On 8 Jan 2012 at 23:33, Aaron Griffith wrote:
> >ro·ḟitir ní·labrathar
> >labrathar, ni·fitir
> I wonder if this argument is valid or not. There is no way to know, but
> ro·ḟitir is clearly relative, so it seems that the meaning _should_ have
> been clear to native speakers. Still, however, the syntax with relative in
> initial position is uncommon. The opposite order would accomplish the same
> thing (ní·labrathar ro·ḟitir, etc.) and it has the advantage of being more
> common. Tomás Ó Cathasaigh has a nice article on these relatives with
> zero-antecedent in Celtica 21 (downloadable for free at
Ó Cathasaigh shows that this construction is - well, not common - but
frequent enough. The next problem that I see with it is not so much
the permissibility of the construction, but it's ambiguity, which in
all likelihood entirely depends on the verbs and verbal forms used.
Suppose we have:
The automatic interpretation of these sentences for me would not be
to supply the subject "int í", but rather the object "a n-í":
"he does not say (what) he knows
he does not know (what) he says"
In most examples collected in Ó Cathasaigh's article, the
relationship between the two clauses is clear from the context and
thus unambiguous, even though grammatically alternative, i.e.
ambiguous readings would be possible. E.g.,
could well be translated "he whom he slays will be slain", but this
is rather non-sensical in the context. Or
ra·fitir as lia
instead of meaning "the majority knows it", could also be "he knows
it, i.e. the majority". Maybe I am blinded at the moment, but in our
sentence I do not see how the ambiguity could be removed without a
lot of explanations that would ultimately remove the desired
conciseness of the expression.