Re: Latin /j/ etc. (was: Latin <h>)

Quoting Ray Brown <ray.brown@...>:

> On Thursday, January 15, 2004, at 12:22 AM, Tristan McLeay wrote:
>
> [snip]
> > My revealing-ignorance question is how do we know that intervocalic /j/
> > was always geminite in Latin?
>
> Briefly:
> 1. The preceding syllable is always reckoned as heavy in Latin prosody.
> This by itself, of course, doesn't prove gemination; an intervocalic /j/
> could, for some reason, cause the preceding vowel to be lengthened; but it'
> s
> difficult to see why and i kow of no parallel examples in other natlangs.
With the exception of some funky loans like _kaviar_  - which is ['kavjar] in
my 'lect - pretty much every stressed vowel with a following /v/ is long in
Swedish. There are people who say ['kA:viar] or similar, too.

BP or Daniel Andreasson may know why, if there's a reason! :)

                                                          Andreas

PS Conversely, you don't normally get a long vowel before /j/. Can't think of
a counterexample that don't have a morpheme boundary before the /j/.