ACE parses sentence, FFTB doesn't display a tree?

I think I’ve asked this before, but I cannot find the answer now…

Suppose I have a sentence which I was using as a test sentence, and I have added a lexical rule to handle one of the words better. ACE currently parses this sentence, yielding 4 trees:

(1, 0, 1, <0:3>, 1, "el" "los", 0, "da0mp0", "da0mp0" 0.99272812) (2, 1, 2, <4:10>, 1, "sordo" "sordos", 0, "ncmp000", "ncmp000" 0.33870630) (3, 2, 3, <11:13>, 1, "lo" "lo", 0, "pp3msa0", "pp3msa0" 0.33476364) (4, 3, 4, <14:18>, 1, "oír" "oyen", 0, "vmip3p0", "vmip3p0" 1.00000000) (5, 4, 5, <19:24>, 1, "mejor" "mejor", 0, "aq0cs00", "aq0cs00" -1.00000000) (6, 4, 5, <19:24>, 1, "mejor" "mejor", 0, "aqccs00", "aqccs00" -1.00000000) (7, 4, 5, <19:24>, 1, "mejor" "mejor", 0, "aqccs00", "aqccs00" -1.00000000) (8, 5, 6, <25:28>, 1, "que" "que", 0, "cs", "cs" 0.44986075) (9, 5, 6, <25:28>, 1, "que" "que", 0, "pr0cn00", "pr0cn00" 0.44986075) (10, 6, 7, <29:37>, 1, "nosotros" "nosotros", 0, "pp1mp00", "pp1mp00" 1.00000000) (11, 7, 8, <37:38>, 1, "." ".", 0, "fp", "fp" 1.00000000)

(ACE LUI doesn’t display inflected forms, only lemmas, but the words are inflected.)

The change with respect to the previous version of the grammar is an additional lexical rule for the word mejor, and that results in two additional trees (there were only two trees before).

Now when I look at this sentence, which is part of a test profile, in FFTB, the sentence appears brown as I would expect:

Screenshot from 2024-04-03 11-49-03

But I cannot get to a tree, no matter what I do:

Start:

End, no matter which choices I make:

The same happens if I don’t do an automatic update and simply try to treebank the sentence “from scratch”.

I suspect this is somehow related to the fact that I added a lexical rule? Note that something is amiss with the word mejor, it is not underlined with green. Would be good to understand what’s going on (and what to do about it).

In the terminal, I see:

gold tree node 'aq0cs00@aj_pp_i-cmp_native_le' conflicts with 'aqccs00@aj_pp_i-cmp_native_le'

OK, but why don’t I see a tree?

This might be the answer:

…except in this case this is not two identical lexical entries but two possible morphological analyses for a word, I guess, which seems two me like that should be working?..

Hi Olga,

As you suspected the symptoms point strongly to a duplicate lexeme. In this case i think it may not be the grammar proper that is at fault however. Your YY input contains 3 “mejor” tokens, of which 2 look identical to me. That would give rise to two copies of any parses based on that token which would be indistinguishable to FFTB, hence the observed behavior. Is there a difference between those tokens that is escaping my eye?

-Woodley

1 Like

Ah! Thanks, Woodley, I didn’t notice. Probably a bug in my YY interface. It shouldn’t be there.