impermissibles in names

[Logical Language Group <lojbab@GREBYN.COM>]

I think Cowan answered Iain and Colin on this, but am not sure.  So I'll
give my own answer:

>la kolin. cusku di'e
>> .uu zoi gy. Bradford gy.
>
>> I just noticed I've been using a lojbanization of Bradford that is not
>> only not a good rendering ("bradfyd" would be better) but is also
>> invalid phonologically.
>
>> Henceforward, I will repair this by going to a really local
>> pronunciation:
>
>>         bratfyd
>
>cu'u mi
>> > I'm curious to know what's phonologically invalid about
>> > "*bradfrd".  It doesn't break any rule I can remember coming
>> > across.
>
>> "df" is not a permissible medial (voiced-unvoiced).
>
>So which is it, guys?
>>
>mi'e .i,n.

The word resolution algorithm (and presumably Lojban speakers) will
presume that names have only permissible medials.  Thus, when they hear
a voiced/unvoiced combination in a name, they will probably presume
sloppy pronunciation and hear it either as 'bratfrd' or 'bradvrd',
depending on whether voicing or devoicing predominates in the speakers'
accent, or perhaps in the listeners native speech (Russian, for example,
presumes devoicing in such situations).  (This presumes they don't
already know what the Lojbanized name refers to).

It would be legal using the local pronunciation (bratfyd), or if that is
too unrecognizable, use a 'y' as a cluster legalizing 'buffer':
bratyfrd/bratyford.

lojbab