[tijlan]

  > When we speak, we frequently start a sentence not knowing how it's going to end.

You can start and continue with a Lojban sentence indefinitely via various means. Metalinguistic markers such as sei ... se'u allows a discursive (on-the-fly) predicate or sentence. You can insert parenthetical notes with to ... toi into anywhere. You can insert the attitudinals (used to express attitudes, emotions, evidentiality, etc.) into anywhere. A construct called tanru allows an endless sequence of predicates (whose form in Lojban do not alter according to the natlang parts of speech such as adjective or adverb, thus imposing less restrictions on the way you keep forming a sentence than natlangs such as English does). With li'o you can omit any quantity of text you don't want in your expressions. With si/sa/su you can 'erase' various mistakes in your utterance. And so forth.

  > We pause and insert filler words to give us time to form our thoughts.

That's what y is for in Lojban.

  > We use ambiguity because sometimes we just don't know precisely what we're trying to say

The unambiguity of Lojban is mostly syntactic, not always semantic. You can be semantically ambiguous in Lojban with e.g. tanru.

  > (Lojban has, if I recall correctly from Okrent's book, over 30 ways to say 'and').

Which includes ju'e, a vague connective for "and".

  > Forcing people to have a complete understanding of what they want to say before saying it

Which is not the case in Lojban.

[DanielStraight]

Syntactic ambiguity is half the point. I still have to decide whether or not I want to use the vague/all-purpose connective for "and." If I want to add parenthetical information, I have to realize I'm doing that and indicate it. If I want to "erase" a mistake, I have to recognize that that's what I want to do and use the proper word. I still have to know exactly what I intend to say. When I'm speaking, I don't think to myself, "Ok, this is parenthetical... this is correcting an error... this is vague and all-purpose." I just speak.

In Python, whether or not a method is private is defined basically by whether or not you call it from another class. In Java, you have to be explicit about it. You CAN just declare everything public (be vague and all-purpose), but aside from being frowned upon, you still have to conciously decide to do this. If you don't care about the protection of scope, you can make everything public, but you're still specifying scope. In Python, you don't specify scope at all. Even if you wanted to, you can't.

The same thing shows up in type checking. If you want, you just call everything an object in Java. It's vague and all-purpose, but you still have to specify a type. You still can't write 'a = 4'. You still have to write 'Object a = new Integer(4);', which it's tough to argue is simpler just because it isn't type-checked.

Explicit ambiguity is not much closer than explicit disambiguity to implicit ambiguity.

I realize at this point, I'm getting beyond my own knowledge of Lojban, but I just want to make the general point that being able to be ambiguous does not automatically afford the advantages of the natural ambiguity in "natlangs."

[tijlan]

  > I still have to decide whether or not I want to use the vague/all-purpose connective for "and."

If you are undetermined, you use the undetermined option, the vague one. You don't really make a decision for that.

  > If I want to add parenthetical information, I have to realize I'm doing that and indicate it. If I want to "erase" a mistake, I have to recognize that that's what I want to do and use the proper word. I still have to know exactly what I intend to say.

Planning is not required for adding parenthetical information. You use it on the spot where you happen to want it.

If you don't recognize that you want to erase a mistake, you just don't use the erasers.

  > When I'm speaking, I don't think to myself, "Ok, this is parenthetical... this is correcting an error... this is vague and all-purpose." I just speak.

I'm not a native English speaker. When I started learning and speaking in English, I would think to myself, "Ok, the word "which", when used after a slight pause, which corresponds to the comma in writing, it means a non-restrictive relative pronoun, which is what I want now, so I'm going to use it that way". As I kept practicing the language, I internalized the rule, becoming less and less actively conscious of it. Lojban is no exception. I have already internalized some parts of the grammar to which I used to pay much attention in my earlier period of learning.

Also important to note is that, what English expresses with non-verbal properties such as intonation, Lojban can do verbally. When you orally say something in English which you would put in a parenthesis in writing, your speech act are still subject to some phonetic principles such as inserting a pause, changing the rhythm, lowering the pitch, using less breath, and so on. You consciously or unconsciously have to be in command of these properties if you are to successfully deliver your utterance. A particular combination of these non-verbal phonetic properties is what Lojban expresses with to ... toi, the parenthetical markers. Whether or not you are actively conscious of them is a matter of internalization. Just like an experienced English speaker not always actively thinks about each noun's grammatical number in their utterance but still manages to add the plural marker "s" where appropriate, an experienced Lojban speaker would be able to correctly start a parenthetical note with "to" but without always actively checking whether or not that's what they really want to say.

> I realize at this point, I'm getting beyond my own knowledge of Lojban, but I just want to make the general point that being able to be ambiguous does not automatically afford the advantages of the natural ambiguity in "natlangs."

Could you give me an English example with such advantages?