Re: Language Sketch: Gogido
|From:||Logan Kearsley <chronosurfer@...>|
|Date:||Monday, August 25, 2008, 23:44|
Thanks. I shall.
>> Also, I wanted lots of distinct vowels to use as word-ending
>> part-of-speech markers (because final consonants are disallowed), but
>> that system is sort of up in the air at the moment. Primarily, I'm
>> wondering whether I should have the word endings be part of the root
>> morphemes which just happen to be standardized, or whether they should
>> be separable type-marking affixes which can be replaced during
> You might take a look at the way I did this in my engelang säb zjed'a ,
> Root words are marked for the semantic category and *default* part
> of speech by the final consonant; an optional final vowel marks change
> of POS, (for nouns) case and (for verbs) transitivity. There's a systematic
> way for roots of each semantic class to derive words of each part of
>>> What do you do with experiencer-verb-focus or agent-verb-focus or
>>> force-verb-patient sentences, etc.? Is there any way for those
>>> arguments to have the preposition elided if they're in a default position relative to
>>> the verb, or do
>>> they always have to have a preposition marking them?
>> There's no explicit verb focusing. Topicality / definiteness /
>> emphasis can be indicated by varying the phrase order. I toyed with
> In this instance by "focus" I meant the typically object argument of sensory or
> thinking verbs whose subject is usually an experiencer, e.g. in
> "She sees him", or "She thinks about him", "she" would be experiencer
> and "him" the focus, according to theta role terminology as I understand it
> -- though "focus" is used to mean other things in other linguistic contexts.
Ah! In that case, experiencer maps to agent and focus maps to patient.
So, yes, they could be left unmarked.
>> the idea of having a default order for a larger number of theta-roles
>> that would allow for eliding prepositions, but I decided that made it
>> too complex. So, in the current system, only agent and patient can
>> have prepositions elided, and everything else *must* be marked with a
>> preposition. Parsing goes like this:
>> 1. Check if there are any un-determined (unmarked) arguments. If so,
>> continue. If not, you're done.
>> 2. Check to see if there's already a determined agent. If so, then the
>> unmarked argument is determined to be the patient. If not, then the
>> first unmarked argument is determined to be the patient. Go back to
>> step one.
> Is there a typo in paragraph #2, with "patient" substituted for "agent"
> one of the two times it appears (perhaps the second)? Because otherwise
> I don't get it.
Oops, yes. The second occurrence should be "agent"- if there isn't a
determined agent yet, the first unmarked argument becomes the agent.
> If I understand correctly, you could have any of VAP, AVP, or APV
> without needing to mark the agent and patient explicitly, since
> the nouns and verbs and modifiers are morphologically distinct
> (right?). Only if the patient precedes the agent, or if you have
> an experiencer instead of an agent or a focus instead of a patient
> or something would you need to have prepositions to mark them.
Right. An alternative system (which results in exactly equivalent
interpretations for my entire corpus so far, since I haven't taken
advantage of the weirder ordering possibilities afforded by this
system) is that anything unmarked before the verb is an agent, and
anything unmarked after the verb is a patient.
Hm. That's probably a better system, actually, since it allows you to
elide any component without changing the rest of the sentence. Well,
I'll decide for sure when I get to writing something where it actually
makes a difference.
> Given your parsing rule, I have to ask if you would explicitly mark
> the patient when there is no agent but a force instead, e.g.
> "The wind opened the door". It seems you would need to.
> force-PREP wind-NOUN open-VERB patient-PREP door-NOUN
> because otherwise "door" would be the first unmarked argument
> and would be parsed as agent.
Yes. As things currently stand, that would be the case.
> Do you mark objects of result distinctly from preexisting patients?
> I.e., in "Hikaru wrote a letter" would "letter" be unmarked if in default
> position or would it need a preposition to mark it no matter its
"Letter" could be left unmarked, but doesn't necessarily have to be.
Generally, I'm trying to get by with as few different required
distinctions and categories as possible, but also allow for maximal
expressiveness if you want it. So, if you leave "letter" unmarked, it
means exactly what you want it to mean, but if you want to emphasize
that the letter was created by the action, you could mark for that;
or, if you wanted to specify that the letter actually did exist
beforehand, you could mark for that, too- in which case the meaning
would change to something like "Hikaru edited a letter" or"Hikaru
amended a letter". And if you want to distinguish between those, then
there're alternate words and more levels of grammatical complexity
that you can layer on top.
>>> What about valency, evidentiality and mood?
>> Valency is unmarked. You can apply any number or type of arguments to
>> any verb, although what that *means* may be rather opaque in many
> You could have a default causative derivation for any normally
> intransitive verb which has no transitive counterpart defined
> in the lexicon, I guess; something like "agent or force causes patient
> to undergo the process denoted by the verb", or "experiencer
> undergoes the process denoted by the verb because of perceiving
> or thinking about the focus".
That's a good idea. Saves a few derivational affixes.
I'll have to think about how far the principle can extended.