Split constituents

[jonathanrobie]

Greek allows constituents to be split. Consider the following clause:

Matthew 1:20 τὸ γὰρ ἐν αὐτῇ γεννηθὲν ἐκ πνεύματός ἐστιν ἁγίου·

The subject constituent τὸ ἐν αὐτῇ γεννηθὲν is interrupted by γὰρ, which is a post-positive. The complement ἐκ πνεύματός ἁγίου is interrupted by the copulative verb ἐστιν. Post-positives not only split a constituent, they often operate on a higher logical level in the constituent structure. The Lowfat and GBI treebanks maintain the logical levels and intact constituents at the expense of word order:

γὰρ*
    s τὸ ἐν αὐτῇ γεννηθὲν
    c  ἐκ πνεύματός ἁγίου·
    v ἐστιν

In Treedown, we want to be able to maintain word order while representing the logical levels and constituents.

When the interrupting constituent is on the same level, I propose the following:

1. If only one constituent in a clause has a given label, repeating the label signals a continuation of the constituent. In the following example, there is only one complement, but it is interrupted by the verb:

c ἐκ πνεύματός 
v ἐστιν 
c ἁγίου

2. If more than one constituent in a clause has a given label - I assume this occurs? - each constituent is given a numeric label, valid within the clause. The exact syntax for this is TBD, but it might look something like this:

c(1) ἐκ πνεύματός 
v ἐστιν 
c(1) ἁγίου

3. For post-positives, we need a way of indicating both the logical level and the occurrence in sentence order. I propose we use a convention like this:

    - μὴ 
    v.part ὁρῶντες 
cj >
    v.part πιστεύοντες
    cj* δὲ 

In this notation, cj > indicates that there is a logical conjunction at that place, but it is found further on in the text. cj* indicates that the conjunction occurs at that position. Are there postpositives that are not conjunctions?

[mwpalmer]

Postpositives are problematic, of course, but I think your proposed notation works well, including in the case of οὖν, which I do analyze as a conjunction.

[jonathanrobie]

An alternative spelling to the postpositive proposal:

    - μὴ 
    v.part ὁρῶντες 
cj>
    v.part πιστεύοντες
    >cj  δὲ 

where cj> is supposed to mean "a conjunction somewhere forward" and >cj is supposed to point directly to the conjunction we are talking about.

[jtauber]

There seems to be a lot of inconsistency in the examples whether this is

cj>
    ...
    >cj δὲ

or

cj >
    ...
    cj* δὲ

or

cj >
    ...
    cj δὲ