Add Turkish wordlist to BIP39 #1613

pull efecini wants to merge 3 commits into bitcoin:master from efecini:master changing 2 files +2064 −0

efecini commented at 8:25 pm on June 8, 2024: none

According to the “EF English Proficiency Index”, Turkish people have low proficiency in English hence we need the Turkish seed words. Link: https://en.wikipedia.org/wiki/EF_English_Proficiency_Index
Create turkish.txt
Add Turkish wordlist to BIP39
96d64db8c3
Update bip-0039-wordlists.md caf1667fe9
Update bip-0039-wordlists.md 93b8b0c87f
murchandamus commented at 2:37 pm on June 9, 2024: contributor

Hi Efe,

This update sounds to me as if the BIP39 authors have generally restricted themselves to focus only on an English wordlist. I am afraid that I do not know where your wordlist should be submitted eventually, since neither BIP39 nor the WLIPs seem to be looking to incorporate further lists. Perhaps this would be a topic to discuss with other Turkish speaking Bitcoin users.

I also noticed that there were at least two prior attempts to create a Turkish word list. You may find #1173 and #1359 useful.
efecini commented at 3:19 pm on June 9, 2024: none

Hi Efe,

This update sounds to me as if the BIP39 authors have generally restricted themselves to focus only on an English wordlist. I am afraid that I do not know where your wordlist should be submitted eventually, since neither BIP39 nor the WLIPs seem to be looking to incorporate further lists. Perhaps this would be a topic to discuss with other Turkish speaking Bitcoin users.

I also noticed that there were at least two prior attempts to create a Turkish word list. You may find #1173 and #1359 useful.

I checked them both and both of them has falsy words. If there is a Portuguese wordlist(population: 10M), there should be a Turkish one too(population: 85M). I am living in Turkey as a resident for 38 years now and have sufficient knowledge on how low an average Turkish citizen English proficiency level is. Again, if a Portuguese person has the ability to save wealth in 12 words in his/her native language, a Turkish person also should.
untreu2 commented at 3:36 pm on June 9, 2024: none

ACK
izelgurbuz commented at 3:37 pm on June 9, 2024: none

ACK
Metonya commented at 3:42 pm on June 9, 2024: none

ACK
konnydacherry commented at 3:46 pm on June 9, 2024: none

ACK
kaanaskin93 commented at 3:51 pm on June 9, 2024: none

ACK
cankasikcii commented at 3:51 pm on June 9, 2024: none

ACK
leonidasbitcoin commented at 3:57 pm on June 9, 2024: none

ACK
cemgolo commented at 4:27 pm on June 9, 2024: none

ACK
yutyo commented at 5:43 pm on June 9, 2024: none

ACK
birolcaglar97 commented at 9:21 am on June 10, 2024: none

ACK
merveaslanturkiyelidemir commented at 9:31 am on June 10, 2024: none

ACK
in bip-0039/turkish.txt:142 in 93b8b0c87f
137+avurt 138+avut 139+aya 140+ayak 141+ayar 142+ayaz
murchandamus commented at 5:51 pm on June 10, 2024:

Scrolling over your list, I noticed that there are a number of words that overlap with others in the first three letters, and in some cases those three letters by themselves are even another word on the list, e.g.

aya ayak ayar ayaz

This feels like a certain pitfall for would-be users. One of the main design goals of these wordlists is to make the words as distinctive as possible. I would recommend that you aim for a difference of at least two letters among the first four letters, and that words should never be prefixes of other words on the list.

ProofOfKeags commented at 6:27 pm on June 10, 2024:

Adding to this, one of the key properties of the word lists is that the words are uniquely identifiable by their first four letters. Projects like crypto steel rely on this behavior.
in bip-0039/turkish.txt:104 in 93b8b0c87f
99+aort 100+apse 101+aptal 102+ara 103+araf 104+arak
murchandamus commented at 5:54 pm on June 10, 2024:

Anagrams seem especially dangerous:

arak … arka

I would recommend that you reconsider the design goals for your wordlist and selection of words from the perspective that someone with a smattering of understanding transcribes their seed words under pressure from some Bitcoin enthusiast and then later proceeds to depend on their seed backup for their life savings without ever verifying that it is accurate.
murchandamus changes_requested
murchandamus commented at 6:07 pm on June 10, 2024: contributor
Hey, it’s great to see that there are a lot of people supporting your project.

As explained above, BIP 39 no longer accepts additional word lists. I encourage you to continue to hone your work and pursue publication, but as BIPs are author documents, I do not see it getting published in BIP 39.

I see a couple possibilities:
- Draft your own BIP to supersede BIP 39 in regard to Turkish
- Publish your wordlist independently per a designated GitHub page or repository
Either way, I would hope that some of the people that appeared here to support your pull request would also support you by reviewing your proposal and helping you disambiguate the words on the list.
jonatack commented at 9:03 pm on June 10, 2024: contributor
As explained above, BIP 39 no longer accepts additional word lists. I encourage you to continue to hone your work and pursue publication, but as BIPs are author documents, I do not see it getting published in BIP 39.
I see a couple possibilities:

Draft your own BIP to supersede BIP 39 in regard to Turkish

Publish your wordlist independently per a designated GitHub page or repository

Either way, I would hope that some of the people that appeared here to support your pull request would also support you by reviewing your proposal and helping you disambiguate the words on the list.
Thank you @efecini for your proposal. Nevertheless, @murchandamus is correct; according to the BIP author’s feedback in #1395 (comment), further wordlists won’t be added to BIP39. This should therefore be closed in favor of one of the alternatives suggested (and bip-0039/bip-0039-wordlists.md updated).
jonatack closed this on Jun 10, 2024

Contributors
efecini murchandamus untreu2 izelgurbuz Metonya konnydacherry kaanaskin93 cankasikcii leonidasbitcoin cemgolo yutyo birolcaglar97 merveaslanturkiyelidemir ProofOfKeags jonatack