45,234
edits
(→See also: Twenty-seven Taiwanese words beginning with the backquote) |
(→Method: update. 1,800 distinct sounds) |
||
Line 2: | Line 2: | ||
== Method == | == Method == | ||
We isolated 2,936 rows from the dictionary that are monosyllables and converted their TRS to MTL. We only considered words from the first section of the dictionary because they appear to be frequently used, and ignored the second section. Then we counted the frequency of each | We isolated 2,936 rows from the dictionary that are monosyllables and converted their TRS to MTL. We only considered words from the first section of the dictionary because they appear to be frequently used, and ignored the second section. We folded in backquoted words, for example ''`lie'' was counted as ''lix''. Then we counted the frequency of each sound with Python's collections.Counter, which tells the number of homophonic dictionary rows, and got 1,800 distinct sounds. Then we used Counter again on those results and found: | ||
* | * 1853 rows (63%) are homophonic, 1083 rows (37%) are not | ||
* | * The most homophonic sounds are: ''{{x|lie}}'', ''{{x|ky}}'', and ''{{x|køf}}'', which match 7 rows each, followed by ''{{x|cie}}'', ''{{x|kafn}}'', ''{{x|kefng}}'', ''{{x|sefng}}'', ''{{x|kaf}}'', ''{{x|kaq}}'', ''{{x|zngf}}'', ''{{x|ti}}'', ''{{x|sw}}'', ''{{x|leeng}}'', and ''{{x|kerng}}'', which match 6 rows each | ||
** most homophones cover two rows: 896 rows (31%), 448 distinct sounds (25%) | |||
** most | ** some sounds cover three rows: 516 rows (18%), 172 sounds (10%) | ||
** some | ** the rest match from four to seven rows: 441 rows (15%), 97 sounds (5%) | ||
** | |||
[[File:rows and matching MTL vs. match level.png|thumb|none]] | [[File:rows and matching MTL vs. match level.png|thumb|none]] | ||
edits