Decodings

Please note that the shown decodings can only be approximations. Quốc Âm Tân Tự is not fully decipherable because the original rhyme table is written in Chữ Nôm, which often has ambiguous pronunciations. Furthermore a lot of the annotation characters don't exist in any dictionary, so the intended pronunciation is completely unknown. This leads to the rhyme table being underdetermined, not allowing for full recovery without asserting additional hypotheses. Another aspect leading to inconsistencies is that the rhyme table assigns at least six types of rhymes (differentiated by coda) to only five symbol groups denoted by 5 markings, shown above as 5 rows. This led to the 6th rhyme type being scattered all over the place where phonotactical restrictions left an empty slot. A small number of these fill-in rhymes have unclear or competing Chữ Nôm pronunciations and can only be deciphered by applying the bijection hypothesis and guesswork. These are mostly the 8 rhymes in the bottom right corner. The positions in the last column and the last row suggest that those are the rhymes the author assigned last.

The following is a list of hypotheses that are used in the decodings shown on this page.

Hypotheses

Bijection hypothesis - Assumption: Even though some of the original annotations have the same rhyme in modern Hán Nôm, all known decodings assume that each rhyme occurs only once. Used by: omniglot.com, qatt.org

Pair hypothesis - Assumption: The rounded variant of a rhyme and its unrounded rhyme occur pairwise. This hypothesis often includes the assumption that the khẩu radical (口) often denotes the roundedness of a rhyme instead of being part of the Chữ Nôm character itself. Used by: qatt.org, to a lesser extent by omniglot.com

Symmetry hypothesis - Assumption: Each rounded rhyme is the mirror image of its unrounded rhyme. This is a stricter variant of the pair hypothesis. This leads to a shift of a number of rhymes that would otherwise violate this condition. Used by: qatt.org

Nucleus hypothesis - Assumption: The 22 base symbols represent the 22 nucleus rhymes of the Vietnamese language. Used by: qatt.org

Minority language hypothesis - Assumption: The superfluous nh/ng rhymes represent certain rhymes of minority languages ending in ng. This hypothesis excludes the vận thư hypothesis. Used by: omniglot.com

Vận thư hypothesis - Assumption: The 6 superfluous rhymes ending in nh/ng reflect the fact that old rhyme dictionaries distinguish between rhymes that modern Quốc Ngữ write with nh coda. E.g. while Quốc Ngữ knows only one anh rhyme, old rhyme dictionaries distinguish two separate anh rhymes, which have merged in modern Vietnamese. The reasoning is that the original document shows strong knowledge in ancient rhyme tables. This hypothesis excludes the minority language hypothesis. Used by: qatt.org

Ư hypothesis - Assumption: The rounded version of ư is ươ. Although in modern Vietnamese phonology this is not true, the original rhyme table suggests that the author pairs these two rhymes together. Used by: qatt.org

Ơ hypothesis - Assumption: The modern "anh" rhyme in Quốc Ngữ is equivalent to ơnh in Quốc Âm Tân Tự. Therefore "anh" is grouped together with rhymes that have the ơ nucleus. Used by: qatt.org

Decodings

The omniglot.com decoding
Bijection hypothesis Pair hypothesis Minority language hypothesis
This decoding optimizes best fit with modern Quốc Ngữ pronunciation. 59% of the rhymes have a pronunciation that match at least one of the modern pronunciations of the original Chữ Nôm annotation. The rest is more or less similar but not exact. This decoding uses the minority language hypothesis and therefore assigns rhymes like ooeng and uưng. It also assigns the rhymes ung, ong and ông even though the inventor of Quốc Âm Tân Tự explicitly described a separate method to write these particular rhymes called Tham Thư.

The qatt.org decoding
Bijection hypothesis Symmetry hypothesis Nucleus hypothesis Vận thư hypothesis Ư hypothesis Ơ hypothesis
This decoding trades exact pronunciation matches for homogeneity. It maintains an overall similarity with modern pronunciation instead of seeking exact matches. 38% of the rhymes exactly match modern pronunciation. The rest is more or less similar but not exact. This decoding uses the nucleus hypothesis and the vận thư hypothesis to handle inconsistencies found in Chữ Nôm annotations of the original rhyme table.