>The Italian/Sicilian/Portuguese connection can't be.
Not connection, but similarity. After all, all of them are
"developments" of the Latin perfect (which was one of the
verbal morphemes).
>The -st- part is 2nd person only. The Romanian fus- perfect
>seems to be back-formed from the pluperfect
I don't know. But I rather suspect that Romanian fost
(probably an earlier fust) stemmed from the perfect fui-st,
with i vanished and later u > o). Fus- on the other hand
is ->in Romanian<- as stable as they come (both in the
"simple perfect" fusei, fusesi, fuse, fuseram, fuserati,
fusera, and in the plusquamperfect tense fusesem, fusesesi
etc. There ain't no fus- > fos- whatsoever.)
>(cf Italian and Portuguese imperfect subjunctive, and
>that seems to derive from the Latin pluperfect subjunctive
Here, no comparison with the Romanian subjunctive possible,
since Romanian has built it in a different way: sã fiu,
fii, fie, fim, fitzi, fie (with regional variants hiu,
hii, hie... and siu, sii, sie...)(and subj. perf.: sã fi
fost).
>and then of course fost may derive from the fus- perfect,
>as Piotr says.
But which could have been the participle vowel before the
-t in order to later on have been dropped?
>But still, it's one of a kind, in an irregular and
>'non-thematic' (-t) verb.
It must've been a contracted form in a time when the
idiom wasn't yet Romanian, but no longer genuine Latin. :o)
Anyway, the existence of the fost/e form in the very
same perfect tense pattern in some of the Romance
languages must be a sign pertaining to some pattern or
fad that was common to old Romanian as well. These can't
be mere coincidences. (My guestimation.)
George