Introduction
This chapter discusses the developments affecting the lateral liquid when it served as a syllabic nucleus. There can be no doubt that the Proto-Ionic reflex of *lÌ¥ was a-colored. The mainstream view is that âλαâ is the regular outcome; the main aim of this chapter is to examine whether *lÌ¥ > âαλâ can be excluded, still keeping in mind that there was an early vocalization to âαλâ in some environments (e.g. before laryngeal plus vowel, cf. section 1.2.1).
The fact that there is much less evidence for *lÌ¥ than for *rÌ¥ makes it difficult to draw clear-cut conclusions. As we will see, many potential examples are inconclusive for various reasons: the etymology is not compelling (section 10.1), the full grade vowel slot may have been levelled (section 10.2), or *lÌ¥ is not reflected directly for another reason (section 10.3). A number of strong pieces of evidence for *lÌ¥ > âλαâ are discussed in section 10.4, and the possibility of a special development *lÌ¥ > âαλâ before nasals is examined in section 10.5. Finally, the scanty evidence from other dialects is treated in section 10.6.
10.1 Unknown, Doubtful, or Uncertain Etymologies
Since the etymology of the following words is doubtful or unknown, they will be left out of consideration:
-
á¼ÏλαÏÏον âcurved poop of a shipâ (Il., Hdt.);
-
γλάμÏν and Î³Î»Î±Î¼Ï Ïá½¹Ï âblear-eyedâ (com.);
-
θάλÏÏ âto heatâ (Od.+);
-
κάλÏη âtrotâ (Paus., Plu.);
-
κλαδαÏá½¹Ï âweak; handicappedâ (late);
-
λάξ adv. âwith the heelâ (Hom.+);
-
λαÏαÏá½¹Ï âslack, hollowâ (Hp. Arist.) and λαÏá½±Ïη âflank of the bodyâ (Il.+);
-
ÏλαδαÏá½¹Ï âhumid, damp; flaccidâ (Hp., A.R.), ÏÎ»Î±Î´á½±Ï âto be flaccidâ (Hp.+);
-
Ïλαδεá¿Î½ âto be rentâ (hapax, A. Choe. 28); ÏÎ»á½±Ï âto bruise, crushâ (Pi.+).
For discussion of these words, I refer to the standard etymological dictionaries.
Various middle perfect forms are analogical creations on the basis of present or aorist stems with a full grade root, e.g. á¼Ïὶ ⦠á¼Ïá½³ÏαλÏο (Hom.) to á¼ÏιÏá½³Î»Î»Ï âto enjoin, give a commandâ and á¼ÏÏαλμαι (Scut.+) to ÏÏá½³Î»Î»Ï âto prepare, equipâ. An analogical origin of á¼Ïá½³ÏαλÏο follows from the fact that Ïá½³Î»Î»Ï etymologically belongs to a root ending in a laryngeal, *telh2â âliftâ.
In other cases, discussed in alphabetical order in the following subsections, there are serious reasons to doubt a reconstruction with *l̥ that has been proposed by previous scholars.
10.1.1 αá½Î»Î±Î¾ and á¼Î»Î¿Î¾
These words are attested in the acc. sg. as αá½Î»Î±ÎºÎ± âfurrowâ (Hes., Pi.+), á¼Î»Î¿ÎºÎ± (trag.), ὦλκα (Hom.).1 The traditional etymology (GEW s.v. á¼Î»Î¿Î¾, LIV2 s.v. *h2u̯elkâ) derives these words from the same root as Lith. vilÌkti (1sg. velkù) âto drawâ, OCS 1sg. vlÄkÇ« âto dragâ, Av. varÉk- âto drawâ, which was reconstructed by Schindler (1972) as *h2u̯elkâ (with *h2â based on the Greek noun). Assuming that Hom. ὦλκα continues *á¼Ïολκα, this form has been derived, together with αá½Î»Î±ÎºÎ±, from a Proto-Greek ablauting paradigm containing the forms acc. sg. *au̯olk-mÌ¥, gen. sg. *au̯lÌ¥k-os. This is theoretically possible, but it would remain unclear why á¼Î»Î¿ÎºÎ±, attested in the tragedians with an alleged Aeolic vocalization to âλοâ, has no trace of digamma (cf. the preserved trace in Hom. ÏαλαύÏÎ¹Î½Î¿Ï < *tala-u̯rÄ«nos). To assume that á¼Î»Î¿Îºâ is a reshaping of *á¼Î¿Î»Îºâ (GEW, l.c.) is unmotivated. Moreover, various dialectal by-forms are attested: Dor. εá½Î»á½±Îºá¾± and the glosses αá½Î»á½±Ïα and á½Î»Î¿ÎºÎµÏ in Hsch. Since it is not possible to reduce these to one proto-form, the word is most probably a borrowing: Beekes (EDG q.v.) views it as Pre-Greek in view of the interchanges κ/Ï and word-initial α/οâ attested in the Hesychius glosses.2
10.1.2 γάλα
Beside γάλα, γάλακÏÎ¿Ï (Il.+) a few by-forms with a different root shape are found: γλακÏοÏá½±Î³Î¿Ï âwho live on dairyâ (Il. 13.6), name of a Scythian people (Hes. fr. 151), Î³Î»á½±Î³Î¿Ï n. âmilkâ (Il. 2.471 = 16.643, Pi. fr. 106.4), ÏεÏÎ¹Î³Î»Î±Î³á½µÏ âoverflowing with milkâ (Il. 16.642).3 There are also some glosses of unclear interpretation: κλάγοÏΠγάλα. ÎÏá¿ÏεÏ; γλακῶνÏεÏΠμεÏÏοὶ γάλακÏÎ¿Ï âfull of milkâ, and γλακκόνΠγαλαθηνόν âsucking milkâ (all Hsch.). The variation between γαλακÏâ and γλακÏâ can be explained as originating in the monosyllabic nominative *glakt > *gla > γάλα.4 The question is, then, whether the Greek forms with γλαâ must be derived from a pre-form with *lÌ¥.
Unfortunately, it is quite uncertain how the âmilkâ-word is to be reconstructed for PIE, and if it can be reconstructed at all. The most obvious comparandum is Lat. lac, lactis âmilkâ, which could be the outcome of a pre-form *glÌ¥gtâ if we assume the validity of Schrijverâs rule *CRDCâ > pre-Lat. *CRaDCâ.5 A second possible cognate is Class. Arm. kaÉ«cÊ¿ âmilkâ, which might reflect a nom. *glÌ¥Kt-s.6 Taken together, these words for âmilkâ could point to a pre-form *glÌ¥Ktâ (Armenian excludes a form with *dlâ). Finally, it has been suggested that this *glÌ¥gtâ was derived from the verbal root of Hitt. kalank-i âto soothe, appeaseâ (cf. also galaktar âa soothing substanceâ).7 Indeed, it is conceivable that milk, as the nourishment given to infants, was referred to as a soothing substance.8 Problematic, however, are the structure of the reconstructed root *glÌ¥gâ with two mediae, and the fact that word-initial *glâ should have been retained in Latin. The first problem could be addressed by reconstructing the root as *gleǵhâ, but in this case the Latin vocalism and the root shape of Greek γλάγοÏ, ÏεÏÎ¹Î³Î»Î±Î³á½µÏ would remain unexplained. The second problem could be resolved by reconstructing a different anlaut (*dlâ or *mlâ), or by assuming a dissimilation *glaktâ > Lat. lactâ.9 In view of these problems, it is best not to base any conclusions regarding the development of *lÌ¥ on the word for âmilkâ.
10.1.3 κλαγγή
The noun κλαγγή âpiercing sound, cryâ (Il.+) is also attested as a root noun (dat. sg.) κλαγγί (Ibyc.), and has given rise to a derived verb ÎºÎ»á½±Î¶Ï < *klang-i̯e/oâ, aor. κλάγξαι. Latin clangÅ âto cryâ (pres. only) has been compared, but if the word is onomatopoeic, it would be unwise to use it as evidence, because in that case the original form may have contained *a rather than *lÌ¥.10 Another possibly related form within Greek is the intensive perfect κέκληγα (Hom.+), with the aor. κλαγεá¿Î½ (B., E.).11 Nothing in this lemma decisively points to a pre-form with *lÌ¥.
10.1.4 λάÏιοÏ
The adjective λάÏÎ¹Î¿Ï (Il.+) means âhairy, shaggyâ (of animals, of the human chest); âovergrown, woodedâ (of land), cf. λαÏιαύÏην âwith hairy neckâ (h. Herm.).12 For the first meaning, the etymological dictionaries compare OIr. folt âhairâ < PClt. *u̯oltoâ; for the second, a Germanic word for âuncultivated field; woodâ (G. Wald, OE weald < *u̯óltuâ).13 In view of these, λάÏÎ¹Î¿Ï has been derived from an inherited noun PIE *u̯lÌ¥toâ with a suffix âιοÏ. There are, however, several issues with this reconstruction: we are dealing with a root etymology, and the zero grade is only attested in Greek. Moreover, the Balto-Slavic word for âpanicleâ (e.g. Lith. váltis f.), whose acute root points to *u̯olHâ and thereby excludes a comparison with λάÏιοÏ, is probably related to the Germanic word.14 It would therefore be unwise to draw conclusions concerning *lÌ¥ from λάÏιοÏ.
10.1.5 λαγαÏá½¹Ï and λαγÏá½¹Ï
The adjective λαγαÏá½¹Ï âhollow, sunken; thin, leanâ (Ion.-Att.; epigraphically at Cos) is clearly related within Greek to λαγών, attested mostly in the plural Î»Î±Î³á½¹Î½ÎµÏ âthe flanks of an animalâ (âsunken spotsâ). Furthermore, it is attractive to reconstruct λαγÏá½¹Ï âhareâ (Hom.) as PGr. *slag-ou̯s-óâ or *slÌ¥g-ou̯s-óâ âslack-eared [animal]â (cf. Peters 1980: 59). Outside of Greek, these forms are to be compared primarily with the Germanic group of ON slakr, OE slæk âweak, floppyâ < PGmc. *slakaâ < PIE *sloǵoâ.15
The further reconstruction of these words is muddled by the multitude of potential cognate forms. First of all, within Greek etymological dictionaries compare λαγαÏá½¹Ï with Î»á½±Î³Î½Î¿Ï âlustful, hornyâ (Arist.+) and its derivations Î»Î±Î³Î½Îµá½»Ï âto have intercourseâ, λαγνεία âintercourseâ.16 However, the semantic connection is weak; in my view, Î»á½±Î³Î½Î¿Ï is better derived from the root PIE *selǵ- âto let goâ (Ved. sarj ârelease, set freeâ, Av. harÉz) together with Cretan λαγαιÏ, aor. λαγαÏαι âto releaseâ and λαγάÏÏαιΠá¼Ïεá¿Î½Î±Î¹ âto let goâ (Hsch. λ 39).17 This root has a different full grade compared to the Germanic words reflecting *slakaâ (and hence also λαγαÏá½¹Ï), suggesting that the latter belong to a different root *slegâ âweak, slackâ.
The reconstruction of λαγαÏá½¹Ï is complicated further by the existence of words in other languages with more or less similar forms, but diverging semantics:
-
Lat. laxus âspacious, wide, looseâ with laxÅ, âÄre âto extend; relax; releaseâ (reflecting pre-Italic *slÌ¥g-s-oâ by Schrijverâs rule *RDC > RaDC),18
-
Lat. langueÅ âto be faint, be languidâ;
-
Ved. Ålaká¹£á¹Ã¡- âsmooth, slippery, softâ (AV+), MoP laÅ¡n âsmoothâ;19
-
Toch. A slÄkkär âsadâ, B slakkare âdartingâ.
Although the Tocharian forms have a similar appearance to Greek λαγαÏá½¹Ï, they are probably unrelated for semantic reasons.20 In my view, the appurtenance of the Indo-Iranian words for âsmoothâ is uncertain in view of the considerable semantic difference. On the other hand, I would propose to derive at least Lat. laxus from *selǵâ âto let goâ, because the derived verb laxÅ, âÄre means âto relax; releaseâ (cf. Cret. λαγαÏαι and Ved. sarj). Lat. langueÅ, however, is semantically close not to laxus, but to the Greek verbs Î»Î±Î³Î³á½±Î¶Ï âto give way, yieldâ and Î»Î¿Î³Î³á½±Î¶Ï âto loiter, waste timeâ, ascribed to Aeschylus, Aristophanes and Antiphanes in the lexicographical tradition.21 In Î»Î¿Î³Î³á½±Î¶Ï one might even see evidence for a different root with an internal nasal (perhaps to be connected with Lat. longus âlongâ, Goth. laggs âid.â).22 It is therefore attractive to reconstruct Lat. laxus as *sl̥ǵ-s-oâ and to disconnect it from langueÅ etymologically.23
It is possible to argue that λαγαÏá½¹Ï âhollow, leanâ and λαγÏá½¹Ï âhareâ must be compared primarily with PGmc. *slakaâ âweak, floppyâ, reflecting a zero grade *slÌ¥gâ, and that forms with an internal nasal (λαγγάζÏ, Lat. langueÅ) are to be derived from a different root. On the other hand, from a semantic perspective this would be arbitrary. Thus, no firm conclusions can be based on λαγαÏá½¹Ï, Î»Î±Î³á½¹Î½ÎµÏ and λαγÏá½¹Ï, as too many problems are involved in the reconstruction of the root. On the other hand, Î»á½±Î³Î½Î¿Ï âhornyâ (and derivatives) and Cretan λαγαιÏ, λαγαÏαι âto releaseâ derive from PIE *selǵâ and are strong pieces of evidence. On these forms, see further sections 10.4.5 and 10.6.1.
10.1.6 λάÏνη
The noun λάÏνη âfrizzy or curly hairâ (e.g. of a sheepâs fleece or the human chest) is traditionally reconstructed as PGr. *u̯lÌ¥k-snÄâ.24 A root *u̯olḱ- âhairâ is indeed attested in Balto-Slavic and Indo-Iranian (e.g. Ru. vólos âhairâ. Ved. válÅa- m. âsprout, twigâ), but the lack of precise cognate formations is disturbing (cf. above on λάÏιοÏ), and assuming a suffix âsnÄâ is an emergency measure. In fact, λάÏνη can be plausibly connected within Greek with the adjective λάÏεια (f.) âwoodedâ (Hom.), λαÏá½»-ÏÎ»Î¿Î¹Î¿Ï âwith a hairy rindâ (v.l. in Nic. Al. 269), and perhaps á¼Î¼ÏιλαÏÎ±á½·Î½Ï âto weedâ (Od.). The etymology of this second group has been extensively discussed by Lamberterie (1975; 1990: 732â742), who plausibly compares λόÏÎ¿Ï âambushâ < *âbush, thicketâ, and relates âνη in λάÏνη to the suffix âνοâ in Î¸á½±Î¼Î½Î¿Ï âthicketâ, ÏÏ ÎºÎ½á½¹Ï âcompact, close, thickâ. Against the reconstruction λάÏνη < *u̯lÌ¥k-snÄâ, he argues that an initial digamma is excluded by the Homeric attestations (1990: 733), and concludes that the Greek evidence points to a root λαÏâ / λοÏâ of unknown origin (1990: 741â742); it is therefore impossible to reconstruct a common PIE pre-form.
10.1.7 μαλθακόÏ
The adjective Î¼Î±Î»Î¸Î±Îºá½¹Ï âsoft, mild, weakâ (class.), Aeol. Î¼á½¹Î»Î¸Î±ÎºÎ¿Ï (Alc.), is supposed to be related within Greek to μάλθη (Hippon., Crat., S.), μάλθᾰ (Ar. fr. 157). The last-mentioned word is a technical term for a mixture of wax and pitch used for caulking ships, but it may also denote wax (S. Ichn. 140). From a semantic point of view, this comparison could work if we start from a basic meaning *âsoft stuffâ, but from a morphological perspective it is less evident. There is a derivative μάλθÏν (ascribed to Socrates by Stobaeus 4.15.16) which perhaps means âsoftieâ, as opposed to á¼ÏγάÏÎ·Ï in the sense of a hard-working man; this may indeed imply that âÎ±ÎºÎ¿Ï was later added as a suffix. However, an adjectival suffix âÎ±ÎºÎ¿Ï is not productive, and although influence of Î¼Î±Î»Î±Îºá½¹Ï âsoftâ on Î¼Î±Î»Î¸Î±Îºá½¹Ï is conceivable, this would be an additional assumption.
The meanings attested for Î¼Î±Î»Î¸Î±Îºá½¹Ï are diverse. It qualifies nouns referring to physical objects like soft soil, cushions, the skin, limbs, etc. More often, however, the word is used metaphoricallyâeither negatively (e.g. cowardly warriors) or positively (e.g. soothing words, mild sleep). In view of this, the often cited connection with the Germanic adjective for âmildâ, e.g. OHG milti âmercifulâ, Goth. *unmilds, is semantically quite attractive.25 However, in view of the problems just discussed, this root etymology is not more than a fairly remote possibility.26 Finally, it is not certain that the dialectal difference between Ion.-Att. Î¼Î±Î»Î¸Î±Îºá½¹Ï and Aeol. Î¼á½¹Î»Î¸Î±ÎºÎ¿Ï must be ascribed to a syllabic liquid: compare the dialectal distribution of καθαÏá½¹Ï and κοθαÏá½¹Ï âpureâ (section 9.7.2).
10.1.8 ÏÎ»á½±Î³Î¹Î¿Ï and ÏλάζÏ
The adjective ÏÎ»á½±Î³Î¹Î¿Ï (Pi.+) âathwart, oblique, sidewaysâ occurs in substantivized form as Ïá½° Ïλάγια âthe flanks/sidesâ, of the body but especially of an army (Hdt., Th.+). It has no established Indo-European etymology, and accordingly there is no unambiguous evidence that Ïλαγâ developed from *plÌ¥gâ. There are two possible cognates within Greek: the root Ïλαγγâ in ÏÎ»á½±Î¶Ï âto go astrayâ, and on the other hand Hom. á¼ÎºÏÎ±Î³Î»Î¿Ï âterrible, outrageousâ (if this was dissimilated from *âplag-loâ).
The verb ÏÎ»á½±Î¶Ï âto turn sth. away from, thwart, make deviateâ (act.), âto go astray, waverâ (mid.-pass.) is the epic and poetic counterpart of the prose form Ïλανάομαι. Frisk (GEW s.v.) compared this to Lat. plangÅ (plÄnxi, plÄnctus) âto beat, strike; mournâ, assuming that the Greek meaning âto drive astrayâ developed from âto beat off trackâ.27 However, the Greek comparandum to Lat. plangÅ is not ÏÎ»á½±Î¶Ï but ÏλήÏÏÏ, which has same the duality of meanings, âto beatâ and âto beat the chest, mournâ (cf. Goth. faiflokun 3pl. âbeat the chestâ). Frisk explains the root-internal nasal of the aorist á¼ÏλάγÏθη as imported from the present stem, but this assumption is gratuitous, as a root Ïλαγγâ underlies all stems of this verb (including ÏÎ»á½±Î¶Ï < *plang-i̯e/oâ, where the nasal disappeared by regular sound change). One could assume that an infixed present stem *plh2-n-gâ developed first to *plÄngâ, and then to the attested plangâ by Osthoffâs Law. However, in view of the semantic gap between âto beatâ and âto deviateâ, this etymology remains uncertain.
A better comparandum for some of the Greek words is a North-Germanic verb meaning âto swerveâ: ON flakka âto rove aboutâ, Far. flakka âto roamâ, which is derived by Kroonen from an o-grade iterative PGmc. *plog-neh2â (EDPG s.v. *flakkÅnâ). It is difficult to include ÏÎ»á½±Î¶Ï in this comparison in view of its root-internal nasal, but ÏÎ»á½±Î³Î¹Î¿Ï âathwartâ could constitute a more serious comparandum for the Germanic words.
There is, however, a second possibility: the interchanges between the various roots meaning âto go astrayâ et sim. can be taken as reflexes of a substrate origin.28 The attested root shapes are:
-
*plangâ > ÏλάζÏ, á¼ÏλάγÏθη âto drive off courseâ;
-
*plagâ > ÏÎ»á½±Î³Î¹Î¿Ï âathwartâ; á¼ÎºÏÎ±Î³Î»Î¿Ï âoutrageousâ
-
*a-m(b)lakâ > Att. á¼Î¼Ïλακεá¿Î½, á¼Î¼Î²Î»Î±ÎºÎµá¿Î½ âto errâ (trag.);
-
*mlÄkâ > βλάξ, gen. Î²Î»á¾±Îºá½¹Ï âstolid, stupidâ.
As there is no way to derive all these forms from an Indo-European root, and in view of the absence of clear cognates, it is a distinct possibility that they were all borrowed. Therefore, I will not use ÏÎ»á½±Î³Î¹Î¿Ï and ÏÎ»á½±Î¶Ï in this discussion.
10.1.9 ÏλάÏÏÏ
The verb ÏλάÏÏÏ âto knead; mold, shape, formâ (Hes.+) has no ascertained etymological comparanda (cf. GEW, DELG s.v.), and Beekes (EDG) even considers a Pre-Greek origin. That the root ended in âθâ is shown by the compound κοÏοÏÎ»á½±Î¸Î¿Ï âmodeler of figurinesâ (Pl., Isoc.) and Ïλάθανον âcake moldâ (Theoc.). Compounds such as καÏαÏλάÏÏÏ âto smear, plaster overâ (Hdt., Ar.+) illustrate that the connection with malleable materials such as dough and plaster is old.
The non-ablauting root Ïλαθâ in combination with a yod-present suggests that the verb is denominative. I would like to propose an etymological connection with the PIE root *bhlendh- âto mingle; become turbidâ which is reflected in Germanic (Goth. blandan (sik) âto mix, mingleâ, ON blanda âto blend, mixâ, cf. EDPG s.v. *blandanâ) as well as in Balto-Slavic (cf. notably Lith. blÄÌsti (1sg. blendžiù) âto sleep, stir flour into soup, talk nonsense, become cloudyâ < *bhlendh-i̯e/oâ). The Germanic strong verb is suggestive of an Indo-European origin.
If we start from an original meaning âto mix flour (dust, sand) through a liquid; make turbidâ (as in the meaning âstir flour into soupâ of Lith. blÄÌsti), we may suppose that an early form of Greek had reflexes of nominal derivatives such as *bhlnÌ¥dh-tóâ (cf. ÏλαÏÏá½¹Ï) or *bhlnÌ¥dh-éh2â (cf. the form Ïλαθά âmodelled figureâ mentioned as Doric in Plutarch) denoting a dough or wall-plaster. Starting from such a form it would be possible to create a denominative verb *bhlnÌ¥dh-i̯e/oâ denoting the process of working dough or plaster, i.e. âto knead; smearâ.
A problem for this reconstruction is the fact that Grassmannâs Law has applied in ÏλάÏÏÏ, ÏλαÏÏá½¹Ï, ÏλάÏμα and all other derivatives (instead of expected xÏλάÏÏÏ, xÏλαÏÏá½¹Ï, xÏλάÏμα, etc.). By itself, a deaspirated word-initial stop spreading through all derivatives containing the root would not be shocking: cf. ÏιÏÏá½¹Ï, Ïá½·ÏÏιÏ, ÏιÏÏ Î½á½¹Ï with the root of Ïείθομαι âto give ear to, obeyâ, and paradigmatic forms like á¼ÏειÏα and Ïá½³ÏειÏμαι. However, in that case there was a clear basis of forms where Grassmannâs Law did operate: the verbal stems ÏείθÏ, Ïείθομαι, á¼Ïιθόμην, as well as the old perfect Ïá½³Ïοιθα.
This issue could be resolved by assuming that the denominative verb was derived from a nominal form such as *Ïλαθή (cf. Dor. Ïλαθά mentioned above) after Grassmannâs Law had applied there. A comparable case seems to be Ïεá¿Ïμα n. âropeâ: its root no doubt reflects PIE *bhendhâ âto bindâ, but the verb is absent from Greek. In fact, this form suggests another possibility: Grassmannâs Law may not have operated in forms where âθμâ was preserved relatively long, such as *ÏεÏÎ»Î±Î¸Î¼á½³Î½Î¿Ï > *ÏεÏÎ»Î±Î¸Î¼á½³Î½Î¿Ï (later >(>) ÏεÏλαÏμένοÏ) or *Ïλάθμα > *Ïλάθμα (later >(>) ÏλάÏμα). If one is prepared to accept this possibility, it is attractive to connect ÏλάÏÏÏ with the root *bhlendhâ in the way just described.
10.1.10 ÏÎºÎ±Î»Î¼á½¹Ï and Ïκάλμη
ÏÎºÎ±Î»Î¼á½¹Ï m. âthole, i.e. the pin by which the oar was fastened to the ÏÏÏÏηÏá½µÏâ (h. Hom., A.+), Ïκάλμη âa type of knife or daggerâ (S. fr. 620, Hsch.). A possible connection with PGmc. *skalmaâ, *skalmÅâ as attested in various concrete meanings (e.g. ON skÇ«lm âtip of a forkâ, OHG scalm âcanoeâ) is mentioned by DELG and GEW (both s.v. ÏκαλμόÏ). Frisk suggests that the Greek words have an identical origin with these Germanic words (i.e. PIE *skol-moâ, *skol-meh2â), but that their vocalism was secondarily influenced by that of the verb ÏÎºá½±Î»Î»Ï âto hoe, stir upâ in a more original meaning such as âto split offâ (âhat sich nach ÏÎºá½±Î»Î»Ï gerichtet, u. zw. in einem ursprünglicheren Sinn von âspaltenâ o. ä.â). Chantraine (DELG s.v. ÏκαλμόÏ) is slightly more vague, but agrees that the vocalism may have been influenced by ÏκάλλÏ. Beekes (EDG s.v. ÏκάλλÏ) apparently views ÏÎºÎ±Î»Î¼á½¹Ï and Ïκάλμη as inner-Greek derivatives of ÏÎºá½±Î»Î»Ï and does not mention the possibility of comparing the Germanic words.
In any case, since the PIE root *skelH- âsplit, slitâ is now reconstructed with a laryngeal on account of Lith. skélti âto split; strike fireâ, Hitt. iÅ¡kalla-i âto slit, splitâ (cf. Kloekhorst, EDHIL s.v. iÅ¡kalla-i, who argues that the laryngeal was *h2 or *h3), the root of ÏÎºÎ±Î»Î¼á½¹Ï and ÏÎºá½±Î»Î»Ï cannot reflect a pre-form with *lÌ¥ (pace LIV2 s.v. *skelâ, where ÏÎºÎ±Î»Î¼á½¹Ï is cited as the main reason to posit a laryngeal-less root). Beekes (EDG s.v. ÏκάλλÏ) envisages whether ÏÎºá½±Î»Î»Ï may reflect an inherited *sklÌ¥H-i̯e/oâ (with loss of laryngeal by the so-called âPinault Effectâ) or a nasal present *sklÌ¥-neHâ (cf. βάλλÏ, on which see section 10.5.1). In my view, it is more likely that ÏÎºá½±Î»Î»Ï is a denominative, cf. forms like ÏÎºÎ±Î»á½·Ï âpickaxeâ (Att. inscr., 4th c. BCE) and the related denominative ÏÎºÎ±Î»Îµá½»Ï âto stir; poke (the fire)â (Ar.+), forms in which Ïκαλâ could well reflect a prevocalic zero grade *sklÌ¥Hâ. Thus, ÏÎºá½±Î»Î»Ï may reflect a denominative PGr. *skal-i̯e/oâ.
Finally, note that ÏÎºÎ±Î»Î¼á½¹Ï has also been compared to PGmc. *helmanâ > OE helma ârudderâ (E. helm), ON hjalm-vÇ«lr âid.â, which is semantically very close. However, this connection is uncertain, as the Germanic words lack the initial sâ and could instead be connected with the word for âstalk, reedâ, *ḱlh2-mâ (Kroonen 2011: 162â163, EDPG s.v. *helmanâ).
10.1.11 ÏÏλάγÏνα
ÏÏλάγÏνα (n. pl.) âentrails, visceraâ (Hom.+) refers to a collection of innards, âespecially heart, lungs, liver, kidneys, which in sacrifices were reserved to be eaten by the sacrificers at the beginning of their feastâ (LSJ). This word is clearly related to YAv. spÉrÉzan- m. âspleenâ, nom. sg. spÉrÉza, Lith. blužnìs âid.â, and within Greek to ÏÏλήν âspleenâ (Il.+). The difficulty to reconstruct a PIE pre-form on the basis of these and other related terms for the spleen is well-known:29 the lack of a root-final velar in Greek ÏÏλήν is mostly assumed to be due to taboo deformations.30
In Friskâs view, ÏÏλάγÏνα stands for earlier *ÏÏλάÏνα, with a secondary internal nasal.31 This collective would reflect a PIE Transponat *spl̥ǵh-n-h2, but is probably not old: the comparative evidence points to a specific denomination of the spleen, so to an original singular form. Therefore, ÏÏλάγÏνα probably contains the weak stem of the PIE paradigm, e.g. gen. sg. *spl̥ǵh-n-ós, and is likely to contain a regular vocalization to âλαâ.32 I see no particular reason to assume that the vowel slot of ÏÏλάγÏνα was influenced by that of ÏÏλήν.33 On the other hand, it would be unwise to base any conclusions on ÏÏλάγÏνα, because most of its cognates in other IE languages have undergone irregular deformations.
10.1.12 ÏαλλόÏ
ÏÎ±Î»Î»á½¹Ï m. âpenisâ is attested in the classical language from Hdt. and Ar. onwards. There are possible cognate forms in two other branches: in Celtic we find OIr. ball âmemberâ, ball ferda gl. membrum virile < PClt. *balnoâ, and possibly also W. balleg âsack, purseâ. Latin has follis âbag; testiclesâ, which may derive from *bholniâ or *bhlÌ¥niâ. It is possible to posit a pre-form PIE *bhlÌ¥nóâ, which would account for Gr. ÏÎ±Î»Î»á½¹Ï as well as the Celtic words.34 While it is true that the word lies in a sphere of taboo, there is no principled reason to doubt the validity of this comparison. However, Greek also has another synonymous word, Ïá½±Î»Î·Ï âηÏÎ¿Ï (also Ïαλá¿Ï âá¿ÏοÏ) m. âpenisâ. Since there is no obvious way to derive this variant from a pre-form in *bhlÌ¥â, it is doubtful whether ÏÎ±Î»Î»á½¹Ï must reflect *bhlÌ¥nóâ.35
10.2 Cases of âλαâ and âαλâ Influenced by a Full Grade Form
The outcome of a number of forms with *l̥ provides evidence for the color of the anaptyctic vowel, but not necessarily for its place, because the full grade slot may have been introduced in the vocalized zero grade.
10.2.1 á¼ÏαλÏνοÏ, á¼ÏÏÎ±Î»á½³Î¿Ï and á¼Î»ÏνιÏÏοÏ
A root shape á¼Î»Ïâ is found in the following forms:
-
á¼ÏαλÏνοÏ, only in Pi. Pyth. 8.84 (modifying νόÏÏοÏ), glossed as âcheerful, happyâ by LSJ but possibly rather meaning âhoped forâ; the adjective seems derived from the verb á¼ÏέλÏομαι âto hopeâ.
-
á¼Î»ÏνιÏÏοÏ, a superlative attested in the scholia to Pi. Isthm. 5.12, where the mss. have the corrupt (while unmetrical) form á¼Î½á½³Î»ÏιÏÏοÏ. The passage reads: âthere are truly two things alone that foster the finest sweetness (á¼ÏÏον ⦠Ïὸν á¼Î»ÏνιÏÏον) of life in blossoming prosperity: (â¦)â. Wackernagel (1910) suggested to correct the form to á¼Î»ÏιÏÏοÏ. This form is indeed found in Aeschylus (Pers. 982), where it was traditionally interpreted as a proper name á¼Î»ÏιÏÏÎ¿Ï carried by a high-ranking Persian officer who is called âeyeâ of the King. In his edition of the Persae, West proposes to read an appellative á¼Î»ÏιÏÏον. This is attractive, because Pindarâs phrase á¼ÏÏον ⦠Ïὸν á¼Î»ÏνιÏÏον is directly mirrored in the Aeschylean passage: ΠεÏÏᾶν Ïὸν á¼ÏÏον, Ïὸν Ïὸν ÏιÏÏὸν ÏάνÏâ á½Ïθαλμόν, Î¼Ï Ïία Î¼Ï Ïία ÏεμÏαÏÏάν, ÎαÏανώÏÎ¿Ï Ïαá¿Î´â á¼Î»ÏιÏÏον (â¦).36 In my view, the âνâ in Pindarâs á¼Î»ÏνιÏÏον could be caused by an attempt (by scholiasts or grammarians?) to explicitly connect á¼Î»ÏιÏÏον to á¼ÏαλÏνοÏ.
-
á¼ÏÏÎ±Î»á½³Î¿Ï âwith pleasureâ (Od.), probably with dissimilation λâ¦Î» > Ïâ¦Î» and folk-etymological aspiration taken from á¼ÏÏá½±Î¶Ï âto rob; snatch awayâ. Indeed, the meaning of á¼ÏÏÎ±Î»á½³Î¿Ï may have been influenced by that of á¼ÏÏá½±Î¶Ï already in Homer, where á¼ÏÏÎ±Î»á½³Î¿Ï occurs three times.37 The non-dissimilated form is attested in the gloss á¼Î»ÏαλέονΠá¼Î³Î±ÏηÏόν âcherishedâ (Hsch.).
As for the etymology of these adjectival forms, it is commonly accepted that their root reflects the zero grade of á¼Î»Ïομαι âto surmise, reckon; expect, hopeâ.38 Possibly, Pindarâs á¼ÏαλÏÎ½Î¿Ï was derived directly from the verb (á¼ÏέλÏομαι âto hopeâ, cf. Hom. á¼ÏιέλÏομαι âto aspire toâ) when the root was still capable of undergoing ablaut; for deverbal âÎ½á½¹Ï cf. ÏεÏÏÎ½á½¹Ï âagreeableâ (Ïá½³ÏÏομαι âenjoyâ). It must also be taken into account that adjectives in âÎ±Î»á½³Î¿Ï and âÎ½á½¹Ï occur more often as a pair: cf. ÏμεÏÎ´Î½á½¹Ï âterribleâ beside ÏμεÏÎ´Î±Î»á½³Î¿Ï âid.â (quasi-opposite in meaning to á¼ÏÏαλέοÏ), and post-Hom. á¼°ÏÏÎ½á½¹Ï âwithered, thin, leanâ beside Hom. á¼°ÏÏÎ±Î»á½³Î¿Ï âwithered, dryâ (hapax).
For these reasons, it is likely that a simplex *á¼Î»ÏÎ½á½¹Ï < *u̯alp-nóâ once existed. Under its influence, an original superlative *u̯elp-istoâ may have been reshaped as *u̯alpistoâ (cf. section 4.1.2). As a deverbal adjective, *u̯alp-nóâ may owe its vocalism (instead of expected *u̯lapâ < *u̯lÌ¥pâ) to the influence of verbal forms with *u̯elpâ, or to the comparative and superlative. For this reason, á¼ÏÏÎ±Î»á½³Î¿Ï and á¼Î»ÏιÏÏÎ¿Ï cannot be used as cogent evidence for a regular *lÌ¥ > âαλâ.39
10.2.2 γλάÏÏα
An Eastern Ionic by-form of γλῶÏÏα âtongueâ is γλάÏÏα, attested in late literary Ionic in Herodas (a Hellenistic, 3rd c. BCE mimographer who imitated the language of Hipponax). The authenticity of γλάÏÏα is guaranteed by its occurrence in inscriptions from Asia Minor, where it denotes the tongue as a part of a sacrificed animal. Possibly, γλάÏÏα was preserved beside γλῶÏÏα in Eastern Ionic because of its semantic specialization. It may continue the original form of the motional feminine *dlÌ¥kh-i̯a, which was derived from the weak stem of a root noun *dlÅǵhâ, *dl̥ǵhâ reflected in γλῶÏÎµÏ âbeard of cornâ (Scut., cf. Hom. γλÏÏá½·Ï âbarb of an arrowâ). Subsequently, γλάÏÏα may have been reshaped, under the influence of γλῶÏÎµÏ or γλÏÏá½·Ï, to γλῶÏÏα, which was the only form to survive in Classical Greek. It cannot be excluded that the outcome âλαâ in γλάÏÏα < *dlÌ¥kh-i̯a was influenced by the vowel slot of cognate words like γλῶÏÎµÏ or γλÏÏá½·Ï. Therefore, Eastern Ionic γλάÏÏα is not a certain example for the regular development of *lÌ¥ in Ionic-Attic.
10.2.3 ÏλαÏá½»Ï
The adjective ÏλαÏá½»Ï âbroad; flatâ is quoted as a prime example for the development of *lÌ¥ in almost every manual. Its forms of comparison are secondary (ÏλαÏá½»ÏεÏÎ¿Ï and âÏαÏοÏ).40 The adjective is also attested in Lesbian poetry (ÏλάÏÏ Alc. fr. 74).41 Other forms attested in Greek with this root are ÏλαÏαμών m. âflat stone or objectâ, ÏλάÏÎ¿Ï n. âbreadth, width; plane surfaceâ (Cypr. fr. 1.2, Simon., Hdt.+), and adjectives in âÏλαÏá½µÏ (X., Th., Arist.). The old form of the adjectival feminine is probably reflected in the toponym ΠλάÏαια. It is possible that âλαâ directly reflects *lÌ¥, but it cannot be excluded that the vocalization was influenced by a now-lost full grade reflex *pleth2- > PGr. *plet(a)â (cf. Ved. práthasâ n., práthate) that was originally present in the forms of comparison, or by the older form *pletos of the neuter abstract. After this, the stem form ÏλαÏâ would have spread from the adjective to all other derivatives. Therefore, ÏλαÏá½»Ï and related âCaland systemâ forms do not offer absolutely compelling evidence for the regular reflex of *lÌ¥.
10.3 The Pre-form Did Not Necessarily Contain *l̥
10.3.1 βλάβομαι, βλάÏÏÏ
As I have argued extensively elsewhere (Van Beek 2017b), the root of βλάÏÏÏ âto hinder; harmâ must be reconstructed as *mlÌ¥kwâ in view of the etymological comparison with Ved. marcáyati âto slander, injureâ (caus.) and OAv. mÉrÉc âto destroyâ (vel sim.).42 This connection is accepted by both GEW and DELG (s.v. βλάβη) and is supported by phraseological material.43 I will now summarize the arguments; for the details the reader is referred to Van Beek 2017b.
There are two old formations in the verbal paradigm: the thematic root present βλάβεÏαι âto be distracted (of a speaker); to give way (of the knees)â, attested only in the Iliad, and the intransitive aor. á¼Î²Î»á½±Î²Î·Î½ âwas impededâ (beside younger á¼Î²Î»á½±Ïθην)44 attested from Homer onwards.45 Compared to these intransitive forms, the causative active paradigm βλάÏÏÏ, βλάÏαι is clearly secondary. As for the nominal forms, there are two archaic-looking formations. First, the compound á¼Î²Î»Î±Î²á½µÏ âunharmed; unwavering, securelyâ is old within Greek (it is also attested in Cretan: see below) and could be compared with the root compound in *âmlÌ¥kwâ underlying Old Avestan a.mÉrÉxÅ¡ âwhich does not harmâ, ahu.mÉrÉxÅ¡ âharming lifeâ.46 Secondly, βλάβη âharm; curseâ (A.+) may reflect an old root noun with later addition of *âÄ in Proto-Greek.47 Other nominal derivatives follow productive patterns and may be relatively recent creations.48
The oldest meanings of βλάÏÏÏ are âto hinder, impedeâ and âto misleadâ (Il.+); the meaning âto damageâ first appears after Homer. A second remarkable use of βλάÏÏÏ is found in Hesiod: the verb means âto slander, pronounce a false oathâ, i.e. it refers to deceiving someone else with crooked words.49 The meaning âto speak falsely or deceptivelyâ arose metaphorically from âto put off track, misleadâ (with words). It is probably of PIE age in view of the corresponding phrase marcáyati dváyena âleads astray with double tongueâ attested in the Rigveda. Similar phraseology must underlie the use of the adverb á¼Î²Î»Î±Î²á½³ÏÏ âsincerely, without deceivingâ in traditional oath formulae in Thucydides and Attic inscriptions (e.g. IG I3 53.13â14).
A well-known problem with this etymology is the root-final âβâ in Ionic-Attic. Interestingly, forms with root-final âÏâ (in harmony with the root reconstruction *melkwâ) are found in Cretan (for the attestations, see Bile 1988):
-
inf. καÏαβλαÏεθαι, rendered as âêtre leséâ (Gortyn, early 5th c. BCE; mid.-pass. inf. âεθαι is regular from âεÏθαι);
-
abstract αβλοÏια (Gortyn, Axos); αÏλοÏια (Lyttos), rendered as âconduite qui ne fait tort à personneâ (DELG s.v. βλάβη);50
-
á¼Î²Î»Î¿Ïá½³ÏÎ á¼Î²Î»Î±Î²á½³Ï. ÎÏá¿ÏÎµÏ (Hsch.).
For αβλοÏια, Chantraine (1933: 79) compared the near-synonym á½ Ïελία âservice; behavior which benefitsâ. Since an older form á½ Ïέλεια (derived from the s-stem forms á½ÏελοÏ, âÏÏελήÏ) is attested beside á½ Ïελία, he suggested that αβλοÏια can be derived from the s-stem attested in á¼Î²Î»Î±Î²á½µÏ and the gloss á¼Î²Î»Î¿Ïá½³ÏÎ á¼Î²Î»Î±Î²á½³Ï. ÎÏá¿ÏÎµÏ (Hsch.). Since all attestations of αβλοÏια are from the 6th or 5th c. BCE and from various different regions of Crete, it is probably a traditional legal term. The form αÏλοÏια at Lyttos may be due to the sound change *Dlâ > Tlâ also observed in ÎºÎ»ÎµÏ ÎºÎ¿Ï ânew wineâ (in the same inscription as αÏλοÏια) and in the Cretan gloss ÎºÎ»á½±Î³Î¿Ï âmilkâ (Hsch.).
In view of this Cretan evidence, it must be asked whether βλαβâ in Ionic-Attic can be secondary. It would be ad hoc to assume a distance assimilation βλαÏâ > βλαβâ for Ionic-Attic.51 An analogical explanation of the âβâ is out of reach, as most verbs in âá½±ÏÏÏ have a stem ending in âÏâ (cf. á¼Ïá½µ, βαÏá½µ, Ïá½±ÏοÏ, etc.). While some remodeling took place in derived verbs with occlusive-final roots (for example, ÏλήÏÏÏ âto strikeâ replacing the reflex of *plÄgi̯e/oâ, cf. Barber 2013: 262â269), it is usually the yod-present stem that adapts its consonantism. Moreover, a labiovelar would have lost its labial feature before yod early on (cf. Î½á½·Î¶Ï âto washâ), so βλάÏÏÏ must have been reshaped by analogy at some point anyway. Therefore, βλάβεÏαι and βλάβη probably preserve an old reflex of the root-final stop.
A second issue is the difference in root vocalism between καÏαβλαÏεθαι and αβλοÏια. Chantraine explains âλοâ as a pre-Doric dialectal reflex of *lÌ¥, but this seems ad hoc (both root shapes are attested in Gortyn). The stem-formation of καÏαβλαÏεθαι must be identical to that of Hom. βλάβεÏαι since word-internal âÏÏâ (< *âpi̯â) was originally preserved in Cretan (it was later assimilated to âÏÏâ).52 There are, then, two issues: the difference between βλοÏâ and βλαÏâ in Gortynian Cretan, and the divergence in the root-final stop between Cretan and Ionic-Attic.
Both issues can be resolved in the same way. In Van Beek 2017b, I proposed to compare Hom. βλάβεÏαι to the athematic nasal infix present *ml-n-kwâ reflected by Old Avestan forms like 3pl. mid. vÄ«-mÉrÉá¹caitÄ.53 The idea is that in PGr. *mlnÌ¥Ìkwe/oâ, the root-final stop was voiced after an accented syllabic nasal (> *mlnÌ¥Ìgwe/oâ), which was later vocalized (> *mlágwe/oâ > βλάβε/οâ). This sound change *ânÌ¥ÌTâ > *ânÌ¥ÌDâ is an extension of the rule *ânÌ¥Ìtâ > *ânÌ¥Ìdâ proposed by Olsen (1989).54 The thematic nasal infix present *mlnÌ¥kw-e/oâ underlying βλάβεÏαι could be compared to λάμÏÏ âto shineâ < *lh2-n-p-e/oâ (for the root *leh2pâ, cf. Lith. lópÄ âtorchâ).55
Table 26
Reflexes of PIE *mlÌ¥kwâ in Greek
|
Proto-Greek |
Ionic-Attic |
Cretan |
|---|---|---|
|
*mln̥kw-e/o- |
βλάβεÏαι |
*βλάβεÏθαι >> βλαÏεθαι |
|
*mlÌ¥kwâ >> *mlÌ¥kw-Ä- |
*βλάÏη >> βλάβη |
|
|
*nÌ¥-mlÌ¥kwâ >> *nÌ¥-mlÌ¥kw-es- |
*á¼Î²Î»Î±Ïá½µÏ >> á¼Î²Î»Î±Î²á½µÏ |
á¼Î²Î»Î¿Ïá½³Ï |
This voicing rule may help us explain the divergences between Ionic-Attic and Cretan in the following way. If Greek inherited both *mlnÌ¥Ìkwe/oâ (> βλάβεÏαι) and a root noun *mlÌ¥kwâ (cf. βλάβη), we may assume that Ionic-Attic preserves the regular outcome of the primary nasal present in Homeric βλάβεÏαι, while the outcome of the root noun *mlÌ¥kwÄ was aligned with the verbal stem, yielding βλάβη for expected *βλάÏη. The aorist βλαβá¿Î½Î±Î¹ may have secondarily taken over the root of the present stem. In Cretan, on the other hand, the root-final consonant of the verb καÏαβλαÏεθαι may have been influenced by the primary noun or other forms without the original nasal infix (cf. αβλοÏια, á¼Î²Î»Î¿Ïá½³Ï); these latter forms may show the regular zero grade reflex. These developments are shown in Table 26.
This scenario may account for the existence of two root allomorphs βλοÏâ and βλαÏâ in Gortyn without resorting to unmotivated borrowing from a pre-Doric (Achaean) substrate (as is done by, e.g., Chnatraine in DELG s.v. βλάβη). For we may now assume that βλαÏâ has the reflex of *nÌ¥, while βλοÏâ directly reflects *mlÌ¥kwâ, with âλοâ as the regular outcome of *lÌ¥ between two labial consonants. Such a reflex of *lÌ¥ in Cretan would be paralleled by the reflex âοÏâ < *rÌ¥ in this dialect after labial consonants (section 3.1.2). Moreover, the reflex âλοâ with an anaptyctic vowel after the liquid would be at variance with the development of *rÌ¥ in Cretan.56 Seen in this light, it is indeed likely that καÏαβλαÏεθαι does not contain the reflex of *mlÌ¥kwâ but that of *mlnÌ¥kwâ, and that the vowel slot of the vocalized zero grade βλοÏâ < *mlÌ¥kwâ was influenced by that of βλαÏâ.
In sum, for Ionic-Attic no definite conclusions can be based on Hom. βλάβεÏαι (because its âαâ may reflect a syllabic nasal), nor on á¼Î²Î»Î±Î²á½µÏ, βλάβη or βλαβá¿Î½Î±Î¹ (because they may have been influenced by βλάβεÏαι). The only significant conclusion to be drawn is that âλοâ or âολâ (in á¼Î²Î»Î¿Ïá½³Ï and αβλοÏια, possibly with analogical vowel slot) was probably the Cretan outcome of *lÌ¥ in a labial environment.
10.3.2 διÏλάÏιοÏ
The adjective διÏλάÏÎ¹Î¿Ï âtwofold, double the size, twice as muchâ is first attested in Solon (fr. 13.73 W), and it is common in Attic prose.57 It may originally be a legal term: cf. διÏλάÏÎ¹Î¿Ï Î¶Î·Î¼á½·Î± âdouble the fineâ, also found in Arcadian (IG V,2 6.35, also in Dubois 1988, Tegea 4.18) and in Elis, where it could be due to Koine influence (Minon 2007, I: 208). The Ionic form διÏλήÏÎ¹Î¿Ï is attested in Herodotus and inscriptions; its âηâ may be analogical after a semantically close form like ÏαÏαÏλήÏÎ¹Î¿Ï âabout the same size, about equalâ (from the root *pelh2â of Ïá½³Î»Î±Ï ânearâ).58
In Classical Greek, the meaning of διÏλάÏÎ¹Î¿Ï âdouble the sizeâ is different from that of διÏλόÏ, διÏλόοÏ, contracted διÏÎ»Î¿á¿¦Ï (Hom., Pi., trag., etc.), which means âdouble, twofoldâ in the sense of âconsisting of two discrete entitiesâ. διÏÎ»á½¹Ï clearly represents older *dui-pl-oâ as in Lat. duplus (< *du-pl-oâ), simplus, also in Goth. tweifls âdoubtâ, Lyc. tbiplẽ âtwice(?)â, OIr. dÃabul âdoubleâ.59 The root is also present in PGmc. *âfalþaâ â-foldâ (Goth. âfalþs, MoG. âfalt < *âpol-toâ). For διÏλάÏιοÏ, on the other hand, the etymological dictionaries (Boisacq 1916, GEW, DELG and EDG) posit an earlier *δίÏλαÏοÏ, enlarged by a suffix âÎ¹Î¿Ï (like e.g. á¼Î¼Î²Ïá½¹ÏÎ¹Î¿Ï beside á¼Î¼Î²ÏοÏοÏ).60 This *δίÏλαÏÎ¿Ï would continue a compound *dui-plÌ¥-toâ from the same root *pelâ âfoldâ as *dui-pl-oâ.
Upon closer scrutiny, however, it appears that διÏλάÏÎ¹Î¿Ï may have been created within the history of Greek, because there is also a verb διÏÎ»á½±Î¶Ï âto be twice as bigâ (S. Aj. 268 Ïá½¹ Ïοι διÏλάζον μεá¿Î¶Î¿Î½ κακόν). This denominative verb may have been derived from διÏÎ»á½¹Ï or its n. pl. διÏλά. For the subsequent derivation of διÏλάÏÎ¹Î¿Ï from διÏλάζÏ, cf. Î¸Î±Ï Î¼á½±Î¶Ï â Î¸Î±Ï Î¼á½±ÏιοÏ, á¼ÏÏάζομαι â á¼ÏÏá½±ÏιοÏ. Thus, the derivational chain is διÏÎ»á½¹Ï âdoubleâ (Hom.+) â διÏÎ»á½±Î¶Ï âto be twice as bigâ (trag.) â διÏλάÏÎ¹Î¿Ï âdouble the size, twice as bigâ (Thgn.+), whence later â διÏλαÏÎ¹á½±Î¶Ï âto doubleâ (Pl. Leg. 920a). If the only old form in Greek is διÏλόÏ, then διÏλάÏÎ¹Î¿Ï must not be compared directly with E. (two)âfold.
10.4 Promising Evidence for *lÌ¥ > âλαâ
10.4.1 βλαδεá¿Ï and βλαδαÏá½¹Ï
A root βλαδâ reflecting *mlÌ¥dâ is attested in the following glosses, all from Hsch.:
-
βλαδεá¿ÏÎ á¼Î´á½»Î½Î±Ïοι. á¼Î¾ á¼Î´Ï νάÏÏν âweakâ;
-
βλαδαÏόνΠá¼ÎºÎ»ÎµÎ»Ï μένον, Ïαῦνον âflaccid, porousâ;
-
βλαδόνΠá¼Î´á½»Î½Î±Ïον âweakâ.
In addition, the same root might be contained in the following glosses from Hsch., even if connecting them is less obvious from a semantic perspective:
-
βλαδαÏάΠá¼ÏÏα. μÏÏá½±. ὠμά âuntimely; dull, stupid; rawâ;
-
βλάδανΠνÏθÏá¿¶Ï âslothfulâ.
Since the PIE full grade was *meldâ (see section 4.4), βλαδâ must be the regular outcome of zero grade *mlÌ¥dâ. The form βλαδεá¿Ï (from an unattested sg. βλαδύÏ) seems the most archaic, as it would directly reflect the PIE adjective *mlÌ¥d-úâ (Ved. mrÌ¥dú- âsoft, delicateâ, Lat. mollis âsoft, gentleâ).
In addition to these forms, á¼Î¼Î±Î»Î´á½»Î½Ï (with secondarily added á¼â) is likely to be based on another reflex of the u-stem adjective, with an alternative vocalization. The problems with the two coexisting vocalizations Î²Î»Î±Î´á½»Ï and *(á¼)Î¼Î±Î»Î´á½»Ï have been discussed in section 4.4. Since ÏλαδαÏá½¹Ï âdamp, weak, flaccidâ (cf. ÏÎ»Î±Î´á½±Ï âto make flaccidâ) looks like the Ionic-Attic vernacular form corresponding to βλαδαÏá½¹Ï, I suggested there that ÏλαδαÏá½¹Ï may have arisen from βλαδαÏá½¹Ï by contamination with a semantically close word, such as ÏλάÏÏÏ âto kneadâ. If one were to assume that βλαδαÏá½¹Ï and Î²Î»Î±Î´á½»Ï stem from a non-Ionic-Attic dialect, *(á¼)Î¼Î±Î»Î´á½»Ï would continue the u-stem adjective, with levelling of the full grade slot; but even in this case, Attic ÏλαδαÏá½¹Ï would indirectly continue the outcome of *mlÌ¥dâ.
10.4.2 βλαÏÏá½¹Ï
According to the etymological dictionaries, the thematic aor. βλαÏÏεá¿Î½ âto sprout, budâ (Pi.+), with the derived pres. βλαÏÏάνÏ, has no etymology. In the meantime, Lamberterie (1990: 358â361) proposed to derive it from the noun βλαÏÏá½¹Ï âsprout, young shootâ (Hdt.+), which he reconstructs as a substantivized adjective *mlÌ¥d-tóâ âtender, youngâ. As a parallel, he points out that PIE *meld- âsoft, weakâ also served as the basis for a word for soft or tender shoots in Slavic (*moldÑ âyoung, tenderâ > OCS mladÑ, Ru. molodój, etc.). The derivation of a thematic aorist βλαÏÏεá¿Î½ from βλαÏÏá½¹Ï yields some difficulties. Lamberterie proposes to compare βλαÏÏá½¹Ï with Hom. Î¸Î±Î»Î»á½¹Ï âid.â, which seems to be derived from the present stem of Î¸á½±Î»Î»Ï âto flourishâ. On this basis, a verb *βλάÏÏÏ, impf. á¼Î²Î»Î±ÏÏον, aor. á¼Î²Î»á½±ÏÏηÏα would have been back-formed, after which the imperfect á¼Î²Î»Î±ÏÏον was reinterpreted as a thematic aorist. The assumed switch of aspect is not without problems, but the idea to derive βλαÏÏá½¹Ï from *mlÌ¥d-tóâ is intuitively attractive. If the etymology is correct, it furnishes another example for a regular outcome âλαâ < *lÌ¥.
10.4.3 γλαÏÏ Ïá½¹Ï
The etymology of Hom. γλαÏÏ Ïá½¹Ï âhollowâ (epithet of ships, caves, and the phorminx, in Od. 14.533 also of a hollow stone that provides shelter) has been evaluated in various ways.61 There are two basic proposals. First, γλαÏÏ Ïá½¹Ï has been derived as an adjective in *âulóâ from the root of γλάÏÏ âto scoop out, dig a holeâ. This verb is attested as a simplex only in ÏοÏÏὶν γλάÏει âhe digs [the earth] with his pawsâ, of a lion (Scut. 431), and with a preverb only in the line εá½Î½á½°Ï δâ á¼Î½ ÏαμάθοιÏι διαγλάÏαÏâ á¼Î»á½·á¿Ïιν âhaving scooped out lairs in the sand of the beachâ (Od. 4.438, the subject is Eidothea). Lamberterie objects to this proposal that the only indication for a PIE verbal root is precisely Greek γλάÏÏ, and that the alleged connections with Slavic (e.g. Bulg. glob âeye socketâ) and Celtic words (MIr. gulba gl. rostrum âbeakâ) are uncertain.
A second proposal is made by Chantraine (DELG s.v. γλαÏÏ Ïá½¹Ï). He argues that γλάÏÏ can hardly be separated from γλύÏÏ âto carve, sculptureâ, a root which does have verbal cognates in other IE languages (Lat. glÅ«bere âto peel, strip the barkâ, OHG klioban âto cleaveâ). This combination is accepted by Lamberterie, who assumes a dissimilation *Î³Î»Ï Ïá½»â > γλαÏá½»â and a semantic development from âstripped offâ to âhollowâ in the adjective. Subsequently, the verbal root, too, would have split into γλαÏâ and Î³Î»Ï Ïâ.62 Lamberterie further suggests that the reconstructed u-stem *Î³Î»Ï Ïá½»â could be deverbal, and that another reflex of this u-stem is perhaps found in the Slavic adjective *gloÌbokÑ (Ru. glubókij) âdeepâ.
The second scenario does not seem plausible to me. First, the assumed dissimilation *Î³Î»Ï Ïá½»â > γλαÏá½»â is not self-evident (as Lamberterie 1990: 316 himself admits); I have not found a convincing parallel in Greek. Moreover, it is unclear how the split into γλύÏÏ and γλάÏÏ should be envisaged: for a factitive verb based on the new adjective *γλαÏá½»Ï, one expects *γλαÏύνÏ. Thirdly, the only proposed cognate is found in Slavic, where the three root variants *gloÌbâ, *glybâ and *glÑbâ could point to non-IE origin.63 Finally, the semantic connection between âto peel off, scaleâ and âto make hollowâ is conceivable, but not evident. The oldest meaning in both Latin and Germanic is âto peel off, scaleâ, which is very close to that of γλύÏÏ âto carveâ, i.e. âto scale off chips of wood or stoneâ. In defense of Chantraine, it is true that the adjective γλαÏÏ Ïá½¹Ï is applied not only to natural cavities (caves, holes), but also to man-made hollow objects (musical instruments, ships). However, the verb γλάÏÏ does not refer to holes that are made by carving, chiseling, or peeling: it means âto dig a hole with the hands or pawsâ in both its attestations.
In view of these problems, I wish to propose an alternative etymology: γλαÏÏ Ïá½¹Ï contains the root of δελÏá½»Ï âwombâ, δελÏá½·Ï âdolphinâ (i.e. â[aquatic animal] with wombâ), and á¼Î´ÎµÎ»ÏÎµá½¹Ï âbrother/sister, born of the same motherâ < *smÌ¥-gwelbh-es-óâ âfrom the same wombâ. In Indo-Iranian, the root *gwelbhâ is reflected in Ved. gárbha- m. âwomb, embryoâ, YAv. garÉβa- m. âwombâ, gÉrÉβuÅ¡- ânewborn lambâ.64 The verb γλάÏÏ would be the only trace of *gwelbhâ as a verbal root, but even if the precise origin of the zero grade thematic root presents (of the type Ved. tudáti) is unclear,65 I see no reason to doubt its etymological connection with the nouns mentioned.
As for the phonological developments, there is a number of clear cases where a Common Greek labiovelar onset dissimilates against a labial stop in the following coda or onset: compare καÏÎ½á½¹Ï âsmokeâ < PGr. *kwapnoâ / *ku̯apnoâ (Lith. kvãpas âid.â), á¼ÏÏοκόÏÎ¿Ï âbakerâ (Hdt.) beside Myc. a-to-po-qo âid.â (PIE *pekwâ âto cook, ripenâ; the Ionic form has undergone metathesis to *âkwopoâ), and Hom. κόλÏÎ¿Ï âbosom, lap; gulfâ < PGr. *kwólpoâ (cf. PGmc. *hwalfaâ n. in ON hvalf, OE hwealf âvaultâ).66 This dissimilation took place relatively late, as it is not yet found in Mycenaean. Moreover, δελÏá½»Ï and relatives illustrate that the palatalization of labiovelars before e precedes the dissimilation. It is possible that the dissimilation took place irregularly, but it seems to be applied in a remarkably consistent fashion.67
The semantic development is relatively straightforward: a meaning âhollow; cavityâ can be posited for the PIE root. Already in the proto-language, nominal formations developed a special meaning âwombâ.68 It deserves attention, however, that γλαÏÏ Ïá½¹Ï would be the only adjective derived from this root, which mainly furnishes substantival derivatives. This brings us to the hapax γλάÏÏ (n.) âcave, shelterâ (Hes. Op. 533). Lamberterie (1990: 313â314), building on Leumann (1953: 223 n. 2), analyzes this as a substantivized form of an adjective *gwlÌ¥bh-úâ, and claims that γλαÏÏ Ïá½¹Ï is an extension in *âlóâ of this adjective.69 However, it cannot be excluded either that γλάÏÏ is an original noun, with a suffix to be compared with δελÏá½»Ï. In this case, γλαÏÏ Ïá½¹Ï can be analyzed as a de-substantival derivation in *âróâ.70
In either case, since the full grade slot of the root for âhollowâ was *gwelbhâ, this etymology furnishes new and compelling evidence for a regular development *lÌ¥ > âλαâ in one of the dialects reflected in Homeric Greek. This etymology also helps to clarify the background of the toponym ÎελÏοί (Boeot. ÎελÏοί). Given a root meaning âhollowâ, this name may be a substantivized adjective which referred to caves or places of shelter, just as in γλάÏÏ âcaveâ; also note the toponym ÎλάÏÏ Ïαι (Il. 2.712).
10.4.4 κλάδοÏ
The thematic noun á½ / Ïὸ ÎºÎ»á½±Î´Î¿Ï âbranchâ (Ibyc., A., B.+), later also attested as a monosyllabic stem κλαδâ (E., Ar.), has been compared with Germanic and Slavic words: ON and OE holt (n.) âwood; forestâ < *klÌ¥doâ, and OCS klada, Ru. kolóda âwooden logâ < *kóldeh2â. The comparison is semantically attractive and phonologically perfect, and I therefore follow the etymological dictionaries in reconstructing a PIE noun *klÌ¥do-.71 Still, the limited distribution of this word and the lack of a good root etymology are reasons for some doubt.72
10.4.5 λάγνοÏ
The adjective Î»á½±Î³Î½Î¿Ï âlascivious, hornyâ (Arist.) and its derivatives λαγνεύÏ, λαγνεία are best derived from the root *selǵ- that is also attested in Ved. sarj âto release, let goâ and in the Cretan verb Î»Î±Î³Î±Î¹Ï (aor. λαγαÏαι) âto releaseâ (on which see section 10.6.1). As argued above, a further possible cognate is Lat. laxus âspacious, wide, looseâ if this reflects *slÌ¥g-s-oâ with Schrijverâs rule *RDC > RaDC. Furthermore, as I have argued in Van Beek 2018: 59â60, á¼ÏÎµÎ»Î³á½µÏ âwantonâ may also be related to λάγνοÏ, reflecting PGr. *ad-selg-esâ. The Greek evidence suggests the existence of a Proto-Greek verb, with an e-grade thematic present *selg-e/oâ (whence á¼ÏελγήÏ) beside a zero grade thematic aorist *slÌ¥g-e/oâ (whence Cret. λαγαÏαι).
The derivation from PIE *selǵâ implies that Î»á½±Î³Î½Î¿Ï is unrelated to Germanic *slakaâ âslackâ (compared by Frisk, GEW s.v. λαγαίÏ) because the latter has a different vowel slot. Therefore, Î»á½±Î³Î½Î¿Ï is a relatively strong piece of evidence for the development of *lÌ¥.
10.4.6 Ïλάξ and δίÏλαξ, ÏÏá½·Ïλαξ
Ïλάξ, gen. ÏÎ»Î±Îºá½¹Ï (S., E.+) denotes a âflat surfaceâ, e.g. that of the sea, or the flank or flat summit of a mountain. This noun is traditionally compared to Germanic words meaning âlayer, surfaceâ, especially ON flær (f. pl.) âstrip of landâ < PGmc. *flahiz and ON flá (f. sg.) âid.â < PGmc. *flahÅ. According to Frisk (GEW s.v. Ïλάξ), this comparison points to an inherited root noun PIE *plakâ, with inherited *a. However, we must note that according to Kroonen (EDPG s.v. *flahÅâ), the root noun inflection in the plural form flær is secondary, and the Ä-stem form reflected in the singular flá is older. Kroonen therefore compares the Germanic words to Latv. plaka âlowland, plainâ, and reconstructs them as quasi PIE *plok-eh2â (EDPG s.v. *flakaâ).73
Does this mean that the comparison of the Germanic words with the Greek root noun Ïλάξ, as a mere root etymology, becomes less plausible? On the contrary, for it appears that a verbal root *plekâ can be reconstructed. In Germanic we find various reflexes of a strong verb *flahanâ, e.g. OE flÄan âto strip, flayâ, ON flá âid.â. This throws an unexpected light on the semantic development to âplain; flat surfaceâ. Features of the landscape are often named by analogy with the body of animals (e.g. ridge, headland, neck, mouth of a river). The identification made in the case of Ïλάξ is that between hair and vegetation: a plain without trees was described using the image of a skinned animal, stripped of its hairy skin. The same image was at work in the Germanic nouns quoted above.74
There are two Greek forms with a 2nd compound member âÏλακâ âlayerâ. Hom. δίÏλαξ (adj.) âtwo-layeredâ is attested in δίÏλακι δημῷ â(wrapped) in a double layer of fatâ (Il. 23.243 and 253), and it occurs in substantivized form in δίÏλακα ÏοÏÏÏ Ïέην âpurple mantleâ (Il. 3.126, 22.441, Od. 19.241). The hapax ÏÏá½·Ïλαξ describes the âthree-layeredâ rim (á¼Î½ÏÏ Î¾) of Achillesâ shield (Il. 18.479â480). What is the etymology of this second member âÏλακ-? It has been derived from the root of ÏÎ»á½³ÎºÏ âto plait, twineâ (PIE *pleḱâ).75 Given the identical formation of Lat. duplex âtwofoldâ and the existence of the verb plicÄre âto fold, windâ in that language, this seems plausible at first sight. In addition, the use of â-foldâ in the Germanic languages seems to offer a good parallel; the phrase δίÏλακι δημῷ would preserve a trace of the original meaning âtwo-fold, wrapped twiceâ.
In reality, δίÏλαξ and ÏÏá½·Ïλαξ must be compounds with Ïλάξ âsurfaceâ; their second member is unrelated to ÏÎ»á½³ÎºÏ âto twineâ.76 The main argument is that Ïλάξ, like other nouns derived from the verbal root *plekâ âto strip, flayâ (compare ON fló âlayerâ < PGmc. *flÅhÅ, EDPG q.v.), must also have had the meaning âlayerâ. It is telling that all Homeric uses of δίÏλαξ and ÏÏá½·Ïλαξ concern layers that may have been obtained in the process of flaying and dissecting an animal: hides (in a shield) and layers of fat.
Thus, Ïλάξ can be plausibly reconstructed as a root noun PIE *pl(o)kâ âsurface, layerâ belonging with a verbal root meaning âstrip, flayâ. There is no reason to assume that an ablauting full grade form *plokâ was preserved in the paradigm sufficiently long to influence the vocalization of *plÌ¥kâ, and the verb has left no traces in Greek. Hence, Ïλάξ < *plÌ¥kâ is an important piece of evidence.
10.4.7 ÏλάÏη
Although ÏλαÏá½»Ï and related âCalandâ-system forms do not offer compelling evidence for the regular reflex of *lÌ¥, this may be different for the cognate form ÏλάÏη âshoulder-blade; blade of an oarâ, which often occurs as a determinative compound ὠμοÏλάÏη when denoting the body-part. It is often maintained that ÏλάÏη may refer to any flat surface, but meanings other than the two just cited (âsheet of papyrusâ, âwinnowing fanâ) are rare and late. This means that ÏλάÏη has very concrete referents. Moreover, it is remarkable that Hitt. paltana- c. âshoulder(blade)â, OIr. leithe âid.â and OCS pleÅ¡te âshoulderâ derive from the same root. The Celtic and Slavic forms both appear to continue a pre-form *pleth2-i̯oâ. One might therefore be inclined to view ÏλάÏη as directly reflecting PIE *plÌ¥th2-eh2â. According to Chantraine (DELG s.v. 1 ÏλαÏá½»Ï), ÏλάÏη was created beside the neuter abstract ÏλάÏοÏ, on the model of βλάβη beside βλάβοÏ; but this does not seem likely to me because the antiquity of Î²Î»á½±Î²Î¿Ï is not guaranteed, and also since the last-mentioned forms retained a connection with the verb βλάÏÏÏ. Thus, although the formation underlying ÏλάÏη is not necessarily of PIE origin, it is an old derivative that is relatively isolated within Greek, and therefore a reasonably strong candidate to display the regular reflex of *lÌ¥.
10.5 The Development of *l̥n
A couple of Ionic-Attic forms suggest that *lÌ¥ developed to âαλâ conditioned by a following nasal plus vowel.77 Indeed, a special development before nasals would not be unexpected, given that the same happened to the syllabic liquids in the prehistory of Celtic (cf. section 9.4). In order to see whether such a development is conceivable for Greek, let us first discuss several present stems for which an original sequence *lÌ¥n can be reconstructed.78
10.5.1 The Presents Î²á½±Î»Î»Ï and θάλλÏ
Consider the three following reconstructions:
-
Î²á½±Î»Î»Ï âto throwâ < *gwlÌ¥ne/oâ << *gwlÌ¥-n-(e)h1â;
-
Î¸á½±Î»Î»Ï âto flourishâ < *dhlÌ¥ne/oâ << *dhlÌ¥-n-(e)h1â;
-
Ïá½±Î»Î»Ï âto toss, sway, brandishâ < *plÌ¥ne/oâ << *plÌ¥-n-(e)h1â.
That Î²á½±Î»Î»Ï continues an original nasal present *gwlÌ¥ne/oâ (PIE *gwlÌ¥-n-h1â, root *gwelh1-) is widely accepted and seems reasonably certain.79 Since the root of Î¸á½±Î»Î»Ï is best reconstructed as *dhelh1-, with Hackstein (2002: 220),80 an inherited nasal present *dhlÌ¥ne/oâ << *dhlÌ¥-n-(e)h1â is also the most likely option. The reconstruction of Ïá½±Î»Î»Ï as *plÌ¥-n-(e)h1â is less certain, but remains a viable possibility, cf. the next section.
There is good evidence showing that intervocalic *âlnâ developed to âλâ with compensatory lengthening of the preceding vowel. As argued by Slings (1975), West Greek and Aeolic forms of the verb âto wish, wantâ (Dor. δήλομαι, Boeot. βειλομη, Thess. βελλομαι) are best reconstructed as a nasal present PGr. *gwelne/oâ. In my view, Ion.-Att. βούλομαι is best analyzed as a contamination of this *gwelne/oâ and the old verb βόλομαι (attested in e.g. Homer, Euboean and Arcadian). The noun Î²Î¿Ï Î»á½µ < *gwolnÄ is best analyzed as a regular deverbal abstract of the type Ïομή to the outcome of PGr. *gwelne/oâ.81 Slings also draws attention to á½ÏÎµá½·Î»Ï âto oweâ, which demands a similar pre-form *ophelne/oâ beside the thematic aorists ὤÏελον, ὦÏλον. To these examples, I would also add the case of Hom. εἴλομαι âto throng togetherâ, which is best derived from *u̯elne/oâ (see below).
If this is true, and if Î²á½±Î»Î»Ï and Î¸á½±Î»Î»Ï indeed continue nasal presents, how can their geminate âλλâ be accounted for? I propose that it reflects *âlÌ¥nâ and that the development of *lÌ¥ to âαλâ took place after the first stages of the 1st compensatory lengthening had affected original post-vocalic *âlnâ.82 However, even if this analysis of Î²á½±Î»Î»Ï and Î¸á½±Î»Î»Ï is correct, the question whether these presents can be used to prove a regular vocalization *âlÌ¥nâ > *âalnâ > âαλλâ (rather than > âλανâ) remains open. In Î²á½±Î»Î»Ï the vocalization may have been influenced by the aorist βαλεá¿Î½, and similarly, it would be possible to argue that the outcome of *dhlÌ¥ne/oâ was influenced by the root allomorph θαλâ in the frequent pf. ptc. ÏÎµÎ¸Î±Î»Ï á¿Î± < *dhe-dhlÌ¥h1-us-ih2, and in derivatives like θαλέθÏ, θαλύÏ, θαλεÏá½¹Ï.
10.5.2 ÏάλλÏ
The case of Ïá½±Î»Î»Ï is more complex. Considering the verb and its derivatives, we have evidence for a non-ablauting root PGr. *palâ. The question is from which Indo-European pre-form this root was generalized. The root is mostly reconstructed as *pelh1â on account of the denominative verb ÏÎµÎ»ÎµÎ¼á½·Î¶Ï âto shake, cause to quiverâ (probably derived from a lost noun *Ïá½³Î»ÎµÎ¼Î¿Ï n.). The LIV2 (s.v. *pelh1â, following Harðarson 1993: 161) reconstructs an inherited nasal present *plÌ¥-n-h1â that is directly reflected in ÏάλλÏ. Frisk (GEW s.v. ÏάλλÏ), however, derives Ïá½±Î»Î»Ï from a yod-present *pal-i̯e/oâ in view of the sigmatic aorist Ïá¿Î»Î±Î¹ < *pal-sâ, which normally does not pair with a nasal present stem. Thus, the reconstruction of the present stem depends on which verbal formation is considered to be primary. The sigmatic aorist Ïá¿Î»Î±Î¹ must be secondary in any case (cf. LIV2 l.c. and Beckwith 1996: 125); the root aorist ÏάλÏο, á¼ÏαλÏο is also widely supposed to be an artificial creation (Leumann 1950: 60â¯ff., followed by Harðarson 1993: 196â197). The only potentially old aorist formation is the reduplicated participle á¼Î¼ÏεÏαλών âswinging up (over the head)â < *pe-plh1-e/oâ, which is exclusively Homeric.
Etymologically, Ïá½±Î»Î»Ï has been connected with Sln. pláti âto waveâ, Ru. dial. polótâ âto winnowâ: see LIV2 (s.v. *pelh1â) and Beckwith (1996: 123â129). On the other hand, several etymological dictionaries (DELL s.v. pellÅ; GEW s.v. ÏάλλÏ) compare Ïá½±Î»Î»Ï primarily with Lat. pellÅ âto beat against, strike; pushâ; in this case the Latin perfect pepulÄ« can be compared directly with the reduplicated aorist á¼Î¼ÏεÏαλών. Indeed, pace LIV2, the comparison with Latin is attractive also from a semantic point of view: Frisk (l.c.) compares ÏÎ±Î»Î¼á½¹Ï âpulseâ with Lat. pulsus âid.â. Although neither of these formations can be inherited, the meaning âto beatâ (of the heart) may well be old: compare ÏάλλεÏαι ἦÏÎ¿Ï (Il. 22.452), Ïαλλομένη κÏαδίην (Il. 22.461). Another meaning shared by Ïá½±Î»Î»Ï and Lat. pellÅ is âto vibrateâ (of the strings of an instrument), cf. Pl. Phd. 94c.83 For this reason, the reconstruction of a nasal present *plÌ¥-n-h1â âto shake, quiver, vibrateâ (tr.) underlying both Ïá½±Î»Î»Ï and Lat. pellÅ deserves full consideration.
Most modern etymological dictionaries,84 however, separate Ïá½±Î»Î»Ï from the root of Lat. pellÅ and U. am-pelust âwill have slainâ because they prefer to connect the Italic words with OIr. ad·ella âvisitsâ and fut. âeblaid âwill driveâ. The root of OIr. ad·ella is reconstructed as *pelh2â on the basis of a comparison with Ïίλναμαι, aor. ÏελάÏαι âto approachâ; the fut. âeblaid âwill driveâ is also included in the comparison, with a supposed semantic development *âto bring nearâ > âto thrust, drive nearâ > âto strikeâ.85 This scenario has been embraced by various scholars, but in my view the assumed semantic development is questionable; in addition, as just argued, it is implausible to separate Ïá½±Î»Î»Ï from Lat. pellÅ.86 The only reason to disconnect these verbs is the assumption that the Irish future âeblaid is derived from the same root as ad·ella. However, in view of the difference in meaning (âdriveâ versus âvisitâ), it is possible to separate ad·ella etymologically from âeblaid, and to regroup the words as follows: *pelh1â âto strike, vibrateâ is reflected in Lat. pellÅ, pepulÄ«, Gr. ÏάλλÏ, á¼Î¼ÏεÏαλών, and OIr. âeblaid, while *pelh2â âto draw closeâ is continued in OIr. ad·ella âvisitâ and Gr. Ïίλναμαι, ÏελάÏαι, Ïλá¿Ïο âto draw nearâ.
Nevertheless, it remains uncertain whether Ïá½±Î»Î»Ï continues an old nasal present. Given that the middle root aorist ÏάλÏο and the sigmatic aorist á¼Ïηλα must both be secondary if the root was indeed PIE *pelh1â, it is certainly possible to assume that the entire verbal system was rebuilt on the basis of an inherited present stem Ïá½±Î»Î»Ï < *plÌ¥-n-h1â. In that case, the development would be comparable to that found in Î²á½±Î»Î»Ï and θάλλÏ. However, it cannot be excluded either that Ïá½±Î»Î»Ï reflects a yod-present beside the aorist á¼Ïηλα, and that both formations are denominative, for instance to Ïá½±Î»Î¿Ï m., which is retained only with the meaning âlot (shaken from a helmet)â but may originally have been a verbal noun denoting the act of tossing.
10.5.3 κάλλοÏ, καλλιâ and Related Forms
The sequence âαλλâ also appears in the lexical root of κάλλοÏ, ÏεÏικαλλήÏ, the first compound member καλλιâ, and the forms of comparison καλλίÏν, κάλλιÏÏοÏ. All these forms belong to the positive ÎºÎ±Î»á½¹Ï âbeautifulâ. The etymology of these forms is mostly considered unclear.87 The only existing proposal is a comparison with Ved. kalyÄÌá¹a- âbeautiful, lovelyâ (f. kalyÄá¹Ä«Ì), assuming an IE adjectival root *kalâ that would appear as *kal-iâ in compounds.88 It is problematic for this comparison that Greek καλλιâ has a geminate;89 moreover, the a-vocalism of the reconstructed root is disturbing.90
The root shape καλλâ can be accounted for if we start from a pre-form containing *lÌ¥n.91 Since adjectives with âCalandâ morphology could be productively derived from primary verbs in Greek, the forms κάλλοÏ, âÎºÎ±Î»Î»á½µÏ and καλλίÏν, κάλλιÏÏÎ¿Ï can be mechanically derived from a verb *κάλλÏ, reflecting a thematicized nasal present PGr. *klÌ¥ne/oâ.92 Noting that âbeautifulâ may easily develop from âexcelling, outstandingâ, this reconstructed form PGr. *klÌ¥ne/oâ may directly correspond to the nasal present attested in Lat. âcellÅ âto stand outâ (cf. also Lith. kìlti âto riseâ, 1sg. pres. kylù).93 Thus, the original meaning of Homeric ÏεÏÎ¹ÎºÎ±Î»Î»á½µÏ would be âstanding out, excellingâ. That a semantic development to âexcel, surpassâ could easily take place in derivatives from this root is illustrated not only by Lat. praecellÅ and excellÅ, but also by Lith. kilnùs âupright; excellent, splendidâ, related to kìlti âto riseâ.
This brings us to the formation of the positive, Att. κᾰλόÏ, Hom. κᾱλόÏ, Boeot. καλÏοÏ. A root *kalâ (with old *a) is excluded because Ionic-Attic âκαλλâ cannot be obtained from this. Now, *kalu̯óâ could theoretically reflect PGr. *klÌ¥u̯óâ, if one supposes a vocalization *lÌ¥ > âαλâ before *u̯.94 However, if the etymology proposed here is correct, the root is to be reconstructed as *kelh1â.95 This would imply that *kalu̯óâ did not derive directly from *klÌ¥u̯óâ, but is a thematicization of PGr. *kalúâ < PIE *klh1-uâ.96 Lith. kilùs and PGr. *kalúâ may theoretically derive from a common pre-form PIE *klh1-uâ âsticking out, rising upâ, but the Lithuanian form is more likely to be an independent, productive creation of that language.
Returning to the vocalization of *lÌ¥, the root καλλâ can be plausibly derived from *klÌ¥n-e/oâ and thus offers suggestive evidence for a regular development of *lÌ¥n to âαλλâ. However, it cannot be entirely excluded that the outcome âαλâ in κάλλοÏ, καλλίÏν and related forms arose under influence of the basic adjective *kalu̯óâ.
10.5.4 Ion. á¼Î»á½µÏ, Hom. á¼Î¿Î»Î»á½³ÎµÏ
The Ionic adjective á¼Î»á½µÏ (with á¾±) (Hdt., Hp.) means âthronged, amassed, in close formation, forming a unityâ, pl. also âall togetherâ.97 This is a potentially important piece of evidence for *lÌ¥n, because it is cognate to Hom. á¼Î¿Î»Î»á½³ÎµÏ (plurale tantum) âin a throng, all togetherâ < zero grade *ha-u̯lÌ¥n-esâ, and probably also á¼ÎµÎ»Î»á½µÏ âthick, denseâ (hapax at Il. 3.13) with a full grade root.98 The zero grade formation is also reflected in West Greek: Elean αÏλανεοÌÏ âall togetherâ, and the gloss á¼Î»Î±Î½á½³ÏÏÎ á½Î»Î¿ÏÏεÏῶÏ. ΤαÏανÏá¿Î½Î¿Î¹ âentirely, completelyâ (Hsch.).99 Since Tarentum was a Spartan colony, the adverb can be reconstructed also for Proto-West Greek.
There are several uncertainties in the reconstruction of this adjective. The dialectal origin of Hom. á¼Î¿Î»Î»á½³ÎµÏ is unclear. The hapax á¼ÎµÎ»Î»á½µÏ must reflect *ha-u̯elnesâ and seems to be of Aeolic origin in view of its geminate reflex of intervocalic *âlnâ. The Ionic prose form á¼Î»á½µÏ could continue a full grade (like Homeric á¼ÎµÎ»Î»á½µÏ) or a zero grade root (like á¼Î¿Î»Î»á½³ÎµÏ, αÏλανεοÌÏ). In the latter case, the development could be reconstructed as *ha-u̯lÌ¥nâ > *hau̯alnâ > *hau̯allâ > *hÄllâ, with loss of digamma followed by simplification of the geminate after a long vowel.100 However, since á¼Î»á½µÏ may also be the regular contraction product of a pre-form *hau̯ẹÌlesâ < *ha-u̯elnesâ with a (secondarily introduced) e-grade root, it cannot serve as evidence for the Ionic-Attic development of *lÌ¥n.
For present purposes, the main question is: which pre-form to reconstruct for Proto-Greek? We must reconstruct *smÌ¥-u̯lÌ¥n-esâ with a zero grade root, because there would be no motivation for introducing a zero grade independently in the West Greek and Homeric forms. It follows that the e-grade was introduced secondarily in Homeric á¼ÎµÎ»Î»á½µÏ âdenseâ (and possibly also in Ionic á¼Î»á½µÏ: see above). The basis for its introduction must have been the verb âto throngâ, which formed a nasal present *u̯elne/oâ reflected in Hom. εἴλομαι âto be throngedâ < *u̯elne/oâ.101 Indeed, this same full grade is found in the Homeric causative present Îµá¼°Î»á½³Ï âto press togetherâ and most other stems of its paradigm (aor. á¼Î»Ïαι, mid.-pass. pf. á¼ÎµÎ»Î¼Î±Î¹).102 A zero grade reflex *u̯alâ is preserved only in the inagentive aorist á¼Î»á¿Î½Î±Î¹.
What was the derivational basis of *smÌ¥-u̯lÌ¥n-es-? A suffix *ânesâ, as assumed by the etymological dictionaries (GEW, DELG, EDG), is difficult to motivate in this compound. Since s-stem adjectives could be directly derived from verbal stems in Greek, I propose that the present stem *u̯elne/oâ (reflected in Hom. εἴλομαι) originally had the shape *u̯lÌ¥n-e/o-.103 In a similar way, the precursors of βούλομαι âto wantâ (West Greek δείλομαι, etc.) and á½ÏÎµá½·Î»Ï âto oweâ must have secondarily introduced their full grade (see above). Moreover, this substitution is paralleled in Latin nasal presents such as pellÅ âto thrustâ, excellÅ âto excelâ. In sum, a Proto-Greek present stem *u̯lÌ¥ne/oâ (or perhaps rather compounded *smÌ¥-u̯lÌ¥ne/oâ) is the most likely derivational basis for an adjective *smÌ¥-u̯lÌ¥n-esâ.104
Let us now consider the possible origins of Hom. á¼Î¿Î»Î»á½³ÎµÏ. At first sight, it seems logical to view this form as Aeolic, in view of the geminate reflex âλλâ < *âlnâ in combination with the o-colored reflex. However, since *âlÌ¥nâ (as opposed to *âVlnâ) may have yielded âαλλâ also in the Ionic words discussed above, the geminate by itself does not tell us much. As for the o-vocalism, one might expect *lÌ¥ > âλοâ in Aeolic on the basis of *rÌ¥ > âÏοâ, but in reality there is no further evidence to support or exclude the idea that *lÌ¥ yielded âολâ (rather than âλοâ) in Aeolic dialects, whether generally or only in the position before n.105 Therefore, Wathelet (1970: 170) is rightly hesitant when he cites á¼Î¿Î»Î»á½µÏ as a possible example for the outcome of *lÌ¥ in Aeolic.
Another option to be taken into serious consideration is an âAchaeanâ origin of á¼Î¿Î»Î»á½³ÎµÏ. We do not know the regular outcome of *lÌ¥ in Mycenaean: there is no convincing evidence for *lÌ¥ generally, let alone in the position before nasals.106 Nevertheless, it is possible to assume that the Mycenaean outcome of PGr. *smÌ¥u̯lÌ¥nÄs was hau̯olnÄs or hau̯ollÄs.107
To sum up, the Proto-Greek form *ha-u̯lÌ¥n-esâ was directly reflected as *hau̯lane(h)â in at least part of West Greek, given El. αÏλανεοÌÏ âall togetherâ and the gloss á¼Î»Î±Î½á½³ÏÏÎ á½Î»Î¿ÏÏεÏῶÏ. ΤαÏανÏá¿Î½Î¿Î¹ in Hesychius. These forms will play an important role in section 10.6. On the other hand, the value of á¼Î»á½µÏ and á¼Î¿Î»Î»á½³ÎµÏ as evidence for the regular outcome of *âlÌ¥nâ is, unfortunately, limited.
10.5.5 Conclusions on *l̥n in Ionic-Attic
The verbs Î²á½±Î»Î»Ï and Î¸á½±Î»Î»Ï (and perhaps ÏάλλÏ) continue old nasal presents. A possible scenario is that these presents directly reflect pre-forms of the type PGr. *ClÌ¥-n-e/oâ. This development is supported by κάλλοÏ, καλλίÏν, κάλλιÏÏοÏ, which I propose to derive from an inherited nasal present *klÌ¥-n-eh1â >> *klÌ¥ne/oâ > *Îºá½±Î»Î»Ï âto excelâ. However, analogical influence on the vowel slot of these forms cannot be entirely excluded.
10.6 Dialectal Evidence
There is only little evidence for the vocalization of *l̥ in the other dialects, but nevertheless, important conclusions can be drawn for two West Greek dialects: Cretan and the dialect of Elis.
10.6.1 Cretan
As we have seen above, the root βλαÏâ ~ βλοÏâ might offer evidence for a conditioned o-colored development of *lÌ¥ in a labial environment in Cretan. The vowel slot of βλοÏâ may be due to leveling if βλαÏâ contains the outcome of a syllabic nasal.
The gloss κλάγοÏΠγάλα. ÎÏá¿ÏÎµÏ (Hsch.) displays a development κλâ < γλâ typical for certain parts of Crete.108 The form ÎºÎ»á½±Î³Î¿Ï suggests a development *lÌ¥ > λα in Cretan after a non-labial consonant, but the reconstruction of the pre-form remains uncertain (see above): a vocalized nasal cannot be excluded.
The verb Î»Î±Î³Î±Î¹Ï âto releaseâ (of persons in custody) has an aor. λαγαÏαι that is well-attested in Gortyn, and is also found as a gloss λαγάÏÏαιΠá¼Ïεá¿Î½Î±Î¹ âto let goâ (Hsch.). The vowel âαâ in the second syllable of this telic lexeme probably originated in the aorist, from which the present stem was derived. Frisk (GEW q.v.) suggests that λαγάÏ(Ï)αι is a reshaping of an older root or thematic aorist after ÏαλάÏαι (ÏÎ±Î»á½±Ï âto loosen; relax; release a prisonerâ), which has a similar meaning. This is reasonable since âαâ can easily be part of the root in ÏαλάÏαι but not in λαγάÏαι. The etymological connection with Ved. sarj âto let go, set freeâ (cf. LIV2 s.v. *selǵâ) is obviously attractive; it shows that Gortynian Cretan underwent a development *hlÌ¥gâ > *hlagâ > λαγâ, or perhaps rather *hlÌ¥gâ > *lÌ¥gâ > λαγâ, depending on the relative chronology.
If Î»Î±Î³Î±Î¹Ï and αβλοÏια are reliable evidence for the Cretan reflex of *lÌ¥, they would mirror the dual reflex of *rÌ¥ in this dialect, which yielded âαÏâ normally, but âοÏâ after labial consonants (âμοÏÏοÏ, ÏοÏÏι, and probably ÎÏοÏδιÏα: see section 3.1). It must be noted, however, that the vowel slot of αβλοÏια might be analogical. The evidence is so scanty that adding one form to the dossier may completely change the picture.
10.6.2 Elean αÏλανεοÌÏ and Tarentine á¼Î»Î±Î½á½³ÏÏ
As we have seen in section 10.5.4, the West Greek cognates of Hom. á¼Î¿Î»Î»á½³ÎµÏ are Elean αÏλανεοÌÏ âall togetherâ and the gloss á¼Î»Î±Î½á½³ÏÏÎ á½Î»Î¿ÏÏεÏῶÏ. ΤαÏανÏá¿Î½Î¿Î¹ âentirely, completelyâ (Hsch.).109 They provide valuable evidence for the regular development of *lÌ¥ in these dialects. The related verb αÏοÏÎµÎ»ÎµÏ /apowÄleÅ/ is also attested in Elis (cf. Minon 2007: 511â513), with a reflex of the first compensatory lengthening. The full grade *u̯elnâ presupposed by αÏοÏÎµÎ»ÎµÏ ensures that αÏλανεοÌÏ displays the regular development of *âu̯lÌ¥nâ in this dialect.110
In this context, the post-labial reflex *lÌ¥ > λο (possibly for ολ) in Cretan αβλοÏια gains new significance. It strongly suggests that the development of *lÌ¥ was later than Proto-West Greek (assuming that it makes sense to reconstruct such an entity), because in αÏλανεοÌÏ we find an a-colored reflex even after digamma. In addition, Elean αÏλανεοÌÏ proves that the development of an anaptyctic vowel in *âlÌ¥nâ was also a matter of the individual dialect groups, or even of the individual dialects: there was no early pan-Greek anaptyxis before the liquid in the sequence *âlÌ¥nâ.
10.6.3 Other Dialects
The Lesbian evidence is as follows. In Mytilene we find the word for âshoulder-bladeâ as ὠμοÏλάÏα[ν (IG XII,2 71.2), and also the abstract Ị̈λ̣α̣ÏÎ¿Ï âbreadthâ (Hodot 1990, MYT 013, 10, 3rd c.). The adjective ÏλαÏá½»Ï is attested in Lesbian poetry as ÏλάÏÏ (Alc. fr. 74). Borrowing from Ionic cannot be easily excluded for any of these instances, and is especially likely in ὠμοÏλάÏα, given that the expected Aeolic outcome of âshoulderâ would be á½Î¼Î¼Î¿â < *Homsoâ (cf. á¼ÏομμάδιοÏ, v.l. in Theoc. 29.29). Finally, ÏÏλανÏνÏν (Hodot 1990, MYT 015.04, 3rd c.) could also be an Ionic borrowing.
In literary Lesbian, two other words with âλαâ < *lÌ¥ are attested: á¼Î²Î»á½±Î²Î·[ν âunscathedâ (Sapph. 5.1) and γλαÏá½»Ïα[ âhollowâ (Alc. 7.8), but both could be borrowings from Ionic or from epic poetry (γλαÏÏ Ïá½¹Ï âhollowâ is a traditional epithet of ships and caverns in Homer, and the adjective á¼Î²Î»Î±Î²á½µÏ belongs to a high register). The adjective Î¼á½¹Î»Î¸Î±ÎºÎ¿Ï occurs as the Aeolic counterpart of class. μαλθακόÏ. However, as argued above, the etymological connection with OHG milti âmercifulâ and other Germanic words is uncertain; as a consequence, it remains uncertain whether the difference in vocalism must be ascribed to the vocalizations of a syllabic liquid or to some other cause (cf. καθαÏá½¹Ï beside κοθαÏá½¹Ï âpureâ, section 9.7.2). Finally, the Homeric word á¼Î¿Î»Î»á½³ÎµÏ âthronged, all togetherâ is attested as á¼á½¹Î»Î»ÎµÎµÏ in Alcaeus, but again an epic origin cannot be excluded. In sum, the evidence for *lÌ¥ in Lesbian is inconclusive.
As for Boeotian, ΠλαÏÎ·Îµá½»Ï is the epichoric term meaning âinhabitant of Plataeaâ. We are dealing, however, with a toponym and it cannot be excluded that Plataea was originally founded by speakers of a different dialect.
In Arcadian, the term ιμÏλαÏια (IG V,2 4.2) is perhaps related to ÏλαÏá½»Ï. Although the meaning is not clear, the following verb ιλαÏκεÏθαι (with dative rection) may suggest that the dat. sg. ιμÏλαÏιαι denotes a sacrificial offering (cf. Dubois 1988 ad loc.). Further, we find a PN ΠλαÏÎ¹Î±Ï (IG V,2 6.57 and 85, Dubois 1988: 45), but it would be unwise to base a conclusion on it because the bearer need not have been an Arcadian. The verb βλάÏÏÏ is attested also in Arcadian: aor. subj. ÏοÏκαÏÏ Î²Î»Î±Ïη (IG V,2 6.37), aor. ptc. Ïο καÏÏ Î²Î»Î±Ïθεν (ibid. 41). The inscription contains regulations concerning construction sites, and the meaning of the verbal forms is simply âto damageâ, like that of Classical βλάÏÏÏ. While the compound καÏαβλάÏÏÏ is not normal in Ionic-Attic, a West Greek Koine form cannot be excluded because a number of clauses and collocations appear in similar inscriptions elsewhere.111 That is, the West Greek verb καÏαβλάÏÏÏ may have been Arcadianized by introducing the preverb καÏÏ â (and ÏοÏâ). Thus, none of the Arcadian forms discussed here informs us about the regular outcome of *lÌ¥ in that dialect.
The Cyprian form po-lo-te-i (ICS2 318 VII, 2) cannot be relied upon. It was interpreted by R. Meister, in his editio princeps of this text, as the dat. sg. of a neuter ÏλόÏοÏ* which he supposed to be the dialectal equivalent of Ionic-Attic ÏλάÏÎ¿Ï âplane surfaceâ. For the Cyprian form, he posited the meaning âtablet, writing surfaceâ. With i te-ka-to-i po-lo-te-i, the ostracon on which the text has been written would then refer to itself as the âtenth pageâ of an archive. However, Massonâs edition and especially his 1966 article make it clear that no definite value can be attached to Meisterâs interpretation.112 Instead of Meisterâs po-lo-te-i, Masson prefers to read pe-lo-te-i. Moreover, the interpretation âtabletâ and the comparison with Attic ÏλάÏοÏ, which is not attested in that meaning, are mere guesses. Therefore, the form can be left out of further consideration.113
10.7 Conclusions on *l̥
The regular slot of the anaptyctic vowel before occlusives in Ionic-Attic was probably âλαâ. Leaving aside the uncertain connection between Î¼Î±Î»Î¸Î±Îºá½¹Ï and the Germanic word for âmildâ, reliable evidence for a reflex âαλâ is completely absent. On the other hand, there are several good candidates for the development to âλαâ: βλαδεá¿Ï < *mlÌ¥dâ and other entries in βλαδâ from Hesychius meaning âweak, porous, flaccidâ, ÏλάÏη âshoulder-bladeâ, possibly also βλαÏÏá½¹Ï âsproutâ (if from *mlÌ¥d-tóâ) and ÎºÎ»á½±Î´Î¿Ï âbranchâ (if < *klÌ¥d-oâ with G. Holz). As new pieces of evidence, I have adduced γλαÏÏ Ïá½¹Ï âhollowâ, from a pre-form *gwlÌ¥bh-u-lóâ âhollowâ, and Ïλάξ âplane surface, plainâ (including δίÏλαξ and ÏÏá½·Ïλαξ, which preserve an older meaning âlayerâ).
This conclusion that *lÌ¥ > âλαâ is remarkable given the evidence for *rÌ¥ > âαÏâ, with a different vowel slot. One could object that, as far as we know, the developments of PIE *lÌ¥ and *rÌ¥ were identical in all other Indo-European daughter languages. However, the evidence that we have must be taken seriously. In addition, if the treatment of *lÌ¥ and *rÌ¥ was indeed different, this may suggest that their developments took place at different chronological stages, i.e. that *lÌ¥ was vocalized earlier, at least in Ionic-Attic.
We have seen that the reflex of *lÌ¥n in Ionic-Attic resulted in a geminate âλλâ that did not take part in the 1st compensatory lengthening, as opposed to older instances of intervocalic *ln which did undergo the 1st compensatory lengthening. The evidence in favor of *lÌ¥n > âαλλâ consists of the verbs Î²á½±Î»Î»Ï and Î¸á½±Î»Î»Ï (which reflect pre-forms of the type PGr. *ClÌ¥-n-e/oâ, from inherited athematic nasal presents), as well as κάλλοÏ, καλλίÏν, κάλλιÏÏοÏ, for which I have proposed a new etymological derivation from a lost verb *klÌ¥n-e/oâ > *κάλλομαι âto stand outâ. It is not excluded that âαλâ in these forms was the regular development of *lÌ¥ before n. This would be analogous to the conditioned reflexes of *lÌ¥ and *rÌ¥ in Celtic, which developed to al and ar before n, but to li and ri before stops and m.
On the other hand, it cannot be excluded either that *lÌ¥n > *Éln is analogical, and that we happen to have no good examples of *lÌ¥n > *lÉn in Ionic-Attic. We do know that the last-mentioned reflex was regular in (part of) West Greek, witness Elean αÏλανεοÌÏ âall togetherâ. Further evidence for the dialectal reflexes of *lÌ¥ is extremely scanty. A potentially important form is Cret. αβλοÏια âcondition of having done no harmâ, which could display an o-colored reflex of *lÌ¥. If this interpretation is correct, it suggests that the vocalization of *lÌ¥, like that of *rÌ¥, took place in the individual West Greek dialects.
The nom. sg. is not attested in archaic and classical Greek.
Schrijver (2019: 369) tentatively proposes to identify the source from which αá½Î»Î±Î¾ and relatives were borrowed as Minoan (Lin. A) au-re âpigâ. For the semantic connection between âpigâ and âplowâ he draws attention to OIr. soc âpigâs snout; plowshareâ as well as French soc âplowâ, which was borrowed from this Celtic word for âpigâ, PClt. *sukkoâ, *sukkÄâ.
After the classical period, Î³Î»á½±Î³Î¿Ï is again found in Hellenistic hexameter poetry (Nic., Mosch.), probably in imitation of Homer. Callimachus has γάλακι (Hec. 1.4.4); Lycophron (4th c. tragedian) attests thematic (â)γλαγοâ in compounds; and ÏÎ¿Î»Ï Î³Î»Î±Î³á½µÏ appears in Aratus (Phaen. 1.1100).
A parallel is Î³Ï Î½á½µ, Boeot. βανά âwomanâ, both from PGr. *gwnÄ (cf. Beekes, EDG s.v. γάλα).
Schrijver (1991: 479â480).
Weitenberg (1985), also apud Kortlandt (2003: 65). Weitenberg derives the dialectial form katʿn from the acc. sg. *gl̥Kt-m.
Puhvel, HED s.v. kala(n)kâ, gala(n)kâ. This connection is not discussed by Kloekhorst (EDHIL s.v. kalank-i), who follows Oettinger in comparing kalank-i with ON kløkkr âweak, softâ, Lith. glẽžnas, gležnùs âid.â, and reconstructs the root as *gleǵhâ because of the non-acute root in Baltic.
Since drugs are often prepared with milk, another idea could be that γάλα originally denoted milk mixed with drugs.
For the latter assumption, see Meiser (1998: 114) and EDL s.v. lac.
For an extensive discussion of this word group, cf. Tichy (1983: 41â48). The Germanic group of ON hlakka âto cry; rejoiceâ is probably related to *hlah(j)anâ âto laughâ (cf. EDPG s.v. *hlakkÅnâ) and has nothing to do with κλαγγή, unless in the sense that both are onomatopoeic.
One could assume that κέκληγα derives from a root *kleh2gâ and is unrelated to κλαγγή.
Bader (see DELG, Supp. q.v.) distinguishes λάÏÎ¹Î¿Ï âhairyâ from λάÏÎ¹Î¿Ï âwillingâ in the formula λάÏιον κá¿Ï, a formal term of address preceded by the genitive of a PN (Il. 2.851 and 16.554), which would originally mean âstrong-willed heartâ. Baderâs reconstruction is questionable, however, because she has to assume an irregular laryngeal metathesis.
For these, and possible Slavic cognates, see GEW, DELG and EDG s.v. λάÏιοÏ.
The reconstruction of this material is further complicated by the existence of another word for âhairâ: *u̯olḱoâ, attested in Skt. válÅaâ âsprout, twigâ, Av. varÉsaâ âhair (on the head)â, Ru. vólos âhairâ, etc. The roots *u̯olHâ and *u̯olḱâ may have influenced each other.
See EDPG s.v. *slakaâ, where it is proposed that OIr. lacc âslackâ (which is clearly related to the Germanic adjective) could reflect *slg-noâ.
See GEW, DELG, and EDG, all s.v. λαγαίÏ.
See also Van Beek (2018: 59â60 with n. 72 and 73), where I have also suggested that á¼ÏÎµÎ»Î³á½µÏ âbrutalâ reflects PGr. *ad-selgesâ, containing a trace of the PIE preverb *h2ed and of an e-grade verbal stem based on PIE *selǵ-.
EDL s.v. laxus. Schrijver himself did not explain laxus with his rule (1991: 136 and 165), as he followed Lubotskyâs proposal that the root contained a laryngeal.
Connected by Mayrhofer (EWAia q.v.) with most of the words listed above: λαγαÏá½¹Ï, λαγαίÏ, Lat. laxus, ON slakr.
Cf. DTB s.v. slakkare and EDL s.v. langueÅ.
Cf. λογγάÏαιΠá¼Î½Î´Î¹Î±ÏÏá¿Ïαι, ÏÏÏαγγεύεÏθαι âto waste time, loiterâ (Hsch. λ 1192), λογγάÏÏÎ ÏÏÏαγγεύÏομαι âloiterâ (Phot. λ 370 = A. fr. 112); λογγάζεινΠÏὸ διαδιδÏá½±Ïκειν Ïὸ á¼Ïγον âto shirkâ (Phryn., = Ar. fr. 811; according to Phrynichus, in the passage in question the subject of λογγάζειν are horses pretending to have lame legs); λαγγάζειΠá¼Î½Ïὶ Ïοῦ á¼Î½Î´á½·Î´ÏÏιν âgives way, yieldsâ. á¼Î½ÏιÏá½±Î½Î·Ï á¼Î½ÏεÏá½½Ïá¿ (= Antiph. fr. 37, in Antiatticist lexica).
Beekes (EDG s.v. Î»Î±Î³Î³á½±Î¶Ï and λαγαίÏ) views the internal nasal of λαγγâ beside the root λαγâ in λαγαÏá½¹Ï as well as Î»Î±Î³Î±á½·Ï as a substrate phenomenon.
A recent treatment of the semantics and etymology of Lat. laxus is Höfler 2017, as I discovered when finalizing the manuscript of this book. Höfler and I converge in criticizing the assumed etymological relation between laxus (laxÄre) and langueÅ, but in my view he is mistaken in rejecting the reconstruction *sl̥ǵ-s-oâ and the connection with PIE *selǵ- âto let goâ. Höflerâs own reconstruction *slÌ¥k-s-oâ to a hypothetical PIE root *slekâ is subject to two problems: the evidence for such a root is very marginal, and *slÌ¥k-s-oâ does not actually yield *laksoâ. Höfler resorts to positing a pre-form *slÉk-s-oâ with shwa secundum, but he gives no clear morphological motivation for this.
Cf. IEW s.v. *u̯elâ 4.
For this comparison and other uncertain suggestions, see NIL 485â¯f. The further connection of this alleged PIE *meldhâ with Ved. márdhati âto neglect, abandonâ is semantically not evident (cf. the remarks in EWAia s.v. mardh); morphologically, márdhati could be viewed as an intransitive present in *âdhe/oâ of the type ÏÎ»á½µÎ¸Ï to the root *merâ âdisappearâ (cf. also the extensions Ved. mará¹£ âforgetâ and marḠâbe mercifulâ).
Kroonen (EDPG s.v. *meldaâ) now reconstructs the Germanic adjective as PIE *melh2-tóâ, comparing Skt. mlÄtaâ âsoftâ and OIr. mláith âid.â. In my view, the Schwebeablaut speaks against this etymology; moreover, I would reconstruct the PIE root for âcrushâ as *melh1â (see chapter 4), although this does not per se affect the connection.
This judgement is taken over by de Vaan (EDL s.v.).
See Beekes (EDG s.vv. á¼Î¼ÏλακίÏκÏ, ÏλάγιοÏ, and ÏλάζÏ), who adopted my suggestion to reconstruct a Pre-Greek verbal root *(a)mplankâ on the basis of these comparisons. I also included Ïλάνη âerrandâ in the comparison, assuming a root-final nasal velar *âÅâ in the substrate language, but that is much more hypothetical.
âDa eine Rekonstruktion im einzelnen nicht möglich ist, müssen wir uns auch für ÏÏλήν und das davon nicht zu trennende ÏÏλάγÏνα auf blosse Vermutungen beschränkenâ (Frisk, GEW s.v. ÏÏλήν).
However, note the proposal of Puhvel (1999: 74) to derive ÏÏήν and ÏÏλήν from *bhreǵh-n-s and *spleǵh-n-s, respectively, by a regular development PIE *âeǵhns > âÄn with compensatory lengthening. It is unclear how Puhvel envisages this development phonetically, but it would have the advantage of providing ÏÏήν with a natural etymology (cf. διάÏÏαγμα) and of explaining why ÏÏλήν coexists with ÏÏλάγÏνα in Greek. An obvious objection is that no structurally comparable PIE sound changes are known: one wonders what was wrong with a vocalization *bhreǵh-nÌ¥(âs) or *spleǵh-nÌ¥(âs).
In this word, deformations took place in other branches too: compare Ved. plÄ«hánâ (AV+) âspleenâ, which may have been influenced by snÄ«hánâ âsnotâ (Mayrhofer, EWAia q.v.).
The secondary zero grade in the Baltic forms (Lith. blužnìs âspleenâ, OPr. blusne âid.â), as well as Slavic material (OCS. slÄzena âid.â, Ru. selezënka) and perhaps also Skt. plÄ«hánâ âid.â (AV+), point to a full grade II. On the other hand, there is Celtic material pointing to a full grade I (MIr. selg, MBret. felch âspleenâ).
Cf. DELG (s.v. ÏÏλήν): âil nââ¯est pas sûr que les Grecs aient senti la parenté entre ÏÏλήν et ÏÏλάγÏνα.â
Cf. EDPC s.v. *balloâ.
In the opinion of Beekes (EDG s.v.), ÏÎ±Î»Î»á½¹Ï could be a substrate word.
For a discussion of the competing hypotheses and their relative merits, see Garvie ad loc. Schmitt (1978) remarks that á¼Î»ÏιÏÏÎ¿Ï cannot be a genuine Iranian name, but this is not judged decisive by Garvie because Aeschylus made up several other Iranian-sounding names in the Persae.
The meaning given in the LfgrE is âerwünscht, angenehmâ (adj.), âfreudig, gernâ (adv.). The etymological connection with á¼ÏαλÏÎ½Î¿Ï and á¼Î»ÏνιÏÏÎ¿Ï is accepted there, because it is favored by the attested inner-Greek semantic development of á¼ÏÏαλέοÏ. On the other hand, â⦠mit einer aus der antiken Etymologie gewonnenen Bedeutung gierig (Adv.) oder zu erraffend, erraft, räuberisch (Adj.) zu rechnen (â¦) ist an keiner Stelle nötig. Auch nachhomerisch tritt á¼ÏÏÎ±Î»á½³Î¿Ï zunächst noch in der etymologisch richtigen Bedeutung auf (â¦), daher ist wahrscheinlich, dass die anfänglich sich nur beim Adv. findende Bedeutung heftig (â¦) auf falscher Interpretation von besonders Od. 6.250 beruht, wo der Zusammenhang eine Umdeutung begünstigt.â (LfgrE s.v. á¼ÏÏαλέοÏ).
DELG comments on the adjectives: âgroupe archaïque altéré ensuite par lââ¯Ã©tymologie populaireâ. The older root meaning of á¼Î»Ïομαι is âto think, surmise, reckonâ, cf. Lachnit (1965). This casts some doubts on the connection with Lat. volup (adv.) âwith pleasureâ, which can be derived from *u̯elp-iâ (de Vaan 2008 s.v.) and would thereby reflect the same formation as Gr. á¼Î»Ïá½·Ï.
This also answers the objection made by Beekes (EDG s.v. á¼Î»ÏνιÏÏοÏ): âIt is doubtful to interpret á¼Î»Ïâ as *ÏαλÏâ, a zero grade of *ÏελÏâ in á¼Î»Ïομαι, á¼Î»Ïá½·Ï (for wouldnât one expect *ÏλαÏ-?)â.
The comparative ÏλαÏίον (Epich. fr. 100 K-A) is probably secondary for expected *ÏλάÏÏον.
It is uncertain whether this is an authentic Lesbian form or a borrowing from Ionic; see section 10.6 on the dialectal evidence.
Since the object of Av. mÉrÉc is often ahuâ ârighteous lifeâ or aÅ¡Ìaâ âorderâ, better translations than âto destroyâ might be available, e.g. âto disturbâ.
The comparison of the Indo-Iranian root with Hitt. markii̯e/a-zi âto disapprove ofâ, as accepted by HED and EDHIL (q.v.), therefore has little to recommend it. Beekesâ view (EDG s.v. βλάβη) that βλάÏÏÏ is of Pre-Greek origin cannot be substantiated either. See Van Beek 2017b: 55â56 for criticism of these and other views.
á¼Î²Î»á½±Ïθην is preferred in Epic Greek for metrical reasons, and is less frequent than á¼Î²Î»á½±Î²Î·Î½ in the classical language. Therefore, á¼Î²Î»á½±Î²Î·Î½ must be older.
Plus a Homeric imitation in Anacreont. 31.26.
For the derivation of an s-stem compound from an intransitive verbal stem in Greek (replacing an original root compound), see Meissner (2006: 186â197).
A root noun is attested in Vedic (RV 8.67.9, ins. sg. mrÌ¥cÄÌ) and YAv. (mÉrÉxÅ¡ âruinâ). Cf. á¼Î»Îºá½µ âfighting spiritâ, ÏÏ Î³á½µ âflightâ beside Homeric á¼Î»Îºá½·, Ïύγαδε, as well as δίκη âverdict; way of conductâ (a quasi-antonym of βλάβη) corresponding to Vedic dÃÅâ âdirectionâ.
As DELG (s.v. βλάβη) remarks, âPar son attestation plus ancienne comme par son sens concret, le thème verbal semble plus archaïque que les formes nominalesâ. The forms Î²Î»á½±Î²Î¿Ï (n.) âharm; curseâ (Hdt.+) and βλαβεÏá½¹Ï âharmfulâ (Hes.+) are either deverbal or backformations to á¼Î²Î»Î±Î²á½µÏ (cf. Schwyzer 1939: 482).
The instances are Hes. Op. 193â194, Op. 258, Op. 282â283, and perhaps also Th. 89.
This translation may have to be modified: if we compare the use of á¼Î²Î»Î±Î²á½µÏ âsincerely, unerringâ in Athenian oath formulae, αβλοÏια may have originally referred to behavior that was conform to the law (or legal procedure).
Schwyzer claims (1939: 257) that the phenomenon of distance assimilation belongs to the âungepflegte Umgangsspracheâ and therefore rarely appears in literary testimonies, but this is unfalsifiable.
Cf. pf. mid. εγÏαÏÏαι âhas been writtenâ, εÏÏα > εÏÏα âsevenâ.
Cf. also OAv. mÉrÉá¹gÉduiiÄ (2pl. mid. pres. ind.), mÉrÉá¹gÉidiiÄi (pres. inf.), and mÉraÍ Å¡ÌiiÄtÌ° (3sg. act. pres. opt.).
Cf. Cretan δεκάδâ âdecadâ and á¼Î¾ÏβάδιαΠá¼Î½á½½Ïια. ÎάκÏÎ½ÎµÏ âearringsâ (Hsch.), adduced by Olsen (1989), which probably continues *eks-ou̯snÌ¥Ìt-ia. In view of these forms, the rule must be dated to Proto-Greek.
A nasal present would also account for the zero grade root vocalism of βλάβομαι: usually, thematic middle root presents have an e-grade root (δέÏκομαι, Ïείθομαι, etc.). It is assumed here that the nasal, not the liquid, was vocalized in the Greek pre-form. In Indo-Iranian, it was the liquid that vocalized in nasal infix presents to *CRC-roots: cf. Ved. krÌ¥ntáti âcutsâ, YAv. kÉrÉá¹taiti, or Ved. ptc. rÌ¥ndhántâ âsucceedingâ (root ardh). However, this consonantal realization of the nasal could be ascribed to the occurrence of ablauting athematic forms like Skt. rÌ¥á¹Ã¡dhâ. This means that the vocalization *mlnÌ¥kw-e/oâ presupposed by the Greek form could be regular. Nasalized verbal stems like ÏλαγÏâ, λαγγâ, κλαγγâ are non-probative in this respect because they have no ascertained IE etymology.
To assume an o-grade root in the pre-form of á¼Î²Î»Î¿Ïá½³Ï would be unmotivated.
The word is unattested, however, in the tragedians.
Cf. Hdn. 3.130.4. The special Ionic form is also attested in inscriptions, e.g. αá½Ïοὶ Ïὴν θÏιὴν διÏληÏίην á½ÏελόνÏÏν IG XII Supp. 347 II, 6 (Thasos).
According to Kretschmer (cf. Frisk GEW s.v. διÏλόοÏ), διÏÎ»á½¹Ï was reshaped as διÏÎ»á½¹Î¿Ï under influence of the word for âsea journeyâ, *plóu̯oâ > ÏÎ»á½¹Î¿Ï > ÏλοῦÏ. This is not immediately convincing, but seems possible in view of the lack of alternatives.
The form δίÏαλÏÎ¿Ï is wrongly cited by Boisacq 1916 s.v. διÏλόÏ; it belongs to Ïá½±Î»Î»Ï âto tossâ.
See Lamberterie (1990: 315â¯ff.) for an extensive treatment.
â⦠la relation, perçue en synchronie, entre lââ¯adjectif et le verbe a entraîné la scission dââ¯une seule et même racine *Î³Î»Ï Ïâ en deux racines, resp. Î³Î»Ï Ïâ et γλαÏâ, la première ayant lââ¯acception technique de âsculpterâ dont la seconde est dépourvue, encore quââ¯on en trouve des traces dans certains emplois de γλαÏÏ Ïá½¹Ïâ (Lamberterie 1990: 315).
Cf. the doubts expressed by Derksen, EDSIL s.v. *gloÌbòkÑ, about the possibility to reconstruct this word.
In spite of doubts concerning the chronology of the attestations (cf. EDL s.v. vulva), it seems to me that Lat. vulva (imperial inscr. vulba) âwombâ can hardly be separated from Ved. gárbhaâ. The meanings âbodily cavityâ and âcavity in the landscapeâ are also found side by side in Gr. κόλÏÎ¿Ï âbosom, lap; gulf of the seaâ. This may have dissimilated from PGr. *kwolpoâ, from a root *kwelpâ also found in Germanic *hwelfanâ âto vault, revolveâ and *hwalfaâ âvaultâ (cf. GEW, EDG, EDPG). That root looks very much like the one under discussion, but we can only speculate about their interrelation (some early borrowing, or substrate phenomenon in the proto-language?). Similarly, Hitt. ḫu̯elpiâ (adj.) ânew, fresh, newbornâ, (n.) ânewborn animal, whelpâ is also semantically close to the other forms just mentioned, but formally irreconcilable.
For a recent discussion of the type tudáti in Greek, doubting its antiquity, see Willi 2018: 351â355 with references.
See Schwyzer (1939: 298â299, 302) for an overview of these cases of dissimilation. The etymology is accepted by Kroonen, EDPG s.v. *hwalfaâ.
Hom. γέÏÏ Ïαι âdams; lines of battleâ, post-Hom. γέÏÏ Ïα âbridgeâ seems not to have undergone dissimilation, as against Boeot. βεÏÏ Ïα, Cret. δεÏÏ Ïα pointing to a reconstruction PGr. *gwephuri̯a. However, since the word cannot be properly reconstructed for PIE, one could also argue (with Beekes, EDG s.v. γέÏÏ Ïα) that it was borrowed in different ways into the different Greek dialects.
For the semantic development, cf. Ved. yóni- âsheltered place; bed, nestâ; also âwombâ: see Van Beek fthc., also on the etymology of Ved. yóniâ. In Classical Sanskrit, the meanings âinside, middle, interiorâ and âadyton, interior of a sanctuaryâ are well-attested for gárbhaâ (see M-W, q.v.).
Leumann is followed also by Frisk (GEW), Chantraine (DELG), and Beekes (EDG). For a discussion of the evidence for adjectives in *âulóâ, see also Lamberterie (1990: 708â714). Clear instances are Î´Î±Ï Î»á½¹Ï âshaggyâ < *dnÌ¥s-u-lóâ beside δαÏá½»Ï < PIE *déns-uâ, *dnÌ¥s-éu̯â (see section 9.1.1) and the adverb ÏαÏÏ Î»á¿¶Ï âroughly, coarselyâ beside ÏαÏá½»Ï âthickâ, corresponding to Ved. bahuláâ âthick, dense, wideâ and bahúâ âmany, frequentâ.
Probert (2006: 284â285) remarks that while some adjectives in âÏ Ïá½¹Ï derive from *âu-loâ, in other cases âu-roâ is old.
Cf. EDPG s.v. *hultaâ, GEW s.v. κλάδοÏ.
Greek speakers may have connected ÎºÎ»á½±Î´Î¿Ï with the verb κλάÏ, aor. âκλάÏαι âto break (also of branches and stalks)â by folk etymology, but a direct etymological connection (as assumed in DELG s.v. κλάδοÏ) is hard to substantiate because it is difficult to see how the present ÎºÎ»á½±Ï could be secondary. Beekesâ comparison with κÏάδη âbranchâ, κÏÎ±Î´á½±Ï âto swingâ (EDG s.v. κλάδοÏ), assuming an interchange Ï/λ which he explains from a substrate origin, clearly goes too far.
Cf. also Icel. flár, Nw. flÃ¥ âflat, wideâ < PGmc. *flahaâ < PIE *plók-oâ (EDPG, s.v. *flahaâ).
In addition, there is also a Baltic verbal root with reflexes in Lith. plà kti âto beatâ, Latv. plakt âto become flatâ; compare also the derivative Lith. (dial.) plãkanas âflatâ, Latv. plakans âid.â. As for the meaning of Lith. plà kti âto beatâ, Derksen (EDBIL s.v. plakti) remarks that this root and *plaHkâ âto beatâ (in Slavic plakati âto cry, lamentâ) may have influenced each other.
Cf. de Vaan, EDL s.v. âplex (following Ernout-Meillet, DELL) and Beekes, EDG s.v. δίÏλαξ.
Thus also Frisk, GEW s.v. δίÏλαξ and W-H, s.v. duplex, but without the argumentation given here.
I leave aside the following forms: (1) Ion.-Att. ÏÏήλη, Dor. ÏÏάλα, Lesb. ÏÏάλλα. The pre-form is not necessarily *stlÌ¥-neh2â, as is often assumed: see section 1.2.5; (2) Î¼Î±Î»Î»á½¹Ï âflock of woolâ: the comparison with Arm. mal âramâ, proposed by Greppin (1981), is doubtful: cf. the discussion in Clackson (1994: 232); (3) ÎºÏ Î»Î»á½¹Ï âcrooked, club-footedâ, which Meier-Brügger (1990) derived from *kwlÌ¥nóâ, with the root *kwelâ âturnâ: see section 1.3.2 for criticism of this etymology; (4) ÏÎ±Î»Î»á½¹Ï âpenisâ, on which see section 10.1.12; (5) ÏÎ»Î±Î½á½±Ï âto drive off track; lead astrayâ (Hom.+), Ïλάνη âlong journey; errorâ (Ion.-Att.) and other related forms, because they have no convincing IE etymology according to the standard etymological dictionaries.
In Ïίλναμαι âto approachâ, âλνâ was restored due to a proportional analogy with its antonym ÏκεδάÏαι : Ïκίδναμαι âto disperseâ (cf. aor. ÏελάÏαι). It may have replaced a morphologically opaque form like *Ïάλλαμαι.
Cf. LIV2 s.v. *gwelh1â with further refs. It has been claimed (GEW and EDG, both s.v. βάλλÏ) that a yod-present cannot be excluded. However, yod-presents were not normally derived from thematic aorists, whereas nasal presents regularly occur beside thematic or root aorists (cf. e.g. Ion. Ïá½±Î¼Î½Ï < *tm-n-eh1â beside Ïαμεá¿Î½ < *tmh1-e/oâ). This pattern is probably inherited from PIE.
Pace LIV2 s.v. *dhalh1â.
Slingsâs proposal (op. cit.) that Ion.-Att. βούλομαι was directly derived from Î²Î¿Ï Î»á½µ is not very attractive.
In the development of original intervocalic *âlnâ, there may have been an intermediate stage *âllâ, after which the geminate was simplified with CL in most dialects. For a different scenario, see Slings (1975: 4â5).
Cf. also the meanings âflounderâ (á¼Î½Î±ÏάλλεÏαι á¼°ÏθύÏ, of a fish in Il. 23.692, also at Hdt. 1.141) and âquiverâ (of the knees of old men, Ar. Ran. 345).
E.g. LIV2 s.v. *pelh1â, EDL s.v. pellÅ.
Strunk (1985: 235).
The connection of Lat. pellÅ with Gr. Ïá½±Î»Î»Ï is now also defended by Willi (2018: 73) as âsemantically more straightforwardâ than a connection with *pelh2â âapproachâ.
See GEW (though judging the comparison with Skt. kalyÄÌá¹aâ to be âbrauchbarâ) and DELG (âétymologie ignoréeâ).
The first proposal to link Skt. kalyÄÌá¹aâ to Greek καλλιâ was made by E. Leumann (1893). Wackernagel (1934: 191â197) subsequently analyzed the Sanskrit word as an old compound containing the word for âelbowâ as a second member. This etymology was never fully embraced by the handbooks, but Pinault (2003) again pleads for it, arguing that the second member of Skt. kalyÄÌá¹aâ (or rather its feminine kalyÄá¹Ä«Ìâ) is a non-IE word for âhaunchâ borrowed independently by both Indo-Aryan and Tocharian. See Pinaultâs article for an overview of previous research on kalyÄÌá¹aâ.
In the view of Wackernagel (1934), καλλιâ replaced an older *καλιâ. Since Pinault (2003) does not deal with this issue, he apparently accepts Wackernagelâs view.
For the problems involved in reconstructing a PIE phoneme *a, see Lubotsky 1989. I will not further deal with this issue here.
I further elaborated this idea, which was presented already in Van Beek 2013, in a paper presented during the workshop Caland in Sicht (Ãsterreichische Linguistiktagung, Graz, 20Â November 2016). A published version of this paper is in preparation.
In the meaning âto stand outâ, the normal Homeric Greek verb was (δια)ÏÏá½³ÏÏ; note also the inherited middle perfect κέκαÏμαι âto excelâ. For âCalandâ systems secondarily derived from an already constituted stem, see also Nussbaum (1976: 98).
Lat. âcellÅ is assumed to have introduced its e-vowel from a prehistoric aorist: see EDL s.v. âcellÅ 1.
See Willi (2017), who derives PGr. *kalu̯óâ / *klÌ¥u̯óâ from the root PIE *ḱleuâ âhearâ. Note, however, that there is no evidence directly supporting such a rule for the vocalization of *lÌ¥u̯: see section 1.2.2.
I assume root-final *âh1â because of the present PGr. *klÌ¥ne/oâ, on a par with other thematicized nasal presents like βάλλÏ, ÏάμνÏ, Î¸á½±Î»Î»Ï that derive from roots in *âh1â. Vine (2006) and Seržant (2008) reconstruct this root as *kelh3â, but this claim is based on reconstructions and etymologies that I consider to be questionable.
Thematicization of u-stem adjectives, whatever its cause, occurs more often in Greek: cf. Hom. ÏÏÎµÎ¹Î½á½¹Ï ânarrowâ beside ÏÏÎµÎ½Ï â in compounds, ÏÎ±Î½Î±á½¹Ï âthinâ and Myc. ta-na-wa beside ÏÎ±Î½Ï â in compounds (cf. Ved. tanúâ), and Î¼Î±Î½á½¹Ï âsparse, thinâ beside μανύ and Arm. manr (u-stem) âsmall, thinâ.
Attic uses á¼Î¸Ïá½¹Î¿Ï (of uncertain etymology) with the same meaning.
The context is as follows. The Achaean and Trojan armies approach each other; the Achaeans are compared to Notos (the South Wind) which blows a gust of mist over the mountains: á½£Ï á¼Ïα Ïῶν á½Ïὸ ÏοÏÏὶ κονίÏÎ±Î»Î¿Ï á½ÏÎ½Ï Ïâ á¼ÎµÎ»Î»á½´Ï á¼ÏÏομένÏνΠμάλα δâ ὦκα διέÏÏηÏÏον Ïεδίοιο (Il. 3.13â14), âLikewise a thick cloud of dust arose from under their feet as they marched: and they crossed the plain very quicklyâ. The idea that á¼ÎµÎ»Î»á½µÏ is related to á¼ÎµÎ»Î»Î± âgust of windâ is difficult to maintain; see Kirk (Comm. Il., ad loc.).
The Elean form is an adverb in âÏÏ based on the s-stem adjective. It modifies the directly preceding numeral â¨Ïâ©ÎµÎ½ÏακαÏιοÌν, denoting the council of 500 in its entirety (Minon 2007: 36, 511â513 translates âau completâ). This excellently fits the semantics of Hom. á¼Î¿Î»Î»á½³ÎµÏ âgathered together, in a groupâ.
Ex hypothesi, the cluster âlnâ (with consonantal realization of the liquid) would have emerged too late from *âlÌ¥nâ to join the first compensatory lengthening.
In Van Beek 2018: 43â44, I have proposed that *smÌ¥-u̯lÌ¥n-esâ was derived from a compounded verb *smÌ¥-u̯lÌ¥ne/oâ âto flock togetherâ, with *smÌ¥â âtogetherâ functioning as a preverb comparable to Vedic sám.
A primary active perfect is perhaps attested as á¼á½¹Î»ÎµÎ¹ (a likely emendation in Pi. Pyth. 4.233) in the meaning âto push backâ, cf. DELG s.v. Îµá¼°Î»á½³Ï 1.
Comparable derivations of an s-stem adjective from a middle present stem are, for instance, âÏÏεÏá½µÏ from ÏÏá½³Ïομαι and âδεÏÎºá½µÏ from δέÏκομαι. Cf. Meissner 2006: 192â193; most recently, Blanc (2018) has given an extensive overview of such deverbal formations.
Interestingly, the forms attested in West Greek are adverbial and have a petrified lexical meaning âcompletelyâ (denoting a total or sum); this may have helped the preservation of their zero grade root (as opposed to the verb âto throngâ, which had a full grade in West Greek, too: cf. Elean αÏοÏελεÏ).
If the Aeolic outcome of *lÌ¥ was âλοâ, one could theoretically assume that the vowel slot in á¼Î¿Î»Î»á½³ÎµÏ was analogically introduced from the verbal root *u̯elnâ.
In Van Beek 2013: 47 n. 131, I suggested that Myc. wo-ne-we (PY Cn 40.2 and 643.1) could be the nom. pl. of a u-stem adjective *u̯lÌ¥n-uâ meaning âcompactâ, related to á¼Î¿Î»Î»á½³ÎµÏ. I have now changed my mind: see section 2.3.1.
Note that in the Mycenaean outcome of original intervocalic *âlnâ, the nasal has been lost or assimilated: cf. o-pe-ro-te /ophellontes/ vel sim.
It has been supposed that Cret. ÎºÎ»á½±Î³Î¿Ï is from *Î³Î»á½±ÎºÎ¿Ï by metathesis of voice (see the older lit. in Frisk s.v.), but this is both unlikely and unnecessary. Another Cretan form, ÎºÎ»ÎµÏ qÎ¿Ï (gen. ÎºÎ»ÎµÏ ÎºÎ¹Î¿Ï) ânew wineâ (Bile 1988, No. 28) beside Myc. de-re-u-ko /dleukos/, Cret. gen. Î³Î»ÎµÏ ÎºÎ¹Î¿Ï (Gortyn) and Att. Î³Î»Îµá¿¦ÎºÎ¿Ï (Arist.), shows that some regions of Crete underwent a devoicing γλâ > κλâ.
Tarentum was a colony of Sparta.
The evidence from Elis for the outcome of *r̥ is minimal and internally contradictory: see section 3.2.3.
Compare IG VIIÂ 3073.29â37 and 3074.9â11 (Lebadeia in Boeotia), which also contain regulations for construction.
âOn a gardé ici sans modification la translittération de Meister (â¦) la lecture de nombreux signes et la présence de beaucoup de diviseurs apparaissent très incertaines, ainsi même que le sens de la lectureâ (Masson, ICS1 ad loc., pp. 317â318). The text was left unchanged in the 1983 second edition of ICS. In his article, Masson comments: âLââ¯interprétation des deux derniers mots est fort incertaine. Meister voulut reconnaître á¼°(ν) δεκάÏῳ ÏλόÏει âsur la dixième tabletteâ, avec (â¦) une forme *ÏλόÏÎ¿Ï correspondant à lââ¯ionien-attique ÏλάÏÎ¿Ï âlargeur, surfaceâ, qui aurait ici le sens matériel non attesté de âTonplatte, Tonscherbeâ; toute lââ¯argumentation concernant ce dernier terme est peu plausible; dââ¯autant plus que nous ne croyons guère au po initial.â (1966: 263â264).
It is not even mentioned by DELG or GEW s.v. ÏλαÏá½»Ï.