Questions on a Hakka song and its particular Hakka dialect

Question

Warning, General Background, Table of Contents

This is going to be a very long post, just like my work on this video was very long. This is why I am breaking it down into sections. It is also going to contain many questions, because I feel they are very closely related and would not benefit from being torn apart by post separation. Also, in a similar occasion, a user commented something along the lines of «Why did you split this?!».

So almost a year ago I started my blog, and sometime in August I had this crazy idea of a post which, in a long time, would come to contain all the Chinese songs I had ever come across. So I dug into my computer, and, among others, this video popped up.

The time on my hands dropped down, so the post was left dormant until, after graduating, I had more time, and added a bunch of songs. I had previously made a list of the songs I came across during my second year studying Chinese, and the above video was item 38.

So I examined it very closely, trying to figure out how to actually write the Hakka in it, both in terms of hanzi and in terms of Phak-fa-sii (the Romanization I kinda use for Hakka). This took me from last Friday till the day before yesterday (11-15/5/18) - well, I had a few other things to do these days, but still, a lot of time. Many problems emerged, and another video with higher-quality subtitles also popped up from Youtube search. This also contained an intro to the singer, which I wanted to have as section 1 below here, but the character limit forced me to move here.

After doing all that was in my power to clear them up, I come here to ask questions. The rest of the post will be divided into the following sections, though sections 2-3 need to be partially moved to a pseudo-answer because of the character limit.

First rough transcription of the sounds and character approximation;
Precise transcription of what I hear in the video, including tonal information (in the form of tone contours) for the spoken part;
Transcription with Southern Sixian Hakka contours as given at hakkadict;
A couple tweaks to the spelling based on data emerged from the previous two sections;
Comments on what seem to be Min loans in this dialect;
Comments on the matching of tones to Southern Sixian or Raoping, or neither;
Comments on tone fluctuations and possible explanations in terms of sandhi;
Final spelling and transliteration, with English translation attempt;
Questions.

With all of that stuff, I guess you can better understand why I would want to keep all the questions in one post. So let's get started, shall we? Let's go.

First rough transcription of the sounds and character approximation

I already gave above the link to the post where I describe my spelling and romanization conventions for Hakka. I would summarize the differences between Phȧk-fà-sìi and my romanization, but the character limit doesn't allow me. However, I need to point out the peculiar accent of the singer, because it induces quirks in the romanization. The speaker's accent has some of its own peculiarities; if you see a final y, it's because I hear what should be a final -i as more of a /y/ (German ü, French u, Pinyin yu); if you see -eoi or -eot, pretend it's Cantonese Jyutping, 'cause that is what I hear; -eot appears in kyeot only, where it seems the w was first fronted because of the front vowel e, and then the /kɥet/ underwent a rounding metathesis to become /kjøt/; there is also the kyun in her name, which in Hakka should be kiun, but she seems to have an accent influenced by Mandarin there; it is also funny how the initial v- is consistently /v/, except in /woj/ (会 | voi); in fact, /woy/; moreover, what should be vung-tshoi is actually fung-tshoi in her accent; but more sound mismatches will come in later sections.

The below is my first attempt at transcribing the video.

What is in '…' is apparently Min words that leaked into this subdialect.
Note that 等 is normally pronounced ten AFAIK, but when used as an -ing form particle, this subdialect has nun instead; Wiktionary reports nún as a Sixian Hakka pronunciation of 等, hence the spelling choice;
Note also that 盖 is reported by Wiktionary as Southern Sixian Hakka for "very", which perfectly fits the context and the Mandarin captions;
What is in "…" is a spelling straight out of the captions of the song; in particular:
- 该俆 is the55-hai53 in the spoken part but te-he in the song; now, 该 is ke55 according to hakakdict, which fits the tone, but not the initial; however, this "the" seems to mean "that", so the meaning matches; 俆 seems to be a nonsense choice, since it is in neither of my Hakka dictionaries, and Wiktionary says Mandarin has it as xú and Cantonese as ceoi4 or syu1, which is very far from hai/he; a better pronunciation match is 系, which would make the two words combine into "that is" or "that was", but "that time" seems to actually be the meaning, given "hip-siong e te-hai"; as far as this goes, 俆 doesn't match the meaning either; but I still keep it because of the captions;
- 㧡 is given as khai (with the appropriate tone) in all dialects on hakkadict, and Wiktionary says even Middle Chinese had the -i final, so how come we have khan?
- 匍背 matches the sounds phu-poi, but the meaning should be "support", which suggests the fist character should mean "support", instead of being the phonetic phu in Putaoya; indeed, the founder of Hakka Verse (Anton Xie a.k.a. 谢可为 | Chia Khobui, henceforth Khobui sifu) says there exists a word 扶背 in his subdialect, which he rarely used, pronounced fu-poi, with the meaning I expect; given that 扶 has an alternate pronunciation which is phu11 in Southern Sixian Hakka, I guess that is the way to go;
- 瓜仔 would be kwa-e, but what is heard is kwabe, which I could analyze as kwap-e or kwa-pe;
The phrase 该彼日 is my spelling guess; I hear "te pi(t) nyit", it's supposed to mean "that day". "nyit" must be that character, the rest seem to be "that"; the captions give 个晡日; apart from the problem with te, which should be the same as the-hai IMO, we have that 晡 is pu, not pi(t), unless we assume this is not just a Min-Hakka cross but a cross of two Hakka varieties (southern Sixian being the dominant one, with Dabu for the "pi") and Min (see loans in '…'), plus some Cantonese influence in her accent (oi comes out like Cantonese eoi and ket sounded like Cantonese gyeot); coming to 个, we've seen that is e in this subdialect, and I don't think the t could be a -t from a tiam-lit, and besides, 礼 is not lit AFAIK, but li

It is also to be noted that tiam-li does not match the first character, which is apparently tien even in southern Sixian.

A couple tweaks to the character spelling

First of all, Mandarin 去年 matches Hakka 旧年, which solves the mismatch problem of 去 | hi55 to the khiu55 of the video, given 旧 is kiu55;
I originally thought the "nge44" was a ngai (I) where the -ai had monophthongized to /e/, but given it has a high tone whereas ngai has a low tone, and ngi (you) has a high tone, and the captions supporte the ngi, I believe it's either a weird pronunciation in the dialect, or just in her personal idiolect; or perhaps it is a nya where the -ya fused to /e/; either way, the 涯 there becomes a 你; oh BTW, another 你 further below is pronounced ng; I would have expected the different pronunciations of 你 to be from different dialects, not mixed in the same dialect, but what do you know :);
The m in khon m to is hard to hear, but it is definitely there, so that part is 看毋着;
Having heard a -t in the et, I posit it's tet without the initial (cfr. ke -> e and ki -> i), and spell it 得;
The "nan" in "Nga nan piong et lok sim" seems to mean "just, only", like 才; the best thing I could come up with in terms of characters was 另, which can be nang55, matching the tone but not the final; the meaning doesn't match;
Also, all those "nga"s are probably "ngai"s with a fleeting -i, given the tone mismatch; so they change from 吾 to 涯;
I used the Mandaring spelling 美浓, but I should have used the Hakka one 弥浓, given that the Mandarin one features a tone mismatch;
I am still unsure whether it is "A55 m53-ko44" or "Am53-ko44"; the former, my first choice, was to match the captions' 不过, but the latter convinces me more in terms of sound, though no reference I have mentions such a word meaning 不过; spelt the way I did, it would mean more of "Nights passed [until etc.]" rather than "But [until etc.]";
"Yin-wi yi" is actually "Yin-vi", so no 伊 there;
As mentioned above, I write phu-poi as 扶背 rather than 匍背;
In view of the "he" in the song, "te-he" is not the the-hai from the spoken part, but just the te (spelt accordingly) plus a 系 | he.

Comments on the apparent Min loans

Two words I deemed Min loans may not be entirely such; one is i11, which is the actual Southern Sixian pronoun, spelt 佢, according to hakkadict; the other one is e55, which seems to resonate with the alternate Hakka pronunciation ê on Wiktionary, though with a tone mismatch; so I assume the tone was a loan from the identical-sounding Min 的, a tone taken up with Sandhi included (and misheard too because mainstream Taiwanese AFAIK sandhies ê to /e21/, not /e11/ and much less /e55/; whoops, tone mismatch still; well hello Shao'an "een11 e55" where the 个 should be gai31! OK, question to be asked about this;
We then have that preposition, which I originally heard "ti" and mapped to Min 佇 | tī, which even matched the tone if sandhi is disregarded; however, it turns out it's actually thi11, which has an extra aspiration w.r.t. Min, and a tone mismatch w.r.t. the thî reported as an alternate pronunciation of 在 by Wiktionary;
Then we have the phrase 如按呢 | lú án-ne, which I assumed to have been come up with in Min, and then loaned as is into this subdialect; the lú was then analyzed as an intensifier, compared to 就, and assimilated to it, or at least that is how I explain the "lu then-nun" and "mi nyit thien mo kong lu oi hong"; if sandhi is accounted for, only the tone of án doesn't match; however, sandhi can be used to justify that: lu55 an55 ne11 means a jump from an55 to ne11, better smooth it out by saying an53 ne22; note that I am sandhi-ing ne as well, from a high tone to a low one;
Then we have the preposition 共 | kā, loaned without sandhi, or perhaps it's just 佮 | kak, the Hakka word, with the final -k gone;
Finally, we have the "ma" dubbed 也 in the captions, which I would have assumed to be the second part of "a-ma", but then came to associate with 嘛 | mā in Min, which unfortunately doesn't match the tone 23 of the video; I ended up assuming the Min loan was reassigned to the tone of the Hakka 也, which happens to be yâ (ya24) in Southern Sixian;
Oh and we have a sound-only loan, perhaps, in 故 being read ko instead of ku, but the tone is Southern Sixian.

Comments on tones

Raoping or Southern Sixian?

Many tones I hear in the video match Southern Sixian, as expected; many more do not. Let me say some things about the mismatches and perhaps about a few of the fuzzy matches. I abbreviate Southern Sixian as NSX (N for Nan) and Raoping as RP.

三 is sam24 in NSX and sam11 (perfect match) in RP;
月 has tone contour 5 in both NSX and RP, but I hear it as ngiat2, which would match the Hailu tone;
阿妈 is a24-ma24 in NSX, but RP a11-ma11 matches better; however, I have reason to believe ma should be a ma55, which matches neither;
每 is mi24 in NSX and mui11 in RP, so it seems we have a tone loan;
日 is ngit2 in both NSX and RP, but the big jump from mi2 to ngit makes it clear we should have ngit5 instead;
无 is mo11 in NSX and mo55 in RP, none of which matches the mo53 of the video, which would match Shao'an;
光 is kong24 in NSX and kong11 in RP, so it seems the tone in the video matches RP;
出门 is chut2-mun55 in RP, which matches no tone in the video, so it's either NSX or guess;
担竿 is tam53-kon11 in RP, no match;
No tone reported for 㧡 matches the khan44 of the video;
番薯 is fan11-shu55 in RP, no match;
庄 is tsong11 in RP, match;
一 is rhit2 in RP, no match;
到 is to53 in RP, worse match than NSX to55;
中门 is tsung11-mun55 in RP, so no match; but 问 would give mun53, a fuzzy match; I'll leave that doubtful because the name is doubtful;
项 is hong24 in RP and hong55 in NSX, so RP matches the video's hong34 better;
会 is voi24 in RP, perfect match;
君 is kiun11 in RP, matching the video's tone;
眼珠 is ngian41-tsu112 in the video, ngan53-cu11 in RP, ngian33-tsu24 in NSX, so RP matches both syllables better;
A couple of "e"s in this transcription are exclamative particles, which I spell as Khobui sifu suggested, and believe are of low tone, e11; such a tone is indeed found in the ai11 of RP, supposedly spelled 唉, which is e in NSX;
所以 is so53-yi11 in RP, matching both tones better; I posit the yi had a 11 tone that sandhied with a so53 to create a continuous drop rather than drop gradually from 5 to 3 and then jump to 1; and then the 3 shifted to 4 because whatever :); the kho22-yi21 further down seems to support this claim;
希望 is hi11-mong24 in RP, a fuzzy match;
可以 probably has NSX kho31 combined with RP yi11;
考上 probably results from a sandhi of the combination of NSX khau31 and song11 from RP;
大学 is thai24-hok5 in RP, a match for syllable 1 and a mismatch for syllable 2; maybe the low tone on hok is a Cantonese influence (daai6-hok6)?
得 and 落 both have the wrong tone in RP;
心 is sim11 in RP, match;
因为 seems to have NSX sounds and RP tone (rhin11) on syllable 1, and NSX vi55 for syllable 2;
这 would be li53 in RP, no match, and sounds don't match either;
句 and 话 have level tones (22) in the video, but RP has non-level 53s, so no match;
As seen above, 中 is cung11 in RP, sound match;
暗 is am53 in RP, matching the video better;
典礼 is tien53-li11 in RP, fuzzy match;
Neither the chim55 of RP nor the tshim11 of NSX match the tshim31 of the video; no sandhi can apply as there is a pause right before it;
Hip2-siong55 (NSX) vs. hip2-siong53 (RP): syllable 2 is NSX, syllable 1 matches neither;
欢喜 is fan11-hi53 in RP, matching the tones of the video better, especially if we posit a sandhi from he53 e55 to hi55 e55, and that the reference level shifted making me take all those 5s for 3s;
感受 is kam53-shiu24 in RP, matching syllable 2 better but syllable 1 just as bad;
最重要 matches nothing, except 最 matches RP, or a sandhied NSX;
phan34-fa32 is now a tone fuzzy match, and as for the missing stop coda, -p is definitely not there because the lips do not close, but a -t may be there, cfr. Shao'an fat24.

Getting creative with sandhi

Some sandhi arguments to explain some tone problems are already in the previous sections. The rest follows.

I justify khon42 m22 to22 e11 as a sandhi for khon55 m11 to31 e11, the NSX tones;
The negative element mo occurs as mo53, mo22, mo31, mo42; I venture the first is a sandhied form, with thiên pushing up, mo53 starting from just above the end of thiên (i.e. at the peak of the up-and-down), and mo53 and kong21 making a continuous descent to the 1 which they both should be on; the mo31 may be sandhied by being between ma23 and yit2, and the mo42 may be sandhied by being between tsai44 and phan34; therefore, I posit the actual tone is mo22 or mo11, which matches or fuzzy matches the NSX mo11;
I posit the an53 in lu55-an53-ne22 is sandhied to connect the 55 to the 11 (or rather 22), and was originally 55, matching Min with sandhi accounted for;
tap3-yang34 is probably a sandhi of tap3-yang44, fuzzy match to tap3-yang55 which is NSX sounds and RP tones;
khau33 to332 is probably sandhi for khau31 to32, a fuzzy match to NSX;
The ngai41 is probably sandhied with the preceding to44;
I posit tui55 nga53 loi31 is a big sandhi for tui55 nga(i)11 loi11;
khon42 m22 to21 e11 is probably a sandhi for khon53 m11 to31 e11, or perhaps even khon55 m11 to31 e11 which matches NSX perfectly (whereas khon53 would be RP tone);
I partially retract my claims about the tones of 欢喜: while the first syllable is indeed in RP tone (fon11), the second one is hi33, which, since followed by a e33, is justified as sandhi from hi31 rather than hi53 (i.e. NSX rather than RP);

Tone fluctuations: what are those tones?

This preposition thi occurs twice, once as thi33 and once as thi11; as such, I argue its tone is thi11, and write it thì accordingly; this matches the Min tone of tī, low level (when out of sandhi);
The particle e occurs sometimes as e33, sometimes as e44, sometimes as e34; I argue that its tone is meant to be level, and the e34 are accidental inflections in the pitch; given the e44, I conclude it's e55, and write it e accordingly; the Min tone would be ê, which is what the Wiktionary reports, but I prefer to derive it from Hakka ke55 by way of cutting off the initial, as happened with ki11, which is reported as i11 in NSX by hakka.dict.edu.tw;
I believe the nun11 reading for 等 is an accident, and the correct tone is nun32, or its fuzzy match nun31, which matches the Wiktionary-reported alternate NSX sound of 等;
In a-ma, the a is consistently a11, but the ma appears as ma33, ma34, ma55, ma24, ma33, ma33; given that I cannot produce any good sandhi arguments for the level forms, the majority wins, so the tone will be 55;
The hong seemed to match RP better, but the preceding syllable is lu34, an accidental inflection, so hong34 is that way too, and the actual tone is hong44 or hong55;
keoi34 is probably an accidental inflection plus strange accent form of koi33, which fuzzy matches koi55, which is NSX;
Why the typical ngai11 changed to ngai31 in the penultimate line, I have no clue; maybe it was just stretched out and an inflection in pitch came natural.

Final transcription and character rendition

It is now time for the final rendition. Here are my diacricits:

No diacritic 55
ˆ 24
´ 31
˝ 53
` 11

Diacritics for checked tones:

No diacritic 2
˙ 5

Superscripts:

¹ Min tone
² Raoping tone
³ Tone base solely on video

口白：
'佇'旧年个三月份 | Thì¹ khiu-ngiàn e sàm² ngiat³-fun
涯又梦着涯个阿妈 | Ngài yu mung tó ngài e à²-ma³
"该俆"系涯盖细盖细个时节 | The³ ha̋y³ he ngài koy se koy sei e sìi-tsiet
每日天无光 | Mì² ngıt³ thiên mò kòng²
如跈等阿妈共下出门去卖菜 | Lu¹ thèn-nún à²-ma³ khiung-ha tshu̇t³-mùn hi mai tshoy
阿妈用担竿"㧡"等番薯番薯叶 | À²-ma³ yung tâm³-kón³ khan³-nún fàn³-su³ fân-sú³ yap
"㧡"等瓮菜 | Khan³-nún fung-tshoy
'如按呢'从弥浓庄 | Lu¹-an¹-ne¹ tshùng Mì-nùng tsòng²
一路行行到中门 | Yıt³-lu-hàng hàng to Tsùng²-műún³
'佇'路项阿妈佢会'共'涯讲 | Thì¹ lu hong à²-ma³ ì wôy² ká¹ ài kóng
姿君涯眼珠看毋着吔 | Tsîi-kyùn² nge³ ngia̋n²-tsù² khőn² m̀ tó è³
故所以涯希望你可以考上大学 | Ko-ső²-yì² ngài hì²-mông² ng̀ khó-yì² kháu-sòng² thâi²-hok³
吾另放得落心 | Ngà³ nam piong ėt³ lok³ sìm²
因为阿妈哩句话 | Yìn²-vi à²-ma³ yà³ kì³ fà³
涯盖搭營 | Ngài keoi tap-yàng²
来考着中国文化大学 | Leòi kháu tó Tsùng²-kyeot vùn-fa Thâi²-hok³
暗过到涯毕业典礼该毕日 | A̋m²-ko to ngài pit-ngiȧp tia̋m²-lì² te pit ngıt³
吾路寻无着阿妈佢个扶背 | Ngà³ lu tshím³ mò tó à²-ma³ ì e phù-poi
翕相个"该俆" | Hıp³-siong e te ha̋i³
吾'嘛'无一种欢喜个感受 | Ngà³ mâ³ mò yit tsúng fòn²-hí e kám-sû²
因为对吾来讲 | Yìn²-vi tui ngà³ lòi kóng
最重要最重要个人 | Tsui tshùng³-yàu³ tsui tshùng³-yàu³ e ngìn
涯再无辦法看着 | Ngài tsai mò phân²-fa(t)* khon tó *fà or fat
佢个扶背 | Ì e phù-poi

有一个暗晡头 | Yû yit ke am-pû-thèu
涯发梦梦着涯个阿妈 | Ngài pot-mung mung tó ngài e à²-ma³
"该"系盖久盖久吔以前 | Te³ he koi kíu koi kíu è³ yit-tshien
涯还盖细盖细个时节 | Ngài hàn koi se koi sei e sìi-tsiet
涯还盖细盖细个时节 | Ngài hàn koi se koi sei e sìi-tsiet
阿妈每日用担竿 | À²-ma³ mì² ngıt³ yung tâm³-kón³
"㧡"着番薯叶"㧡"瓮菜 | Khan³ tó fân-sú³ yap khan³ fung tshoi
带涯共下去卖菜 | Tai ngài khiung-ha hi mai tshoy
从弥浓行到中门 | Tshùng Mì-nùng hàng to Tsùng²-műún³
有么人爱买无 | Yû mak-ngìn oi mâi mò?
有么人爱买无 | Yû mak-ngìn oi mâi mò?
阿妈希望涯读大学 | À²-ma³ hì²-mông² ngài thu̇k thâi²-hok³
毋使像佢爱去卖菜 | M̀-síi tshiong ì oi hi mai tshoi
每日天无光如爱行 | Mì² ngıt³ thiên mò kòng² lu¹ oi hòng
"㧡"菜"㧡"啊变扶背 | Khan³ tshoi khan³ ah pien pù-poi
"㧡"菜"㧡"啊变扶背 | Khan³ tshoi khan³ ah pien pù-poi
有一个暗晡头 | Yû yit ke am-pû-thèu
涯发梦梦着涯个阿妈 | Ngài pot-mung mung tó ngài e à²-ma³
"该俆"盖久盖久吔以前 | Te³ he̋³ koi kíu koi kíu è³ yit-tshien
涯还盖细盖细个时节 | Ngài hàn koi se koi sei e sìi-tsiet
涯还盖细盖细个时节 | Ngài hàn koi se koi sei e sìi-tsiet
阿妈每日用担竿 | À²-ma³ mì² ngıt³ yung tâm³-kón³
"㧡"着瓜仔"㧡"吊菜 | Khan³ tó kwâ[p]-é khan³ tiau-tshoi
带涯共下去卖菜 | Tai ngài khiung-ha hi mai tshoy
从弥浓行到中门 | Tshùng Mì-nùng hàng to Tsùng²-műún³
有么人爱买无 | Yû mak-ngìn oi mâi mò?
有么人爱买无 | Yû mak-ngìn oi mâi mò?
一年一年过一年 | Yıt³ ngiàn yıt³ ngiàn ko yıt³ ngiàn
涯从大学爱出社会 | Ngài tshùng thâi²-hok³ oi tshu̇t³ sa-fi
毕业典礼该晡日 | Pit-ngiȧp tia̋m²-lì² te pit ngit
寻无阿妈个扶背 | Tshím³ mò à²-ma³ e pù-poi
看无阿妈个扶背 | Khőn² mò à²-ma³ e pù-poi

Questions

Am I right in breaking that "the-hai" into "the", some form of "that", and "hai", something like "time, moment"? What are the tones, and how should I spell it?
Is that "khan" actually a variant of 㧡 with the matching tone?
Why is 瓜仔 heard as kwap-e / kwa-pe? Is that stray consonant an evolution of a glottal stop to separate the two syllables?
Any comments on what I wrote about Min loans, Raoping tonal loans, creative sandhies and tone fluctuations?
That "te pit nyit" part really seems to have a "pit", so I would write 该毕日, with "te" being that famous "the"; should I really be writing 晡日 when that is usually pu-nyit and the pu part is pu in am-pu-theu in the song? Am I right in seeing the "the" from "the-hai" in this phrase?
Where does that "tiam" come from, when all Chinese dialects I can see, and even Middle Chinese, have -ien?
Is that nge44 actually a 你, or maybe a 涯 as I first thought, or something else?
What is going on with this loss of initials?
Is it "a m-ko" or "am-ko"? And does it mean "However"?
Am I right in reading that "nam" as an equivalent of 才? How should I spell it in characters, and what is its tone?
The place name is in the captions as 中壇, which simplifies to 中坛, and should be Tsung-tham in Hakka; the video has Tsung-mun, which would suggest 中门, except 门 doesn't have a matching tone in NSX or in RP, and 问, which has a matching tone in RP, seems inappropriate here; I asked on Quora, and @justinrleung gave me this link, where it seems the place name was once spelled 中坛 and pronounced Tsun'en and later Tsun'un, transliterating the kana, which would seem to be Tsung-ien and Tsung-wun respectively, but old people now have either Tsung-tham or Tsung-iun; so what is going on here? What are the tones of that name? How should I spell it ij characters?
The possessive particle should be ke55 in NSX and kai53 in RP, but I found it as e55 here; Wiktionary has an example giving it as ê, aka e24, spelled 的; what is then the right tone and character for this?

Update

As @justinrleung points out in a comment, I misread 係 as 俆 in the captions. This doesn't solve any problem because it's supposed to be he, not hai, and indeed right after the first the-hai we have a he written that way. I will leave things as they are. If the spelling is correct, this could be a loan from another subdialect or even another dialect family.

Update 2

As justinrleung answered on Quora, the answer to question 1 is that the "the-hai"/"te-hai" should be 该下仔 | ke-ha-é, a Southern Sixian form of the 该下 reported on hakkadict, and the initial t- or th- is an accidental aspiration plus an idiolectal shift in the place of articulation that is not documented anywhere and that justinrleung personally has never heard. So this leaves "only" 11 questions here.

Update 3

About question 10, I just thought it could be 另毋 | nang m̀ | otherwise not? Oh my, that tone though… completely mismatched, it should be 55 11 and it's heard as 45…

Update 4

I isolated spelling questions in the pseudo-answer, and the one about phù-poi was answered by @justinrleung on Quora. Turns out it's supposed to be 匍背 | hunchbacked.

Final UPDATE

This Q has become a big mess of a post, with a huge body; a huge pseudoanswer (PA 1) with the sections 2-3 removed above, some isolated Qs, some of the answers to those; another huge pseudoanswer (PA 2) with tone stuff; numerous offshoots to Quora. Now I clearly cannot possibly include all that info here, but I want to at least sum up what's left to answer. First of all, a few comments on this body's questions:

All the tone Qs are superseded by PA 2, which I'll extract a revised transliteration from, whether here or in PA 1;
Most Qs about Min loans have been answered either in the updates above or in PA 1; only a couple remained, both included in PA 2 and outsourced to Quora; this takes care of Q 4 above;
Qs 1 5 7 9 10 and 12 above, along with Qs 1 2 5 6 8 and 11 in PA 1, are answered either in PA 1, in its comments, in an update above, or in comments to the question;
Q 8 here has a simple answer: erosion of common words; ki->i, ke->e, cang->nang, tu->lu, tshiu->lu, tet->et;
For Q 4 ("ma") and Q 7 ("ka") in PA 1, see Min loan Qs item above;
For all other Qs, see list at end of PA 1.

29.329 characters

Wow, this is really a very long question, but kudos for your efforts. It'll probably take me a while before I can formula a decent answer to this. Before I give a full answer, there are some obvious errors: (1) You have mistaken 係 (traditional form of 系) for 俆 in the captions. (2) You don't have an 11th question, do you? — justinrleung, May 18, 2018 at 15:11
@justinrleung (1) I realized that at some point but forgot to mention it in the post. Will make an update. (2) Actually, I have at least two more: 11) that place name and that character which should be the simplified form of tán but was used before simplified with strange sound, and 12) "OK, question to be asked about that" in the post (item 1 in Min loans section). Will add them soon. — MickG, May 18, 2018 at 15:23
Actually one more thing... ê is not given as a Hakka pronunciation at the Wiktionary entry for 的. — justinrleung, May 18, 2018 at 15:28
Ah ok, I added that example in Wiktionary, and now that I'm more familiar with the Meinong accent, I think I got the tone wrong (you can't really be sure of the tones in singing). I've corrected the entry accordingly. — justinrleung, May 18, 2018 at 15:44
The tones are definitely not from the song, but based on the actual words (with the help of dictionaries and other sources, like videos) and how they would be said if not sung. For example, there isn’t really a real way to distinguish Mì-nùng from Mî-nùng, so either could work. — justinrleung, May 18, 2018 at 16:01

MickG · Accepted Answer · 2019-01-15 22:41:54Z

Here is the pseudo-answer I promised to include the info that doesn't fit in the question body.

Precise transcription of what I hear

So I picked this video, and tore it apart in 0.25 speed to figure out exactly how the tones and sounds went. And this is my transcription.

Thi33 khiu55-ngian51 e334 sam11 ngiat2-fun33
Ngai24 yu55 mung55 to331 ngai11 e33 a11-ma33
The55 hai51 he33 ngai21 koy33 se334 koy33 sei33 e34 sii21-tsiet2
Mi22 ngit51 thien24 mo53 kong21
Lu55 then11-nun11 a11-ma34 khiung44-ha44 tshut4-mun21 hi33 mai33 tshoi334
A11-ma55 yung55 tam35-kon31 khan44-nun32 fan11-su4 fan24-su31-yap34
Khan44-nun32 fung33-tshoi334
Lu55-an53-ne22 tshung21 Mi11-nung11-tsong11
Yit5-lu45 hang21 hang21 to34 Tsung214-mun51
Thi11 lu34 hong34 a11-ma24 i11 woi24 ka31 ai11 kong31
Tsii34-kyun11 nge44 ngian41-tsu112 khon42 m22 to22 e11
Ko55-so54-yi41 ngai12 hi21-mong35 ng22 kho22-yi21 khau33 song31 thai33-hok3
Nga22 nam45 piong5t et4 lok3 sim11
Yin22-vi55 a11-ma33 ya22 ki22 fa223
Ngai31 keoi34 tap3-yang34
Loi22 khau33 to332 Tsung22-kyeot3 vun21-fa34 thai34-hok34
Am53-ko44 to44 ngai41 pit3-ngiap4 tiam43-li11 te33 pit32 ngit3
Nga33 lu55 tshim31 mo22 to221 a22-ma33 i11 e33 phu11-poi334
Hip4-siong55 e55 te44-hai41
Nga11 ma23 mo31 yit2 tsung21 fon21-hi33 e33 kam32-su34
Yin22-vi55 tui55 nga53 loi31 kong21
Tsui53 tshung22-yau22 tsui32 tshung22-yau22 e22 ngin21
Ngai31 tsai44 mo42 phan34-fa32 khon44 to31
I22 e44 phu21-poi33

Yu yit ke am-pu-theu
Ngai pot-mung mung to ngai e a-ma
Te-he koi kiu koi kiu e yit-tshien
Ngai han koi se koi sei e sii-tsiet
Ngai han koi se koi sei e sii-tsiet
A-ma mi ngit yung tam-kon
Khan to fa-su yap kan fung tshoi
Tai ngai khiung-ha hi mai tshoy
Tshung Mi-nung hang to Tsung-mun
Yiu mak-ngin oi mai mo?
Yiu mak-ngin oi mai mo?
A-ma hi-mong ngai thuk thai-hok
M-sii tshiong i oi hi mai tshoi
Mi ngit thien mo kong lu oi hong
Khan tshoi khan a pien pu-poi
Khan tshoi khan a pien pu-poi
Yu yit ke am-phu-theu
Ngai pot-mung mung to ngai e a-ma
Te-he koi kiu koi kiu e yit-tshien
Ngai han koi se koi sei e sii-tsiet
Ngai han koi se koi sei e sii-tsiet
A-ma mi ngit yung tam-kon
Khan to kwap-e khan tiau-tshoi
Tai ngai khiung-ha hi mai tshoi
Tshung Mi-nung hang to Tsung-mun
Yiu mak-ngin oi mai mo?
Yiu mak-ngin oi mai mo?
Yit ngian yit ngian ko yit ngian
Ngai tshung thai-hok oi tshut sa-fi
Pit-ngiap tiam-li te pit ngit
Tshim mo a-ma e pu-poi
Khon mo a-ma e pu-poi

Transcription with Southern Sixian Hakka tones as found in hakkadict

The sounds make this seem to be Southern Sixian Hakka, and the mention of Meinong also points in that direction, so I proceeded to give a transcription with the Southern Sixian Hakka tones from hakkadict, leaving those that matched perfectly unmarked, marking those that made fuzzy matches with ', and marking mismatches with ". I also put asterisks where the sounds didn't match perfectly and for other side notes, with explanations at line ends.

Thi khiu55*-ngian11 (k)e55' sam24'' ngiat5''-fun55' (actually hi55)
Ngai11'' yu55 mung55 to31 ngai11 (k)e55' a24''-ma24''
The55* hay he55' ngai11 koy55' se55' koy55' sei55' (k)e55' sii11-tsiet2 (actually ke55)
Mi24'' ngit2'' thien24 mo11'' kong24''
Lu then11-nun31''* a24''-ma24' khiung55-ha55 tshut2''-mun11' hi55' mai55' tshoy55' (actually nen31)
A24''-ma24'' yung55 tam55''-kon24'' khan24''-nun31* fan24''-su11' fan24-su11'' yap5‘ (actually khai24-ten31)
Khan24‘’-nun31* fung55' tshoy55' (actually khai24-ten31)
Lu-an-ne tshung11'* Mi11-nung11 tsong24'' (actually tshiung11)
Yit2''-lu55'-hang11' hang11' to55'' Tsung24''-mun11''* (问 would give better matches with Raoping tones)
Thi lu55‘ hong55'' a24''-ma24 i11 woy55'' ka ai11 kong31
Tsii24'-kyun24'' nge11'' ngian33''-tsu24'' khon55'' to31'' e24''* (actually spelled 唉)
Ko55*-so31'-yi24'' ngai11' hi24''-mong55'' ng11' kho31''-yi24'' khau31''-song24'' thai55'-hok5'' (actually ku55)
Nga24'' nam55'* piong55 (t)et2'' lok5'' sim24'' (actually nang55)
Yin24''-vi55 a24''-ma24'' ya31'' ki55'' fa55''
Ngai11'' keoi55'' tap2'-yang11''
Leoi11' khau31'' to31' Tsung24''-kyeot2'* vun11'-fa55' Thai55''-hok5'' (actually kwet2)
Am55''-ko55' to55' ngai11'' pit2'-ngiap5' tiam31''-li24'' te55''* pit2' ngit2' (actually ke55) (pit2' from 毕 or pu24'' from 晡)
Nga24'' lu55 tshim11'' mo11' to31' a24''-ma24'' i11 (k)e55' phu11-poi55'
Hip2''-siong55 (k)e55 te55'* hai (actually ke55)
Nga24'' ma mo11'' yit2 tsung31' fon24''-hi31'' (k)e55' kam31'-su55'' Yin24''-vi55 tui55 nga24'' loi11'' kong31'
Tsui55'' tshung55''-yau55''* tsui55'' tshung55''-yau55'' (k)e55' ngin11' (actually yeu55)
Ngai11'' tsai55' mo11'' phan55-fa* khon55' to31 (fap2 vs fa43)
I11' (k)e55' phu11'-poi55'

Yu24 yit2 ke55 am55-pu24-theu11
Ngai11 pot2-mung55 mung55 to31 ngai11 (k)e55 a24-ma24
Te-he ko55i kiu31 koi55 kiu31 e yi[t]24-tshien11
Ngai11 han11* koi55 se55 koi55 sei55 (k)e55 sii11-tsiet2 (actually van11)
Ngai11 han11* koi55 se55 koi55 sei55 (k)e55 sii11-tsiet2 (actually van11)
A24-ma24 mi24 ngit2 yung55 tam55-kon24
Khan24 to31 fan24-su11 yap5 khan24 fung55 tshoi55
Tai55 ngai11 khiung55-ha55 hi55 mai55 tshoy55
Tsh(i)ung11 Mi24-nung11 hang11 to55 Tsung24-mun11
Yu24 ma[k]31-ngin11 oi55 mai24 mo11?
Yu24 ma[k]31-ngin11 oi55 mai24 mo11?
A24-ma24 hi24-mong55 ngai11 thuk5 thai55-hok5
M11-sii31 tshiong55 i11 oi55 hi55 mai55 tshoi55
Mi24 ngit2 thien24 mo53 kong24 lu oi55 hong11
Khan24 tshoi55 khan24 a55 pien55 p(h)u11-poi55
Khan24 tshoi55 khan24 a55 pien55 p(h)u11-poi55
Yu24 yit2 ke55 am55-pu24-theu11
Ngai11 pot2-mung55 mung55 to31 ngai11 (k)e55 a24-ma24
Te-he ko55i kiu31 koi55 kiu31 e yi[t]24-tshien11
Ngai11 han11* koi55 se55 koi55 sei55 (k)e55 sii11-tsiet2 (actually van11)
Ngai11 han11* koi55 se55 koi55 sei55 (k)e55 sii11-tsiet2 (actually van11)
A24-ma24 mi24 ngit2 yung55 tam55-kon24
Khan24 to31 kwa[p]24-e31 khan24 tiau55-tshoi55
Tai55 ngai11 khiung55-ha55 hi55 mai55 tshoy55
Tshung11 Mi24-nung11 hang11 to55 Tsung24-mun11
Yu24 ma[k]31-ngin11 oi55 mai24 mo24?
Yu24 ma[k]31-ngin11 oi55 mai24 mo24?
Yit2 ngian11 yit2 ngian11 ko55 yit2 ngian11
Ngai11 tsh(i)ung11 thai55-hok5 oi55 tshut2 sa55-fi55
Pit2-ngiap5 tiam31-li24 te pit ngit (see spoken part about te pit ngit)
Tshim11 mo53 a24-ma24 (k)e55 p(h)u11-poi55
Khon55 mo53 a24-ma24 (k)e55 p(h)u11-poi55

Isolating some questions

For reasons I won't go into now, I wish to isolate a few questions from above down here, with the rest of the questions being a bunch of tone and sound mismatches.

In the first line of the spoken part, I'm pretty sure "thi" is an equivalent of 在, but the Hakka sounds for this character given in the Wiktionary include a "ti" sound, that is unaspirated with high tone, and a "thî", aspirated with raising tone, and what I hear is a low tone, which would be thì. My original explanation as a Min loan doesn't work, because that one is tī, unaspirated. So am I right about the meaning, and what is up with the tone?
One line of the spoken part features what I hear as either "Am ko to" or "A m ko to". In the pseudo-answer I went for the former, giving Am53-ko44 to44 | 暗过到, but IIRC the captions suggested the meaning of "But", which is why I thought of "A m ko", that is "A", exclamative, and "M-ko", but. Which one is the case, and how do I spell and transliterate it?
In the same line, and in the analogue line of the song, we have 毕业典礼该晡日, or at least that should be it, except a) 典礼 oddly sounds "tiam-li" whereas any source gives me "tien-li", and b) 晡 sounds "pit" instead of "pu", whereas in am-pu-theu it is indeed pu. So are those characters correct, and if so what is up with the sounds, and if not how should I spell this?
There is a line sounding "Nga ma mo yit tsung fon-hi e kam-su", which should be ？？无一种欢喜个感受, the first two characters being however you prefer to write "nga", the should-be possessive form of "ngai", and the second one I don't know. It is odd that nga is used as non-possessive, but that happens elsewhere in the video. The second character should mean something like 也 or 都, but "ma" is not something I know from Hakka in that meaning, I know of a "me" but it's not the one used here, or doesn't seem to. Min has a word 嘛 | má for this, which is why I thought of a Min loan. Is it so, or did Hakka independently develop a "ma" for this meaning? And how should I spell it and transliterate it?
In one line, there is something like "Nga nam piong et lok sim". "nam" seems to have to mean 才. I thought of this recently: maybe it's 另毋 | nang m | otherwise not. Is that so?
There appears to be a word "lu" which is mapped to 就 by the captions. My thought was 如按呢 could be a Min phrase for "just this way" or "just", loaned as is into Hakka, and then cut down to just "lu" in some cases. Could that be? And in any case, how do I spell the "lu" in "lu then-nun a-ma" and the "Lu-an-ne", and how do I transliterate them? Note that this question also tangentially touches the "Nga lu tshim mo to", where the "lu" may be the same problematic lu.
In one line, we have "I woi ka ai kong". This should mean "She would tell me" (used to tell me). I is usually "ki", third-person pronoun. woi is 会. kong is 讲. "ai" should be "ngai", "I", which somehow lost its ng-. The "ka" sounds like it should be "kak", which I know can be used this way, but where is that -k? Could that be a Min influence, where we have 共 | kā? And how should I spell and transliterate this preposition?
There is one line sounding like "Tsii-kyun, nge ngian-tsu khon m to e", where "nge" should mean "my", but that is "ngai" or "nga", so what is up?
What is up with the kwa-e sounding like kwap-e in the song? Is it a glottal stop thing?
In the penultimate spoken line, the wannabe 办法 sounds like phan-fa or phan-fat, but it should be phan-fap, is it me mishearing or is something else up? Idiolectal thing, perhaps?
Is the title 扶背 | phù-poi | support, as I thought, or 匍背 | phû-poi | hunchback, as the captions suggest? If the former, what does the line "㧡"菜"㧡"啊变扶背 mean? And if the latter, what is up with the tone? And in any case, is 匍背 a word for "hunchback" in any Chinese dialects, and if so how is it pronounced in them?

Update

@justinrleung produced a source for 匍背 | phù-poi | huchbacked on Quora. It is a Southern Sixian Hakka - only word.

Concerning question 1, I looked up 在 in hakkadict, and the closest pronunciation I could find was «di55/di11/coi33/di53/da55/te24» or «du55/di11/to33/du53/da55/te24» (sounds for Sixian/Hailu/Dabu/Raoping/Shaoan/Southern Sixian), both of which are thê in Southern Sixian. The tone is mismatched, and the vowel also, but could this be it? Alternately, in Wiktionary, I found a Meixian synonym 喺, which oddly and unfortunately has no Hakka sound in the entry, could that be it?

Concerning question 4, I know of a particle "me" meaning more or less 也, though I don't remember the tone. Could this be a variant of that?

Update 2

I asekd 1 and 2, and @justinrleung answered both:

That is indeed a thî, the aspiration is there, the tone doesn't match, either because the Meinong dialect actually pronounces that tone (which is the Yin Ping tone) as 33 instead of 24, or because of Sandhi, though the 24->33 sandhi for the Yin Ping tone appears to be a Northern Sixian thing, and this song appears to be Meinong, so I'd favor the Meinong Yin Ping explanation, given that this song explicitly says the singer and her A-ma (grandma) set off from Meinong at dawn and went all the way to "Tsung-mun" (aka Zhongtan, but that pronunciation is a whole 'nother question) to sell the vegetables her A-ma carried with tam-kon (shoulder-poles).
It is indeed A-m̀-ko I'm hearing, where the a- is essentially meaningless (this is equivalent to just m̀-ko), and is probably an influence of Hokkien 猶毋過, which among its various pronunciations has á-m̄-kò.

Speaking of tones, Justin provided me with A Comparative Study of the Tones of the Sixian Hakka Accent in the Liudui Region (六堆地區四縣腔客語聲調比較研究, which I will definitely try reading to see if I can answer some of those half-a-billion sandhi and tone mismatch questions by myself. By the way, I just found out Hakka had Sandhi, after being convinced of the opposite for ages :).

Update 3

Concerning 6, I just found 恁呢 meaning this in the dialectal synonyms of 这样 on the Wiktionary. The pages for the characters give án and nè/nó/nê. nè matches the video almost perfectly, given it's ne11. án would be an31, not the an53 heard in the video, but the tone 53 isn't distinct AFAIK, and given we're between a 55 and a 11, a Sandhi explanation would be very convincing. Still have to find this at that dictionary Justin suggested, and to see if I can decode the lu... Here is án-nè/án-nê... I also found 都 read lû in Southern Sixian on Wiktionary, but that is lu24, here I have lu55... I found 乜 as a dialectal synonym of 都 but no such meaning is given in the entry and no Hakka pronunciation is found for the given meanings... 熟讀唐詩三百首，不會做詩也會吟。 is rendered here with 也 ya55 for Sixian, 乜 me11 for Hailu, and 嘛 ma11 for Shaoan. So I guess I'll use 嘛, though I have ma55 in the video... But wait, I have ma23, followed by 31, so Sandhi again! And 4 is linda answered. I mean, the word is found, but with a Sandhied tone and in the wrong subdialect.

About 8 and 5, a Hakka native I just met agreed with the translations, so I'll go with my guesses. Also, Sixian has 你的 as ng11 ke55, which isn't too difficult to mistake for nge55, considering Southern Sixian has e55 instead of ke55. Actually, I'll go with ng11 e55 for that, writing 你个.

So I have answers for 1 and 2, none for 3, a hint of correctness for my guess at 4, a native agreeing with my translation at 5, the an-ne part of 6 with a partial explanation (requiring a rather creative Sandhi) for the lu part (admitting it always means 都 and the captions got the 就s wrong), nothing for 7, a new idea for 8, nothing for 9 and 10, and the answer for 11.

Final (maybe?) UPDATE

With the notes at the end of the question, all questions above and in the body are either superseded by the other pseudo-answer, or summed up in the following list, where BQ is a question from the body, and PAQ is a question from this pseudoanswer.

[Leftover of BQ1] Why is 该 | ke pronounced te? Idiolectal thing?
[BQ2] Why is 㧡 | khâi pronounced khân? Idiolectal thing?
Why is 故 | ku pronounced ko in the word 故所以? Idiolectal thing?
[BQ6 and PAQ3] Why is it tiam-li when 典礼 starts with tien in basically every Chinese dialect I can find documented, Middle Chinese included? Idiolectal thing? Oh well, Wiktionary says 典禮 is tián-lî but 典 alone is tién, WTH?
[BQ11] If it's supposed to be Tsung-yun, why does she say Tsung-mun? Idiolectal thing?
[BQ3 and PAQ9] Why is 瓜仔 | kwa-é kwap-e or kwab-e or actually more like kwa-be? Is it a glottal stop thing? Could it be that it's actually supposed to be kwa-ve, another word, maybe spelt 瓜薈 or 瓜棫?
[PAQ10] Why does 办法 | phan-fap sound more like phan-fa or phan-fat, or more accurately why is it definitely not phan-fap given her lips don't close for the -p?

Now I hope to add here a revised character+PFS transcription, but the character limit may get in the way… The superscripts indicate what question in pseudoanswer 2 contains the problem with the tone of the marked syllable, and stars mean I followed my intuitions from the pseudoanswer in terms of tone names (and therefore diacritics) though the dictionaries say otherwise. Since I'm gonna read that out loud when I make the video of this song (I have a Sicilian and an English version, and part of a Min version), I'll indicate pauses (whether actual breaths or just slowdowns in the speech) with ^.

口白：
在旧年个三月份 | Thî khiu-ngiàn¹ e ^ sâm ngiȧt-fun
涯又梦着涯个阿妈 | Ngâi*^2A yu mung tô*^2C ^ ngâi^2A e â-ma*^2B
该下仔系涯盖细盖细个时节 | Te-ha-é ^ he ngâi*^2A koy se ^ koy sei e sìi-tsiet
每日天无光 | Mî ngit^4A ^ thiên mò^4B kông
就跈等阿妈共下出门去卖菜 | Lu thèn-nún^5A ^ â-ma*^2B ^ khiung-ha ^ tshut^5B-mùn ^ hi mai tshoy
阿妈用担竿㧡等番薯番薯叶 | Â-ma*¹ yung tam-kôn ^ khan^6A-nún^5A ^ fân-su fân-sù yȧp^6B
㧡等瓮菜 | Khan^6A-nún^5A ^ fung-tshoy
就恁呢从弥浓庄 | Lu án-nê ^ tshùng ^ Mî-nûng*^{end of PA} tsông
一路行行到中坛 | Yit^9A-lu-hàng ^ hàng to Tsûng-mùn
在路项阿妈佢会佮涯讲 | Thî lu hong ^ â-ma*^2B ^ î^10A woi ka*^10C âi*^2A kóng
姿君你个眼珠看毋着吔 | Tsîi-kyûn ^ n̂g*¹¹ e ngián-tsû ^ khon m̀ tô*^2C ê
故所以涯希望你可以考上大学 | Ko-só-yî ^ ngâi*^2A hî-mong^12A ^ n̂g*^2C khô*^12B-yì² kháu^12C-sông thai-hȯk
吾正放得落心 | Ngâ nang ^ piong et⁽¹³⁾ lȯk⁽¹³⁾ sîm
因为阿妈哩句话 | Yîn-vi ^ â-ma*^2B ^ ya*¹⁴ ki fa
涯盖搭營 | Ngâi*^2A ^ keoi tap-yàng^15B
来考着中国文化大学 | Leòi kháu tô*^2C ^ Tsûng-kyeot ^ Vùn-fa Thai-hȯk
啊毋过到涯毕业典礼该晡日 | A-m̀^17A-ko ^ to ngâi*^2A ^ pit-ngiȧp tián-lî ^ te pîn-ngit^4A
吾都寻无着阿妈佢个匍背 | Ngâ lu*¹⁸ ^ tshìm mò^4B tô*^2C ^ â-ma*^2B ^ î*^10A e phù-poi
翕相个该下仔 | Hip-siong e te ha-é
吾嘛无一种欢喜个感受 | Ngâ mâ*^20A mò^4B yit^9A tsùng*^20B ^ fôn-hí e kám-su
因为对吾来讲 | Yîn-vi ^ tui ngâ lòi kóng²⁴
最重要最重要个人 | Tsui tshung-yau ^ tsui ^ tshung-yau e ngìn
涯再无辦法看着 | Ngâi•^2A ^ tsai mò^4B phan-fat ^ khon tô*^2C
佢个匍背 | Î*^10A e ^ phù-poi

有一个暗晡头 | Yû yit^9A ke am-pû-thèu
涯发梦梦着涯个阿妈 | Ngâi*^2A pot-mung mung tô*^2C ngâi*^2A e â-ma*^2B
该系盖久盖久吔以前 | Te he koi kíu koi kíu ê yî-tshièn
涯还盖细盖细个时节 | Ngâi*^2A hàn koi se koi sei e sìi-tsiet
涯还盖细盖细个时节 | Ngâi*^2A hàn koi se koi sei e sìi-tsiet

阿妈每日用担竿 | Â-ma*^2B mî ngit^4A yung tam-kôn
㧡着番薯叶㧡瓮菜 | Khan^6A tô*^2C fân-sù yȧp^6B khan^6A fung-tshoi
带涯共下去卖菜 | Tai ngài khiung-ha hi mai tshoy
从弥浓行着中坛 | Tshùng Mî-nùng hàng tô*^2C Tsûng-mún
有么人爱买无 | Yû mak-ngìn oi mâi mò^4B?
有么人爱买无 | Yû mak-ngìn oi mâi mò^4B?

阿妈希望涯读大学 | Â-ma^2B hî-mong^12A ngâi*^2A thu̇k thai-hȯk
毋使像佢爱去卖菜 | M̀-síi tshiong î*^10A oi hi mai tshoi
每日天无光就爱行 | Mî ngit^4A thiên mò^4B kông lu oi hòng
㧡菜㧡啊变匍背 | Khan*^6A tshoi khan*^6A ạ pien phù-poi
㧡菜㧡啊变匍背 | Khan*^6A tshoi khan*^6A ạ pien phù-poi

有一个暗晡头 | Yû yit^9A ke am-pû-thèu
涯发梦梦着涯个阿妈 | Ngâi*^2A pot-mung mung tô*^2C ngâi*^2A e â-ma*^2B
该系盖久盖久吔以前 | Te he koi kíu koi kíu ê yî-tshièn
涯还盖细盖细个时节 | Ngâi*^2A hàn koi se koi sei e sìi-tsiet
涯还盖细盖细个时节 | Ngâi*^2A hàn koi se koi sei e sìi-tsiet

阿妈每日用担竿 | Â-ma*^2B mî ngit^4A yung tam-kôn
㧡着瓜仔㧡吊菜 | Khan^6A tô*^2C kwâ[p]-é khan^6A tiau-tshoi
带涯共下去卖菜 | Tai ngài khiung-ha hi mai tshoy
从弥浓行着中坛 | Tshùng Mî-nùng hàng tô*^2C Tsûng-mún
有么人爱买无 | Yû mak-ngìn oi mâi mò^4B?
有么人爱买无 | Yû mak-ngìn oi mâi mò^4B?

一年一年过一年 | Yit^20B ngiàn¹ yit^20B ngiàn¹ ko yit^20B ngiàn¹
涯从大学爱出社会 | Ngâi*^2A tshùng thai-hȯk oi tshut^5B sa-fi
毕业典礼该晡日 | Pit-ngiȧp tián-lî te pîn-ngit^4A
寻无阿妈个扶背 | Tshìm mò^4B â-ma*^2B e phù-poi
看无阿妈个扶背 | Khon mò^4B â-ma*^2B e phù-poi

Doing this, I realized several leftover questions are not in the above list. Let me put them here:

瓮菜 is supposed to be vung-tshoi, why does she say fung-tshoi? Idiolectal thing?
从 is supposed to be tshiùng, why does she say tshùng? Idiolectal thing? Influence from other dialects?
君 is supposed to be /kjun/, why does she say /kyn/? Idiolectal thing? Influence from e.g. Cantonese?
Why does she front so many /oj/s to /øj/s?
Why does she read /kjøt/ instead of /kwet/ (kwet) for 国?
Why does she read every v- as /v/ except the one in 会 | voi which she read woi?
重要 is supposed to be tshung-yeu (modulo tones) in Southern Sixian, why does she say tshung-yau (modulo tones)? Idiolectal thing? Influence of e.g. Mandarin?
How come 暗晡头 is am-PU-theu while 晡日 is PIN-ngit? Why not am-PIN-theu or PU-ngit?
Why do I hear tiam-di instead of tian-li?

So we have a grand total of:

7+9=15 pronunciation questions;
24 tone questions in the other pseudoanswer;
I guess no more doubts on character spelling? Apart from the unfindable words "ka" and "ma", of course, which I wrote with the character for the "kak" I know from Hailu Hakka with that meaning, and with the character of the Min mā = also, which this could be a loan of…

For question 5, given the Kana name "Tsun'un", I think the name used to be Tsung-wun, and maybe it stayed like that in Meinong while evolving to Tsung-yun in Zhongtan. From Tsung-wun to Tsung-mun the road is small: the rounding of lips of the u starts early, on the end of the -ng, et voilà an m-. From Tsung-yun to Tsung-mun, it seems harder to go.

UPDATE once more :)

Outsourced all these questions (except for 4 6 8 in list 2) to Quora, with list 1 here and the outsourced part of list 2 here.

23.674 characters

Update again

The closest I can get to that ka/kak is 风合水 | fung kak sui, but that's a long way from sourcing the usage of kak that seems to be found here.

Yet another update

I'm led to believe she is using 擐 | khan instead of the khâi: both mean "carry" (though khâi is with shoulder-poles and khan is hanging from the hands), and the way she says it matches khan perfectly and only matches khâi inasmuch as it sounds similar to khan. Yes, 擐's root pronunciation is khuan, but hakkadict reports alternate khan. Could she actually have been carrying two shoulder-poles with her two hands, instead of hoisting the pole on her shoulders? Because that would perfectly fit the use of 擐 instead of khâi.

Update 8

The question about the ma23 has been answered by @justinrleung. There is a dictionary which reports 嗎 with the desired meaning. The tone doesn't fit -- it's reported as ma (Qu tone) for Northern Sixian and mà Yang Ping) for Southern -- but that may be an influence of the synonym yâ | 也. It is indeed a loan from Min 嘛.

Update 9

From here:

Vung-tshoi should be spelled 蕹菜, and there is a dictionary reporting fung-tshoi, besides Wiktionary;
從 as tshung (whatever the tone) is probably a Mandarin influence, says @justinrleung, as I had suspected;
The final of 君 should be -yun, but it varies between -yun -ün -yün, with -ün being found here for Sixian and hakkadict giving (can't verify) kyun for Northern and kün for Southern;
国 has varying pronunciations in Meinong, the most common being ket; kyöt may be idiolect or even a one-off;
要 is yau in Meixian, so it may be that Meinong speakers retained that, though hakkadict gives yeu for Sixian, both Northern and Southern; she says something like /jǝw/.

I'm awaiting a reply on what "different pronunciations" Meinong speakers "seem to have".

And I sure hope I didn't break the char limit here :).

Update 10

Pronunciations for 国: kwet, ket, kyet, kyat.

For that weird ngit51, listen to fat-ngà: same thing. Northern Sixian, but whatever. Would like to find that huge drop in isolation… Hm, getting close, again Northern…

woi could be due to a w allophone of v, but patterns are elusive and a recording at the Hakka cloud has v. And here comes labiodental approximant, as if we didn't have enough sounds already :). I mean, fro that to v the path is short (some devoicing and we're there), but why would it velarize to get to /w/?

And with this we find kün for Southern and kyun for Northern, so we can conclude this is a variation internal to the dialects. Perhaps even single speakers oscillate, I mean, we have only one instance of this syllable.

Concerning 3, this gives coˊbinˊzeuˊxinˇ as a possible reading of 昨晡朝晨, and a bin55 with creative Sandhi could easily be taken for a pi(t)23 as I had. So that might be it. — MickG, Dec 1, 2018 at 23:42
Just hear the tsho55-pin55-ngit1 recording here and how it's like completely reversed to tsho11-pin14-ngit5. Also, I'm thoroughly confused about those contours: I thought the acute marked "tone 1" aka 55, but it's actually 24 and sandhies to 11 so the only strange tone is the ngit5 :). — MickG, Dec 1, 2018 at 23:50

MickG · Accepted Answer · 2019-01-12 16:29:00Z

This is an attempt to figure out the tones using the description in this study pointed to by @justinrleung. I should add it to the question, but of course it doesn't fit, so here I am. I don't know how much this will solve, but let's see.

First of all, the description. From the introduction, which I translated entirely, I get the following contour table, where YP=Yin Ping, Y'P=Yang Ping, Q=Qu, S=Shang, YR=Yin Ru, and Y'R=Yang Ru:

Tone Contour YP 322 Y'P 31 S 42 Q 55 YR 43 Y'R 55

Now, the contours found on both hakkadict and the Wiktionary are somewhat different. There is a 55 which develops from historical Qu tones at least in one case, so that will be the Qu tone. The study mentions a mis-analysis of YP as 324, which matches the contour 24, so that will be our YP on the dictionaries. There is a 31 which develops from Shang tones in at least one cases, so that will be S. A checked tone 5 exists, and that will be our Yang Ru. The other checked tone, 2, will have to be the Yin Ru, and the remaining contour 11 will have to be Y'P. So I will assume the following matches, the diacritics coming from matching Wiktionary contours to Wiktionary diacritics on PFS:

Tone Contour Dict contour Diacritic Hakka Romanization System Diacritic YP 322 24 circumflex acute Y'P 31 11 grave circumflex S 42 31 acute grave Q 55 55 none none YR 43 2 none grave Y'R 55 5 vertical line none

The study also mentions the following Sandhi rules:

YP+Y'P=324+Y'P;
YP+{YP/Q/Y'R}: YP=322;
YP+{S/YR}: YP=322;
X+YP=X+322;
Y'P+YP=YP+YP;
Y'P+{S/YR}: Y'P=32;
Y'P+{Q/Y'R}: Y'P=3->1.5;
The paragraph below

The above all come from the intro, except for 8, which refers to the following two:

（3）由上聲和陰入組合的二字詞，前字起始音會高於後字
諸如「上聲+上聲」（廠長），「上聲+陰入」（粉筆），「陰入+陰入」（竹節），「陰入+上聲」（屋頂）四種組合時，會産生前字起始音高於後字，後字音程變小，音量也變輕的現象，調型則維持不變。

（4）陰平，陽平分別後接上聲或陰人時，上聲或陰入調值會產生差異
差異的原因乃是因爲陰平，陽平與上聲或陰入分別有著同樣音程的相對語感;陰，陽平結束音高112（調值）的差異即成爲後接上聲或陰入音高（调值）差異了，例：「豬骨」與「牛骨」，因前字r豬」，r牛」結束音高相差半度左右，在「豬骨」與「牛骨」有著相同的相對語感之情況下，兩詞的後字雖同爲「骨」但前者音者會高於後者半度左右。

From what I can understand of this, it seems both (3) and (4) imply one tone slightly raises and the other one slightly compresses, so fuzzy matches should see through these rules. Therefore, I will ignore them for the sake of this attempt.

With this in mind, I proceed to infer tone names from contours found on hakkadict (see other answer).

Not that I incorporated some of the new info I got since the other pseudo-answer into the above, so the match isn't perfect. Now I give you the famous precise transcription with both the contours resulting from the video and those resulting from the above tone names plus the study's contours and sandhi. I give it line by line, commenting on the problems of each line separately.

Thi33|322' khiu55|55-ngian51|3:1.5'' e334|55'* sam11|322' ngiat2|55''-fun33|55'': The last two are probably just a problem of defining the 5 levels, maybe if I relisten I can justify reanalyzing as 44 and 55, giving fuzzy match and match; the only big problem is ngian51 in the video vs. ngian3:1.5 or ngian31 in the study's terms, but maybe it's just starting from 55 and jumping to 3 is discontinuous? UPDATE: I relistened, and can reanalyze as ngiat33-fun22, still both mismatches, but if I decide the 5 has been lowered to the former 3 then it becomes ngiat55-fun44, fuzzy matches, so I'll go with that, leaving only the wannabe sandhi rule that Q+S makes the S a 51 to avoid the jump from the Q's 5 to the S's 3; UPDATE 2: If ngai is Y'P, the proposed rule is negated by the ngai in l. 3, but the suspicion I have is that it's actually YP, which would leave the rule standing; but of course l. 3 has e|Q sii|Y'P-tsiet|?R whose sii21 has no such rule applied, so the rule wouldn't apply there;
Ngai24|3:1.5'' yu55|55 mung55|55 to331|42' ngai11|3:1.5'' e33|55'' a11|322'-ma33|322': I guess the e33 vs. e55 is just a problem of tone level definition; I'm highly suspicious of ngai being Y'P, considering making it YP would solve one of the problems, but still, the first ngai24 is not justified; UPDATE: Maybe when ngai meets a y- the boundary between the two gets confused and the ngai can seem to rise up if the y- is in e.g. Qu tone like here? That would produce a justification for ngai24 if ngai were YP; a-ma seems fine here, but the change in pitch suggests the tones should be different…
Te55|55-ha55|55-e11|42'' he33|55' ngai21|3:1.5' koy33|55'' se334|55'' koy33|55'' sei33|55'' e34|55'' sii21|32'-tsiet2|43'': a shifting reference frame for the levels accounts for all the 33|55, the e11 is guessed and could in fact be e51 (I had hay51 originally, so either ha51-e11 or ha55-e51 or ha53-e31), which makes it a sandhi fuzzy match, and that leaves us with the 2|43 of the Yin Ru; UPDATE: Note that -e is Shang tone, so 42, meaning we have 42|51 or maybe 52, decidedly a fuzzy match; the final word is more like sii21-tsiet21, which if we assume the reference for the tone levels has lowered could be sii32-tsiet32, giving us a fuzzy match; so the line would be OK, even with a Y'P on ngai;
Mi22|322' ngit51|43'' thien24|324' mo53|322'' kong21|322': Pretty sure the ngit51 is BS, such a fall on a Ru? Maybe 53 or 54, fuzzy match; as for mo53|322, let me relisten… Nope, I agree with myself on both counts, the fall is huge in the ngit51 and the mo starts higher than the thien ends, so we have two problems;
Lu55|55 then11|32''-nun11|42'' a11|322''-ma34|322'' khiung44|55'-ha44|55' tshut4|43'-mun21|31' hi33|55'' mai33|55'' tshoi334|55'': Let me change the tone levels at some points: Lu55|55 then22|32'-nun22|42'' a22|322'-ma34|322'' khiung55|55-ha55|55 tshut4|43'-mun21|31' hi55|55 mai55|55 tshoi556|55: that's much better; the elephants in the room are the nun22|42 and the ma34|322; the latter suggests maybe YP+Q makes a 324, but that would screw up the very beginning, so I guess not; as for the nun, it looks like the tone should actually be a YP, so we'd get a fuzzy match; Wiktionary does indeed report a ten24 for Meinong, which would be Yin Ping, so maybe that is the solution? UPDATE: I think a-ma is a|?P-ma|Q, and that would produce matches or fuzzy matches to all the a-ma's down to this line; the tshut of tshut-mun, however, has no level difference from the ha of the khiung-ha before it, which is a 55, and this would make it a Y'R, but it's supposed to be a YR… it's too short to hear a fall in pitch, so I can't do much;
A11|322''-ma55|322'' yung55|55 tam35|55''-kon31|322' khan44|322''-nun32|31' fan11|324''-su4|31'' fan24|324'-su31|31-yap34|55'': Every time we see a-ma, the ma has the wrong tone: maybe it should be a Qu tone? Why is tam a 35 when it's supposed to be a Qu aka 55? The first fan-su she gets the tones wrong and then corrects herself, and those match alright; yap sounds like it should be a Yin Ru, but maybe it's just a shifting reference for the tone levels? UPDATE: tam-kon is more like tam45-kon21 or tam45-kon32, so the kon is a fuzzy YP, and the tam could be a Q which fluctuates a little to put some motion into a sequence of 55s; I see now that khan (or rather khai) is given as /khai⁴⁴/ for Meixian on Wiktionary, which matches the video; but Meixian is in Mainland China while Meinong is in Taiwan… could Meixian have influenced Meinong or just the singer?
Khan44|322''-nun32|42' fung33|55''-tshoi334|55'': our friend khan is back with the same problem as before; nun surprisingly fuzzy matches, but really it fuzzy matched in the previous line too, so maybe l. 5 where the mismatch was had a problem of tone levels and a small pitch fall I didn't hear? Fung-tshoi mismatches, but again, a shifting reference for the tone levels does away with that. It's anyway higher than the nun, so it couldn't be a level-1, and the only level tone (in fact, the only one altogether, not just the only level one higher than 1) is 55;
Lu55|55-an53|42'-ne22|31'' tshung21|31' Mi11|31''-nung11|322'-tsong11|322': we have a few rogue Yang Pings who decided to sandhi despite being in fromt of Yang Pings and not Yin Pings, apparently; maybe a rule should be added that Y'P+Y'P=322+Y'P? For the moment, nothing contradicts it;
Yit5|43''-lu45|55' hang21|31' hang21|3:1.5' to34|55'' Tsung214|324'-mun51|31'': the mun starts lower than the to ends, so at most a 3, so that's great; but the Tsung ends lower than the mun starts, making it 212… ah no, maybe 21:2.5 or even 213, so 324 is still a fuzzy match with a reference level problem; if we assume the to is another rogue Yang Ping, we can write it as 著/着 and solve all problems here (except that yit…), because the description of hakkadict, «嚎啕大哭[]附在動詞後面，表示動作已達結果的語氣，今國語習慣用「到」或「著」字。例：寒著、冷著。», sounds like it could fit here; the yit has no pitch fall, but then again it's very short, so I can't really distinguish; maybe I can assume it is 54~43?
Thi11|322' lu34|55'' hong34|55'' a11|322'-ma24|324' i11|31'' woi24|55'' ka31|?? ai11|31'' kong31|42': Lift the reference for the first few tones, and they become fuzzy matches; curiously, a-ma is fuzzy on both tones this time; but wait: there is a break there, so no sandhi can apply, so it's actually ma24|322'', and the mismatch is still there; gotta say, reading that as Qu tone requires a rather big upward drift for it, but the jump from a to ma is there and prevents taking it as the same tone; honestly, there is a small pitch change in the i, so it could indeed be a i31, though i32 seems like a better match; Yang Ping? The woi is more of a one-degree change, maybe woi23, or woi45, so Qu seems like a good guess; ka and ai are almost fused, like kaai31, so it's hard to tell where to split them; if we think of woi as woi34, then it's kaai42, maybe a Qu plus Yin Ping, maybe kaai52 = ka54 + ai42, or in fact ka43 + ai32, and then the kong sounds like a 41 ~ 42;
Tsii34|322''-kyun11|322' nge44~ng11|3:1.5''e55|55 ngian41|42'-tsu112|322' khon42|55'' m22|32' to22|42'' e11|322': Tsii is probably just a fluctuation, I mean, YP is basically level, so from 3 to 4 instead of to 2 can happen, and maybe it was actually a 2:2.5~22; the ng bothers me, it sounds like it should be a YP, not a Y'P; khon and m were originally not distinguished, and the m does seem to have a pitch drop, while khon seems not to, so it's more like khon44 m42, a fuzzy match; we have already discussed the to above, and the Yang Ping would sandhi to 322, thus giving a fuzzy match; except the alternate spelling gives me a Shang, not a Yang Ping, and a Shang is definitely out of the question;
Ko55|55-so54|42''-yi41|324'' ngai12|322'' hi21|322'-mong35|55'' ng22|32' kho22|42''-yi21|322' khau33|42'' song31|322' thai33|55''-hok3|55'': so-yi is more of so53-yi31, fuzzy match + we undo the sandhi to get another fuzzy match; why sandhi would not apply there without an evident pause is not evident; or else we posit that this is in fact a Yang Ping; I can't say the syllable was short and chopped the raise off; if yi is yi322, then ngai is ngai23, which modulo a little fluctuation would match a Yin Ping; and if ngai is indeed a Yin Ping, we have one further reason for the yi322, given that the sandhi problem is removed from the root; the mong raises a bit too much to blame it on fluctuations, so I cannot explain it; kho really seems to be a Yin Ping but it should be a Shang, Idk; khau song is really more like khau33 song11, maybe khau55 song322, but khau should be 42 as a Shang, which is another mismatch; if song is song22, thai-hok is thai44-hok4, which fuzzy matches the thai55-hok55 we'd have from the study;
Nga22|31'' nam45|55' piong5|55 et4|43' lok3|55'' sim11|322': Again, ngai seems to be a Yin Ping ngai322; the (t)et is higher than the lok, yet the reverse should happen; positing a reference frame shift seems weird, but given that (t)et has a pitch fall and lok doesn't, it seems like that is the case; how confusing…
Yin22|322'-vi55|55 a11|322'-ma33|322' ya22|42'' ki22|55'' fa223|55'': despite the fuzzy matches, the difference in pitch between a and ma in a-ma makes the tone difference undeniable, and suggests, once again, a|YP-ma|Q; in that case, it would be a11-ma55 ya44 ki44 fa445, which makes three mismatches into a fuzzy match: a very compelling argument for that Qu;
Ngai31|3:1.5' keoi34|55'' tap3|43'-yang34|31'': this is the first time ngai behaves decidedly like a Yang Ping, given the IMO too big pitch drop for a 322; keoi may in fact be 45, a fuzzy match; the tap seems to raise a little, in opposition to the Yin Ru contour, but it is lower than keoi, which supports a low checked tone, aka Yin Ru, so a little fluctuation may have played a role here; yang seems to almost dip, like 323 or something, maybe even 324, and the next syllable just so happens to be a Yang Ping, which suggests a remote sandhi to a Yin Ping, but this is supposed to be a Yang Ping…
Loi22|32' khau33|42'' to332|42' Tsung22|322'-kyeot3|43' vun21|3:1.5'-fa34|55' thai34|55''-hok34|55'': before I forget, kyeot actually sounds like it's falling slightly, which matches even better; khau is back just like before; definitely not khau55 this time, or else loi is 44 and a mismatch; to is a fuzzy match this time: surprise! 322 would still be better though; I'm not sure if thai starts at the 3 or the 4 of fa, so maybe it's thai45-hok45, which would make it a fuzzy match;
A55|324''-m533:1.5''-ko445|55' to44|55' ngai41|32'' pit3|43'-ngiap4|55' tiam43|42'-li11|322' te33|55'' pit32|324'' ngit3|43': to is more like 45, still fuzzy match, ngai is a definite mismatch whether YP or Y'P, tiam is more like a 21 or 32, still fuzzy match, though a Ping would match better, the end is solved by a shifting reference for the tone levels and noticing ngit has a very small pitch drop;
Nga33|322' lu55|322'' tshim31|31 mo22|31'' to221|42'' a22|322'-ma33|322' i11|3:1.5'' e33|55'' phu11|3:1.5''-poi334|55'': so nga is YP and ngai is Y'P? Maybe they're being tonally confused then! And the two lu's may also have merged, let me see before… yep, this is the only dōu and it sounds identical to the other two jiùs; the to starts a little highet than the mo ends, so maybe a 2.5:1, Y'P? again, a-ma has two different level tones, so â-ma or à-ma seem opportune; i and phu seem to have a slight drop in pitch, so maybe fuzzy matches, though the slight drop matches 322 better than 31; poi might be 445, fuzzy match, and e might be hovering around 4, again a fuzzy match; tshim ends as low as mo starts, and to starts a little highwer, so maybe tshim32|31' mo22|31'' to21|42'', to seems to be a Ping tone, and mo a YP, maybe Y'P+YP or YP+YP; wait though, to is actually 31, so let me raise the reference: tshim43 mo33 to42, that screws up tshim, and the to has always been a S disguised as a YP, so let's not raise the reference :);
Hip4|55'-siong55|55 e55|55 te44|55'-hai44|55'-e41|42': when will I have another problem-free line like this :)?
Nga11|322' ma23|? mo31|31 yit2|43'' tsung21|42'' fon21|322'-hi33|55'' e33|55'' kam21|42''-su23|55'': I'd guess ma is a YP, so it would have to be 324, fuzzy match; actually, it's perhaps a bigger rise in pitch than just 24, so yep, YP; mo is more of a mo41 or mo42 then, closer to a Shang… low checked tone can only be a YR, despite the pitch not dropping, though that can be explained away with the shortness of the syllable; tsung is really more of a Y'P, which would have no sandhi given the pause after it; unfortunately, documented options are Shang and Qu, both mismatches; for the end, let's raise the reference: fon32|322'-hi44|55' e44|55' kam32|42*-su34|55', where kam is a fuzzy match to both a Y'P and a Shang;
Yin22|322'-vi55|55 tui55|55 nga53|324'' loi31|31 kong21|42'': kong is more of a 221, even less matching for a Shang, sounds more like a Yin Ping;
Tsui53|55'' tshung22|55''-yau22|55'' tsui32|55'' tshung22|55''-yau22|55'' e22|55'' ngin21|31': more like tsui54 tshung32-yau22 tsui31 tshung22-yau22 e22 nyin20, still quite a disaster; tsui seems like it should be a Shang or Yang Ping, given the big drop in pitch the second time; ho-ho, hakkadict reports an alternate Nan Sixian tsui31, which would be a Shang! So that is basically solved; for the end, I'll assume the reference was dropped after the second tsui, to have tshung55-yau55 e55 nyin54, or maybe a little less, so I have nyin43|42'… wait, it has to be a Yang Ping… the other tshung-yau is again a lower reference: the tsui is emphasised, the tshung-yau can have a lower reference, then maybe ngin gets emphasised again?
Ngai31|3:1.5' tsai44|55' mo42|31'' phan34|55''-fa32|43' khon44|55' to31|42'': ngai is for once perf… wait, sandhi with Qu tone, almost perfect :); mo is problematic as always, matching Shang perfectly and Yang Ping only fuzzily; for the rest, we raise the reference, and the matches are all perfect except for phan45, fuzzy match;
I22|3:1.5'' e44|55' phu21|3:1.5'-poi33|55'': raise the reference for everything, and phu-poi has two fuzzy matches, and i33 is a mismatch, more of a Yin Ping.

Finally got through this. Time to isolate the problems and formulate questions. I'll literally copypaste everything and work my way through it, leaving a part of it untouched after the "I must get back to maths" warning.

The ngian51|3:1.5 suggests a sandhi rule that Q+S makes the S a 51 to avoid the jump from the Q's 5 to the S's 3; if ngai is Y'P, that rule is negated by the ngai in l. 3, but I suspect (see next item) it's a YP, which would leave the rule standing; but of course l. 3 has e|Q sii|Y'P-tsiet|?R whose sii21 has no such rule applied, so the rule wouldn't apply there; what is up with this rule then?
A: Ngai is almost always a problem, behaving like a Yin Ping instead of the Yang Ping it should be; in this line, ngai24 is a 322 which gets confused with the y- of the yu55 after it, and the other ngai fits a 322; in the next line it fuzzy matches both; in lines 10 and 13 it's a fine YP; in l. 12 it's a YP with some upward fluctuation; in l. 15 it's basically a Y'P, in l. 17 it's almost a Shang; in l. 23 it's basically a Y'P; could it be on its way to merging with nga, which is indeed a YP (except when it tries to play Shang in l. 21 :)?
B: a-ma seems fine here, but YP+YP should be 322+322, not 11+33, so raising a little we may think of this as a33-ma55, matching â-ma (YP+Q); and such is the situation in almost all other instances of the word in this video; the only instance where a YP would fit better is in l. 10, where it appears to sandhi with the Y'P of i, but that tone is also doubtful to me, and the same situation presents itself in l. 18, with no sandhi, so maybe l. 10 is an accidental upward drift and the tone is indeed a Qu?
C: I made a bit of confusion about the to; first of all, it can be 到 or 着, which are respectively to|Q and to|S; in the video, we find 331 in this line, to34 in l. 9, to22 in l. 11, to332 in l. 16, to44 in l. 17, to221 in l. 18, to31 in l. 23; apart from to44, which is probably 到, the rest seem to be near-level low tones, so YP, except the to31 which would be a perfect Yang Ping; so could this actually be a Yin Ping rather than a Shang?
L. 3 presents no further problems: yay!
A: ngit here has a very high pitch, definitely not a 4, and even in the entries here it seems to oscillate between high and low tone, so what's going on here?
B: mo, looking here and on other entries, seems to have a sandhi with a preceding raising tone, as if YP+Y'P in at least the case of mo as the Y'P gives 24 53, raising the Y'P and making the YP raise; we find Y'P+mo22+S in l. 18, ?+mo31+YR in l. 20, and Q+mo42+Q in l. 23; the ? seems to be a sandhied YP, being ma23; so it would seem this word's tone just adapts to its surroundings; is there a better explanation (like a sandhi rule I'm missing) or is that the best bet?
A: nun looks like the tone should actually be a YP, so we'd get a fuzzy match; Wiktionary does indeed report a ten24 for Meinong, which would be Yin Ping, so maybe that is the solution? In the following two lines we find nun32, which would fuzzy match both; so either we posit a different tone than reported, or we assume I didn't hear a pitch drop in this one nun but it was there, or maybe there is another explanation?
B: The tshut of tshut-mun has no level difference from the ha of the khiung-ha before it, which is a 55, and this would make it a Y'R, but it's supposed to be a YR… it's too short to hear a fall in pitch, so I can't do much; how can I explain this?
A: khan is consistenly a fuzzy Qu when it should be a YP; I see now that khan (or rather khai) is given as /khai⁴⁴/ for Meixian on Wiktionary, which matches the video; but Meixian is in Mainland China while Meinong is in Taiwan… could Meixian have influenced Meinong or just the singer?
B: yap sounds like it should be a Yin Ru, but maybe it's just a shifting reference for the tone levels?
No further problems here, yay!
an53|42'-ne22|31'' and Mi11|31''-nung322'-tsong11|322': we have a few rogue Yang Pings who decided to sandhi despite being in fromt of Yang Pings and not Yin Pings, apparently; maybe a rule should be added that Y'P+Y'P=322+Y'P? For the moment, nothing contradicts it; it is true that an-ne is reported here as either án-nè or án-nê, so that wouldn't be a problem; and the Wiktionary solves the problem by reporting Mî-nùng as well as Mì-nùng, so yay!
A: Yit seems to have an oscillating tone, since this line features it as yit5, and l. 20 features it as yit2; here it is as high as the following Qu-tone lu, but maybe a reference shift can place it at 4, and we can assume a chopped-off 43 is what we have before our eyes, but the yit2 is in a stranger position where raising the reference is problematic, so what is going on here?
B: Tsung21:2.5-mun31 is finally tonally OK, though the mun is still funky business since it's supposed to be a yun… anyways, no further problems here as far as this post is concerned, the sounds are for another place;
A: i should be a Yang Ping, but it's i11, more like the Yin Ping i22; a Qu is supposed to follow, so no sandhi would happen, though woi is also weird tonally; elsewhere, it is consistently level, whether i11 as in "a-ma i e phu-poi", or i22 as in the last line; could it be a Yin Ping? I mean, here the example sounds like i22 mo21, so a level tone and a falling tone, and if mo is Yang Ping (and thus reported here), then how could i22 be Y'P if mo21 is? Though I wrote it could be i32… yeah well, i322 fuzzy matches that too;
B: From what I wrote, it sounds like the woi2 could be explained away as a drifting of a Qu tone;
C: ka31: ka and ai are almost fused, like kaai31, so it's hard to tell where to split them; if we think of woi as woi34, then it's kaai42, maybe a Qu plus Yin Ping, maybe kaai52 = ka54 + ai42, or in fact ka43 + ai32, and then the kong sounds like a 41 ~ 42; I outsourced the matter of this ka to Quora anyway;
ng11: supposed to be Yang, level like a Yin, and the ng22 of next line is an even better match for a Yin Ping, so is this actually a Yin Ping?
A: hi21|322'-mong35|55'': the mong raises a bit too much to blame it on fluctuations, so what is up?
B: kho22|42''-yi21|322': kho really seems to be a Yin Ping but it should be a Shang, what gives?
C: khau44|42'' song322|322': khau should be Shang, but seems to be Qu, what's up? And three lines below, khau33, maybe a fuzzy Qu or a fuzzy Yin Ping… actually definitely a Yin Ping or we screw up another tone which we cannot justify being raised above normal levels;
piong5|55 et4|43' lok3|55'': the (t)et is higher than the lok, yet the reverse should happen; positing a reference frame shift seems weird, but given that (t)et has a pitch fall and lok doesn't, it seems like that is the case; how confusing… but I guess we consider this solved, so no further problems here, yay!
ya22|42'' ki22|55'' fa223|55'': the mismatches are solved by assuming a-ma ends in a Qu with a reference shift, since ma55 there makes these 44 44 445, fuzzy matches; so if â-ma, then no problems, otherwise why are these so low in tone?
A: keoi34|55'' is probably a reference shift;
B: tap3|43'-yang34|31'': the tap seems to raise a little, in opposition to the Yin Ru contour, but it is lower than keoi, which supports a low checked tone, aka Yin Ru, so a little fluctuation may have played a role here; yang seems to almost dip, like 323 or something, maybe even 324, and the next syllable just so happens to be a Yang Ping, which suggests a remote sandhi to a Yin Ping, but this is supposed to be a Yang Ping… what is going on?
No further problems here, yay!
A: A55|324''-m533:1.5''-ko445|55': the tone of A is due to a Min loan, so it matches that one, and we will make it a Qu tone (because of contours) and spell it 啊, as opposed to 阿 which gave the mismatching YP; as for m, maybe the Yang Ping has a raised brother when after Qus?
B: te33|55'' pit32|324'' ngit3|43': the end is solved by a shifting reference for the tone levels and noticing ngit has a very small pitch drop; so no more problems, yay!
lu55|322'': the two lu's may have merged, given the tones are so similar; or maybe it's just emphasis?
No more problems here, yay!
A: ma23|?: I'd guess ma is a YP, so it would have to be 324, fuzzy match; actually, it's perhaps a bigger rise in pitch than just 24, so yep, YP; this is anyway outsourced to Quora;
B: tsung21|42'': such a low tone seems to be a Yang Ping, how is that a Shang or a Qu?
kong21|42'': kong is more of a 221, even less matching for a Shang, sounds more like a Yin Ping; in l. 10, however, a reference raised, encouraged by a 11 posited Yin Ping, brought it to a Shang; what is up with this word?
Tsui54 tshung32-yau22 tsui31 tshung22-yau22 e22 nyin20: tsui is Shang with much emphasis; with some funky reference shiftings, everything is fine, except that nyin, supposed to have a Yang Ping, which seems to have a Shang… would it make sense that the tsui's AND the nyin are emphasised, but NOT the tsung-yau?
No further problems here, yay!
I33|3:1.5'' e55|55 phu31|3:1.5'-poi44|55', and i is already tackled above, so no further problems.

Who would have thought I would have made it to the end? Welp, now we have… let me count… 1 2A 2B 2C 4A 4B 5A 5B 6 9A 10A 10C 11 12A 12B 12C 14 15B 17A 18 20A 20B 21 22… 24 precise tone questions for Justin :) or whomever else decides to answer any of these. That is a lot of questions! And then there are some weird pronunciations still left to clear up, and possibly some spelling and meaning problems… this video is a never-ending question mine :)!

29.249 characters

Positing a mó would explain 3/4 lines where mo is found.

Mî-nùng doesn't really work because it would give a 324, which isn't there. But Mì-nùng tsòng doesn't work either, because they'd be all 31s and we have 11s. So that is still a problem: it looks like these should really be Mî-nûng tsông. OK tsông is actually that way, but Mi-nung should be Mî-nûng, which contradicts etymology and dictionary info.

MickG · Accepted Answer · 2019-01-30 22:29:44Z

As the saying goes, non c'è 2 senza 3 (there's no 2 without a 3), so here is pseudo-answer 3 :). Here is the content of the spoiler that will be in the blog post about this song.

First of all, this song is probably in Meinong Hakka. That it is Hakka is clear. That it is Meinong I infer from when she says she and granny set off from Meinong to sell groceries. After lots and lots and lots of work at SE, and 1 2 3 4 5 6 7 8 9 10 11 12 13 Quora questions (wow!), here is the current state of the art. First of all, the transliteration obeys the conventions of this post - except I found out Minhakka is a bastard who upturned the tone diacritics, and I've always used Minhakka's wrong version of the marks, whereas here I use the correct one. To discuss this, let me first mention this study, which is where I get the tone contours. Here is a table of what I got:

Before we tackle sandhi rule and all the tone problems I have with this video, there are a few quirks in the singer's pronunciation, and a few new words in her dialect, that need to be addressed:

在 is not the usual tshoi (or rather tshôi), it is pronounced thî; this is documented on the Wiktionary for Southern Sixian Hakka, so no problem there;

个 is not kài or kè (or rather kai or ke), but e; this is documented in the Wiktionary, and possibly as an alternate pronunciation in hakkadict, which won't load now; this is not the only example of common words being eroded in her dialect:

就 and 都 are both pronounced lu; the latter is documented at the Hakka cloud in the example, and the former I can't find documented anymore; that Hakka cloud is highly search unfriendly;

佢 is pronounced ì (well, the tone is to be treated carefully, see below) instead of kì;

正 is pronounced nang instead of cang; this is documented at hakkadict as an alternate Southern Sixian pronunciation;

得 is pronounced et instead of tet; this is undocumented, but with all this erosion business it's easy to accept;

会 is pronounced woi, whereas all other v- initials are pronounced as vs, except the vun in vun-fa which sounds like /vwun/; my guess here is that, before rounded vowels, the v labialized to vw (hence vwun and *vwoi), and then erosion set in and simplified *vwoi>woi;

该 | ke is consistently pronounced te; at the moment, it is unsure whether this is idiolectal or more widespread;

㧡 | khâi seems to be consistently pronounced khan, as though it were 擐; I don't know if the two words are merging, if she's misusing the second one, or if granny actually lifted the poles with her hands instead of carrying them on her shoulders (or any other situation took place that would warrant the use of khan instead of khâi);

The particle 等 (-ing in English) is nún (tone problem!) instead of the tén I'm used to; this is documented in the Wiktionary;

She says fan-su twice because she gets the tones wrong on the first one; the tones on the first instance are adapted from the video according to the study's tonal description;

Instead of the familiar 恁样 | an-nyong (don't remember the tones right now), we find 恁呢 | án-nê, documented as such on I think the Wiktionary;

The place name 中坛, which should be Tsung-tham or Tsung-iun according to this, is heard as Tsung-mun; given that source tells us Kana spellings of the old name were Tsun'en and Tsun'un, which seem to match Tsung-ien and Tsung-wun, I guess the name could be Tsung-vun, with vun being rid of the v- and pronounced wun, and the nazality of -ng carrying over to make the w- into a m-;

There is one instance of a word "ka", which seems to be equivalent to the 佮 we have oftimes seen in Hakka songs, but I can't document it, and the known word is kak;

君 is read kün instead of kiun; this, despite being undocumented in writing, is documented via recordings of the character's pronunciation at the Hakka cloud and at hakkadict, which show variation between kün and kiun;

A variant of 毋过, 啊毋过, is found; I cannot document it, but it seems Justin Richard Leung (@justinrleung on Stack Exchange) can; he said the a- is probably borrowed from Min 猶毋過, one of whose alternate pronunciations is á-m̄-kò;

典礼 is documented as tián-lî, but heard as tián-dî in the sung part; my take is that the -m is an artefact of singing, while the -di could be an alternate pronunciation *tián-dî, or another artefact of singing;

晡日 is pîn-ngit instead of pû-ngit, but this is (indirectly) documented at Wiktionary and the Hakka cloud as an alternate pronunciation; in 暗晡头, however, it's pû we hear;

Phù-poi is not, as I thought, a pronunciation of 扶背 | support, but the (apparently Southern Sixian only) word 匍背 | hunchback / hunched back, documented at the Hakka cloud;

There is one instance of a word mâ, seemingly meaning "also, even"; this appears to be a conflation of a Min loan 嗎 (mà) and the synonym yâ which gave the tone;

Nga and ngài seem to get confused, and in fact it seems ngài is taking the tone of nga, except for the single nga; but more on the tone below;

办法 | phan-fap sounds like phan-fat, and the -p is not there because the lips don't close; no idea whether this can be sourced or not;

瓜仔 | kwâ-é sounds like kwâ-bé, no idea why; I am, in fact, entertaining the idea that another word was meant, perhaps 瓜薈 | kwâ-ve, or 瓜棫 | kwâ-vet, though the e is held long so the latter seems unlikely;

If you see -oy, -ay, -eoi, or -eot, it's because in many -oi she rounds (or seems to round) the -i to a -ü, in many other -oi she seems to front the o and round the i, resulting in an equivalent to Cantonese Jyutping -eoi final, and in Tsung-kwet, she reads kwet as kyeot would be in Cantonese; while the -oy are probably just the rounding lasting too long (a "trailing roundedness" if you will), the -eoi are unexplained because one of them is after k-, where fronting is not warranted by minimization of movement, and kyeot is even more pozzling because it seems the character is usually ket in Meinong, so one could at most guess *kwet>*küet>kyeot or *kwet>*kɰøt>kyeot; actually, -ay and -oy may just be her ending the diphthongs with /ɪ/ as in English instead of /i/ as in Italian;

Remember the 恁 | àn (or rather an) from other songs? Well, as documented on the Wiktionary, Southern Sixian has 盖 | koi instead;

I mentioned ngai and ki, what about nyi? Well, it's ǹg here, which I spell 你, choosing to reserve the spelling 汝 for the nyî (or rather nyì) I'm used to;

The word 蕹菜 is pronounced fung-tshoi, which is reported as an alternate pronunciation in an old dictionary I have no access to; this pronunciation warrants dropping the spelling of the captions, since 瓮菜 would only lead to vung-tshoi;

I choose to use the original Hakka spelling of Meinong, 弥浓, instead of the borrowed Japanese respelling 美浓, but use the latter in Min because the original spelling is a Hakka thing;

故所以 is heard as ko-só-yî instead of ku-só-yî, for no known reason;

从 is heard as tshùng instead of the tshùng it's recorded as in dictionaries; this is perhaps a new-style pronunciation influenced by Mandarin's cóng;

重要 is reported as tshung-yeu for Southern Sixian, but is heard as ending in something of a /jǝw/; Justin would expect a -yau, since Southern Sixian should be closer to Meixian than to Northern Sixian; if the should-be is -yeu, maybe it was retracted to /jɵw/ for smoothness of motion, and if the original was -yau, maybe the a was raised to reduce the movement of the tongue; in any case, we don't know if it's idiolectal or more widespread.

Now that those are tackled (and I hope I didn't forget anything), it's time to tackle the biggest problem: tones. Apart from the mismatches between the study and hakkadict, which is illustrated in the above table, neither match what I hear perfectly. But now, let's discuss the sandhi rules mentioned by the study:

Yin Ping+Yang Ping=324+Yang Ping; this is the reason I matched the contour 24 to this tone; the Yin Ping never sandhies otherwise;

Yang Ping+Yin Ping=Yin Ping+Yin Ping, that is 322+322;

Yang Ping+{Shang/Yin Ru}: Yang Ping=32, that is, the Yang Ping is shortened a little before Shang and Yang Ru, making it harder to tell apart from a Yin Ping;

Yang Ping+{Qu/Yang Ru}: Yang Ping=3->1.5, that is, instead of going all the way down to level 1, it stops half a level higher;

由上聲和陰入組合的二字詞，前字起始音會高於後字: 諸如「上聲+上聲」（廠長），「上聲+陰入」（粉筆），「陰入+陰入」（竹節），「陰入+上聲」（屋頂）四種組合時，會産生前字起始音高於後字，後字音程變小，音量也變輕的現象，調型則維持不變; basically, when a word is made of Shang+Shang, Shang+Yin Ru, Yin Ru+Yin Ru, or Yin Ru+Shang, the first syllable is slightly higher than the second one, and the second one is shorter and lower in volume than usual;

陰平，陽平分別後接上聲或陰人時，上聲或陰入調值會產生差異: 差異的原因乃是因爲陰平，陽平與上聲或陰入分別有著同樣音程的相對語感;陰，陽平結束音高112（調值）的差異即成爲後接上聲或陰入音高（调值）差異了，例：「豬骨」與「牛骨」，因前字「豬」，「牛」結束音高相差半度左右，在「豬骨」與「牛骨」有著相同的相對語感之情況下，兩詞的後字雖同爲「骨」但前者音者會高於後者半度左右; essentially, to distinguish Yin and Yang Ping before Shang or Yin Ru, the second tone is raised half a degree (but the table later in the study gives a full degree, 53 for Shang and 54 for Yin Ru) after a Yin Ping and left intact after a Yin Ping; in numbers, as per said table, Yin Ping + Shang = 322-53, Yang Ping + Shang = 31-42, Yin Ping + Yin Ru = 322-54, Yang Ping + Yin Ru = 31-43.

The last two sound like they should be easily ignorable if we allow "fuzzy matches" to be enough for us. So I set out to get the tone name of each syllable, and this is what I got:

Thi|YP khiu|Q-ngian|Y'P e|Q sam|YP ngiat|Y'R-fun|Q
Ngai|Y'P yu|Q mung|Q to|S ngai|Y'P e|Q a|YP-ma|YP
Te|Q-ha|Q-e|S he|Q ngai|Y'P koy|Q se|Q koy|Q se|Q e|Q sii|Y'P-tsiet|YR
Mi|YP ngit|YR thien|YP mo|Y'P kong|YP
Lu|Q then|Y'P-nun|S a|YP-ma|YP khiung|Q-ha|Q tshut|YR-mun|Y'P hi|Q mai|Q tshoy|Q
A|YP-ma|YP yung|Q tam|Q-kon|YP khai|YP-nun|S fan|YP-su|Y'P fan|YP-su|Y'P yap|Y'R
Khai|YP-nun|S fung|Q tshoy|Q
Lu|Q an|S-ne|Y'P tshung|Y'P Mi|Y'P-nung|Y'P tsong|YP
Yit|YR-lu|Q-hang|Y'P hang|Y'P to|Q Tsung|YP-mun|Y'P
Thi|YP lu|Q hong|Q a|YP-ma|YP i|~Y'P woy|Q ka ai|Y'P kong|S
Tsii|YP-kyun|YP ng|Y'P e|Q ngian|S-tsu|YP khon|Q m|Y'P to|S e*|YP (actually spelled 唉)
Ko|Q-so|S-yi|YP ngai|Y'P hi|YP-mong|Q ng|Y'P kho|S-yi|YP khau|S-song|YP thai|Q-hok|Y'R
Nga|YP nang|Q piong|Q et|YR lok|Y'R sim|YP
Yin|YP-vi|Q a|YP-ma|YP ya|S ki|Q fa|Q
Ngai|Y'P keoi|Q tap|YR-yang|Y'P
Leoi|Y'P khau|S to|S Tsung|YP-kyeot|YR vun|Y'P-fa|Q Thai|Q-hok|Y'R
A|YP-m|Y'P-ko|Q to|Q ngai|Y'P pit|YR-ngiap|Y'R tiam|S-li|YP te|Q pin|YP-ngit|YR
Nga|YP lu|YP tshim|Y'P mo|Y'P to|S a|YP-ma|YP i|Y'P e|Q phu|Y'P-poi|Q
Hip|YR-siong|Q e|Q te|Q-ha|Q-e|S
Nga|YP ma mo|Y'P yit|YR tsung|S fon|YP-hi|S e|Q kam|S-su|Q
Yin|YP-vi|Q tui|Q nga|YP loi|Y'P kong|S
Tsui|Q tshung|Q-yau|Q tsui|Q tshung|Q-yau|Q e|Q ngin|Y'P
Ngai|Y'P tsai|Q mo|Y'P phan|Q-fa|YR khon|Q to|S
I|Y'P e|Q phu|Y'P-poi|Q

Of course, the ma and ka, which are doubtful, do not have a tone from hakkadict. Again of course, this only deals with the spoken part, since tones are neglected in singing. I ignored the pronunciation problems mentioned above for this transcription. The below two-column section shows the resulting tone contours according to the study last, and the contours I hear first.

Thi33 khiu55-ngian51 e334 sam11* ngiat33*-fun22*
Ngai24 yu55 mung55 to442 ngai11 e33* a11-ma33
Te55 ha55-e52 he33* ngai21* koy33* se334* koy33* sei33* e34* sii21*-tsiet2*
Mi22 ngit51 thien24 mo53 kong21
Lu55 then11*-nun11 a11*-ma34 khiung44*-ha44* tshut4-mun21 hi33* mai33* tshoi334*
A11-ma55 yung55 tam45-kon32 khan44*-nun32 fan11-su4 fan24-su31-yap34
Khan44*-nun32* fung33*-tshoi334*
Lu55-an53-ne22 tshung21 Mi11-nung11-tsong11
Yit5-lu45 hang21 hang21 to34* Tsung213*-mun21
Thi11* lu34* hong34* a11-ma24 i21* woi34* ka43 ai32 kong41
Tsii2:2.5-kün11 ng11 e55 ngian41-tsu112 khon44 m43 to32 e21
Ko55-so53-yi31 ngai23 hi21-mong35 ng22 kho22-yi21 khau33 song11* thai33*-hok3*
Nga22 nam45 piong5t et43 lok3* sim11
Yin22-vi55 a11-ma34* ya42 ki22* fa223*
Ngai31 keoi45* tap324-yang324
Loi22 khau33 to332 Tsung22-kyeot32 vun21-fa34* thai34*-hok4*
A55-m53-ko44* to34* ngai41 pit3*-ngiap4* tian32*-li11* te33* pin32 ngit32*
Nga33 lu55 tshim31 mo22 to2.5:1 a22-ma33 i21 e33 phu21-poi334
Hip4-siong55 e55 te44-ha44-e41
Nga11 ma23 mo42 yit2* tsung21* fon21*-hi33* e33* kam32*-su34*
Yin22-vi55 tui55 nga53 loi31 kong442
Tsui54 tshung33*-yau33* tsui32* tshung22*-yau22* e22* ngin21*
Ngai31 tsai44* mo42 phan34*-fa32* khon44* to31*
I33 e55 phu32-poi44

Thi|322 khiu|55-ngian|3:1.5 e|55 sam|322 ngiat|55-fun|55
Ngai|3:1.5 yu|55 mung|55 to|42 ngai|3:1.5 e|55 a|322-ma|322
Te|55-ha|55-e|42 he|55 ngai|3:1.5 koy|55 se|55 koy|55 se|55 e|55 sii|32-tsiet|43
Mi|322 ngit|43 thien|324 mo|322 kong|322
Lu|55 then|32-nun|42 a|322-ma|322 khiung|55-ha|55 tshut|43-mun|3:1.5 hi|55 mai|55 tshoy|55
A|322-ma|322 yung|55 tam|55-kon|322 khai|322-nun|42 fan|324-su|31 fan|324-su|3:1.5 yap|55
Khai|322-nun|42 fung|55 tshoy|55
Lu|55 an|42-ne|31 tshung|31 Mi|31-nung|322 tsong|322
Yit|43-lu|55-hang|31 hang|3:1.5 to|55 Tsung|324-mun|31
Thi|322 lu|55 hong|55 a|322-ma|324 i|3:1.5 woy|55 ka|? ai|32 kong|42
Tsii|322-kyun|324 ng|3:1.5 e|55 ngian|42-tsu|322 khon|55 m|32 to|42 e*|322 (actually spelled 唉)
Ko|55-so|42-yi|324 ngai|322 hi|322-mong|55 ng|32 kho|42-yi|322 khau|42-song|322 thai|55-hok|55
Nga|322 nang|55 piong|55 et|43 lok|55 sim|322
Yin|322-vi|55 a|322-ma|322 ya|42 ki|55 fa|55
Ngai|3:1.5 keoi|55 tap|43-yang|31
Leoi|32 khau|42 to|42 Tsung|322-kyeot|43 vun|3:1.5-fa|55 Thai|55-hok|55
A|324-m|3:1.5-ko|55 to|55 ngai|32 pit|43-ngiap|55 tian|42-li|322 te|55 pin|322-ngit|43
Nga|322 lu|324 tshim|31 mo|32 to|42 a|322-ma|324 i|3:1.5 e|55 phu|3:1.5-poi|55
Hip|43-siong|55 e|55 te|55-ha|55-e|42
Nga|322 ma|? mo|32 yit|43 tsung|42 fon|322-hi|42 e|55 kam|42-su|55
Yin|322-vi|55 tui|55 nga|324 loi|32 kong|42
Tsui|55 tshung|55-yau|55 tsui|55 tshung|55-yau|55 e|55 ngin|31
Ngai|3:1.5 tsai|55 mo|3:1.5 phan|55-fa(t)|43 khon|55 to|42
I|3:1.5 e|55 phu|3:1.5-poi|55

Asterisks denote where a shift of the frame of reference for the tone levels would improve the match between heard contours and expected contours. The below tackles whatever I don't think I can explain away as fuzzy matches or tone fluctuations.

The first problem is obviously the ngian, which starts way too high; I think at least some of the recordings for 舊年 at the Hakka cloud do have a ngian51 (or ngien51), but I can't slow them down, so I can't be sure; talking speed probably cuts the jump short;

Then we see the ngai24; now, I think the -i and the following y- merge, and continuously grow from 2 to 5, so I think I can reread the ngai as a 23 or 22 or 33, which still don't match a Yang Ping, but do match a Yin Ping, and many other ngais later on strongly suggest it's ngâi, not ngài: the next one is 11, then we have 21, then there is that ai32, then ngai23, which is more of an accidental fluctuation from an intended ngai22 or ngai33, and these are more Yin Pings than Yang Pings, then there is a ngai31 which would be a Yang Ping, then ngai41 which is just weird but more of a Yang Ping, then ngai53 which is literally no tone, and finally a ngai31; OK, looking at this list, and at the speed of the speech in each instance, it seems that the Yin Ping appears when ngai is spoken fast, and Yang Ping stays when it's spoken slower (with some funky 41 and 53 businesses); I think we can explain this as a combination of speed cutting the drop short, and conflation with nga, which is indeed ngâ, and is probably merging with ngài; in fact, the ngai53 I reported was a nga53, meaning the two are being tonally and semantically conflated (with funky business, again); so I guess I'll keep the "root tones" ngài and ngâ, with this caveat, and then the nga53 is just weird;

The next problem is obviously a-ma; this should be â-mâ, but almost every time it appears as two level tones of different heights; in only one instance does it behave like two Yin Pings: in front of ì woi kạ ài kóng, where the sandhi rule applies and mâ is ma24~ma324; however, the sandhi should not occur, because there is a pause after â-mâ; also, in another instance where it's followed by ì, it doesn't sandhi; so we either assume it's actually â-ma, and explain that weird sandhi as an even weirder huge upward drift in pitch, or we stick to â-mâ, and have way more problems; I go for â-ma; what do you think @justinrleung?

To saves itself by a hair thanks to raising the reference :);

What is going on with that ngit51 which is supposed to be ngit43?

Mo is weird; it seems its tone is adaptive: it's supposed to be mo31 + sandhi, but mo53 is not explained by that, nor is mo22 or mo33, so I guess we can only say its tone is rendered unimportant by how common it is, and what contexts it appears in: in all three instances of the word, it is clear that it must be a negative element, so even if the tone adapts, the message is clear, and adapting the tone allows the pitch to vary continuously instead of being "jumpy"; in view of that, we just leave the etymological tone: mò;

The particle nún appears three times; in two of those instances, it is apparently lower than a Yin Ping, and in the other one it is just as low; however, the first two instances are preceded by another doubly problematic word; see below for that, but my conclusion is that that word is a Qu tone (khan), which makes raising the reference a trick to get a nun43~nun42, so nún confirmed; what about the other instance? Well, it seems to be nun21 or nun31, which sounds Yang Ping, but the preceding Yin Ping would have to sandhi, which it doesn't, so it must be nún; I guess it's just de-emphasised so the reference lowers;

A mention must be made of how the first fan-su has wrong tones that she corrects afterwards; essentially, the wrongness is fân-su instead of fân-sù, with the different sandhi that follows from this; hence my transcription;

I naturally have to re-mention the khâi/khan thing at this point; it is evident that she is saying khan: high level tone, nún starting below it; so that is how I will transcribe it; I will wait for a comment by @justinrleung on whether the captions have the wrong word and it should be the actual khan that is used here;

The yap is a raising checked tone, which shouldn't exist, and it's like a reversed Yin Ru; raising the reference risks screwing up the preceding Yang Ping, making it a 42 Shang; but maybe it's just de-emphasised as it ends a three-syllable word, something like the neutral tone in Mandarin;

an-ne is reported as án-nè or án-nê, and the level tone of the ne makes it clear we have the latter; as for an53, well, remember Shang+Yin Ping=53-322? Well, that's another reason for positing á-nê :);

Mi11-nung11 could only be explained as Mî-nûng, but that is not reported anywhere that I know of; so I either have to posit this Mî-nûng sound, or take Mì-nùng and assume the mì got cut short by speech speed; Mî-nùng definitely doesn't fit, because the Yin Ping would sandhi to 324 and that doesn't happen; @justinrleung what do you think?

I should be ì, but it is always level; or rather, it seems like it drops like half a degree; that would suggest a Yin Ping î; of course, there is that questionable sandhi that happens once out of two instances of the same word combination, which would make it a Yang Ping; @justinrleung, what say you?

Tsii-kün seems to have a double problem; the kün shouldn't actually sandhi to 324, because between Tsii-kün and ǹg there is a pause; still, we should have the same tone, which we don't; besides positing a tsii-kü̂n, so assuming tsii is a Qu tone, the only explanation I have is that a sequence of level tones takes a decreasing pitch curve; I would do that in Mandarin with Zījūn, probably; @justinrleung what do you think?

The m43 supposed to be m32 is probably another blatant negative whose tone adapts to its surroundings because it's not essential to comprehension;

If we raise the reference and read khon55 m54 to43 e32, it's all fuzzy matches, except m disucssed above, so e is also OK;

The so in ko-so-yi sounds like it's being sandhied as if in front of a Yin Ping, except the following is a clear Yang Ping; WTF?

The mong has a strange big pitch raise which I'm reluctant to explain away as a drift; @justinrleung?

The ng does seem to have a slight drop in pitch, and given it's very short, it could be ǹg; the other ng in the line before, however, definitely sounds like a Yin Ping n̂g; @justinrleung?

kho in kho-yi is a level tone and should be a Shang; seems to be a Yin Ping; @justinrleung?

khau has the same problem; again, a Yin Ping would fit, in both instances of the word; @justinrleung?

The reference raise for the lok could be justified by end-of-phrase tone drop, something like a neutral tone; @justinrleung what do you think?

Then we have ya, which should be yá but sounds like yâ; now, the recording here clearly has ya22 ha44 for yá-ha, in both versions, which would sound like yâ-ha, not yá-ha, so it seems the tone varies; maybe this is another form of erosion (speaking of which, the second recording for Sixian has khon-e-to for khon tet tó, pretty extreme erosion); here it's even weirder: ya53 ha33; moedict doesn't play audios, and hakkadict doesn't even display play buttons, so I can't hear those recordings; oh wait, relistening that is not ya22, but ya42 as expected, and the following Qu tones are 22, perhaps because they end the sentence;

In tap324-yang324, tap32 is a fuzzy Yin Ru, and the 4 is a drift, perhaps 323 actually; as for yang324, there is an ending drift, but the rest is more of a yâng than a yàng, it's a level tone; @justinrleung how come?

Â is definitely not what is here, but then again 啊 allows essentially any tone, and this comes from á-m̀-kò in Min where the first tone is level by sandhi, so I'll go for a-m̀-ko;

The ngai shouldn't actually sandhi because it's followed by a pause, so it should be a 31, add in some speech speed that cuts the jump from to34 down to the 3 of ngai31 short, and ngài is saved;

Not raising the reference on pin and raising it on the whole rest is probably justified by the exact tone being unimportant given the collocation right before ngit which means "day" in and of itself;

I posit lû (once tû) and lu (once tshiu) are merging to a single lu, and that is what we find here; the sandhi shouldn't apply because there is a break after it, so it should be lu322, still not lu55;

As mentioned above, what was once ma (min loan) is probably getting conflated with the synonym yâ, and becoming mâ, which fits the ma23 here thanks to sandhi;

We raise the reference from yit on because yit tsung ends the first part of the sentence, and the rest keeps the lowered reference;

That nga53 is unsandhied and too high, perhaps because of how common it is, so another "tonal erosion" to achieve continuous pitch; @justinrleung what do you think?

ngin needing less reference raising than the rest is probably because the tsui tshung-yau before it is a repetition, and the e is a common word; @justinrleung?

Finally, the hi in fon-hi sounds like a Qu or Yin Ping, but should be a Shang; @justinrleung?

Stack Exchange Network

Questions on a Hakka song and its particular Hakka dialect

Warning, General Background, Table of Contents

First rough transcription of the sounds and character approximation

A couple tweaks to the character spelling

Comments on the apparent Min loans

Comments on tones

Raoping or Southern Sixian?

Getting creative with sandhi

Tone fluctuations: what are those tones?

Final transcription and character rendition

Questions

3 Answers 3

Precise transcription of what I hear

Transcription with Southern Sixian Hakka tones as found in hakkadict

Your Answer

Not the answer you're looking for? Browse other questions tagged
pronunciation
topolect
spelling
hakka
or ask your own question.

Linked

Hot Network Questions

Questions on a Hakka song and its particular Hakka dialect

Warning, General Background, Table of Contents

First rough transcription of the sounds and character approximation

A couple tweaks to the character spelling

Comments on the apparent Min loans

Comments on tones

Raoping or Southern Sixian?

Getting creative with sandhi

Tone fluctuations: what are those tones?

Final transcription and character rendition

Questions

3 Answers 3

Precise transcription of what I hear

Transcription with Southern Sixian Hakka tones as found in hakkadict

Your Answer

Sign up or log in

Post as a guest

Not the answer you're looking for? Browse other questions tagged pronunciationtopolectspellinghakka or ask your own question.

Linked

Related

Hot Network Questions

Not the answer you're looking for? Browse other questions tagged
pronunciation
topolect
spelling
hakka
or ask your own question.