"Self-Segregating Syntax"? (Unique Trees, Recoverable Uniquely From The String)
|From:||Eldin Raigmore <eldin_raigmore@...>|
|Date:||Saturday, April 15, 2006, 17:19|
There has been much since Dec 2005, and also recently, about "self-
segregating" or "auto-isolating" morphology on this list. (This is a
morphology in which each segment/phoneme in a word goes with a particular
morpheme -- syllable boundaries, morpheme boundaries, and word boundaries
are uniquely recoverable from the utterance.)
I am interested in the analogous situation in re syntax.
Any string-rewriting grammar, including Context-Free Grammars, when parsing
an utterance, in effect assigns a tree structure to it (with the words, or
perhaps the morphemes, as leaves). The same is true of any tree-rewriting
grammar, including any Tree-Adjoining Grammar and any Ranked-Node-Rewriting
Grammar. (In tree-rewriting grammars, the produced tree is not necessarily
isomorphic to the production tree, as it is in string-rewriting grammars;
nevertheless there is a tree.)
I'd like list members to tell me two things;
1) What do you call (auto- or self-)(-isolating or -segregating) syntaxes --
those in which at most one tree structure can be assigned to any string?
2) What are the most popular systems for doing so?
I have though of one collection of such systems; it is "sort of prosodic".
Suppose each word could be assigned a unique score from, say, 0 to, say,
9. (Phonologically this could be indicated by tone, by length, by "weight"
(moraically), by volume, by lenis/fortis or lax/tense, by sonority, or
perhaps by other ways.)
If a word's "score" is an even digit, then that word is the first word in a
group of words that contains all subsequent words of lesser "score" forward
to, but not including, the next word with an equal or higher "score", or
the end of the utterance.
If a word's "score" is an odd digit, then that word is the last word in a
group of words that contains all previous words of lesser "score" back to,
but not including, the last word with an equal or higher "score", or the
beginning of the utterance.
I can post, or e-mail offlist, examples if anyone asks.