Wiktionary talk:About Chinese/Hakka

Meixian gi and ki

 * Hi. For Hakka (Meixian), "gi" and "ki" are pronounced as [ci] and [cʰi] but it's not what I hear in MFCCD, e.g. 斤 (gin1) and 其 (ki2). Their transcription also uses "k" and "kʰ". Could you unconfuse me, please? --Anatoli T. (обсудить/вклад) 22:39, 8 March 2020 (UTC)
 * The sources I have say that velar stops become palatal before /i/, which should be pretty clear from the tables on this page. — justin(r)leung { (t...) 22:43, 8 March 2020 (UTC)
 * Sorry, it's not clear to me. Palatal doesn't necessarily mean a different sound, for example [kʲ] is palatal as in Russian but 其 is supposed to be /cʰi¹¹/. That's not what is heard in the audio or not how I would describe the initial. Please confirm that the IPA is right. --Anatoli T. (обсудить/вклад) 22:58, 8 March 2020 (UTC)
 * They specifically use those symbols. They're definitely not palatalized like the Russian consonants. There may be speaker variation though. From what I've heard from videos on Youtube, I know /h/ before /i/ is a pretty clear [ç], but I'm not sure about /k/ and /kʰ/. This study actually treats /c/ and /cʰ/ as separate phonemes from /k/ and /kʰ/ (although it seems like something like /kʰien/ [cʰian] in other sources is treated as /cʰan/). — justin(r)leung { (t...) 23:07, 8 March 2020 (UTC)
 * OK, thanks. We have a bit of discrepancy between MFCCD (transcription and audio) and ours (IPA). Do you agree? In this document palatal [k] is presented just like the Russian [kʲ], e.g. 機 as [kʲi] on page 2. It seems Hakka (Meixian dialect) palatalisation is close to the Japanese (and the Russian). --Anatoli T. (обсудить/вклад) 08:13, 9 March 2020 (UTC)
 * There are definitely differences between MFCCD and ours because ours is a narrower transcription (probably should be in square brackets). Thanks for that proceeding: it's the first time I've encountered something that documents both [kʲ] and [c] in Meixian. Anyhow, these are probably both allophones of /k/ under most traditional analyses. Again, I'll have to take a look at how we should present this information, i.e. phonemic (broad) vs. phonetic (narrow). — justin(r)leung { (t...) 15:46, 9 March 2020 (UTC)
 * Thank you very much. It's OK if [kʲ] and [c] are allophones. I thought they were not. I'll let you decide if any change is required. --Anatoli T. (обсудить/вклад) 21:03, 9 March 2020 (UTC)

Proposal for switching to Taiwanese Hakka Romanization System from Pha̍k-fa-sṳ
Background

Currently, Wiktionary uses Pha̍k-fa-sṳ (客語白話字) for Taiwanese Hakka. Pha̍k-fa-sṳ is a system intended for writing the Sixian dialect of Taiwanese Hakka. Literature in Pha̍k-fa-sṳ include the 客語聖經：現代台灣客語譯本. However, Pha̍k-fa-sṳ was not intended for the other varieties of Hakka spoken in Taiwan, including Hailu, Dabu, Raoping and Zhao'an.

On the other hand, Taiwan's Ministry of Education (MoE) uses its Hakka Romanization System (客家語拼音方案) for the six varieties of Hakka currently recorded by the 臺灣客家語常用詞辭典. The system differs significantly from Pha̍k-fa-sṳ and uses tone mark suffixes instead of diacritics to mark the tones of characters. The details of the system can be found here.

At the moment, the module covers only Sixian and Meixian Hakka. However, we may support other Taiwanese Hakka varieties in the module in the future. Given that they are documented by 臺灣客家語常用詞辭典 and 客語認證詞彙資料庫, they can be without a doubt considered better documented than most other varieties of Hakka spoken worldwide.

Proposal

I propose switching to using Taiwanese Hakka Romanization System as the romanization system of Taiwanese Hakka from Pha̍k-fa-sṳ. To convert all the Pha̍k-fa-sṳ forms to Hakka Romanization System forms in an automated process, a bot can be written and run. It should not be very difficult since we already have code that automatically generates Hakka Romanization System forms from Pha̍k-fa-sṳ. I do no have the skills necessary for writing the code, so I will have to ask other editors for help on this.

We will have to separately discuss how to deal with tone mark suffixes. Aside from using them as they are in the official system, we can also replace them with numbers to allow for easier input.

Advantages
 * We can directly copy and paste the romanization from 臺灣客家語常用詞辭典 and 客語認證詞彙資料庫 when we are adding the Hakka pronunciation of words instead of converting the romanization into Pha̍k-fa-sṳ first, which saves time and reduces errors.
 * Diacritics are more difficult to type. Pha̍k-fa-sṳ uses diacritics while Hakka Romanization System uses tone mark suffixes, although one can argue they are also difficult to type.
 * If we support Taiwanese Hakka varieties other than Sixian in, we will likely be using Hakka Romanization System for those varieties. It will be consistent to also use the Hakka Romanization System for Sixian Hakka.
 * Pha̍k-fa-sṳ is used only in Christian circles in Taiwan, whereas Hakka Romanization System is used by a wider audience. It is also used in material published by the Taiwan MoE. RcAlex36 (talk) 11:01, 29 May 2021 (UTC)

Disadvantages
 * It will take some effort to convert all the current Pha̍k-fa-sṳ forms to Hakka Romanization System forms. Employing a bot to do so may result in bugs which break and have to be fixed manually by editors.
 * Hakka Romanization System, being a recently invented romanization system, is less well-established than Pha̍k-fa-sṳ. Note however that Hakka Romanization System is also gaining currency in Taiwan due to promotion efforts by the Taiwan government.
 * Some people online comment that Hakka Romanization System looks ugly compared to Pha̍k-fa-sṳ.

Fellow editors, may I have your opinion on this proposal? RcAlex36 (talk) 11:01, 29 May 2021 (UTC)


 * . Maybe HRS can also be used for other Hakka dialects such as Guoxing Hakka (page 1: “bb可用於雲林詔安腔、南投國姓鄉及部分南部客家地區”). --沈澄心✉ 14:00, 29 May 2021 (UTC)


 * I don't speak Hakka, so I'll abstain. The dog2 (talk) 15:06, 29 May 2021 (UTC)


 * overall. Several steps we need to consider in addition to the things you mentioned:
 * We need to first make some functions that convert from HRS to PFS (in the case of Sixian) and from HRS to IPA.
 * We would need to convert PFS to HRS when we automatically pull from our data modules (Module:zh/data/hak-pron) when making new entries with or when showing examples with . This shouldn't be hard to implement.
 * Should we also perhaps take the chance to split Guangdong Romanization out as a separate parameter? Currently Sixian and Meixian share one note box and one parameter for audio, which makes it a little inconvenient. — justin(r)leung { (t...) 19:47, 29 May 2021 (UTC)
 * Yes, I think we should split them into separate parameters since those are separate dialects of Hakka. This along the lines of how we split Hokkien and Teochew under Min Nan. The dog2 (talk) 21:52, 29 May 2021 (UTC)
 * Yep, I think they should be split. In the future when we have the other varieties of Hakka, they should also be separate parameters. RcAlex36 (talk) 05:10, 30 May 2021 (UTC)
 * —Suzukaze-c (talk) 18:54, 30 May 2021 (UTC)