Karlos Arregi: gender

[Martha McGinnis]

I agree with both Alec and Martha that gender should be viewed as a
property of n, rather than roots.

As pointed out by Jonathan for Italian and Russian, in Spanish many roots
seem to belong to a specific gender. For instance, the root in caj-a `box'
is always feminine (*el caj-a, *el caj-o). However, much more than this
needs to be said. Many roots can appear with both genders. In some cases,
the meaning difference is completely predictable from natural gender, as in
gat-o/a `cat' or chic-o/a `boy/girl'. In others, the meaning difference is
totally unpredictable, as in libr-o/a `book/pound' or pal-o/a
`stick/shovel'. Finally, there are cases in which the two words are
related, but the meaning difference is not reducible to natural gender, as
in naranj-o/a `orange tree / orange'.

A very simple way to handle all these cases is to say that gender is a
property of n, and that all these roots can be inserted in the context of n
regardless of gender. Then, their encyclopedic entries are specified for
different meanings according to gender. To say that there are two different
roots for each gender would be to complicate things unnecessarily. If this
view is tenable, then, I see no problem in extending it to roots which only
can appear in the context of one gender. For instance, we can simply say
that the root caj- means `box' in the context of feminine n, and that no
meaning is assigned to it in the context of masculine n.

Best,

--Karlos