I guess the way I think of it is basically as follows: Written Chinese tends to suit Mandarin well. Since all modern varieties of Chinese essentially share the same ancestor (Middle Chinese) and borrowing between varieties (especially from Mandarin into other varieties) is common, it suits the other varieties of Chinese to varying degrees as well. However, there is usually no (standardized or widely accepted) way of encoding non-Mandarin dialect-specific features in written Chinese (with some notable exceptions, e.g. Cantonese can and is actually written with characters that represent Cantonese-specific words).
It's relatively easy to write Mandarin in Chinese characters; people do this all the time. To some extent, even dialect variation within Mandarin can be represented using Chinese characters. However, representing spoken Cantonese using Chinese characters is much harder. Representing Qingtianese using Chinese characters is almost impossible since the etymologies of Qingtianese words are often rather unclear. When people do try to write Qingtianese with Chinese characters, what they mostly end up doing AFAICT is just using characters whose Mandarin pronunciation sounds similar to the Qingtianese words.
佛手瓜 is literally Buddha + hand + melon, so I guess a melon that looks like Buddha's hand? In general, in Chinese, it's very common to put nouns together to form what may seem like very long compounds sometimes. Sometimes, these can be broken up using the possessive particle 的 (pronounced
de in Mandarin; I'm not sure to what extent other varieties of Chinese use this particular particle), but (Mandarin) Chinese-speakers avoid using too many 的's in succession. 佛手瓜, however, is a fixed expression. In Mandarin, it's pronounced
fóshǒuguā. In Cantonese, it's
fat6 sau2 gwaa1. In Taiwanese, it's
hu̍t-chhiú-koe.
That totally made sense and was easy to understand, right?