10

The description from Wikipedia:

Lojban (pronounced [ˈloʒban]) is a constructed, syntactically unambiguous human language based on predicate logic, succeeding the project of Loglan.

I understand that this is the goal of the language. Has it succeeded? I skimmed through the Wikipedia article but wasn't able to quickly find a clarification on its ambiguity. I did notice this in the "Future Goals" header:

Eliminating ambiguity in language

I am having trouble understanding the distinction here.

MrHen
  • 6,113
  • 37
  • 59

1 Answers1

15

The Lojban site seems to clarify, indicating that yes, it has succeeded in at least being syntactically unambiguous (emphasis mine):

Lojban has an unambiguous grammar (proven by computer analysis of a formal grammar with YACC), pronunciation, and morphology (word forms). [1] The person who reads or hears a Lojban sentence is never in doubt as to what words it contains or what roles they play in the sentence. Lojban has no words that sound alike but have different meanings (like "herd" and "heard"), that have multiple unrelated meanings ("set"), or that differ only in punctuation but not in sound (like the abominable "its" and "it's"). There is never any doubt about where words begin and end ("cargo shipment" can be heard as 2, 3, or 4 words). The function of each word is clear; there is nothing like the English "Time flies like an arrow.", in which any of the first three words could be the verb. Precision in no way confines the meaning of a Lojban sentence. It is possible to speak nonsense, to tell a lie, or to be misunderstood. You can be very specific, or you can be intentionally vague. Your hearer may not understand what you meant, but will always understand what you said.

SOURCE

So, without being 100% clear on your question, it appears that one could take this two different ways:

  • Yes, Lojban has succeeded in making an unambiguous language with respect to word roles (verb, adjective, noun, etc.), word used (as in nothing like "witch/which" when listening or "flies (bugs)/flies (flight)" when reading) and the like.
  • No, the human intentionality component still appears to allow the speaker to be vague or intentionally misleading, which still allows for ambiguity with respect to the content.

It appears that the goal was to eliminate common hangups with misunderstandings that occur as a result of syntax, not as a result of being imprecise or misleading. Thus, again, the answer may depend on what you're asking. It appears that their goal was always to eliminate syntactic sources of ambiguity and that they have succeeded in doing so, as shown through the success of parsing the language with a compiler.

[1] YACC

Hendy
  • 21,665
  • 17
  • 116
  • 179
  • +1; This gets to the heart of what I was wondering about. Any more details on "proven by computer analysis of a formal grammar with YACC" would be great, but I may have to take that elsewhere. :) – MrHen Jun 07 '11 at 20:45
  • Well, my assumption (perhaps incorrect) was that if a compiler could parse the language and output the correct result, the language is syntactically unambiguous. In other words, for non-unambiguous languages, such a straightforward parsing would not be possible -- hence needing [Watson](http://www-03.ibm.com/innovation/us/watson/index.html) to understand things like Jeopardy questions (though I realize that there's more than simply ambiguous word meanings going on in that example). Does that help? – Hendy Jun 07 '11 at 21:05
  • @Hendy: Not really. What does "parsing" a language mean in the sense of outputting the correct result? As in, if Jeopardy questions were in Lojban, how would that help Watson? – MrHen Jun 07 '11 at 23:11
  • 1
    If Jeopardy questions were in Lojban, Watson would not need the immense amount of programming required to distinguish "which" from "witch", "its" from "it's", etc. There's an immense amount of context-awareness that is required just to disambiguate syntactically, and Lojban eliminates that. As their site says, "Your hearer...will always understand what you said." Thus, a parser hearing some Lojban word will not wonder whether a question word, "which," was meant or whether a black-clad/long nosed broom rider, "witch" was meant. – Hendy Jun 07 '11 at 23:17
  • Okay, that makes sense. Thanks. I was getting "processing words" mixed up with "processing meaning" in the sense that "correct" implies identification of the meaning of each part of what was said (as opposed to processing the meaning of the whole.) At least, that is how I would word it. – MrHen Jun 08 '11 at 00:15
  • @MrHen: Yes, that's how I'd take it according to their definition of "unambiguous" -- unambiguous parts has been attained; ambiguous interpretation/intended meaning is still possible. – Hendy Jun 08 '11 at 01:59
  • 2
    @Hendy: YACC is a parser generator, and generates LALR(1) parser tables. (The details of LALR(1) are arcane and not of great interest here; a language with a LALR(1) grammar is context-free in Chomsky's hierarchy.) If there are no conflicts, then the grammar given is unambiguous and not really all that complicated, given the limits of LALR(1) parsing. The fact that sentences can be parsed by the grammar means that each word can be assigned its role. There can be no doubt about what role each word plays, and it should be readily human-parsable. – David Thornley Jun 08 '11 at 02:50
  • @David: I'd seen that lingo when I looked at what YACC was, but didn't understand it entirely. "Parse" is jiving with what I took to mean, and you put it well: "The fact that sentences can be parsed by the grammar means that each word can be assigned its role." Perfect -- 1:1 "lookup table" rather than multiple potential roles/meanings per word. – Hendy Jun 08 '11 at 11:11