>>How do others with conlangs with the familiar masc-fem-neut gender system
>>assign gender to words?

Cwendaso (or Índumom Tovlaugadóis, as perhaps I should call it, because
that is what its speakers call it) has four genders: masculine, feminine,
epicine, and inanimate.  It's pretty obvious what gender almost anything
should belong to, with a couple of quirks.  The quirks that I know of seem
to involve the epicene gender.

Epicene words such as "parent," "child," "person" and "sheep" can easily be
made masculine or feminine if they are the subject, direct object, or
indirect object of a verb, since the verbs have agreement markers for
these.  You can talk about someone's mother without using <bámam>, the word
for mother, by using the word for "parent" with a feminine agreement marker
on the verb to say, essentially, "female parent."

The other little quirk is that there are a number of words which *we* would
probably guess to be of inanimate gender, which are actually epicene to
speakers of the language.  Among these are <párem> 'life', <índumom>
'speech' or 'language', <aigdngm> 'song', <tovlm> 'teaching' or
'instruction', <tádhwad> 'lamentation', <khángem> 'knowledge', <ngatí>
'head' (together with any other (attatched) body parts), and possibly
<dóvde> 'fire' and <áisadh> 'flute'.  All of these things are perceived as
having some sort of lifelike qualies to them, and so will not be marked
with inanimate agreement markers, but with epicene ones.

Many of the above epicine nouns are abstract.  I suspect that a very large
proportion of abstract nouns in Cwendaso will turn out to be
epicine.  However, I know of one exception: <úmam> 'death' is inanimate in
gender because it has absolutely no lifelike qualities to it.  I suspect
that certain other abstract nouns with the right semantics will be assigned
the inanimate gender.  Examples of such nouns would be "emptiness,"
"darkness," and "cold."

The question that I still have not been able to answer for myself is
whether a human corpse would be epicine or inanimate.  There are good
arguments for each side, and I think that the answer to the question is
culturally significant.