xoi and new soi as bridi relative clause

[guskant]

The current highest score of {xoi} (or {fi'oi}) 

http://jbovlaste.lojban.org/dict/xoi

and new {soi} 

http://mw.lojban.org/papri/new_soi

as bridi relative clause are defined with assertion about their (semantic) scope.

xoi:

Right-scoping adverbial clause: encloses a bridi and turns it into an adverbial term; the antecedent (ke'a) of the enclosed bridi stands for the outer bridi {lo su'u no'a ku} (the bridi in which this xoi term appears), including all the other adverbial terms (tags...) within this bridi located on the right of this xoi term (rightward scope).

soi:

soi terms always have top-scope, meaning they scope over any quantifiers, negations or tenses that precede them. When multiple soi are used in the same bridi, they are understood to have equal scope.

{xoi} and new {soi} are currently in selma'o SOI of la ilmentufa, and SOI_clause is treated as tag_term:

http://mw.lojban.org/extensions/ilmentufa/camxes-exp.js.peg

tag_term = expr:((!gek tag free* (sumti / KU_elidible free*) / NA_clause free* KU_clause free* / !gek !ek !joik_jek !gihek NA_clause free*  KU_elidible free* / SOI_clause free* subsentence SEhU_elidible free*) (joik_jek tag_term)*) {return _node("tag_term", expr);}

Generally, Lojban terms have scope that spans over all the following terms in a sentence. The scope of terms is modified by a prenex of a statement. If {xoi}-clause and new {soi}-clause are defined as having scope, it should not be a term. If they are defined as terms, {xoi} has harmony with scope of other terms, but {soi} becomes an exception of scope of terms, otherwise loses its power as left-scope. 

Which do you prefer:

1. keep the current situiation, and regard {soi} as an exception of scope of terms;

2. keep the current situiation, and let {soi} loose the power of left-scope ({soi} becomes the same as {xoi});

3. create a new grammatical property for {soi}, and keep {xoi} as it is;

4. allow {soi}-clause become a "free", that has the same property as {sei}-clause (easier than 3.);

5. other.

re'i

[Gleki Arxokuna]

2015-07-20 17:09 GMT+03:00 guskant <gusni...@gmail.com>:

1. keep the current situiation, and regard {soi} as an exception of scope of terms;

2. keep the current situiation, and let {soi} loose the power of left-scope ({soi} becomes the same as {xoi});

3. create a new grammatical property for {soi}, and keep {xoi} as it is;

4. allow {soi}-clause become a "free", that has the same property as {sei}-clause (easier than 3.);

5. other.

I choose 4. (soi is exactly the same in grammar as sei) or 5. (retain the original meaning from CLL formalizing it first)

For other meanings intended in 1. and 2. I can only propose extending existing grammar like what's done in "altatufa parser, stable version":

Namely, {i broda noi mo} under this app parses as

(NIhO [I {FA ZOhE} {CU <FASNU (¹FA [LO {NU <(²FA mi²) (²CU [broda VAU]²)> KEI} KU] [noi {<FA ZOhE> <CU (²mo VAU²)>} KUhO]¹) VAU>}]) 

Can be tested at http://mw.lojban.org/extensions/ilmentufa/altatufa-stodi.html

Note that focus is not preserved in such parsing.

This also has an advantage of having {poi} / {noi} distinction for bridi relative clauses instead of proliferating cmavo over and over again making the language unnecessarily complex.

[selpa'i]

la .guskant. cu cusku di'e

> Which do you prefer:
>
> 1. keep the current situiation, and regard {soi} as an exception of
> scope of terms;
> 2. keep the current situiation, and let {soi} loose the power of
> left-scope ({soi} becomes the same as {xoi});
> 3. create a new grammatical property for {soi}, and keep {xoi} as it is;
> 4. allow {soi}-clause become a "free", that has the same property as
> {sei}-clause (easier than 3.);
> 5. other.

New-{soi} originally started out as a free modifier (before xoi became
popular), but then I thought that being a term is more useful because it
allows us to use it in termsets:

(A) mi ge lo xamsi soi cafne gi lo cmana soi to'e cafne cu klama
"I often go to the ocean, but rarely to the mountain."

Such a construction used to be impossible before. However, now that
{xoi} exists, it could easily replace {soi} in that sentence, and {soi}
*could* go back to being a free modifier (option 4).

I think this would depend on how much support {xoi} has (it has
experimental cmavo form which might bother some people?). If {xoi} is
going to become part of standard Lojban then I think option 4 works
(sentence (A) needs to remain expressable somehow).

I'd like to hear from other users of both {soi} and {xoi}.

mi'e la selpa'i mu'o

[guskant]

je'e 

In order to keep (A) in which {soi} is replaced by {xoi} be valid, new selma'o XOI should be substituted for the new SOI so that {xoi}-clause be a term.

new SOI will become almost the same as SEI. The difference is only the bridi-tail (and pseudo-{zo'u} if {soi}-clause encloses a "subsentence").

Considering the simplicity of grammar and the advantage of new SOI compared with SEI, it would be better to modify SEI so that it encloses a sentence, and merge {soi} to selma'o SEI, though this change will require more {se'u}. 

Even if {sei} and {soi} are in the same selma'o, they can be semantically different: {sei} will have the same scope as UI, while {soi} will have the broadest scope over a sentence, and may take the "signified" of the sentence with {ke'a} in the clause.

By the way, on the "new soi" page, "subsentence" is suggested in {soi}-clause. Do you intend to use {zo'u} in {soi}-clause, or it simply inherited the official grammar of NU/NOI? {zo'u} in NOI- or new SOI/XOI- clauses may produce logical problem, and I want to avoid it if possible. (la zantufa-0.2 allowed "statement" including {zo'u} in NOI-clause, but it will be changed to "sentence" in the future version, and then {zo'u} in NOI-clause will be banned.)

[Gleki Arxokuna]

2015-07-21 9:44 GMT+03:00 guskant <gusni...@gmail.com>:

Considering the simplicity of grammar and the advantage of new SOI compared with SEI, it would be better to modify SEI so that it encloses a sentence, and merge {soi} to selma'o SEI, though this change will require more {se'u}. 

This would be absolutely devastating. {sei} is very similar to LE and that's the simplicity of the grammar.

I don't see any need in changing it, what is more the current grammar of SEI is more advantageous to me.

[selpa'i]

di'e voi cusku ki fa la .guskant. ( :P )

> In order to keep (A) in which {soi} is replaced by {xoi} be valid, new
> selma'o XOI should be substituted for the new SOI so that {xoi}-clause
> be a term.
>
> new SOI will become almost the same as SEI. The difference is only the
> bridi-tail (and pseudo-{zo'u} if {soi}-clause encloses a "subsentence").
>
> Considering the simplicity of grammar and the advantage of new SOI
> compared with SEI, it would be better to modify SEI so that it encloses
> a sentence, and merge {soi} to selma'o SEI, though this change will
> require more {se'u}.
>
> Even if {sei} and {soi} are in the same selma'o, they can be
> semantically different: {sei} will have the same scope as UI, while
> {soi} will have the broadest scope over a sentence, and may take the
> "signified" of the sentence with {ke'a} in the clause.

Let's not keep mixing topics just yet. We can make {soi} a free modifier
before worrying about doing anything to {sei}.

> By the way, on the "new soi" page, "subsentence" is suggested in
> {soi}-clause. Do you intend to use {zo'u} in {soi}-clause, or it simply
> inherited the official grammar of NU/NOI? {zo'u} in NOI- or new SOI/XOI-
> clauses may produce logical problem, and I want to avoid it if possible.
> (la zantufa-0.2 allowed "statement" including {zo'u} in NOI-clause, but
> it will be changed to "sentence" in the future version, and then {zo'u}
> in NOI-clause will be banned.)

What are the logical problems when allowing a prenex in NU/NOI/SOI? The
prenex has scope over the NU/NOI/SOI, which in turn has scope over the
main bridi.

(B) lo prenu poi ro da zo'u ke'a djica lo nu ke'a viska da
"people that are such that for all X, they want to see X"


(C) ra troci lo ka ro da zo'u lo nu da viska ce'u cu rinka lo nu da
cisma
"She attempts that for all X, X seeing her causes X to smile."


(D) ma'a ca ro xavdei lo ka vokta'a cu simxu, soi ku'i na ku ro da
poi jbopre zo'u lo nu da pagzu'e ke'a cu dikni
"On every Saturday we have vocal chats, which however is such
that not every Lojbanist is such that their taking part in them occurs
regularly."


Which, if any, of these are problematic to you? (Sorry for the long
example (D), it was the first thing that came to mind)

[selpa'i]

mi cusku di'e

> What are the logical problems when allowing a prenex in NU/NOI/SOI? The
> prenex has scope over the NU/NOI/SOI, which in turn has scope over the
> main bridi.

Clarification: I meant SOI has scope over the main bridi, not NU and
NOI, of course.

[guskant]

Le mardi 21 juillet 2015 07:45:12 UTC, la gleki a écrit :

2015-07-21 9:44 GMT+03:00 guskant <gusni...@gmail.com>:

Considering the simplicity of grammar and the advantage of new SOI compared with SEI, it would be better to modify SEI so that it encloses a sentence, and merge {soi} to selma'o SEI, though this change will require more {se'u}. 

This would be absolutely devastating. {sei} is very similar to LE and that's the simplicity of the grammar.

I don't see any need in changing it, what is more the current grammar of SEI is more advantageous to me.

It's not simplicity of grammar but similarity of form of description sumti and inserted bridi. If {sei}-clause could enclose a sentence, the grammar would be 23 bytes smaller. If we add {soi}-clause to "free", the structure and the usage will be very similar to {sei}-clause. If we had the structure of "SOI sentence SEhU" from the beginning of Lojban, I would have never used "SEI terms selbri SEhU". 

Le mardi 21 juillet 2015 09:30:41 UTC, selpa'i a écrit :

di'e voi cusku ki fa la .guskant. ( :P )
> In order to keep (A) in which {soi} is replaced by {xoi} be valid, new
> selma'o XOI should be substituted for the new SOI so that {xoi}-clause
> be a term.
>
> new SOI will become almost the same as SEI. The difference is only the
> bridi-tail (and pseudo-{zo'u} if {soi}-clause encloses a "subsentence").
>
> Considering the simplicity of grammar and the advantage of new SOI
> compared with SEI, it would be better to modify SEI so that it encloses
> a sentence, and merge {soi} to selma'o SEI, though this change will
> require more {se'u}.
>
> Even if {sei} and {soi} are in the same selma'o, they can be
> semantically different: {sei} will have the same scope as UI, while
> {soi} will have the broadest scope over a sentence, and may take the
> "signified" of the sentence with {ke'a} in the clause.

Let's not keep mixing topics just yet. We can make {soi} a free modifier
before worrying about doing anything to {sei}.

 

OK, I should stop by pointing out the similarity of SEI and SOI, I should not talk about my ideal. I'm very sorry for that.

> By the way, on the "new soi" page, "subsentence" is suggested in
> {soi}-clause. Do you intend to use {zo'u} in {soi}-clause, or it simply
> inherited the official grammar of NU/NOI? {zo'u} in NOI- or new SOI/XOI-
> clauses may produce logical problem, and I want to avoid it if possible.
> (la zantufa-0.2 allowed "statement" including {zo'u} in NOI-clause, but
> it will be changed to "sentence" in the future version, and then {zo'u}
> in NOI-clause will be banned.)

What are the logical problems when allowing a prenex in NU/NOI/SOI? The
prenex has scope over the NU/NOI/SOI, which in turn has scope over the
main bridi.

I said only NOI and SOI/XOI, not NU. From a logical point of view, a prenex is unnecessary or rather problematic in NOI-clause, while NU clause must be able to enclose a prenex with full logical connectives.

 

    (B) lo prenu poi ro da zo'u ke'a djica lo nu ke'a viska da
        "people that are such that for all X, they want to see X"

zo'u in this fragment is logically meaningless because of lack of main bridi. If there were main bridi, the prenex could be put out:

roda zo'u ko'a prenu ije ko'a djica lo nu ko'a viska da

And then it becomes logically analyzable.

Prenex in noi-clause is only a pseudo-prenex that is logically meaningless.

 

    (C) ra troci lo ka ro da zo'u lo nu da viska ce'u cu rinka lo nu da
cisma
        "She attempts that for all X, X seeing her causes X to smile."

I said {zo'u} in NU-clause is necessary. No problem here.

 

    (D) ma'a ca ro xavdei lo ka vokta'a cu simxu, soi ku'i na ku ro da
poi jbopre zo'u lo nu da pagzu'e ke'a cu dikni
        "On every Saturday we have vocal chats, which however is such
that not every Lojbanist is such that their taking part in them occurs
regularly."

No problem here. My main problem was this: 

what if some xoi-clauses and soi-clauses in a sentence have each prenex? which prenex will be regarded as outmost? 

However, considering (D), I understood the logical property of xoi/soi-clause.

They are statements independent of the main bridi. Logically, {soi}, {xoi} and {se'u} plays the same role as {to} {toi}. Then a full statement should be allowed to xoi/soi-clauses.

 

[Gleki Arxokuna]

2015-07-21 17:18 GMT+03:00 guskant <gusni...@gmail.com>:

It's not simplicity of grammar but similarity of form of description sumti and inserted bridi. If {sei}-clause could enclose a sentence, the grammar would be 23 bytes smaller. If we add {soi}-clause to "free", the structure and the usage will be very similar to {sei}-clause. If we had the structure of "SOI sentence SEhU" from the beginning of Lojban, I would have never used "SEI terms selbri SEhU". 

Similarly, you can eliminate LE and move it to LOhOI saving even more rules.

[guskant]

Thank you for the advice. It seems apparently a nice idea, but I see more importance in LE-clause than SEI-clause, so I would rather keep it: A LE-clause automatically picks up the first argument from predication, while LOhOI can change the argument to pick up by {ke'a} afterwords. They have more different aspect compared with the similarity between SEI and new SOI.

By the way, I have no intention to force any change to ilmentufa or camxes, so don't be so much worry. You can continue being illogical and inconsistent as you like. 

[selpa'i]

la .guskant. cu cusku di'e

> Le mardi 21 juillet 2015 09:30:41 UTC, selpa'i a écrit :
>
> di'e voi cusku ki fa la .guskant. ( :P )

> > Even if {sei} and {soi} are in the same selma'o, they can be
> > semantically different: {sei} will have the same scope as UI, while
> > {soi} will have the broadest scope over a sentence, and may take the
> > "signified" of the sentence with {ke'a} in the clause.
>
> Let's not keep mixing topics just yet. We can make {soi} a free
> modifier
> before worrying about doing anything to {sei}.
>
> OK, I should stop by pointing out the similarity of SEI and SOI, I
> should not talk about my ideal. I'm very sorry for that.

I was trying to keep the thread on topic, because threads often go on
too many tangents, which delays getting the main topic settled. Of
course you should talk about your ideals. It's just that SEI and SOI,
similar or not, can be treated individually.

> (B) lo prenu poi ro da zo'u ke'a djica lo nu ke'a viska da
> "people that are such that for all X, they want to see X"
>
> zo'u in this fragment is logically meaningless because of lack of main
> bridi. If there were main bridi, the prenex could be put out:
>
> roda zo'u ko'a prenu ije ko'a djica lo nu ko'a viska da
>
> And then it becomes logically analyzable.
> Prenex in noi-clause is only a pseudo-prenex that is logically meaningless.

Why should the relative clause care about the main bridi? The relative
clause in (B) is like a predicate that attaches to {prenu} with {je}.
{poi ro da zo'u ...} could be rewritten {poi ckaji lo ka ro da zo'u
...}, and (B) could be rewritten as {lo prenu je ckaji be lo ka ...}.

At what point do you think does it stop being equivalent?

> (C) ra troci lo ka ro da zo'u lo nu da viska ce'u cu rinka lo
> nu da
> cisma
> "She attempts that for all X, X seeing her causes X to smile."
>
> I said {zo'u} in NU-clause is necessary. No problem here.

Good.

> (D) ma'a ca ro xavdei lo ka vokta'a cu simxu, soi ku'i na ku ro da
> poi jbopre zo'u lo nu da pagzu'e ke'a cu dikni
> "On every Saturday we have vocal chats, which however is such
> that not every Lojbanist is such that their taking part in them occurs
> regularly."
>
> No problem here. My main problem was this:
> what if some xoi-clauses and soi-clauses in a sentence have each prenex?
> which prenex will be regarded as outmost?
>
> However, considering (D), I understood the logical property of
> xoi/soi-clause.
> They are statements independent of the main bridi. Logically, {soi},
> {xoi} and {se'u} plays the same role as {to} {toi}.

This may be true for {soi}, but I'm not at all sure it's true for {xoi}.
There are two options for {xoi}: it's either restrictive or
non-restrictive. If it is one of the two, then we don't have a word for
the other and vice versa. There should really be two {xoi}. Let's call
them {Pxoi} and {Nxoi}. There is an important difference between (E) and
(F):

(E) so'i verba cu krixa Pxoi fanza
"Many children are yelling annoyingly."
(There may be children there whose yelling isn't annoying)

(F) so'i verba cu krixa Nxoi fanza
"Many children are yelling, which is annoying."
(Every yelling child is annoying)

Which one is {xoi} supposed to be?

[Spheniscine (la zipcpi)]

Pxoi and Nxoi sounds a lot like {po'oi} and {no'oi}, except their grammar is different. NOhOI is even implemented in camxes.

[selpa'i]

la zipcpi cu cusku di'e

> Pxoi and Nxoi sounds a lot like {po'oi} and {no'oi}, except their
> grammar is different. NOhOI is even implemented in camxes.

NOhOI attaches to the selbri and restricts or comments on it. It is not
really related to {xoi} in any relevant way.

[la durka]

To be honest, I don't see a big problem with #1. We don't usually claim that everything in the same selma'o has the same scoping implications -- think about KOhA. And we can't even claim it for terms -- {ca da} is a scoping term while {lo broda} is not. (Please correct me if I am abusing the term "term".)

#2 seems awkward because {soi} would then be redundant, so what's the point. #3 seems awkward because I'm not sure how this new grammatical category would look, but it seems like it'd be quite similar to a term, so it would "infect" a lot of grammar rules.

#4 would work too, I think.

- mu'o mi'e durkavore

[guskant]

Le mardi 21 juillet 2015 15:12:47 UTC, selpa'i a écrit :

la .guskant. cu cusku di'e

>          (B) lo prenu poi ro da zo'u ke'a djica lo nu ke'a viska da
>              "people that are such that for all X, they want to see X"
>
> zo'u in this fragment is logically meaningless because of lack of main
> bridi. If there were main bridi, the prenex could be put out:
>
> roda zo'u ko'a prenu ije ko'a djica lo nu ko'a viska da
>
> And then it becomes logically analyzable.
> Prenex in noi-clause is only a pseudo-prenex that is logically meaningless.

Why should the relative clause care about the main bridi? The relative
clause in (B) is like a predicate that attaches to {prenu} with {je}.
{poi ro da zo'u ...} could be rewritten {poi ckaji lo ka ro da zo'u
...}, and (B) could be rewritten as {lo prenu je ckaji be lo ka ...}.

At what point do you think does it stop being equivalent?

Relative clauses should care about the main bridi, because the bridi in relative clauses share their universe of discourse with the main bridi. About this example of fragment, I said only "logically meaningless", but it does not bring any problem by itself. 

The problem occurs in the case that plural relative clauses appears with their own prenex, sometimes nested in logical connectives. Which prenex is the outmost? How can the negations, numbers and logical connectives are transformed into a prenex normal form?

If (E), xoi-clause encloses a sentence, not a statement. If (F) can be expressed by soi-clause, then (E) for xoi is useful, though I don't know which the creator of the word think of.

 

[Jorge Llambías]

On Wed, Jul 22, 2015 at 12:12 AM, guskant <gusni...@gmail.com> wrote:

Relative clauses should care about the main bridi, because the bridi in relative clauses share their universe of discourse with the main bridi. About this example of fragment, I said only "logically meaningless", but it does not bring any problem by itself. 

The problem occurs in the case that plural relative clauses appears with their own prenex, sometimes nested in logical connectives. Which prenex is the outmost? How can the negations, numbers and logical connectives are transformed into a prenex normal form?

I don't understand what you mean. Could you give an example where you find a prenex in a relative clause problematic? In general, a prenex in a non-restrictive relative clause cannot be exported to the main bridi prenex:

   ro da zo'u da prami la alis noi no de zo'u ke'a prami de

   = ro da zo'u da prami la alis .i ta'o no de zo'u la alis prami de

   =/= ro da no de zo'u da prami la alis .i je la alis prami de

mu'o mi'e xorxes

[guskant]

non-restrictive relative clause is not related to the problem.

How do we transform {poi}s with {zo'u} in prenex into prenex normal form? 

ro da 

poi su'o de 

poi su'o di poi rirni ke'a zo'u di cliva ke'a 

zi'e poi ro daxire poi ctuca ke'a zo'u da .onai de tavla ke'a daxire

zo'u de .a di xendo ke'a 

zo'u ko'a poi su'o daxici zo'u daxici tavla de ke'a cu penmi di 

[guskant]

2015-07-23 0:51 GMT+00:00 Jorge Llambías <jjlla...@gmail.com>:

>
>
> On Wed, Jul 22, 2015 at 8:28 PM, guskant <gusni...@gmail.com> wrote:
>>
>>
>> non-restrictive relative clause is not related to the problem.
>
>

> They have the same grammar as restrictive clauses, so if you disallow the
> prenex for one you disallow it for the other as well, unless you separate
> their grammars.
>  

I agree.

>>
>> How do we transform {poi}s with {zo'u} in prenex into prenex normal form? 
>> ro da 
>> poi su'o de 
>> poi su'o di poi rirni ke'a zo'u di cliva ke'a 
>> zi'e poi ro daxire poi ctuca ke'a zo'u da .onai de tavla ke'a daxire
>> zo'u de .a di xendo ke'a 
>> zo'u ko'a poi su'o daxici zo'u daxici tavla de ke'a cu penmi di 
>
>

> Not sure what that is supposed to mean, it's very confusing. The same
> variable names are being used inside and outside the scope of their binders,
> which means they have to be bound with new implicit "su'o". Also, is "su'o
> di poi rirni ke'a" really meant to mean just "su'o se rirni", for example?
> Embedded relative clauses can get confusing very fast, so it's not a good
> idea to use "ke'a" when you could use "da", "de", "di". "ke'a" will always
> correspond to the tightest "poi".
>
> Could you make a simpler example where your problem shows up, preferrably
> with the fewest possible variables?

>
> mu'o mi'e xorxes
>

ex.1)

ro da 

poi su'o de 

poi ro di zo'u di broda de da

zo'u de brode da di 

zo'u da brodi de di

Do we have a rule to unify {zo'u}s nested in {poi}? Maybe the preceding ones are outer, but I have never seen the rule written.

ex.2)

ro da 

poi su'o de zo'u de broda da

zi'e poi ro di zo'u di brode da

zo'u da brodi de di

Do we have a rule to unify {zo'u}s in {poi} connected with {zi'e}? Maybe we can define the preceding ones are outer. If so, ex.2 seems apparently the same with

ex.2-1)

ro da 

poi ro di zo'u di brode da

zi'e poi su'o de zo'u de broda da

zo'u da brodi de di

but actually they will have different meaning.

ex.3)

su'o de zo'u ko'a poi ro da zo'u da broda ke'a cu brode de

When {zo'u} in {poi} is covered with a constant {ko'a}, is it considered outer than {su'o de} or inner?

[Gleki Arxokuna]

So simply put we have:

1. bridi = terms? CU_elidible selbri terms?

2. sei-clause = terms? CU_elidible selbri

3. sumti-tail = selbri

Of course, we ignore side constructs like quantifiers and relative clauses.

So are you asserting that it's better to align 1. and 2. thus losing backward compatibility with previous grammar of 2.?

Are you asserting that it's the best way to minify the grammar?

If you answer yes to both questions then I disagree since it's similarly possible to align 2. and 3. and then you wont lost any compatibility.

For you I aligned 2. and 3. in altatufa parser: http://mw.lojban.org/extensions/ilmentufa/altatufa-stodi.html

Here, we get 

{sei gau mo}, {sei gau mi CU mo}, {sei mo}

and

{lo gau mo}, {lo gau mi CU mo}, {lo mo}

both sumti tail and sei-clause use the same bridi

We can also say that x1 of sumti tail has the default value of {zo'e} while {sei}-clause can have the default value e.g. {mi} so that

{sei gleki} = {sei mi gleki}

x1 then can be overriden explicitly

{sei do gleki}

and 

{lo fa do gleki cu sipna} ~= {lo gleki noi du do cu sipna}

Of course, we will still be "illogical and inconsistent" since {lo do gleki} would mean not what one could expect after learning {sei do gleki} and 1. is not aligned but you yourself crossed out this reasoning by saying "I see more importance in LE-clause than SEI-clause, so I would rather keep it"

It's a road to nowhere since in the end one can come and say "There is still the lack of logic here, I assert that usage is insignificant so let's change this rule".

 

--
You received this message because you are subscribed to the Google Groups "lojban" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lojban+un...@googlegroups.com.
To post to this group, send email to loj...@googlegroups.com.
Visit this group at http://groups.google.com/group/lojban.
For more options, visit https://groups.google.com/d/optout.

[guskant]

2015-07-23 20:28 GMT+00:00 Jorge Llambías <jjlla...@gmail.com>:

>
> On Wed, Jul 22, 2015 at 10:43 PM, guskant <gusni...@gmail.com> wrote:
>>
>>
>> ex.1)
>> ro da 
>> poi su'o de 
>> poi ro di zo'u di broda de da
>> zo'u de brode da di 
>> zo'u da brodi de di 
>>
>>
>> Do we have a rule to unify {zo'u}s nested in {poi}? Maybe the preceding
>> ones are outer, but I have never seen the rule written.
>
>

>  The second and third "di" are not bound by "ro di", they are outside its
> scope. and the third "de" is not bound by "su'o de". So filling in the
> missing implicit quantifiers, your sentence becomes:

>
> ro da 
> poi su'o de 

> poi ro di zo'u di broda de da [ku'o su'o di]
> zo'u de brode da di [ku'o su'o de su'o di]

> zo'u da brodi de di 
>

> The variable "di" is bound three times independently, and the variable "de"
> twice. 
>
> A simpler example is "ro da poi su'o de zo'u da de broda zo'u da de brode".
> That's enough to show the issue you're talking about. Linguists call these
> sentences "donkey sentences", after "Every farmer who owns a donkey beats
> it". https://en.wikipedia.org/wiki/Donkey_sentence
>
> A variable bound within a relative clause is not available to be used
> outside the relative clause. The scope of the quantifier is limited to the
> relative clause only.
>

This understanding would solve all my problems. However, my problems come exactly from doubt on this understanding. Could you explain the reason for that a variable bound within a relative clause is not available to be used outside the relative clause in spite of they stand on the same universe of discourse on the outer bridi?

 

>> ex.2)
>> ro da 
>> poi su'o de zo'u de broda da
>> zi'e poi ro di zo'u di brode da
>> zo'u da brodi de di
>>
>> Do we have a rule to unify {zo'u}s in {poi} connected with {zi'e}? Maybe
>> we can define the preceding ones are outer. If so, ex.2 seems apparently the
>> same with
>>
>> ex.2-1)
>> ro da 
>> poi ro di zo'u di brode da
>> zi'e poi su'o de zo'u de broda da
>> zo'u da brodi de di
>>
>> but actually they will have different meaning.
>
>

> The two have the same meaning, but the final "de" and "di" are independent
> of the ones inside the relative clauses. 

>
>> ex.3)
>> su'o de zo'u ko'a poi ro da zo'u da broda ke'a cu brode de
>>
>> When {zo'u} in {poi} is covered with a constant {ko'a}, is it considered
>> outer than {su'o de} or inner?
>
>

> This one is not problematic (other than the issue of what exactly a
> restrictive clause does to a constant, but this has nothing to do with
> whether or not the restrictive clause contains bound variables). Presumably
> "ko'a poi ro da zo'u da broda ke'a" are the ko'as that everyone brodas. So
> the sentence says that there is someone that the ko'as that everyone broda,
> brode. 

[Jorge Llambías]

On Thu, Jul 23, 2015 at 5:52 PM, guskant <gusni...@gmail.com> wrote:

2015-07-23 20:28 GMT+00:00 Jorge Llambías <jjlla...@gmail.com>:
>
> A variable bound within a relative clause is not available to be used
> outside the relative clause. The scope of the quantifier is limited to the
> relative clause only.

This understanding would solve all my problems. However, my problems come exactly from doubt on this understanding. Could you explain the reason for that a variable bound within a relative clause is not available to be used outside the relative clause in spite of they stand on the same universe of discourse on the outer bridi?

That's just how first order predicate logic works. A quantifier quantifies a bridi. The variable is just a place-keeping device, internal to the mechanism of quantification, it doesn't refer to anything, so it can't be used from outside the bridi as if it referred to something. It has nothing to do with universe of discourse. 

[Ilmen]

Please correctly if I made any mistake:

ro da poi ke'a su'o de viska cu se kanla su'o di
= ro da poi su'o de zo'u ke'a de viska ku'o su'o di zo'u di da kanla
= ro da zo'u ganai su'o de zo'u da de viska gi su'o di zo'u di da kanla
= roldza fa loka ko'a ce'ai ganai suzdza fa loka ko'a ce'u viska gi suzdza fa loka ce'u ko'a kanla
= roldza fa loka ko'a ce'ai zilvlina fa lodu'u jitfa fa lodu'u suzdza fa loka ko'a ce'u viska kei kei kei fe lodu'u suzdza fa loka ce'u ko'a kanla

With this expansion, you can see that actually da/de/di outside of a prenex expand to {ce'u}, so it is an open slot used for making properties of which we will quantify how many entities they are satisfied with, using logical quantification predicates (roldza, suzdza).

mi'e la .ilmen. mu'o

[Ilmen]

On 23/07/2015 23:56, Ilmen wrote:

Please correctly if I made any mistake:

*Please correct me if I made any mistake.

[guskant]

Le jeudi 23 juillet 2015 21:15:26 UTC, xorxes a écrit :

On Thu, Jul 23, 2015 at 5:52 PM, guskant <gusni...@gmail.com> wrote:

2015-07-23 20:28 GMT+00:00 Jorge Llambías <jjlla...@gmail.com>:
>
> A variable bound within a relative clause is not available to be used
> outside the relative clause. The scope of the quantifier is limited to the
> relative clause only.

This understanding would solve all my problems. However, my problems come exactly from doubt on this understanding. Could you explain the reason for that a variable bound within a relative clause is not available to be used outside the relative clause in spite of they stand on the same universe of discourse on the outer bridi?

That's just how first order predicate logic works. A quantifier quantifies a bridi. The variable is just a place-keeping device, internal to the mechanism of quantification, it doesn't refer to anything,

I know up to this.

so it can't be used from outside the bridi as if it referred to something. It has nothing to do with universe of discourse. 

mu'o mi'e xorxes

I didn't say the variable bound by a quantifier refers to something. When a variable is bound by a quantifier, the variable has a domain on or in a universe of discourse, otherwise the binding is meaningless. As long as a bridi including bound variables has a truth value, there should be a universe of discourse.

{poi}-clause attached to a variable restricts the domain of variable to a certain part of universe of discourse of the outer bridi. The universe should be shared by the inner and outer of {poi}.

 

[guskant]

It seems correct. And then how do you transform it into a prenex normal form? 

 

[Ilmen]

Once again, please correct me if I made any mistake:

{ro da zo'u ganai su'o de zo'u da de viska gi su'o di zo'u di da kanla}
= ∀x((∃y(viska(x,y))) → (∃z(kanla(z,x)))

# Application of the "(∃xϕ) → ψ" ≡ "∀x(ϕ → ψ)" transformation:
≡ ∀x∀y(viska(x,y) → (∃z(kanla(z,x)))

# Application of the "ϕ → (∃xψ)" ≡ "∃x(ϕ → ψ)" transformation:
≡ ∀x∀y(∃z(viska(x,y) → kanla(z,x))

= {ro da ro de su'o di zo'u ganai da de viska gi di da kanla} (prenex normal form)

[guskant]

I appreciate your formation. One point to confirm is about your first formation using {roldza} and {suzdza}. That is one of a reasonable interpretation of a bridi nested in {poi}-clause, and I agree to that interpretation. Is that a common interpretation among lojban speakers? In fact, la xorxes seems to be against it, and the prenex in {poi}-clause cannot be put out according to his interpretation, and his interpretation is also consistent.

 

[Jorge Llambías]

On Sun, Jul 26, 2015 at 7:04 PM, guskant <gusni...@gmail.com> wrote:

Le dimanche 26 juillet 2015 21:34:54 UTC, Ilmen a écrit :

On 26/07/2015 15:29, guskant wrote: 

Le jeudi 23 juillet 2015 21:56:28 UTC, Ilmen a écrit :

ro da poi ke'a su'o de viska cu se kanla su'o di
= ro da poi su'o de zo'u ke'a de viska ku'o su'o di zo'u di da kanla
= ro da zo'u ganai su'o de zo'u da de viska gi su'o di zo'u di da kanla
= roldza fa loka ko'a ce'ai ganai suzdza fa loka ko'a ce'u viska gi suzdza fa loka ce'u ko'a kanla
= roldza fa loka ko'a ce'ai zilvlina fa lodu'u jitfa fa lodu'u suzdza fa loka ko'a ce'u viska kei kei kei fe lodu'u suzdza fa loka ce'u ko'a kanla

It seems correct. And then how do you transform it into a prenex normal form?

{ro da zo'u ganai su'o de zo'u da de viska gi su'o di zo'u di da kanla}

= ∀x((∃y(viska(x,y))) → (∃z(kanla(z,x)))

# Application of the "(∃xϕ) → ψ" ≡ "∀x(ϕ → ψ)" transformation:
≡ ∀x∀y(viska(x,y) → (∃z(kanla(z,x)))

# Application of the "ϕ → (∃xψ)" ≡ "∃x(ϕ → ψ)" transformation:
≡ ∀x∀y(∃z(viska(x,y) → kanla(z,x))

= {ro da ro de su'o di zo'u ganai da de viska gi di da kanla} (prenex normal form)

I appreciate your formation. One point to confirm is about your first formation using {roldza} and {suzdza}. That is one of a reasonable interpretation of a bridi nested in {poi}-clause, and I agree to that interpretation. Is that a common interpretation among lojban speakers? In fact, la xorxes seems to be against it, and the prenex in {poi}-clause cannot be put out according to his interpretation, and his interpretation is also consistent.

Why do I seem to be against it? I agree with Ilmen's expansions, and notice that he does have a prenex in a poi clause in one of the steps.

[guskant]

Le lundi 27 juillet 2015 00:02:32 UTC, xorxes a écrit :

Why do I seem to be against it? I agree with Ilmen's expansions, and notice that he does have a prenex in a poi clause in one of the steps.

mu'o mi'e xorxes

What Ilmen showed in the first formation is giving a model to the nested {poi}-clause: that is a meta-linguistic formation mentioning truth value. You said that the internal bound variables in {poi}-clause have nothing to do with universe of discourse, and I thought you meant the truth value of the internal bridi of {poi}-clause has nothing to do with the truth value of the outer bridi. Then it is against the formation of Ilmen that connects the truth value of inner and outer of {poi}-clause. 

[Spheniscine (la zipcpi)]

About Pxoi vs Nxoi:

We *could* make say {ni'oi} Nxoi, while xoi = Pxoi (either that or assign {pi'oi}, but I'm reluctant to use that cmavo space if we aren't gonna kill {xoi}, which might be impossible to do at this point).

On the other hand though, there might be a need for a UI to mark "non-restrictive" information, since {xoi} is proposed to be an expansion for {fi'o}, BAI, and sumtcita anyway. Thus this UI can then be useful to mark any sumtcita place (or even, say, seltau) as incidental information.

I *think* {ta'o} works for that purpose, but others might disagree. There's also {ra'unai}, but I'm not sure if it's suitable, and I'd prefer an unnegated form because I believe that information should be considered restrictive by default.

[Alex Burka]

I much prefer a UI marker for [non-]restrictiveness. That unifies a lot of stuff, providing a way to think about {ne} in terms of {pe}, {noi} in terms of {poi}, etc and reduces cmavo proliferation (refer to the Cmavo Non-Proliferation Act of 2014, zo'o). And {ta'o} seems to fit.

mu'o mi'e la durkavore

--
You received this message because you are subscribed to a topic in the Google Groups "lojban" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/lojban/ylLq6Urt3N4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to lojban+un...@googlegroups.com.

[guskant]

ie zo ta'o mapti i ji'a lu ra'unai li'u na'e ka'e se smuni lo du'u na'e jimte

[Ilmen]

Actually I'm not sure anymore which of those {xoi} should be. Initially
I made this particle as a better and more precise alternative to {fi'o},
that would parallel NOI in syntax and enable rewording any sumtcita tag
as a xoi-phrase. However it's turning out that maybe there are two
distinct and mutually exclusive categories of sumtcita cmavo, namely
subordinating sumtcita (ka'e, ka'enai…), whose host bridi's truth value
depends on the sumtcita's meaning, versus realis/non-subortinating
sumtcita (ri'a, ri'anai…), whose host bridi's truth value is independent
from the sumtcita's, so that there couldn't exist an all-encompassing
cmavo for turning any predicate into the semantic equivalent of any
sumtcita clause.

Here is the basis for supposing that sumtcita can be split into two
categories, depending on whether they affect the truth value of the
outer bridi that hosts them so they couldn't be removed without changing
the sentence's meaning:

NON-SUBORDINATING/INDEPENDENT TAGS:
• {mi sipna ri'a lo nu tatpi} entails {mi sipna}.
• {mi sipna ri'a nai lo nu tatpi} entails {mi sipna}.
• {mi citka se pi'o lo forca} entails {mi citka}.
• {snime carvi ca nai lo nu critu} entails {snime carvi}.

SUBORDINATING TAGS:
• {mi sipna ka'e} doesn't entail {mi sipna}.
• {mi sipna ka'e nai} doesn't entail {mi sipna}.
• {mi sipna na ku} doesn't entail {mi sipna}.
• {mi sipna va'o lo nu tatpi} doesn't seem to entail {mi sipna} in
modern mainstream usage.

So it seems two versions of {xoi} would be needed for covering all these
cases.

This also raises the question of whether {fi'o} is subordinating, i.e.
whether {mi sipna fi'o se cumki} entails {mi sipna} or not.

mi'e la .ilmen. mu'o

On 21/07/2015 17:12, selpa'i wrote:
> la .guskant. cu cusku di'e

[Jorge Llambías]

On Tue, Aug 4, 2015 at 6:47 PM, Ilmen <ilmen....@gmail.com> wrote:

NON-SUBORDINATING/INDEPENDENT TAGS:
• {mi sipna ri'a lo nu tatpi} entails {mi sipna}.
• {mi sipna ri'a nai lo nu tatpi} entails {mi sipna}.
• {mi citka se pi'o lo forca} entails {mi citka}.
• {snime carvi ca nai lo nu critu} entails {snime carvi}.

I'm not sure I agree with this part. I admit they all appear to be strong pragmatic entailments, especially without context, but are they logical entailments? How would you translate, for example: "I (only) eat with a fork. There are no forks. Therefore, I do not eat."

[Gleki Arxokuna]

2015-08-05 0:47 GMT+03:00 Ilmen <ilmen....@gmail.com>:

• {mi sipna va'o lo nu tatpi} doesn't seem to entail {mi sipna} in modern mainstream usage.

this can be just a shortcut for {va'o da'i}. I don't think {da'i} should be implied.

Similarly, {ko'a va'o broda} in modern usage does seem to imply {mi broda}.

And I don't think that {ko'a ka'e broda} immediately excludes {mi ca'a broda}.

[Ilmen]

On 05/08/2015 08:26, Gleki Arxokuna wrote:
> And I don't think that {ko'a ka'e broda} immediately excludes {mi ca'a
> broda}.

Indeed, {ko'a ka'e broda} neither entails {ko'a ja'a broda} nor {ko'a na
broda}; it just claims {lo nu broda cu cumki}, as far as I can tell.

[selpa'i]

la .ilmen. cu cusku di'e

> NON-SUBORDINATING/INDEPENDENT TAGS: • {mi sipna ri'a lo nu tatpi}
> entails {mi sipna}. • {mi sipna ri'a nai lo nu tatpi} entails {mi
> sipna}. • {mi citka se pi'o lo forca} entails {mi citka}. • {snime
> carvi ca nai lo nu critu} entails {snime carvi}.

(Quoting Ilmen, but making a general comment)

One issue with any sentence containing abstractions is that they can
carry an implicit {da'i}, which sometimes gets (again, implicitly)
applied to the main bridi. With certain tags this is more common than
with others, but the possibility is always there.

However, assuming {da'i nai} in all the above abstractions (and main
bridi), the entailment should be a logical one.

For example, I understand {mi sipna ri'a lo nu tatpi} as expanding as
follows:

mi sipna ri'a lo nu tatpi

lo nu mi sipna cu fasnu gi'e se rinka lo nu tatpi

In other words, the tag operates on the same level as the main bridi,
and their relationship is that of logical AND. The entire sentence thus
claims the conjunction of the two connnectands and nothing else. (and
this is also how Pxoi would work)

When a negation precedes the sentence, then the negation negates the
conjunction only:

na ku mi sipna ri'a lo nu tatpi
na ku lo nu mi sipna cu fasnu gi'e se rinka lo nu tatpi

Which leaves open whether or not any sleeping occured. All we know is
that if sleeping occured it was not caused by being tired.

Of course, *if* there is a {da'i}, then it becomes less clear, and the
realis-ness of the main bridi then might depend on the semantics of the
tag. In usage there is usually a difference in interpretation between
{mi pacna lo nu da'i ricfu} ({mi pacna} is asserted) and {mi klama va'o
lo nu da'i do kansa mi} ({mi klama} is often not understood to be
asserted), even though both have a {da'i}.