> This is a nonsense.
> No Indoeuropean language has really the masculine gender;
> what is called the masculine gender is really the common gender.
> There is the -H2 suffix marking feminine gender
> but no corresponding suffix to mark masculine gender.
> So it is the feminine gender which is privileged by the grammar.
When I finally reach an insight, I tend to overuse it; in casu
Miguel's *-k# > *-x#.
I think that that suffix was never a "collective". It's an
individuating suffix, which generates names of individual
persons/object from names of non-countable
person-categories/matter/stuff/.
I think it survives in Nordwestblock -k. eg. Danish 'padde' amphibian,
English 'paddock'. Also in the NWGermanic -ke/-tje 'diminutive' suffix.
As to its use, I will use an example from the little Dutch I know:
bier "beer", non-countable stuff
biertje "glass/bottle of beer", countable stuff
pils "lager", non-countable stuff
pilsje "glass/bottle of lager", countable stuff
Those Dutch 'diminutives' ('individuals'?) do have perfectly normal
plurals. But even in,
'bier drinken'
vs.
??'biertje drinken'
where the last phrase is very bad Dutch, you get a feeling of
different semantics: "drinking beer" vs. "drinking glasses of beer"
therefore, in PIE, there was a semantic difference between
*<root>-om : non-countable stuff amount of <root>
(remember *-om is in its origin a partitive)
and
*<root>-ex : individual(s)/piece(s) of <root>
and then the context decided whether you chose to understand it as one
or several, in the first case as 'feminine' (mostly, but not always),
in the last case as 'plural of the corresponding <root> stuff (non-
countable 'stuff' being by nature neuter)
Also, I think the verbal factitive *-x is an individuating,
singulative suffix: *nove-x- "one piece of newness, ein Stück Neues".
With verbal suffix it becomes "makes one case/instance of newness".
Torsten