Message-ID: <4722415B.2000106@lojban.org>
Date: Fri, 26 Oct 2007 15:34:51 -0400
From: Robert LeChevalier <lojbab@lojban.org>
User-Agent: Mozilla Thunderbird 1.0.7 (Windows/20050923)
MIME-Version: 1.0
To:  lojban-beginners@lojban.org
Subject: [lojban-beginners] Re: geodesic dome
References: <821531.97363.qm@web27709.mail.ukl.yahoo.com>	 <2204fa080710250218g1d01c396gc2cffe2594094d0a@mail.gmail.com>	 <47209CB8.8070703@lojban.org>	 <975a94850710250956t2cd0e855ncce7c1681ff719e2@mail.gmail.com>	 <4720DEC4.3040806@lojban.org>	 <975a94850710251146g2f1cb00eka98d0ec87c02ae3b@mail.gmail.com>	 <47210CF9.9010906@lojban.org>	 <975a94850710251510h57f6d0d1ue3d900ca8b7369f9@mail.gmail.com>	 <4721E54F.7070001@lojban.org> <975a94850710260905u5d0dc9ddw77a85a78aa082c04@mail.gmail.com>
In-Reply-To: <975a94850710260905u5d0dc9ddw77a85a78aa082c04@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Sender: lojban-beginners-bounce@lojban.org
Errors-to: lojban-beginners-bounce@lojban.org
Precedence: bulk
Reply-to: lojban-beginners@lojban.org
Content-Length: 7764

Joel Shellman wrote:
>>That is one philosophy of lujvo-making, but it leads to very long lujvo,
>>possibly with lots of extraneous places if you follow jvojva rules.
>>Zipf's law implies that such is inappropriate for concepts that see a
>>lot of use.
> 
> Yes, but don't we do this sort of thing all the time:
> 
> "So here on our right is a nice geodesic dome. The dome is blah blah.
> When someone goes in the dome blah blah."

You are attempting to apply English rules to Lojban.  That might indeed 
happen in Lojban, but if so it should not be because we do so in English.

In Lojban, insofar as this is done, we tend to more explicitly use 
anaphora (the generalized set of usages that in English are called 
"pronouns).

> So, the specific word is used initially to create context, and the
> general term is then used after that since it's obvious what is being
> referenced. And that initial "word" might be a whole phrase (as it is
> in English).

That can work in Lojban only so long as the shortening has the same 
place structure as the longer expression.  In general, that will NOT be 
true when the more specific word is a lujvo - to crib from your later 
example, the place structure for a single word meaning "motocross bike" 
will not generally be the same as the place structure for a word meaning 
"bike".  Likewise, all the different lujvo for the different flavors of 
geodesic dome might not have the same place structures, especially for 
someone trying to follow the jvojva rules (which don't really consider 
the phenomenon you are talking about).

> That said, I think we might be talking past each other a little. I
> think you're specifically talking about word creation,

Yes.

> and I'm just
> thinking a little more generally. So... just like motorcross biking is
> two words, maybe in that example, you wouldn't create a lujvo that
> means the specific "motorcross biking" concept, but rather use a
> phrase.

If you use a phrase, especially as a beginner, you will likely make 
fewer errors.  But for people who do a lot of motocross biking, a 
long-winded phrase likely won't satisfy them.

> Perhaps I'm thinking of someone creating a word on the fly... I'm
> gathering that is unlikely, though, so probably not a useful point of
> discussion.

It depends on what sort of lojbanist you are.  At my level of skill, I 
coin words on the fly all the time, enough so that last week another 
experienced Lojbanist had to query me about several of my coinings that 
aren't in any dictionary (I don't use the dictionary much when I do 
Lojban either).

>>I do favor the use of short general lujvo when indeed there are likely
>>to be multiple longer concepts that will be made from it.  I don't think
>>that applies here.  There are rather limited numbers of dome structures,
> 
> I'd say there's a fair number of domes...

Kinds of dome structure, is what I meant.  Distinct concepts based on 
spherical buildings.

>> > If context is insufficient, the speaker
>>
>>>should use a more specific lujvo, or explain further.
>>
>>That is fine for a lengthy academic discussion, but not for ordinary
>>conversation.
> 
> But as I mentioned earlier--don't we do this in conversation all the
> time. Use a specific term/phrase initially and then a shorter general
> term that obviously refers to it.

In English, we do that sometimes.  We also introduce figures of speech 
that save us a lot of words.  After all, since buildings don't talk, it 
really is rather strange to have "The White House announced yesterday 
...".

> "Fred is the fastest motorcross biker in the world. Biking is blah
> blah and his bike is blah blah..."
> 
> Without the initial more specific term, we wouldn't know what kind of
> biking we're talking about--but later on, we can use the short general
> term and it's still clear.

But of course in a motorcross club, you would just say "biking" and no 
one would assume you might mean a pedal-powered bicycle.  At that point, 
what you are describing becomes "jargon" or "insider talk".

You can do that in Lojban or another foreign language, but be forewarned 
that it is much more likely that you will be misunderstood.  Lojban has 
explicit techniques for indicating the use of a short form, and it is 
safer to use them.

>>Obviously, one would have to build a dictionary into a program.
>>Programmatic lujvo would mean you would only need gismu and fu'ivla in
>>that dictionary, but since fu'ivla are an open-ended set, you haven't
>>really simplified the problem.  So there is no advantage for the
>>computability aspect.
> 
> I would expect lujvo generation to occur much more often than fu'ivla addition.

We hope it will.  But there a several million different species of 
animal life, and there won't be a lujvo for all of them.  There are 
several thousand languages and associated cultures, and yet people 
complain about the several dozen that we chose to associate with gismu.

> Oh, and fu'ivla follow specific rules, right? So being open ended
> doesn't matter--they're still recognizable, aren't they?

There are 4 kinds of borrowing, the first two being quoting foreign text 
and converting to a name.  Type III fu'ivla are words with a classifier 
stuck on the front, which makes them easier to make on the fly, and also 
means that they will be more understandable to someone with less than 
complete contextual knowledge.  Some people are making Type IV fu'ivla, 
which are pure borrowings, and that is extremely tricky - there are only 
rules on what you CAN'T do, and they aren't easy to follow on-the-fly.
I don't use Type IV fu'ivla and I do my best to ignore those that others 
have created.

A Type IV fu'ivla is only recognizable as such after carefully ruling 
out that it is something else.  Type III fu'ivla have a distinctive 
structure that cannot occur in lujvo.

> It just seemed to me really interesting that with a very small core
> dictionary, one could potentially comprehend to a certain level
> (obviously not perfectly) nearly the whole language. Have to think
> about that some more...

Many other languages use compounding much more than English does. 
German is noted for its long words.  Russian productively uses a number 
of roots extremely productively.  The Amerind language Nootka often says 
an entire sentence in a single very long word.

>>If most people follow jvojva completely or approximately for most
>>lujvo-making, then it does reduce the learning curve.  But learning to
>>use the jvojva themselves is hardly a beginner concept, since even the
>>most skilled users of the language haven't mastered the practice.
> 
> Okay, one more--if one did follow the lujvo-making rules/conventions,
> would the place structure be deterministic then?

Because the rules were not specified that specifically, there is some 
judgement involved.  Perhaps someone who could follow the rules exactly 
would always get the same answer, but the most skilled lujvo place 
analyzer so far, Nick Nicholas, still makes inconsistent decisions about 
place structures.

 > If so... once could
> consider that there is a "form" of the language (that which follows
> those rules) that is completely computable from no more than the root
> words (excluding fu'ivla which are inherently exceptions, so that's
> ok).

I don't think so.  It was not part of the design that lujvo be 
"computable" as to their semantics.  Decomposition was added to make 
them much more learnable (TLI Loglan until 1982 allowed lujvo that 
resembled gismu, and had no requirement for decomposition, nor any rules 
at all governing place structures - great ridicule occurred when it was 
discovered that the roots of the word for "manning a ship" was the 
incredibly sexist metaphor "man-do").

lojbab