Re: Some Thoughts on Word Classes
Saul Epstein 
Thu, 20 Aug 1998 20:06:48 -0500

Quotes from: Rob Zook <rzook*informix,com>
Date: Thursday, August 20, 1998 5:03 PM
 
> Ok, that sound like a pretty precise version of the nebulous
> thoughts I was trying to describe regarding how the entities
> and action/states, so naturally it looks fine to me.

Magnificent!

> However, I would like you to explain how you see 4. above
> rolling into 1b. States? Should there be a affix to
> distinguish between state, action and quantity?

Quality. Quantity, being a different kind of specification -- with an
explicit morphology, no less -- should rightly stay separate.

As for quality merging with state... Whl'prala has two words which,
unmodified, represent entities: <sbah>, meaning "red" (or "redness");
and <plak>, meaning "blood." But, also unmodified, these words can
represent a quality or qualities posessed by the corresponding
entities.

     <sbah plak>     "red blood"
     <plak sbah>     "blood red"

The first term refers to an entity (blood) which is characterized by
the sole quality of an abstract entity (redness). The second term
refers to an abstract entity (redness) characterized by one quality
of a concrete entity (blood). (Of course, Vulcans are unlikely to
think of blood as red in general. But they might have occasion to
describe something as <qomiplak sbah>, or "Terranblood red.")

So, entities are characterized by qualities, and a state seems to be
the characterization of one entity in terms of another entity's
qualities. And since using a word to modify another in this way
doesn't currently involve any marking (as with affixes, etc.) there
doesn't seem to be a need to put words so used in a separate class.
Although... It just occured to me that we could go the other way,
lumping qualities and states together as 4, and making 1 consist only
of actions. That would eliminate a need to come up with a label
grouping actions with states...

> Also, I do agree that in Formal Mode 6a and 6b will
> likely be exactly the same thing, but I suspect Informal
> Mode will not require that much precision.

Probably not. I just don't know whether that will mean that Informal
has two sets or that it will use one set with less precision.
 
> I'm not sure however, that I see the difficulty in describing
> a relation as distinct from an entity.

Relation is not a satisfactory label, I'm afraid. Having thought of
this, I find the prospect of separating actions from states as
"major" classes very appealing. In any case, everything you say here
pretty well illustrates the difficulty of description in terms other
than those of entity...

> One thing I think we should codify in the grammar rules of
> Whl'q'n, that we should class entities by their "level" of
> abstraction. Simply, does said point to an extensional
> entity? Measurable, or experienceable by you and others?
> This is sort of releated to the idea of evidential affixes,
> and this might be a simply way to codify what I mean. Perhaps
> by adding the affixes, below one could express this
> extensional vs intensional nature of the entity one speaks
> about.

     [snip]

These are worth-while distinctions, especially for the Formal mode.

> So, if I understand your proposed grammar additions,
> th'dhahqte would be roughly, "my song" as an entity, as
> opposed to the action, so th'dhahqte'aq would mean roughly,
> "my song (that you're experiencing right now)".

Mmm, I'm not sure. <th'dhahqte> would represent... ah... the class of
lyric compositions or performances possessed by me. So, my "complete
works," also my skill or style, perhaps, as a lyricist or singer.
This is an entity. (Not to be confused with <th'dhahqe>, the
more-than-ten plural of <th'dhahq>, a particular composition or
performance.)

> But how would you specifically distinguish "I sing" from "my song",
just
> th'dhahqte?

Not exactly. <th'dhahqte>, being a class, can't be interpreted as an
action, that's true. <th'dhahq> can be interpreted as an action ("I
sing (right now).") or an entity ("my (specific) song"). And I think,
in a sense, neither interpretation is dominant, even in a sentence
like

     en'[use] whl'pralahi         th'dhahq
     it-use   Vulcan-language-ACC I-song
     "My song is in Vulcan." OR "I'm singing in Vulcan."

but that the Formal mode would include ways of forcing one
interpretation. (In addition, including tense and aspect information
would further disambiguate the action-or-entity role of a word...)

> If Marketa buys the grammar additions it would make the
> lexicon a little easier to maintain, we could simply supply
> a list of root words, and one can form the verb-like,
> gerund-like and other forms with affixes.

I thought so, too. Such a lexicon would also be easier to use and, I
think, better reflect the given morphology.

So here's another possible set of classes...

    1. Action
    2. Entity
    3. Deixis
    4. Quality (or State)
    5. Quantity
    6. Operation

--
from Saul Epstein
locus*planetkc,com - www,planetkc,com/locus
"Surakri' ow'phahcur the's'hi the's'cha'; the's'phahrka
     the's'hi surakecha'."
                                        -- K'dvarin Urswhl'at