Tim May wrote:

(quoting someone else, I think)
 Roots differ in terms of which voice/focus/nominalization
>  > > affixes they can take, but they don't split cleanly into verbs and
>  > > nouns.  Nearly all roots can take at least some, and very few can
> take
>  > > all.
>  >
>  > That's very interesting. If nearly all, however, any idea why not all
> can
>  > take at least one?
>  >
Tim May:
> I've no idea.  Himmelmann says "practically all", and I don't think he
> gives any counterexamples - it's possible that he doesn't know any,
> but can't confidently make a universal claim.
>
It's probably not possible to come up with a "rule"; the same is pretty much
true of Indonesian and other "Western" AN languages. ("Eastern"-- Moluccan
and Oceanic-- langs. have lost so much of the active morphology that they
_may_ have developed more clearly defined word classes). But notice that
Engl. isn't entirely clear-cut either.

In Indo., no matter how "noun-y" a word may be, most can at minimum take the
ber- prefix, meaning "to have..." or "to do ...habitually or as a
profession" or "to have the quality of..." et.al.  It seems that in Tagalog,
almost any "noun" can take the mag- prefix, which interestingly is cognate
with the Indo. one, though I don't think the meanings are quite the same.

OTOH, Indo. seems to have more words (comparing the dictionary defs. at
least) that are defined as "verbs/adjectives" and which occur with the
clearly verbal prefixes/suffixes, and if they have a noun forms, they're
derived. But again, their base forms can almost always co-occur with the
possessive markers, which essentially turns them into nouns-- pikir/ku
'think-my'= my thought (is...), what I think is...; datang/nya
'(his/her/its/the)coming/arrival'

> Other papers by the same author [Himmelmann] may also be of interest; in
> particular, this sketch of Tagalog:
> http://web.archive.org/web/20040728131501/www.linguistics.ruhr-uni-bochum.de/~himmelmann/tagalog_Curzon.pdf

He does good work. That's an excellent little description, very concise.
>
> (This is one of several papers recently removed from his web page (I
> think they're finally being published) but which can still be accessed
> from the copy at the Internet Archive:
http://web.archive.org/web/20040508005935/http://www.linguistics.ruhr-uni-bochum.de/~himmelmann/publications.html

Hmmm-- most of the links for the .pdf's don't work-- I hope just a temporary
glitch.