ACE LUI visualizing lemmas instead of surface forms (YY)

[olzama]

Not sure if this was asked before… I didn’t find anything.

I use YY-input mode for the SRG, for example (scroll to the right):

(1, 0, 1, <0:4>, 1, "poco" "poco", 0, "rg", "rg" 0.6631234668847097) (2, 1, 2, <5:12>, 1, "importar" "importa", 0, "vmip3s0", "vmip3s0" 0.9929577464788732) (3, 2, 3, <12:13>, 1, "." ".", 0, "fp", "fp" 1.0)

LKB-FOS displays the correct surface forms:

But ACE, whe the -l (LUI) switch, displays the lemmas:

Can I make ACE display the surface forms also? The LKB relies on some YY interface; is it doing something additional to ensure the surface forms are displayed? @johnca

[johnca]

The LKB always displays surface forms in trees (whether that’s through its native GUI or via LUI). Therefore, when I added the new YY interface to LKB-FOS I made sure that the LKB picks up the surface field from each token.

[olzama]

I wonder if I am doing something wrong with ACE, that it doesn’t do that…

[sweaglesw]

I hunted through the ACE code and discovered that ACE’s YY input reader just discard’s the 2nd string in the YY format, with a comment “another string, e.g. surface form for the SRG”. I suppose when I wrote that, there weren’t any usable grammars available to play with that required a different value for the 2nd string there, so I didn’t implement it. It doesn’t look like a one-liner to figure the right way to carry that data through the system, either. Does that string have any other use?

Upshot: there’s no way to display those surface forms in LUI driven by ACE at the moment.

[olzama]

@sweaglesw

Does that string have any other use?

I can’t think of any other use immediately, so perhaps this is fine (especially since I can now get that from the LKB!) I think fftb displays the surface form (can’t check right now); that would be important…