Suggestion for a new animacy marker in Lojban.

142 views
Skip to first unread message

Gleki Arxokuna

unread,
Aug 4, 2012, 2:35:59 AM8/4/12
to loj...@googlegroups.com
Suggestion for a new animacy marker in Lojban.

Many if not most languages divide all predicates into levels of animacy.
English, for instance, has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

This allows quickly determine agents of most actions.
Example:
The woman was looking at the mirror. It was ugly.
Let's try it in Lojban.
{lo ninmu pu ca'o catlu lo minra .i ta pu tolmelbi}
No, too ambiguous. And I opine that counting two sumti back in order to use {ra} is much trickier for human brain than just understanding semantic roles of sumti.
Therefore, I suggest introducing a new marker reflecting animacy of any object. I'll use {xoi} which currently bears no official meaning.

xoi - marks preceding construct as animate
xoinai - marks preceding construct as inanimate

{lo ninmu pu ca'o catlu lo minra. i ta xoinai pu tolmelbi}

However, some languages have more levels of animacy.
The father was looking at his son. He was beautiful.
{lo patfu pu catlu lo bersa .i ta xoixime'i pu melbi}
The author of this sentence probably thinks that children are less animate than grown-ups. 
So we can build a scale ranging from most animate objects to inanimate.
It's only the speaker who decides what level of animacy this or that object has.

Gender-specific pronouns.
You might argue why not add more specific markers reflecting for instance the gender of the object described.
Let's repeat once again.

English has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

In other words, English has two pronouns expressing sex but only one pronoun expressing inanimate objects.
There might be languages that split inanimate levels into other specific classes (furniture, houses, weapons).
Therefore, it would be stupid to try to import all those quirks of natlangs. {ta poi nakni} is fine.

Unsettled issues.
Some languages have "abstractions" in their lowest level of animacy hierarchy.
Lojban is pretty strict when dealing with objects and abstractions. The issue with the scale "su'unai - su'u" that one might imagine remains unsettled.

Jonathan Jones

unread,
Aug 4, 2012, 3:42:21 AM8/4/12
to loj...@googlegroups.com
On Sat, Aug 4, 2012 at 12:35 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
Suggestion for a new animacy marker in Lojban.

Many if not most languages divide all predicates into levels of animacy.
English, for instance, has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

This allows quickly determine agents of most actions.
Example:
The woman was looking at the mirror. It was ugly.
Let's try it in Lojban.
{lo ninmu pu ca'o catlu lo minra .i ta pu tolmelbi}

{my. pu tolmelbi}
 
No, too ambiguous. And I opine that counting two sumti back in order to use {ra} is much trickier for human brain than just understanding semantic roles of sumti.
Therefore, I suggest introducing a new marker reflecting animacy of any object. I'll use {xoi} which currently bears no official meaning.

xoi - marks preceding construct as animate
xoinai - marks preceding construct as inanimate

{lo ninmu pu ca'o catlu lo minra. i ta xoinai pu tolmelbi}

However, some languages have more levels of animacy.
The father was looking at his son. He was beautiful.
{lo patfu pu catlu lo bersa .i ta xoixime'i pu melbi}
The author of this sentence probably thinks that children are less animate than grown-ups. 
So we can build a scale ranging from most animate objects to inanimate.
It's only the speaker who decides what level of animacy this or that object has.

Gender-specific pronouns.
You might argue why not add more specific markers reflecting for instance the gender of the object described.
Let's repeat once again.

English has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

In other words, English has two pronouns expressing sex but only one pronoun expressing inanimate objects.
There might be languages that split inanimate levels into other specific classes (furniture, houses, weapons).
Therefore, it would be stupid to try to import all those quirks of natlangs. {ta poi nakni} is fine.

Unsettled issues.
Some languages have "abstractions" in their lowest level of animacy hierarchy.
Lojban is pretty strict when dealing with objects and abstractions. The issue with the scale "su'unai - su'u" that one might imagine remains unsettled.

I'm one of those weirdos that thinks that the restriction of "it" to only inanimate objects is a bad idea. It is precisely because of that evolution that we have people trying to replace what "it" used to mean with "Singular They" and "ey, eir, em", and "somepony", all of which are, at least in my opinion, absolutely horrid and ineffectual substitutes for just saying frakking IT.

Further, we don't actually have that distinction. It's really more of a very fuzzy line.

The celebration-of-a-new-birth balloons/cards/etc. all say "It's a boy!" and "It's a girl".

We call our ships, cars, motorcycles, planes, ..., "she". ("Oh, a vintage 1950 Thunderbird?!" "Yeah, the old girl's still a beauty, ain't she?")

On a show I watch called "New Girl", the theme song goes "Who's that girl? It's Jess!" Not "Their Jess!", not even "She's Jess!".

Arguably any human knows the difference between the animate and the inanimate, and from what I've seen, words that explicitly provide that distinction just makes things harder. We don't need to have an "(in)animate marker". If you want to be unambiguous as to what you are referring to, don't use ambiguous reference.

--
mu'o mi'e .aionys.

.i.e'ucai ko cmima lo pilno be denpa bu .i doi.luk. mi patfu do zo'o
(Come to the Dot Side! Luke, I am your father. :D )

Gleki Arxokuna

unread,
Aug 4, 2012, 3:48:18 AM8/4/12
to loj...@googlegroups.com
It's all based on the analysis of existing languages.
Sure, English example is not fine.
But Navajo for instance has the following hierarchy.

humans/lightning → infants/big animals → med-size animals → small animals → insects → natural forces → inanimate objects/plants → abstractions

Therefore, lightning is somewhat za'e su'unai.

Jonathan Jones

unread,
Aug 4, 2012, 4:01:16 AM8/4/12
to loj...@googlegroups.com
On Sat, Aug 4, 2012 at 1:48 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
It's all based on the analysis of existing languages.
Sure, English example is not fine.
But Navajo for instance has the following hierarchy.

humans/lightning → infants/big animals → med-size animals → small animals → insects → natural forces → inanimate objects/plants → abstractions

Therefore, lightning is somewhat za'e su'unai.

It doesn't matter what language you use for examples. My position is still the same. The less arbitrary distinctions there are, the better. And merely by the simple fact that every language splits things differently, (although I'll grant that language families tend to be tcesimsa if not mintu,) is enough evidence for me at least that it is arbitrary.

I know that is human nature to classify things. We do it all the time. But not all classifications are useful or necessary.

We have semantically unambiguous ways to refer to things, whether they be people, animals, objects, or what-have-you. We DON'T just have ra/ri/ru, ta/ti/tu, and va/vi/vu. We have the KOhA and VOhA, and the BY, which are more than enough in nearly any circumstance.

Honestly, how would you rather refer to the mirror in your example? {ta xoinai}, or {my.}?
 
--
You received this message because you are subscribed to the Google Groups "lojban" group.
To view this discussion on the web visit https://groups.google.com/d/msg/lojban/-/z98Dn0wZA-cJ.

To post to this group, send email to loj...@googlegroups.com.
To unsubscribe from this group, send email to lojban+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban?hl=en.

Veijo Vilva

unread,
Aug 4, 2012, 4:15:30 AM8/4/12
to loj...@googlegroups.com
On 4 August 2012 10:48, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
It's all based on the analysis of existing languages.
Sure, English example is not fine.
But Navajo for instance has the following hierarchy.
humans/lightning → infants/big animals → med-size animals → small animals → insects → natural forces → inanimate objects/plants → abstractions

In Finnish we don't have separate pronouns for the two sexes, and in informal language these days it's more like a norm to use an inanimate pronoun for everything. Japanese has no "real" pronouns and quite often does just fine by eliding the subject or the object or both of them and letting the context resolve the meaning. We have a scale from nil to what ever complexity the human mind has deemed culturally necessary.
 

On Sat, Aug 4, 2012 at 12:35 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
Suggestion for a new animacy marker in Lojban.

Many if not most languages divide all predicates into levels of animacy.
English, for instance, has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

This allows quickly determine agents of most actions.
Example:
The woman was looking at the mirror. It was ugly.
Let's try it in Lojban.
{lo ninmu pu ca'o catlu lo minra .i ta pu tolmelbi}

 
Here I'd tend towards the Japanese solution of eliding the subject or rather go a step further and elide even the tense:

  {lo ninmu pu ca'o catlu lo minra .i tolmelbi}

  mu'o mi'e veion

--

iesk

unread,
Aug 4, 2012, 7:34:40 AM8/4/12
to loj...@googlegroups.com
Le samedi 4 août 2012 10:01:16 UTC+2, aionys a écrit :
The less arbitrary distinctions there are, the better. And merely by the simple fact that every language splits things differently, (although I'll grant that language families tend to be tcesimsa if not mintu,) is enough evidence for me at least that it is arbitrary.

I know that is human nature to classify things. We do it all the time. But not all classifications are useful or necessary.


ni'o ti'e lo pe'a catni trusi'o befi la lobgu'e ku lei clira kakpa zo'u bau la lojban. cusku be ji'i ro da zifre gi'e cusku be ji'i no da bilga .i sidbo lo du'u ta'u nai kulnu nutli¹ .i go'i lo du'u ta'u nai terzasti naldukse²

ni'o ja'o ru'e bilga lo du'u tu'a lo si'o zo'e zukte ja nalzukte kei ka'e frili se cusku bau ly. kei le bi'u nai trusi'o pe'a (to .i zu'u nai na'e mukti lonu ba'e mi cusku toi)

__
¹ cultural neutrality?
² metaphysical parsimony?

What I am trying to say is that I think that, according to what I perceive to be Lojban's ideology, there should be a relatively easy way to make the animacy/inanimacy distinction (which I, by the way, feel no inclination of making).

mu'o mi'e .iesk.

Gleki Arxokuna

unread,
Aug 4, 2012, 7:44:15 AM8/4/12
to loj...@googlegroups.com


On Saturday, August 4, 2012 12:01:16 PM UTC+4, aionys wrote:
On Sat, Aug 4, 2012 at 1:48 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
It's all based on the analysis of existing languages.
Sure, English example is not fine.
But Navajo for instance has the following hierarchy.

humans/lightning → infants/big animals → med-size animals → small animals → insects → natural forces → inanimate objects/plants → abstractions

Therefore, lightning is somewhat za'e su'unai.

It doesn't matter what language you use for examples. My position is still the same. The less arbitrary distinctions there are, the better. And merely by the simple fact that every language splits things differently, (although I'll grant that language families tend to be tcesimsa if not mintu,) is enough evidence for me at least that it is arbitrary.

I know that is human nature to classify things. We do it all the time. But not all classifications are useful or necessary.

We have semantically unambiguous ways to refer to things, whether they be people, animals, objects, or what-have-you. We DON'T just have ra/ri/ru, ta/ti/tu, and va/vi/vu. We have the KOhA and VOhA, and the BY, which are more than enough in nearly any circumstance.

Honestly, how would you rather refer to the mirror in your example? {ta xoinai}, or {my.}?
Honestly, we both speak European languages. That's why our opinion means nothing as we can't remove our cultural bias.
We need someone from another culture (like Navajo speaker).
 
To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/lojban?hl=en.

selpa'i

unread,
Aug 4, 2012, 7:52:05 AM8/4/12
to loj...@googlegroups.com
Am 04.08.2012 08:35, schrieb Gleki Arxokuna:
Suggestion for a new animacy marker in Lojban.

Many if not most languages divide all predicates into levels of animacy.
English, for instance, has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

This allows quickly determine agents of most actions.
Example:
The woman was looking at the mirror. It was ugly.
Let's try it in Lojban.
{lo ninmu pu ca'o catlu lo minra .i ta pu tolmelbi}

"ta" does not work for back-referencing. The above sentence would most simply be expressed as either (1) or (2):

(1) lo ninmu pu ca'o catlu lo minra .i my tolmelbi
    "The woman was looking at the mirror. It was ugly."

(2) lo ninmu pu ca'o catlu lo minra .i ny tolmelbi
    "The woman was looking at the mirror. She was ugly."

There is absolutely no need to use (in-)animacity or any other arbitrary hierarchy. If you have to, you can always use existing words to specify such things, but you don't have to invent new cmavo.

Gender-specific pronouns.
You might argue why not add more specific markers reflecting for instance the gender of the object described.
Let's repeat once again.

English has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

In other words, English has two pronouns expressing sex but only one pronoun expressing inanimate objects.
There might be languages that split inanimate levels into other specific classes (furniture, houses, weapons).
Therefore, it would be stupid to try to import all those quirks of natlangs. {ta poi nakni} is fine.

Yes. If you absolutely have to, you can specify gender/sex through various techniques, but forcing the speaker to always do so would imply that sex/gender is of primary importance, which in turn would potentially support a sexist world-view. If one is not able to talk about something without knowing its gender or sex, then that is a definite short-coming of the language.


Unsettled issues.
Some languages have "abstractions" in their lowest level of animacy hierarchy.
Lojban is pretty strict when dealing with objects and abstractions. The issue with the scale "su'unai - su'u" that one might imagine remains unsettled.

You either have a NU or you don't. What scale are you imagining?

mu'o mi'e la selpa'i

-- 
pilno zo le xu .i lo dei bangu cu se cmene zo lojbo .e nai zo lejbo

Gleki Arxokuna

unread,
Aug 4, 2012, 8:07:45 AM8/4/12
to loj...@googlegroups.com


On Saturday, August 4, 2012 3:52:05 PM UTC+4, selpa'i wrote:
Am 04.08.2012 08:35, schrieb Gleki Arxokuna:
Suggestion for a new animacy marker in Lojban.

Many if not most languages divide all predicates into levels of animacy.
English, for instance, has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

This allows quickly determine agents of most actions.
Example:
The woman was looking at the mirror. It was ugly.
Let's try it in Lojban.
{lo ninmu pu ca'o catlu lo minra .i ta pu tolmelbi}

"ta" does not work for back-referencing.
ta - pro-sumti: that there; nearby demonstrative it; indicated thing/place near listener.
So I showed an example where exactly that meaning was intended.
 
The above sentence would most simply be expressed as either (1) or (2):

(1) lo ninmu pu ca'o catlu lo minra .i my tolmelbi
    "The woman was looking at the mirror. It was ugly."

(2) lo ninmu pu ca'o catlu lo minra .i ny tolmelbi
    "The woman was looking at the mirror. She was ugly."

There is absolutely no need to use (in-)animacity or any other arbitrary hierarchy. If you have to, you can always use existing words to specify such things,
By the way how would you express "x1 is animate of level x2"?
 
but you don't have to invent new cmavo.

Gender-specific pronouns.
You might argue why not add more specific markers reflecting for instance the gender of the object described.
Let's repeat once again.

English has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

In other words, English has two pronouns expressing sex but only one pronoun expressing inanimate objects.
There might be languages that split inanimate levels into other specific classes (furniture, houses, weapons).
Therefore, it would be stupid to try to import all those quirks of natlangs. {ta poi nakni} is fine.

Yes. If you absolutely have to, you can specify gender/sex through various techniques, but forcing the speaker to always do so
No. I'm not forcing. Everything must be optional. 
would imply that sex/gender is of primary importance, which in turn would potentially support a sexist world-view. If one is not able to talk about something without knowing its gender or sex, then that is a definite short-coming of the language.
Sure. 


Unsettled issues.
Some languages have "abstractions" in their lowest level of animacy hierarchy.
Lojban is pretty strict when dealing with objects and abstractions. The issue with the scale "su'unai - su'u" that one might imagine remains unsettled.

You either have a NU or you don't. What scale are you imagining?
Look at the list of the levels of animacy in Navajo. 

selpa'i

unread,
Aug 4, 2012, 8:16:51 AM8/4/12
to loj...@googlegroups.com
Am 04.08.2012 14:07, schrieb Gleki Arxokuna:


On Saturday, August 4, 2012 3:52:05 PM UTC+4, selpa'i wrote:
Am 04.08.2012 08:35, schrieb Gleki Arxokuna:
Suggestion for a new animacy marker in Lojban.

Many if not most languages divide all predicates into levels of animacy.
English, for instance, has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

This allows quickly determine agents of most actions.
Example:
The woman was looking at the mirror. It was ugly.
Let's try it in Lojban.
{lo ninmu pu ca'o catlu lo minra .i ta pu tolmelbi}

"ta" does not work for back-referencing.
ta - pro-sumti: that there; nearby demonstrative it; indicated thing/place near listener.
So I showed an example where exactly that meaning was intended.

You showed an example where [ta] does not at all apply. ta requires you to be able to point at the thing. But the sentence is in the past.


 
The above sentence would most simply be expressed as either (1) or (2):

(1) lo ninmu pu ca'o catlu lo minra .i my tolmelbi
    "The woman was looking at the mirror. It was ugly."

(2) lo ninmu pu ca'o catlu lo minra .i ny tolmelbi
    "The woman was looking at the mirror. She was ugly."

There is absolutely no need to use (in-)animacity or any other arbitrary hierarchy. If you have to, you can always use existing words to specify such things,
By the way how would you express "x1 is animate of level x2"?

zukte.
I have no idea what "level 2" is supposed to mean, but it sounds derogatory. I do not want to sort different species into different levels. I consider all living things equal.


 
but you don't have to invent new cmavo.

Gender-specific pronouns.
You might argue why not add more specific markers reflecting for instance the gender of the object described.
Let's repeat once again.

English has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

In other words, English has two pronouns expressing sex but only one pronoun expressing inanimate objects.
There might be languages that split inanimate levels into other specific classes (furniture, houses, weapons).
Therefore, it would be stupid to try to import all those quirks of natlangs. {ta poi nakni} is fine.

Yes. If you absolutely have to, you can specify gender/sex through various techniques, but forcing the speaker to always do so
No. I'm not forcing. Everything must be optional. 
would imply that sex/gender is of primary importance, which in turn would potentially support a sexist world-view. If one is not able to talk about something without knowing its gender or sex, then that is a definite short-coming of the language.
Sure. 


Unsettled issues.
Some languages have "abstractions" in their lowest level of animacy hierarchy.
Lojban is pretty strict when dealing with objects and abstractions. The issue with the scale "su'unai - su'u" that one might imagine remains unsettled.

You either have a NU or you don't. What scale are you imagining?
Look at the list of the levels of animacy in Navajo.

The list is: humans/lightning → infants/big animals → med-size animals → small animals → insects → natural forces → inanimate objects/plants → abstractions

You can sort all of them into either NU or not-NU. All the animals including humans are objects. Objects are obviously objects too. Natural forces can be either, depending on how you express them. lo lindi  vs lo nu lindi. And finally, abstractions are clearly NU.

mu'o mi'e la selpa'i

Gleki Arxokuna

unread,
Aug 4, 2012, 9:35:14 AM8/4/12
to loj...@googlegroups.com
Excuse me, YOU consider??? Probably I would do the same. But it's not the opinion that others might have.
One of common replies is "Lojban doesn't have the goal to emulate every language". Probably it's true.
But why UI were imported from Amerindian languages? It was the same process of raising expressive power. 
I'm sure that if someone suggested introducing UI now they would be immediately attacked by purists.

Animacy is marked in most languages in the world.
That's why even if English adopts Spivak pronoun "ey" it won't replace "it".
"ey" and "it" usually refer to absolutely different objects playing distinct roles in bridi.
"ey" is animate, usually an agent.
"it" is usually not an agent.
 


 
but you don't have to invent new cmavo.

Gender-specific pronouns.
You might argue why not add more specific markers reflecting for instance the gender of the object described.
Let's repeat once again.

English has at least two levels. These are the pronouns for them
1. Animate. He/she 
2. Inanimate. It

In other words, English has two pronouns expressing sex but only one pronoun expressing inanimate objects.
There might be languages that split inanimate levels into other specific classes (furniture, houses, weapons).
Therefore, it would be stupid to try to import all those quirks of natlangs. {ta poi nakni} is fine.

Yes. If you absolutely have to, you can specify gender/sex through various techniques, but forcing the speaker to always do so
No. I'm not forcing. Everything must be optional. 
would imply that sex/gender is of primary importance, which in turn would potentially support a sexist world-view. If one is not able to talk about something without knowing its gender or sex, then that is a definite short-coming of the language.
Sure. 


Unsettled issues.
Some languages have "abstractions" in their lowest level of animacy hierarchy.
Lojban is pretty strict when dealing with objects and abstractions. The issue with the scale "su'unai - su'u" that one might imagine remains unsettled.

You either have a NU or you don't. What scale are you imagining?
Look at the list of the levels of animacy in Navajo.

The list is: humans/lightning → infants/big animals → med-size animals → small animals → insects → natural forces → inanimate objects/plants → abstractions

You can sort all of them into either NU or not-NU. All the animals including humans are objects. Objects are obviously objects too. Natural forces can be either, depending on how you express them. lo lindi  vs lo nu lindi. And finally, abstractions are clearly NU.
Clearly. But notice that Lojban abruptly splits everything into either NU or not-NU whereas Navajo has a scale.
We should pay more attention to this fact.
I love Lojban for not mixing up objects and abstractions.
But some languages seem to be even more precise.
It's all about expressive power.

Jorge Llambías

unread,
Aug 4, 2012, 12:22:17 PM8/4/12
to loj...@googlegroups.com
On Sat, Aug 4, 2012 at 9:07 AM, Gleki Arxokuna
<gleki.is...@gmail.com> wrote:
>
> By the way how would you express "x1 is animate of level x2"?

You can use "va'e" for scales of any sort. So if we use "se ruxse'i"
for "animate", you could say: "ko'a ci va'e lo ka se ruxse'i": "ko'a
is level 3 on the scale of animacy".

Alternatively, you can use "klani": "ko'a klani li ci lo ka se
ruxse'i", which would then allow us to make the lujvo "ruxsezlai" for
"x1 is animate of level x2", maybe even just "ruxlai" or "genra
ruxlai" if it's not clear you are talking about grammar.

mu'o mi'e xorxes

Veijo Vilva

unread,
Aug 4, 2012, 12:59:06 PM8/4/12
to loj...@googlegroups.com
On 4 August 2012 14:44, Gleki Arxokuna <gleki.is...@gmail.com> wrote:

Honestly, we both speak European languages. That's why our opinion means nothing as we can't remove our cultural bias.
We need someone from another culture (like Navajo speaker).

Speaking about me as another participant in this conversation, my mother tongue is Finnish, which isn't an Indo-European language, and I used to know Japanese pretty well, another non-European language. These two languages differ greatly from each other, even ignoring the writing systems, but in principle English is for me just as alien and different as Japanese. Finnish culture has, of course, during the last few centuries been greatly affected by the general European culture and lately especially by Anglo-American culture, but this hasn't too much affected the basic structure of the language. Personally I'm reasonably fluent (at least in reading) in three Indo-European languages, i.e., Swedish, German and English, and have also studied Russian and French, which gives me a rather good grasp of the differences between my mother tongue and a variety of European languages, between the fundamental distinctions these languages make. Finnish doesn't make a gender distinction, definite/indefinite distinction is rather tricky to express in a proper, native way, and part/total distinction is perhaps the most essential and for foreigners the hardest to learn distinction affecting even our approach to basic quantification. The old, aboriginal Finnish culture was quite animistic, but this didn't affect the set of pronouns available, which, however, didn't limit our ability to express everything deemed necessary within that culture.

No language can in a natural way express every fine shade of meaning some other language can, not at the level of grammar, not at the level of native vocabulary, not at the level of inherited cultural understanding, not at a level every user of the language can fully grasp. Every translation from one language to another has to bridge a gap, sometimes a narrow one, sometimes an almost impossibly wide one. Considered against the overall background of natural languages, a constructed language always represents, of needs, a set of more or less arbitrary choices made by the developers, choices giving the language its distinct flavor. Most constructed languages have tried to be as natural as possible, just trying to remove some perceived "weaknesses" of some arbitrary set of natural languages or trying to be as easy to learn as possible. OTOH,  Loglan/Lojban was intentionally designed to be in a way as different as possible, and thus more weight has been given to being able to express naturally some things and ideas awkward to express either in the natural languages in general or in the Indo-European languages familiar to the developers in particular -- the keyword here is "some" as the total set of possible features is way too large. Lojban, like every language, will always have limitations, and more important than being able to express things the way they are expressed in other languages is to develop a native idiomatic way of expression using the existing machinery with all its possibilities and native implications. The time for any extension comes when it is felt that something important enough and frequent enough cannot be expressed simply enough in the world that is then.

mu'o mi'e veion

--

MorphemeAddict

unread,
Aug 4, 2012, 1:10:05 PM8/4/12
to loj...@googlegroups.com
What sort of lack do the fluent speakers of Lojban feel regardng this issue? If they don't have a problem, then there is no problem. Problem solved.
 
stevo

--
You received this message because you are subscribed to the Google Groups "lojban" group.
To view this discussion on the web visit https://groups.google.com/d/msg/lojban/-/xCz0FxKdifoJ.

To post to this group, send email to loj...@googlegroups.com.
To unsubscribe from this group, send email to lojban+un...@googlegroups.com.

Jonathan Jones

unread,
Aug 4, 2012, 3:23:00 PM8/4/12
to loj...@googlegroups.com
On Sat, Aug 4, 2012 at 5:44 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
On Saturday, August 4, 2012 12:01:16 PM UTC+4, aionys wrote:
On Sat, Aug 4, 2012 at 1:48 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
It's all based on the analysis of existing languages.
Sure, English example is not fine.
But Navajo for instance has the following hierarchy.

humans/lightning → infants/big animals → med-size animals → small animals → insects → natural forces → inanimate objects/plants → abstractions

Therefore, lightning is somewhat za'e su'unai.

It doesn't matter what language you use for examples. My position is still the same. The less arbitrary distinctions there are, the better. And merely by the simple fact that every language splits things differently, (although I'll grant that language families tend to be tcesimsa if not mintu,) is enough evidence for me at least that it is arbitrary.

I know that is human nature to classify things. We do it all the time. But not all classifications are useful or necessary.

We have semantically unambiguous ways to refer to things, whether they be people, animals, objects, or what-have-you. We DON'T just have ra/ri/ru, ta/ti/tu, and va/vi/vu. We have the KOhA and VOhA, and the BY, which are more than enough in nearly any circumstance.

Honestly, how would you rather refer to the mirror in your example? {ta xoinai}, or {my.}?
Honestly, we both speak European languages. That's why our opinion means nothing as we can't remove our cultural bias.
We need someone from another culture (like Navajo speaker).

Well, I would rather refer to it with {my.}, and it has nothing to do with the fact that I speak English, and everything to do with the fact that it's 2 syllables and 6 characters shorter. Like most humans, I'm lazy in my speech, and I prefer shorter.
 
To view this discussion on the web visit https://groups.google.com/d/msg/lojban/-/SZLAWPmIc_8J.

To post to this group, send email to loj...@googlegroups.com.
To unsubscribe from this group, send email to lojban+un...@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/lojban?hl=en.

selpa'i

unread,
Aug 4, 2012, 3:28:36 PM8/4/12
to loj...@googlegroups.com
coi ro do
^:i !ji /glu (to mi rinsa toi)

First of all, sorry if this doesn't belong here, but I'm trying to find
more info about gua\spi, and I know for a fact that at least some jbopre
had some involvement with it. (And it's a sister language of lojban)
I read the "Introduction to Gua\spi" and the "Gua\spi Reference Manual",
but I felt that those texts were not enough to really get to a point
where I could even construct a single bridi. In fact, had I not had a
lojbanic background, I would probably not have understood any of it.
I read that the inventor James Carter is not working on the language
anymore, but maybe he'd be interested in getting back into it, if there
was an audience of interested people.
I know that John Cowan came to Lojban *from* Gua\spi, so please, could
you share some of your experiences with the langauge and maybe even
point me in some direction? (I really hope the language is hopelessly
dead, like another conlang I once wanted to learn, only to find out that
the author had abandonned it.)
If anybody knows of any other learning materials for this langauge, and
a more easily searchable dictionary, I'd be very grateful to be made
aware of them.

ki'e mu'o mi'e la selpa'i

Jonathan Jones

unread,
Aug 4, 2012, 3:35:15 PM8/4/12
to loj...@googlegroups.com

I disagree, because unlike your animacy tirade, UI are obviously useful. cuvyjbopre do not immediately attack any proposed change out of some love for "Lojban as it is now". Even the most fundamnetalist of us knows that Lojban is not perfect, probably never will be, and does need improvements and useful changes. The question we ask ourselves for EVERY proposal is, "Is this useful? Will it make my beloved language better? Do we need this, or even something like this?"

And in my opinion, the answer to those questions with regard to making a new word to distinguish between (in)animate "levels" is a profound NO.
 

An arbitrary scale.
 
We should pay more attention to this fact.

Why?
 
I love Lojban for not mixing up objects and abstractions.
But some languages seem to be even more precise.
It's all about expressive power.

mu'o mi'e la selpa'i



-- 
pilno zo le xu .i lo dei bangu cu se cmene zo lojbo .e nai zo lejbo

--
You received this message because you are subscribed to the Google Groups "lojban" group.
To view this discussion on the web visit https://groups.google.com/d/msg/lojban/-/cQatnuYy5DYJ.

To post to this group, send email to loj...@googlegroups.com.
To unsubscribe from this group, send email to lojban+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban?hl=en.

selpa'i

unread,
Aug 4, 2012, 3:51:02 PM8/4/12
to loj...@googlegroups.com
.oi se'i

Am 04.08.2012 21:28, schrieb selpa'i:
coi ro do
^:i !ji /glu (to mi rinsa toi)

First of all, sorry if this doesn't belong here, but I'm trying to find more info about gua\spi, and I know for a fact that at least some jbopre had some involvement with it. (And it's a sister language of lojban)
I read the "Introduction to Gua\spi" and the "Gua\spi Reference Manual", but I felt that those texts were not enough to really get to a point where I could even construct a single bridi. In fact, had I not had a lojbanic background, I would probably not have understood any of it.
I read that the inventor James Carter is not working on the language anymore, but maybe he'd be interested in getting back into it, if there was an audience of interested people.
I know that John Cowan came to Lojban *from* Gua\spi, so please, could you share some of your experiences with the langauge and maybe even point me in some direction? (I really hope the language is *not* hopelessly dead, like another conlang I once wanted to learn, only to find out that the author had abandonned it.)
If anybody knows of any other learning materials for this langauge, and a more easily searchable dictionary, I'd be very grateful to be made aware of them.

ki'e mu'o mi'e la selpa'i

John E Clifford

unread,
Aug 4, 2012, 7:20:29 PM8/4/12
to loj...@googlegroups.com
No conlang is dead if someone is still reading up on it, though it seems like a couple decades since I last heard of this one.  But Carter was still on some lists a couple of years ago, so keep looking.  (I confess I never looked at it, since it was politically incorrect when it first started out and I forgot all about it later.)


From: selpa'i <sel...@gmx.de>
To: loj...@googlegroups.com
Sent: Saturday, August 4, 2012 2:51 PM
Subject: Re: [lojban] gua\spi

--
You received this message because you are subscribed to the Google Groups "lojban" group.

John E Clifford

unread,
Aug 4, 2012, 7:25:20 PM8/4/12
to loj...@googlegroups.com
The last address I have for him (2009) is James F. Carter          Voice 310 825 2897    FAX 310 206 6673
UCLA-Mathnet;  6115 MSA; 405 Hilgard Ave.; Los Angeles, CA, USA 90095-1555
Email: ji...@math.ucla.edu  http://www.math.ucla.edu/~jimc (q.v. for PGP key)


From: John E Clifford <kali9...@yahoo.com>
To: "loj...@googlegroups.com" <loj...@googlegroups.com>
Sent: Saturday, August 4, 2012 6:20 PM

MorphemeAddict

unread,
Aug 4, 2012, 7:40:43 PM8/4/12
to loj...@googlegroups.com
I find that guaspi already has a lot of features (tones for grammar, very short words) that I'd like to see in LoCCan3, but the orthography, especially for those tones, is so hard for me to grasp that I never pursued it. 

stevo

--
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to loj...@googlegroups.com.
To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.

Jacob Errington

unread,
Aug 4, 2012, 11:45:24 PM8/4/12
to loj...@googlegroups.com
It probably wouldn't be complicated at all to create an alternate orthography using diacritics similar to those used in pinyin. These symbols were probably used by Carter for technological reasons at the time (this language is incredibly old at this point).
I'd personally be interested in learning gua\spi, and should it be dead, working on it maybe. Perhaps we can use it to test some ideas for loccan3 ?

To unsubscribe from this group, send email to lojban+un...@googlegroups.com.

Jim Carter

unread,
Aug 5, 2012, 1:11:08 AM8/5/12
to loj...@googlegroups.com
On Sat, 4 Aug 2012, selpa'i wrote:

> First of all, sorry if this doesn't belong here, but I'm trying to find more
> info about gua\spi, and I know for a fact that at least some jbopre had some
> involvement with it. (And it's a sister language of lojban)

Well, maybe "bastard offspring" might be a better term.

> I read the "Introduction to Gua\spi" and the "Gua\spi Reference Manual", but
> I felt that those texts were not enough to really get to a point where I
> could even construct a single bridi. In fact, had I not had a lojbanic
> background, I would probably not have understood any of it.
> I read that the inventor James Carter is not working on the language anymore,
> but maybe he'd be interested in getting back into it, if there was an
> audience of interested people.

Unfortunately I've had to make some choices about what to commit time
to, and other issues pushed ahead of gua\spi.

> I know that John Cowan came to Lojban *from* Gua\spi,

Well, actually he was very active in Lojban at the beginning, and
commented occasionally on gua\spi.

> so please, could you
> share some of your experiences with the langauge and maybe even point me in
> some direction? (I really hope the language is hopelessly dead, like another
> conlang I once wanted to learn, only to find out that the author had
> abandonned it.)

I hope it could be of some use to the Lojban community at this point;
I hope it's not hopelessly dead.

> If anybody knows of any other learning materials for this langauge, and a
> more easily searchable dictionary, I'd be very grateful to be made aware of
> them.

Sorry about the vocabulary lookup CGI. The database package had a major
version jump and the script was not finding anything. I rebuilt the
database and that brought it back to life. I've also learned a whole
lot about writing CGI scripts since 2001, and I gave it a little love.

For people not familiar with gua\spi, here are some major points where
it could be interesting to the lojban community:

* The grammar is radically simple. Of course, replacing the grammar is
out of the question for lojban, but possibly someone could spot small
tweaks, or shifting grammar into semantics, that could be helpful for
lojban's problem areas.

* I followed John Clifford's lead in defining what a sentence actually
means, and I did my best to specify how the words join to create
meanings, rather than relying on pragmatics and wishful thinking.

* Virtually all compound words can be transformed algorithmically into
phrases containing only primitive words (lujvo), and the connection
between arguments in such phrases follows regular patterns rather than
being handcrafted for each lujvo. This feature was a badly needed
(but not politically correct) improvement in Old Loglan, and I think
it would still be attractive in lojban.

* The vocabulary words are grouped in "thesaurus categories", an aid
to learning, and common argument patterns in these categories, already
present in Old Loglan, were strengthened, also an aid to learning.

* One of my major goals for gua\spi was to bring some sanity to
mathematical expressions (MEX) which were being designed at that time.
You might say that MEX grammar was tacked onto the core of Lojban,
whereas MEX *is* the core of gua\spi, and I didn't have to tack on
anything to express non-MEX meanings. Particularly, MEX has to
express a meaning because the semantic specifications say it does,
not because words are somehow similar to the conventional way of
writing mathematics, and the same clarity of meaning carries over to
everyday life. One could consider how the semantics and philosopy
of MEX in gua\spi could contribute to lojban, though of course not
the grammar and vocabulary.

Jim Carter

unread,
Aug 5, 2012, 1:17:52 AM8/5/12
to loj...@googlegroups.com
On Sat, 4 Aug 2012, Jacob Errington wrote:

> It probably wouldn't be complicated at all to create an alternate
> orthography using diacritics similar to those used in pinyin. These symbols
> were probably used by Carter for technological reasons at the time (this
> language is incredibly old at this point).

Yes, support for ISO-8859-1 was just beginning and Unicode was not
available. In TeX I could make overstruck accents happen, but this was
impossible in serial media, so I just did everything with -/|\^=.

Gleki Arxokuna

unread,
Aug 5, 2012, 4:32:49 AM8/5/12
to loj...@googlegroups.com
Pardon, but how is it possible? Lujvo with internal predicate structure, no? 

Jonathan Jones

unread,
Aug 5, 2012, 5:23:25 AM8/5/12
to loj...@googlegroups.com
On Sun, Aug 5, 2012 at 2:32 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
On Sunday, August 5, 2012 9:11:08 AM UTC+4, JimCarter wrote:
* Virtually all compound words can be transformed algorithmically into
phrases containing only primitive words (lujvo), and the connection
between arguments in such phrases follows regular patterns rather than
being handcrafted for each lujvo.  This feature was a badly needed
(but not politically correct) improvement in Old Loglan, and I think
it would still be attractive in lojban.
Pardon, but how is it possible? Lujvo with internal predicate structure, no?

I would assume in similar manner to how we make "regular" lujvo.

For instance, any {tcebroda} means {ko'a mutce lo ka broda}.

Jacob Errington

unread,
Aug 5, 2012, 8:31:19 AM8/5/12
to loj...@googlegroups.com
On 5 August 2012 05:23, Jonathan Jones <eye...@gmail.com> wrote:
On Sun, Aug 5, 2012 at 2:32 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
On Sunday, August 5, 2012 9:11:08 AM UTC+4, JimCarter wrote:
* Virtually all compound words can be transformed algorithmically into
phrases containing only primitive words (lujvo), and the connection
between arguments in such phrases follows regular patterns rather than
being handcrafted for each lujvo.  This feature was a badly needed
(but not politically correct) improvement in Old Loglan, and I think
it would still be attractive in lojban.
Pardon, but how is it possible? Lujvo with internal predicate structure, no?

I would assume in similar manner to how we make "regular" lujvo.

For instance, any {tcebroda} means {ko'a mutce lo ka broda}.


Picking nits (in case nintadni ever read this) but I think you mean "{rodytce} means {ko'a mutce lo ka broda}".
{tcero'e} is naljvajvo for about 80% of possible brode. The only meaningful counterexample is {tcemau} (and its counterpart {tceme'a}) which gets taken over by the usual structure of -mau and -me'a lujvo: {ko'a zmadu ko'e lo ka ce'u mutce ko'i ko'o kei ko'u}. Ah, another one would be {tcemlu}.

mu'o mi'e la tsani
 
--
mu'o mi'e .aionys.

.i.e'ucai ko cmima lo pilno be denpa bu .i doi.luk. mi patfu do zo'o
(Come to the Dot Side! Luke, I am your father. :D )

--
You received this message because you are subscribed to the Google Groups "lojban" group.

selpa'i

unread,
Aug 5, 2012, 10:09:28 AM8/5/12
to loj...@googlegroups.com
Am 05.08.2012 07:11, schrieb Jim Carter:
> On Sat, 4 Aug 2012, selpa'i wrote:
>
>> I read that the inventor James Carter is not working on the language
>> anymore, but maybe he'd be interested in getting back into it, if
>> there was an audience of interested people.
>
> Unfortunately I've had to make some choices about what to commit time
> to, and other issues pushed ahead of gua\spi.

How complete is gua\spi? If it's as usable as Lojban currently is, I'd
say it's very close to complete. So if you have some other texts about
the language, more grammar explanations etc, it would be great to be
able to read them.
Also, you mention that you translated quite a bit of text into gua\spi.
Could you upload that text somewhere? I would like to see some actual
usage of the language.

> * One of my major goals for gua\spi was to bring some sanity to
> mathematical expressions (MEX) which were being designed at that time.
> You might say that MEX grammar was tacked onto the core of Lojban,
> whereas MEX *is* the core of gua\spi,

I don't have a problem with that, being that I'm a big proponent of
mekso in lojban.

So bottom line, how realistic would you consider it to be that people
could start learning gua\spi? Would you be able to support the learners
if some bigger problems arose? Once more than one person (you) is at a
good level in the language, the burden gets distributed and eventually,
the language will get a life of its own.

In hope, mu'o mi'e la selpa'i

Jonathan Jones

unread,
Aug 5, 2012, 2:25:15 PM8/5/12
to loj...@googlegroups.com
On Sun, Aug 5, 2012 at 6:31 AM, Jacob Errington <nict...@gmail.com> wrote:
On 5 August 2012 05:23, Jonathan Jones <eye...@gmail.com> wrote:
On Sun, Aug 5, 2012 at 2:32 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
On Sunday, August 5, 2012 9:11:08 AM UTC+4, JimCarter wrote:
* Virtually all compound words can be transformed algorithmically into
phrases containing only primitive words (lujvo), and the connection
between arguments in such phrases follows regular patterns rather than
being handcrafted for each lujvo.  This feature was a badly needed
(but not politically correct) improvement in Old Loglan, and I think
it would still be attractive in lojban.
Pardon, but how is it possible? Lujvo with internal predicate structure, no?

I would assume in similar manner to how we make "regular" lujvo.

For instance, any {tcebroda} means {ko'a mutce lo ka broda}.


Picking nits (in case nintadni ever read this) but I think you mean "{rodytce} means {ko'a mutce lo ka broda}".
{tcero'e} is naljvajvo for about 80% of possible brode. The only meaningful counterexample is {tcemau} (and its counterpart {tceme'a}) which gets taken over by the usual structure of -mau and -me'a lujvo: {ko'a zmadu ko'e lo ka ce'u mutce ko'i ko'o kei ko'u}. Ah, another one would be {tcemlu}.

Right. I always get that backwards.
 

mu'o mi'e la tsani
 
--
mu'o mi'e .aionys.

.i.e'ucai ko cmima lo pilno be denpa bu .i doi.luk. mi patfu do zo'o
(Come to the Dot Side! Luke, I am your father. :D )


--
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to loj...@googlegroups.com.
To unsubscribe from this group, send email to lojban+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban?hl=en.

--
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to loj...@googlegroups.com.
To unsubscribe from this group, send email to lojban+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban?hl=en.

selpa'i

unread,
Aug 5, 2012, 7:25:02 PM8/5/12
to loj...@googlegroups.com
Am 04.08.2012 21:28, schrieb selpa'i:
> I read the "Introduction to Gua\spi" and the "Gua\spi Reference
> Manual", but I felt that those texts were not enough to really get to
> a point where I could even construct a single bridi. In fact, had I
> not had a lojbanic background, I would probably not have understood
> any of it.

I'm reading the reference manual again, taking my time this time around,
and with the additional motivation I have now, I'm beginning to
understand more of what I'm reading. Is there really nobody else who has
gone through the process of learning the language (to a point where they
could produce some simple texts maybe)?

Gleki Arxokuna

unread,
Aug 7, 2012, 1:24:02 PM8/7/12
to loj...@googlegroups.com


On Sunday, August 5, 2012 4:31:19 PM UTC+4, tsani wrote:
On 5 August 2012 05:23, Jonathan Jones <eye...@gmail.com> wrote:
On Sun, Aug 5, 2012 at 2:32 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
On Sunday, August 5, 2012 9:11:08 AM UTC+4, JimCarter wrote:
* Virtually all compound words can be transformed algorithmically into
phrases containing only primitive words (lujvo), and the connection
between arguments in such phrases follows regular patterns rather than
being handcrafted for each lujvo.  This feature was a badly needed
(but not politically correct) improvement in Old Loglan, and I think
it would still be attractive in lojban.
Pardon, but how is it possible? Lujvo with internal predicate structure, no?

I would assume in similar manner to how we make "regular" lujvo.

For instance, any {tcebroda} means {ko'a mutce lo ka broda}.
That makes sense. Sorry it's too late to make lujvo in Lojban "regular" only.



Picking nits (in case nintadni ever read this) but I think you mean "{rodytce} means {ko'a mutce lo ka broda}".
{tcero'e} is naljvajvo for about 80% of possible brode. The only meaningful counterexample is {tcemau} (and its counterpart {tceme'a}) which gets taken over by the usual structure of -mau and -me'a lujvo: {ko'a zmadu ko'e lo ka ce'u mutce ko'i ko'o kei ko'u}. Ah, another one would be {tcemlu}.

mu'o mi'e la tsani
 
--
mu'o mi'e .aionys.

.i.e'ucai ko cmima lo pilno be denpa bu .i doi.luk. mi patfu do zo'o
(Come to the Dot Side! Luke, I am your father. :D )


--
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to loj...@googlegroups.com.
To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.

Jim Carter

unread,
Aug 8, 2012, 1:56:54 PM8/8/12
to loj...@googlegroups.com
On Sun, 5 Aug 2012, selpa'i wrote:

> How complete is gua\spi? If it's as usable as Lojban currently is, I'd say
> it's very close to complete.

I can see this discussion flying off on multiple tangents: what does it
mean for a language to be complete? I'd say that there is an equivalence
class of languages that can express predicate logic, including the natural
languages I know, Lojban/Loglan, and gua\spi. Gua\spi's vocabulary is a
slight superset of Lojban (I may have thrown out one or two words, and I
did add maybe 5% more words), so in that sense gua\spi covers just about
the same semantic space that Lojban does.

(Tangent: hard-edged Aristotlean logic is not typical of everyday usage,
and "predicate logic" should be interpreted with fuzzy values of the
predicates.)

"Usable" refers to how easily and accurately the speakers can express what
they want (versus whether it can be done even if kludges are needed). The
vendor hypes the product as being usable.

> So if you have some other texts about the language, more grammar
> explanations etc, it would be great to be able to read them.

Unfortunately, the material you've already seen in
http://www.math.ucla.edu/~jimc/guaspi/index.html
is all that I've done. Sorry.

> Also, you mention that you translated quite a bit of text into gua\spi.
> Could you upload that text somewhere? I would like to see some actual
> usage of the language.

Actually I said that I'd translated about 20,000 words into Old Loglan.
Sorry again.

> So bottom line, how realistic would you consider it to be that people could
> start learning gua\spi? Would you be able to support the learners if some
> bigger problems arose? Once more than one person (you) is at a good level in
> the language, the burden gets distributed and eventually, the language will
> get a life of its own.

I'm ambivalent here -- I would be happy to support someone (you) working on
gua\spi, but I need to warn up front that it could not be high-bandwidth
support, e.g. like what Robin Powell does for Lojban. May I squeeze in a
support tidbit (which could be of interest in Lojban also)? From the
reference manual, vocabulary chapter intro: "Frequently I have thought that
some form or meaning required a new primitive word, or even a change in the
grammar, but it has turned out that existing words were more than adequate
if creatively used."

Occasionally someone stumbles across gua\spi and sends me mail about it,
but I think you (selpa'i) are the first one to dig into it deeply.

I think the major value of gua\spi for the Lojban community is in seeing
how a similar language does things differently, particularly in the areas
of compound words and in MEX.

In about an hour I'm going to be on an airplane to Prague, so I may have a
little trouble holding up my end of this thread, but I'll try.

James F. Carter Voice 310 825 2897 FAX 310 206 6673
UCLA-Mathnet; 6115 MSA; 520 Portola Plaza; Los Angeles, CA, USA 90095-1555

selpa'i

unread,
Aug 8, 2012, 7:39:02 PM8/8/12
to loj...@googlegroups.com
Am 08.08.2012 19:56, schrieb Jim Carter:
> On Sun, 5 Aug 2012, selpa'i wrote:
>
>> How complete is gua\spi? If it's as usable as Lojban currently is,
>> I'd say it's very close to complete.
>
> I can see this discussion flying off on multiple tangents: what does
> it mean for a language to be complete? I'd say that there is an
> equivalence class of languages that can express predicate logic,
> including the natural languages I know, Lojban/Loglan, and gua\spi.
> Gua\spi's vocabulary is a slight superset of Lojban (I may have thrown
> out one or two words, and I did add maybe 5% more words), so in that
> sense gua\spi covers just about the same semantic space that Lojban does.

I realized this reading through the reference grammar again. It seems
like it can do everything (not an educated opinion, I'm still a
beginner) that Lojban can do, and I consider Lojban complete enough to
be used in daily life and in literature.

> "Usable" refers to how easily and accurately the speakers can express
> what they want (versus whether it can be done even if kludges are
> needed). The vendor hypes the product as being usable.

True, but I didn't get the impression you were hyping the product more
than anybody else would hype their personal conlang.

>
>> So if you have some other texts about the language, more grammar
>> explanations etc, it would be great to be able to read them.
>
> Unfortunately, the material you've already seen in
> http://www.math.ucla.edu/~jimc/guaspi/index.html
> is all that I've done. Sorry.

No problem. I'm still thankful for the documents on that page. I just
wish that they went at a slower pace.

>
>> Also, you mention that you translated quite a bit of text into
>> gua\spi. Could you upload that text somewhere? I would like to see
>> some actual usage of the language.
>
> Actually I said that I'd translated about 20,000 words into Old Loglan.
> Sorry again.

Ah, no problem again. Then I guess I'll have to write my own texts
instead. :)

>
>> So bottom line, how realistic would you consider it to be that people
>> could start learning gua\spi? Would you be able to support the
>> learners if some bigger problems arose? Once more than one person
>> (you) is at a good level in the language, the burden gets distributed
>> and eventually, the language will get a life of its own.
>
> I'm ambivalent here -- I would be happy to support someone (you)
> working on gua\spi, but I need to warn up front that it could not be
> high-bandwidth support, e.g. like what Robin Powell does for Lojban.
> May I squeeze in a support tidbit (which could be of interest in
> Lojban also)?

Any help is appreciated. I'm currently wrestling with subordinate
clauses in gua\spi, and if I can't figure them out by myself, I will
have to ask you for clarification. I expect that after subordinate
clauses, the rest is straightforward as it's very similar to Lojban
(numbers, articles, pronouns etc), so I hope to achieve some level of
independence after this stumbling block. (I'm also still struggling with
the 5th tone, the others I can easily produce, but I'm making progress).

> From the reference manual, vocabulary chapter intro: "Frequently I
> have thought that some form or meaning required a new primitive word,
> or even a change in the grammar, but it has turned out that existing
> words were more than adequate if creatively used."

I believe that 100%. I won't be trying to invent anything new until I'm
fluent *fingers crossed*.

>
> Occasionally someone stumbles across gua\spi and sends me mail about
> it, but I think you (selpa'i) are the first one to dig into it deeply.

Good to know. It motivates me to be the first one to try this.

>
> I think the major value of gua\spi for the Lojban community is in
> seeing how a similar language does things differently, particularly in
> the areas of compound words and in MEX.

Yes, I'm currently discovering all the differences and similarities, and
it's really interesting. I'm not sure how much of it can be used in
Lojban, but I'm just doing this for its own sake, because I am
interested in gua\spi itself.

> In about an hour I'm going to be on an airplane to Prague, so I may
> have a little trouble holding up my end of this thread, but I'll try.

Okay. This thread doesn't seem to be going at a fast pace, anyway. If I
can't figure out the subordinates, I'll turn to you, if that's okay.
Though, now that I think about it, I'm not sure I can figure it out
alone. I've tried for two days and it's still confusing.
For example, in, say:

^:i \ji /crw \kseo ^ve tum \qnou [\ji]
"I eat the cheese with [my] hands."

I can't tell whether the ve-clause is really a clause or if it's just a
sort of ad-hoc case link which gets filled by "qnou". I was relatively
sure it's the former. The third case of "tum" gets moved to the first
position automatically, making "qnou" go into the second case.
But ^ve tum \qnou == \vo X1 \zu tum \qnou, so now I think it's the
latter. But this would mean that everything gets reduced to an infinitive...

But I'm afraid that if I start asking questions now, I will remember
more and more things I would like to have clarified, so maybe we should
do this off-list.

Gleki Arxokuna

unread,
Aug 9, 2012, 1:31:02 AM8/9/12
to loj...@googlegroups.com
Two ideas.
1. Why not change tone notation to pinyin style with numbers? (May be it'll be easier for future Chinese gua\spi users?)
2. If not much difference between Lojban and gua\spi may be create a two-way converter Lojban <=> gua\spi? If such converter exists gua\spi can  quickly accumulate the same corpus as Lojban.

MorphemeAddict

unread,
Aug 9, 2012, 2:49:49 AM8/9/12
to loj...@googlegroups.com
On Thu, Aug 9, 2012 at 1:31 AM, Gleki Arxokuna <gleki.is...@gmail.com> wrote:
Two ideas.
1. Why not change tone notation to pinyin style with numbers? (May be it'll be easier for future Chinese gua\spi users?)

Pinyin uses accent marks far more often than numerals. (Wade-Giles uses numerals.) And Chinese almost never use pinyin, except on road signs, and then without tone marks. 

stevo
 
--
You received this message because you are subscribed to the Google Groups "lojban" group.
To view this discussion on the web visit https://groups.google.com/d/msg/lojban/-/lKieaxUS6m0J.

Gleki Arxokuna

unread,
Aug 9, 2012, 3:39:12 AM8/9/12
to loj...@googlegroups.com
I meant that if tones are hard for se ke ropno bangu then they might be easier for se ke jungo bangu. After all Lojban is not popular in China for some reason.

And gua\spi input might use chinese input system. It's fine if your application can convert numerals to accent marks.

selpa'i

unread,
Aug 9, 2012, 9:08:23 AM8/9/12
to loj...@googlegroups.com
Am 09.08.2012 07:31, schrieb Gleki Arxokuna:
> Two ideas.
> 1. Why not change tone notation to pinyin style with numbers? (May be
> it'll be easier for future Chinese gua\spi users?)

Numbers are not better than the current tone marks, but it's obviously
possible to use pinyin-like diacritics: :î qnǔ qò jan tára jún ksèo zěy jù

> 2. If not much difference between Lojban and gua\spi may be create a
> two-way converter Lojban <=> gua\spi? If such converter exists gua\spi
> can quickly accumulate the same corpus as Lojban.

That's an interesting idea, but who shall write such a converter? :)
It's probably possible to make a basic converter, but it won't be able
to carry over a particular writing style. gua\spi excels at different
things than Lojban, so a direct translation won't yield an optimal result.

Gleki Arxokuna

unread,
Aug 9, 2012, 10:03:31 AM8/9/12
to loj...@googlegroups.com


On Thursday, August 9, 2012 5:08:23 PM UTC+4, selpa'i wrote:
Am 09.08.2012 07:31, schrieb Gleki Arxokuna:
> Two ideas.
> 1. Why not change tone notation to pinyin style with numbers? (May be
> it'll be easier for future Chinese gua\spi users?)

Numbers are not better than the current tone marks, but it's obviously
possible to use pinyin-like diacritics: :î qnǔ qò jan tára jún ksèo zěy jù

> 2. If not much difference between Lojban and gua\spi may be create a
> two-way converter Lojban <=> gua\spi? If such converter exists gua\spi
> can quickly accumulate the same corpus as Lojban.

That's an interesting idea, but who shall write such a converter? :)
It's probably possible to make a basic converter, but it won't be able
to carry over a particular writing style. gua\spi excels at different
things than Lojban, so a direct translation won't yield an optimal result.
Lojban is syntactically unambiguous. 
If gua\spi is unambiguous too and their gismu coincide in semantics then there are no problems.
Except technical :)
What is more important that almost every conlanger views other conlangs as rival projects.
It shouldn't be the case in Lojbanistan.
No problem if we have numerous LoCCans. Let's see what will be accepted more widely.
This is another level of backward compatibility of LoCCans: through such converter.

selpa'i

unread,
Aug 9, 2012, 4:03:30 PM8/9/12
to loj...@googlegroups.com
Am 09.08.2012 01:39, schrieb selpa'i:
> But I'm afraid that if I start asking questions now, I will remember
> more and more things I would like to have clarified, so maybe we
> should do this off-list.

Would you be able to come on IRC (@ jimc)? It would probably be more
time-efficient to talk about things in real-time. Maybe we can even get
everything sorted out in one sitting.

selpa'i

unread,
Aug 15, 2012, 8:34:52 PM8/15/12
to loj...@googlegroups.com
Am 09.08.2012 22:03, schrieb selpa'i:
> Am 09.08.2012 01:39, schrieb selpa'i:
>> But I'm afraid that if I start asking questions now, I will remember
>> more and more things I would like to have clarified, so maybe we
>> should do this off-list.
>
> Would you be able to come on IRC (@ jimc)? It would probably be more
> time-efficient to talk about things in real-time. Maybe we can even
> get everything sorted out in one sitting.
>
> mu'o mi'e la selpa'i
>

You've been gone for quite long now, I hope you are fine. I just wanted
to inform you that all my major confusions have disappeared and that I
now have a rather good grasp of this language and I'm still actively
working on it. I also managed to motivate one other person to pick it
up. There are still some minor questions I would really like to discuss
with you (to make sure my understanding is correct, for example), but
all in all, I can now manage on my own. The reference grammar wasn't so
bad afterall, even though I had to read it dozens of times and tear my
hair out over it. =)
Anyway, I hope everything's okay

Best wishes

selpa'i

unread,
Jan 11, 2013, 6:30:34 PM1/11/13
to loj...@googlegroups.com
I'm giving this thread a bump for two reasons.

1. If there are any lurkers on this list that have some interest in
Gua\spi but don't know where to start, you are more than welcome to join
our humble Gua\spi mailing list and get help or inspiration there.

https://groups.google.com/forum/#!forum/guaspi

2. If anyone here knows Jim Carter personally, I would be glad to know
what happened to him. We last heard from him in August, and he is not
responding to our emails. We are worrying something might have happened
to him. Any pointers are appreciated.

la gleki

unread,
Jan 12, 2013, 12:51:27 AM1/12/13
to loj...@googlegroups.com


On Saturday, August 4, 2012 10:35:59 AM UTC+4, la gleki wrote:
Suggestion for a new animacy marker in Lojban.

Many if not most languages divide all predicates into levels of animacy.
English, for instance, has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

This allows quickly determine agents of most actions.
Example:
The woman was looking at the mirror. It was ugly.
Let's try it in Lojban.
{lo ninmu pu ca'o catlu lo minra .i ta pu tolmelbi}
No, too ambiguous. And I opine that counting two sumti back in order to use {ra} is much trickier for human brain than just understanding semantic roles of sumti.
Therefore, I suggest introducing a new marker reflecting animacy of any object. I'll use {xoi} which currently bears no official meaning.

xoi - marks preceding construct as animate
xoinai - marks preceding construct as inanimate

{lo ninmu pu ca'o catlu lo minra. i ta xoinai pu tolmelbi}

However, some languages have more levels of animacy.
The father was looking at his son. He was beautiful.
{lo patfu pu catlu lo bersa .i ta xoixime'i pu melbi}
The author of this sentence probably thinks that children are less animate than grown-ups. 
So we can build a scale ranging from most animate objects to inanimate.
It's only the speaker who decides what level of animacy this or that object has.

Gender-specific pronouns.
You might argue why not add more specific markers reflecting for instance the gender of the object described.
Let's repeat once again.

English has at least two levels. These are the pronouns for them
1. Animate. He/she
2. Inanimate. It

In other words, English has two pronouns expressing sex but only one pronoun expressing inanimate objects.
There might be languages that split inanimate levels into other specific classes (furniture, houses, weapons).
Therefore, it would be stupid to try to import all those quirks of natlangs. {ta poi nakni} is fine.

Unsettled issues.
Some languages have "abstractions" in their lowest level of animacy hierarchy.
Lojban is pretty strict when dealing with objects and abstractions. The issue with the scale "su'unai - su'u" that one might imagine remains unsettled.

I started this thread because in the list of semantics primitives reacted by A.Wierzbicka "who" and "what" are separate concepts.
However, the list is far from perfection if you read her books really thoroughly. {ma poi prenu} {ma poi dacti} (even when objects are persons) might be enough. Anyway it's a highly speculative topic.
Reply all
Reply to author
Forward
0 new messages