Linguistics 567: Grammar Engineering

Lab 9 Due 5/26 (11:59 pm)

Preliminaries

The goal for this lab is to create an "accommodation" transfer grammar for your language by using it as the target language in two translation pairs, with English and Chadian Arabic as the inputs. Along the way, you will be cleaning up your grammar so that it generates for as many of the MMT sentences as possible (ideally all of them), and generates as few outputs as are motivated.

As usual, I'll be asking for before and after tsdb profiles and a write up (see directions below).

Full credit for the "tasks" portion of the lab will be given if you can translate 12 of the the MMT sentences into your languages from English and 10 of the MMT sentences into your language from Chadian Arabic Chadian Arabic (without excessive spurious realizations). At least 30/50 points if you can translate at least two MMT sentence into your language from both English and Chadian Arabic (without excessive spurious realizations).

Running the translation system

I assume you got the basic translation system running last week in Lab 8. If not, look at the Lab 8 instructions and/or post to Canvas for help.

Collect your expected outputs

Include a plain text file (called iso.txt, with iso replaced by your iso language code) with your 24 MMT sentences (in your language only) in the format your grammar expects (morpheme segmented or not), one per line, with a blank line between each. NB: I'm not looking for IGT here, just the actual strings your grammar expects.

It's possible that more than one of the English MMT strings translates as the same string in your language. That's fine. Just repeat it so the alignment between the English MMT file and your MMT file works.

Attempt to translate into your language

  1. You should already have the eng and shu grammars from Lab 8.
  2. Make sure *translate-grid* is being set in your lkb/globals.lsp file (per Lab 8): 
    (setf *translate-grid* '(:iso :eng :shu))
  3. Similarly, edit the flie lkb/globals.lsp in the English and Chadian Arabic grammars so that the line for *translate-grid* now looks like the appropriate one of the lines below (again replacing "iso" with the code for your language). 
    (setf *translate-grid* '(:eng :shu :iso))
(setf *translate-grid* '(:shu :eng :iso))
  4. Now load your grammar into the "target" lkb.
  5. Parse Dogs sleep with the English grammar in the "source" lkb and select "rephrase". Ideally, this should be working from Lab 8 :-).
  6. Observe what happens: Do you get generation outputs? Some error in the emacs buffer in the "target" emacs?
  7. If you get an error, you'll need to compare the MRSs to to see what the difference is. I expect that for Dogs sleep you won't need any transfer rules, and thus any errors should be addressed through harmonization (aka cleaning up your MRS) and/or work on your semi.vpm file.

Comparing MRSs

To compare the MRSs, you can look at the MRS from the English grammar directly, but this can be a bit misleading, since you really want to look at the input to the generator (i.e., the transfer output). To do this, you can select "Generate | Display Input MRS" or "Generate | Display Internal MRS" from the "target" LKB Top menu.

  1. Generate | Display Internal MRS
  2. Parse the expected output
  3. Choose Indexed MRS from the pop-up menu

There are a number of things that could be wrong:

  1. Missing RELS or HCONS (broken diff-list append).
  2. Misspelled PRED values (look carefully at the underscores).
  3. Misspelled/differently spelled feature values (e.g. sing instead of sg). Recall that the right hand side of the VPM rules should match the right hand sides from the eng and shu versions of this file.
  4. Misspelled/differently spelled feature names (e.g., PERS instead of PER). Recall that the right hand side of the VPM rules should match the right hand sides from the eng and shu versions of this file.
  5. Incompatible variable properties (features and values).

Update semi.vpm, if necessary

(You may have done all there is to do here in Lab 8.)

The file semi.vpm provides a mapping between grammar-external features of indices (referential indices and events) and their values, and grammar-internal ones. For background on VPM, see the DELPH-IN wiki.

  1. If your grammar uses a PERNUM feature, you'll need to map separate PER and NUM features from the external (right-hand side) of the VPM to a single PRENUM feature on the internal (left-hand side). See the example under "Properties: An Example" on the DELPH-IN wiki page. (There is also a an example in the semi.vpm file in the eng grammar.)
  2. If your language has aspect marked in some sentences but other forms that are just underspecified for aspect, you'll want to have the default aspect be no-aspect. Define this as a subtype of aspect in your grammar, but don't have anything other than the semi.vpm mention it otherwise.

Create a transfer grammar

Once you have Dogs sleep translating, it's time to try a broader range of the MMT sentences, as well as both English and Chadian Arabic as input to see what kinds of transfer rules you will need.

Note that you will be modifying the English and Chadian Arabic grammars for this part of the lab. The transfer rules types are in mt-mrs.tdl, mtr.tdl and acm.tdl. Of those, acm.tdl should be the most interesting. You'll want to edit the file acm.mtr to create instances of the transfer rules that you need for your grammar. It will be simplest to edit this file in one grammar (say the English one) and create a symbolic link to it in the other grammar, so that you have one transfer grammar for your language.

  1. Try translating all of the MMT sentences from English to your language and Chadian Arabic to your language.
  2. For each one that doesn't go through, compare the input MRS to the MRS your expected output is giving.
  3. Do any harmonization that is warranted. If it's not clear whether you should try to do MRS harmonization or try to build a transfer rule, post to Canvas.
  4. For the remaining differences, look to see if one of the existing transfer rule types in acm.tdl will do the trick. If so, create an instance of that transfer rule type in acm.mtr, e.g.,: 
    pro-drop := pronoun-delete-mtr.
  5. If you need a different transfer rule, post on Canvas about what you need, and we'll work out how to formulate it.
  6. Reload the "source" grammar and try translating again.
  7. Rinse and repeat.

Write up

  1. Include your iso.txt file with your translations of the MMT sentences.
  2. Describe any clean up you did to your grammar.
  3. Describe the transfer rules you instantiated, and why.
  4. Describe any further transfer rules you needed to develop, and why.
  5. Document your current coverage on translating the MMT sentences from English and Chadian Arabic into your language. If you are generating more than one output for each input, explain the sources of variation.
  6. If you don't have full coverage, describe why not.
  7. Describe the current coverage of your grammar over your test suite (using numbers you can get from Analyze | Coverage and Analyze | Overgeneration in [incr tsdb()]) and a comparison between your baseline test suite run and your final one for this lab (see Compare | Competence).

Submit your assignment

Back to main course page
Last modified: