Re: replies re. ka & mamta be ma

[jorge@PHYAST.PITT.EDU]

And:
> To allow for knowing noone comes, I suggested:
>
>    da zohu koha djuno le duhu da du lohi klama
>
> If you reject this, I'll disagree with your rejection but I see
> that all your claims about the crucial need for Q-kau follow from
> the alternative q-kau-less reformulations you've offered.

I don't reject it, but it is not trivial that it is equivalent.
The kau-form makes no recourse to sets, which you need to do with
this formulation.

And I could argue that it is not necessarily equivalent. Say that
the following is true:

        ko'a djuno le du'u lo prenu ba klama
        She knows that at least someone will come.

Now, it is not clear to me that I could not claim from that that:

        ko'a djuno le du'u makau ba klama
        For some word(s) X, it is true that she knows X will come.

My replacement formulation would allow it. Your set formulation
would not, because she doesn't necessarily know the full set
of comers.

In the same sense, the question

        ma ba klama

admits {lo prenu} as an answer, which isn't enough to say what is
{lo'i ba klama}.


> > Just as in the case of questions we ask our interlocutor to fill
> > with the appropriate sumti, here we claim that there is a _sumti_
> > that makes the statement true, not a referent of da.
>
> This is surely wrong if you mean sumti in the syntactic sense.
> "Who went" doesn't ask "which words could replace 'who'?".

In Lojban it does. At least that is how questions in Lojban are
explained. I am not saying that that is how "who" works in English,
but in Lojban, all you are asking of your interlocutor is to fill
in the blank.

> If you mean sumti in the semantic sense, I agree. Note that if
> Tim and Tom went, then "she knows who went" entails that she knows
> Tim went and she knows Tom went. But replacing "who" by Tom makes
> the statement true. So your method may need refining.

Are you saying that {la tom} is not a valid answer to {ma klama}?
I think whatever is valid for {ma} should also be valid for {makau}
even though in English this is not the case. Why should we follow
the English model?

> >        ko'a djuno le du'u xokau prenu cu klama
> >        She knows how many people come.
> >        There is a <number> (i.e. a string of PAs) such that
> >        she knows that {<number>prenu cu klama} is true.
>
> I think that works. I don't think that your metalinguistic approach
> captures the way we really understand indirect interrogatives, but
> who cares.

But that is the way even direct interrogatives work in Lojban. I don't
know if it's the way we really understand them or not.

> > In general, "...Q-kau..." can be reworded as "there is an <X>
> > with the same grammar of Q, such that  "...<X>..." is true.
>
> Agreed. (At least for the time being.)

Good. I'm pretty happy with that. Notice that I don't require to know
a full answer (Tim & Tom). That would be practically impossible in the
case of {mokau}.

> > Just as questions say "replace Q with appropriate thing of same
> > grammar so as to make the sentence true", indirect questions say
> > "there is an appropriate replacement for Q-kau that makes the
> > sentence true".
>
> I accept this is one way of describing what happens, but not necessarily
> the optimal way. There are ways of doing semantics of all interrogatives
> without invoking this notion of "appropriate replacement of same grammar".

Could you expand on that? It sounds interesting.

> Also, refer to my Tim/Tom reservation above.

I think that is precisely what fails in your {lo'i} formulation.
Even if for {makau} it could make sense to require full knowledge, for
{mokau} it would be impossible, and the rule should be consistent.

> > > To the miniscule extent that any of this discussion is relevant to
> > > Lojban, my point is that ease-of-use & flexibility justifications
> > > for some grammatical device should be distinguished from other
> > > justifications.
> > Not that I like to disagree, but I think there are no other
> > justifications. "Reflecting cognition" doesn't tell me anything.
> What do you think of the lojbo notion of malglico?

{malglico} is Lojbanic nonsense modeled on something that makes sense
in English. It comes after the language design. It shouldn't affect the
justifications for anything in the design.

> When you look at
> Lojban do you see nothing but a language optimized for ease of use?

No. I see a language based on predicate logic, with many accesories
provided for ease of use. That doesn't mean that I believe there to be
anything primitively cognitive about predicate logic, or that there
is anything that is absolutely essential. {ro} and {su'o} are redundant,
but you would make things extremely difficult if you only had one of
them.

> Or do you discern other criteria motivating its design? If you
> do discern other criteria, do you think them mistaken?

Not especially. There are lots of things that with the benefit of
hindsight could have been done differently e.g. collapse all the
logical connector series (serieses? serii?) into one. But I wouldn't
say it was based on a mistaken criterion.

> > What I'm trying to say is this:
> >        <sumti1> A <sumti2> <selbri>
> > means the same as:
> >        <sumti1> <selbri> IJA <sumti2> <selbri>
> > for any <sumti1>, <sumti2>, and <selbri> and any corresponding A-JA pair.
> > I call that a syntactic rule, because it is independent of meaning.
>
> It is a syntactic rule, but a wholly unnecessary one.

And yet so much easier to teach to the parser than a purely semantic rule.

> If this is the case, I would still dissent from:
> > > > Not always. {mi e do klama} is truth-conditionally equivalent to
> > > > {mi klama ije do klama}, but this is independent of the semantics.
> > > > They are sintactically truth-conditionally equivalent.
>
> since I prefer to think of duhu and not seduhu as having truth
> conditions.

So do I, but it is a natural step to say that two sedu'u are t-c.e. when
their du'u are. I am not saying that they have a truth value, I'm just
defining an equivalence class for them.

Jorge