Actually, I would start by following up the suggestions made in another thread
on this topic earlier this year:
esp. re prosodic features and tonal patterns. Though this is not my area, I
personally thought that Celso's comments on how code-switched sentences can
'fit' melodic and rhythmic patterns particularly insightful (see his example of
trilingual sentences in a two-beat trochaic rhythm found in Galazian music). As
these accord with Sergio's comments on the conflict between Italian throcaic
endings and musical styles with a strong back-beat, I would think that using
rhythm as a starting point would be a good idea.
Unfortunately, this really isn't my area, and I am afraid suggesting a
more-packaged theory to apply is beyond me at this point. However, this topic
has been discussed here before, and I am sure options will present themselves...
Good luck, Ray, with the very interesting work!
Peace and Love,
[Non-text portions of this message have been removed]