Introduction
In chapter 2, it was argued that an analogical account of the vocalism of Ïá½³ÏαÏÏÎ¿Ï runs into problems. Moreover, in chapter 1 καÏÏεÏá½¹Ï and ÏαÏÏá½»Ï have been identified as problematic forms for the idea that âÏαâ is the regular reflex of word-internal *rÌ¥ in Ionic-Attic. I therefore hypothesize that the regular Ionic-Attic reflex is âαÏâ. This means that a considerable number of forms with âÏαâ < *rÌ¥ must be accounted for. This is the objective of chapters 4 till 9.
Within this context, it is of the utmost importance to systematically examine the analogical processes that may have influenced forms with âÏαâ and âαÏâ. In this chapter, I discuss forms with âÏαâ and âαÏâ belonging the so-called âCaland systemâ.1 I will first give a descriptive overview of this system from a Greek point of view, focusing on reconstructible ablaut patterns (section 4.1) and the productivity of derivations (section 4.2). After that, a detailed account of the reflexes of *rÌ¥ and *lÌ¥ in individual formations, notably u-stem adjectives (sections 4.3 to 4.5), will be given. The etymological family of καÏÏεÏá½¹Ï, κÏá½±ÏÎ¿Ï and related forms is treated separately in chapter 5.
4.1 The Root Vocalism of âCalandâ Formations in Greek and PIE
According to Calandâs original formulation of the phenomenon that bears his name, Indo-Iranian adjectives in âraâ, âmaâ, or âantâ replace these suffixes with âiâ when appearing as a first compound member.2 Calandâs prime examples from Avestan included dÉrÉzraâ âfirmâ beside dÉrÉzi-raθaâ âhaving a firm chariotâ, and xrÅ«raâ âbloodyâ beside xruui-druâ âhaving a bloody weaponâ. Wackernagel (1897) then extended the substitution rule from Indo-Iranian to Greek, adducing cases like ÎºÏ Î´Ïá½¹Ï âgloriousâ beside ÎºÏ Î´Î¹-άνειÏα âbringing fame to menâ. Crucially, he argued that the phenomenon is inherited from PIE in view of the equation Ved. rÌ¥jÃ-Åvanâ PN âwho has swift dogsâ beside á¼Ïγι-κέÏÎ±Ï Î½Î¿Ï âwith bright/swift lightningâ (Il.+) and á¼ÏγίÏÎ¿Î´Î±Ï Îºá½»Î½Î±Ï âswift-footed dogsâ (Il. 24.211; cf. also Hom. Îºá½»Î½ÎµÏ á¼Ïγοί and Îºá½»Î½ÎµÏ Ïá½¹Î´Î±Ï á¼Ïγοί).3 Furthermore, Wackernagel remarked that other suffixes participated in the alternation as well: notably s-stems (á¼Î½-αÏÎ³á½µÏ âbrightâ beside á¼Ïγιâ, Îºá¿¦Î´Î¿Ï âgloryâ beside ÎºÏ Î´Ïá½¹Ï) and u-stem adjectives (Ved. rÌ¥júâ âstraightâ beside rÌ¥jÃ-pyaâ âflying straightâ, epithet of the eagle).4
Although the use of *âiâ as a compounding allomorph of *âroâ played a key role in the discovery of âCalandâ morphology, this archaic substitution rule has lost its central place in more recent discussions. Scholars like Risch and Nussbaum have stressed that âCalandâ morphology is primarily a system of regularly alternating affixes that must be studied as a historically developing entity, with its own dynamics in the individual languages. Thus, Meissner speaks of the âCaland systemâ as a âregular and well-defined set of correspondences of derivational affixesâ (2006: 3).5 What is remarkable about these correspondences is that roots which combine with âCaland suffixesâ do not normally take other derivational suffixes (such as *âtiâ, *âmnÌ¥â).
In Greek, a model âCaland systemâ consists of a primary adjective (often in âá½»Ï or âÏá½¹Ï, though other suffixes are possible too), forms of comparison in âá½·Ïν and âιÏÏοÏ, compounded adjectives in âηÏ, and a neuter noun in âοÏ.6 Sometimes, a compounding first member in âιâ is found. Thus, the following Greek forms containing the root ÎºÏ Î´â are attested:
-
Adj. ÎºÏ Î´-Ïá½¹Ï
-
Comp. κῡδ-á½·Ïν, superl. κύδ-ιÏÏοÏ
-
Neuter abstract κῦδ-οÏ
-
Cpd. á¼Ïι-κῡδ-á½µÏ
-
Cpd. κῡδ-ι-άνειÏα.
From a Greek point of view, these suffixes (as well as adverbial âα, on which see section 5.2.9) can be called âcentralâ, as opposed to âmarginalâ suffixes (such as âνόâ and âμοâ).7 In addition, several verbal formations have close ties to these nominal forms. In Greek, these are notably the stative verbs in âá½³Ï (with a Homeric aorist in âηÏâ) and the factitive verbs in âá½»Î½Ï and/or âαίνÏ. The only Greek root to attest all nominal and verbal formations just mentioned is that of κÏá½±ÏÎ¿Ï âmightâ; derivations from this root will be extensively discussed in chapter 5. Most Greek âCalandâ roots, however, have one or several gaps in their âsystemâ.
Already before our first attestations, many Greek âCalandâ roots generalized one root vowel throughout the entire system of derivations.8 Thus, beside the adjective ÏαÏá½»Ï âquick, swiftâ, we find a comparative θάÏÏÏν, a superlative Ïá½±ÏιÏÏοÏ, a neuter abstract Ïá½±ÏÎ¿Ï âspeedâ, and an adverb Ïá½±Ïα, all of which are attested from Homer onwards. As we will see below, there are good arguments for reconstructing an original non-ablauting e-grade root in the forms of comparison and in the neuter abstract, at least for the variety of late PIE from which Proto-Greek developed.9 This suggests that the forms θάÏÏÏν, Ïá½±ÏιÏÏοÏ, and Ïá½±ÏÎ¿Ï were influenced in their root vocalism by the positive ÏαÏá½»Ï, which can be considered the basic formation.
A second example is furnished by the following Greek forms derived from the PIE root *pleth2â: ÏλαÏá½»Ï âwideâ, ÏλάÏÎ¿Ï âflat open surfaceâ, and compounds in âÏλαÏá½µÏ. An e-grade root would be expected in the s-stem noun and adjective (cf. Ved. práthas-),10 but again, Greek has forms with a-vocalism. The system has clearly been reshaped on the basis of the adjective.11
These fairly trivial examples teach us that nominal forms with âÏαâ, âλαâ or âαÏâ, âαλâ need not (or do not) directly continue a pre-form with *rÌ¥ or *lÌ¥. Thus, in order to judge the provenance of Caland forms and their pertinence as evidence for the regular reflex of *rÌ¥ or *lÌ¥, we must first obtain a clearer picture of the expected ablaut paradigms, in PIE and in early Greek, of the formations involved. Before turning to the reconstruction of these paradigms, however, an important caveat must be made. Many Caland roots are considered to be primarily adjectival or nominal, either because verbal forms are unattested or because they are morphologically marked (with suffixes like *âeh1â).12 The root of ÏαÏá½»Ï offers an illustrative example. However, as stressed by Lamberterie (1990: 38â39), many IE languages have examples of deverbal u-stem adjectives; in these cases, influence of verbal forms on the root shape of the adjective (and other nominal fomations) must be reckoned with. For instance, Lith. platùs âextendedâ beside iÅ¡-pleÄiù âI stretch outâ follows the model of e.g. badùs âsharpâ beside bedù âI stingâ. An important consequence of this observation is that forms like Lith. platùs do not allow us to reconstruct an o-grade root allomorph in the PIE u-stem adjectives.13
Within Greek, too, there is some evidence for the derivation of u-stem adjectives from intransitive verbs. Lamberterie (1990: 414â417 and 542â544, cf. 957) adduces the examples Hom. βÏÎ¹Î¸á½»Ï âheavyâ (from Hom. βÏίθÏ, βέβÏιθα âto be heavyâ) and Hom.+ ÏÏηÏá½»Ï âroughâ (beside Hom. Ïá½±ÏαÏÏÏ âto stir upâ, pf. Ïá½³ÏÏηÏα âto be in upheavalâ). Risch (1974: 65), too, observed that Caland forms are often derived from verbs, quoting á¼Î»á½³Î³ÏÏ âto put to shameâ, á¼Î»ÎµÎ³ÏÎ¿Ï âdisgraceâ, superl. á¼Î»á½³Î³ÏιÏÏÎ¿Ï âmost shamefulâ. More recently, Meissner (2006: 186â197) has demonstrated that Greek s-stem adjectives are frequently derived from intransitive verbal formations like the aorist in âην and (following Tucker 1990) from stative verbs in *âÄâ, and Blanc (2018) has adduced abundant evidence for âÎ·Ï as a general deverbal compound suffix.14
It was thought for a long time that such deverbal Caland formations were innovations of Greek. However, Rau shows that a deverbal origin can also be assumed for a large number of Caland adjectives in Indo-Iranian, where âthe vast majority of Caland system adjectives (â¦) pair with stative/inchoative and factitive formations that are to all appearances primaryâ (2009: 138â139). Frequently, the primary verb is a full-grade thematic formation, mostly an intransitive middle, sometimes accompanied by a secondary causative active form. In the example used earlier, PIE *pleth2â, Vedic has an intransitive primary verb práthate âspreadsâ. In Greek, too, many individual Caland systems stand beside primary thematic middle presents (see the overview in Rau 2009: 152â155). Cases of interest in the present context are primary ÏÏá½³Ïομαι âto coagulate, grow fatâ (cf. ÏαÏÏá½»Ï âthick, numerousâ, ÏÏαÏεÏá½¹Ï âthick, solidâ) and μέλδομαι âto become softâ (cf. βλαδύÏ* âweakâ).15 Clearly, such cases are archaisms: it was no longer possible to derive an adjective in âá½»Ï within Greek.
After these preliminary remarks, let us consider in more detail the ablaut paradigm of the most important primary formations: u-stem adjectives, forms of comparison, and s-stem nouns and adjectives.
4.1.1 The u-stem Adjectives
In Greek u-stem adjectives we generally find a zero-grade root and suffixal ablaut. For instance, the suffix of βαÏá½»Ï âheavyâ is âÏ â in the nom.-acc. sg. of the masculine and neuter, and reflects *âeu̯â elsewhere in the paradigm. A similar situation is reflected in cognate Indo-Iranian adjectives of the type urúâ âwideâ, prÌ¥thúâ âbroadâ, in which the suffix of the m. and n. oblique stem derives from *âeu̯â. Since this type of suffix ablaut is at home in the proterodynamic (PD) accent/ablaut-paradigm, the reconstruction of a regular PD u-stem adjectival paradigm for PIE (including root ablaut, i.e. strong stem *CéRC-uâ, weak stem *CRC-éuâ) is widely accepted.16
Since this ablaut pattern could play an important role in accounting for the root vocalism of adjectives like κÏαÏá½»Ï, ÏλαÏá½»Ï, etc., let us first review the reasons for reconstructing it. Within Greek, an important piece of evidence is δαÏá½»Ï âhairy; densely grownâ. The main question here is how to explain the retention of intervocalic âÏâ; this has previously been ascribed to expressive gemination (Szemerényi 1954: 261) or to a âdouble treatmentâ of *âNÌ¥sVâ (DELG s.v. Î´Î±Ï Î»á½¹Ï), but neither of these proposals offers a satisfactory solution.17 The formation is clearly inherited, because a stem *dnÌ¥s-uâ is also presupposed by the near-synonym Î´Î±Ï Î»á½¹Ï (or perhaps rather δαῦλοÏ) âdense, hairy, shaggyâ < *dnÌ¥s-u-lóâ.18 In Latin we find dÄnsus âthick, denseâ, and Hittite has daÅ¡Å¡u- âstrong, powerful; heavy, well-fed; difficult, importantâ (among other meanings). The latter form is important because its geminate âÅ¡Å¡â can only be explained if we reconstruct *déns-uâ rather than *dnÌ¥s-uâ (EDHIL, s.v. dasÌsÌuâ). Like the verbal forms δαá¿Î½Î±Î¹ âto learnâ, δέδαε âtaughtâ, and the relic first compound member of δαá¿ÏÏÏν âprudentâ, Î´Î±Ï Î»á½¹Ï shows that PGr. *âNÌ¥sVâ was regularly lenited to *âNÌ¥hVâ. This makes the retention of âÏâ in δαÏá½»Ï an even more urgent problem.
The retention can be explained, however, if we suppose that δαÏá½»Ï continues an ablauting paradigm *déns-uâ, *dnÌ¥s-éu̯â,19 and that intervocalic *âsâ was lenited earlier than intervocalic *âNsâ. In this case, *dnÌ¥h-eu̯â could be restored to *dnÌ¥s-eu̯â on the basis of the strong stem *dens-uâ. In Î´Î±Ï Î»á½¹Ï, on the other hand, the âsâ was not restored, presumably because the paradigm did not have ablaut. Thus, the pair δαÏá½»Ï beside Î´Î±Ï Î»á½¹Ï provides indirect evidence that u-stem adjectives preserved paradigmatic root ablaut in Proto-Greek; in addition, the presence of an e-grade in the PIE paradigm is proven by Hittite daÅ¡Å¡u-.
There is some further suggestive Greek evidence for the presence of forms with an e-grade root within the original paradigm. Willi (2002) attractively derives Att. εá½Î¸á½»Ï âstraight atâ and Ion. (Hom.+) á¼°Î¸á½»Ï âid.â from a single pre-form PGr. *i̯euthuâ.20 He compares this Proto-Greek form with Lith. judùs âbelligerentâ21 and derives both from the verbal root *(H)i̯eudhâ âto go straight atâ, reflected in Ved. yudh âto fightâ, Lat. iubeÅ âto orderâ (OLat. ioubeÅ âto sanctionâ), Lith. jùsti âto get movingâ, judÄÌti âto be agile, stir (intr.)â.22 Since this verbal root is unattested in Greek, it is likely that forms with e-grade root were originally present in the paradigm, i.e. PIE *(H)i̯éudh-uâ, *(H)i̯udh-éuâ.
As for PIE *su̯eh2d-u- âagreeable, tastyâ, all IE languages that continue this formation agree in showing the reflex of a full grade root: Gr. ἡδύÏ, Ved. svÄdúâ, Lat. suÄvis, OE swÅt, etc. Since zero grade forms of this root are found in Vedic (caus. sÅ«dáyati âto make acceptableâ, sÅ«Ìdaâ âsweetnessâ), we know that it could undergo ablaut in the proto-language.23 The question remains whether one can exclude that the lexical entry *su̯eh2d-uâ had a non-ablauting root already in the particular chronological phase that corresponds to reconstructed PIE. In my view, the most natural scenario would be to reconstruct a PD paradigm for PIE itself, with subsequent generalization of the full grade root in the daughter languages. Indeed, there would have been a clear motive for this generalization. After the loss of laryngeals, the outcome of the zero grade *suh2dâ was *sÅ«dâ in most languages; and since the resulting ablaut *su̯Ädâ : *sÅ«dâ was anomalous, it would not be surprising if all daughter languages eliminated it independently.24 One also wonders whether the vocalism of á¼¡Î´á½»Ï was perhaps influenced by the primary thematic verb underlying ἥδομαι âto enjoy oneselfâ (cf. Ved. svádate, svÄÌdate âto become tastyâ). This explanation is conceivable for Greek, but it is less evident for most other branches that have a trace of *su̯eh2d-uâ, as they show no trace of the primary verb. We must therefore assume that the PIE adjective contained an e-grade root at least in the strong case forms.
In various different daughter languages, there are scattered remains of original u-stem adjectives with an e-grade root. Examples:
-
Lat. brevis âshortâ < *mreǵhu̯iâ << *mréǵh-uâ;25
-
Lat. gravis âheavy; importantâ < *gwrau̯â plus âiâ << *gwreh2-uâ;26
-
Arm. meÉ«k âsoftâ < *meldu̯iâ << *meld-uâ âweak; softâ;27
-
Hitt. tÄÌpuâ âlittle, fewâ < *dhébh-uâ.28
Like εá½Î¸á½»Ï in Greek, these forms can be taken as evidence for the presence of an e-grade root somewhere in the original u-stem paradigm. It must be granted, however, that the evidence is indirect. As for Lat. brevis and gravis, the exact derivational history of this type of Latin i-stems is disputed,29 but in my view it cannot be doubted that the type ultimately developed from PIE u-stem adjectives, as there are too many precise lexical correspondences between Latin and the other languages. Thus, the Latin forms can be taken as evidence for proterodynamic ablaut (with Fischer 1991: 7).
In sum, the case of δαÏá½»Ï (and Î´Î±Ï Î»á½¹Ï) beside Hitt. daÅ¡Å¡uâ suggests that u-stem adjectives had PD inflection with root ablaut in PIE. Additional corroboration comes from εá½Î¸á½»Ï, *su̯eh2d-uâ, and outside Greek from cases like Lat. brevis. If the explanation proposed here for the retained sibilant in δαÏá½»Ï is correct, it proves that the root ablaut was retained up till Proto-Greek.
In Van Beek 2013, I went one step further, assuming that the root ablaut was retained as late as Proto-Ionic. I used this to argue for an analogical origin of the reflex âÏαâ in Greek adjectives like βÏαÏá½»Ï. When the weak stem *mrÌ¥khéu̯â vocalized as *markhéu̯â, it would have been analogically changed into *mrakhéu̯â under influence of the strong stem *mrékhuâ, after which it would have ousted the latter. This assumption is rather costly, however, and in section 4.3.3 I will propose to account for the leveled zero-grade reflex in an alternative way, by means of influence of the forms of comparison.
4.1.2 Primary Comparatives and Superlatives
The reconstruction of the so-called primary comparatives and superlatives is important for our purposes for more than one reason: the formations are unproductive in Greek, and their root vocalism shows traces of ablaut with respect to the positive. As we will see, a fair amount of analogical reshaping must have taken place in these formations at a relatively shallow time-depth.
According to the most widespread view,30 PIE primary comparatives had an e-grade root with ablaut in the suffix, whereas primary superlatives regularly had a zero grade root. This is motivated as follows by Schwyzer (1939: 538):
Die Wurzel hatte ursprünglich bei den Komparativen mit âá½·Ïν Starkstufe, bei den Superlativen auf âιÏÏÎ¿Ï Schwachstufe, z.B. κÏαÏá½»Ï, κÏá½³ÏÏÏν (wie κÏá½³ÏοÏ) κÏá½±ÏιÏÏÎ¿Ï (â¦). Doch zeigen die Superlative schon früh auch Starkstufe und Anfangsakzent: Ïá½³ÏιÏÏÎ¿Ï (â¦); umgekehrt wurden die Komparative früh dem Superlativ bzw. Positiv angeglichen, z.B. dor. κάÏÏÏν kret. κάÏÏÏν statt κÏá½³ÏÏÏν nach κάÏÏιÏÏοÏ; μάÏÏÏν wie μακÏá½¹Ï trotz μήκιÏÏÎ¿Ï Î¼á¿ÎºÎ¿Ï (â¦).
In Schwyzerâs view, the case of κÏá½³ÏÏÏν : κÏá½±ÏιÏÏÎ¿Ï proves an original difference in root vocalism between the PIE comparative and superlative. In order to maintain this, it must be assumed that some superlatives acquired the e-grade root of the comparative at an early date.31 To this, one may object that the reconstruction of a zero grade root in the superlative is based mainly on κÏá½³ÏÏÏν : κÏá½±ÏιÏÏοÏ. In Vedic, there is no clear-cut evidence for an ablaut difference between comparative and superlative: both formations regularly have a full grade root.32 Moreover, the superlative κÏá½±ÏιÏÏÎ¿Ï (from earlier *kretistoâ) may in fact have taken over the root vocalism of the positive κÏαÏá½»Ï, while the comparative κÏá½³ÏÏÏν remained untouched by this development.33 Thus, the only direct piece of evidence for the alleged difference in root ablaut between the PIE comparative and superlative falls away.
Apart from κÏá½³ÏÏÏν : κÏá½±ÏιÏÏοÏ, Greek has only one other clear instance of an ablaut difference between primary comparative and superlative: μήκιÏÏÎ¿Ï âlongestâ beside μακÏá½¹Ï, comp. μάÏÏÏν.34 The root vocalism of μάÏÏÏν may well have been influenced by the positive μακÏá½¹Ï, replacing an older *mÄki̯osâ or *mÄkiosâ (or its outcome).35 The superlative μήκιÏÏοÏ, on the other hand, is suggestive evidence for an original e-grade root in this formation.36
It is much more attractive, then, to reconstruct an e-grade root for both the comparative and the superlative paradigm. This not only directly explains the Vedic forms, but also accounts for isolated Greek cases like κεÏδίÏν âmore profitable, betterâ, κέÏδιÏÏÎ¿Ï âmost craftyâ and ÏλείÏν âmoreâ, Ïλεá¿ÏÏÎ¿Ï âmostâ < *pléh1-iosâ, *pléh1-ist(H)oâ (beside ÏÎ¿Î»á½»Ï âmany, muchâ). These e-grade formations were preserved because they were not (or could not be) influenced by a positive with different root vocalism. The case of κεÏδίÏν, κέÏδιÏÏÎ¿Ï is telling: the older zero-grade root is preserved only in the non-Ionic-Attic form κοÏδύÏÎ ÏανοῦÏÎ³Î¿Ï âwicked; cunningâ (Hsch.), while Ionic-Attic only has κεÏÎ´Î±Î»á½³Î¿Ï âwily; profitableâ (Hom.+), with a secondary Caland suffix and an e-grade that was probably taken from κέÏδοÏ. These examples corroborate the idea that zero grade root vocalism normally spread from the positive to the forms of comparison (cf. Î²Î±Î¸á½»Ï âdeepâ, superl. βάθιÏÏοÏ, as opposed to the archaism Î²á½³Î½Î¸Î¿Ï âdeep placeâ).
A remaining issue concerns the possibility that the PIE comparative had paradigmatic root ablaut. This paradigm is often reconstructed with an accented, non-ablauting e-grade root, and its weak stem is supposed to underlie the superlative formation.37 This matches the situation in Indo-Iranian, e.g. Ved. ugráâ âstrongâ, comp. ójÄ«yasâ, superl. ójiá¹£á¹haâ, or yúvanâ âyoungâ, superl. yáviá¹£á¹haâ. Root accentuation in the comparative is also required for a pre-stage of Germanic, as indicated by the preserved reflexes of Vernerâs Law in Goth. juggs âyoungâ ~ comp. juhiza, from PGmc. *jungáâ, *júnh-izâ.
As for suffixal ablaut, it is common to reconstruct the oldest paradigm as having qualitative ablaut: *CéC-i̯osâ versus C(e)C-i̯ésâ (and, in Beekesâ view, also *C(e)C-isâ). Indeed, an e-grade suffix somewhere in the paradigm would help to explain Lithuanian comparatives of the type gerèsnis âbetterâ (to gẽras âgoodâ).38 Another argument is Lat. maiestÄs âpowerâ, which seems based on a Proto-Italic comparative stem *mag-i̯esâ. Finally, the Skt. comparative suffix âÄ«yasâ does not show the effects of Brugmannâs Law and may therefore have to be traced back to *âi̯esâ as well.39 Still, whatever the exact reconstruction of the PIE paradigm, the Greek comparatives are understood best from a (post-PIE) paradigm nom. *CéC-i̯Ås, acc. *CéC-i̯os-m, gen. *CéC-is-os,40 because only in this way do the distributions discussed above receive an account.41
In conclusion, the vocalism of Greek primary comparative and superlative formations is explained most economically on the assumption that both had a non-ablauting e-grade root in (early) Proto-Greek. In most adjectives, the zero grade reflex of the positive subsequently ousted this e-grade, which was retained only in a number of relic forms.
4.1.3 The s-stem Nouns and Adjectives
As is well known, Schindler (1975) argued that neuter s-stem nouns originally had proterodynamic inflection in pre-PIE, i.e. a strong stem *CéC-s beside a weak stem *CC-ésâ. He also sketched a way to derive the standard paradigm to be reconstructed for PIE (nom. *CéC-os, gen. *CéC-es-os) from this earlier paradigm. In the late proto-language, the full grade root would have been generalized in most individual s-stem neuters, and the root accent was also generalized. Following this reasoning, Stüber (2002: 19) concluded that âfür die Grundsprache ein intakter Wurzelablaut angenommen werden mussâ. Her main additional arguments for this claim are inherited s-stems with a zero-grade root that are attested in more than one daughter language (e.g. á¿¥á¿Î³Î¿Ï âshiverâ beside Lat. frÄ«gus âcoldâ), and the word for âmouthâ in Hittite.
Within Greek, however, there is no direct proof of root ablaut in the neuter s-stems. Important observations on this issue have been made by Meissner (2006). For instance, it has been argued since Brugmann (for references see Meissner 2006: 72) that the coexistence of s-stem neuters like Ïá½±Î¸Î¿Ï âexperienceâ and Ïá½³Î½Î¸Î¿Ï âsuffering, griefâ must reflect a PIE paradigm with root ablaut.42 However, Meissner convincingly shows (2006: 65â68) how post-Homeric Î²á½±Î¸Î¿Ï âdepthâ replaces Homeric βένθοÏ, and how Ïá½±Î¸Î¿Ï starts to appear at the side of the relic form ÏένθοÏ. His chronological observations are strengthened by his semantic analysis of the forms: ÏάθοÏ, derived from the aorist Ïαθεá¿Î½ (aided by s-stem adjectives like Hom. αἰνοÏÎ±Î¸á½µÏ âwho has experienced terrible thingsâ), has the same broad range of meanings as the verb, whereas Ïá½³Î½Î¸Î¿Ï only means âsufferingâ; and Î²á½±Î¸Î¿Ï functions as a deadjectival abstract to βαθύÏ, whereas Î²á½³Î½Î¸Î¿Ï is a noun with concrete referents. Since Ïá½±Î¸Î¿Ï and Î²á½±Î¸Î¿Ï are secondary creations, Ïá½³Î½Î¸Î¿Ï and Î²á½³Î½Î¸Î¿Ï may simply reflect a Proto-Greek paradigm with a non-ablauting root.
In similar cases, it must be borne in mind that neuter s-stems could be synchronically derived from verbal roots throughout Greek prehistory. For instance, Stüber (2002: 199â200) discusses the etymology that derives κá¿Î´Î¿Ï âworry; thing to take care ofâ (Hom.+) with Goth. hatis âhateâ and Welsh cawdd ârage, griefâ from an inherited root-ablauting neuter *kéh2d-os, *kh2d-ésâ. This analysis is unfounded because the Greek form can be derived from the verb κήδομαι âto mourn; take care ofâ synchronically: such a derivation dispenses us from the task of accounting for the semantic and morphological divergence between κá¿Î´Î¿Ï and the alleged cognate formations. A similar explanation can be given for the zero grade root in á¿¥á¿Î³Î¿Ï, which stands beside an intransitive verb á¿¥á¿Î³á½³Ï, pf. á¼ÏÏιγα âto shudder atâ.
Stüber (2002: 199â200) discusses the case of κá¿Î´Î¿Ï together with two other examples: Indo-Iranian *u̯árH-asâ âwidthâ beside *úrH-asâ âbreastâ, and the word for âmouthâ (Hitt. aiÅ¡, gen. sg. iÅ¡Å¡aÅ¡, Ved. ins. sg. ÄsÄÌ, Lat. Ås, OIr. á). But: âDamit ist allerdings die Zahl derjenigen neutralen s-Stämme, für die Wurzelablaut gesichert ist, auch schon erschöpftâ.43 Moreover, even if *úrH-asâ âbreastâ evidently has a claim to antiquity, it would be difficult to exclude that the form *Hu̯árH-asâ arose beside the adj. *HurH-úâ as a productive adjectival abstract within Proto-Indo-Iranian. Finally, the reconstruction of the Anatolian word for âmouthâ is beset with difficulties (see EDHIL s.v. aiÅ¡). It is therefore likely that PIE (that is, nuclear PIE at least) had already eliminated most, if not all traces of root ablaut in the s-stem neuters.
In PIE, possessive s-stem compounds could be regularly derived from s-stem neuters. Well-known examples such as εá½-μενήÏ, Î´Ï Ï-Î¼ÎµÎ½á½µÏ ~ μένοÏ, beside Ved. su-mánasâ, dur-mánasâ ~ mánas show that this procedure was inherited. Another example from Greek is ÏÎ¿Î»Ï -ÏÎµÎ½Î¸á½µÏ beside Ïá½³Î½Î¸Î¿Ï âsufferingâ (both Hom.). The comparative evidence suggests that the s-stem compound had a non-ablauting e-grade root, just like the simplex. But there are also compounds with a zero grade root, e.g. Hom. αἰνοÏÎ±Î¸á½µÏ âwho has suffered terrible thingsâ. As Tucker (1990), Meissner (2006) and recently Blanc (2018) have shown in detail, the derivation of s-stem compounds from intransitive verbs was productive in Greek. In such derivations, the second compound member naturally took the vocalism of the synchronic verbal stem: thus, αἰνοÏÎ±Î¸á½µÏ was derived from the aor. Ïαθεá¿Î½ âto experienceâ. This is also the origin of forms with zero grade vocalism in the simplex: Ïá½±Î¸Î¿Ï originated as a backformation from compounds like αἰνοÏÎ±Î¸á½µÏ (Meissner 2006: 88). There is no reason, then, to reconstruct root ablaut for PIE s-stem adjectives either.
4.2 Analogical Reshaping and Re-derivation
Having reviewed the ablaut schemes to be reconstructed for the relevant formations, we may now embark on a more detailed discussion of forms that have a bearing on the development of the syllabic liquids. The main issue to be resolved in the remaining part of this chapter concerns the outcome of *rÌ¥ in the u-stem adjectives (sections 4.3 to 4.5). But first, I will illustrate in more detail how a-vocalism of the root spread from these adjectives through entire derivational systems (section 4.2.1), discuss examples of the general tendency of Homeric Greek to avoid u-stem adjectives and replace them with new formations (section 4.2.2), and comment on the origin of factitive verbs of the type θαÏÏá½»Î½Ï (section 4.2.3). The main purpose of these preparatory discussions is to show that many apparent zero grade forms cannot be used to determine the regular outcome of liquid vocalization. Furthermore, it appears that alternations between âαÏâ and âÏαâ in formations derived from the same root are never random reshufflings: usually a precise model can be indicated for newly formed derivatives.
4.2.1 The Spread of a-vocalism across âCalandâ System Formations
As we have already seen, not every form with âαÏâ or âÏαâ can be used as evidence for the outcome of *rÌ¥, because many of them contain a generalized a-vowel. The question to be answered in this section is how exactly the a-vocalism started to proliferate in Ionic-Attic. Two important examples are Aeolic κÏá½³ÏÎ¿Ï (Alc.), which was replaced by κÏá½±ÏÎ¿Ï (Hom.+), and θέÏÏÎ¿Ï (Alc. fr. 206.2), which was replaced by θάÏÏÎ¿Ï (Hom.+). The same replacement took place in the derivationally connected s-stem compounds and âstativeâ verbs in âá½³Ï (Tucker 1990: 54). See Table 3.
Table 3
Replacement of e-grade âCalandâ forms in Ionic-Attic
|
Original eâgrade |
Replaced with aâvocalism |
|---|---|
|
κÏá½³ÏÎ¿Ï (Alc.) |
κÏá½±ÏÎ¿Ï (Hom.+) |
|
-κÏá½³ÏÎ·Ï in names (Aeol., Arc.-Cypr.) |
á¼ÏικÏαÏá½³ÏÏ, á¼ÎºÏαÏá½µÏ; Î Î¿Î»Ï ÎºÏá½±ÏηÏ, etc. (Ion.-Att., Hom.+) |
|
κÏá½³ÏηÏαι (Sapph. fr. 20.5, Alc. fr. 351) |
κÏαÏá½³Ï (Hom.+) |
|
θέÏÏÎ¿Ï (Alc. fr. 206.2) |
θάÏÏÎ¿Ï (Hom.+) |
|
á¼Î»Î¹Î¸á½³ÏÏηÏ, Î Î¿Î»Ï Î¸ÎµÏÏεá¿Î´Î·Ï (Hom.) |
ÏÎ¿Î»Ï Î¸Î±ÏÏá½µÏ (Hom.+) |
|
θέÏÏειÏâ (Theoc. 28.3) |
θαÏÏá½³Ï (Hom.+) |
It is generally agreed that the root vocalism of s-stem neuters was influenced by that of u-stem adjectives. Meissnerâs comment on this replacement deserves to be quoted in its entirety (Meissner 2006: 71):
⦠of all words with full grade, only Ïá½³Î½Î¸Î¿Ï really remains in use while κÏá½³ÏοÏ, θέÏÏοÏ, and Î²á½³Î½Î¸Î¿Ï seem to have disappeared from common Attic-Ionic usage at a very early stage, being replaced by the zero grade forms. The first consequence of this secondary emergence of the zero grade forms is that these cannot be considered reflexes of an old paradigmatic ablaut variation in the root. The motivation for this replacement is not hard to find. κÏá½³ÏοÏ, θέÏÏοÏ, and Î²á½³Î½Î¸Î¿Ï are all abstract nouns and correspond to the u-stem adjectives κÏαÏá½»Ï, θÏαÏá½»Ï, and Î²Î±Î¸á½»Ï that have generalized (in the positive) the zero grade. These adjectives can be conceived as the more âbasicâ form and it is easy to accept Rischâs suggestion that the full grade was eliminated in favour of the zero grade under the pressure of the adjectives. In fact, what we see happening here is only the final stage of this regularization for in a number of cases this change was already complete at the time of our earliest attestations (cf. among others ÏαÏá½»Ï : Ïá½±ÏοÏ, ÏαÏá½»Ï : Ïá½±ÏοÏ). Moreover, the trend is [almost] universally towards the vocalism of the adjective.
Not only may the adjectives be considered as more basic than neuter abstract nouns; it is difficult to indicate another source of the a-vocalism in the relevant s-stem neuters.44 Meissner makes the important observation that Ïá½³Î½Î¸Î¿Ï could be preserved in Homer (and even later) because it was not accompanied by an adjective. As we have seen above, it was eventually replaced by ÏάθοÏ, but this may have happened as late as the tragedians, perhaps as a backformation to compounds in âÏαθήÏ. Another crucial example is κεÏÎ´Î±Î»á½³Î¿Ï âwily; profitableâ (κεÏδίÏν, κέÏδιÏÏοÏ) beside κέÏÎ´Î¿Ï âruse; profitâ, where the u-stem adjective (preserved only in the gloss κοÏδύÏÎ ÏανοῦÏÎ³Î¿Ï âwicked; cunningâ in Hsch.) was apparently eliminated so early from Proto-Ionic that it could not influence the root vocalism of the other forms.
In my view, the zero grade reflex of the u-stem adjective first spread to other adjectival forms (the forms of comparison), and later to the noun. The pair βαθύÏ : βάθιÏÏÎ¿Ï is illustrative. While the pre-form *βένθιÏÏÎ¿Ï of the superlative has apparently been replaced already in Homer, the neuter Î²á½³Î½Î¸Î¿Ï was preserved. Although this may be due to its occurrence in formulaic material (e.g. βένθεÏι λίμνηÏ, see Meissner 2006: 65â66), the absence of Î²á½±Î¸Î¿Ï (first encountered after Homer) is noteworthy. When the vocalism of s-stem neuters like κÏá½±ÏÎ¿Ï had been levelled, the same replacement took place in derived s-stem adjectives (in casu âκÏαÏá½µÏ).45 Indeed, for all roots with an internal liquid, s-stem compounds are attested whenever an s-stem abstract is affected by the replacement: âÏλαÏá½µÏ, âκÏαÏá½µÏ, âθαÏÏá½µÏ, and âÏαÏβήÏ.
The âstativeâ verbs in âá½³Ï, âηÏα (with suffix PIE *âeh1â) appear to behave as primary (underived) formations. They differ from denominative verbs derived from neuters in âοÏ, in that the latter have a different type of aorist in âεÏ(Ï)α (e.g. ÏελέÏ, ÏÎµÎ»Îµá½·Ï âto accomplish, fulfilâ, aor. á¼ÏέλεÏ(Ï)α derived from ÏέλοÏ).46 Originally the âstativeâ verbs in âá½³Ï had a zero grade root, cf. Lat. rubÄre âto be redâ as opposed to the full grade in Gr. á¼ÏÎµá½»Î¸Ï âto make redâ. In Homeric verbs like βαÏá½³Ï (pf. ptc. βεβαÏηόÏâ), θαÏÏá½³Ï, κÏαÏá½³Ï and ÏαÏβέÏ, the root vocalism is equal to that of the accompanying s-stem nouns and adjectives (cf. Tucker 1990: 57â63). However, since these s-stems originally had e-grade, the question arises whether the archaic category of âstativesâ in âá½³Ï may have exerted analogical influence on derivationally associated formations such as s-stem adjectives. This indeed seems plausible,47 and the implication is that âstativeâ verbs in âá½³Ï (just like the adjectives in âá½»Ï) are candidates to show the regular outcome of *rÌ¥.
On the other hand, as appears from Table 3, in Lesbian the leveling seems to have gone the other way: cf. κÏεÏá½³Ï âto have powerâ with the vocalism of Ïὸ κÏá½³ÏÎ¿Ï âpowerâ. In this connection, the root ÏαÏβâ requires further comment. The forms Ïá½±ÏÎ²Î¿Ï âfear, apprehensionâ, á¼ÏαÏÎ²á½µÏ âintrepidâ, and ÏαÏÎ²á½³Ï âto fearâ are usually derived from a PIE verbal root *tergwâ, continued in epic Skt. tarjati âto threatenâ (whose active voice may reflect an oppositional causative), Lat. torvus âgrimâ, and perhaps Hitt. tarkuu̯ant- âlooking angrilyâ.48 Since no e-grade forms of this root are directly attested in Greek, we must ask to what extent ÏαÏβâ can be seen as an analogical vocalization replacing the older e-grade allomorph; for instance, Ïá½±ÏÎ²Î¿Ï might be viewed as a remodeling of older *Ïá½³ÏÎ²Î¿Ï (thus Tucker 1990: 43). However, in this case we must ask in which formation the a-vocalism originated. The adjective attested to this root is ÏαÏÎ²Î±Î»á½³Î¿Ï âfearfulâ (h. Herm. 165, S. Tr. 957), but its formation is most probably secondary after the antonym θαÏÏαλέοÏ.49
One might try to resolve this issue by positing an original adjective *trÌ¥gwâúâ âfearfulâ, which would have yielded *ÏαÏÎ²á½»Ï and then imposed its a-vocalism on the other forms.50 However, this remains speculation, as there is no further evidence for such a form. Moreover, it is quite uncertain that the neuter Ïá½±ÏÎ²Î¿Ï is a primary formation, and that the verb ÏαÏÎ²á½³Ï is derived from it.51 Stüber (2002: 47â48) argues that ÏαÏÎ²á½³Ï is the oldest formation, and that zero grade root was introduced from there into the s-stem noun. In support of this, Meissner (2006: 94) suggests that Ïá½±ÏβοÏ, which in Homer occurs only in Book 24 of the Iliad, may actually be a backformation from á¼ÏαÏÎ²á½µÏ (or ÏαÏβέÏ) because these forms are much more frequent in Homer. Thus, we may suspect that ÏαÏÎ²á½³Ï is the regular outcome of *trÌ¥gw-Äâ, and that e-grade forms of this root had been lost before the vocalization of *rÌ¥.
In sum, the s-stem nouns ÏλάÏοÏ, κÏá½±ÏοÏ, θάÏÏοÏ, Ïá½±ÏÎ²Î¿Ï and their counterparts in second compound members cannot be used as evidence for the regular development of the syllabic liquids, as their vocalism may have been influenced by that of u-stem adjectives (ÏλαÏá½»Ï, κÏαÏá½»Ï, cf. also θÏαÏá½»Ï) or inchoative verbs in âá½³Ï (θαÏÏá½³Ï, ÏαÏβέÏ). These primary and unproductive formations constitute the main body of evidence for the vocalization of *rÌ¥. In addition to u-stem adjectives, there are also productive thematic adjective types in âεÏá½¹Ï and âαλέοÏ. Establishing the derivational prehistory of these forms will put us in a better position to judge their relevance for the vocalization of *rÌ¥.
4.2.2 Replacement of u-stem Adjectives
In Greek, u-stem adjectives are unproductive generally, and in Epic Greek they even seem to be avoided.52 For instance, the inherited form βÏαÏá½»Ï âshortâ is unattested in Homer, who uses ÏμικÏá½¹Ï and á½Î»á½·Î³Î¿Ï instead. Moreover, beside an expected adjective in âá½»Ï or even in place of it, we find adjectives in âεÏá½¹Ï (after light root syllables) or in âÎ±Î»á½³Î¿Ï and âá½±Î»Î¹Î¼Î¿Ï (after heavy root syllables). Many such forms are found in Epic Greek only, and they occasionally penetrated into other poetic genres.
A key factor accounting for the underrepresentation of adjectives in âá½»Ï in Epic Greek is related to meter and verse composition. Let us consider some instances of suffix competition. While κÏαÏá½»Ï appears only in one single name-epithet formula (verse-final κÏαÏá½ºÏ á¼ÏγεÏÏόνÏηÏ), the form κÏαÏεÏá½¹Ï âfierce, vehement, strong, etc.â is extremely frequent. Another well-known pair is Î³Î»Ï Îºá½»Ï âsweet, pleasantâ beside Î³Î»Ï ÎºÎµÏá½¹Ï.53 In Classical prose there is no trace of Î³Î»Ï ÎºÎµÏá½¹Ï; the only current form is Î³Î»Ï Îºá½»Ï. Since the poetic variant Î³Î»Ï ÎºÎµÏá½¹Ï is highly convenient in dactylic rhythms, it may well have an artificial origin within Epic Greek. Indeed, in Homer Î³Î»Ï Îºá½»Ï occurs exclusively in the strong stem Î³Î»Ï Îºá½»â (Î³Î»Ï Îºá½»Ï, Î³Î»Ï Îºá½»Î½, Î³Î»Ï Îºá½»), while Î³Î»Ï ÎºÎµÏá½¹Ï is used in many different cases. This is related in part to the problematic shape of various case forms of Î³Î»Ï Îºá½»Ï: for instance, the feminine Î³Î»Ï ÎºÎµá¿Î± could be used in hexameter verse only with tautosyllabic scansion of muta cum liquida, which is still relatively rare in Homer and was probably avoided to a large degree (see sections 6.5 and 6.6). A more general problem with adjectives in âá½»Ï is the fact that the metrical shape of the feminine forms is different from the masculine/neuter for every single case form. Poets frequently resorted to inflection and/or transformation of phraseological material, and in such cases (for instance when an adjective had to modify a noun with a different gender) it was convenient to keep it in the same metrical position. This means that using adjectives in âεÏá½¹Ï or âÎ±Î»á½³Î¿Ï gave epic poets much more flexibility than using adjectives in âá½»Ï, with their suffix ablaut and metrically different feminine formation.54 Given this metrical incentive, it is likely that Î³Î»Ï ÎºÎµÏá½¹Ï was created analogically within Epic Greek, possibly on the model of κÏαÏá½»Ï : κÏαÏεÏá½¹Ï or Î¸Î±Î»á½»Ï : θαλεÏá½¹Ï.55
As for the adjectives in âαλέοÏ, although their origin remains hard to establish, it is widely accepted that they were productive synchronically beside s-stem nouns.56 In Homer, we find examples like κεÏÎ´Î±Î»á½³Î¿Ï âwilyâ beside κέÏÎ´Î¿Ï âruse; profitâ and the frequent á¼ÏÎ³Î±Î»á½³Î¿Ï âpainful; difficultâ (dissimilated from *algaléoâ) beside á¼Î»Î³Î¿Ï âpain; hardshipâ. Some adjectives in âÎ±Î»á½³Î¿Ï took the place of an original u-stem adjective, or were created in order to supply for its loss. For instance, the gloss κοÏδύÏÎ ÏανοῦÏÎ³Î¿Ï âwicked; cunningâ (Hsch.) is clearly of non-Ionic origin, and an archaism. The lack of a corresponding adjective *καÏÎ´á½»Ï in Ionic-Attic suggests that the u-stem form was lost before κεÏÎ´Î±Î»á½³Î¿Ï was derived from κέÏδοÏ, or at least before *καÏÎ´á½»Ï had the chance to influence the vocalism of the other forms.57
There are three forms in âÎ±Î»á½³Î¿Ï with an apparent zero grade reflex: θαÏÏαλέοÏ, ÏαÏÎ²Î±Î»á½³Î¿Ï and á¼ÏÏαλέοÏ. We have already encountered ÏαÏÎ²Î±Î»á½³Î¿Ï âfearfulâ (h. Herm.) beside ÏαÏÎ²á½³Ï and Ïá½±ÏβοÏ, á¼ÏαÏβήÏ. Most probably, ÏαÏÎ²Î±Î»á½³Î¿Ï is not an old adjective (pace DELG s.v.). Instead, it was derived from Ïá½±ÏβοÏ, and the latterâs vocalism was taken over from the stative verb ÏαÏÎ²á½³Ï and/or the deverbal adjective á¼ÏαÏβήÏ.58 The form á¼ÏÏÎ±Î»á½³Î¿Ï âwith pleasure, eagerâ (3â¯Ã Hom.) is from *u̯alpaleoâ by liquid dissimilation (cf. á¼Î»ÏαλέονΠá¼Î³Î±ÏηÏόν âcherishedâ, Hsch.), containing the root of á¼Î»Ïομαι âto reckon; hope, expectâ. In this case, no neuter abstract is attested from which *u̯alpaleoâ could be derived. However, a-vocalism is also found in the superlative á¼Î»ÏιÏÏÎ¿Ï (A., Pi.) and may stem from a primary adjectival formation, such as a u-stem adjective.59
An important form is θαÏÏÎ±Î»á½³Î¿Ï âpersevering, audacious; confidentâ (Hom.+; Att. θαÏÏαλέοÏ). The vocalism of θαÏÏÎ±Î»á½³Î¿Ï was influenced by that of θάÏÏÎ¿Ï (from which it was derived) after the latter had replaced the older form *θέÏÏοÏ. The question is how the last-mentioned replacement could take place if θαÏÏÎ±Î»á½³Î¿Ï did not yet exist. An archaic adjective formation is θÏαÏá½»Ï âbold, recklessâ, but this has a different root shape θÏαÏâ. As I will argue below, it is likely that another form *θαÏÏá½»Ï âdaring, confidentâ once existed in Proto-Ionic, and that this form influenced the vocalism of θάÏÏÎ¿Ï before it lost currency and was ousted by θαÏÏαλέοÏ.60
In sum, the evidence suggests that the vocalism of forms in âÎ±Î»á½³Î¿Ï was adopted from their base forms: ÏαÏβέÏ, θάÏÏÎ¿Ï (θαÏÏá½³Ï), and κέÏÎ´Î¿Ï (in the case of á¼ÏÏαλέοÏ, the base form is unknown). It is therefore not possible to use the vocalism of ÏαÏβαλέοÏ, á¼ÏÏÎ±Î»á½³Î¿Ï and θαÏÏÎ±Î»á½³Î¿Ï as evidence for the regular outcome of liquid vocalization.
4.2.3 Derivational History of the Factitives in âύνÏ
Homer has a remarkable asymmetry in root shape between the stative verb κÏαÏá½³Ï âto be mightyâ (with the vocalism of κÏá½±ÏοÏ, âκÏαÏá½µÏ) and the factitive καÏÏá½»Î½Ï âto make firmâ. Later Ionic prose writers do not have the same asymmetry: they use κÏαÏύνÏ, derived from the original adjective κÏαÏá½»Ï. The same derivation cannot explain καÏÏύνÏ: as far as we know, there was never a by-form καÏÏá½»Ï*. This, in combination with the fact that κÏαÏá½»Î½Ï was metrically inconvenient in Epic Greek (it necessitated tautosyllabic scansion of muta cum liquida), apparently provided the motive for creating καÏÏύνÏ.
However, was it possible to replace âÏαâ with âαÏâ just like that? Most scholars seem to make this assumption.61 I suspect that mere metrical convenience was not a sufficient reason for swapping the liquid and the vowel. The reason is that the occurrence or non-occurrence of doublets often cannot be predicted, as appears from the following examples:
-
κÏαÏÎ±Î¹á½¹Ï âviolentâ was not avoided, nor reshaped to *καÏÏαιόÏ, but simply tolerated (with its aberrant scansion) in the old formula Îοá¿Ïα κÏαÏαιή.
-
The superlative κÏá½±ÏιÏÏοÏ, on the other hand, is avoided in Homer and replaced by κάÏÏιÏÏÎ¿Ï âfiercestâ.62
-
There is no by-form *καÏÏá½»Ï to κÏαÏá½»Ï, and there are no compounds in *âκαÏÏá½µÏ accompanying those in âκÏαÏá½µÏ.
-
The aorist (á¼)κÏá½±ÏηÏα âgained victory/the upper handâ is absent from Homer. Since this form is common in later poetry, and given that other members of the small group of Homeric stative verbs in âá½³Ï are frequent especially in the aorist stem (cf. Tucker 1990: 39), it is natural to suppose that epic poets avoided (á¼)κÏá½±ÏηÏα for metrical reasons. They never created an alternative form κάÏÏηÏα*, even if this would have been metrically useful.63
Apparently, simply replacing âÏαâ with âαÏâ was not always a viable option. My working hypothesis is that by-forms with âÏαâ or âαÏâ could be created only if they were the product of an inner-epic proportional analogy or derivation. In other words, in order to account for καÏÏá½»Î½Ï and κάÏÏιÏÏÎ¿Ï we must determine a model and a motive.64
Given that καÏÏá½»Î½Ï cannot be derived directly from the adjective κÏαÏá½»Ï, we have to ask whether the derivation of verbs in âá½»Î½Ï from neuter abstracts was already productive in Homer. Tucker (1981) discusses the spread of the Greek factitive verbs in âá½¹Ï, âύνÏ, and âαίνÏ.65 Among the factitive verbs in âá½»Î½Ï she distinguishes three types according to the base form:
-
based on u-stem adjectives (βαÏá½»Ï âheavyâ â Hom. βαÏá½»Î½Ï âto weigh down onâ)
-
based on s-stem nouns (μá¿ÎºÎ¿Ï âlengthâ â Att. Î¼Î·Îºá½»Î½Ï âto lengthenâ)
-
based on o-stem adjectives (λεÏÏá½¹Ï âthin, delicateâ â Att. λεÏÏá½»Î½Ï âto make thinâ).
As the chronology of the attestations confirms, the âá½»Î½Ï factitives originated beside u-stem adjectives (cf. also Hom. βαθύνÏ, ἰθύνÏ). Therefore, types (2) and (3) are generally considered to be later derivational patterns.
Derivation type (2), factitives in âá½»Î½Ï beside s-stem neuters, is clearly productive in Attic. Tucker (1990) argues that this type started to become productive already in Homer. She remarks (1990: 47) that in Homer there are âtwo âá½»Î½Ï verbs for which the only clear connection is with s-stem nominal formsâ: á¼Î½Ïá½»Î½Ï beside Ïá½° á¼Î½Ïεα, and á¼Î»ÎµÎ³á½»Î½Ï beside á¼Î»ÎµÎ³ÎµÎ¹Î½á½¹Ï, Î´Ï Ï-ηλεγήÏ. She also points to θαÏÏá½»Î½Ï and καÏÏá½»Î½Ï as further possible Homeric examples of derivations from s-stems, as the adjectives θÏαÏá½»Ï and κÏαÏá½»Ï have a different vowel slot, while the correct vowel slot is found in the corresponding s-stem neuters θάÏÏÎ¿Ï and Hom. κάÏÏοÏ. If it is possible to derive θαÏÏá½»Î½Ï and καÏÏá½»Î½Ï from these s-stem nouns, the distribution of âÏαâ and âαÏâ over the attested forms would indeed make sense.66
Although this is definitely an improvement over the view that the interchange between âÏαâ and âαÏâ is randomly induced by metrical utility, there are serious problems with Tuckerâs concrete suggestions. First of all, the evidence for derivation (2) as early as Homer is not clear-cut: á¼Î½Ïá½»Î½Ï âto prepare (a meal)â cannot have been reanalyzed as derived from Ïá½° á¼Î½Ïεα, which is a lexicalized form with a concrete meaning âgear, tools, equipmentâ, especially âarmsâ.67 The second example á¼Î»ÎµÎ³á½»Î½Ï âto attend a mealâ cannot count as evidence either, because á¼Î»ÎµÎ³ÎµÎ¹Î½á½¹Ï and Î´Ï Ï-Î·Î»ÎµÎ³á½µÏ have a markedly different meaning, âhard to deal withâ. It is more likely that á¼Î»ÎµÎ³á½»Î½Ï is a contamination between á¼Î»á½³Î³Ï âto take care of, attend toâ and á¼Î½ÏύνÏ, á¼ÏÏá½»Î½Ï âto prepare a mealâ.68 The main problem is that Tucker is unable to point out a convincing pivotal form, i.e. a verb in âá½»Î½Ï beside an s-stem neuter and a u-stem adjective. Her best example is Ïὸ εá½ÏÎ¿Ï âbreadthâ beside εá½Ïá½»Ï and εá½ÏύνÏ, where the idea seems to be that εá½Ïá½»Î½Ï was originally derived from εá½Ïá½»Ï, but secondarily reanalyzed as derived from εá½ÏοÏ. The problem is that εá½ÏÎ¿Ï occurs only once in Homer (Od. 11.312), whereas εá½Ïá½»Ï is frequent. It is questionable whether a transparent derivation εá½Ïá½»Ï â εá½Ïá½»Î½Ï could fall into disuse as long as εá½Ïá½»Ï existed.
Notwithstanding these issues, Tucker is right to emphasize that καÏÏá½»Î½Ï was not derived directly from κÏαÏá½»Ï (because the latter has a different vowel slot), but from κάÏÏοÏ. Given the semantic proximity of θάÏÏÎ¿Ï and κάÏÏοÏ, the pair θαÏÏá½»Î½Ï : θάÏÏÎ¿Ï would provide an excellent model for the derivation of καÏÏύνÏ. But what about the pair θαÏÏá½»Î½Ï : θάÏÏÎ¿Ï itself? A priori, one expects a secondary association of s-stem nouns and âá½»Î½Ï verbs to have started in one or two (preferably frequent) cases where an original u-stem adjective has been lost or replaced by a different form. Subsequently, a derivational relation between a neuter abstract noun âX-nessâ beside a factitive in âá½»Î½Ï âto provide with X-nessâ could easily be established. The root θαÏÏâ would be an excellent candidate for this reanalysis, because the base form *thrÌ¥súâ had turned into θÏαÏá½»Ï, with the wrong vowel slot. Moreover, there are independent indications that θαÏÏÎ±Î»á½³Î¿Ï ousted an older form *θαÏÏá½»Ï (see section 4.5). Finally, the derivation θάÏÏÎ¿Ï â θαÏÏá½»Î½Ï âto encourage, reassureâ is transparent in Homeric Greek, and θαÏÏá½»Î½Ï is frequent and semantically close to καÏÏύνÏ.
In sum: the later, Classical derivational pattern μá¿ÎºÎ¿Ï â Î¼Î·Îºá½»Î½Ï has not yet acquired full productivity in Homeric Greek, but it is already present in an embryonic stage in the pair θάÏÏÎ¿Ï : θαÏÏύνÏ, combined with the absence of an adjective *θαÏÏá½»Ï.
4.3 Reflexes of *r̥ and *l̥ in the u-stem Adjectives
The following u-stem adjectives with a root shape CLaCâ are attested in Homeric Greek and/or Classical Ionic-Attic: βÏÎ±Î´á½»Ï âslowâ, βÏαÏá½»Ï âshortâ, θÏαÏá½»Ï âboldâ, κÏαÏá½»Ï âfirmâ (vel sim.), and ÏλαÏá½»Ï in its distinct meanings âbroadâ and âsaltyâ. In Homer, none of these adjectives is frequent, and βÏαÏá½»Ï is even absent; as we have seen, this tendency can be ascribed to the metrical inconvenience of ablauting paradigms. An adjective Î²Î»Î±Î´á½»Ï âweakâ is frequently cited; it is attested only as a gloss βλαδεá¿Ï (Hsch.), along with a few other glosses with βλαδâ.
In the handbooks, a number of these forms are adduced as evidence for the regular development of the syllabic liquids.69 Given that the evidence is so meagre, it is remarkable that these treatments systematically ignore another u-stem adjective with an original syllabic liquid: ÏαÏÏá½»Ï ânumerousâ. Derived from the same root (that of ÏÏá½³ÏÏ) is the rare adjective ÏÏαÏεÏá½¹Ï âsolid, thickâ (Hom.+), with an interchange that calls to mind the doublet κÏαÏεÏá½¹Ï ~ καÏÏεÏá½¹Ï. In my view, a correct understanding of the origin of ÏαÏÏá½»Ï and ÏÏαÏεÏá½¹Ï is crucial for establishing the regular development of *rÌ¥, and we will therefore turn to these forms first. There are two questions to be answered: Why do we find âαÏâ in ÏαÏÏá½»Ï, as against âÏαâ or âλαâ in all other u-stem forms?70 And: What was the derivational basis for ÏÏαÏεÏá½¹Ï?
4.3.1 *rÌ¥ > Î±Ï is Regular in ÏαÏÏá½»Ï
The Homeric adjective ÏαÏÏá½»Ï âthick, dense, close togetherâ modifies nouns for âsnowflakesâ (e.g. ÏαÏÏειαὶ νιÏá½±Î´ÎµÏ Il. 19.357) and âarrowsâ (e.g. ÏαÏÏá½³ÎµÏ á¼°Î¿á½· Il. 11.387).71 Its acc. pl. n. ÏαÏÏέα is used as a temporal adverb meaning âagain and again, one right after the otherâ, e.g. ÏαÏÏέα Ïε ÏÏÏá½³ÏεÏαι ÏÏá½·ÏÎ±Ï á¼Î½Î´Ïῶν ÏειÏηÏίζÏν, â(the boar) turns round again and again, putting the ranks of men to the testâ (Il. 12.47). Lamberterie (1990: 676â680) gives solid arguments for deriving ÏαÏÏá½³ÎµÏ from the intransitive verb ÏÏá½³Ïομαι (pf. ÏεÏιÏá½³ÏÏοÏε) in its older meaning âto form a layer, become thick, coagulateâ (Hom.+).72 The development of meaning from âthickâ to âfrequentâ is common, too. For instance, English âthickâ may also be used as an adjective or adverb denoting a frequent occurrence, as in thick and fast; in Dutch, dikwijls means âfrequently, oftenâ.
A striking fact about the attested forms and their meanings is that Homer appears to have used ÏαÏÏέεÏ, ÏαÏÏειαί as a plurale tantum. This usage is continued in post-Homeric poetry.73 The singular form ÏαÏÏá½»Ï is attested in literary texts twice in Aeschylus, and much later once in Lucian, in a parody; other Hellenistic poets again use only the plural form.74 Another remarkable feature is the accentuation of the feminine ÏαÏÏειαί, which is paralleled in the synonymous plurale tantum θαμέεÏ, θαμειαί. This accentuation is clearly a retained archaism: presumably, it was not aligned with the productive type of accentuation (βαÏá½»Ï Î²Î±Ïεá¿Î± βαÏá½») because the forms ÏαÏÏειαί and θαμειαί were no longer current in the spoken language.
Most dictionaries (e.g. LSJ) cite another form with the root shape ÏαÏÏâ: the neuter s-stem Ïá½±ÏÏοÏ. However, as Meissner has demonstrated (2006: 110â111), the singular Ïá½±ÏÏÎ¿Ï is only found in ancient grammarians, commentaries and scholia; all real attestations in primary sources are in the plural.75 It is therefore possible to assume that these forms are substantivizations of the u-stem adjective, with a corresponding accent retraction.76 This hypothesis is corroborated, as Meissner remarks, by the parallel case of Ïá½° βÏá½±Ïεα, attested from Thucydides and Herodotus onwards in the lexicalized meaning âshoal, sandbankâ. The absence of contraction of âεα in the Attic form proves that we are dealing with an old u-stem form, with subsequent retraction of the accent accompanying the lexicalization as a substantive.77 Moreover, the expression á¼Î½ Ïá½±ÏÏεÏιν á½Î»Î·Ï âin the thick (= dense parts) of the forestâ (Hom.) has a neat phraseological parallel in á¼Î½ βÏá½±ÏεÏι Î»á½·Î¼Î½Î·Ï âin the shallows (= shallow parts) of the lagoonâ, attested in Hdt. 4.179.
We may conclude that Ïá½±ÏÏÎ¿Ï can be ignored for purposes of reconstruction. This puts us in a better position to judge the origin of âαÏâ in ÏαÏÏá½»Ï. In section 1.4, a number of previous attempts to explain the reflex ÏαÏÏâ were discussed, such as secondary ablaut (KuryÅowicz), or metrically-induced metathesis (Güntert). All such proposals illustrate the embarrassment of earlier scholars concerning the reflex âαÏâ. In reality, in view of the full-grade slot of the root (ÏÏá½³Ïομαι, PIE *dhrebhâ), the outcome ÏαÏÏâ cannot have an analogical origin. That the vowel slot of ÏαÏÏá½»Ï was not aligned with that of ÏÏá½³Ïομαι is not surprising, given the lexicalized nature of this adjective.78 Apparently, it did not undergo the influence of comparative or superlative forms: no such forms are attested for ÏαÏÏá½»Ï, which may well have to do with its lexical semantics.79
We may conclude that ÏαÏÏá½³ÎµÏ (and its feminine ÏαÏÏειαί, with relic accentuation) is the regular outcome of PGr. *thrÌ¥phéu̯es. It is a prime piece of evidence for a regular development *rÌ¥ > âαÏâ in Proto-Ionic.80
4.3.2 Derivation of Hom. ÏÏαÏεÏá½¹Ï
Before the end of the classical period, the adjective ÏÏαÏεÏá½¹Ï is attested only in the formula á¼Ïὶ ÏÏαÏεÏήν Ïε καὶ á½Î³Ïήν (Il. 14.308, Od. 20.98, h. Dem. 43) âboth over the solid land and the waters of the seaâ (Wyatt 1999), literally âover the solid and the liquidâ. After that, ÏÏαÏεÏá½¹Ï first reappears in Hellenistic poetry,81 and Oppian is especially fond of the word in his Halieutica.
Ancient scholia and lexica explain ÏÏαÏεÏá½µ by remarking that θÏá½³Ïαι is another word for Ïá¿Î¾Î±Î¹, which may mean âmake solid or stiff, esp. of liquids: freeze, ⦠curdle, â¦â (LSJ mg. III).82 That the juxtaposed forms á½Î³Ïá½¹Ï and ÏÏαÏεÏá½¹Ï are antonyms is confirmed by the following Homeric simile, which illustrates how Paeëon heals wounded Ares:
Il. 5.902â904á½¡Ï Î´â á½ Ïâ á½Ïá½¸Ï Î³á½±Î»Î± Î»ÎµÏ Îºá½¸Î½ á¼ÏÎµÎ¹Î³á½¹Î¼ÎµÎ½Î¿Ï ÏÏ Î½á½³Ïηξεν,á½Î³Ïὸν á¼á½¹Î½, μάλα δâ ὦκα ÏεÏιÏÏá½³ÏεÏαι ÎºÏ Îºá½¹ÏνÏι,á½£Ï á¼Ïα καÏÏαλίμÏÏ á¼°á½µÏαÏο θοῦÏον á¼Ïηα.
Even as the juice of the fig speedily makes to grow thick the white milk that is liquid, but is quickly curdled as a man stirs it, even so swiftly healed he furious Ares.
tr. Wyatt 1999
This secures the etymological connection between ÏÏá½³Ïομαι, ÏÏαÏεÏá½¹Ï and ÏαÏÏá½»Ï. We now have to explain why âÏαâ is found in ÏÏαÏεÏá½¹Ï, as against âαÏâ in ÏαÏÏá½»Ï. What was the model for creating ÏÏαÏεÏá½¹Ï? At first sight, the most logical option would be a proportional analogy with the u-stem adjective, given the existence of other similar pairs: κÏαÏεÏá½¹Ï beside κÏαÏá½»Ï, θαλεÏá½¹Ï beside (*)θαλύÏ, and Î³Î»Ï ÎºÎµÏá½¹Ï beside Î³Î»Ï Îºá½»Ï. However, the shape of the u-stem adjective is ÏαÏÏá½»Ï, not *ÏÏαÏá½»Ï, and this means that the model breaks down. Even if one were to assume a prolonged retention of root ablaut in u-stem adjectives, it would not be feasible to argue that the root allomorphs ÏαÏÏâ and ÏÏαÏâ once coexisted within the same paradigm.
Fortunately, an alternative base form for the creation of ÏÏαÏεÏá½¹Ï can be pointed out: the verbal stem. A number of adjectives in âεÏá½¹Ï pair with primary verbs: apart from κÏαÏεÏá½¹Ï âstrongâ beside κÏαÏá½³Ï âto be strongâ, cf. the Homeric cases ÏÏÏ Î³ÎµÏá½¹Ï âhorribleâ beside ÏÏÏ Î³á½³Ï âto abhorâ, and θαλεÏá½¹Ï âabundantâ beside Î¸á½±Î»Î»Ï âto be abundantâ. Remarkably, after Homer we find a couple of cases where âεÏá½¹Ï pairs with an inagentive aorist in âá¿Î½Î±Î¹: e.g. ÏακεÏá½¹Ï âsoft, tenderâ (Alcm.+) from Ïήκομαι, Ïακá¿Î½Î±Î¹ âmeltâ, βλαβεÏá½¹Ï âdamagingâ (Hes.+) from βλάÏÏÏ, βλαβá¿Î½Î±Î¹ âhinder, damageâ, ÏανεÏá½¹Ï âclear, evidentâ (Pi.+) from Ïαίνομαι, Ïανá¿Î½Î±Î¹ âappearâ, and ÏÏαλεÏá½¹Ï âthat makes one stumbleâ (A.+) from ÏÏάλλÏ, ÏÏαλá¿Î½Î±Î¹ â(make) stumbleâ (Hom.+). Apparently, the adjectival suffix âεÏá½¹Ï could be added to the verbal root (in its weak form, when available) with some productivity.
The derivation of ÏÏαÏεÏá½¹Ï from ÏÏá½³Ïομαι âto curdleâ fits well in this series, as the verb has an old intransitive aorist á¼ÏÏá½±Ïην. From a semantic perspective, too, this derivation of ÏÏαÏεÏá½¹Ï âsolidâ works better than a connection with ÏαÏÏέεÏ: the verb ÏÏá½³Ïομαι actually has the meanings âto become solid, form a crustâ, while ÏαÏÏá½³ÎµÏ had probably lexicalized its metaphorical meaning âthickâ > âfrequent, in large numbersâ early on (before the vocalization of *rÌ¥).
In conclusion, while ÏαÏÏέεÏ, ÏαÏÏειαί contains a precious vestige of the regular development of *rÌ¥, the adjective ÏÏαÏεÏá½¹Ï has a different root shape because it was derived from the verb ÏÏá½³Ïομαι, á¼ÏÏá½±Ïην at a later time.
4.3.3 Analogical Root Vocalism in the Structure *CraCuâ
If ÏαÏÏá½³ÎµÏ provides compelling evidence for *rÌ¥ > αÏ, we are left with the other u-stem adjectives. How to explain the vowel slot of βÏαδύÏ, βÏαÏá½»Ï, κÏαÏá½»Ï, ÏλαÏá½»Ï, θÏαÏá½»Ï (beside θέÏÏοÏ) and βλαδεá¿Ï (beside μέλδομαι)? While the more problematic cases θÏαÏá½»Ï and βλαδεá¿Ï will be dealt with later, the vocalization in βÏαδύÏ, βÏαÏá½»Ï, κÏαÏá½»Ï, and ÏλαÏá½»Ï can be analogical after full grade forms. Before discussing the evidence for these forms and their cognates, let us pause and ask in which ways such analogical influence may have taken place.
As we have seen above, there is some evidence for the reconstruction of proterodynamic root ablaut in PIE u-stem adjectives. The retention of âÏâ in δαÏá½»Ï gives reason to assume that this ablaut remained intact as late as Proto-Greek. In Van Beek 2013, I assumed that it was even preserved as late as Proto-Ionic, and that the outcome âÏαâ in βÏαÏá½»Ï and κÏαÏá½»Ï is due to inner-paradigmatic levelling of the vowel slot (*markheu̯â >> *mrakheu̯â after *mrekhuâ). While this assumption gives us some leeway in explaining the vocalized zero grades, it is a rather costly assumption in the absence of further positive evidence, and especially when viewed against the general trend in Greek to eliminate inner-paradigmatic root ablaut. Moreover, if we assume that βÏαÏá½»Ï and κÏαÏá½»Ï underwent reshaping, we also have to explain why θÏαÏá½»Ï and βλαδεá¿Ï escaped this alleged influence of the strong stem.
As an alternative, one could assume that the full grade root attested in other formations (e.g. the neuter abstract) influenced the vocalization of the adjective. For instance, κÏαÏá½»Ï (*krÉtus) would have supplanted *καÏÏá½»Ï (*kÉrtus), with the regular vowel slot, under the influence of κÏá½³ÏÎ¿Ï as attested in Lesbian. However, we must take into account that the root vocalism of the adjective usually spreads to derived formations (hence, κÏá½³ÏÎ¿Ï was replaced by κÏá½±ÏοÏ). Moreover, θÏαÏá½»Ï did not undergo the influence of θέÏÏοÏ.
For these reasons, I prefer to ascribe the analogical reshuffling of the root vowel to the forms of comparison. Indeed, most adjectives stand in a close relation with their forms of comparison.83 It is straightforward to assume that the adjective *mrÌ¥khuâ was vocalized as *mrÉkhuâ rather than *mÉrkhuâ because the comparative and superlative were originally *mrékhi̯osâ (or its outcome) and *mrékhistoâ. Given the lexical meaning âshortâ, the superlative would be frequent enough to exert such influence. In the case of *krÌ¥tuâ there is actual evidence that the comparative and superlative retained the original root shape *kretâ longer (κÏá½³ÏÏÏν, κÏá½±ÏιÏÏÎ¿Ï << *krétistos; see chapter 5 for further discussion).
With this in mind, let us now discuss in more detail the development of several individual adjectives in âá½»Ï, together with their cognate formations.
κÏαÏá½»Ï is attested exclusively in the formula |H κÏαÏá½ºÏ á¼ÏγεÏÏόνÏÎ·Ï (4â¯Ã Hom., 5â¯Ã h. Herm.), which refers to Hermes. Its meaning is therefore somewhat uncertain, but the etymological connection with κÏείÏÏÏν, κÏαÏεÏá½¹Ï (etc.) cannot be doubted. The reflex âÏαâ may have been influenced by the comparative κÏá½³ÏÏÏν âstronger, superiorâ (*krét-i̯osâ). The cognate form κÏαÏεÏá½¹Ï cannot be used as evidence for the development of *rÌ¥ because the variant καÏÏεÏá½¹Ï, which is also the Ionic-Attic prose form, displays the regular reflex. This means that κÏαÏεÏá½¹Ï may well have been influenced by κÏαÏá½»Ï, but not the other way around. More extensive argumentation and discussion of these claims will be provided in chapter 5; for the reflex âÏαâ in Homeric κÏαÏαιόÏ, κÏαÏαιâ, see also section 6.8.3.
βÏαÏá½»Ï is the normal word for âshortâ (of time) in Classical Greek, but it is unattested in Homer.84 The adjective is first attested in its Aeolic form as an adverb βÏá½¹Ïεα n.pl. (Sappho fr. 31.7),85 and from Pindar onwards in its Ionic-Attic and West Greek form. The primary superlative βÏá½±ÏιÏÏÎ¿Ï is attested a number of times in poetry, but in classical prose the forms of comparison have generally been replaced by βÏαÏá½»ÏεÏοÏ, βÏαÏá½»ÏαÏÎ¿Ï (Hdt.+).86 In view of the full grade in Lat. brevis âshortâ, the PIE root was *mreǵhâ.87 As just explained, it is a distinct possibility that a zero grade PGr. *mrÌ¥kh-uâ adapted its vocalization to the root shape *mrekhâ to be reconstructed for the forms of comparison. In Aeolic, βÏοÏâ < mrokhâ has the regular dialectal reflex; note that names like Thess. ÎÏοÏÎ¿Ì (SEG 24: 406.1, ca. 460â450â¯BCE) prove that initial μÏâ was preserved until a relatively late date.88
βÏÎ±Î´á½»Ï âslowâ is rare in Homer but normal in the classical language, both in Attic prose and in poetry.89 Primary forms of comparison are only marginally attested. The comparative βÏάδιον (Hes. Op. 528) is probably analogical, because in an inherited form one would expect to find âζâ < *âdi̯â after a light root syllable. An inherited *βÏάζÏν may underlie the hapax βÏá½±ÏÏÏν (Il. 10.226), if this form acquired its âÏÏâ from the antonym θάÏÏÏν âfasterâ.90 The primary superlative is attested only twice as βάÏδιÏÏÎ¿Ï (Il. 23.310 and 530); after Homer, βÏαδύÏεÏÎ¿Ï and βÏαδύÏαÏÎ¿Ï are generalized. The neuter s-stem βÏá½±Î´Î¿Ï is a nonce formation based on Ïá½±ÏοÏ, and can be left aside for purposes of reconstruction.91
Since βÏÎ±Î´á½»Ï may refer both to physical slowness (in running or racing) and to lack of mental alertness, it probably has a cognate in Baltic: Lith. gurdùs âweak, slow, uncommunicativeâ, Latv. gurÌds âtired, wearyâ, both from *gwrÌ¥d-ú-.92 The reconstructed form is peculiar because it violates the constraint that a single PIE root may not contain two mediae. Nevertheless, given the perfect formal and semantic match between Greek and Baltic and the relic status of u-stem adjectives in Greek, it cannot be doubted that the form is inherited.93 Moreover, since no other adjective denoting physical slowness can be reconstructed for PIE (as far as I am able to discern), it seems probable that *gwrÌ¥d-úâ fulfilled this function.
In view of its isolation, one could be tempted to take βÏÎ±Î´á½»Ï as a key example for the regular vocalization of *rÌ¥. It is difficult, however, to establish the original full grade slot of the root. In Proto-Ionic, a full grade may have been around in the forms of comparison. At first sight, the Homeric superlative βάÏδιÏÏÎ¿Ï could be taken as evidence for *gwerdâ. However, βάÏδιÏÏÎ¿Ï could be an artificial epic creation because, as Chantraine (1958: 24) already remarked, *βÏάδιÏÏÎ¿Ï âne pouvait entrer à aucune place du vers homériqueâ. While metrical utility alone was not a sufficient reason to substitute Î±Ï for Ïα (see section 4.2.3 above), a model for an analogy is available: βάÏδιÏÏÎ¿Ï may have been formed to βÏÎ±Î´á½»Ï on the model of another artificial Homeric superlative, κάÏÏιÏÏÎ¿Ï to κÏαÏá½»Ï. This idea receives support from the fact that both roots are used in the context of horse-racing: κάÏÏÎ¿Ï denotes the stamina or endurance of horses in Od. 3.370 (they are á¼Î»Î±ÏÏá½¹ÏαÏοι θείειν καὶ κάÏÏÎ¿Ï á¼ÏιÏÏοι),94 and the horses of Antilochus are called βάÏδιÏÏοι θείειν (Il. 23.310) by his father Nestor. Finally, if βÏá½±ÏÏÏν indeed replaced *βÏάζÏν, this form probably recovers an earlier *gwred-i̯osâ.95
As for the abstract βÏÎ±Î´Ï Ïá½µÏ (Hom.+), its accented suffix only occurs in four Greek abstracts in âÏÎ·Ï (Pike 2011: 148). Since the s-stem abstract Ïá½±ÏÎ¿Ï may denote both speed and swiftness, whereas the âÏÎ·Ï abstract βÏÎ±Î´Ï Ïá½µÏ is the regular form to refer to slowness (cf. Lamberterie 1989), it is likely that βÏÎ±Î´Ï Ïá½µÏ is older than ÏαÏÏ Ïá½µÏ âswiftnessâ, which has the same accentuation. However, this does not imply that βÏÎ±Î´Ï Ïá½µÏ is the regular outcome of a PGr. *gwrÌ¥du-tÄtâ: it is possible that the form was secondarily re-derived from (or influenced by) its base form βÏαδύÏ.
Thus, none of the forms βÏαδύÏ, βÏá½±ÏÏÏν, βάÏδιÏÏÎ¿Ï provides unambiguous evidence for the regular vocalization of *rÌ¥, because the original full grade slot of the root is not known with certainty. If this was *gwredâ, it may have influenced the outcome of * gwrÌ¥d-úâ.
The adjective ÏλαÏá½»Ï âbroad, extended; flatâ is cited as a prime example of the development of the syllabic liquids in most manuals.96 It is well-attested from Homer onwards, and also attested in Lesbian poetry (ÏλάÏÏ , Alc. fr. 74). The forms of comparison were secondarily rebuilt as ÏλαÏá½»ÏεÏοÏ, âÏαÏοÏ. Related forms attested in Greek are ÏλαÏαμών âflat stone or rockâ (h. Hom. +), ÏλάÏÎ¿Ï âbreadth, width; plane surfaceâ (Cypr. fr. 1.2, Simon., Hdt.+), and adjectives in âÏλαÏá½µÏ (X., Th., Arist.). As will be discussed in chapter 10, âλαâ may well be the regular reflex of *lÌ¥, but in ÏλαÏá½»Ï it could also be explained in the same way as in κÏαÏá½»Ï and βÏαÏá½»Ï, i.e. as an adaptation to the original full grade slot (PIE *pleth2â).97 Though no reflex of this full grade is attested in Greek, it may have been eliminated in the s-stem noun at a relatively recent date. Outside of Greek, the same formation is attested in Ved. práthasâ, Av. fraθahâ âbreadthâ, and OIr. leth âsideâ; cf. also the primary verb Ved. práthate âextendsâ.
In sum, the forms βÏαδύÏ, βÏαÏá½»Ï, κÏαÏá½»Ï, and ÏλαÏá½»Ï can no longer be viewed as compelling evidence for the regular reflex of the syllabic liquid.98 The forms θÏαÏá½»Ï and βλαδεá¿Ï, on the other hand, constitute serious evidence for âÏαâ and âλαâ as the regular vocalizations of the respective syllabic liquids: they cannot have been influenced by cognate full grade forms. There is, however, also evidence for a different reflex of the zero grade: the factitive verbs á¼Î¼Î±Î»Î´á½»Î½Ï âto erode, weakenâ and θαÏÏá½»Î½Ï âto encourageâ. As we have seen, verbs in âá½»Î½Ï were productively derived only from u-stem adjectives until a relatively recent date (section 4.2.3). Therefore, á¼Î¼Î±Î»Î´á½»Î½Ï and θαÏÏá½»Î½Ï seem to imply the earlier existence of adjectives *(á¼)Î¼Î±Î»Î´á½»Ï and *θαÏÏá½»Ï. We have to account for the coexistence of both vocalizations.
4.4 *Î²Î»Î±Î´á½»Ï versus á¼Î¼Î±Î»Î´á½»Î½Ï
Traces of the zero grade reflex βλαδâ < *mlÌ¥dâ are attested only in glosses (Hsch., β 54â59):
-
βλαδάΠá¼ÏÏα, μÏÏá½± AS.99 ὠμά (âuntimely; dull, stupidâ; âraw, uncookedâ)
-
βλάδανΠνÏθÏá¿¶Ï (âslothfulâ)
-
βλαδαÏάΠá¼ÏÏα AS. μÏÏá½±. ὠμά AS
-
βλαδαÏόνΠá¼ÎºÎ»ÎµÎ»Ï μένον, Ïαῦνον (âflaccid, porousâ)
-
βλαδόνΠá¼Î´á½»Î½Î±Ïον (âpowerless, weakâ)
-
βλαδεá¿ÏÎ á¼Î´á½»Î½Î±Ïοι. á¼Î¾ á¼Î´Ï νάÏÏν.100
Thus, an adjective Î²Î»Î±Î´á½»Ï is only attested in the plural form βλαδεá¿Ï.101 The appurtenance of the first three glosses is not straightforward: á½ Î¼á½¹Ï ârawâ means almost the opposite of âweak, softâ, and the meanings contained in the glosses á¼ÏÏα, μÏÏá½± and νÏθÏá¿¶Ï could have developed from âweakâ, but this is not evident. The glosses á¼ÎºÎ»ÎµÎ»Ï μένον, Ïαῦνον and á¼Î´á½»Î½Î±Ïον, on the other hand, can be understood quite well as having developed from âweak, softâ.
The verb á¼Î¼Î±Î»Î´á½»Î½Ï is attested from Homer onwards, but in various different meanings. The philological evidence for this verb has been extensively discussed by Lamberterie (1990). From his discussion, I retain the following conclusions:
-
In Homer, á¼Î¼Î±Î»Î´á½»Î½Ï occurs three times in a similar context. On each occasion, the Achaean wall is reduced to dust, corroded, by the erosion of wind and water. The meaning âto make invisibleâ, found in post-Homeric poetry, is ultimately based on reinterpretations of the Homeric passages. In the Hippocratic Corpus, á¼Î¼Î±Î»Î´á½»Î½Ï means âto weakenâ (vel sim.); sometimes, á¼Î¼Î±Î»Î´á½»Î½Ï is even used as an equivalent of á¼Î¼Î²Î»á½»Î½Ï âto make bluntâ (cf. Lamberterie 1990: 364â368).
-
The á¼â of á¼Î¼Î±Î»Î´á½»Î½Ï may be due to contamination with á¼Î¼Î±Î¸á½»Î½Ï âreduce to sandâ (Hom.+), which may itself owe its factitive suffix âÏ Î½â to á¼Î¼Î±Î»Î´á½»Î½Ï (cf. Lamberterie 1990: 363).102 It is noteworthy that the á¼â of the gloss á¼Î¼á½³Î»Î´ÎµÎ¹Î½Î Ïήκειν. ÏÏεÏá½·Ïκειν âto melt; deprive ofâ (Hsch.) was also secondarily added: cf. μέλδομαι âto become soft by boiling or heatingâ (Il. 21.363, Nic.), from the same root as á¼Î¼Î±Î»Î´á½»Î½Ï.103
-
As Lamberterie (1990: 372â373) shows, the PIE root was *meld- rather than *mledâ in view of Gr. μέλδομαι, PGmc. *(s)meltanâ âto meltâ, Arm. meÉ«k âsoftâ < *meldu̯iâ.104 The full grade of Ved. và mradÄ (RV, hapax) and Å«Ìrá¹a-mradas- âsoft like woolâ is an innovation of Indo-Aryan.105
-
á¼Î¼Î±Î»Î´á½»Î½Ï presupposes the earlier existence of an adjectival stem *(á¼)μαλδύâ âreduced to dustâ (Lamberterie 1990: 364).
A semantic problem must now be taken into consideration. All reflexes of the adjective PIE *mlÌ¥d-úâ carry the meaning âweak, soft, tenderâ, but á¼Î¼Î±Î»Î´á½»Î½Ï means âto corrodeâ. At first sight, then, the meaning of á¼Î¼Î±Î»Î´á½»Î½Ï seems to match that of Vedic mard âto crushâ, which is both etymologically and synchronically distinct from the root mrad âsoftenâ. Lamberterie answers this problem by assuming that the meaning âto reduce to dustâ displayed by the Homeric factitive is old, claiming that it âreflète directement le sens fondamental de la racine *mel(H2)â âbroyer, moudreâ (â¦)â (1990: 364). This forces him to consider the meaning âweak, soft, tenderâ, attested in all branches that have a reflex of the u-stem adjective, as a secondary development from âcrushed, pulverizedâ. It does not seem very likely, however, that this semantic development took place independently in various different branches. Although it is possible that the roots *meld- and *melh1- were identical at a pre-stage of PIE (via the âeffet Kortlandtâ), the meaning of *meldâ âto become weak or softâ was clearly distinct from that of *melh1â âto crushâ in PIE itself.106 We may assume that the meaning of *(á¼)Î¼Î±Î»Î´á½»Ï developed from âweak, softâ to âflaccid, porousâ (as in βλαδαÏόνΠá¼ÎºÎ»ÎµÎ»Ï μένον, Ïαῦνον Hsch. discussed above);107 and from *(á¼)Î¼Î±Î»Î´á½»Ï the factitive verb á¼Î¼Î±Î»Î´á½»Î½Ï âto make porous, corrodeâ could be derived.
It remains to explain the different vocalizations in *(á¼)μαλδύâ and *βλαδύÏ. It would not help to start from an adjective *méld-uâ, *mlÌ¥d-éuâ with root ablaut, for it would be difficult to derive both *μαλδâ and *βλαδâ from it within the same dialect.108 Moreover, it is uncertain whether root ablaut was preserved in adjectives in âá½»Ï until the vocalization of *lÌ¥.
As a way out of this dilemma, one might surmise that the forms with βλαδâ are not from Ionic-Attic, but from a different dialect. The glosses provide no clue about their provenance, but since lexical meanings such as âflabbyâ and âporousâ would be compatible with medical terminology, one could hypothesize that forms with βλαδâ are from the Hippocratic Corpus (a considerable number of treatises belonging to this corpus are known to have been lost). In that case, it may be wondered whether these forms could be of Doric origin: Hippocrates and his pupils lived and worked on the island of Cos. This speculation may receive some support from the adjective ÏλαδαÏá½¹Ï: one of its meanings is âflaccidâ, which is also how βλαδαÏá½¹Ï is glossed, and ÏλαδαÏá½¹Ï is mainly attested in the Hippocratic corpus. It is not unthinkable that ÏλαδαÏá½¹Ï is a secondary reshaping of βλαδαÏá½¹Ï in Ionic, perhaps under the influence of ÏλάÏÏÏ âto kneadâ. In this case, βλαδαÏá½¹Ï could well stem from a different dialect, and the same might then hold for βλαδύÏ.
In conclusion, the adjective *(á¼)Î¼Î±Î»Î´á½»Ï that seems to be presupposed by the factitive verb á¼Î¼Î±Î»Î´á½»Î½Ï âto corrodeâ would be the expected Ionic-Attic continuant of PIE *mlÌ¥d-úâ âweak, flaccidâ, provided that its vocalization was influenced by the full grade (as found e.g. in the forms of comparison and the primary verb μέλδομαι). On the other hand, in glosses with βλαδâ we find a direct reflex of *lÌ¥. The main problem is posed by the gloss βλαδεá¿Ï, which seems to be the plural of an adjective βλαδύÏ: why did this form coexist with *(á¼)μαλδύÏ? There is no obvious explanation, but it is conceivable that one of these forms is of non-Ionic-Attic origin. In any case, βλαδâ reflects a zero grade *mlÌ¥dâ; this conclusion will be bolstered with further arguments in chapter 10.
4.5 θÏαÏá½»Ï versus θαÏÏύνÏ
The adjective θÏαÏá½»Ï âboldâ < *dhrÌ¥s-úâ is attested from Homer onwards, both in poetry and in prose. Given that the root had a full grade θεÏÏâ, it seems a strong counterexample against *rÌ¥ > âαÏâ as the regular Ionic-Attic development. However, θÏαÏá½»Ï is different from other u-stem adjectives with a similar root structure in that its zero grade reflex does not show the influence of the original full grade root, θεÏÏâ. If κÏαÏá½»Ï, βÏαÏá½»Ï, and βÏÎ±Î´á½»Ï are indeed due to leveling, one would expect *dhrÌ¥s-úâ to end up as θαÏÏá½»Ï* under influence of *dhers-. Although some historical grammars cite a form θαÏÏá½»Ï,109 it is not attested as an appellative, nor as a simplex, but only as a first compound member ÎαÏÏÏ -, ÎαÏÏÏ â in personal names; moreover, these names occur in West Greek dialects, not in Ionic-Attic.110 A key question is whether an adjective *θαÏÏá½»Ï indeed existed at some pre-stage of Ionic-Attic.
In the following pages I will therefore consider all derivatives of this root, first in Homer, then in Classical Attic and Ionic. The semantic values of the attested formations play a key role: they may help us establish the historical and synchronic derivational relationships.
4.5.1 The Roots θÏÎ±Ï â and θαÏÏâ in Homer: Attestations
Table 4 (next page) contains all forms containing the root shapes θÏαÏâ and θαÏÏâ as attested in Homeric Greek.
It appears that there are no true doublets in Homer. The only exception is the neuter abstract θάÏÏÎ¿Ï ~ θÏá½±ÏοÏ, but the variant θÏá½±ÏÎ¿Ï may well be a one-off creation, analogical to κÏá½±ÏÎ¿Ï ~ κάÏÏοÏ.111 Leaving aside this hapax θÏá½±ÏοÏ, the root shape θÏαÏâ only occurs in θÏαÏá½»Ï and compounds with θÏαÏÏ -, for which no variant with θαÏÏâ exists. It is therefore not true that âαÏâ could always be substituted for âÏαâ in Homer, or vice versa (pace Lamberterie 1990: 849 and 852).
The allomorph θαÏÏâ can be due to the secondary introduction of a-vocalism in a pre-form with *θεÏÏâ. Thus, the full grade root of θέÏÏοÏ, attested in Alcaeus, has been replaced by θάÏÏÎ¿Ï in Ionic-Attic. Similarly, in ÏÎ¿Î»Ï Î¸Î±ÏÏá½µÏ the second member replaces âθεÏÏá½µÏ, which is preserved in Homer only in the personal names á¼Î»Î¹Î¸á½³ÏÏÎ·Ï and Î Î¿Î»Ï Î¸ÎµÏÏεá¿Î´Î·Ï. The question then remains where θαÏÏâ originated: does it also reflect an older zero grade in some forms, whether by regular sound change or analogical reshaping?
In the first member of compounds (including personal names), the two variants available in poetry were θÏαÏÏ - and θεÏÏι-. They serve as counterparts of both θÏαÏá½»Ï and θαÏÏαλέοÏ, θάÏÏοÏ. While θεÏÏι- is a clear archaism, it seems as if θÏαÏÏ â may have been introduced in compounds at any time. However, since the distinction between these first members was utilized for metrical variation, and since both θεÏÏιâ and the reflex of *thrÌ¥suâ are widespread in epigraphic onomastic material,112 the coexistence of θεÏÏιâ and θÏαÏÏ â (earlier *thrÌ¥suâ) is bound to be old as well.
Table 4
Forms with the root shapes θÏαÏâ and θαÏÏâ in Homer
|
θÏαÏâ |
θαÏÏâ |
|---|---|
|
θÏαÏá½»Ï âdauntless, brave, recklessâ113 |
|
|
θÏαÏÏ Îºá½±ÏÎ´Î¹Î¿Ï âbrave-heartedâ θÏαÏÏ Î¼á½³Î¼Î½Î¿Î½Î± âid.â PNâ¯s ÎÏαÏÏ Î¼á½µÎ´Î·Ï, ÎÏαÏύμηλον |
|
|
θαÏÏá½»Î½Ï âto instill courageâ |
|
|
θάÏÏÏ Î½Î¿Ï âconfidentâ114 |
|
|
θαÏÏÎ±Î»á½³Î¿Ï âdauntless, brave; persevering, audacious; confidentâ |
|
|
ÏÎ¿Î»Ï Î¸Î±ÏÏá½µÏ âdauntlessâ |
|
|
θÏá½±ÏÎ¿Ï âcourageâ (only Il. 14.416) |
θάÏÏÎ¿Ï âperseverance, stamina; courageâ |
|
θαÏÏá½³Ï âto hold on; gain courageâ |
In sum, leaving aside the hapax θÏá½±ÏοÏ, the root allomorph θÏαÏâ is limited in Homer to the adjective θÏαÏá½»Ï and the compounds with first member θÏαÏÏ â (including personal names). This distribution calls for an explanation. In what follows, I argue that there is evidence for a lost adjective *θαÏÏá½»Ï in Proto-Ionic, and suggest that θÏαÏá½»Ï reflects an archaism not of the spoken language, but of the epic tradition. Not only is θÏαÏá½»Ï morphologically isolated, but it is also semantically detached from most forms with θαÏÏâ.
4.5.2 The Roots θÏαÏâ and θαÏÏâ in Homer: Semantics
It is usually thought that there was not yet a tangible semantic or lexical distinction between θαÏÏÎ±Î»á½³Î¿Ï and θÏαÏá½»Ï in Homer. In his dictionary treatment of this etymon, Chantraine (DELG s.v. θάÏÏοÏ) claims that Homeric θÏαÏá½»Ï means âbraveâ as an epithet of Hector and other heroes, âcourageousâ in the phrase Ïόλεμον θÏαÏύν, and âintrepidâ as an epithet of arms that throw spears. A lexical split allegedly first occurs in Classical Attic, which (generally speaking) makes a distinction between θαÏÏÎ±Î»á½³Î¿Ï âconfidentâ and θάÏÏÎ¿Ï âcourage, confidenceâ on the one hand, and θÏαÏá½»Ï âaudacious, recklessâ, θÏá½±ÏÎ¿Ï âarroganceâ on the other. This semantic specialization is thought to be of post-Homeric date and supposed to have developed by the lexicalization of a pragmatic difference between a pejorative sense ârecklessâ and a laudatory meaning âconfident, courageousâ.115
In reality, the Homeric evidence may point in a different direction. In his extensive discussion of the semantics of this root, Lamberterie (1990: 854) shows that θαÏÏαλέοÏ, not θÏαÏá½»Ï, serves as the productive adjectival counterpart of θάÏÏοÏ, θαÏÏá½³Ï, and θαÏÏύνÏ. He compares the following items of Homeric phraseology:
-
θαÏÏαλέον νύ οἱ ἦÏÎ¿Ï á¼Î½á½¶ ÏÏεÏίν (Il. 19.169);
-
θάÏÏÏ Î½Î¿Î½ δέ οἱ ἦÏÎ¿Ï á¼Î½á½¶ ÏÏεÏίν (Il. 16.242, θάÏÏÏ Î½Î¿Î½ = 2sg. impv. aor.);
-
θαÏÏύνονθâ á¼Ïá½±ÏÎ¿Ï Ï ÎºÎ±á½¶ á¼ÏοÏÏύνονÏα μάÏεÏθαι (Il. 13.767; 17.117, 683).
This observation is corroborated by a closer consideration of the two Homeric forms and their semantics. In Homer θαÏÏÎ±Î»á½³Î¿Ï has the same range of meanings as the abstract noun θάÏÏοÏ, from which it was probably derived (cf. section 4.3.2):
(1) âpersistenceâ (whether in a positive sense âstaminaâ, or pejorative âobstinacy, perseverance, audacityâ), cf.
Il. 17.570â573καί οἱ Î¼Ï á½·Î·Ï Î¸á½±ÏÏÎ¿Ï á¼Î½á½¶ ÏÏήθεÏÏιν á¼Î½á¿ÎºÎµÎ½,á¼¥ Ïε καὶ á¼Ïγομένη μάλα ÏÎµÏ ÏÏÎ¿á½¸Ï á¼Î½Î´ÏομέοιοἰÏÏανάᾳ δακέειν, λαÏόν Ïá½³ οἱ αἷμâ á¼Î½Î¸Ïá½½ÏÎ¿Ï ÎÏÎ¿á½·Î¿Ï Î¼Î¹Î½ θάÏÏÎµÏ Ï Ïλá¿Ïε ÏÏá½³Î½Î±Ï á¼Î¼Ïὶ μελαίναÏ
and she [Athena] put into his [Menelaus] heart the perseverance of a mosquito, which even when brushed off keeps trying to bite in the human skin; it likes the taste of human blood; with a similar endurance did she fill him in his dark lungs.
Od. 7.50â52Ïὺ δâ á¼ÏÏ Îºá½·Îµ μηδέ Ïι Î¸Ï Î¼á¿·Ïá½±ÏβειΠθαÏÏÎ±Î»á½³Î¿Ï Î³á½°Ï á¼Î½á½´Ï á¼Î½ ÏᾶÏιν á¼Î¼Îµá½·Î½Ïνá¼ÏγοιÏιν Ïελέθει, εἰ καί Ïοθεν á¼Î»Î»Î¿Î¸ÎµÎ½ á¼Î»Î¸Î¿Î¹.
But do you enter the palace and do not be timid at heart: for a man who perseveres has more success in all matters, even if he comes from somewhere else.
Od. 17.449á½¥Ï ÏÎ¹Ï Î¸Î±ÏÏÎ±Î»á½³Î¿Ï ÎºÎ±á½¶ á¼Î½Î±Î¹Î´á½µÏ á¼ÏÏι ÏÏοá¿ÎºÏηÏ
such an obstinate and shameless beggar you are
(2) âcourage, confidenceâ, cf.:
Od. 6.140Ïá¿ Î³á½°Ï á¼Î¸á½µÎ½Î·Î¸á½±ÏÏÎ¿Ï á¼Î½á½¶ ÏÏεÏὶ θá¿ÎºÎµ καὶ á¼Îº Î´á½³Î¿Ï Îµá¼µÎ»ÎµÏο Î³Ï á½·Ïν.
and in her [Nausicaä] heart Athena put courage and she took fear out of her legs.
Il. 19.169θαÏÏαλέον νύ οἱ ἦÏÎ¿Ï á¼Î½á½¶ ÏÏεÏίν, οá½Î´á½³ Ïι Î³Ï á¿Î±ÏÏὶν κάμνει ÏÏὶν ÏάνÏÎ±Ï á¼ÏÏá¿Ïαι Ïολέμοιο.
The heart in his chest is courageous, and his limbs do not get tired before everyone else has stopped fighting.
On the other hand, in pre-classical poetry θÏαÏá½»Ï usually means âbold, intrepidâ, and this is clearly an older meaning than âcourageous, confidentâ or âaudaciousâ.116 However, already in Homer there are restrictions on the use of θÏαÏá½»Ï. The positive nuance âintrepid, dauntlessâ is retained in compounds and in the archaic formula θÏαÏειάÏν á¼Ïὸ ÏειÏῶν, where it qualifies the persevering arms of warriors (that keep throwing spears). The negative nuance ârecklessâ is predominant when θÏαÏá½»Ï qualifies human beings: Hector (whose recklessness is thematic throughout the Iliad), his charioteers, and Odysseus in a passage where his foolhardy eagerness to confront the Cyclops is criticized by one of his companions.117 Finally, the phrase Ïόλεμον θÏαÏύν (3â¯Ã) is used twice when Helenâs abduction is mentioned as the cause of the Trojan war (cf. also Il. 10.27â28):
Od. 4.145â146á½ Ïâ á¼Î¼Îµá¿Î¿ ÎºÏ Î½á½½ÏÎ¹Î´Î¿Ï Îµá¼µÎ½ÎµÎºâ á¼Ïαιοὶἤλθεθâ á½Ïὸ ΤÏοίην, Ïόλεμον θÏαÏὺν á½ÏμαίνονÏεÏ
when you Achaeans came to the walls of Troy on account of me, bitch-face, waging a stout-hearted/reckless war
Thus, in spite of some potential overlap between θαÏÏÎ±Î»á½³Î¿Ï and θÏαÏá½»Ï in Homer, there are in fact clear differences between the two in meaning and use. While θÏαÏá½»Ï never means âconfident, courageousâ, θαÏÏÎ±Î»á½³Î¿Ï and θαÏÏá½»Î½Ï are readily used in this sense. If the phrase θαÏÏαλέον ⦠ἦÏÎ¿Ï is paralleled by the compound θÏαÏÏ Îºá½±ÏδιοÏ, this is due to the fact that θÏαÏÏ - is still the productive 1st compound member corresponding to θαÏÏÎ±Î»á½³Î¿Ï and θάÏÏοÏ, θαÏÏá½³Ï. Furthermore, θαÏÏαλέοÏ, θάÏÏοÏ, θαÏÏá½³Ï and θαÏÏá½»Î½Ï are frequently opposed to words for fear (Ïá½±ÏβοÏ, δέοÏ; ÏαÏβέÏ, δείδÏ) or restraint (αἰδώÏ), but θÏαÏá½»Ï is never used in such oppositions.118
In sum, the derivational connection in Homer between θαÏÏÎ±Î»á½³Î¿Ï (but not θÏαÏá½»Ï) and θάÏÏοÏ, θαÏÏá½³Ï, θαÏÏá½»Î½Ï suggests that θÏαÏá½»Ï is a poetic archaism. The use of θÏαÏÏ â as a first compound member corresponding to both θαÏÏÎ±Î»á½³Î¿Ï and θÏαÏá½»Ï confirms this. The relic status of θÏαÏá½»Ï and θÏαÏÏ â may explain why their root was not adapted to θαÏÏâ.
On the other hand, θαÏÏÎ±Î»á½³Î¿Ï was probably derived from θάÏÏοÏ. But what caused the replacement *θέÏÏÎ¿Ï >> θάÏÏοÏ? It would be problematic to assume that the vocalism of θÏαÏá½»Ï was responsible, as one expects analogical influence to reduce the number of different root shapes, not to increase them. Therefore, the root vowel of θάÏÏÎ¿Ï must have been introduced from a different form, preferably from an adjective.119 We must now pose the question: is it possible that an earlier stage of Ionic-Attic had an adjective *θαÏÏá½»Ï? This would immediately account for the vocalism of θάÏÏÎ¿Ï and for the vowel slot of the factitive verb θαÏÏá½»Î½Ï in one time (cf. section 4.2.3). Before further discussing this issue, let us first consider the situation in Classical Greek.
4.5.3 The Roots θÏαÏâ and θαÏÏâ in Classical Greek
The attested formations and the distribution of the allomorphs θÏαÏâ and θαÏÏâ in Classical Greek are listed in Table 5 (on the following page). It distinguishes Ionic from Attic, and poetic forms from prose forms.
In Classical Greek, the allomorph θÏαÏâ is no longer limited to the forms θÏαÏá½»Ï and θÏαÏÏ â: we also find the forms of comparison θÏαÏá½»ÏεÏοÏ, θÏαÏá½»ÏαÏοÏ, a denominative verb θÏαÏύνÏ, and an abstract θÏá½±ÏοÏ. Thus, unlike Homeric Greek, the Classical language has the variants θÏαÏá½»Î½Ï ~ θαÏÏá½»Î½Ï (Attic θαÏÏύνÏ) and θÏá½±ÏÎ¿Ï ~ θάÏÏÎ¿Ï (Attic θάÏÏοÏ).120 As we will see in the next chapter, the situation for κÏαÏâ ~ καÏÏâ is exactly the reverse: doublets κÏαÏεÏá½¹Ï ~ καÏÏεÏá½¹Ï and κÏá½±ÏÎ¿Ï ~ κάÏÏÎ¿Ï are found in Epic Greek, but not in Classical prose.
Table 5
θÏαÏâ vs. θαÏÏâ in Classical Greek prose and poetry
|
θÏαÏâ âbold, brave, recklessâ |
θαÏÏâ âcourageous, assertiveâ |
|---|---|
|
θÏαÏá½»Ï âbold, reckless; audaciousâ (poetry and prose) |
|
|
1st CM θÏαÏÏ â, ÎÏαÏÏ â (poetry, Pi.+)121 |
|
|
θÏαÏá½»ÏεÏοÏ, âÏαÏÎ¿Ï (never in poetry) |
|
|
θÏαÏá½»Î½Ï âto emboldenâ (Attic, Hp.) |
θαÏÏá½»Î½Ï âto encourageâ (Ion. prose, Att. poetry, Th.), θαÏÏá½»Î½Ï (X.) |
|
θαÏÏÎ±Î»á½³Î¿Ï âself-assured, assertive, confident; audaciousâ (Attic prose; θαÏÏâ in Th., Hp. and poetry) |
|
|
θÏá½±ÏÎ¿Ï âaudacity, insolenceâ (Attic, e.g. Ar., X., Pl.; never in Ionic) |
θάÏÏÎ¿Ï âcourage, confidenceâ (Pi., trag., Hdt., Th., Pl.), Att. θάÏÏÎ¿Ï (X., Pl.) |
|
ÎºÏ Î½Î¿-θÏαÏá½µÏ âimpudent as a dogâ (A.)122 |
εá½-θαÏÏá½µÏ âcourageousâ (A.) |
|
θαÏÏá½³Ï (Att. θαÏÏá½³Ï) especially in impv. θάÏÏει âhold on!â |
The alternation θÏαÏâ ~ θαÏÏâ in Classical Greek reflects a phenomenon of the spoken language, while that between κÏαÏâ and καÏÏâ in Homer belongs to the artificial language of epic. In Homer the alternation is utilized for metrical purposes, but in Classical Attic the two roots θÏαÏâ and θαÏÏâ are semantically distinct. As is well-known,123 Attic prose generally makes a distinction between θÏá½±ÏÎ¿Ï âaudacity, boldness, recklessnessâ and θάÏÏÎ¿Ï âcourage, (selfâ)confidenceâ, and also between θÏαÏá½»Î½Ï âto emboldenâ and θαÏÏá½»Î½Ï âto encourage, give confidenceâ.124 Generally speaking, this distinction is respected in Classical poetry, too, even if there are some instances where θÏαÏâ is used instead of expected θαÏÏâ, perhaps for metrical reasons.125 It is plausible that θÏá½±ÏÎ¿Ï and θÏαÏá½»Î½Ï were productively created to the old adjective θÏαÏá½»Ï as an adjectival abstract and factitive verb, respectively.
For θÏαÏá½»Ï, the Homeric meaning âbold, daringâ continues to be the normal one in early Classical poetry, also in poetic compounds with θÏαÏÏ â (see Lamberterie 1990: 851). In Classical prose the predominant meaning becomes âaudaciousâ, but we must note, with Huart (1968: 430), that θÏαÏá½»Ï and θÏá½±ÏÎ¿Ï do not have an exclusively pejorative meaning in Thucydides.126 This does not imply, however, that Thucydides made no distinction between θÏαÏá½»Ï and θαÏÏÎ±Î»á½³Î¿Ï (as Huart claims): θÏαÏá½»Ï means âbold; recklessâ as against θαÏÏÎ±Î»á½³Î¿Ï âconfident, self-assuredâ.127
Both Ionic and Attic retain θαÏÏá½»Î½Ï (already Homeric), but we also find θÏαÏύνÏ, based on θÏαÏá½»Ï or on θÏá½±ÏοÏ. It is noteworthy that the âá½³Ï verb only appears in the form θαÏÏá½³Ï âto gain courage; hold onâ; the stative-inchoative verb corresponding to θÏá½±ÏÎ¿Ï was not *θÏαÏá½³Ï, but expressed by means of the middle of the factitive, θÏαÏύνομαι âto be(come) bold or audaciousâ. Thus, the only old verbs are θαÏÏá½»Î½Ï and θαÏÏá½³Ï; θÏαÏá½»Î½Ï is a more recent creation. This means that Homeric θαÏÏá½»Î½Ï is not a metrical replacement of a vernacular form *θÏαÏá½»Î½Ï âto encourageâ, but that it was linguistically real already at an early date. In view of the difference in root vocalism, we may conclude with some confidence that θαÏÏá½»Î½Ï was not directly derived from θÏαÏá½»Ï. The derivation of θαÏÏá½»Î½Ï probably took place before the vocalization of *rÌ¥, and before the ancestor of θαÏÏÎ±Î»á½³Î¿Ï ousted that of θÏαÏá½»Ï.
In sum: in Homer θÏαÏá½»Ï retains its archaic meaning âbold, recklessâ; it did not take part in the semantic development to âcourageous, self-assuredâ that θαÏÏαλέοÏ, θαÏÏá½»Î½Ï and other derivations with θαÏÏâ had already undergone in Homer. In Classical Greek, θÏαÏá½»Ï further specializes in pejorative meanings like âaudacious; arrogant, insolentâ and serves as the basis for new derivations: a factitive verb θÏαÏá½»Î½Ï and an abstract θÏá½±ÏοÏ. See Table 6.
Table 6
The oldest distribution of the root shapes θÏαÏâ and θαÏÏâ
|
Archaic forms and meanings |
Innovative forms and meanings (Homer) |
Innovative forms and meanings (Attic) |
|---|---|---|
|
θÏαÏá½»Ï âbold, daring, recklessâ |
θαÏÏÎ±Î»á½³Î¿Ï âaudacious; courageousâ |
θÏαÏá½»Ï âaudacious, arrogantâ |
|
θάÏÏÎ¿Ï âpersistenceâ |
θάÏÏÎ¿Ï âcourage, confidenceâ |
θÏá½±ÏÎ¿Ï âaudacity, insolenceâ |
|
θαÏÏá½³Ï âpersevereâ |
||
|
θαÏÏá½»Î½Ï âgive courageâ |
θÏαÏá½»Î½Ï âemboldenâ, θÏαÏύνομαι âbe bold, take courageâ |
4.5.4 Reconstruction
Let us now review the arguments for positing an older adjective *θαÏÏá½»Ï.
First of all, the shape of the factitive verb θαÏÏá½»Î½Ï seems to presuppose a base form *θαÏÏá½»Ï for the adjective.128 Against this, both Tucker (1990) and Strunk (1975) have objected that θαÏÏá½»Î½Ï may have been derived from the abstract θάÏÏÎ¿Ï already in Homer. However, we have seen (section 4.2.3) that the basis for this derivation, as early as Homer, is very slim. The main question is: how did the derivational pattern originate which links factitives in âá½»Î½Ï to neuter abstracts? Homeric καÏÏá½»Î½Ï was derived from κάÏÏÎ¿Ï (a form *καÏÏá½»Ï never existed), but this derivation presupposes the existence of a model. Given the absence of alternatives, it is attractive to think that the pair θαÏÏá½»Î½Ï : θάÏÏÎ¿Ï was pivotal in the emergence of the new derivational pattern, i.e. that the original base form *θαÏÏá½»Ï of θαÏÏá½»Î½Ï was lost before our first attestations.129 This is corroborated by the pair á½ÏÏαλέÏÏ : á½ÏÏá½»Î½Ï âto inciteâ, which is clearly based on θαÏÏÎ±Î»á½³Î¿Ï : θαÏÏá½»Î½Ï âto encourageâ.
Secondly, except for the fact that θÏαÏá½»Ï is actually attested, there is every reason to believe that *dhrÌ¥s-úâ would indeed have resulted in *θαÏÏá½»Ï, whether its vocalism arose by analogy with the full grade root *dhersâ or by regular sound change. If we suppose that this *θαÏÏá½»Ï was supplanted by θαÏÏÎ±Î»á½³Î¿Ï (derived from θάÏÏοÏ), all pieces suddenly fall into place. First, *θαÏÏá½»Ï (perhaps assisted by θαÏÏá½³Ï) induced the reshaping θέÏÏÎ¿Ï >> θάÏÏοÏ. Next, after *θαÏÏá½»Ï had fallen in disuse and was replaced by θαÏÏαλέοÏ, a new derivational pattern θάÏÏÎ¿Ï â θαÏÏá½»Î½Ï emerged.
If this account is accepted, it still remains to account for the form θÏαÏá½»Ï. Its deviant root shape can only be explained as the regular phonetic reflex of a pre-form *thrÌ¥súâ. It can be excluded, however, that both θÏαÏá½»Ï and *θαÏÏá½»Ï resulted from the same paradigm in the same dialect. In my view, a promising solution is that θÏαÏá½»Ï has a special epic reflex of *rÌ¥; it was borrowed from epic into the spoken language with its archaic meaning âbold, intrepidâ (whence âaudacious; recklessâ). It would not be unexpected if an adjective meaning âbold, daring, recklessâ was borrowed from heroic poetry. This scenario will be bolstered with further arguments in chapter 6, where I propose that âÏαâ was a regular outcome of what I shall call âEpic *rÌ¥â, i.e. *rÌ¥ which was retained in Epic Greek longer than in the vernaculars and then underwent its proper vocalization.
Such a suggestion may appear random at this point, especially since the isolated adjective ÏαÏÏá½³ÎµÏ (with its reflex âαÏâ) is also limited to Epic Greek. In order to effectively counter this objection, we must analyze more material. I will start with a discussion of the root of κÏαÏá½»Ï and καÏÏεÏá½¹Ï in chapter 5, and then turn to the Homeric evidence for muta cum liquida scansion in forms with âÏαâ or âÏοâ in chapter 6. At this point, we may already take into account the fact that θÏαÏá½»Ï occurs in Homeric material that is clearly traditional: cf. the formulaic verse-end θÏαÏειάÏν á¼Ïὸ ÏειÏῶν, the phrase Ïόλεμον θÏαÏύν, and the metrically governed alternation between θÏαÏÏ â and θεÏÏιâ as first compound members and in names.
4.6 Conclusions
Starting out from a discussion of the expected ablaut grades in PIE and Proto-Greek âCalandâ formations, we have seen that many forms with âαÏâ (âαλâ) and âÏαâ (âλαâ) cannot be used as evidence for the regular reflex of *rÌ¥ or *lÌ¥. This holds for most forms belonging to the following categories:
-
s-stem nouns (e.g. ÏλάÏοÏ, θάÏÏοÏ, κÏá½±ÏοÏ) and compounded adjectives (e.g. âÏλαÏá½µÏ, âθαÏÏá½µÏ, âκÏαÏá½µÏ); these originally had a full grade root and secondarily introduced the zero grade reflex of a simplex adjective;
-
the u-stem adjectives ÏλαÏá½»Ï, κÏαÏá½»Ï, βÏαÏá½»Ï, βÏαδύÏ, whose vocalization may have been influenced by the full grade slot of the root (as in the forms of comparison);
-
a number of adjectives in âÎ±Î»á½³Î¿Ï like θαÏÏαλέοÏ, á¼ÏÏαλέοÏ, ÏαÏβαλέοÏ, which may owe their vocalism to earlier s-stem abstracts or stative verbs.
The so-called âstativeâ verbs in âá½³Ï (e.g. κÏαÏá½³Ï, θαÏÏá½³Ï, ÏαÏβέÏ) play an ambiguous role. Etymologically, they have a zero grade root, but synchronically they have derivational ties with s-stem nouns and adjectives, witness the fact that the Lesbian poets use κÏεÏá½³Ï beside Ïὸ κÏá½³ÏÎ¿Ï (contrast Ionic-Attic κÏαÏá½³Ï). For this reason a form like θαÏÏá½³Ï is difficult to use for the purpose of reconstruction, although it may in theory display the direct reflex of zero-grade *thrÌ¥sâ. In the case of ÏαÏβέÏ, Ïá½±ÏβηÏα it is quite plausible that its aorist directly reflects *trÌ¥gw-Ä-sâ.
Returning to the u-stem adjectives, we have seen that three forms show the regular vocalization of a syllabic liquid. PIE *mlÌ¥d-úâ âsoftâ is reflected in the plural form βλαδεá¿Ï, attested as a gloss in Hesychius. The factitive verb á¼Î¼Î±Î»Î´á½»Î½Ï âto corrodeâ is a denominative based on another reflex of *mlÌ¥d-úâ, *á¼Î¼Î±Î»Î´Ï â with secondary á¼â, whose vocalization was influenced by the full grade *meldâ (cf. μέλδομαι âto meltâ). It is uncertain how this divergence is to be explained (possibly a dialectal difference). For further evidence for *lÌ¥ > âλαâ, see chapter 10.
We have discussed θÏαÏá½»Ï and ÏαÏÏá½³ÎµÏ as the only two cases of a reflex of *rÌ¥ for which analogical reshaping is excluded. Ex hypothesi, the two reflexes cannot both be regular in the same variety of Greek. The form ÏαÏÏέεÏ, an archaic plurale tantum in Homer, with an aberrant accentuation of the feminine ÏαÏÏειαί, shows the regular Proto-Ionic reflex. The cognate adjective ÏÏαÏεÏá½¹Ï âsolidâ, with its alternative vocalization, was productively derived from the verb ÏÏá½³Ïομαι, á¼ÏÏá½±Ïην âto thickenâ. In chapter 5, we will see that *rÌ¥ > âαÏâ in the Ionic-Attic vernaculars is also supported by καÏÏεÏá½¹Ï âfirmâ and κάÏÏα âveryâ, while *rÌ¥ > âÏαâ is found in epic vocabulary derived from this root (κÏαÏεÏá½¹Ï, κÏαÏαιόÏ).
This leaves us with the task of accounting for θÏαÏá½»Ï. Excepting some cases in poetry, there was never a free allomorphy between θÏαÏâ and θαÏÏâ, neither in Homeric Greek nor in Classical prose. I have argued that an alternative form *θαÏÏá½»Ï once existed: this underlies the factitive verb θαÏÏύνÏ, and crucially, it would be hard to understand why and how the allomorph θαÏÏâ spread through all other derivatives (replacing θεÏÏâ) if θÏαÏá½»Ï had always been the current form of the adjective. The adjective corresponding functionally to θάÏÏÎ¿Ï and θαÏÏá½»Î½Ï is θαÏÏÎ±Î»á½³Î¿Ï (not θÏαÏá½»Ï), and it would make good sense if *θαÏÏá½»Ï was ousted by θαÏÏαλέοÏ. The pre-Homeric loss of *θαÏÏá½»Ï would also explain how the derivation of factitives in âá½»Î½Ï from s-stem neuter abstracts started (namely from the pair θαÏÏá½»Î½Ï : θάÏÏοÏ).
In view of these considerations, I suppose that *θαÏÏá½»Ï is the regular Proto-Ionic reflex of *thrÌ¥suâ, and that θÏαÏá½»Ï did not develop in spoken varieties of Ionic-Attic. In chapter 6, I will further elaborate the idea that its reflex âÏαâ arose within the prehistory of Epic Greek.
For introductions to the Caland system of derivational morphology, the reader is referred to Meissner (2006) and Rau (2009).
The following paragraphs on the history of scholarship on Calandâs Law and the Caland system closely follow Meissnerâs overview (2006: 14â16). See there for a full discussion of all scholars who significantly contributed to the subject.
Wackernagel also claimed that á¼ÏÎ³á½¹Ï originated from *arg-róâ by dissimilation; this has in more recent times been doubted by Nussbaum; see Vine (2011) for discussion.
As Meissner (2006: 11â14) shows, this discovery had already been made by Parmentier (1889), who was actually the first to recognize the systematic nature of the alternations involved, but only failed to see that *âiâ in compounding first members also took part in the alternations.
The term âCaland systemâ was coined by Nussbaum (1976: 5). Rau, closely following Nussbaum, speaks of âa certain subset of IE roots that take a more or less well-defined subset of IE nominal suffixes that stand in a close derivational relationship and can be thought of as mutually implying one anotherâ (2009: 70).
Cf. Meissner (2006: 18), although he does not include the comparative and superlative forms in the Caland alternations, because âthe more we go back in time the more universal the use of the inherited suffix *âi̯osâ for the comparative becomesâ.
The terminology is that of Nussbaum (1976: 6). For the root ÎºÏ Î´â in question, another positive ÎºÏ Î´Î½á½¹Ï is found beside ÎºÏ Î´Ïá½¹Ï, but first in Hesiod. In the following discussion, I will leave most âmarginalâ Caland suffixes out of consideration, though some instances (such as âαλέοÏ) will treated in more detail.
See Meissner (2006: 71).
The old strong stem of ÏαÏá½»Ï may well have been *thÄkhâ, in view of the Eretrian PN ΤήÏιÏÏÎ¿Ï âwith swift horsesâ (first connected with ÏαÏá½»Ï by Bechtel; cf. GEW, DELG). This interpretation is appealing in view of the Homeric phrases ÏαÏá½³â á¼µÏÏÏ and ÏαÏá½³ÎµÏ Î´â á¼±ÏÏá¿ÎµÏ. The alternative etymology advocated by Lamberterie (1990: 584â590) seems less attractive to me.
The form ÏλαÏαμών âflat stone or rockâ (with root-final âαâ < *h2) does not belong to the Caland system in Greek, but it has an immediate formal counterpart in Ved. prathimánâ âextensionâ. As a morphologically isolated and lexicalized item, ÏλαÏαμών must be of considerable antiquity. On the basis of the comparison with Ved. prathimánâ, a PIE pre-form *pleth2-mónâ has been reconstructed (cf. NIL 564). However, as Jesse Lundquist points out to me, the latter form may well be of inner-Vedic date (created as an alternative for the older abstract formation práthasâ; the Vedic form is discussed by Rau 2009: 121, 133). Consequently, one might also derive ÏλαÏαμών from PIE *plÌ¥th2-mónâ.
A possible reflex of the full grade *pleth2â in Greek has been identified by Blanc (2012) in á¼ÏλεÏÎ¿Ï âimmenseâ, which he derives from *smÌ¥-pleth2-etoâ, a doublet of *smÌ¥-pleth2-esâ as reflected in Ved. sáprathas- âextendedâ.
Balles, who is followed in this by Rau (2009) and various other scholars, has introduced into Indo-European Linguistics the descriptive framework developed by Dixon (1982). According to this view, most Caland adjectives belong to the class of âproperty concept adjectivesâ, i.e. they âpredicate some non-verbal and non-relational property concept stateâ (Rau 2009: 78). Typical examples are adjectives for dimensions, physical properties, and speed.
As was done by Kuiper (1942: 55), who compared neuter nouns of the type *dór-u, *dr-éu-s.
For the deverbal nature of adjectives in âÏá½¹Ï, see now Van Beek 2021a.
See sections 4.3 and 4.4 on these words.
See e.g. Fischer (1991), Rix (1992: 123 and 147), Meissner (2006: 35), Beekes (2011: 221). However, the acceptance is not universal: a different view is expressed by Lamberterie (1990, e.g. 953), who argues that instances of a full grade root in u-stem adjectives were introduced from coexisting verbal forms.
Szemerényi accepts Meilletâs view âthat âÏâ, earlier âÏÏâ, is due to expressivityâ, while deriving Î´Î±Ï Î»á½¹Ï from *dnÌ¥suloâ. This view is accepted by Lamberterie (1990: 702).
On Î´Î±Ï Î»á½¹Ï vs. δαῦλοÏ, see Radt (1982; 1994), who argues that the barytone accentuation is old; but according to Probert (2006: 368) âthe case is by no means clear-cutâ. For the reconstruction, see Lamberterie (1990: 702), GEW and DELG (both s.v. Î´Î±Ï Î»á½¹Ï).
For this idea, see also Nikolaev (2010: 238â239, 241, with references to earlier literature).
With dissimilation to *eithuâ in Ionic, the intermediary stage *eithuâ being attested in the derivative ÎµÎ¹Î¸Ï [νη]ν âfineâ (Chios, 5th c.); see Willi (2002: 129).
The correctness of this identification with Lith. judùs is suggested by Homeric phraseology: á¼°Î¸á½ºÏ Î¼á½±ÏεÏθαι means âto fight face to faceâ.
Willi, however, explains the full grade of PGr. *i̯euthuâ by assuming that it replaced the (in his view expected) outcome *á½Î¸á½»Ï < *(H)i̯udhúâ on euphonic grounds. This seems an emergency solution to me.
The Vedic forms with guá¹a root (pres. svádati, caus. svadáyati) can be explained by Lubotskyâs Law (Lubotsky 1981).
It is even possible that a trace of *suh2d-uâ is found in Goth. sutis âquiet, peacefulâ, but there are various problems with this idea: see Lamberterie (1990: 487â489) with further literature.
See Fischer 1982 and 1991. I do not consider it likely that the root vocalism of brevis was taken over from the comparative brevior (Sihler 1995: 358); one expects the base form (adjective) to influence the derivative (comparative). This is in fact precisely what happened in Latin, because comparatives like brevior contain the *âuâ of the positive.
For this reconstruction of Lat. gravis, see Fischer (1982), Nussbaum (1976: 41, 68). Greek βαÏá½»Ï, Ved. gurúâ, Av. gouruâ, Goth. kaurus (etc.) all point to a zero grade root *gwrH-uâ.
For this reconstruction and for different possibilities, see NIL 483â484. A zero grade root is found in Ved. mrÌ¥dúâ âsoft, weakâ, Gr. βλαδεá¿Ï (see section 4.4) and other cognates. The reconstruction of Lat. mollis âweak; softâ is debated. Nussbaum (1976: 67) claims that *mlÌ¥du̯iâ would turn up as Latin *molluis, and concludes from this that the pre-form of mollis must have been *mlÌ¥d-iâ. I fail to see, however, how the alternative reconstruction *melduâ plus âiâ can be excluded: in *meldu̯iâ, with a larger consonant cluster, the d may well have been lost before the development of intervocalic âdu̯â to âu̯â (as in suÄvis) took place. In that case, the development was *meldu̯iâ > *melu̯iâ > *melliâ > *molliâ. Thus, Lat. mollis possibly represents an old e-grade, too.
See Kloekhorst 2014: 184 on the distribution between plene and non-plene spellings in tepuâ.
Nussbaum (1976: 67â68) explained Lat. âviâ as a complex suffix consisting of the âcentralâ suffixes âuâ and âiâ. In more recent times, however, he has revoked this idea (cf. Weiss 2009: 315).
It is found, for instance, in Meier-Brügger (1992a: 84, less explicitly 2010: 357â358), Rix (1992: 168), Chantraine (1961: 109â110).
See Ruijgh (1992: 91 n. 50), who speculates that this leveling of ablaut grades could start when instances like comp. *h2er-iosâ : superl. *h2r-is-tHoâ developed to PGr. *ari̯osâ, *aristoâ by regular sound change.
A review of the problem and the history of early research on it can be found in Seiler (1950: 21â22), who does not reach a definite conclusion. The idea that the PIE superlative had a zero grade root comes from Osthoff (1910), who drew attention to the oxytone accentuation of some Vedic superlatives (e.g. kaniá¹£á¹há- âyoungestâ, daviá¹£á¹hám âfar awayâ). Such forms are probably archaic: in Classical Sanskrit, they are lost or replaced by root-accented forms. However, the final accent of these superlatives is not necessarily related to their root vocalism: even if the suffix was accented (PIE âtHóâ), the root may have had a full grade, because the PIE superlative (*CeC-is-tHoâ or *âmHoâ) was in all probability derived from the weak stem of the comparative. The latter can be reconstructed as *CeC-isâ on account of forms like Goth. mais âmoreâ < *meh2-isâ.
See chapter 5 for further discussion of these forms.
The Attic pairing of a comparative á½Î»Îµá½·Î¶Ïν beside a superlative á½Î»á½·Î³Î¹ÏÏÎ¿Ï also looks archaic at first sight. However, given that Homer has á½Î»á½·Î¶Ïν, Attic á½Î»Îµá½·Î¶Ïν is rather to be analyzed as secondarily influenced by its counterpart μείζÏν (which itself replaces older μέζÏν). This analysis is confirmed by the fact that Att. á½Î»Îµá½·Î¶Ïν, like μείζÏν, has a spurious diphthong (see Seiler 1950: 101â103).
The long root vowel of the neuter and adverb μᾶÏÏον does not derive from *mÄki̯Ån, but is secondary; cf. Barber (2013: 169â170).
Seiler (1950: 75â76, following Brugmann) thought that μήκιÏÏÎ¿Ï was influenced by the neuter s-stem μá¿ÎºÎ¿Ï, but this assumption is both unmotivated and unnecessary. My main objection is that a replacement of a putative *μάκιÏÏÎ¿Ï by μήκιÏÏÎ¿Ï would distantiate the superlative from comparative μάÏÏÏν and positive μακÏá½¹Ï (i.e. this analogy would lack all motivation). Besides, it would entail a change in prosodic structure, a problem which does not exist if we assume a replacement of *mÄki̯Ån by *maki̯Ån. It is quite possible that the neuter μá¿ÎºÎ¿Ï exerted influence on the superlative μήκιÏÏοÏ, but only in the sense that it helped to preserve it against the pressure of μακÏá½¹Ï, μάÏÏÏν.
So strong stem *CéC-i̯osâ vs. weak stem *CéC-isâ in the comparative, and *CéC-is-t(H)oâ for the Greek and Indo-Iranian superlative: see Seiler (1950: 21: âNiemand bestreitet diesâ, with reference to Meillet and Brugmann); Sihler (1995: 358); Weiss (2009: 356).
Slavic has âÑsâ (continuing zero grade *âisâ) and Old Prussian has forms deriving from *âisâ, too (cf. Stang 1966: 267â268).
See Barber (2013: 157).
That is, a paradigm with non-ablauting e-grade root and o/zero suffix ablaut. In Greek, the suffix allomorph *âisâ was subsequently eliminated in favor of *âiosâ, a process paralleled in other 3rd declension paradigms.
In other words, the appearance of e-grades and zero grades in Greek forms of comparison is not at all random. I therefore do not share Barberâs pessimism (2013: 157â158) regarding the possibility to draw conclusions about the original root ablaut of specific comparative formations in Greek. For instance, when Barber remarks concerning Ion. κÏá½³ÏÏÏν vs. Doric κάÏÏÏν that âit seems arbitrary to give one historical priority over the other on the basis of ablaut grade alone, if there is a good chance that there was some sort of root ablaut alternation in the first instanceâ (2013: 158), he fails to note that the vocalism of κÏá½³ÏÏÏν is aberrant within the adjectival paradigm (and is therefore an archaism), whereas that of κάÏÏÏν can be easily explained by analogical leveling (see chapter 5 for further details). In other words, while it is theoretically possible to explain κÏá½³ÏÏÏν beside κάÏÏÏν as reflecting original root ablaut in the comparative paradigm, it is much less costly to explain this difference from a paradigm without root ablaut.
Likewise, it has been claimed that Î¸á½±Î¼Î²Î¿Ï âstupefactionâ beside Ïá½±ÏÎ¿Ï âid.â reflects a paradigm containing the root shapes *dhembhâ and *dhmÌ¥bhâ: cf. Hackstein 2002: 237. On this problematic word, see below.
Beside the inherited s-stem νέÏÎ¿Ï âcloudâ (cf. OCS nebo, Hitt. nÄpiÅ¡, Ved. nábhasâ), forms like Ved. ámbhas âwaterâ, Arm. amb âcloudâ have been interpreted as containing the outcome of a zero grade root *nÌ¥bhâ with a re-introduced nasal. However, since an ablauting root existed in PIE (cf. Ved. abhráâ â(thunderâ)cloudâ, Av. aβraâ ârain-cloudâ < *nÌ¥bh-róâ, Lat. imber ârain (shower)â < *nÌ¥bh-riâ), it cannot be proven that the s-stem paradigm originally harbored a zero-grade root allomorph.
The adverb in âα may have played a role in the case of μάλα : μᾶλλον : μάλιÏÏα, where we only have adverbial forms.
Meissner (2006: 182â186) argues that there was no derivational link between adjectives in âá½»Ï and compounds in âηÏ.
See Tucker (1990: 28â33), following Watkins (1971).
Cf. also Î¸á½±Î¼Î²Î¿Ï âamazement, stupefactionâ beside θαμβέÏ, θαμβá¿Ïαι âto be stupefiedâ. Tucker (1990: 42â43) proposes to derive the verb Î¸Î±Î¼Î²á½³Ï from the s-stem noun. The argument is that the root shape θαμβâ cannot represent the development of a syllabic nasal (as in aor. ptc. ÏαÏών âstupefiedâ), and that it must therefore be a remodeling of *θεμβâ. That root shape, however, cannot be original in the stative verb in âá½³Ï, but it would be at home in a neuter noun. However, a problem with this argument based on θαμβέÏ, θαμβá¿Ïαι is that the reconstruction of the verbâs phonetic developments remains uncertain. Szemerényi (1954) argued for comparing θαμβέÏ, aor. ÏαÏών and pf. ÏέθηÏα âto be stupefiedâ with Goth. dumbs âmuteâ and cognates (PGmc. *dumbaâ, PIE *dhembhâ). In subsequent discussions, Barton (1993) and Hackstein (2002: 237â238) have tried to account for the difference between θαμβâ, ÏαÏâ and θηÏâ. The etymology is semantically plausible, but Szemerényiâs reconstruction of Î¸Î±Î¼Î²á½³Ï presupposes that Greek âμβâ may derive from PIE *âmbhâ, a development rendered uncertain by á½Î¼ÏÎ±Î»á½¹Ï ânavel, center; hubâ < PIE *h3nbh-l-óâ and á¼ÏÏεμÏá½³Ï âfirmlyâ (cf. Ved. stambh, and Van Beek 2018 for the reconstruction of á¼â). The explanation of ÏέθηÏα instead of expected Ïá½³ÏηÏα (attested in Hsch.) given by Barton (1993) is not without problems either. Beekes (EDG s.v. θαμβέÏ) argues that the variation between θαμβâ, ÏαÏâ and θηÏâ can be understood if the words were borrowed from the Pre-Greek substrate: in such words, interchanges between stops and pre-nasalization are more frequently attested. Although this no longer seems likely to me in view of the archaic morphology of these verbs (cf. also Barton 1993: 2 with n. 3), the reconstruction of Î¸Î±Î¼Î²á½³Ï involves too many problems to base an account of the derivational history of statives in âá½³Ï on it.
Hitt. tarkuu̯ant- may alternatively be derived from *dr̥ḱ-u̯entâ âwith [angry] gazeâ, cf. Hom. á½ÏόδÏα âlooking askanceâ. The comparison of Middle Welsh tarfu âto disturb, trouble, scareâ with ÏαÏβâ is not without problems (cf. EDPC q.v.).
Pace Chantraine (DELG s.v. ÏαÏβέÏ), who thinks that ÏαÏÎ²Î±Î»á½³Î¿Ï âpourrait être ancienneâ.
As I did in Van Beek 2013: 92.
LIV2 632, Anm. 1.
In Classical Attic prose (and presumably in the spoken vernacular), only the following 14 u-stem adjectives were current: βαθύÏ, βαÏá½»Ï, βÏαδύÏ, βÏαÏá½»Ï, δαÏá½»Ï, δÏιμύÏ, εá½Ïá½»Ï, εá½Î¸á½»Ï, θÏαÏá½»Ï, ÏαÏá½»Ï, ÏλαÏá½»Ï, ÏÏαΰÏ, ÏαÏá½»Ï, ÏÏαÏá½»Ï.
Lamberterie (1990: 470) shows that there is no clear semantic difference between Î³Î»Ï Îºá½»Ï and Î³Î»Ï ÎºÎµÏá½¹Ï.
Some Homeric u-stem adjectives with a heavy root syllable are very numerous as tokens, e.g. á½Î¾á½»Ï and ὠκύÏ, but this is mainly due to their frequent occurrence in formulaic cola (cf. á½Î¾á½ºÎ½ á¼Ïηα, á½Î¾á½ºÎ½ á¼ÎºÎ¿Î½Ïα, ξίÏÎ¿Ï á½Î¾á½», Ïá½±Ïγανον á½Î¾á½», á½Î¾á½³Ï Î´Î¿Ï Ïá½·, á½Î¾á½³Ï Ïαλκῷ, all of which belong to traditional war narrative).
See chapter 5 for the antiquity of κÏαÏεÏá½¹Ï, and note that Î³Î»Ï ÎºÎµÏá½¹Ï (denoting quiet, i.e. non-violent, activities) was more or less its antonym. Also, note that θαλεÏá½¹Ï âabundantâ is probably an inner-Greek innovation, too: the correspondence with Arm. dalar âgreen, freshâ is rightly criticized by Clackson (1994: 118â120), who notes the semantic distance and points out that dalar cannot be derived from a pre-form in *âeroâ. Thus, the pair κÏαÏá½»Ï : κÏαÏεÏá½¹Ï may also have induced the creation of θαλεÏá½¹Ï beside Î¸Î±Î»á½»Ï (of which Homer preserves relics, f. θάλεια and probably gen. pl. n. θαλέÏν). On the other hand, ÏÏαÏεÏá½¹Ï âthick, solidâ was derived not from the adjective ÏαÏÏá½»Ï (which has a different vowel slot and meaning), but directly from the verbal root of ÏÏá½³Ïομαι (á¼ÏÏá½±Ïην) âcoagulate; form a crustâ: see section 4.3.2 below.
E.g. Tucker (1990: 55â56): âtheir vocalism or phonological shape suggests that they were created on the basis of s-stem nounsâ. Rau (2009: 128 n. 9) observes that this process âgenerally results from the derivational association of morphologically unrelated formationsâ. A number of adjectives in âÎ±Î»á½³Î¿Ï acquired the suffix by contamination with semantically close forms, such as á¼ÏÏÏÎ±Î»á½³Î¿Ï âdryâ, á¼Î¶Î±Î»á½³Î¿Ï âid.â (after καÏÏαλέοÏ) and their antonym Î¼Ï Î´Î±Î»á½³Î¿Ï âmoistâ; cf. also á½ÏÏÎ±Î»á½³Î¿Ï and λεÏÏαλέοÏ. For an overview of such forms, see Debrunner (1917: 165â168).
For this interpretation of κοÏδύÏ, and on the question why the adjective was lost so early, see Lamberterie (1990: 867â874), following R. Schmitt. The reflex âοÏâ points to an Aeolic or Arcado-Cyprian origin.
The adverb á½ÏÏαλέÏÏ âquicklyâ (after Homer also adj. á½ÏÏαλέοÏ) was formed beside á½ÏÏá½»Î½Ï âto spur onâ after the semantically close model of θαÏÏÎ±Î»á½³Î¿Ï : θαÏÏύνÏ. Note that á½ÏÏá½»Î½Ï has no convincing etymology (the traditional comparison with Skt. tvarate and OHG dweran is criticized by Beekes, EDG s.v. á½ÏÏαλέÏÏ).
These forms and their reconstruction are further discussed in section 10.2.1.
It is not excluded that the vocalism of θαÏÏÎ±Î»á½³Î¿Ï and θάÏÏÎ¿Ï was influenced by that of θαÏÏá½³Ï, which could directly reflect an inherited âstativeâ *thrÌ¥s-Äâ. However, the derivation of adjectives in âÎ±Î»á½³Î¿Ï from nouns in âÎ¿Ï seems more robust (cf. κεÏδαλέοÏ, κέÏδοÏ) than that from verbs in âá½³Ï.
This view is widespread, see e.g. the casual remark by Strunk (1975: 286), regarding καÏÏύνÏ, that âinlautendes âαÏâ < *ârÌ¥â vor Konsonant (â¦) auch sonst gelegentlich statt oder neben âÏαâ vorkommt.â
An explanation for these distributions is proposed in chapter 6.
However, creating *(á¼)κάÏÏηÏα was not strictly necessary from the viewpoint of verse composition, as the meaning âto obtain victoryâ was expressed already by metrically equivalent νίκηÏα (cf. formulaic νικήÏῠκÏείÏÏÏν Ïε γένηÏαι, 4â¯Ã Hom.).
The absence of *(á¼)κάÏÏηÏα can be explained if we accept that κÏαÏá½³Ï functioned, from a synchronic viewpoint, as a primary formation. I now disagree with Tuckerâs suggestion (1990: 62â67) that statives in âá½³Ï could be derived from s-stem compounds, and revoke what I said about this in Van Beek 2013: 92. In fact, if compounded adjectives in âÎ·Ï are derived from statives in âá½³Ï, as I now think, the absence of *καÏÏá½³Ï accounts directly for the absence of compounds in *âκαÏÏá½µÏ.
The classic study of Greek denominatives is Fraenkel (1906).
For καÏÏύνÏ, the same suggestion was made by Strunk (1975: 296): âVermutlich ist κάÏÏÎ¿Ï sogar die wirkliche morphologische Basis für die epische Verbalableitung.â
The root of á¼Î½Ïεα has no convincing etymology. In spite of DELG (s.v. á¼Î½ÏοÏ), the connection with á¼Î½á½»Ï âto accomplishâ < *snh2-uâ is difficult to maintain because that root ended in a laryngeal.
Compare Chantraineâs judgment, âaprès á¼Î½ÏύνÏâ (DELG s.v. á¼Î»á½³Î³Ï).
Examples are Lejeune (1972: 196, citing βλαδαÏá½¹Ï ~ á¼Î¼Î±Î»Î´á½»Î½Ï and âθαÏÏá½»Ï, θÏαÏá½»Ïâ; but note that the first form θαÏÏá½»Ï is not directly attested), Rix (1992: 65, citing ÏλαÏá½»Ï), and Sihler (1995: 94â95, citing θÏαÏá½»Ï, βλαδύÏ, and ÏλαÏá½»Ï). DELG (s.v. θάÏÏοÏ) remarks that a form θαÏÏá½»Ï âa dû exister comme lââ¯indiquent divers composés et le verbe dénominatif en âύνÏ.â As we will see below, this may well be true, but the form should be asterisked in any case. Sihler not only cites θαÏÏá½»Ï, but also âLesb. θοÏÏá½³ÏÏâ, which does not exist at all (the form is correctly cited as θÏοÏá½³ÏÏ by Lejeune (1972: l.c.), and is attested as such only in Joh. Gramm. 2.11, where it is labelled Aeolic).
An exception could be made for the gloss κοÏδύÏÎ ÏανοῦÏÎ³Î¿Ï (Hsch.), but this is of non-Ionic-Attic origin: see above. As I have argued in section 2.2, the Mycenaean form ta-pa-e-o-te is too uncertain to be used in this discussion.
Lamberterie (1990: 665â666) remarks that this use is intimately related with that as a qualification of snowflakes, the image being that of a rain of arrows.
An extensive study of the semantics of ÏÏá½³ÏÏ is Demont (1978).
The adverb ÏαÏÏá½³ÏÏ (only B. 13.53) is a trivial reshaping of the Homeric acc. n. pl. ÏαÏÏέα.
The two Aeschylean passages are discussed by Lamberterie (1990: 671). In Pers. 926, ÏαÏÏá½»Ï ÏÎ¹Ï is a plausible conjecture for Î³á½°Ï Ïá½»ÏÏιÏ, because Ïá½»ÏÏÎ¹Ï would be a vox nihili. Lucian uses the gen. sg. ÏαÏÏá½³Î¿Ï modifying á¼Ïá½³ÏληÏ. Remarkably, in all three cases the adjective modifies a feminine noun. Moreover, note that in both ÏαÏÏá½ºÏ â¦ Î¸Ïίξ âthick hairâ (A. Sept. 535) and the conjecture ÏαÏÏá½»Ï ÏÎ¹Ï Î¼Ï ÏÎ¹á½°Ï á¼Î½Î´Ïῶν âthronging myriads of menâ (A. Pers. 926), ÏαÏÏá½»Ï qualifies a singular noun with plural or collective sense. I therefore claimed in Van Beek 2013: 101 that the singular of ÏαÏÏá½»Ï was a secondary creation by Aeschylus. I still consider this plausible, but there is no need to insist on it.
Î²Î±Î¸Îµá½·Î·Ï Ïá½±ÏÏεÏιν á½Î»Î·Ï âin the thicket of the deep forestâ (Il. 5.555), Î²Î±Î¸á½³Î·Ï á¼Î½ Ïá½±ÏÏεÏιν á½Î»Î·Ï (Il. 15.606), and μνιόενÏα Î²Ï Î¸Î¿á¿Î¿ Ïá½±ÏÏεα âthe mossy thicket of the depthâ (A.R. 4.1238).
Already before Homer, following the loss of intervocalic digamma, the u-stem adjectives in Ionic-Attic had generalized the dat. pl. ending âεÏι by a proportional analogy with the s-stems, with which they shared the nom. pl. n. in âεα.
From an older s-stem form, one would expect Att. βÏá½±Ïη (Meissner 2006: 108â109).
In Van Beek 2013: 101, I claimed that ÏαÏÏâ was protected from analogical replacement by ÏÏαÏâ because ÏαÏÏá½³ÎµÏ would have become a plurale tantum early on. I explained the difference with other u-stem adjectives (e.g. κÏαÏá½»Ï, βÏαÏá½»Ï), in which analogical influence of the full grade did take place, with the assumption that paradigmatic root ablaut was still preserved in the singular paradigm when *rÌ¥ was eliminated in Proto-Ionic. However, while I still consider it plausible that the singular form ÏαÏÏá½»Ï was created anew beside ÏαÏÏá½³ÎµÏ after Homer (cf. above), I now renounce on the idea that the adjectives in âá½»Ï retained paradigmatic root ablaut until as late as Proto-Ionic.
This is not to say that no comparative or superlative forms of ÏαÏÏá½»Ï ever existed, but only that they were apparently too marginal (at the relevant time) to influence the vocalization of the adjectival root. Since being clotted or curdled is an aggregation state, its meaning does not easily lend itself to the expression of degree: what matters most is the binary opposition between âthickâ and âliquidâ.
Incidentally, note that the absence of forms like *ÏÏαÏειαί with muta cum liquida scansion (which would be the expected outcome of *thrÌ¥pheu̯i̯ai in Epic Greek, according to the scenario to be put forward in chapter 6) can be ascribed to the availability of the synonymous form θαμειαί.
á½Î³Ïá½µ Ïε ÏÏαÏεÏá½µ Ïε (A.R. 2.545/6), á½Î³Ïá¿Ï Ïε ÏÏαÏεÏá¿Ï Ïε (A.R. 4.281), further in Arat. 1.1027, Theoc. 21.18 and 44, in the last passage in the meaning âwell-fed, thick, fatâ.
LSJ (s.v. ÏÏαÏεÏá½¹Ï) remarks that ÏÏαÏεÏá½¹Ï is from ÏÏá½³ÏÏ in the meaning âto make thickâ. The connection with ÏÏá½³Ïομαι âto curdleâ is also corroborated by glosses like ÏÏαÏεÏόνΠÏηκÏόν. ÏÏá½¹Ïιμον. Î»ÎµÏ Îºá½¹Î½. ξηÏόν. ÏεÏηγμένον (Hsch. Ï 1284).
Whether the relation between the forms of comparison and the basic adjective is seen as inflectional or derivational is a theoretical issue that need not detain us here.
On the hapax comparative βÏá½±ÏÏÏν Il. 10.226, which may belong to βÏαδύÏ, see immediately below. It seems that (Ï)μικÏá½¹Ï âshortâ, á½Î»á½·Î³Î¿Ï âsmallâ, and the adverb Î¼á½·Î½Ï Î½Î¸Î± âfor a short timeâ are used instead of βÏαÏá½»Ï in Epic Greek. This may be due to a general preference for using thematic forms in hexameter Greek (see section 4.2.2); besides, the superlative βÏá½±ÏιÏÏÎ¿Ï was metrically unfit.
Cf. also βÏá½¹ÏÏονοÏΠβÏαÏÏ Ïá½³ÏÎ¿Ï (Hsch. β 1193), probably of Aeolic origin.
On the question whether βÏαÏá½·Ïν â(upper) armâ is related, see section 6.9.5.
A reflex of the zero grade root is found in Ved. múhur âinstantlyâ, Av. mÉrÉzu-jÄ«tiâ âshort-livedâ, OHG murg(i) âshortâ, Goth. gamaurgjan âto shortenâ (denom. verb).
For this point, see further section 7.2.1.
Only 3 attestations in Homer: βÏÎ±Î´á½»Ï (Od. 8.329 and 330), βÏÎ±Î´á½³ÎµÏ â¦ á¼µÏÏοι (Il. 8.104).
Cf. Seiler (1950: 43 and 56â¯f.); differently Barber (2013: 160 with n. 23). The text at Il. 10.226 runs: βÏá½±ÏÏÏν Ïε Î½á½¹Î¿Ï Î»ÎµÏÏá½´ δέ Ïε μá¿ÏιÏ. Normally, βÏá½±ÏÏÏν is taken to be a comparative of βÏαÏá½»Ï âshortâ on phonological grounds, but semantically βÏÎ±Î´á½»Ï would fit much better (cf. Nordheider, LfgrE s.v. βÏá½±ÏÏÏν).
Cf. Lamberterie (1989) and Meissner (2006: 102â103).
Perhaps, Slavic *gÑrdÑ (> Ru. górdyj âproud, haughtyâ) is related to the Baltic forms: the o-stem may replace an earlier u-stem. A thematic noun would be presupposed by Lat. gurdus âblockheadâ, but its appurtenance is not certain (cf. Lamberterie 1990: 594â595). For further literature on these etymologies, see NIL 195â196.
One may compare the situation with that of PIE *b, for which there is hardly any evidence and where the reduction of certain clusters may perhaps explain the occurrence of *b in initial position, as Sasha Lubotsky has suggested in unpublished conference papers. Although we do not know what actually happened in the prehistory of *gwredâ, I would not exclude a similar reduction of an initial cluster in this case. There is, therefore, no reason to doubt the Indo-European pedigree of this root.
Cf. also the application of the formula κÏαÏεÏὸν Î¼á½³Î½Î¿Ï to the stamina of mules (Il. 17.742). For further connections of the root κÏαÏâ with horse-riding, see chapter 5.
I assume that the comparative βαÏδύÏεÏοÏ, attested in Theocritus, is also an artificial poetic creation.
It is possible that ÏλαÏá½»Ï âbrackishâ is a different adjective, both synchronically and historically: see Lamberterie (1990: 452â463). Proponents of the historical identity of both lexemes believe that ÏλαÏá½»Ï âbroadâ, as an epithet of the Hellespont, was misunderstood to mean âsaltyâ, an important argument being that Herodotus also calls the Hellespont á¼Î»Î¼Ï Ïá½¹Ï âsaltyâ. Cf. GEW s.v. ÏλαÏá½»Ï 2. and Mayrhofer EWAia s.v. paá¹uâ (both embracing this view), DELG s.v. 2 ÏλαÏá½»Ï (doubting it). Against this, Lamberterie remarks that ÏλαÏá½»Ï only denotes brackish, never salty water.
According to Blanc (2012), this full grade is reflected in á¼ÏλεÏÎ¿Ï âimmenseâ, which would have arisen from *smÌ¥-pleth2-etoâ by haplology.
For the same reason, the form κοÏδύÏÎ ÏανοῦÏÎ³Î¿Ï (Hsch.) beside κεÏδίÏν, κέÏδιÏÏÎ¿Ï can play no role in this discussion: κοÏÎ´á½»Ï may have replaced the regular zero grade outcome *κÏοδâ, in an Aeolic dialect, under the influence of an older strong stem *κεÏδύÏ. Theoretically, however, κοÏÎ´á½»Ï could also stem from Arcadian or Cypriot (see section 3.4).
Words provided with these sigla may have been incorporated later into the text of Hesychius from the lexicon of Cyrillus of Alexandria (5th c.); A and S denote two manuscripts of that lexicon. See Cunningham (2018: X).
The alphabetical order of these glosses (βλαδεá¿Ï following βλαδόν) may confirm that βλαδεá¿Ï is from a u-stem adjective *βλαδύÏ. Sometimes βλαδαÏá½¹Ï is quoted as attested in Galen (e.g. Rau 2009: 153), but there the form is actually a conjecture for transmitted βλαβεÏαί (cf. Lamberterie 1990: 356). Lamberterie also includes the gloss βλάζεινΠμÏÏαίνειν âto act foolishlyâ, but its appurtenance to *Î²Î»Î±Î´á½»Ï seems uncertain to me.
It has been suggested that βλαδόνΠá¼Î´á½»Î½Î±Ïον should be corrected to βλαδύν (GEW, DELG), but this is rejected by Lamberterie (1990: 356 n. 4), who also convincingly argues against the necessity to restitute Î²Î»Î±Î´á½»Ï for the transmitted form βÏÎ±Î´á½»Ï at Hp. Aër. 20 (o.c. 356â358).
I would add to this that other adjectives like á¼Î¼Î±Î»á½¹Ï âweakâ (Il.+, no etymology), á¼ÏÎ±Î»á½¹Ï âsoft, tenderâ (Hom.+), á¼Î¼Î±Ï Ïá½¹Ï âdark, unseen, invisibleâ may have played a part in the reshaping of an earlier *μαλδύνÏ. Lamberterie (1990: 362) also draws attention to the gloss á¼Î²Î»Î±Î´á½³ÏÏΠἡδέÏÏ âin a pleasant way, agreeablyâ (Hsch.). The meaning of the gloss may derive from âsoftâ, in which case the addition of á¼â would have to be secondary (as with á¼Î¼Î±Î»Î´á½»Î½Ï). For semantic reasons this seems preferable over assuming a privative s-stem compound *nÌ¥-mlÌ¥d-esâ; besides, a zero grade root would be unexpected as a second compound member.
The second gloss ÏÏεÏá½·Ïκειν suggests that the glossator was confused by the similarity with á¼Î¼á½³ÏÎ´Ï âto deprive (of eyesight)â.
Armenian meÉ«k confirms that the root did not have an initial laryngeal. Unclear to me is the claim of Hamp (1988: 89) that Arm. meÉ«k is a ârevocalizationâ of *maÉ«k.
Within early Vedic, the root mrad âto softenâ is semantically distinct from mard âto crushâ, even if both roots were confused early on (Mayrhofer EWAia s.v. MRAD). Mayrhofer reconstructs PIE *(h2)mledâ âto become weak, dissolveâ as against *merd(H)â âzerdrücken, zerreibenâ (comparing this with Lat. mordeÅ âto biteâ). However, it would not be wise to base the reconstruction *(h2)mledâ solely on the Indo-Aryan evidence, because this branch lost not only the difference between l and r, but also has no reflexes of word-initial preconsonantal laryngeals. As a result, the full grade slot could become a useful means to distinguish between two roots that had merged as a result of regular sound change. In this case, the roots *meldâ âto become weakâ (~ Lat. mollis) and *h2merdâ âto crush, biteâ (~ Lat. mordeÅ) could be kept apart by the creation of a novel full grade in mrad âto softenâ, once full-grade forms deriving from *meldâ had been lost.
Cf. LIV2, *meldâ âweich werdenâ versus *melh2â âzerreiben, mahlenâ. I reconstruct the latter root as *melh1â on account of Myc. me-re-ti-ri-ja âfemale corn grindersâ, among other reasons.
Note that German weich may mean both âweak, softâ and âflaccidâ.
After *mlÌ¥déu̯â > *mladéu̯â, one would expect an analogical reshaping either to (1) *mélduâ, *maldéu̯â and hence levelling to *maldúâ, *maldéu̯â, or directly to (2) *mladús, *mladéu̯â. Neither scenario can explain Î²Î»Î±Î´á½»Ï and á¼Î¼Î±Î»Î´á½»â at the same time.
Lejeune (1972), Sihler (1995); Chantraineâs reference to a âθαÏÏá½»Ï (attesté en composition)â (DELG s.v. θάÏÏοÏ) is more precise but may still give rise to confusion.
I regard the interpretation of the Mycenaean manâs name ta-su as uncertain.
As we will see in chapter 5, the alternations κÏá½±ÏÎ¿Ï ~ κάÏÏÎ¿Ï and κÏαÏεÏá½¹Ï ~ καÏÏεÏá½¹Ï in Epic Greek are structural and have a real linguistic basis.
In epigraphic onomastic material we also encounter a variant ÎαÏÏÏ - or ÎαÏÏÏ â, especially on Crete and in archaic Theran graffiti. In these dialects these forms may well regularly reflect *thrÌ¥suâ, but in other dialects it is difficult to exclude that ÎαÏÏÏ â replaced ÎÏαÏÏ â under the influence of θαÏÏâ in related formations such as θάÏÏοÏ. Cf. sections 3.1.2 and 3.4.3 on names in ÎοÏÏÏ â attested in Cretan and Arcadian inscriptions, and section 3.4.1 on Cyprian to-ro-su-ta-mo.
Forms of comparison of θÏαÏá½»Ï are not attested in early Greek epic. If the Classical forms θÏαÏá½»ÏεÏÎ¿Ï and θÏαÏá½»ÏαÏÎ¿Ï already existed, they could not have been used in epic meter. The hapax θÏá½±Ïιον (Alcm. fr. 87) is a secondary formation (cf. Barber 2013: 161).
θάÏÏÏ Î½Î¿Ï âconfidentâ only occurs twice in Homer. Its derivational morphology is unclear. According to Nussbaum (1976: 76) it is a composite Caland formation, with ânoâ stacked onto *dhrÌ¥s-uâ; older suggestions are listed in Risch (1974: 150â151), with further literature. Since the meaning of θάÏÏÏ Î½Î¿Ï matches that of the other θαÏÏâ forms, and since it only occurs in Homer, I suspect that it was created as a metrical alternative for Ïá½·ÏÏ Î½Î¿Ï âconfidentâ (Hom.+), the only other adjective in âÏ Î½Î¿Ï and one with an almost identical lexical meaning. Since the root meaning âconfidenceâ of θαÏÏâ is innovative, it is unwarranted to use θάÏÏÏ Î½Î¿Ï as evidence for *rÌ¥ > âαÏâ.
In the words of Chantraine (DELG s.v. θάÏÏοÏ), âCette spécialisation est secondaire comme le prouvent les faits homériques et les composés anciens avec θÏαÏá½»Ï au premier membre (â¦)â; cf. also Lamberterie (1990: 849 and 855â859). Meissner (2006: 71), however, adds a different nuance: âIt is important to note that θÏá½±ÏÎ¿Ï has the same negative connotation that the basic adjective θÏαÏá½»Ï had developed much earlier. Already in Homer θÏαÏá½»Ï is attested in the meaning âover-boldâ, ârashâ (cf. Od. 10.436, where Eurylokhos tries to hold back the companions, warning them against rash Odysseus through whose á¼ÏαÏθαλίαι many have perished), though θÏαÏá½»Ï can, of course, be positive as well.â In my view, however, the distinction between positive and negative connotations is epiphenomenal.
I disagree with Chantraineâs claim (DELG s.v. θάÏÏοÏ) that the original root meaning is âto be confidentâ.
Pace Lamberterie (1990: 850), who thinks that θÏαÏá½»Ï, with the exception of Od. 10.436, always has a positive nuance in Homer. Cf. also Meissner (2006: 71).
For Thucydides, Huart (1968: 426) reached the same conclusion concerning θάÏÏÎ¿Ï and θαÏÏá½³Ï: âtoujours θαÏÏεá¿Î½ est en rapport direct avec lââ¯actionâ, and âla confiance sââ¯oppose ainsi à lââ¯appréhension et cette opposition, assez souvent implicite, est parfois clairement formuléeâ.
I have played with the idea that θαÏÏá½³Ï, which could well be an inherited âstativeâ verb, directly reflects PGr. *thrÌ¥s-Äâ and, after its vocalization, influenced the vocalism of θάÏÏοÏ. Although such a scenario is possible, I see no clear way to rigorously prove it at present. Moreover, the earlier existence of an adjective *θαÏÏá½»Ï must be posited in any case: see below.
Excepting the one-off instance of θÏá½±ÏÎ¿Ï in Homer, see above. In the remainder of this chapter, I will refrain from citing the Attic forms with their proper dialectal outcome âÏÏâ and refer to them only in the form with âÏÏâ.
In the extant Odes of Pindar, we find 14 compounds with θÏαÏÏ â (including 7â¯Ã a proper name), as against 7 attestations of the adjective θÏαÏá½»Ï. Names in ÎÏαÏÏ â are common in inscriptions and compete with names in ÎεÏÏιâ (for an outdated overview, see Bechtel 1917: 207 and 211â213). The latter form also occurs in the appellative compound θεÏÏι-εÏá½µÏ âwith audacious wordsâ (in Bacchylides). The distribution between θεÏÏιâ and θÏαÏÏ â was probably metrically conditioned (see above); names in ÎαÏÏÏ â and ÎοÏÏÏ â are found only in certain West Greek dialects and Arcadian.
In view of its meaning, ÎºÏ Î½Î¿Î¸ÏαÏá½µÏ was based on θÏá½±ÏÎ¿Ï (Meissner 2006: 185).
See Lamberterie (1990: 849â859), and also Huart (1968: 426â431), with a special focus on Thucydides; for a concise discussion, see Meissner (2006: 70â71).
This semantic distinction has generally been interpreted as a difference between pejorative (θÏá½±ÏοÏ) and laudatory (θάÏÏοÏ) values. Cf. Huart (1968: 428): âOn admet généralement que θάÏÏÎ¿Ï et θÏá½±ÏÎ¿Ï sââ¯emploient assez indifféremment en poésie, tandis que, dans la prose, θάÏÏÎ¿Ï est utilisé de préférence en bonne part, et θÏá½±ÏÎ¿Ï en mauvaise part.â As Meissner (2006: 71) formulates, âThe negative connotation [of θÏαÏâ] becomes more frequent in Attic, and as early as in tragedy the meaning âarrogantâ, âaudaciousâ prevails (â¦). Thus, θÏá½±ÏÎ¿Ï follows θÏαÏá½»Ï not only in form but in meaning as well.â Lamberterie (1990: 856â857) argues that the pejorative connotation may have developed, in particular, in the frequent use of θÏαÏá½»Ï qualifying impudent words. Yet, as Huart (1968) convincingly shows, this distinction between pejorative and laudatory uses does not hold for Thucydides. On the other hand, Huart goes too far when he concludes that Thucydides made no distinction at all between the two forms. In my view, Thucydides observes a difference between θÏá½±ÏÎ¿Ï âboldness, audacityâ (frequently, but not necessarily, with the connotation of surplus: ârecklessnessâ), whereas θάÏÏÎ¿Ï means âconfidence, assuranceâ (either justified or unjustified: see Huart 1968: 427).
Lamberterie (1990: 856) mentions A. Supp. 772 ÏÏὶν á½ Ïμῳ ναῦν θÏαÏÏ Î½Î¸á¿Î½Î±Î¹ âbefore the ship has reached a safe havenâ, where one would expect a form in θαÏÏâ. It is perhaps on the basis of such instances that LSJ (s.v. θÏá½±ÏοÏ) remarks that âθαÏÏá½»Î½Ï and θÏαÏá½»Î½Ï are used indifferentlyâ, but that clearly goes too far.
About θÏαÏá½»Ï, Huart (1968: 430) remarks: â⦠chez les prosateurs postérieurs à Thucydide, le mot est généralement de valeur péjorative: Thucydide, lui, reste fidèle à lââ¯usage ancienâcelui de la poésieâoù le terme est pris en bonne, ou en mauvaise part.â Herodotus (7.49) also attests the meaning âboldâ without any negative connotations.
As for θαÏÏαλέοÏ, see Lamberterie (1990: 855): âelle désigne toujours lââ¯assurance, la confiance en soi, par opposition à la crainteâ. That θαÏÏÎ±Î»á½³Î¿Ï is semantically distinct from θÏαÏá½»Ï is explicitly noted by Plato, Leg. 649c (see DELG s.v. θάÏÏοÏ), although θαÏÏÎ±Î»á½³Î¿Ï does not mean âself-confidentâ there, but is rendered with á¼Î½Î±á½·ÏÏÏ Î½ÏÎ¿Ï âaudaciousâ.
This was first proposed by Chantraine (DELG, s.v. θάÏÏοÏ): âLe verbe dénominatif confirmerait lââ¯existence de *θαÏÏá½»Ï et se présente sous deux formes: θαÏÏá½»Î½Ï (att. θαÏÏâ) «â¯encourager, donner confianceâ¯Â», etc. (Hom., ion.-att., etc.) et θÏαÏá½»Î½Ï Â«â¯encouragerâ¯Â», qui se dit généralement dââ¯une audace imprudente ou impudente (Aesch. Ag. 222, Th. 1.142), surtout employé au passif et au moyen, le plus souvent au mauvaise part, cf. Ar. Gren. 846, etc.â Although I concur with Chantraine concerning the priority of *θαÏÏá½»Ï, I disagree with his claim that *θαÏÏá½»Ï and θÏαÏá½»Ï were simply doublets: â⦠la forme [θÏαÏá½»Ï] pouvant être analogique de θαÏÏá½»Ï (attesté en composition) qui présente le traitement âαÏâ de *rÌ¥, θέÏÏοÏ, etc. (â¦).â For this, Chantraine refers to Lejeune (1972), who ascribes the double reflex of *rÌ¥ to liquid metathesis.
The alternative would be to assume that the expected form *θÏαÏá½»Î½Ï was replaced by θαÏÏá½»Î½Ï under the influence of the neuter θάÏÏÎ¿Ï and the stative-inchoative verb θαÏÏá½³Ï. This is unlikely because (i) usually the root shape of the adjective (as the basic form for derivation) wins out, and (ii) as long as the adjective continued to exist in the shape θÏαÏá½»Ï, a reshaping *θÏαÏá½»Î½Ï >> θαÏÏá½»Î½Ï would be unlikely.