Preparing datasets - Githubissues

LinguList commented 8 years ago

Each dataset (that is: one source, one reference in our bibtex file) will have an extra folder, and we will add the following things (if available);

inventories.tsv (we start from the literature, but later may add corrections to those)
profile.tsv (the orthography profile we use to parse the data_
words.tsv (the lexical data which is generated by using the profile)
morphemes.tsv (the lexical morpheme data which is generated, containing potential information on morpheme transitions, that is, partial colexifications etc.)
concepts.tsv (the concept list, that links the data to the concepticon, and thus allow us to compare things across languages)
languages.tsv (a short tsv-file indicating specifics about the language and mapping it's name to our master-list)

Inventories is thus a first step towards consistently representing the data. If we have already the concept list, as in the case of Inglis2003, we can annotate this, so no need to do things multiple times (same as for orthography profiles).

LinguList commented 8 years ago

Okay, here's the list for this dataset:

[x] link to concepticon (identical with Mann-1988-406)
[x] add to bibliography
[x] create phoneme inventory file Attached Inglis 2003 Ngochang.ods.zip
[x] provide small description of data This is a working paper that describes the phonology of Ngochang (Achang). It is more detailed than a lot of what we have, but still not as sophisticated as Lustig 2010. The underlying data is from a tape that Wannemacher took in 1994, but the analysis was checked with other native speakers. The vocabulary at the back is in very strict phonetic notation, rather than being adjusted to match the analysis given in the paper.
[x] digitize data
[ ] create orthography profile I am not sure how to do an orthography profile, because in this case the transcription is extremely narrow. Rather than assigning orthographic forms to IPA like usual, we will have to obliterate some of the phonetic noise to get the phonemic orthography, if that is what we want.

nh36 commented 7 years ago

As far as I am concerned this data is now ready to go. Note that we have another source by the same group that has a longer word list.

nh36 commented 7 years ago

Please send an update on this thread.

LinguList commented 7 years ago

Ah, comparing with #34, I realize what you mean or may mean by "update". I'll have a look at this later.

nh36 commented 7 years ago

Sounds good. I will look forward to hearing more.

digling / burmish

Preparing datasets #76