On 1 February 2015 at 21:59, J S Jones <[log in to unmask]> wrote:
> On Sun, 1 Feb 2015 21:15:37 -0700, Logan Kearsley <[log in to unmask]> wrote:
>>On 1 February 2015 at 16:08, J S Jones <[log in to unmask]> wrote:
>>> In Jan24, I have bivalent predicates instead of verbs, adjectives, nouns, and prepositions. All of these can take a person-number prefix (for the more agent-like argument) and a person-number suffix (for the more patient-like argument). However, these can be 0. When an affix is 0, there's a default value depending on the predicate stem's class. The values are 3AnimS (A), 3InanS (I), and Null (U). There are 6 classes WRT default person-number:
>>> A--I: most transitive situations, body parts;
>>> A--U: agentive situations, animate entities;
>>> U--I: scalar and other patientive situations, inanimate entities;
>>> U--U: impersonal situations;
>>> A--A: words like "give" & "tell", kinship, some derived stems, such as "school of fish";
>>> I--I: some derived stems, such as "grove of trees" and "peel of fruit";
>>> The stem also determines other things, such as the argument roles.
>>> What I'm trying to come up with are a name and single-letter abbreviation for each class. Let me know what's not clear.
>>What other parts of speech are there? How do you form complete
>>clauses? Can there be more than one of these stems per clause, and if
>>so, how are they combined?
>>Since there tends to be correlation between semantic classes (as
>>you've laid out here) and syntactic classes, I'm thinking it might
>>make sense to derive at least some of the class names from the
>>syntactic roles that they most typically occupy, if there is in fact
>>any way to distinguish them.
> The other word classes are pronouns, quantity words, determiners, conjunctions, case prepositions, and other particles. The first 2 of these can be inflected (pronouns are like predicates but are univalent). A predicate or pronoun form can constitute a clause by itself; otherwise, the first acts as the head of the clause, with the others organized into phrases (the case prepositions and determiners help there). Non-head predicates and pronouns must be marked as having a 3rd person argument. I'll have to come up with some examples later.

Hm. This seems very alien to my intuition of what "pronoun" means.

On 3 February 2015 at 06:57, J S Jones <[log in to unmask]> wrote:
> Unfortunately, that scheme no longer works -- there are 7 classes now:
> U--U impersonal: rain
> U--A animate: water(1), cat, happy, school of fish
> U--I inanimate: water(2), cup, hot, break, flow, orange grove
> A--U agentive: die, run
> A--A ???: mother, friend, give, flock
> A--I transitive: hand, inhabit, kick, see, in
> I--I ???: fruit, grove
> I'll probably just use numbers or the letter pairs for the abbreviations, but I still need some names.
> The kinship system may turn out to be interesting: I need only one word for each relationship, not two, e.g. "child/parent" or "younger/older sibling".

It seems to me you've done a pretty good job of naming the classes
right here already.

UU: Impersonal class - I class
UA: Animate class - A class
UI: Inanimate class - N class (I was already taken)
AU: Agentive class - G class (A was taken)
AI: Transitive class - T class

That just leaves AA and II, which you've marked with question marks.
AA seems like it's for relations (kinship terms, group membership,
with "give" as a less prototypical member), so how about Relational /
For II, I don't see, based on your example words, what significant
difference there is between that and UI. So, maybe just call them both
Intransitive / N-class, distinguished by a sub-type? E.g., the UI
class could be class N1 or class Na, while the II class could be class
N2 or class Nb.