Open jonathanrobie opened 7 years ago
Luke 1:10 illustrates a case not shown in the previous examples.
Here is the current representation, which does not maintain sentence order:
This involves two constituents, each interrupted by the other. Let's try this with the proposed notation:
καὶ
s πᾶν τὸ πλῆθος
v* ἦν
s* τοῦ λαοῦ
v προσευχόμενον
adv ἔξω
adv τῇ ὥρᾳ τοῦ θυμιάματος.
I'm inclined to say that is clear enough, and simple enough given the complexity of the example.
I"m going forward with this approach. Leaving open because (1) we do not have separate documentation for Treedown yet, and (2) the treebanks do not currently display discontinuity this way.
One obstacle to using sentence order in all sentences is finding the right way to represent the result in Treedown. I would like to propose some Treedown conventions to make that possible. Here are some cases I have noticed so far, I am sure there are others that I have missed.
I think we need to handle at least two distinct kinds of discontinuity. Postpositive conjunctions, which are frequent, imply a discontinuity both of sequence and of hierarchy. Most other instances seem to involve discontinuity of sequence only. I suggest we use two different conventions to distinguish these cases.
Postpositives (discontinuity in both hierarchy and sequence)
In the current trees, postpositives are represented out of sequence, like this:
I would like to change this using two equivalent conventions, one primarily intended for easier editing, the other for clearer display. When marking down postpositives by hand, a postpositive can be flagged by prepending ** to it, as in this example:
This is equivalent to the following display-oriented representation:
If the discontinuity splits a constituent, the label of the constituent is repeated:
Which is equivalent to:
If two distinct constituents share a label, they are distinguished using numbers: s.1, s.2, etc.
Simple discontinuity (discontinuity of sequence only)
When discontinuity involves sequence only, as in split focus or some uses of enclitics, a * before the word or constituent is used.