Tier type dependencies for interlinearization

ELAN
1

Hi,

Interlinearization (5.4 Mac) isn’t working as expected on me. Parse Analyser copies the result from White Space Analyser, without cutting the words into morphemes, and Gloss Analyser doesn’t create any annotations on the corresponding tier.

I thought the problem was my tier type dependencies. Here’s what I have:

utterance (time-aligned)

  • word (symbolic subdivision of utterance)
    • morpheme (symbolic subdivision of word)
      • gloss (symbolic association of morpheme)

Am I doing it wrong? Or might it be some other reason?

Thanks a lot! –Weijian

2

Configuration of the interlinearization system is still a bit dispersed. For the analyzers it is important that:

  • the source and target are set correctly (I guess here the word tier is the source for the Parse Analyzer and the morpheme tier the target etc.)

  • that the relevant tier types are linked to fields in the lexicon. So e.g., in Type->Change Type… the tier type of the morpheme tier(s) is probably linked in the “Lexicon connection” field with the “lexical-unit” field of the lexicon and the gloss tier type probably with the “sense/gloss” field of the lexicon.

It might be that after changes in the tier types you have to save the file and open it again to see the effect, I’m not sure.

3

Hi!
I am also having trouble getting the interlinearization to work (5.5). Right now I’m still trying to get just the Parse Analyzer to work. I’ve imported a .lift dictionary from FLEX. I’ve set up Type for the morpheme (target) as lexical-unit, as specified above. My Type for the words (source) is “word”. Does it need to be different?
My output into the target tier is just whole, unparsed words.

Thanks for any help!

D

4

So, if I understand it correctly you do get some output (suggestions) but it is always the same as the input?
Hmm, should the parser be able to identify prefixes and suffixes in the lexicon, based on an affix marker (e.g. ‘-’, a hyphen) and/or “prefix” or “suffix” as the value of the “morph type” field?

5

Hi Han,
Yes, the dictionary includes morphemes with hyphens and labels as prefix/suffix/stem.
Is it worth emailing you the .lift and .eaf files? Would that be helpful in assessing the issue?

D

6

Hi D,
Yes, that was what I planned to suggest next, it might be very difficult to find the nature of the problem otherwise.
If you send an .eaf, be sure to also send the accompanying .pfsx file. I don’t need the media files (han.sloetjes AT mpi.nl).