Anyone who is learning Chinese has probably come across Pleco, a Chinese dictionary App for your phone. It offers a lot of different features, some of my favourites are offline access, clipboard reader, and integration with Anki.
However, while it is a great resource for learning Mandarin and Cantonese, it doesn't offer much when it comes to other Chinese topolects. You may ask, what are topolects? Well, basically dialects, but they are in most cases mutually unintelligible. It corresponds to the Chinese word 方言1.
OK, so what can I do about it? My first thought would be to just fork the project and implement support for other topolects myself. But you see, Pleco is not Free Software, so I don't even have access to the source code… But after some web-searching, I found out that you can actually create your custom dictionary in Pleco, but you first need to purchase their 10-dollar flashcard add-on.
Before I actually sent them my money, I obviously had to figure out
how custom dictionaries work, and, well it ain't pretty. You can either
import .pqb files, or plain .txt files. The first option is only really
useful for importing dictionaries that other people have already made,
because .pqb seems to be Pleco's own
database file format, which I don't know anything about, and is not
human-readable at all. So my only real option was to create a plain
.txt file. The format works like this:
SIMPLIFIED[TRADITIONAL]<tab>PINYIN<tab>DEFINITION<newline>
I found out about this on their
forum2, but there doesn't seem to be any
official documentation about this. But here comes the really ugly part,
to insert newlines in the definition section, you need to use some weird
private unicode symbol (UEAB1), which my browser isn't even able to
display. There are also four or five other weird unicode symbols for
making things bold, coloured, etc. It would be much better if they used
JSON or CSV, or something already well-known, instead of making up their
own weird format.
OK, rant over. So what did I end up doing? First, I payed for the flashcard add-on. I came a cross a project called Wikihan, where they extracted pronounciation of characters in different topolects from Wiktionary, and stored the result in a TSV file. So kind of exactly what I was looking for, great! They have a GitHub repo with some scripts to generate everything, and I modified it to suit my needs3. Then I wrote a Haskell program to convert it to Pleco's weird plain-text format.
Then importing the file in Pleco, and now you know how to pronounce your characters in Hakka, Hokkien, Gan, etc. However, not all characters have data in all topolects, but better than nothing of course.
They used a Python library called Epitran which isn't packaged in Nixpkgs, and since it was only used to convert different Romanizations to IPA, which I didn't really care about, I removed the dependency on this it to be able to use Nix.↩︎
They used a Python library called Epitran which isn't packaged in Nixpkgs, and since it was only used to convert different Romanizations to IPA, which I didn't really care about, I removed the dependency on this it to be able to use Nix.↩︎