Subumpamese languages

The Subumpamese languages are the languages spoken in the eleven states of Subumpam. They split off from the parent language, called Tapilula, around 600 AD and continued to be spoken until the defeat of Subumpam in the Vegetable War of 2668 AD.

NEW IDEA. THERE WAS NO WESTERN SUBUMPAMESE AT ALL; THIS WAS ACTUALLY PART OF ANDANESE TERRITORY. THIS MEANS, HOWVVEER, THAT THERE IS NO SPREACHBUMND THERE. BUT THAT IS IN KEEPING WITH THE IDEA THAT THE "KALPTA" TERRITORY WAS IN FACT ANDANESE.

Early history

Update notes

NOTE, THIS WILL BE REWORKED HEAVILY SOON, TO THE POINT OF STARTEING FROM SCREATCH. MOST NOTES ARE ON PAPER ONLY NOW.

Proto-Subumpamese will be a "Raspberry Wine" type of language, with heavy reliance on coarticulated consonants, both labial and palatal. It will split early on into five branches: Kava, Vuʒi, West, Central, and East. These five will be mutually intelligible for the most part, and most sound changes will be shared among the different groups.

Fortition

If Proto-Sub loses tones, it could fortify fricatives after high tones as happened in many other languages (though mostly at a later date). Also, syllabic nasals probably all fortify to /un/ ... that is, a three-way merger of ṁ ṅ ŋ̇ as /un/. This works because unlike Tapilula, syllable-final /n/ occurs commonly with other vowels. the actual shift would probably be ṅ > ən > un.

Early changes common to all Subumpamese languages

Proto-Subumpamese split off from the Gold branch of the Tapilula family around the year 600 AD. At this time, the only other Gold speakers were those living in Paba; those who settled Nama spoke different languages. Proto-Subumpamese had already gone through most of the vowel changes that characterized the Gold language of the year 1900. Note that the vowel changes above are responsible for the growth of closed syllables, and that proto-Subumpamese therefore has closed syllables wherever Gold also has them.

However, in Subumpamese the vowel qualities [e] and [o] (on all tones) did not merge to [a] as they did in Gold. There are a few other conditional differences between the two branches of the family, but these are dependent on consonants and occur only on the Subumpamese side. THere may also be /oè/ > /wè/ > /ʷè/, etc.

Subumpamese branched off from Gold before the deletion of all word-initial vowels, and therefore there are some words that were one syllable longer in proto-Subumpamese than in Gold. This also means that proto-Subumpamese retained the Tapilula noun classifier system, which in the Gold branch was wrecked by the deletion of initial vowels.

The only consonant change that occurred between Tapilula and proto-Subumpamese was the palatalization of /k/ (not /ḳ/) before /i/. All of the other changes that define the two branches occurred after the split.

Sound changes common to all Subumpamese languages (part II)

1. First stage of vowel shifts is on paper only and mostly resembles Gold except for the lack of loss of /e o/. Subumpamese contracted its vowel inventory to just /a i u ə/, as in Gold, but did so in very different ways. Whereas Gold did an unconditional shift of [a e o] > [a] regardless of tone, length, stress, or environment, in Subumpamese the vowels first left effects on surrounding consonants and then coalesced according to conditional rules, which differed in each branch.
2. All [u] stained consonants it touched by making them labialized. This happened before even the original vowel shift, so that e.g. /ū/ does not labialize a following consonant since it comes from various squences such as /ùa ùe ùə/ but never */ùu/. The gap of */ūʷ/ was repaired when /ə̄/ later became /ū/.
3. Likewise, all [i] infected consonants with palatalization. Furthermore, all [e] caused velar cosnoants to become postpalatal ( or "prevelar"), a change unique to this one series. (c.f. Japanese; velars move more than others since they are the same PoA as vowels already).
4. Syllable-final /h/ > /x/, forking off a new phoneme. However, both /h/ and /x/ were affected in identical ways by the vowel stains.

NOTE ON POLITICS: AT THIS POINT, THE SUBUMPAMESE COMMON STOCK SPLITS INTO FIVE: KAVA, VUDŽI, WEST, CENTRAL, AND EAST. THE REST OF THE CHANGES ON THIS LIST ARE FOR KAVA ONLY.

Sound changes unique to proto-Kava

This section is presented separately because it is needed by the spreadsheet and may be useful for loanwords. It ends at 1700 AD, a time of no political significance for Kava but near the time of greater significance for the areas around Kava.

1. Three-way merger of ṁ ṅ ŋ̇ as /ən/. (later /un/)
2. ḳ > q.
3. kq qk > qq.
4. xhʷ > xʷ.
5. śʷ s̀ʷ > sʷ . These sounds when a primordial /h/ sound was both palatalized and labialized. Although there was no plain counterpart, this /sʷ/ sound still retained its labialization.
6. ə > u.
7. 'e o > i ə. (thus phonemicizing palatalized consonants) This produces the exact same four-vowel system found in Gold: /a i u ə/.
8. [i] > [ə] if touching a Q in eihte direction.
9. Primordial f > p (spelled /ṗ/ in wordlists).
10. Primordial hʷ w > f v.

Proto-Kava evolves a post-palatal stop and fricative, and also a postvelar (but not uvular) stop. Thus it could be said that there are three "k" sounds: front, middle, and back. However, all Subumpamese daughter languages shift the consonant inventory strongly towards the front, resulting in a family of languages that sound somewhat like watered-down Pabappa.

• All of this happened while there sitll was not an /s/ in the language, which was one of the triggers for the very strong frontward shift. At one point, early proto-Subumapmese had a voiceless fricative inventory consisting of /fʷ ś s̀ x h/ but no [s], even allophonically. All five of these had originated simpyl as allophones of [h]. There was also a corresponding voiced series.

Proto-Subumpamese can be stated to have been spoken around 1700 AD, not 1200 AD as its short sound c hange list might imply.

Therefore the consonant phonology of proto-Kava (~1700 AD?) was:

Bilabials:              p  b  m  
Rounded bilabials:      pʷ bʷ mʷ fʷ w
Alveolars:              t  d  n  l 
Rounded alveolars:      tʷ dʷ nʷ lʷ
Palatalized alveolars:  č  ǯ  ň  ł
Palatals:               ć     ń  ś  ź  y
Prevelars:              c̀     ŋ̀  s̀  z̀
Velars:                 k     ŋ  x  g
Labialized velars:      kʷ    ŋʷ 
Postvelars:             q        h
Labialized postvelars:  qʷ
  

Note: /ň/ = /ñ/; the latter's used because the former isnt on all keyboards.

There still was not an [s], even as an allophone.

Changes common to Western, Central, and eastern Subumpamese languages

After the loss of Vuʒi, the remaining Subumpamese peoples continued to speak a single common language. The ejective series was preserved, and is here considered to be postvelar because it does not obey the sound changes that involve velars. /l/ is preserved in this stage; it disappeared only in the PES branch which was influenced by Pabappa.

1. three-way merger of ṁ ṅ ŋ̇ as /ən/. (later /in/)
2. ḳ > q.
3. kq qk > qq.
4. xhʷ > xʷ. Probably also śʷ s̀ʷ > xʷ as well. These latter two examples occurred when a primordial /h/ sound was both palatalized and labialized.
5. gʷ > w.
6. The high central vowel ə changed to i unconditionally.
7. hʷ > f, thus merging with the /f/ that is in the Kava dialect spelled /ṗ/.
8. Syllable-final ŋ changed to match the place of a following consonant, and changed to n if word-final. This probably also affects ǹ and ń.

NOTE ON POLITICS: THIS IS WHERE THE THREE DIALECTS SPLIT UP INTO WEST VS CENTRAL VS EAST.

At this point the consonant inventory was

Labials:                p  b  m  f
Rounded bilabials:      pʷ bʷ mʷ       w
Alveolars:              t  d  n        l  c  ʒ
Palatalized alveolars:  č  ǯ  ň        ł  
Palatals:               ć     ń  ś  ź  y
Prevelars:              c̀     ǹ  s̀  z̀
Velars:                 k     ŋ  x  g
Labialized velars:      kʷ    ŋʷ xʷ
Postvelars:             q        h
Labialized postvelars:  qʷ

All labialized consonants could be considered to be clusters of consonant + /w/ now, however, unlike earlier. This is because they could no longer occur in syllable finally position except before another labialiezd consonant. By contrast, the palatalized consonants now only existed for PoA's in which the primary PoA was contiguous with the palate; that is, they were all descriptible as some sort of "prepalatal" or "postpalatal" etc.

/ǯ/ and /ł/ might be resticted tyo only occurring before [i].

Changes common to Western Subumpamese

These languages form a sprachbund with the West Andanic languages. Labial consonants are less common than in Central and Eastern Subumpamese, but the general sound of the language is otherwise similar, with the same three tones, and about the same ratio of open to closed syllables.

Changes unique to Central Subumpamese

1. the Palatalized alveolar consonants č ǯ ň ł, which had originated from alveolars bordering /i/, are turned back into plain alveolars c ʒ n l.
2. The labialized coronals tʷ dʷ čʷ ǯʷ nʷ became the velars kʷ kʷ kʷ kʷ ŋʷ.
3. The palatals ć ʒ́ ś ź are shifted forward to č ǯ š ž.^???
4. The true velars k ŋ x g are fronted to ć ń ś ź when before [e] or [i], including any [i] that had been recently created from the schwa.

NOTE ON POLITICS: THIS IS ABOUT 1700-1900 AD. CONTEMPORANEOUS WITH OLD ANDANESE, HOWEVER, THIS BRANCH IS THE ONE WITH THE LEAST INFLUENCE FROM ANDANESE.

Changes common to Eastern Subumpamese

No more than about 20 changes in any language are possible, and it should be more like 15. That is the total for the changes on this list and the individual language lists below. This branch is influenced heavily by Pabappa. It is spoken in Paba too, and the entire branch could perhaps be better named "Lenian languages" even though the much larger number of Pabaps who spoke Pabappa also considered themselves Lenians.

At the time of breakup the consonant inventory was

Labials:                p  b  m  f
Rounded bilabials:      pʷ bʷ mʷ       w
Alveolars:              t  d  n        l
Labialized alveolars:   tʷ dʷ nʷ       lʷ   
Palatalized alveolars:  č  ǯ  ň        ł  
Palatals:               ć     ń  ś  ź  y
Prevelars:              c̀     ǹ  s̀  z̀
Velars:                 k     ŋ  x  g
Labialized velars:      kʷ    ŋʷ xʷ
Postvelars:             q        h
Labialized postvelars:  qʷ

1. Delabialization of all non-dorsal sounds occurred in a few steps. w > v. f > f. (That is, the sound changed from the bilabial of common Subumpamese to a labiodental, but the speakers retained the same symbol.)

   pʷ bʷ mʷ > p b m. (Possibly /ə/ > /o/ when facing a labialized consonant before this shift.)

   tʷ dʷ nʷ lʷ > tl ġʷ ŋʷ ll. (Consider making all four of these into labiovelars.)

2. Labialized palatals and prevelars become labiovelars, the only remaining labialized set in the inventory.
3. ź z̀ > j (spelled /i/). ai əi (on any tone) became ē (perhaps not always long).
4. Palatals shifted doubly forward:

   č ǯ ň ł > c ʒ n l and

   ć ń ś > c n s (no intermediate stage as postalveolars).

5. Velars (but not labiovelars) shifted doubly forward:

   c̀ ǹ s̀ z̀ > č ň š ž and

   k ŋ x g > č ň š ž. (Possibly velars remain in some positions, as in early Proto-Indo-European. This would best be explained as labialization.)

6. The uvular stop q shifted to k. /h/ became /x/ in most positions, but the spelling remained.
7. In syllable-final position, f c shifted to p t. (Thus /k/>/t č/, /h/>/s š/, even though the shifts were not related.)
8. Probably ŋʷ gʷ ġʷ > ŋ g g, but /kʷ/ remains untouched. This shift might need to be earlier on. In the dictionary, these sounds will be written without labialization regardless of their actual pronounciation, since there is no conflict. The shift of ġʷ > g went through an intermediate, either /gʷ/ or /ġ/, but in either case merged unconditionally with the reflex of the original fricative.

These changes left the language with a consonant inventory of at least

Labials:                p  b  m  f  v  w
Alveolars:              t  d  n  s  l     c  ʒ  
Palatalized alveolars:  č  ǯ  ň  š  ž        
Velars:                 k  ġ  ŋ  h  g 
Labiovelars:            kʷ

The PES speakers were in intimate contact with speakers of Old Andanese, whose consonant inventory was /p m w f t n l k g h q kʷ/, and with the Pabappa dialect of the Gold language, whose inventory was /p b m w t d n s z l č ǯ j k ġ ŋ h g ḳ ʕ/. Both of these analyses treat labialization as nondistinctive except for Andanese /kʷ/.

This is ~1900 AD or a little bit afterwards. The last few changes likely spread areally through the PES languages, however, so they can be treated as canonical for the sake of comparing words.

PES soon redeveloped phonemic /q/, due to Andanese influence, interpreting it as an allophone of /h/ after a high tone. This is the same process that occurred in early Thaoa. Likewise, the frequent Gold diphthongs ai əi were interpreted as a new vowel, /e/, while au əu were interpreted as /o/. These diphthongs occurred only on the long tone in Gold, since Gold did not contrast tone in diphthongs, but Old Andanese provided other tones for those vowels. Lastly, Old Andanese /i/ was pronounced [ə] in the vicinity of [q], and this practice was also borrowed (as it was in Thaoa). Thus for example, Old Andanese kòhi "flag, sign" was loaned as kòqy.

Contact with Paleo-Pabappa

Paleo-Pabappa is a hypothetical language spoken in Paba which was replaced by the Gold language sometime between 1103 AD and 2057 AD, on the assumption that Gold was not native to Paba after all, but was actually introduced there from AlphaLeap.

Paleo-Pabappa branched off from Thaoa around 1085, possibly slightly earlier. Thus, it is not descended from proto-Macro-Subumpamese. However, "proto-Thaoa" can be derived from proto-macro-Subumpamese by reversing the labialization and palatalization shifts, which were unconditional, and by merging all [e] and [o] into /a/. This, coupled with grammatical similarities and various areal shifts, helped establish proto-Pabappa in the minds of its speakers as part of the Subumpamese family, just as was the case with the supposed West Subumpamese branch that is actually derived from Old Andanese.

Later developments

Proto-Kava (1700 AD?) to Kava

The consonant phonology of proto-Kava (~1700 AD?) was:

Bilabials:              p  b  m  
Rounded bilabials:      pʷ bʷ mʷ f  w
Labiodentals:           ṗ
Alveolars:              t  d  n  l 
Rounded alveolars:      tʷ dʷ nʷ lʷ
Palatalized alveolars:  č  ǯ  ň  ł
Palatals:               ć     ń  ś  ź  y
Prevelars:              c̀     ŋ̀  s̀  z̀
Velars:                 k     ŋ  x  g
Labialized velars:      kʷ    ŋʷ 
Postvelars:             q        h
Labialized postvelars:  qʷ

Syllable-final consonants were

q qʷ 
k kʷ c̀ ć 
n nʷ ň 
l lʷ ł
x fʷ s̀ ś 
g    z̀ ź

Although these groups were highly restricted in distribution.

The vowel inventory was /a i u ə/, but unlike Gold, the schwa was a full vowel and could be long.

1. The postalveolar affricates č ǯ became c ʒ unconditionally.
2. The palatals ć ń ś ź became č ň š ž except in syllable-final position.
3. The voiceless uvular stop q changed to k when syllable-final.
4. The happy couple c̀ ć got married and changed their name to ć. Thus the four-way contrast of ć/c̀/k/q in syllable-final position had now been resolved as ć/ć/k/k.
5. Syllable-final stops s̀ ś locked arms and kissed their way across the syllable boundary to become aspiration sanitizers, losing their palatalificitosity as they did so. Any voiced consonant they stepped on turned voiceless, which if nasal meant a resurgence of sounds like /,mp/.
6. The alveolar lateral lʷ changed to w unconditionally.
7. Syllable-final ć switchivizes itself to become the same PoA as a following consonant, except that it hugged a velar or postvelar stop and brought them up to its level. (Possibly also to /t/ in standalone position.)
8. Labialization hydrosporoplasmolysticalifactionalistificaticized across a syllabe broundary, spooting out a new series of bilabializated consos, namely all of them: čʷ ǯʷ ňʷ łʷ ćʷ ńʷ śʷ źʷ c̀ʷ ŋ̀ʷ s̀ʷ z̀ʷ xʷ gʷ hʷ. OBVIOUSLY though, things like łʷ didnt really exist.
9. Syllable-final fricatives z̀ ź painfully stung the next consonant, spreading a toxic venom as they tunneled all the way through, but not turning it into a voiced consonant.
10. Any stop/aff after a stop turned into a fricative. That is, after /k kʷ ć/, the stops p b t d č ǯ ć k q became ṗ v s z š ž ś x h.
11. The affricates c ʒ changed to s z when not after a high tone.
12. Voiced stops became voiceless when occurring before a high tone.
13. The postalveolar consonants č ǯ ň š ž ł became c ʒ n s z l unconditionally. (They actually remained before /i/, but this is now allophonic and can be ignored.)
14. The labialized velar nasal ŋʷ changed to w before /i/.
15. Labialization disappeared before /i/. (This probably kills the allophony above.)
16. All remaining labialized consonants became rounded bilabials.
17. Sporadically, the initial consonants of nouns changed to rounded bilabials due to analogy with their plural forms, which began with the prefix /u-/ which in the past had caused labialization. That is, for example, an alternation of t-/up- would change to an alternation of p-/up-.
18. The true palatals ć ń ś ź became č ň š ž unconditionally.
19. Labialization was removed.

The consonant inventory at this time was

Labials:                p  b  m  f  v  w
Alveolars:              t  d  n  s  z  l  c  ʒ  
Palatalized alveolars:  č  ǯ  ň  š  ž          
Palatals:               ć     ń  ś  ź  y
Velars:                 k     ŋ  x  g
Postvelars:             q        h

Syllable-final consonants were

k ć 
n nʷ ň 
l lʷ ł
x fʷ s̀ ś 
g    z̀ ź

The voiced labial fricative /v/ comes from /gʷ/. The prevelars above are given here as palatals for simplicity, but there was no explicit shift of them.

Sound changes affecting the core Subumpamese languages

Unassigned Subumpamese language #1

This probably will go to Central.

1. Syllable-final ś ź s̀ z̀ changed to j (spelled /i/), or lengthened the preceding vowel. Possibly źz̀>j unconditionally.
2. The true palatals ć ʒ́ ś ź ń migrate forwards into č ǯ š ž ň.
3. The prevelars c̀ g̀ s̀ z̀ migrate forwards to become ć ʒ́ ś ź.
4. All sibiliants change to frics except after a high tone. c ʒ č ǯ ć ʒ́ > s z š ž ś ź. Thus most /s/ borders [i].
5. Next, the voiced stop d changed to /r/, except after a nasal, a high tones, or word-initially.
6. Labialization bleeds throughj clusters. e.g. kʷm > kʷmʷ.
7. Then, voiceless stops and fricatives became voiced after a low tone or a long falling vowel.
8. Labialized consos in syllable final posiosion become bilabials. Thus pʷ bʷ mʷ fʷ w > p b m f w; kʷ ŋʷ > p m. These were not rounded.
9. Palatalized consos in syllable final posiosion become mostly coronoals. Thus č ǯ ň ł > c ʒ N l; ć ń ś ź y > c N s z y; c̀ ŋ̀ s̀ z̀ > t N s z. But NOTE that 1) the N assimilates to all fdollowing consos. 2) most of this /s/ is still actually [h] and thus the shift occurs only in transparent compounds.
10. Palatalization also blleds though. This is sort of a compensatory shift to make up for the last one. since /ś/ is really [h], it metathesizes and therefore it is the other consonant that gets palatalized.

    śm śn śŋ > (i)mps (i)nc (i)ńć. źm źn źŋ > MAYBE (i)mbz (i)nʒ (i)ńź (there is still no [ġ]).

    Word-initially, ń > z. Possibly also mʲ > (i)z if it existed.

11. s̀m s̀n s̀ŋ > (ə)mp (ə)nt (ə)ŋk. c̀m c̀n c̀ŋ > (ə)mp (ə)nt (ə)ŋc̀.

Separation of proto-Vuʒi

After these first few changes, the Vuʒi language split off from the rest. This language marked the westernmost limit of the Zenith people during the entire period of Subumpam's existence. At the time of the split, Vuʒi did not have an /s/.

Paleo-Pabappa

This sections is kept separate since it belongs to a different subfamily that got adopted and became more like Subumpamese. It may consist of just a single language, analogous to Russian, or of many, as in the "Seven Sisters" setup.

Starting pohonology

Tapilula's starting phonology was

CONSONANTS
/p b m f t d n l tʷ dʷ nʷ j k ḳ ŋ h g hʷ gʷ/

VOWELS
/a e i o u ə/

On high or low tone. Syllabic consonants were

m n ŋ

And the language was otherwise entirely CV. However, Paleo-Pabappa underwent the same contractions that Thaoa to its east and the other Subumpamese languages to the west did, and therefore Paleo-Pabappa, even by 1085 AD, was already incompatible with its protolanguage and although the people were literate almost nobody could read the old texts.

The early sound changes common to Thaoa, Paleo-Pabappa, and proto-Macro-Subumpamese did not affect the consonant inventory at all; in fact there was only one change, /k/ > /ć/ before [i], which could be considered mere allophony despite that it could delete a vowel.

Try for an absolute maximum of 20 sound changes to reach 1900 AD, the last date at which proto-Pabappa could still be the majority language of Paba. These 20 sound changes will include those shared with PMS and Thaoa. Even this rate of change is high for the family.

Limit Thaoa influence, since the speakers of Paleo-Pabappa are likely to be seen as "hyper-Pabaps", and be even more hostile to Thaoa than the upper class which was mostly safe from slavery and was the quickest to switch to speaking Gold. It might borrow the trait of /tʷ dʷ nʷ/ > /tl dl nl/ from Gold, but since PES and Litila no longer have this it's more likely that this feature is simply not a part of the sound changes of any non-Gold language.

Sound changes from Tapilula to Paleo-Pabappa

Note that the sound changes preceded by a (d) can be ignored since they are all neutralized by a later change devoicing all stops and fricatives. However, they are still important for loanwords.

1. On-paper sound changes in which vowels are metathesized and then the resulting diphthongs are mostly or entirely monophthongized. (PP will likely be more conservative than those to the east and west, and focus mostly on consonants.)
2. All consonants touching a primordial [u] become labialized. (May need to rethink this, because /u/ rarely merges with /o/ and this change would not release anything.)
3. (d) After long vowels, all consonants became voiced. Also, consonants occurring after initial vowels also became voiced.
4. In absolute initial position, t >s. (Whether tʷ > sʷ or not is moot, except for loanwords.)
5. (d) Voiced stops and fricatives become voiceless in initial position.
6. Velar consonants move up: k ŋ h g > č ñ š ž, probably unconditionally.
7. q>k.
8. In absolute final position, š ž ñ > s z n (?)
9. In absolute final position, hʷ č> p t.
10. The labial fricative f became p word-initially, and v became b everywhere. (/v/ was rare; it is not /w/)
11. Frics became stops after a high tone.
12. All labialized consonants change to plain bilabials.
13. Voicing distinction disappears entirely. This was actually triggered by a new voicing of stops after low tones, but because this change removed the last remaining environment that could host a minimal pair, there was no longer any phonemic contast.

Possibly make /g/ >>> /s/, and /h/ >>> /š/, by drawing /g/ into palatalization first and only later doing it with /h/.

Later history of Paleo-Pabappa

Paleo-Pabappa did not evolve into Pabappa. Instead, Paba's people shifted to speaking the Gold language as it was introduced from AlphaLeap, and quickly developed a distinctive dialect of it that ultimately came to be called Pabappa. However, Paleo-Pabappa still survived inside Subumpam, since the two nations of northeastern Subumpam had joined the Subumpamese Union after seceding from Paba. The other Subumpamese people considered Paleo-Pabappa to be just another of the many Subumpamese languages, as it shared similarities with neighboring languages such as Galà and proto-Eastern Subumpamese.

Paleo-Pabappa split into three languages: one for each of the two Pabap nations in Subumpam, and one for people in Paba who had not yet shifted to speaking Gold. However, Subumpam was soon wiped out in a catastrophic war, and Paba's strong national government drove the remnant speakers of Paleo-Pabappa into learning what was to eventually become Pabappa.

Grammar

The grammar of Proto-Subumpamese resembled that of Old Andanese, even though Subumpamese was more closely related to Gold than to Andanese.

Proto-Subumpamese early on shifted from CV classifier prefixes into CVC, but with the second C always the same as the first whenever the original morpheme had been a pure CV. In fact, these reduplicated classifiers were more common than the innovated type with two differing consonants. These consonants interacted with the initial consonant of the stem of the word, so that for example the classifier huh- attached to the word stem mìna produced humpìna "newt". TENTATIVE! the development of CV > CVC is solid, though.

Nouns

Subumpamese nouns have a true noun class system, not a gender system like that of the Gold language, and it is very similar to that of Andanese. Subumpam is a fairly diverse empire. The climate ranges from subtropical and nearly tropical in the south to the cold and rugged mountains of the north, whose people are much poorer than those of the tropics but also much better protected from foreign invasions. In the mountains, most people speak Andanic languages, a family which is related to Subumpamese but much more conservative.

The richer natural environment of the south has led its people to prosper and bring cultural innovations into the north, as well as a more diverse cuisine flavored with tropical fruits such as pineapples and coconuts as well as large, deep-water fish such as tuna.

The peoples of the north borrowed the words for these concepts from the various Subumpamese peoples of the warmer climates, and in most cases, the words were borrowed with their classifier prefixes intact. Thus the already large classifier prefix inventory of the various Andanic languages became even further enlarged as new noun classes were created to mirror the noun classes of the various Subumpamese languages providing the loanwords. In many cases, the noun classes were already homophonous with an Andanic noun class of the same or similar meaning, simply due to the conservatism of both language families and their distant shared ancestry. However, in most cases the Subumpamese languages have diverged much further from the parent language.

For example, most Andanic languages distinguish between the noun class pe- for crustaceans, he- for most other sea life, and pi- for water and weather itself. But the Subumpamese language of Kava has merged all three of these into a single noun class pi- "water". Thus the Kava word pipùna "starfish" was loaned into the Andanic languages as part of the noun class for water and weather, rather than being incorporated into the noun class for most other sea life.

In other cases, a Subumpamese loanword would be borrowed into a noun class with an entirely different semantic meaning, or into a class that did not previously exist in the Andanic languages. For example, the names of many fruits in the eastern Subumpamese language of Pačēpus began with ŋu-. Most Andanese speakers realized that this noun class was cognate to their own nu-, which contained words for fruits and also words for buildings. This is because the parent language's /ŋ/ phoneme had changed unconditionally to /n/ in the early common Andanic language. But a new /ŋ/ had arisen from the sequence /nw/, and thus Andanic languages were able to loan foreign words with /ŋ/ without replacing the phoneme. And so new words for fruits introduced by the Subumpamese people were borrowed with the ŋu- prefix intact, and ŋu- came to specifically denote tropical fruits, even those introduced from other tropical nations.

Some borrowed prefixes were extended to native words. For thousands of years, Andanic and Subumpamese languages had been actively coining new words by shifting existing words from one noun class to another. For example, the words for body parts could also serve as words for edible animal parts, by changing the prefix from one of the human classes to i- "meat". The same root, hìqi, meant "arrow" with the tu- prefix and "key" with the yo- prefix.

But now some long-established Andanic words were given new prefixes of Subumpamese origin to denote that the objects they described were of foreign origin. An arrow produced in Andanic territory would be called tuhìqi as it had always been, but if produced in Subumpam or by Subumpamese people in Andanic territory it would be called kʷuhìqi, borrowing the Subumpamese classifier, but preserving the native root for the noun itself.

Rarely, words moved in the opposite direction. The Subumpamese people generally did not need heavy clothes, as their climate was fairly warm even in winter. Heavy clothes and long pants were most often used to protect their wearers from thorns and thistles, not from winter cold. Thus, some Subumpamese languages borrowed the Andanic classifier prefix ho- to denote heavy clothes or clothes made in the Andanic style. (The native Subumpamese cognate of this prefix varied from one Subumpamese language to another.)

Interaction with other languages

Proto-Eastern-Subumpamese also bordered the Andanic nation of Galà, leading to vocabulary exchange between PES and Old Andanese around 1900 AD, and between Puripup and Galà later on. Even though Subumpamese is more closely related to Pabappa than to any Andanic languages, the grammar of most Subumpamese languages is much more similar to the Andanic languages than to Pabappa, easing the exchange of vocabulary.

The Galà language was fairly similar to Proto-Eastern-Subumpamese around 1900 AD when they first met up; Galà was much more guttural, and also had a much stronger preference for open syllables than PES.

At least one Subumpamese language also survives to develop into Meromo.