[conlang_learners] IE vs non IE

[Jim Henry]

2009/6/15 Olivier Simon <cafaristeir at yahoo.com>:
> So, I told there are two criteria that can be used to characterize an IE
> grammar:
> - Presence of flexion (this criterium is common with Semitic). Even Modern
> IE languages still retain some traces.

That's too vague by itself to characterise a language as IE-like,
though in combination with other features, maybe.

> - The neutral gender has similar forms at the nominative and the accusative.

Yes, that's common AFAIK to all IE languages that still have distinct
case forms (either just in pronouns, as in English (invariant "it" vs.
he/him, she/her) or in nouns and perhaps adjectives as well), and
perhaps rare outside IE (that I don't know, but I would suspect so).
Noun classes that correlate even roughly with sex ("grammatical
gender" in the strict sense) is also an unusual feature that would
characterize a conlang as IE-like (or Semitic-like, depending on what
other features it co-occurs with); IE and Semitic are the only
language families I know of that have this correlation between sex and
the grammatical gender/noun class system.  Most other languages, if
they have a morphological noun class system at all, have different
categories -- a simple animate/inanimate or human/nonhuman split, or a
larger set of categories where one of the categories is probably human
or animate and others but there isn't a masculine or feminine category
as such.   (One of my early sketchy artlangs had this unreflectively
IE-like feature in an otherwise non-IE-like language: it had half a
dozen or so noun classes, with the nonhuman classes similar to those
of Swahili, but there were three human classes, masculine, feminine
and epicene.)

> This may come from the fact that Primeval IE had no Nominative for the
> neutral gender.

There's a good discussion of this issue somewhere in John Lyons
_Introduction to Theoretical Linguistics_ -- he talks about the fact
that inanimate things (roughly correlated with neuter gender in IE
languages, and perhaps more strongly so in Proto-IE) are rarely
agents, so there's less need to distinctly mark agent vs. patient for
nouns denoting such things.

-- 
Jim Henry
http://www.pobox.com/~jimhenry/