From: Trond Engen
Message: 68321
Date: 2011-12-29
> --- In cybalist@yahoogroups.com, "dgkilday57" <dgkilday57@...> wrote:I enjoy these long reasonings evoking analogical levellings and
>
>> [...]
>>
>> As I noted elsewhere, <reord>, <leort>, and <on-dreord> belong to one
>> subclass of verbs, and <leolc> and <heht> to another. I believe
>> their formations are distinct: the three former are /r/-preterits,
>> while the latter two are reduplicated.
>
> The Old English reduplicated preterits <leolc> 'jumped' and <heht>
> 'called, ordered, effected' are sometimes equated directly with
> Gothic<laíláik> [lelaik] and <haíháit> [hehait]. The principal
> difficultywith this is that a long vowel (OE -a:- corresponding to
> Go. -ai-) mustbe abnormally syncopated in the second syllable of
> these preterits.
>
> Gothic <láikan> 'to jump', with its Germanic cognates meaning 'to
> jump,play, be active, fly, swim', etc., is traditionally referred to
> aProto-Indo-European *loig-, the /o/-grade of Pokorny's root *leig-
> 'hüpfen, beben, beben machen' (IEW 667-8). Greek <elelízein> 'to
> causeto shake' and <elelíkhtho:n> 'earth-shaking' are usually placed
> herealso. Mallory and Adams thus cite the IE root as *h1leig-
> 'jump', buttheir implicit analysis is faulty. A root of this shape
> would form theGreek weak perfect stem *ele:lig- (*h1le-h1lig-) by
> so-called Atticreduplication, just as *h2/4leibH- 'smear, anoint'
> forms theAttic-Ionic perfect <alé:lipha> (*h2/4le-h2/4libH-) to
> <aléipho:> 'Ianoint'. Instead, <elelízein> is based on an isolated
> Epic Greekthird-person singular pluperfect passive <elélikto> used
> inan aoristic sense. From this <elelig-> was extracted and used to
> make a new aoristactive <eléliksa> and passive <elelíkhthe:n>, and a
> new /j/-present<elelízo:>. But the original weak perfect stem in
> <elélikto> is*lelig-, which indicates a root with no initial
> laryngeal.
>
> On the other hand, the accent of Lithuanian <láigyti> 'to run around
> wildly' points to a root-internal laryngeal. In this connection it
> is useful to examine the vocalism of the Germanic forms. The present
> system continues Proto-Gmc. *-ai- in Old Norse <leika>, Old Swedish
> <le:ka>, and Old English <la:can>, as do the quotable past
> participles, ON <leikenn> and OE <la:cen>, and the Gothic preterit
> 1/3sg. <laíláik>. Proto-Gmc. *-e:- is reflected by the pret. 1/3sg.
> ON <le:k>, OSw <læ:k>, and OE <le:c> (more usual than <leolc>). As
> is typical for class VII strong verbs, the pret. pl. has the same
> stem as the pret. sg., with the exception of the OSw pret. pl.<liko>.
>
> This OSw archaism suggests a possible explanation for OE <leolc>
> which does not require ad-hoc syncope of the long syllable in a form
> parallel to Go. <laíláik>. In this and similar reduplicated forms
> Gothic has evidently generalized the /o/-grade of the PIE perfect
> singular to the plural (e.g. <haíháit> 1/3sg. 'called', <haíháitum>
> 1pl.). Prehistoric OE (or Proto-Anglo-Frisian or whatever) could
> have generalized the zero-grade, corresponding to OSw <liko>, from
> the pl. to the sg. That is, <leolc> (in the southern dialects of OE,
> since Northumbrian kept -io- and -eo- distinct) could continue
> earlier *liolc, following the old pret. pl. *liolcun, regularly
> broken and syncopated from the inherited 3pl. *lilicun. With this
> root, the */e/ of the reduplicator would have undergone /i/-umlaut in
> Proto-Germanic; the perf. pl. stem *le-lik- would have regularly
> become *li-lik-. This might well have been restored to *le-lik- by
> analogy with the perf. sg. stem *le-laik- and with other perf. pl.
> stems having a root-nucleus other than */i/. (In Gothic, the
> reduplicator evidently escaped the general raising of */e/ to */i/ by
> piecewise analogical restoration. It is too much to believe that all
> Go. redup. preterits follow the purported example of those with
> root-initial /r/, /h/, and /hW/, and have lowered their */i/ back to
> /e/.) But in prehistoric OE, a restored plural stem *lelik- would
> again have become *lilik- during the stage of specifically OE
> /i/-umlaut, when reduplication was moribund and no significant
> morphological pressure existed to restore the */e/.
>
> The PGmc *-e:- (i.e. close */ê/) reflected in the unreduplicated
> preterits evidently arose from Paleo-Gmc. *-e:i- (cf. Sievers, PBB
> 18:409-10, 1894) in closed position, and this is actually the PIE
> /e/-grade *-eh1i-. This type of preterit is thus a root-aorist in
> origin, not a perfect. Lith. <láigyti> must then continue the
> /o/-grade *-oh1i-, with the acute accent as the laryngeal's shadow.
> But this sequence *-oh1i- in closed position did not become *-o:i- on
> its way to Germanic. Instead the laryngeal simply vanished here, and
> Paleo-Gmc. *-oi- became PGmc *-ai- in the /o/-grade. (E.g. PIE
> *meh1- 'great', *móh1-is 'greater, more' (neuter as adverb),
> *móh1-is-on- 'greater' (adjective)> PGmc *maiz adv., *maizan- wk.
> adj. > Go. <máis>, <máiza>, OE <ma:>, <ma:ra>.) In the zero-grade,
> the postconsonantal laryngeal vanished before */i/. We thus have the
> following scheme:
>
> Root-aorist stem: PIE *leh1ig-> PGmc *le:k-
> Present stem: PIE *loh1ig-> PGmc *laik-
> Perf. sg. stem: PIE *le-loh1ig-> PGmc *le-laik-
> Perf. pl. stem: PIE *le-lh1ig-> PGmc *le-lik-> *li-lik-
>
> M&A's root *h1leig- 'jump' should thus be corrected to *leh1ig-.
> The fact that Sanskrit <ré:jate:> 'hops, shakes, quakes' and its
> relatives show no sign of the root-internal laryngeal is no problem.
> Lubotsky in "Reflexes of intervocalic laryngeals in Sanskrit"
> (Kuryl/owicz Mem. Vol. 1:213-33, 1995), has shown that
> Proto-Indo-Iranian *-ah{x}i- became Skt. -e:- within a single
> morpheme, or more generally where there was no morphological pressure
> to restore the laryngeal. An example is <dhe:nú-> 'dairy cow'
> (occurring 122 times in the Rig-Veda) from PIE *dHeh1i-nu-. (In
> open position, *-oh1i- was actually *-oh1j- which became Paleo-Gmc.
> *-ojj-> PGmc *-ajj-. E.g. PIE *dHeh1i- 'suck', prevocalic /o/-grade
> *dHoh1j-> PGmc *dajjanaN, Go.<daddjan>, ON<deggia> 'to suck'.)
>
> Similar considerations apply to Go. <háitan>, ON <heita>, OSw
> <he:ta>,OE <ha:tan> 'to call, name, command'. If the PIE root is
> taken as *kei-d- (extended from *kei- 'to set in motion'), there is
> no principled way of getting Gmc. *-e:- into the unreduplicated
> preterits, ON 1/3sg. <he:t> (beside <heit> with pres. vocalism), OSw
> <hæ:t>, OE <he:t> (beside less usual <heht>). The PIE root should be
> *keh1id-. For zero-grade I do not have an archaic pret. pl. like OSw
> <liko> above, but I believe that the OE pret. pl. <hehton> continues
> a zero-grade form. It is striking that <heht> and <hehton> do not
> show breaking in West Saxon like <feoh>, <feohtan>, etc. This
> suggests that they are hypercorrect forms, borrowed after the stage
> of /e/-breaking from a dialect which frequently had -eo- for WS -e-,
> so that word-borrowers would be likely to implement a knee-jerk
> rule: When borrowing from this dialect, always replace -eo- with
> -e-. In both Kentish and Mercian, back-vowel umlaut of -e- to -eo-
> (involving -a-, -o-, or -u- in the next syllable) was more extensive
> than in pure West Saxon, where it was usual only after a simple
> liquid or labial, e.g.<heorot> 'hart',<heofon> 'heaven'.
>
> Thus I propose that the Anglo-Frisian 3pl. pret. *hihitun (parallel
> to *lilicun above) was retained in Kentish (or another dialect with
> extensive back-vowel umlaut) and was regularly syncopated to *hihtun,
> broken (or umlauted) to *hiohtun, and lowered to *heohtun. As
> assumed above for <leolc>, the pret. sg. *heoht followed the pret.
> pl. These forms were then hypercorrectly West-Saxonized as <hehton>,
> <heht>.
>
> Root-aorist stem: PIE *keh1id-> PGmc *he:t-
> Present stem: PIE *koh1id-> PGmc *hait-
> Perf. sg. stem: PIE *ke-koh1id-> PGmc *he-hait-
> Perf. pl. stem: PIE *ke-kh1id-> PGmc *he-hit-> *hi-hit-
>
> I have no problem taking the *-d- as a root-extension, but M&A's root
> *kei- 'set in motion' should be corrected to *keh1i-. Pokorny had
> the right idea, citing it as *ke:i- 'in Bewegung setzen, in Bewegung
> sein' (IEW 538-9), not as *kei-. While Latin <citus> 'quick, speedy'
> has the expected zero-grade *kh1i-, Greek <ki:né(w)o:> 'I set in
> motion' apparently has a metathetic zero-grade *kih1-. This type of
> zero-grade also occurs in Germanic. Old Swedish <di:a> 'to suck'
> evidently reflects metathetic *dHih1-, not *dHh1i-, to the prevocalic
> full grade *dHeh1j- in *dHéh1jos, Skt. <dhá:yas> n. 'act of sucking'
> (and the /o/-grade *dHoh1j- in Go. <daddjan>, ON <deggia> above). I
> could give several paragraphs of similar metatheses involving roots
> in *-eh{x}i- and *-eh{x}u-. Unfortunately, I cannot fully state the
> conditions under which laryngeal metathesis occurred.