Rolf Noyer: Gender in DM (reply to Claire Bowern)

[Martha McGinnis]

Hello again, I'm back on the soapbox...

I think Claire is correct that the terminology as it is used in the
literature is confusing.  This is why it is important to agree on what we
are talking about before we talk about what we are talking about.  When I
(or Aronoff, for that matter) define _gender_ and _class_ in a certain way,
it is not my intention to characterize the way in which the term is used in
the literature by other authors.  It is my intention to define what *I*
mean by the term and to propose a definition of this term for the sake of
discussion.

There is nothing wrong with a language with 15 genders, where "gender" is
defined as a property of a substantive which propagates syntactically such
that distinct syntactic terminals may be required to "agree" in gender.  Of
course, this doesn't mean that people who have written about the language
will use the term "gender" to refer to this property; they may very well
use the term "class" where I propose we use the term "gender."  Suppose for
some reason -- any reason -- the author doesn't like the term "gender".
They might just as easily have called these subdivisions of the vocabulary
"wugs."

Personally, I prefer the term "inflectional class" when referring to
properties which do not propagate syntactically but can affect (within some
to-be-ascertained local domain) the spell-out of morphological formatives
like "case" and "number" of a stem.  Gender may condition inflectional
class.  For example, the default value for [fem] gender nouns in Spanish is
inflectional class I, characterized by "word-marker" /-a/ (Harris).  In
simple cases, the relationship between gender and inflection class may be
so transparent as to make the postulation of distinct inflectional classes
unnecessary: in such cases rules of allomorphy refer directly to gender and
not to inflection class diacritics.  This may be the case in languages
where the descriptions frequently use "class" as a cover term for both
gender and inflectional class.

One should also be careful not to confuse semantic "classes" and
grammatical categories.  An interesting discussion of the distinction
between semantic "classes" and grammatical categories can be found in
Robert Beard's book _Lexeme-Morpheme Base Moprhology_.  There are lots of
languages where the semantic properties of shape, say, affect grammatical
categorization (e.g. "long thin" objects generally go in one category, "fat
round" objects in another; or flora may _tend_ to go in one category, fauna
in another, in the default instance).  To my mind, these relationships are
on par with the way in which natural gender influences grammatical gender
(which in turn may influence inflectional class). All three notions should
be kept distinct.

One has to approach the use of terminology in the literature with some
caution. To take another example, "case/Case" means many things to many
people. A priori we have no reason to believe that all instances of
phenomena referred to as "case" in the literature are the same.  Likewise,
not all "clitics" are the same; indeed I suspect "clitic" is not
ontologically a primitive at all but a descriptive convenience.  That is
the essential point: in descriptive literature (and by descriptive
literature I mean most of what gets written in linguistics), terminology is
chosen for desriptive convenience rather than in the context of general
theory of grammar.

Rolf