Theiling Online    Sitemap    Conlang Mailing List HQ   

ciantwo class system, verbs, & semantic roles

From:Ed Heil <edheil@...>
Date:Wednesday, June 23, 1999, 22:39
OK, now I'm trying to decide how to do verbs.  I'd like it to work
like this:

The classes fit into an animacy heirarchy which goes like this:

   1st person > 2nd person > HUMAN > ANIMAL > PLANT >
     CONSTRUCT, THING, STUFF, PLACE, POINT

Any given verb has a role heirarchy.  For example, "nun", the verb
for "to see," has the role heirarchy: EXPERIENCER > PERCEPTUM.

Nouns in the sentence are assigned to the role heirarchy according to
the following plan:

1. For nouns of equal animacy, the first one is assigned to the
higher role in the role heirarchy.

e.g. "Bob Dave see" translates as "Bob sees Dave."

2. For nouns of different animacy, the higher-animacy one is assigned
the higher place in the role heirarchy.

e.g. "Man dog see" = "Dog man see" -- both translate as "The man sees
the dog."

Under this system, of course, there is no way to say "The dog sees
the man" (at least with the verb for "see" which has the role
heirarchy experiencer > perceptum)  Obviously there has to be a way to
get around this (other than suppletion of verbs, though I want to
include some of that).  These are the things I've been tossing around:

3a. If clitic pronouns are attached to the verb, roles are assigned
to them in order (high-role first); the nouns in the sentence which
correlate with them get those roles.

So you could say "Man dog ANIMAL.HUMAN.see" meaning "the dog sees the
man."

The difficulty here is that it means you never attach less than two
clitics to the verb, which seems wasteful.

3b. If a clitic pronoun is attached to the verb, it is promoted to
the highest semantically possible argument role.

Thus, "man dog ANIMAL.see" means "the dog sees the man."

3c.  The flip side of 3b: if a clitic pronoun is attached to the
verb, it is *demoted* to the lowest semantically possible argument
role.  This seems good to me in general because "low-level" semantic
roles tend to be patients, and I tend to see patients as more closely
tied to the verb than agents.  (direct objects being part of a VP and
all that jazz.)

Thus, "man dog HUMAN.see" means "the dog sees the man."

But what if you only want to move something one level up or down in
the heirarchy?  For example, "the father offers the dog a master" with
a verb for "offering someone something" with the roles OFFERER >
RECIPIENT > THING-OFFERED.

That would work with 3c but not 3b:

Father master dog HUMAN.offer

No, wait, which human gets demoted?  Probably the logical rule would
be that the last one in line gets demoted.  Oh, wait: the HUMAN clitic
has an obviative form, so you could just use that.  Still, it seems as
if this kind of problem could crop up.

3c. Another possibility would be a verb affix which basically means,
"ignore animacy heirarchy for the purpose of verb assignment and go
with strict word order instead."  But this seems, I don't know, kind
of artificial and extreme to me.

Any comments or ideas?


+ Ed Heil ---------------------- edheil@postmark.net +
|    "What matter that you understood no word!       |
|    Doubtless I spoke or sang what I had heard      |
|           In broken sentences."  --Yeats           |
+----------------------------------------------------+

Ed Heil wrote:

> Having picked up Campbell's _Concise Compendium of the World's > Languages_ as an anniversary present, and read a bit about Bantu > classes, I decided I want to make a class-based language. > > The language is Ciantwo, and it is spoken by a people who lived in > Atlantean times, in tree-house-towns. They are very adept at > breeding/creating 'designer' plants and animals, and they have a > considerable technology based on that. It doesn't particularly affect > the language except for the fact that certain plants and possibly even > animals will enter the 'human construct' class because of it, but it's > a bit of background. > > This is my first draft of the class system: > (c= /tS/; j= /dZ/ ([which is phonemically distinct from dy=/dj/]; > ds=/ds/=[dz]; y=/j/) > > PRONOUN CLITIC OBVIATIVE MEANING > doen de je PERSON > soen se ce PERSON-PLURAL > dsuam dsu dsyu ANIMAL > suam su syu ANIMAL-PLURAL > wain wa -- PLANT > toes te tye CONSTRUCT > tuas tu tyu CONSTRUCT-PLURAL > goen ge gwe THING > moen me --- THING-PLURAL/STUFF > kau ka kya PLACE > ki ki kwi POINT > > Notes: > CONSTRUCT is prototypically an object which humans have shaped for > their uses. > THING is anything that is not alive and not a CONSTRUCT. > STUFF is rather like English mass nouns: cheese, grain, anything not > thought of as structured and differentiated. > PLACE is a location thought of as a container -- something you can be > in. > POINT is a location thought of as a point -- someplace you can be at. > > How's that sound for a start? > > + Ed Heil ---------------------- edheil@postmark.net + > | "What matter that you understood no word! | > | Doubtless I spoke or sang what I had heard | > | In broken sentences." --Yeats | > +----------------------------------------------------+ >