A rationale for custom notation from Merriam-Webster's Twitter:[1]
> IPA has some problems if you grew up speaking English: some of the symbols can be deeply confusing.
> For example: the symbol 'i' (lower-case 'i') stands for the sound /ee/.
> In IPA, the symbol 'j' (lower-case 'j') stands for the sound of /y/ as in 'yes.'
I find it amusing how the Great Vowel Shift sets English apart from other Latin-alphabet languages, where long vowels are overall very alike compared together to modern English: The names of the vowels, 'A E I O U', are also their long sounds; typically (IANALinguist) outside of English they're instead close to something like (spelled Ephonetically) 'ah eh ee aw ooh', which is also reflected in the IPA assignment of /a/, /e/, /i/, /o/, and /u/. Funnily, the same even applies to languages in alphabets and even non-alphabetic writing systems, since the transliterations and transcriptions into the Latin alphabet are also based on vowel qualities from the Latin language itself.
Together other English peculiarities such as 'j' and silent 'e', it seems that when "fake-pronouncing" words from an unknown language by naïvely applying the pronunciation of a known language, using English guarantees being far off the actual pronunciation, with a choice of any other language being significantly more likely than not to result in a closer pronunciation.
English scored the worst on reading by far at 31% (the next lowest was Dutch, at 55%). It also has one of the lowest writing scores at 36%, beating only French (28%) and Chinese (20%).
It also shows what a well-designed phonemic orthography can do: Serbian scored 99% for both reading & writing, indistinguishable from Esperanto.
> IPA has some problems if you grew up speaking English: some of the symbols can be deeply confusing.
> For example: the symbol 'i' (lower-case 'i') stands for the sound /ee/.
> In IPA, the symbol 'j' (lower-case 'j') stands for the sound of /y/ as in 'yes.'
I find it amusing how the Great Vowel Shift sets English apart from other Latin-alphabet languages, where long vowels are overall very alike compared together to modern English: The names of the vowels, 'A E I O U', are also their long sounds; typically (IANALinguist) outside of English they're instead close to something like (spelled Ephonetically) 'ah eh ee aw ooh', which is also reflected in the IPA assignment of /a/, /e/, /i/, /o/, and /u/. Funnily, the same even applies to languages in alphabets and even non-alphabetic writing systems, since the transliterations and transcriptions into the Latin alphabet are also based on vowel qualities from the Latin language itself.
Together other English peculiarities such as 'j' and silent 'e', it seems that when "fake-pronouncing" words from an unknown language by naïvely applying the pronunciation of a known language, using English guarantees being far off the actual pronunciation, with a choice of any other language being significantly more likely than not to result in a closer pronunciation.
[1] https://twitter.com/MerriamWebster/status/130415192278247015...