#kshafa

[paizau]

inside me there are two wolves. one wants to create the most elaborate and overcrowded vowel system you've ever seen. the other thinks it's time to give a more conventional system a try.

[paizau]

Analogy and Sacrificing Complexity

I've been thinking about Kshafa nominal declension (the Evil) again, and came to the conclusion that in some parts it is just too complicated to be kept that way in the regular system. Maybe for select irregular words, but not in general. So I've done some reworking (the Curse).

The Problem

The main problem is that of the indefinite ergative, locative singular, and nominative plural. They are the yellow cells in this table showing the paradigm of liman (I wrote a detailed post on how this paradigm evolved here):

As you can probably see, they have vastly different stems: li(m)V(n)- vs zhaen-, basically only the final nasal is the same. Now I support stem alternations wholeheartedly, but idk. This just feels a bit two much. These aren't even that used forms. like yes the ergative is a pretty important case, but the indefinite ergative, specifically? it just feels too underused to survive this. And so! I propose a single step of sweeping analogical leveling!

The Solution

For this example I will use the dummy form *rikep. The state of affairs just before the analogy takes place is like this:

There are 4 stems in the paradigm atp:

*rikap- the bare stem is the form used when the noun is followed by modifiers, so it's very common

*ricap- the stem used for most definite forms, so also a very dominant stem

*rikip- a stem that appears in a fair number of cells, both definite and indefinite

*rʲàcép- the talked about problem child, a stem that appears in only 1/5 cells, all of them of the less used indefinite.

The indefinite paradigm is the most varied, even though it is less common, and so it gets completely remodeled. The stem *rikip- gets extended into all indefinite forms barring the nominative singular, which keeps its special stem because of how common it is. The singular dative analogizes because its stem is seen as "the definite stem", and the yellow cells analogize because they are just too different.

Now we have a nice simple system of only three stems! a definite stem *ricap-, an indefinite stem *rikip- (which appears in three definite forms), and a basic stem which is derived from the indefinite stem through a vowel change ❤️ yay ❤️

But things don't stop here. After this period of analogy, sound changes continue to happen, and the final paradigm in modern Kshafa is this:

A new forth stem rikhi- is created for the indefinite dative and plural locative. It's the same as the regular general stem with just the stem final consonant dropping because of the prohibition of clusters, and these forms have consonant initial suffixes.

For those wondering, the updated paradigm of liman is this:

It now has only five stem instead of six, with three of them being easily derived from a forth using simple morphophonological rules: limon- > liman, limo-, limu-.

[paizau]

Possible grammatical historical spelling in Kshafa

So I want Kshafa orthography to be pretty consistent. Like not entirely phonemic, but to be 90% of the way there. So for example I will still have silent <NW>'s where old /ŋ/ was later dropped, and I'll keep /ʂ, ʐ/ that originate from palatalized /r/ and /l/ distinct from each other and from original /ʂ/, but vowel length and diphthongs have got to go. I think. I don't have a fully settled system yet.

ANYWAYS. One thing that I do want to keep as a solely grammatical spelling is the distinction between original final *-u > -o and final -o that resulted from diphthong simplification of original *-au, so say the form thefo "plam tree (INDF SG NOM/LOC)" can be spelled two different ways - <TXePQaW> (= thefaw) for the nominative, and <TXePQoQ> (= thefo) for the locative.

Here in the native orthography:

[paizau]

Current palatalization situation of Kshafa stops:

C+j clusters evolved into palatalized consonants twice.

In the first palatalization the velar stops *k and *g palatalized and became true palatals /c/ and /ɟ/, while *j was simply dropped after other stops.

The second palatalization occured after high vowels were lost, creating new C+j clusters. This time the alveolar stops *t and *d palatalized into alveolo-palatal affricates that later simplify to fricatives /ɕ/ and /ʑ/.

The problem:

I want more alveolo-palatals.

Proposed solution:

have the velars palatalize in both cases.

The question:

How should I go about it?

Should I go with option 1:

1st palatalization: *kj *gj > c ɟ

2nd palatalization: *kj, *tj, *gj, *dj > ɕ, ʑ

Or 2:

1st palatalization: *kj, *gj > *tɕ *dʑ > ɕ ʑ

2nd palatalization: *kj, *tj, *gj, *dj > c, ɕ, ɟ, ʑ

currently I'm leaning towards option 1, because it feels more right, but option 2 has more merging and complication potential which I do enjoy.

[paizau]

It's really fun when an originally regular paradigm just goes nuts through sound change. For example the noun *mitkon (no meaning yet) in proto-Kshafa develops 2 very distinct stems:

mikven-

ngun-

(the singular absolutive in some consonant stem nouns is "irregularly" formed by ablauting the final stem vowel to /a/)

[paizau]

The word frá "full", its form, shape at time of orthographic standardisation, and its spelling in transliterated Akka-tą́.

Here are the orthographic forms:

[paizau]

That moment when a root is completely eroded through sound change and all that's left is the inflectional ending:

*rilo-ntad > *rilwonted > *riwonde > *rwōnde > *wōnde > /vo.ⁿde/ - /∅/ + /vo.ⁿde/

compare

*kato-ntad > *ketwonted > *kʰetʰwonde > /kʰe.tʰvo.ⁿde/ - /kʰetʰ-/ + /vo.ⁿde/

[paizau]

verbs is hard :(