雙倒水別莊

The phonemists are short-sighted in at least three aspects. Firstly, they forget that the limitations of the roman alphabet do not allowe one-to-one correspondences, for there are more phonemes in TW than letters in roman alphabet. As a result, what they designed is not what they proclaimed it to be. For instance, TLPA does not alter POJ's use of h for both initial glottal fricative and final glottal stop. [omitted] Secondly, they are not aware of the non-uniqueness of phonological analysis (Chao Y. 1934). That is, they do not know that there can be more than one solutions to a phonemic problem, that phonemics is a 'hocus-pocus'. For instance, the syllabic inventory written as am, iam, an, ian, ang, iang, eng, im, im, un, om, ong, iong in POJ is rearranged by designers of TLPA as am, iam, an, ian, ang, iang, ing, im, in, un, om, ong, iong for what they consider to be in pattern congruity. Nevertheless, the TLPA activists are unaware of better solutions in the 'hocus-pocus' phonemic game. Since ian is phonetically [en] in the majority speech, one could arrange the inventory above as am, iam, an, en, ang, iang, eng, im, in, un, om, ong, iong, for instance, for even better congruity. The game of phonemicization could even change un to on or change om, ong, iong to um, ung, iung for economy's sake. Thirdly, they overlook the fact that, in romanized writings as well, discreteness in forms enhances reading and helps to disambiguate them (Chao Y. 1923).     (Tiuⁿ 2002:3)