Reverse Polish Notation vs. Strictly Left-Branching Language

Discussion of natural languages, or language in general.
Post Reply
User avatar
darkworldsawake
Niš
Niš
Posts: 4
Joined: Sun Jan 31, 2010 4:57 pm

Reverse Polish Notation vs. Strictly Left-Branching Language

Post by darkworldsawake »

Hello, this is my first post--I would have introduced myself somewhere else, and sans wall o' text below, but there didn't seem to be a thread for that.

Reverse Polish Notation is a self-parsing syntax where operators (which would include things like verbs, adpositions--any modifier, basically) succeed their operands (nouns, substantives). It's a bit easier to explain using arithmetical notation: "+" usually means, "take the sum of what comes before and what comes after", but would, in RPN mean, "take the sum of the two preceding items".
This is what makes it self-parsing. Without the order of operations, "2+6/3-1" is intractably ambiguous. With the RPN statement 263/+1-*, however, you know that the expression must be evaluated as "3".

And yes, I am familiar with Fith.

So my question is: is this different from a strictly left-branching language? It seems to me that it is, because the head, which comes last in a LBR, doesn't have to be a modifier.

*2, 6, 3, /{divide two previous entries, "6", "3", return the result "2" to the stack}, +{add two previous entries, "2", "2", return result "4" to the stack}, 1, -{subtract two previous entries, "4", "1" and return result "3" to the stack}, since "3" is the only entry now in the stack, and since there are no more modifiers, this is taken to be the final result.

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: Reverse Polish Notation vs. Strictly Left-Branching Lang

Post by WeepingElf »

darkworldsawake wrote:Hello, this is my first post--I would have introduced myself somewhere else, and sans wall o' text below, but there didn't seem to be a thread for that.

Reverse Polish Notation is a self-parsing syntax where operators (which would include things like verbs, adpositions--any modifier, basically) succeed their operands (nouns, substantives). It's a bit easier to explain using arithmetical notation: "+" usually means, "take the sum of what comes before and what comes after", but would, in RPN mean, "take the sum of the two preceding items".
This is what makes it self-parsing. Without the order of operations, "2+6/3-1" is intractably ambiguous. With the RPN statement 263/+1-*, however, you know that the expression must be evaluated as "3".

And yes, I am familiar with Fith.

So my question is: is this different from a strictly left-branching language? It seems to me that it is, because the head, which comes last in a LBR, doesn't have to be a modifier.

*2, 6, 3, /{divide two previous entries, "6", "3", return the result "2" to the stack}, +{add two previous entries, "2", "2", return result "4" to the stack}, 1, -{subtract two previous entries, "4", "1" and return result "3" to the stack}, since "3" is the only entry now in the stack, and since there are no more modifiers, this is taken to be the final result.
I'd say that a left-branching human language such as Turkish or Japanese has at most a superficial similarity with Fith. Sure, a simple clause looks similar, but actually, the organization of the grammar is entirely different. Fith allows for massive amounts of center-embedding and other weird syntactic tricks utilizing the stack, which lie wholly outside the human mind's real-time comprehension abilities. It is not even meaningful to describe Fith using the syntax-tree approach used to describe human languages.
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
finlay
Sumerul
Sumerul
Posts: 3600
Joined: Mon Dec 22, 2003 12:35 pm
Location: Tokyo

Post by finlay »

Fith is a bit like an SOV language if you restrict the stack length to something like 2 or 3 and don't allow stack-jumbling operators. Those are the two things that differentiate it, IIRC. You can potentially have words appearing the equivalent of several clauses early with Fith.

As for the syntax tree metaphor, we don't really know for certain how the brain deals with sentences. The trees are useful for showing when things are embedded in one another, but I've always seen them as more of an analytical tool.

User avatar
darkworldsawake
Niš
Niš
Posts: 4
Joined: Sun Jan 31, 2010 4:57 pm

Post by darkworldsawake »

Ah, thank you, that's quite helpful.

Also: given that the arguments must precede the predicate in RPN, it seems that it would be impossible to do what lojban does and skip unimportant arguments without using placeholders. Can anyone think of an elegant syntactical solution to this?

Post Reply