For starters, Lojban is a constructed language (just like Esperanto) but the main goal of it is not to be an international auxiliary language, but to prove or disprove the so-called Sapir-Whorf hypothesis. In general this hypothesis says, that the structure of a language affects the ways in which its respective speakers conceptualize their world. Through the years it’s own community and inventors added optional usage goals and fields it could be used in just like in AI-Human interactions, or even to be an Esperanto-like auxiliary language.
The problem with Lojban is quite same with Chinese, it’s about understanding (visual) concepts . Kanjis in the case of that asian language, and grammar in the case of Lojban.
The strongest aspect in Lojban is that all gismu (root words) and compound words give a full concept, it encodes a whole, as objective idea as grammatically possible. For a quick example, xislu means wheel, but more specifically it means “x1 is a wheel [tool] of device/vehicle x2, made of materials/having properties x3.” The arguments and usage rules are fixed just like a dictionary entry, but this also means that you need to learn the whole “entry” even though most of the argument list for example is kind of easy to guess in most of the cases. Just as the widespread xkcd comic strip hints, this is not really different from actually making people learn dictionary entries. Personally I found it really challenging to learn words this way, and I haven’t even mentioned all the pro-sumtis, abstractors and other types of words. Instead of using your own ideas and learning words with words, you are forced to learn a whole dictionary entry for everything. This is highly useful in the terms of disambiguity but makes studying a nightmare.
Learning materials mention tanru concepts in the first few chapters usually, and it becomes quite obvious already, that a fully unambigous language is not possible. Anyone trying to find the final semantics, a Holy Grail in Lojban will be already surprised as even with all the tools to make it a safe environment for concepts, Lojban cannot provide perfection. In certain cases you need to give a description of what you mean, just like in natural languages. Some authors call the attitudinals innovative, but with a more skeptical look it’s not really different than for example using emoticons.
For human usage you need to give up some grammatical parts like terminators that make it fully possible to parse (by a machine or software) but make the language really inconvenient to use. Lots of features are not innovative or new, and can be found in natural languages like the space-time system is very similar to the ko-so-a-do system in japanese. Natural languages don’t work in bridis, you may think, but as you advance in your lojban studies you can be quite surprised that natural grammar can be perceived in a really similar way. Do people need a constructed language that use natural language tools quite efficiently, and forces a framework within the language that only exist in some people’s mind usually? The lojban framework is quite similar to how people can perceive grammar as whole, it’s closer to some kind of meta-language just like some Chomsky-like universal grammar.
The conclusion from this adventure for me is, that Lojban needs more attention. People should study it. But it’s not more than a beautiful thought experiment. It’s not really suited for common usage, because despite the really easy grammar it’s a pain to learn and a pain to use. You can make difficult structures in no time with a dictionary, but actually using it is just mind-numbing and hard. Overall I’d suggest everyone to study some Lojban, but wouldn’t recommend to actually learn it. It’s quite like an amazing thought experiment, useful in many ways but way too much impractical.
Kind of an edited and extended translation of this blog entry from January.