{tu'e...tu'u} in NU

[Spheniscine (la zipcpi)]

One problem that crops up now and then concerns how to include more than one bridi within a NU clause. An example is my translation of Deuteronomy 8:12~14, which I have rendered:

.i tezu'ebo rivbi va'o lo nu do mo'u citka jebo cu tolxagji je cu zbasu lo zabna zdani je cu xabju ri
ju'ei loi cagda'u be do ferti se panzi ju'ei loi rijno joi loi solji vu'o po do zei'a sormei ju'ei loi ro se ponse be do zei'asai sormei kei
lo nu lo menli be do co'a se'ijgi ju'ei do tolmo'i tu'a la .iauex. ku noi cevni do je noi vimcu do la .misr. ku no'u le tutra pe lo ka bapse'u

(note, uses simplified connectives, the experimental tag {zei'a} which is basically {fi'o te zenba}: {zei'a sormei} means "increasingly many" / "to increase in number", and {ju'ei}, which shall be discussed. Yes, I'm pretty terrible with abusing experimental constructs.)

There are several solutions:

official-Lojban solution #1: use {ju'e ... gi ...}. Requires forethought, and only takes two bridi. Chaining them to allow more is impractical.

official-Lojban solution #2: close each NU clause then join it to a new abstractor sumti. e.g. {lo nu broda kei jo'u lo nu brode}. Pretty wordy.

experimental-grammar solution #1: {ju'ei}; essentially acts like a tight-scope {.i} that doesn't close sub-clauses. {lo nu broda ju'ei brode}. Has the advantage of not requiring forethought, but has the disadvantage of not automatically closing any bridi tails; if one of your abstraction-bridi happens to contain a lot of nested NU or POI, they must be closed manually.

experimental-grammar solution #2: Allow {tu'e...tu'u} to act as a single bridi in NU clauses: {lo nu tu'e broda .i brode tu'u}. Does require forethought, but gives a new "level zero" for {.i} to automatically reset to until closed by {tu'u}.

And bonus official-Lojban solution #3, with an experimental shorthand that doesn't require new grammar: {tu'a la'e lu broda .i brode li'u}, with {tu'ai} (selma'o LU) being shorthand for {tu'a la'e lu}. Useful, but may be semantically vague, given lack of LO NU qualification, and {tu'a} possibly implying missing information.

Bonus problem: A related problem is how to make several sentences share a sumtcita-term. e.g: *{ca lo nu mi stuvi'e le sralygu'e kei tu'e lo zabna cu fasnu .i lo mabla cu fasnu tu'u}. {ju'e...gi} also works here, but retains the chaining problem, while I don't know if any of the other listed solutions work.

[Spheniscine (la zipcpi)]

Correction: It's {ju'e gi ... gi...}; e.g. {ju'e gi broda gi brode}

[Spheniscine (la zipcpi)]

Related to the "bonus problem" is goi'e; it really "wants" to be a sumtcita, as COI is awkward and undesirably also accepts selbri, but then it won't work with {tu'e...tu'u}.

[guskant]

Le mercredi 15 juillet 2015 04:26:50 UTC, Spheniscine (la zipcpi) a écrit :

One problem that crops up now and then concerns how to include more than one bridi within a NU clause. An example is my translation of Deuteronomy 8:12~14, which I have rendered:

.i tezu'ebo rivbi va'o lo nu do mo'u citka jebo cu tolxagji je cu zbasu lo zabna zdani je cu xabju ri
ju'ei loi cagda'u be do ferti se panzi ju'ei loi rijno joi loi solji vu'o po do zei'a sormei ju'ei loi ro se ponse be do zei'asai sormei kei
lo nu lo menli be do co'a se'ijgi ju'ei do tolmo'i tu'a la .iauex. ku noi cevni do je noi vimcu do la .misr. ku no'u le tutra pe lo ka bapse'u

(note, uses simplified connectives, the experimental tag {zei'a} which is basically {fi'o te zenba}: {zei'a sormei} means "increasingly many" / "to increase in number", and {ju'ei}, which shall be discussed. Yes, I'm pretty terrible with abusing experimental constructs.)

There are several solutions:

official-Lojban solution #1: use {ju'e ... gi ...}. Requires forethought, and only takes two bridi. Chaining them to allow more is impractical.

official-Lojban solution #2: close each NU clause then join it to a new abstractor sumti. e.g. {lo nu broda kei jo'u lo nu brode}. Pretty wordy.

If you need a structure similar to your translation above, official-Lojban solution #2 must be rather {lo nu broda kei ju'e nu brode}.

 

experimental-grammar solution #1: {ju'ei}; essentially acts like a tight-scope {.i} that doesn't close sub-clauses. {lo nu broda ju'ei brode}. Has the advantage of not requiring forethought, but has the disadvantage of not automatically closing any bridi tails; if one of your abstraction-bridi happens to contain a lot of nested NU or POI, they must be closed manually.

experimental-grammar solution #2: Allow {tu'e...tu'u} to act as a single bridi in NU clauses: {lo nu tu'e broda .i brode tu'u}. Does require forethought, but gives a new "level zero" for {.i} to automatically reset to until closed by {tu'u}.

La zantufa-0.2 ( http://guskant.github.io/gerna_cipra/zantufa-0.2.html ) allows a full statement in noi-clause or nu-clause. It means the following texts are parsed by la zantufa-0.2 as follows:

1. sentences connected with {i JOI} or {i (tag) bo}:

{nu broda i ju'e brode}

(CU [nu {<CU (¹broda VAU¹)> <i ju'e (¹CU [brode VAU]¹)>} KEI] VAU) 

{nu broda i bo brode}

(CU [nu {<CU (¹broda VAU¹)> <i bo (¹CU [brode VAU]¹)>} KEI] VAU) 

{nu broda i ba bo brode}

(CU [nu {<CU (¹broda VAU¹)> <i ba bo (¹CU [brode VAU]¹)>} KEI] VAU) 

Use KEI or KUhO to cut the statement:

{nu broda kei i ju'e brode}

([CU {nu <CU (¹broda VAU¹)> kei} VAU] [i ju'e {CU <brode VAU>}]) 

{brodi ra noi broda ku'o i ju'e brode}

([CU {brodi <ra (¹noi [CU {broda VAU}] ku'o¹)> VAU}] [i ju'e {CU <brode VAU>}]) 

({brodi da} is required here because this {ju'e} connects two sentences, not a fragment and a sentence. It might be allowed in the future version of zantufa, though not yet decided.)

2. sentences between {tu'e ... tu'u}:

{nu tu'e broda i brode tu'u}

(CU [nu {tu'e <(¹CU [broda VAU]¹) (¹i [CU {brode VAU}]¹)> tu'u} KEI] VAU) 

3. sentences connected with forethought connective:

{nu ju'e gi broda gi brode gi brodi gi brodo gi brodu}

(CU [nu {CU <(¹ju'e gi¹) broda (¹[gi brode] [gi brodi] [gi brodo] [gi brodu]¹) GIhI> VAU} KEI] VAU) 

(Forethought connectives of la zantufa-0.2 can connect more than three "statements (not only sentences)", and {gi'i} is used as the elidible terminator GIhI, not as GIhA. See 

http://mw.lojban.org/papri/zantufa_jonma%27o#lo_li.27erla.27i_jonma.27o

for more info.)

As 1 is allowed, your solution #1 is unnecessary, and ma'oi JUhEI in that usage will never implemented to la zantufa. 

 

And bonus official-Lojban solution #3, with an experimental shorthand that doesn't require new grammar: {tu'a la'e lu broda .i brode li'u}, with {tu'ai} (selma'o LU) being shorthand for {tu'a la'e lu}. Useful, but may be semantically vague, given lack of LO NU qualification, and {tu'a} possibly implying missing information.

Bonus problem: A related problem is how to make several sentences share a sumtcita-term. e.g: *{ca lo nu mi stuvi'e le sralygu'e kei tu'e lo zabna cu fasnu .i lo mabla cu fasnu tu'u}. {ju'e...gi} also works here, but retains the chaining problem, while I don't know if any of the other listed solutions work.

Use {zo'u}:

{ca lo nu mi stuvi'e le sralygu'e zo'u tu'e lo zabna cu fasnu .i lo mabla cu fasnu tu'u}

It is parsed even by the official parser:

([{ca <lo (¹nu [mi {CU <stuvi'e (²le sralygu'e KU²) VAU>}] KEI¹) KU>} zo'u] [tu'e {<(¹lo zabna KU¹) (¹cu [fasnu VAU]¹)> <i (¹lo mabla KU¹) (¹cu [fasnu VAU]¹)>} tu'u]) 

zo'u-clause spans over a statement, that is to say, sentences connected with {i JA/JOI}, {i (tag) bo}, {JOI/tag gi ... gi ...} or {GA ... gi ...}. These connectives can be used instead of {tu'e ... tu'u} in many cases. For example,

{ca lo nu mi stuvi'e le sralygu'e zo'u lo zabna cu fasnu .ibo lo mabla cu fasnu}

is parsed by the official parser as follows:

([{ca <lo (¹nu [{mi CU} {stuvi'e <le sralygu'e KU> VAU}] KEI¹) KU>} zo'u] [{<lo zabna KU> cu} {fasnu VAU}] [i bo {lo mabla KU} cu] [fasnu VAU]) 

 

[la durka]

In general I think it's a bad idea to change {.i} so that it doesn't always start a new sentence. The major problems are that it deals a heavy blow to the whole concept of elidible terminators, if you can't "just start a new sentence" to get out of a deeply nested pit of abstractions; and it means it's no longer possible to quickly scan a text for {.i} to separate the sentences -- you need to parse the entire text to find the sentence boundaries. So I would definitely favor a solution for the first problem that doesn't change the grammar so drastically. (Not to say I'm against new grammar in general, of course! But I find severe disadvantages with this particular proposal.)

The bonus problem is not so hard -- you can use {zo'u} as la .guskant. said, or indeed plain tu'e/tu'u, or connect the sentences with {.ije} or {.ibo} to suggest close binding.

- mu'o mi'e la durkavore

P.S. It occurs to me that la zipcpi's "super-terminator" {.i'au}/{.iau}, which I previously dismissed, could cover some of my above objections to changing the meaning of {.i}. However such a super-terminator has yet to be formalized and it seems silly to introduce the change if we need an escape hatch cmavo for the common case!

[guskant]

Le mercredi 15 juillet 2015 13:51:59 UTC, la durka a écrit :

In general I think it's a bad idea to change {.i} so that it doesn't always start a new sentence. The major problems are that it deals a heavy blow to the whole concept of

Officially, {i} always start a new "statement", while {i JOI} or {i (tag) BO} start a new sentence. La zantufa does not differs in this point.

 

elidible terminators, if you can't "just start a new sentence" to get out of a deeply nested pit of abstractions; and it means it's no longer possible to quickly scan a text for {.i} to separate the sentences -- you need to parse the entire text to find the sentence boundaries. So I would definitely favor a solution for the first problem that doesn't change the grammar so drastically. (Not to say I'm against new grammar in general, of course! But I find severe disadvantages with this particular proposal.)

Allowing a full statement in nu-clause or noi-clause requires KEI or KUhO in more cases than the official grammar, and it is just like cmevla as selbrisle requires CU in more cases. This change is required mainly from a logical point of view, and incidentally for grammatical simplicity.

A statement in nu-clause has its own universe of discourse independent from the outer bridi. A full logical reasoning must be possible in that universe, but the official grammar limits the freedom of logical expression in nu-clause to so-called "subsentence". I just gave the full liberty of logical expression to that universe in nu-clause. 

If Lojban is called "logji bangu", I think it's structure related to Logic should be more refined, otherwise "logji" should be removed from its sub-name.

As for noi-clause, the universe of discourse is the same as that of outer bridi. I don't think noi-clause must have full liberty of logical expression, but it is also liberated only for simplicity of grammar.

If you don't like this proposal, just don't use la zantufa. It is defined as "zabna fi la guskant" parser, not "fi do".

[Jacob Thomas Errington]

On 07/15/2015 09:51 AM, la durka wrote:

In general I think it's a bad idea to change {.i} so that it doesn't always start a new sentence. The major problems are that it deals a heavy blow to the whole concept of elidible terminators, if you can't "just start a new sentence" to get out of a deeply nested pit of abstractions; and it means it's no longer possible to quickly scan a text for {.i} to separate the sentences -- you need to parse the entire text to find the sentence boundaries.

We can't really just scan the text for {.i} as it may be quoted, specifically by {lu...li'u}. Indeed, lu-quotes already form a construct that creates a new "level zero" for sentences; giving us this flexibility in any bridi context would merely act as a generalization of this existing capability.

That said, I'm in favour of allowing tu'e..tu'u within NU, and really, at the beginning of any bridi.

.i mi'e la tsani mu'o

[Jorge Llambías]

On Wed, Jul 15, 2015 at 4:29 AM, guskant <gusni...@gmail.com> wrote:

3. sentences connected with forethought connective:

{nu ju'e gi broda gi brode gi brodi gi brodo gi brodu}

(CU [nu {CU <(¹ju'e gi¹) broda (¹[gi brode] [gi brodi] [gi brodo] [gi brodu]¹) GIhI> VAU} KEI] VAU) 

(Forethought connectives of la zantufa-0.2 can connect more than three "statements (not only sentences)", and {gi'i} is used as the elidible terminator GIhI, not as GIhA. See 

http://mw.lojban.org/papri/zantufa_jonma%27o#lo_li.27erla.27i_jonma.27o

for more info.)

I like this use of "gi" because connected lists are by far more common than embedded binary connectives. The meanings of "ge ... gi ... gi ..." and "ga ... gi ... gi ..." are fairly obvious, but what are the proposed generalizations for "go" and "gu"? Is "go" all true or all false? What are "gu", "se gu", "te gu", etc?

mu'o mi'e xorxes

[Jorge Llambías]

On Wed, Jul 15, 2015 at 3:35 PM, Jacob Thomas Errington <ts...@mail.jerrington.me> wrote:

On 07/15/2015 09:51 AM, la durka wrote:

 and it means it's no longer possible to quickly scan a text for {.i} to separate the sentences -- you need to parse the entire text to find the sentence boundaries.

We can't really just scan the text for {.i} as it may be quoted, specifically by {lu...li'u}.

Or it could be grabbed by a magic word: ".i bu" ".i zei da", etc. Because of magic words, there's not much you can do with a Lojban text by just scanning for words, you always need to parse first in order to know whether the words are fulfilling their usual functions or not. 

[Alex Burka]

True. Most magic words are local, though. Only ZOI and LOhU (and LU) can have a large extent. And magic is comparatively rare, whereas NU/NOI is extremely common and I fear a large amount of text would be broken by changing the semantics of .i after abstractions.

--
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/NRUIud_OCj8/unsubscribe.
To unsubscribe from this group and all its topics, 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.

[Jacob Thomas Errington]

On 07/15/2015 06:21 PM, Alex Burka wrote:
>
> True. Most magic words are local, though. Only ZOI and LOhU (and LU)
> can have a large extent. And magic is comparatively rare, whereas
> NU/NOI is extremely common and I fear a large amount of text would be
> broken by changing the semantics of .i after abstractions.
>

Changing the semantics of {.i} after abstractions would be disastrous,
which is why you'd need to explicitly use {tu'e}.

.i mi djica lo nu tu'e do klama lo mi zdani .ibabo mi'o samkei kansi'u tu'u

Like this, {tu'e} is essentially acting as LU, but the result of which
is not a sumti, but a bridi.

[Spheniscine (la zipcpi)]

Another discovered use of tu'e...tu'u in NU: Algorithms, like in computer programs, or instructional texts: e.g.

ifle fa lo du'u ko'a dubmau li no kei

lo nu tu'e

[something]

tu'u kei lo nu tu'e

[something else]

tu'u

Oh yeah, yet another take on the "bonus problem":

*{tu'e lo mabla cu fasnu .i lo zabna cu fasnu tu'u ca lo nu mi stuvi'e le sralygu'e}

[Spheniscine (la zipcpi)]

*{tu'e lo mabla cu fasnu .i lo zabna cu fasnu tu'u ca lo nu mi stuvi'e le sralygu'e}

i.e. How to fix for standard Lojban? Is this a construct that should be supported? etc. 

[Spheniscine (la zipcpi)]

found official-solution #1 for bonus-bonus problem:

tu'e lo zabna cu fasnu .i lo mabla cu fasnu tu'u sei ca lo nu mi stuvi'e le sralygu'e cu co'e

Kinda ugly though.

[guskant]

La zantufa gives only syntactic structure, and I have not yet suggested the semantic structure. From any approach, it is reasonable to interpret that {gu A gi B gi C} has the same truth value as A.

As for {se gu}, one of possible interpretations is grouping of binary transitive connective: that is, {segu A gi B gi C} = {segu (segu A gi B gi'i) gi C} = {segu A gi (segu B gi C)}, and then the truth value of {segu A gi B gi C} is the same as C. However, as you suggested, we may give another interpretation that {se gu} draws the truth value of B, {te gu} draws the truth value of C, and so on. I like the latter interpretation, because it cannot be easily represented by grouping system of binary connectives, and therefore profits from the n-ary forethought connective system.

For the same reason, I prefer the interpretation of "all true or all false" for {go} and "one and only one of them is true" for {gonai} to the interpretation of grouping of binary connectives.

 

[Spheniscine (la zipcpi)]

n-ary gi is interesting - it may perhaps be a step toward another possible solution to the "extended logical connectives" problem I noted here: https://groups.google.com/forum/#!topic/lojban/Jz6bw4zKn7k

It does however break my definition of {ifle} as {ge ganai me'au bo'a* gi me'au bo'e giga me'au bo'a gi me'au bo'i}, but I prefer {ganai me'au bo'a gi me'au bo'e .ije ga me'au bo'a gi me'au bo'i} over it anyway, and with your n-ary gi, {ge ganai me'au bo'a gi me'au bo'e gi'i giga me'au bo'a gi me'au bo'i} works...

(* Yes, I took bo'V for x1 ~ x5. Additionally, {ka'ei} is a special variant of {ka} that preassigns {bo'V} to its ce'u-variables.)

[Spheniscine (la zipcpi)]

On n-ary gi / forethought connectives: What about sumtcita? {pu je ca} works, but {ge pu gi ca} doesn't.{ge puku gi ca(ku)} works though, in camxes-exp and zantufa.

[Alex Burka]

So, to the question of that {gu} and {go} mean with more than 2 arguments, we kind of find ourselves with tradeoff between logic and ease of use.

First I get nervous extending them to >2-ary at all, since it becomes harder to reason about the negations ({gonai X gi Y} is XOR, but {gonai X gi Y gi Z} is not, it's something strange instead -- and nobody can do truth tables in their head, even fluent speakers).

The interpretations you list are undoubtedly useful (particularly the "which one out of these X alternatives"-connective is often asked for, though there is a workable solution using {moi}). But they do not correspond to the mathematical interpretation. For example, n-ary XOR is supposed to be true when an odd number of the arguments are true, which is not obvious (or often useful in speech). Changing this can cause difficulties with composition, for example logical transformations like De Morgans' laws will not work. And {go ko'a gi ko'e gi ko'e [gi'i]} would be different from {go go ko'a gi ko'e gi'i gi ko'i [gi'i]}, which is kind of annoying. Basically we would want to stop defining {go} as iff and {gonai} as xor, because people coming from a math/logic background would have the wrong expectations.

mu'o mi'e la durkavore

[Alex Burka]

Well, zantufa currently goes further than that.

[Jacob Thomas Errington]

On 07/16/2015 06:31 AM, Alex Burka wrote:

Well, zantufa currently goes further than that.

Ah, I had missed that. Sorry.

Then I suppose I'm not a big fan of zantufa, .u'i

[Jorge Llambías]

On Thu, Jul 16, 2015 at 1:23 AM, guskant <gusni...@gmail.com> wrote:

For the same reason, I prefer the interpretation of "all true or all false" for {go} and "one and only one of them is true" for {gonai} to the interpretation of grouping of binary connectives.

One problem with that interpretation for "gonai" is that it doesn't go well with "go ... ginai ... gi ..." and such. The natural Lojbanic interpretation is for "nai" to negate the following connectand. So "gonai A gi B gi C" would be true when A differs in truth value from B and from C.

"One and only one of them" is more useful, but hard to reconcile with the general pattern.