Print

Print


On Sat, 5 May 2012 14:37:17 -0400, Alex Fink <[log in to unmask]> wrote:

>Back from my travels, now.  A couple of these collated here.  
>
>On Sun, 29 Apr 2012 12:47:21 -0400, neo gu <[log in to unmask]> wrote:
>
>>On Sun, 29 Apr 2012 10:41:02 -0400, Alex Fink <[log in to unmask]> wrote:
>>
>>> And I guess you have no punctuals, grammatically.  
>>
>>No. Maybe I need to change that??
>
>Naw, I don't imagine you need to.  (There aren't really any _instantaneous_ happenings in the real world, it's just an approximation certain languages make.)

ok.

>>>> In the absence of an explicitly inherited index, restricted predicates and secondary predicates inherit both the hosts inherited index and its action interval.
>>>
>>> I wonder about potential problems with multiple inheritance, if you have a restrictive predicate with two indices.  
>>
>>It can't have 2 different restrictive arguments, if that's what you mean. Unless of course you have an example.
>
>Well, if it is a grammatical rule of Apr20 that a word doesn't have two different restrictive arguments, then that's that.  But I could envision them having a use: you know, "those two party guests, _the one who's the father of the other_, have eaten all the hors d'oeuvres."  Although that's not an example of conflicting time inheritance yet.

I haven't been able to translate that, with or without 2 restrictive arguments; care to attempt it for me?

>>>>    John-i UPj-o-eat-Di stale-j pancake-j.
>>>>    "John is eating stale pancakes."
>>>>
>>>> The explicitly inherited index is absolute present, which is included in the interval of "eat". The interval of "stale" is a superrange of that of "eat", which is inherited implicitly via index j. 
>>>
>>> Why a superrange?  As I read what you've written it sounds like the subrange rule (above) should apply here.  Is the difference between assertives and restrictives also relevant here?
>>
>>It _should_ be a superrange, because the pancakes are stale before John starts eating them. Shouldn't it? I'm going to have to think more carefully about the restricted predicate inheritance rule.
>
>Yes, a superrange is what seems to make sense here.  I was trying to ask what the consequences of this for the inheritance rule were.  

I guess I'll have to rewrite the rule, but I haven't done it yet.

>>>> The interval of a depictive secondary predicate is a superrange of that of its host predicate's.
>>>[...]
>>>> The starting time of a resultative secondary predicate is the ending time of the host's action.
>>>
>>> It looks like there's no morphosyntactic way to tell depictives from resultatives, so I guess you mean this to fall to pragmatics.  
>>
>>Maybe I could add an inceptive or inchoative (I forget which is which) to the resultative; otherwise, I'd need to use additional link types.
>
>Would you really?  I was expecting there would be some way to do this using the tense morphology you were describing here, along the lines of my "interpreting the A and B operators too loosely" suggestion below.  Adding a morphological inceptive / inchoative feels unnecessary for Apr20 as a whole, given that for main clauses you could just use a structure based on a predicate  C-begin-Di  or the like.  

I probably will have a "begin" phase predicate, but that's different from entry-to-state.

>>>Hm, I wonder if it'd be natural for this language to have more kinds of secondary predication, corresponding to the other kinds of temporal operators.  If you can have
>>>"She painted the barn (A-)red" = "she painted the barn such that it _ended up_ being red"
>>>then why mightn't you have
>>>"She painted the barn (B-)red" = "she painted the barn such that it _stopped_ being red"?
>>>
>>>Alex
>>
>>I think that might be interpreting the A and B operators too loosely. But the 2nd example might be done using red-NEG:
>>
>>    Dj-b-paint-Di Ethel-i barn-j red-NEG-Dj.
>>    "Ethel painted the barn unred."
>
>Fair enough.  But even if one wouldn't use the A and B operators here, one can still step back and look at the set of kinds of secondary predication available.  You've invoked
>  depictives, for which the time of the secondary predicate _is a superrange_ of the main one; and
>  resultatives, for which the time of the secondary predicate _extends out from the endpoint_ of the main one (?or contains such a range).
>This isn't a time-symmetric set of operations, so I wonder if there mightn't be more kinds.  

There probably are, but I wouldn't know what to call them!

>On Sun, 29 Apr 2012 22:39:47 -0400, neo gu <[log in to unmask]> wrote:
>
>>AND can often be omitted. 
>
>Not always?  

Now that I think about it, it can always be omitted here; the indexes should take care of scope, unless there's some problem with complement clauses.

>Anyway, here again, seeing a new word of an anomalous behaviour with a low yield, my impulse would be to work to eliminate it.  If the structures where AND isn't eliminable are, e.g., ones where it's necessary for disambiguation to scope under some other conjunction, might there instead be some way to do it with a different strategy, like your parenthesisation below?
>
>And/or, maybe AND is identical with an adverb just meaning "indeed" or the like.

That sounds suspiciously natlangy.

>>Note that the copula can parenthesize a phrase; the example
>>
>>    Fido-i Dj-COP-NEG-i big-j orange-j cat-j.
>>    "Fido is not a big orange cat."
>>
>>would otherwise have to be rendered:
>>
>>    Fido-i IOR big-NEG-j IOR orange-NEG-j IOR cat-NEG-j.
>
>Ah, good, that's a(?nother) satisfying reason to retain a copula.  

yes

>>IOR _might_ be used for conditional sentences.
>>
>>    "If Fluffy isn't big, she's orange."
>>    "If Fluffy isn't orange, she's big."
>
>Well, I'm sure you're aware of the material conditional not being a good match for natural-language implication (e.g. https://en.wikipedia.org/wiki/Material_conditional#Philosophical_problems_with_material_conditional ).  I for one don't think I'd be satisfied to do it this way.  

Well, I wouldn't use it for a naturalistic language. More work is needed.

>(In UNLWS, "or" is just rendered as "one of" using the morphology for groups, and when the distinction between SEL and IOR is needed we say "at least one of" vs. "exactly one of".  But we mean to have something else for "if", built on our irrealis mood; that's not designed yet.)
>
>On Wed, 2 May 2012 02:33:08 -0400, neo gu <[log in to unmask]> wrote:
>
>>Apr20 Adverbs ....
>>
>>The following adverbs require proximity to the predicate modified, even though it's not very satisfactory.
>
>I wonder if you could instead do adverbs using a restrictive version of the C link type.  If that were called  c , then one might construct your two examples as
>
>  three-k day-k PUk-have_duration-c b-walk-1P
>  "We walked for 3 days."
>
>  John-i house-j Dj-b-leave-Di angry-k Uk-have_manner-c
>  "John left the house angrily."
>(or just 
>  ... Dj-b-leave-Di angry-c  ?)

I'm not sure what this -c link type does, and better to call it something else (-c is already a temporal index).

>
>Alex

Well, my answers are probably no better than they would have been days ago; I'm still working on the _Learn TIAL_ pages (the grammar's a mess).

--
neogu