Hi, here my results. A * means with RC, y means perceivable.
The columns are the 3 possible intensions. Y means very perceivable.

Marga* ynn
Marieke YYn
Masha* nnn
Jacob* nnn
Donijs* ynY
Auke2 YYn
Suzanne nYn
JosMirte* nnn
MargaMirte* YYY
RemcoMirte YYY
MariekeMirte YYn
LennekeMirte YYY
AukeMirte YYn
Lies YYY
JorritSanne nnn
LennekeSarah YYY
JorritSarah* nnn
RemcoSarah yYY
SuzanneSarah nYn (the first NP stressed has F = 2.0 but this still is n)
LiesSarah YYY
DonijsSarah nnY
JacobSarah* nnn (first: F = 2.2) (whoops, Donijs was also *, I forgot)
JosSarah* Ynn
MargaSarah* YYn

So what does this tell us? If there is no RC, then if anything, the NP3
perception troubled us. Except for Suzanne, of course. Sarah had no ear
for the NP3/Marga, but ? did not perceive NP2/Marga either.
Jacob was kind of not perceivable. For Masha, the ?only? data set tells the
same. Donijs could "do" NP1 and 3, but for Sarah, only NP3. Auke had no RC
but could not "do" NP3 anyhow. Marga (with RC) was easy, but not for NP3 for
Sarah. Remco was completely easy, but without RC. Marieke had no RC but did
not manage to make NP3 prominent. Lies had no RC and was easy. Same for
Remco.

So with RC we have: Marga: Ynn/YYY, Jos Ynn/nnn, Jacob nnn/nnn,
Jorrit nnn/nnn, Masha nnn/???, Donijs nnY/ynY.

Pretty much of "n" in the second column of the "with RC" summary, thus!

The details can be found in metaperccalc.tgz in our data area online,
but I guess you do not really want to read the metaperccalc.log ...

CC to ChSc in case he is bored :-).

Eric - metaperccalc.tgz

PS: I know there is a big fat sparse data problem with this, but I think
the results are still useful.

PS Sanne: Tell me when you need that HTML version of my paper sections!
the DOC or RTF version is no real option: StarOffice runs just as bad on
16 MB RAM as Word 2000 does. If you do not believe, just remove all but
16 MB of RAM from your PC and try to run Word...

Greetings, Eric