cutting, sharp’. Hitt. and/or Luv. wid- and/or wida/i-, ‘water’ and/or ‘wet’, appear to be multiply ambiguous, but either match the Gmc. adjective (substantivized and not) or reflect a *u̯ed(o)- ‘water’ and/or ‘wet’. See Kloekhorst 2008: 1015-16 with references. 37 There is also, however, a class of R(o)-ó- forms that are quite clearly property concept adjectives and are made to roots that are cleanly classifiable as “stative” in Schindler’s conception and/or have “adjective verbs” as their derivatives in Watkins’. It is consequently reasonable to think of them as ultimately representing R(o)-ó- possessive derivatives of the R(ó)-o- AA type touched upon in §7.3 just above. An example of such a thing would be a derivative like: *dʰoubó- > OLi. daũbas (4) References Balles, Irene 2003. “Die lateinischen Adjektive auf -idus und das Calandsystem.” Indogermanisches Nomen. Derivation, Flexion und Ablaut, ed. Eva Tichy, Dagmar S. Wodtko, Britta Irslinger. Bremen: Hempen, 9-29. Balles, Irene 2006. Die altindische Cvi-Konstruktion, Form, Funktion, Ursprung. Bremen: Hempen. Balles, Irene 2009. “The Old Indic cvi construction, the Caland system, and the PIE adjective.” Internal reconstruction in Indo-European: Methods, results, problems, ed. Jens Elmegård Rasmussen and Thomas Olander. Copenhagen: Museum Tusculanum Press, 1-15. Bozzone, Chiara 2016. “The Origin of the Caland System and the Typology of Adjectives.” IndoEuropean Linguistics 4: 15-52. Debrunner, Albert 1954. Altindische Grammatik, vol 2.2. Die Nominalsuffixe. Göttingen : Vandenhoeck & Ruprecht. DELG2. Pierre Chantraine. Dictionnaire étymologique de la langue grecque. Histoire des mots. Achevé par Jean Taillardat, Olivier Masson et Jean-Louis Perpillou; avec en supplément, les Chroniques d'étymologie grecque (1-10) rassemblées par Alain Blanc, Charles de Lamberterie et Jean-Louis Perpillou. Paris: Klincksieck, 2009. Derksen, Rick 2008. Etymological dictionary of the Slavic inherited lexicon. Leiden-Boston: Brill. Derksen, Rick 2015. Etymological dictionary of the Baltic inherited lexicon. Leiden-Boston: Brill. Dixon, R. M. W. 1982. Where have all the adjectives gone? And other essays in semantics and syntax. Berlin-New York: Mouton. Höfler, Stefan 2017. Der Stier, der Stärke hat. Possessive Adjektive und ihre Substantivierung im Indogermanischen. Dissertation (University of Vienna). Unpublished. Jasanoff, Jay 2003. Hittite and the Indo-European verb. Oxford-New York: Oxford University Press. Kellens, Jean 1974. Les noms-racines de l’Avesta. Wiesbaden: Reichert. Kloekhorst, Alwin 2008. Etymological dictionary of the Hittite inherited lexicon. Leiden - Boston : Brill. Nussbaum, Alan J. 2014. “The PIE Proprietor and His Goods.” Munus amicitiae Norbert Oettinger a collegis et amicis dicatum, ed. H. Craig Melchert, Elisabeth Rieken, Thomas Steer. Ann Arbor: Beech Stave, 228-254. Rau, Jeremy 2009. Indo-European Nominal Morphology: The Decads and the Caland System.. Innsbruck: Institut für Sprachwissenschaft der Universität. Schindler, Jochem 1972. “L’ apophonie des noms-racines indo-européens.” BSL 67: 31-38. Watkins, Calvert 1971. “Hittite and Indo-European studies: The denominative statives in -ē-.” TPS: 5193. Watkins, Calvert 1973. “An Indo-European agricultural term: Latin ador, Hittite ḫat-.” HSCP 77: 187-93.