Parallel query leads to spurious context tokens in tree view

[amir-zeldes]

When querying a parallel corpus, the KWIC view may show the correct context, but in the tree visualizer, some extra tokens are shown (see image - context is +/-5 in KWIC, but tree shows more tokens which are probably not even the next ones in the text).

To reproduce use in SMULTRON:

english:cat="NP" ->align_e-d[type="good"] german:cat="PP"

[amir-zeldes]

Correction: I just noticed the tokens in the right tree are actually the left context in both hits. Not sure why they are displayed in a tree to the right of the end of the right context.