Re: pe'a & tanru

[ucleaar <ucleaar@UCL.AC.UK>]

Andruc to Skot:
> >Hold on.  First, I disagree that we've established that it's a bad
> >tanru.  I can give several examples of "good tanru" ("good" because
> >they appear in official LLG papers) where {broda brode} isn't a real
> >{brode}.  How about {rokcu cinfo}?  What species (selcinfo) is a stone
> >lion?  Given that the notion of species has to do with genetic
> >compatibility, I'd argue that any "species" one might claim for a stone
> >lion is purely metaphorical.
> Well, let's agree to disagree on whether it's a good tanru.
> In my opininon, {rokci cinfo} must have a species, because that place
> exists in the structure of the gismu. Tanru take the place structure of
> the last element, so {rokci cinfo} is a rocky-type of lion of species x2.
> Crucially, if it doesn't have a species, then it's not a cinfo.
> And of course a stone lion isn't a lion, is it? It's a stone.
> I would say that a stone lion should be translated as {cinfo rokci},
> i.e. a lion-like stone of type x2 from location x3.

What you say certainly, pehi, makes for good clear usage, and I would
humbly commend it. But technically, I think you're wrong. Syntactically,
{cinfo} still has an x2, but since the nature of the cinfo relationship
dissolves when it becomes part of a tanru, there is no guarantee that
semantically the x2 will continue to express the species of x1. But of
course one would generally suppose that it did so continue, and
consequently one would do well to follow your advice.

coo, mie and