How come did the readings of the xing-category appear for 行?
They make no sense. The Early Middle Chinese for “to walk” is γæng, which should give héng – and so it does, in 胻, 蘅, 桁 – even with the same phonetic component. How come did the aberrant xíng appear from nowhere? Same for γængH “conduct”, which even has the Taiwan reading xìng, woth the correct tone, but not the correct… whole of syllable!
And why does the hang stay unaffected? As hang “trade” is the purely expected háng!
TL;DR: vowel shift + possible colloquial-literary split.
行 has the reading 戸庚切 relevant to this meaning of "to walk", reconstructed as /ɦaɨjŋ/ by Pulleybank and /ɦˠæŋ/ by Zhengzhang. The vowel in Middle Chinese 庚 is open division II (開二等). This 庚, as well as the other characters mentioned 桁 and 衡, end up with Mandarin -eng, which is the regular development.
However, at some point in Mandarin (but not Cantonese), the vowel closed, making 行 homophonous with 刑 (Middle Chinese reading 戸經切, open division IV 開四等, Pulleyblank: /ɦɛjŋ/, Zhengzhang: /ɦeŋ/). This merger was already attested by the time of the 中原音韻 Zhōngyuán Yīnyùn from the 1324, in the Yuan period. However, of the above set, only 行 is the only one of the set from the 中原音韻 to complete the merge, with 桁衡珩 being 'regular' and ending up with -eng. This may be due to a literary-colloquial split, with 行 having a more innovative colloquial pronunciation, though it does retain a separate marginal "regular" pronunciation in the standard Mainland Mandarin pronunciation of 道行 dàohéng.
When 行 changed its vowel to /e/ or /i/, the now devoiced /h~x/ initial in the right environment to regularly palatalise, complete in Beijing by at latest the mid-Qing period, though not in Nanjing. By 1815, when Morrison compiles his dictionary, 衡 is split from 行, which has decisively moved into 刑. After the 尖團合流 palatalised the alveolars too (e.g. merging 餳 into xíng), we arrive at the modern situation.
to start-- there is no official answer, because we don't actually know how middle chinese sounded. we have educated guesses, but its always good to keep in mind that at least some of the world's theories are probably wrong. middle chinese has never been heard even by the oldest living people's great great great grandparents.
行 has five recognized pronunciations in standard mandarin-- háng; hàng; héng; xíng; xìng. I did not find any prevaling theory on the specfics, but we can talk about why these kind of changes happen in general.
originally it meant crossroads, which slowly changed over time to mean roads itself, and then walking/travelling itself. it can also have a variety of other meanings- the meanings themselves aren't relavant but the fact they exist may be.
language is not a clean and tidy thing. sometimes there are very logical changes to a words use, and sometimes it is not, it just changes organically. some examples of what it could have been:
a separate word was merged together and combined meaning.
the word it used to be was often used for something else, and so there was a natural change in vocabulary.
some other language influenced chinese at that time and was brought into the language.
and many many more. none of these are chinese specific, but things that happen every day in languages around the world. they also happened in the past too.
please note, it is also possible that it was always there to begin with and we don't have a good record/theory for it. the definitions associated with xing have been documented since at least early middle chinese period, if not older than middle chinese itself.
its far less conclusive, but also note that both cantonese and mandarin have two completely different pronunciations of this character, depending on definitions. so regardless of the reason that happened, the chance of it existing in a shared past language (middle chinese or older) is very plausible.
