--- In
cybalist@yahoogroups.com, "dgkilday57" <dgkilday57@...> wrote:
>
> [...]
>
> My explanation above is not particularly good, and in the meantime I ran across another /r/-form of 'Hunsegau', namely <Hunergo> cited by van den Bergh with no MS. date, but <Hunesco> is dated 996 (and the <c> is not necessarily an error, but may represent progressive devoicing). Our MS. of the Traditiones Fuldenses is from the 11th cent., but the forms in -gewe make it likely that the material came from an 8th-cent. text. The passage is "in pago Hunergewe in regione Fresonum". Now, if the forms <Hunergewe> and <Emergewe> are genuine old forms, not resulting from bad copying or reading, beside the simple river-names which ALWAYS have <s>, Hunesa, Hunese; Emesa, Emese, etc., we are most likely dealing with Verner doublets. The only way I can account for this is by presuming that the Gau-Formen were originally adjectival bahuvrihi-compounds rather than determinative tatpurus.a-compounds, or were modelled after bah.-cpds. That way the pre-shifted Proto-Germanic accent would have been on the stem-vowel of the first element, and Verner would have voiced the preceding /s/ in the Gau-Komposita. But the simple river-names would require the accent on the vowel BEFORE the /s/, which would prevent its voicing. Naturally, during historical times the Gau-Komposita would tend to level their first elements to the simple forms, so if this mechanism works, we must postulate that this little corner of East Friesland preserved some very archaic compounds by accident.
My comments were somewhat opaque. First, MacDonell's example of the contrast in Vedic accent between determinative (tatpurus.a) and possessive (bahuvri:hi) compounds is this: determinative, <ra:ja-putrá-> 'king's son'; possessive, <ra:já-putra-> 'having kings as sons'.
Second, if the forms in '-gau' were originally possessive compound adjectives, the Common Germanic strong neuter *Gawja- (attested in Gothic, acc. sg. <gawi>, gen. sg. <gaujis>, and in West Gmc. but not North Gmc.) must have been extracted from substantivized neuter forms, presumably agreeing with the neuter *lanDa- 'land' understood.
The etymology and original meaning of *Gawja- are obscure. Köbler suggests for the latter 'Landschaft, Gegend, Gau, Gesamtheit von Dörfern(?), Umgebung eines Gewässers(?)', the last being very plausible due to its high frequency of attachment to river-names. Moreover if the original sense was 'watershed, drainage basin', I may conjecture a derivation from PIE *g^Heu- 'to pour'. The formation would be an /o/-grade deverbative *g^Hou-yó- 'passively connected with pouring', as a substantive 'that into which, or from which, one pours, e.g. a basin or trough'. As an example of this formation we have Latin <folium> 'leaf' from *bHol-yó- 'passively connected with blooming', from *bHel- 'to bloom'.
In Sanskrit this suffix *-yó- has become <-yá-> and acquired gerundive force, and the root *g^Heu- has become <hu-> (class III <juhoti>) 'to sacrifice', from the specialized sense 'to pour into the fire'. If my conjecture is correct, Skt. <havyá-> 'that which should be sacrificed' is fully cognate with Gmc. *Gawja- 'trough' vel sim., despite the difference in meaning.
Now, in order to explain <Hunergewe>, <Hunergo>, and <Emergewe> beside <Hunese>, <Emese>, etc., pre-shifted forms of the river-names *Kuneísi- and *Ameísi- (or the like) must be postulated, so that the simple names retain the /s/ in Gmc. *Huni:si, *Ami:si- (assuming parallel formation). Possessive compounds *Kuneisíg^Houyo-, *Ameisíg^Houyo- 'having the Hunse/Ems as a trough, i.e. draining into the Hunse/Ems' would however become Gmc. *Huni:riGawja-, *Ami:riGawja- leading to the presumed 8th-century forms <Hunergewe> and <Emergewe> naming the drainage basins. As is typical in such situations, the first element of these compounds would tend to be levelled to the simplex, and so forms like <Hunusga>, <Hunesga> and <Emisgowe>, <Emisga> are more common.
Of course it may well be that the /r/-forms are erroneous or arise from secondary phonology (and I have already argued for both positions in previous postings). My conjecture about *Gawja- should still work, but with the proviso that all surviving old forms in '-gau' have levelled their first elements, if applicable.
DGK