Pronunciation variants

[stannam]

An example csv file (csv_pron_var.txt) is in 'csv_sample'.

• Unwanted transcription column (only 'Canonical' is expected but another column 'Transcription' gets added when creating a corpus) -- edit: solved
  

• PCT doesn't autogenerate the inventory chart (all segments go under 'uncategorized') -- edit: this issue doesn't arise in the most recent version of 'master'.

• Question: Phonotactic probability doesn't allow '... as separate entry' as variant options? (cf. this)
  

• Error when calculating MI, FL, PrOD, etc by pronunciation variants. -- edit: solved?

Traceback (most recent call last):
File "D:\PycharmProjects\CorpusTools\corpustools\gui\migui.py", line 44, in run
call_back = kwargs['call_back'])
File "D:\PycharmProjects\CorpusTools\corpustools\mutualinfo\mutual_information.py", line 197, in pointwise_mi
probability=True, need_wb=need_wd)
File "D:\PycharmProjects\CorpusTools\corpustools\contextmanagers.py", line 98, in get_frequency_base
tier = getattr(word, self.sequence_type)
AttributeError: 'Word' object has no attribute 'Transcription'

[YuHsiangLo]

Hmmm I think this is caused by the mixing of attribute _transcription, Transcription, _transcription_name, and the transcription getter and setter... This issue is related to #756, and we'll need to do some fundamental refactoring of the Word class to solve this type of problems once and for all.

[stannam]

cf. we have a separate branch for this: 'pronunciation_variants'

[stannam]

The recent commit, the one that forces the column name (92113c5) gets rid of the 'Unwanted transcription column' issue (i.e., no 'canonical' column to start with). However, for an independent reason, "List pronunciation variants" is acting out again 😳.
I have added an example file csv_pron_var.txt to the 'csv_sample' folder.