Penn Treebank:

[jonathanrobie]

We need to consider what conventions to use for the categories discussed in "Penn Treebank: Empty Categories and Resumptive Elements":

https://www.ling.upenn.edu/ppche/ppche-release-2016/annotation/

[ErwinKomen]

The empty categories used in the (historical!) Penn treebanks are:

1. subjects elided under conjunction - *con*
2. General empty subjects - *pro*
3. dislocated constituents - *ICH*-n
4. expletive subjects - *exp*
5. ECM infinitive subjects - *arb*
6. Traces - *T*-n

Relevant for Greek are probably only 1, 2 and 3.

Type 3 occurs in the 'surfaced' variants of the lowfat GNT and the lowfat HebOT, so visualizing that in lowfat should perhaps have priority.

If Greek (or Hebrew) really has 'subjects elided under conjunction' (type 1) or 'general empty subjects' (type 2), then these kinds of empty categories would need to be added to the texts. However, I have two questions:

1. Do these types really exist in Greek/Hebrew?
2. Would it be possible to add them automatically?

As for (1), their existence: Greek and Hebrew have subject person/number marking on the finite verb. English has that too. But the difference between these languages is that English normally requires an overt subject, whereas Greek/Hebrew have 'no pronoun' as default in situations where the subject is the same as in a previous clause. Here is an example from Luke 4:1

cl
   s Ἰησοῦς δὲ 
   cl πλήρης πνεύματος ἁγίου 
   v ὑπέστρεψεν
   pp ἀπὸ τοῦ Ἰορδάνου , 
conj καὶ 
cl
   s *con*
   v ἤγετο
   pp ἐν τῷ πνεύματι
   loc ἐν τῇ ἐρήμῳ 
   time ἡμέρας τεσσεράκοντα
   cl πειραζόμενος ὑπὸ τοῦ διαβόλου .

The grammatical subject stays the same from clause 1 to clause 2.