Re: [tied] Re: Verner's Law for *-ns-á > *-n z-á

Yes.  Both of these are morphologically difficult.  Pokorny (IEW 412) regarded OE _gan(d)ra_ and MLG _ganre_ 'gander' as syncopated forms of a weak masculine derived from PIE *g^Han- 'to gape', Gmc. *gan-, parallel to OHG _kat-aro_ 'tomcat', NHG _Kater_.  I consider it more plausible that no syncope occurred in these 'gander' forms, which instead reflect Verner's Law shifting *-ns- to *-nz- (WGmc *-nr-) in the middle cases of a weak masculine noun derived from the PIE C-stem *g^Hh2ans- 'goose'.

Dialectal NHG _Ganser_ (Bavarian etc. against standard _Gänserich_ formed like _Enterich_ 'male duck') is explained in the Grimm-DWB as continuing MHG *gan3er, equivalent to LG _ganter_ from the Gmc. stem *gant- (already in Pliny as _ganta_ 'wild goose', also OHG _ganzo_, MLG _gante_, Du. _gent_ 'gander').  I prefer to see _Ganser_ as a normalization (probably following _Kater_ and the like) of MHG _ganse_ 'gander', OHG *ganso, standing in the same relation to OE _gan(d)ra_ as OHG _haso_, NHG _Hase_, Du. _haas_ to OE _hara_, NE _hare_.  MHG _ganse_ was obsolescent, having only marginal attestation in the DWB with the 14th-cent. Swabian acc. sg. _ganssen_ and 15th-cent. Austrian acc. sg. _gansen_; this form is also cited from the 16th-cent. Moselland.

In this view, the PIE C-stem *g^Hh2ans- 'goose' (originally 'gaper', since geese often have their bills open) gave rise to a thematic oxytone *g^Hh2ansó-, reflected as Skt. _ham.sá-_ 'gander; male swan', and continued as Gmc. *ganza-, WGmc *ganra-, SwG _ganner_ (with _gann_ back-formed) 'diving bird'.  Borrowed into Baltic, this WGmc word yielded Lith. _gañdras_ and Old Pruss. _gandrus_ (MS. _gandams_) 'stork'.  Beside this thematic formation in PIE stood an amphikinetic masculine /n/-stem, parallel to 'hare' from *k^h2as- 'gray' (originally 'lacking pigment, hoary').  I reconstruct the principal singular cases as follows in PIE, Old Western IE (the centumized ancestor of Germanic, Celtic, Italic, Venetic, and Ligurian), Early Proto-Germanic (just after Grimm's Law),  Middle PGmc (showing replacement of the weak stem by the middle stem in the gen. sg.), and Late PGmc (with Verner's Law and other sound-changes applied):

'Gander'  PIE            OWIE        EPGmc      MPGmc      LPGmc

Nom. Sg.  *g^Hh2ánso:n   *gHánso:n   *gánso:n   *gánso:n   *gánso:N

Acc. Sg.  *g^Hh2ánsonm.  *gHánsonm.  *gánsonm.  *gánsonum  *gánsanuN

Gen. Sg.  *g^H&2nsnés    *gH@...   *g@...   *g@...  *gánziniz

Loc. Sg.  *g^H&2nséni    *gH@...   *g@...   *g@...   *gánzini

'Hare'    PIE            OWIE        EPGmc      MPGmc      LPGmc

Nom. Sg.  *k^h2áso:n     *káso:n     *xáso:n    *xáso:n    *xáso:N

Acc. Sg.  *k^h2ásonm.    *kásonm.    *xásonm.   *xásonum   *xásanuN

Gen. Sg.  *k^&2snós      *k@...     *x@...    *x@...   *xáziniz

Loc. Sg.  *k^&2séni      *k@...     *x@...    *x@...    *xázini

In this declension, the loc. sg. form took over the function of the dative.  In West Germanic, *z became *r, and the alternation was not levelled out of the paradigm until distinct dialects had developed.  Some of these dialects generalized -r-, others -s-.

*****

I overlooked the forms of 'hare' reflecting /e/-vocalism in the root, namely Icelandic _héri_ (*hjeri < *hjazi < *hezan-) and dialectal Norwegian _jase_ (*hesan-) discussed by Kroonen (EDPG 223-4).  These require the laryngeal to be *h1, and the paradigm to be corrected as follows:

'Hare'    PIE            OWIE        EPGmc      MPGmc      LPGmc

Nom. Sg.  *k^h1éso:n     *késo:n     *xéso:n    *xéso:n    *xéso:N

Acc. Sg.  *k^h1ésonm.    *késonm.    *xésonm.   *xésonum   *xésanuN

Gen. Sg.  *k^&1snós      *k@...     *x@...    *x@...   *xáziniz

Loc. Sg.  *k^&1séni      *k@...     *x@...    *x@...    *xázini

The Late PGmc alternation was levelled out in four different ways.  Dial. Norw. _jase_ continues the generalization of the strong stem *xesan-, while OE _hara_ reflects *xazan- with generalization from the weak cases.  OHG _haso_ continues a compromise-form *xasan-, and Icel. _héri_ the opposite compromise-form *xezan-.

*****

'Gannet' (OE _ganot_ st. m.) presupposes Gmc. *gano:t- against *ganat- in OHG _ganazzo_ wk. m. 'gander; male coot' and *gant- in Pliny's _ganta_ 'wild goose', OHG _ganzo_ wk. m. 'gander', and other forms.  Gannets dive into the sea with their bills open to catch fish, so I propose a PIE compound root-noun *g^Hh2an-h1o:d-, *-h1od-, *-h1d- 'gaping eater'.  The second element is simply *h1ed- 'eat' in the /o/-grade appropriate for second elements.  Formally, the first element may be an archaic endingless locative 'with a gape'.  Late PGmc would inherit *gano:t-, *ganat-, and *gant- (with absorption of *h1 in this position).  'Gannet' has generalized the strong stem to all the cases like 'foot'.  The middle stem underlies _ganazzo_, while the weak stem gives _ganzo_ and Pliny's _ganta_ (probably a Gmc. st. f. *ganto:).  The original sense of 'diving fish-eating bird' preserved in NE _gannet_ was not maintained well in continental West Germanic.

DGK