#nominal morphology

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:

"you're probably wondering how I got here"

Kshafa nominal declension can get pretty complicated, especially for consonant final roots. In the history of the language there were various vowel and consonant shifts which cooked up some pretty obscure declention patterns. It's not that uncommon for two stems of the same root to seem suppletive, even though they arose from the same form. Case in point - liman "no meaning yet lol":

The different stems are color coded, and here they are listed out:

liman

limoN-

limo-

limu-

lio-

zhaen-

There are three major steps that got us here.

1. The Proto-Vowel System Has Its Guts Rearranged

A major sound change that happens pretty early on is the rearrangement of the vowel system. Through a series of vowel raising breaking and lowering, the proto language's simple classic five *a *i *u *e *o becomes a short-lived four vowel system *i *u *ə *a which stablizes after some more shifting and merging into a system with no-central vowels and brand new long vowels and diphthongs:

Now back to our word, after this great vowel shift (:O) the basic single root stem broke into 4 stems:

*liman-, the suffixless form

*limun-, with a CV suffix, where V is a non-low vowel

*limwon-, with a CA suffix, where A is a low vowel.

*lyamwen-, with a V initial suffix.

2. The Further Breakdown of *limun-

*limun-, unlike the other stems, further split into three sub-stems, depending on the initial consonant of the following suffix. These consonants were: *w, *s, and a stop.

*limuŋ-wV, the final *n velarizing

*limū-zV, the final *n voicing the following *s, dropping and lenthening the vowel.

*limun-TV, before the stops it simply assimilated to their point of articulation

3. all other changes

There are a few other sound changes, like the lowering of short *o to /u/, the delabialization and then dropping of *ŋw > ŋ > ∅, the frication of palatalized liquids, and the loss of vowel length which gives us the final six stems in all their glory:

*limon > liman > /li.man/

*limonwub > *limuŋwu > /li.mo.o/

*limonke > limunɟe > /li.mu.ᶮɟe/

*limonsod > *limūza > /li.mu.za/

*limonso > *liŋwōzwe > /li.o.ve/

*limoni > *lʲaŋwenai > /ʐa.e.ne/

The method I've been operating under for nominal sub-declensions for the past while was to just figure them out as they come up, as to not fry my brain. unfortunately I have come to this conclusion only after my brain had already been fried. oh well you live and learn.

The sub-declension I worked on today is named "2Ac", meaning 2nd declension nouns (2a - final non-high root vowel (*-e, o, a), that contracts (c). Here is an example paradigm of a regular 2Au noun (2nd declension with a final non-high round vowel, i.e. non-contracting *-o stems):

As opposed to a 2Acu (contracting *-o stem) noun:

The first and most noticeable thing is the stem patterning. Regular 2Au/i nouns can have up to 4 stems, who differ in the stem vowel and either undergo palatalization or have a -v- added, exemplified in hakve:

hakv-

hikv-

hak-

hik- (the indf.acc is a special case)

As opposed to 2Ac nouns which only have 2 stems, like 1st conj nouns, and they are the same, except the second has an additional final -o-, exemplified in krí:

kr-

kro-

The second has to do with the theme vowel. In some forms, there is a theme vowel between the root and ending. For singular 2nd declension nouns this theme vowel is -e-, but for 2Ac nouns, this theme vowel is instead -i-:

2Au (INDF.DAT.SG): hakveche, {hakv}-e-che.

2Au (INDF.DAT.SG): kriche: {kr}-i-che.

The third and final difference is in 4 suffixes - the NOM.PL and LOC.SG, both indefinite and definite. For regular 2Au and 2Ai nouns they are -a(n) in the indefinite and -u(n) or -i(n) respectively for the definite. For 2Ac the suffixes are instead -o(n)* in the indefinite and -a(n) in the definite:

hik●a, hik●an, hík●u, hík●un

kr●ó, kr●ón, kr●â, kr●ân

*There is actually another sub-sub-pattern here, where in the indefinite 2Acu have -o(n) but 2Aci have -i(n) like their non-contracting counterparts.

before: sad, too grid-like, a single lonely case of syncretism (sg. erg. and acc.)

As opposed to after: sexy, beautiful, two more columns added, no shortage of syncretized cells - some across the entire system, some only in certain patterns, full! tripartite marking for definite plural nouns!!

Kshafa has a mid-sized system of five cases - nominativve, accusative, dative, locative, and genetive, with a singular and plural destinction in all cases other than the genetive, which originally was just the unmarked stem.

There are 3 conjugation classes:

The first conjugation, or the I/O-Stems, have a nominative in -e and an oblique stem ending in either a final -i or -o.

The second conjugation, or the E-Stems, have a nominative in -e and an oblique stem ending in an -e

The third conjugation, or the consonant-stems, have eight simple sub-classes, for each broad consonant type.

When making this system I decided that I don't really want to have tonnes of sub-paradigms for every possible word final two syllable sequence, and also one of the major sound change ispirations for this conlang for me is ancient to modern Greek, so I employed some tricks to stir the system into a shape I wanted with the sound changes I have already come up with. There are overall seven changes, presented in their order of chronological occurence. Three in this post, and four in one I will write in the future.

1. Irregular voicing assimilation

This, as opposed to all the other changes is something I did to complicate the system a bit. The first sound change that proto-Kshafa undergoes is voicing assimilation in stop clustes, where the first stop in a cluster assimilates to the voicing of the second:

*gta > kate /ká.té/ (gen.)

However, in order to make the declentions of stems with final voiced and unvoiced stops more distinct, the *k- and *d- of the dative and locative suffixes instead merge into the voicing of the preceding stem final consonant:

*lok- > vèkhó /vèkʰó/ (loc.sg), lokán /lòkán/ (dat.sg), lotin /lòtìn/ (dat.pl)

*log- > vègò /vègò/ (loc.sg), logán /lògán/ (dat.sg), lodìn /lòdìn/ (dat.pl)

Instead of this slightly more boring paradigm where they merge:

*lok- > vèkhó /vèkʰó/ (loc.sg), lokán /lòkán/ (dat.sg), lodìn /lòdìn/ (dat.pl)

*log- > vègò /vègò/ (loc.sg), lokán /lòkán/ (dat.sg), lodìn /lòdìn/ (dat.pl)

(There is also an element of analogy of the tone patterns, with the dative having a high tone, and the locative having a low tone, even though the voicing assimilation happened before tonogenesis and so the expected result would be for the voiced suffixes to have a low tone and vice versa, but that's not the case)

2. Irregular ordering of sound change, or maybe analogy?

This has to do with a monophthongization rule where *iji and *ijə sequences monophthongized to a long *iː.

Normally this rule applies from the start of the word onwards (because that's how lexurgy does it), so *ijijə becomes *iːjə. However in order to prevent a declention subclass of *-ij- final nouns in the acc.pl I decided that in this case, the ending will monophthongize in analogy to other nouns, so -ijijən > *-ijiːn.

*kmiy- > káme /ká.mè/ (gen.), kéme /kémè/ (nom.sg), kámìn /ká.min/ (acc.pl) instead of the expected *kamían /ká.mí.àn/

Hooray! crisis averted.

3. Another case of analogy

One of the biggest sound changes in the history of Kshafa is its own great vowel shift "Vowel Breaking", where vowels raised, lowered, and broke in various ways to turn the classic 5 vowel system - *a, *i, *u, *e, *o into a four vowel system - *i, *u, *ə, *a. One its components was the high vowels *i and *u breaking to *jə and *wə if an *a followed in the next syllable.

The dative suffix in the proto language was *-ke, which became *-kja in the vowel breaking, and caused a preceding stem final *-i or *-u to break. I didn't want this to happen, because this would have added another stem to the paradigm which only appears in the dative singular. This has the opportunity to be fun, but it doesn't fit the vibe I'm going for here. So this sound change in this instence was undone by analogizing the stem into its unbroken form from a different cell in the paradigm: