Also, this is the Wikipedia link for the Hanyu Pinyin-based braille (the above one is wrongly-encoded).

1) Cantonese braille doesn’t use space for word division, rather it appears only when there is a null initial or the syllable is at the high level tone since tone indication is obligatory. Hence, reading Cantonese braille is not very different from reading Thai in print: as each syllable takes 3 cells, you read it continuously without word break. I also recalled that the capital symbol is used to signify English or other latin alphabets.

As you can see, it is based on older Cantonese transliteration so a p is a [p] rather than a [pÊ°] in Hanyu Pinyin and [pÊ°] is derived from the symbol of p. It seems that Hadley’s information is in disagreement to the one in Wikipedia: Hadley’s code utilises braille b, d andg for [p], [t] and [k], and I can’t tell which is right or wrong unless I see Cantonese braille transcription which exists in my apartment’s lift back in Hong Kong.

However, some Cantonese braille codepoints may have been done a triple job: depending to its position in the syllable representation, a cantonese braille cell can be representing an initial, a final and a tone!

2) Chinese Mandarin braille also comes in two flavours just like Chinese characters comes in two flavour in the Chinese region: Zhuyin-based braille used in Taiwan and Hanyu Pinyin-based braille used in Mainland China.

Hanyu Pinyin-based braille merges g/j, k/q, h/x as they are in complementary distribution. Initials and Finals don’t share codepoints unlike Cantonese braille, and the syllables are not in fixed width: the shortest syllable can be in one cell (silibant syllables or null-initial syllables in first or light tone) and the longest in three (with tone indication). All punctuations except ecplisis uses two cells (by shifting the original representation in ordinary braille by one column).

Zhuyin-based braille is slightly different: the codepoints are distributed differently, sometimes with bewilderment if you know only English Braille. For example instead of merging k and q into one codepoint, q and c are merged, and instead of merging h and x, x and s are merged. All “empty rhymes” have to be indicated by a cell that is the same as “er”, and all tones are indicated not matter what: even light tone is indicated, not like Hanyu Pin.

3) Two-cell Chinese Braille invented in the 1970s is great stuff: it’s concise and it’s uniform as each syllable takes two cells: initial and medial are encoded in the first cell, final and tones are encoded in the second cell. More important, because of this conciseness, it also allow word-defining codes: they are not meant to be pronounced but they would indicate which word it is, more like the way one might disambiguating the syllable [ɹaɪt] on a phone: write (as in writing) vs right (as in opposite to left). The words in the brackets defines the word, with a rather high homophony like in Chinese, and allow introduction of classical literature, word-defining codes are great literate aid for the Blind. This is not well-used in the common Chinese braille because it adds even more bulk.