The last post about the predictive power of Japanese to guess the differences between Japanese and Qingyuan Cantonese might give one the wrong idea. The Japanese sound of the words might still be less similar than the sounds of Mandarin or any other Chinese dialect to the sounds that were being predicted. Guangzhou Cantonese (QC) sounds are most similar to GC but when there is a difference between the two one wonders where it comes from. Perhaps to understand why I would spend time poring over other Asian languages it might be helpful to explain my previous discarded assumptions.
Despite some familiarity with GC, when I arrived in Qingyuan I was not comprehensible to them nor they to me. It was like a student who'd only learnt basic NZ English arriving in Glasgow. Qingyuan was one of the Guangdong cities with large Hakka populations; in fact, the family I've been with are descendants of Hakka. The grandfather's mother tongue is Hakka yet raised his family in QC (which he also speaks natively). My first thought was that Hakka might have influenced the Cantonese of the city. There are some word choices that were similar (they have the same non-GC non-Mandarin way of saying peanuts, di-daau, for example). But further checking of some of the GC/QC differences came up short: Hakka could not explain some of the sounds or variance.
QC has several word choices that differ from GC. Take off (shoes/clothes) in GC is cheui, whereas in QC it is tyut; this is the same character as tuo in Mandarin. Yesterday in GC is kamyat but zoyat in QC; this is same characters as Mandarin again, zuori. So I wondered whether it might have been moderated by Mandarin in some way. But even more strongly, Mandarin if anything is the least predictive of any dialect or language to explain some of the pronunciation differences, less so than Hakka.
Why shouldn't neighboring or influential languages predict these? Perhaps, word choices can travel between dialects but actual pronunciation is like the genes of a language. If the languages are different breeds, they don't often mix. (I can think of only one circumstance where they would feasibly do so but it isn't the case for GC/QY so I'll keep it to myself.) Just like German wouldn't absorb French or English vowels, so wouldn't QC accept the sounds of Hakka or Mandarin, even if word choices were transferrable.
Which of course brings me onto Korean and Japanese, which both predict many features of Qingyuan effectively. I've got enough basic fluency and awareness of both to be able to crank out the analysis easily. Korean nailed (100% predictivity) a frustrating difference between GC/QC. QC speakers say sounds ending -i in GC as either -i or -yu; for example, zi (to know) is the same in the two dialects; yet zi (self) in GC is zyu in QC. At first, Mandarin seem to predict some but not all (about 70%). Then I recalled a strange difference when learning Korean.
I learnt Korean after Mandarin had became a strong second language for me. When learning Korean (a phonetic alphabet) I'd often notice the words that came from Chinese and guess the characters that they came from. Like Japanese, it had "Cantonese features": end sounds that Mandarin no longer has. But there were some sounds in Korean that didn't make sense to my Mandarin-mind. In Mandarin two words might be said "shi" but in Korean it would be either "si" or "sa". So I'd have to make sure I learnt the different ones well so not to overgeneralise from Mandarin (letting my L2 affect my L3, a common problem for many students with other languages). Recalling this, suddenly I realised that all the Korean "-i" were "-i" in QC, and all the "-a" sounds that corresponded to "-i" words in GC were "-yu" in QC. I'll be doing the final check when I go back for May Day holidays. Another mystery solved!
Having the same thoughts solved another anomaly, this time Japanese predicts when QC would use an ng- initial versus no initial. There are only a couple of other particular distinctions that aren't explained (yet) by going back, or is that across, to Japanese and Korean. Which really makes you wonder why these two languages predict so well.
Both Korean and Japanese have phonetic alphabets. Phonetic alphabets are like film in a camera when exposed to a foreign language. Both Japanese and Korean were exposed to a form of Ancient Chinese, absorbed many of the words and found a way to write down those sounds with the alphabets they had. 60% of Korean vocabulary originates from Ancient Chinese. Japanese may have a similar quantity mostly borrowed in the 5th - 9th centuries CE. This ancient sound interpretation process noted differences and attempted to reflect that. That these sound differences don't exist in Mandarin or GC, yet do in QC is an interesting surprise.
Though most of the energy of discovery has gone from my rampant comparative phonology, this field trip into the depths of this language and these languages has yielded some interesting common sense that are useful as a lens to understand any language, its evolution and the curiosities within. It has been a trip worthwhile.