Okay this is way overdue and it won't be very satisfying but I'm going to try and get the big elements of Kangshuic diachronics out of the way here. So, first things first, the approximate shape of Proto-Kangshuic. Phonologically, Proto-Kangshuic had a consonant inventory of approximately something like this:
Code:
p t k q
b d
s ʃ x
z ʒ ɣ
m n ŋ
w r l y
It allowed a maximum syllable of CCVC and had seven tones, which were partly distinguished by pitch and partly by phonation (breathy/modal/glottalized). I am significantly less certain on the vowel inventory, which may have had a length distinction or contained diphthongs, but probably one of them. In any event the pure vowel qualities were probably something like /i ɯ u e o a/, but until I've finalized all the sound changes, who knows.
Morphosyntactically, the broad strokes of PK are something like this: PK was a moderately synthetic language with about an equal amount of prefixation and suffixation. Person marking, that was already weakening, was prefixal, as were certain markers of mood and nonfinite markers, while the bulk of TA marking happened suffixally. The language was split-ergative and had a nominal case system of five items: Absolutive, Genitive-Ergative, Accusative(-Allative), Dative(-Comitative) and Locative. There was already a noun class system, derived from a system of numeral classifiers, which is continued in most branches of Kangshuic. The strongly variable morphosyntax in the various descendants is a legacy of the more clitic-like nature of the class markers in PK. There was some amount of nonconcatenative morphology: some combinations of suffixes, due to contraction of complex codas in PPK, produced a marked stem change, and the third-person object marker was a tone change.
The categories distinguished in the verb are person marking for both agent and patient (agent marking is identical to S marking for all verbs), aspect (unmarked, progressive-imperfective, completive, inchoative), tense-mood (unmarked, past, remote-irrealis) and two participle forms that are otherwise unmarked (approximately present and past). Voice was indicated either through simple detransitivization or through periphrastic constructions.
Obviously, there were many changes from this to descendant languages. The Kangshuic family divides itself into around four branches, although the identity of the branches with actual genealogical points of divergence is by no means clear and several isoglosses transcend branch boundaries. In the absence of geographical information other than some MS Paint maps, I will refer to them as Kangshuic A, B, and C, as well as Pirka, a single language that forms a branch of its own.
Pirka is distinguished from all other Kangshuic languages by the simple fact that it did not undergo Kangshuic final loss, a sound change affecting all other families in which final obstruents and nasals were lost, with fricatives vanishing altogether, all stops merging into the glottal stop and final nasals leaving behind vowel nasality. Pirka, however, almost fully retains the original PK inventory of finals, modulo its global loss of the voicing distinction. However, the distribution of finals is significantly different from PK, depending largely on whether the syllable tone was originally glottalized or not. Pirka also only has two tones, merging the original seven first to four, with two binary distinctions of high vs. low and glottal vs. modal, and later losing the glottalization distinction with a concomitant consonant split. It shares the innovation of aspirated stops from fricative-stop sequences with large parts of Kangshuic A and B. Its vowel inventory is /i ĩ e ə ɛ a ã ɔ o u ũ/.
Morphosyntactically, Pirka is extremely innovative, having lost basically all inherited verbal morphology and being highly isolating compared to other Kangshuic languages. The noun class system in Pirka is expressed through determiners that obligatorily accompany nouns, and unusually for Kangshuic languages express a definite-indefinite distinction. The case system is completely gone and has been replaced through various serial verb constructions, like in large parts of Kangshuic A.
Kangshuic A is the best-attested and most widely spoken branch of Kangshuic. Old Kangshi, ancestor of Court Kangshi and several related dialects, is the earliest (and in continuity with the modern forms, basically only) written Kangshuic language, and already shows several distinctly Kangshuic A features, especially in phonology and morphology. Kangshuic A has full loss of final consonants, reflecting their influence in vowel quality and tone, and all languages in the branch have undergone a change of *w > b, *y > ɟ, *r l > dʐ. Old Kangshi reflects these as /b z ʐ/, respectively. The vowel inventories of Kangshuic A languages are generally quite large, including generally six to eight monophthongs, with at least one back unrounded vowel, and a large array of diphthongs. Significant parts of modern Kangshuic A languages also underwent a sound change in which alveolar sibilants were lateralized, with Court Kangshi one of the notable exceptions.
Morphosyntactically, Old Kangshi still preserves the PK case system, although especially oblique case information is often redundantly encoded by verbs even in early texts. Modern Kangshuic A generally does not, although the Ergative has become a topic marker in some of them. The verbal paradigm is generally completely restructured, since PK suffixation generally either fossilized as stem change or was lost entirely in favor of serial verb constructions. Court Kangshi is an example of both developments, indicating the categories of version and subjunctive by various forms of stem change, with principal mood marking occurring by a realis clitic. TAM, however, is handled by derivation or serial verbs. The loss of the case system and person marking on the verb generally also made word order less variable than it was in older stages of the language.
Kangshuic B and C are both more scattered and internally varied branches. The principal change affecting Kangshuic B was widespread vowel epenthesis, followed in many cases by loss of final vowels, creating a distinctively unkangshuic CVCVC(V) word pattern. They generally preserve cases, but have innovated a much more regular verbal TAM paradigm. Kangshuic C languages, being the farthest from the core Kangshuic area and in contact with both other Qoic as well as Zotic languages, make heavy use of phonation distinctions at the expense of tone, and preserve the front rounded vowels that result from fronting before alveolar finals. Morphologically, they are both more heavily head-marking and prone to univerbation than other branches of Kangshuic, including consistent verbal person marking, often a preservation of some verbal aspect-pairs and the nonfinite forms of PK, but also an innovation of new distinctions.
I'll have to go into more detail on each of these in separate posts.