Re: More Þrjótran

> I've been thinking of this.  In this particular word
> EGÔ the G probably disappeared early in Romance, since
> all Romance languages point to *EÔ or *IÔ, so you
> would probably end up with _jó_, but I don't think
> intervocalic G should disappear generally: in North
> Germanic *x disappears in most positions (e.g.
> *slaxan > _slá_) but intervocalic *g remains:
I don't know where I had the information from -- the
corresponding rule was not commented.  The rule was
a devoicing rule for G that shifts it to x (written h),
which then disappears.  When trying to check for
which PG example word I needed this, I came to the
conclusion that it was an overgeneralised rule in the GMP.
Probably accidentally, it worked for all words I used
for testing.  I added 'slo:gum' to the suite and, wrongly,
it came out as 'slo:m'.

I.e., the rule was probably wrong -- thanks for noting!
I fixed it and the test suite still works plus 'slo:gum'
also works.

The new rules produce 'jaga' for 'ego:' now.
We'll see. :-)  I think it could reasonably become 'ja:'.

> (e.g.
> *slaginaz > _slaginn_, *dagaz > _dagr_.
Those were correctly handled before, too, they are part
of my test suite.

> If you are thinking of Verner's law I don't think
> you should apply that, since it clearly antedates
> your timeframe.
No, no, of course I don't.  The only cheating I do in
this respect is voicing Latin final -s -- otherwise the
inflection would look nothing like Norse. :-)

**Henrik

Reply