Lojban is an artificial spoken language. It has a formally defined grammar and is based on predicate logic, but is a pretty complex beast to try to tame.

 

It's pretty hard to learn. It could certainly be made much easier. The proposals below actually describe a language which wouldn't be la lojban, but would be a new language based on la lojban that would be much easier to learn and work with.

 

The Virtues of Lojban

• It is based on logic.
• Lojban goes beyond logic, and is a codable form for representing anything a human could express that a computer could potentially parse.
• The consonents and vowels are mostly okay, with the exception of "." (glottal stop (!)), "'" (breathy thing) and "x" (throat disease).
• The emotives are awesome. I wish English had them.
• The tense and space system is superior to English. Not that making a language that's better than English is particularly hard.
• Commands using "ko" and questions using "ma" just make sense.

 

Easier Words

 

In Lojban, you have about 1400 base words of five letters each, called gismu. These have zero to three shortened alternate forms of three letters each, called rafsi, which can be combined together into compound words called lujvo. Added to this is another thousand or so cmavo which glue all these words together.

 

The gismu have a lot of overlap. For example, above I just used the lojban words: "rafsi", "lujvo" and "cmavo", each describing something quite similar, but they don't have any obvious connection with each other if you looked at the spelling. They could, for example, share a common prefix or suffix which meant "sentence component". This applies in general; there are large groups of gismu which have no apparant morphological relationship when they describe related concepts.

 

What I propose is by starting with the 23 letters and giving each letter a set of very basic connotations, such as "alive", "emotion", "solid", "large", "not" and so forth.

 

Then each letter can be combined with others, using the connotations, to make simple base wordlets (of three letters or so) that have actual meanings. For example, a base wordlet for animal could be made of the letters with the connotations of "alive" and "moves". I'd expect only a few hundred wordlets.

 

Once the wordlets are established, they could be used in sentences by themselves to describe very general concepts (e.g. "food") or simply concatentated together to form more specific concepts (e.g. "dinner" = "food" + "large" + "last"or something).

 

If individual letters have connotations, even a beginner could make an educated guess as to what a word means.

 

Easier Grammar

 

Word places

 

Lojban's grammar is well defined, but is perhaps more complex than it needs to be.

 

Word places in Lojban are very important. The words have places attached to them, numbered from one to five, where parts of the sentence are "plugged in" to add more meaning. For example:

 

lisri story x1 is a story/tale/yarn/narrative about plot/subject/moral x2 by storyteller x3 to audience x4

 

Here, x1 through to x4 are the places which you would fill in to make a sentence.

 

The fault I find with numbering these places is that you need to memorise them for every single gismu and lujvo (i.e. the whole vocabulary), and then carefully decode sentences to work out what belongs in each place of each word. A better mechanism would be to have one word per concept, and then more words which determine what aspect of that concept that part of the sentence relates to.

 

For example, you could instead define lisri simply as "story", and then have very general words for "is", "is about", "gives" and "receives". For example, a sentence could now be:

 

<is> la .cinderelas cu lisri <about> le nu nixli klama <gives> mi <receives> do

(excuse the bad tanru :-) )

 

Note that lojban does have these tags (BAI?), but they're normally not used.

 

Sentences using only postfix or prefix grammars

 

An idea from functional programming is to make sentences out of concepts which are applied to each other.

 

For example, you could start with a basic concept:

 

lisri

 

and then, assuming a postfix grammar, apply another concept to it. This would work much like a stack-based programming machine. In this case, I'm applying the word "pu" which in lojban marks the sentence as being past tense:

 

lisru pu

 

Here, "pu" takes the concept "lisri", and outputs the concept "lisri, but in the past".

 

Say for example, I was the person reading the story. Then we could tack on a modifier which takes a concept, and returns the same concept but with me as the primary agent:

 

lisru pu (mi <primary agent>).   "I have read a story"

 

The parenthesis are needed here for grouping. Here, the <primary agent> would be a word that modifies the concept "mi" and returns a curried function that could be applied to something to make me the primary agent of it.

 

Say then, that I had a complex sentence, and the above sentence was going to be a lojban-like event. Then I could convert my sentence to an event by tacking on "nu".

 

<I enjoy the event of:> (lisru pu (mi <primary agent) nu).

 

Some way of making the parenthesis optional would be needed.

 

I suspect that all of the words in lojban could be redefined to be functions. Typically they'd return the value they've been given after it has had a little bit of modification.

 

One problem is how to specify which arguments apply to which words. In most stack-based machines, the number of arguments is defined in a very concrete way. In a spoken language, the number of arguments is variable depending on how a concept is used. Some grouping mechansm would be needed to resolve this ambiguity in compound sentence structure.

 

Better Typesetting

 

Lojban enthusiasts don't tend to be typesetters. They often use a fixed-width font.

 

Lojban has several issues which make text difficult to read:

• Sentences sometimes start with ".i" or a variant. This is less easy to pick out of text than a period and a space that put a lot of white area between sentences. English also uses a capital letter to make it easier to find the start of a new sentence.
• Only lower case is typically used, making sentences shapeless. Word shapes are very important; somehow, capital letters could be introduced to make the sentence structure more apparant.
• It's common to have ' in the middle of a word. This is fine, but a'unai the apostrophe tends to take up too much whitespace on the fonts that lojbanists use. It should not break the word up visually.