Hipatal

Hipatal is the name of the ocean covering the majority of the planet Teppala, and containing few islands. Those islands that do exist are volcanic, and have a range of diverse populations. Most inhabitants are descendants of the Mumba people who migrated eastward from Laba to Fojy, a journey of more than 10000 miles, and settled various tropical islands along the way. They are thus blonde and blue-eyed, like other Lenians, despite living near the Equator. They are the ancestors of the Dreamers, and all Dreamlandic languages are in fact an early-branching variety of HP-2.

Most settlements are in the tropics; there is no significant monsoon, so rainfall is concentrated along the Equator, and even at 10°N and °S the weather is dry for most of the year and forest growth is impossible.

There are no sprachbunds; each language grows and develops independently.

The Proto-Hipatal language is nearly identical with Tapilula and can be considered a dialect of Tapilula. Its original consonant inventory was

Rounded bilabials:     pʷ  bʷ  mʷ  mbʷ mpʷ     w
Spread bilabials:      p   b   m   mb  mp     (Ø)         
Alveolars:             t       n   nd          l
Rounded alveolars:     tʷ      nʷ  ndʷ     
Velars:                k   ḳ   ŋ   ŋġ      h   g

Here, the consonants /b bʷ/ correspond to standard Tapilula pharngralized voiceless stops, and /mp mpʷ/ correspond to standard /mf mfʷ/. The prenasalized stops can be eliminated from the phonology if they are considered as clusters; however, the analysis would need to be different for different stops.

The vowel inventory was /a e i o u ə/. The schwa vowel here is a high vowel, not a true schwa. Only four of the vowels can follow a labialized consonant: /ʷe ʷi ʷo ʷu/, with /ʷo ʷu/ being most common.

Proto-Hipatal (0) to HP-1 (???)

This is a group set up for the blonde Lenians settlers of the large islands who left early on and were mostly swallowed up by an invasion of more powerful Lenians around 2600 AD. The existence of HP-4 proves that HP-1 must exist.

Proto-Hipatal (0) to HP-2 (???)

Culturally, this branch of the family is confined to smaller islands where the sea can be heard from any point on the island. Genetically, however, it also includes the Dreamlandic languages spoken on the mainland. The initial phonology was slightly different from that of Tapilula, generated by the following sound shifts:

1. Before a low tone, the fricatives h g were fortified to kʷ ḳʷ. Before a high tone, they disappeared. Thus the language became entirely free of fricatives.
2. Before a low tone, the lateral approximant l shifted to r.

At this stage the consonant inventory was

Rounded bilabials:     pʷ      mʷ  w     
Spread bilabials:      p       m   b       
Alveolars:             t       n   l   r
Rounded alveolars:     tʷ      nʷ       
Velars:                k   ḳ   ŋ  (Ø)        
Labiovelars:           kʷ  ḳʷ

There were six vowels, /a e i o u ə/, of which the last was a high vowel, not a true schwa. There were two tones. After a low tone, the stops were sometimes pronounced as fricatives in quick speech, but there was no phonemic contrast. The prenasalized stops /mbʷ mb nd ndʷ ŋġ mpʷ mp/ all occurred in root-initial position, but no classifiers began with a prenasalized stop, so very few words with initial prenasals were used. These are considered allophones of a homorganic nasal followed by a stop; however, they are of mixed origins.

The sequences /ʷe ʷi ʷo ʷu/ occurred, with the latter two being the most common. /kʷe kʷi/ were far more common than /nʷe nʷi/, etc, because of their origins.

There may still have been a rare /bʷ/.

If the labialized consonants are treated as clusters, the phonology reduces to

Bilabials:             p       m   w   b       
Alveolars:             t       n   l   r
Velars:                k   ḳ   ŋ      

1. The consonants t n l r k ḳ ŋ become palatalized to č ň y y č č ň before any /e/ or /i/.
2. Labialization was eliminated.

Syllabary

The language was written in a 14x15 syllabary in which tone was considered a property of the vowel. There were 154 syllables.

   LOW                          HIGH 
     a    e    i    o    u    ə    á    é    í    ó    ú    ə́               
    pa   pe   pi   po   pu   pə   pá   pé   pí   pó   pú   pə́        pṅ   pŋ̇
    ba   be   bi   bo   bu   bə   bá   bé   bí   bó   bú   bə́      
    ma   me   mi   mo   mu   mə   má   mé   mí   mó   mú   mə́    ṁ   mṅ   mŋ̇
                   wo   wu                       wó   wú
    ta   te   ti   to   tu   tə   tá   té   tí   tó   tú   tə́   tṁ        tŋ̇
    na   ne   ni   no   nu   nə   ná   né   ní   nó   nú   nə́         ṅ   nŋ̇
    ra             ro   ru   rə   lá   lé   lí   ló   lú   lə́   lṁ   lṅ   lŋ̇      
         če   či                       čé   čí             
         ňe   ňi                       ňé   ňí  
    ya   ye   yi   yo   yu   yə   yá   yé   yí   yó   yú   yə́
    ka   ke   ki   ko   ku   kə   ká             kó   kú   kə́   kṁ   kṅ   kŋ̇
    ḳa   ḳe   ḳi   ḳo   ḳu   ḳə   ḳá             ḳó   ḳú   ḳə́
    ŋa             ŋo   ŋu   ŋə   ŋá   ŋé   ŋí   ŋó   ŋú   ŋə́   ŋṁ   ŋṅ    ŋ̇

The sequences /ke ki ḳe ḳi/ only occur on a low tone; they may be in complementary distribution with the null-onset /e i/, but as there are two sets opposed to one, they cannot be eliminated from the syllabary.

Note that this language was spoken only a few hundred years after the breakup of Tapilula and Hipatal.

Proto-Hipatal (0) to HP-3 (???)

This is a group set up for the dark-skinned settlers of the large islands who left early on and were mostly swallowed up by the HP-4 invasion around 2600 AD. The existence of HP-4 proves that HP-3 must exist.

Grammar

HP-3 (and maybe HP-4) may have had gendered consonant insertion to break up hiatus in verbs, and possibly also nouns (i.e. a verb /yà/ becomes /pìpʷo/ when used with the epicene (/pʷ/) gender). Also, verbs may take a dislocated gendered suffix, such as -pʷ-o for verbs ending in /a/, due to analogy with compressed sequences like /aa/ > /a/. But note that these are two separate events; /aa/ > /a/ was very early in the parent language, thousands of years before the creation of the new hiatus sequences. A similar analogy happened in Dreamlandic, but not in the Gold branch. This could give rise to inherently gendered verbs both for humans and for animals.

Proto-Hipatal (0) to HP-4 (2600 AD)

Early HP-4 sound changes

This language is spoken in tropical rainforests of a chain of larger islands. It is one of the few groups to contain people who live more than a mile away from the seashore. It is the descendant of the language of Saffaslujje, the city from which earlier migrations had long ago set out. These people were Lenians who identified as Labans, and they swarmed over the previous three groups, but did not invade Dreamland because Dreamland was far stronger than Saffaslujje.

This may in fact be the only true use of the name Pejo, but the name was adopted by many other groups as well.

The original consonant inventory was

Rounded bilabials:     pʷ  bʷ  mʷ  mbʷ mpʷ     w
Spread bilabials:      p   b   m   mb  mp     (Ø)         
Alveolars:             t       n   nd          l
Rounded alveolars:     tʷ      nʷ  ndʷ     
Velars:                k   ḳ   ŋ   ŋġ      h   g

1. The high central vowel ə shifted to match the next vowel in the word. This also included the labialization of the initial consonant; thus, for example, /təpʷu/ > /tʷupʷu/.
2. The voiceless aspirated velar stop k shifted to h unconditionally. Ejectives and labialized forms were unaffected by this change.
3. The labialized consonants lʷ gʷ shifted to w .
4. The labialized alveolars tʷ nʷ ndʷ shifted to kʷ mʷ mmʷ.
5. Schwa disappeared between a nasal and a following stop or fricative; if there was a fricative, it became a stop.
6. Initial schwas disappeared.
7. The ejective stops ḳ ḳʷ shifted to k kʷ.
8. Any remaining schwa ə shifted to i.
9. The rounded vowel o shifted to a unconditionally. /u/ became unrounded, but there was no change in spelling.
10. The mid vowel e shifted to ə unconditionally.
11. The prenasalized voiced stops mbʷ mb nd ŋġ ŋġʷ shifted to the double nasals mmʷ mm nn ŋŋ ŋŋʷ.
12. The voiced velar sounds ŋ g disappeared to Ø Ø. This did not affect the geminate /ŋŋ/.

    If the Sun languages diverge before the Rain languages, this would be a good spot to do it, because it is necessary that they lose the inherited /g/ in order to make a new one.

13. The labialized approximant bʷ shifted to w .
14. Double nasals were reduced to singles.

Thus the consonant inventory of HP-4 was

Rounded bilabials:     pʷ  mʷ  w    
Spread bilabials:      p   m   b       
Alveolars:             t   n   l    
Palatals:                      y 
Velars:                k   ŋ  (Ø)  h
Labiovelars:           kʷ             

The vowel inventory was /a i u ə/, with the labialized stops /pʷ kʷ/ appearing before all four vowels. There were syllabic nasals /ṁ ṅ ŋ̇/. The voiceless fricative /h/, the only fricative in the language, was highly variable in pronunciation, often being labialized or palatalized or both.

An alternative analysis removes the labialized consonants and analyzes /y w/ as vowels. This leaves a consonant inventory of just

Labials:     p   m   b
Alveolars:   t   n   l
Velars:      k   ŋ   h

Although the /h/ is voiceless, it may have been evolving towards a voiced pronunciation, meaning the consonant inventory can be analyzed as a matrix where the only variables are place of articulation (labial/alveolar/velar) and manner of articulation (stop/nasal/approximant). The syllable structure was entirely CV except for the syllabic nasals, so the stop allophone of /b/ would have been rare. The velar nasal /ŋ/ is rare enough that it could shift to /n/ in every daughter language.

Rain languages

HP-4 daughter languages are grouped into Rain and Sun branches, corresponding to phonology, not climate. The Rainy languages are the ones in which /h/ remained voiceless and then went on to create sibilants. These languages diverged into four (or more) branches at the same time, with no shared traits at all, because the four groups of migrants settled new islands at the same time (or possibly four areas on a larger island, etc).

This means that the Sun branch is either nested with the Rain branch, or the Sun branch is actually slightly older than the Rain branch and a new proto-language will need to be drawn up. Below assumes that the Rain-Sun division is older than the internal Rain splits.

The presence of /s/ and in some cases persistence of /h/ gave these languages a "wet" sound.

It is possible that one of these languages is actually the Laban mainland dialect, just as Rain/Sun (HP-4) was the language of the people who had never left the Laban mainland during the earlier migration.

Rain (2600 AD) to Umbrella (Pamā)

This is the numerically dominant group among the conquerors, unless Hunting Vest is; Hunting Vest in any case survives to eventually reach Dreamland.

This branch shifts all of its labialized consonants to pure labials, and then grows new labialized consonants from sequences like /awa/ and /ua/.

The original consonant inventory was

Rounded bilabials:     pʷ  mʷ  w    
Spread bilabials:      p   m   b       
Alveolars:             t   n   l    
Palatals:                      y 
Velars:                k   ŋ  (Ø)  h
Labiovelars:           kʷ             

The vowel inventory was /a i u ə/, with labialized consonants appearing before all four vowels. There were syllabic nasals /ṁ ṅ ŋ̇/.

1. The labialized consonants pʷ mʷ kʷ shifted to p m p. The labial approximant w may have also shifted to b.

   NOTE, this is probably not the first shift in the history, but I am putting it here to fill out the sketch.

2. The vowel sequences ua ui uə shifted to wa wi wə.

Thus the final consonant inventory MAY HAVE BEEN

Labials:       p   m   b
Alveolars:     t   n   l
Palatals:              y
Velars:        k   ŋ  (Ø)  h

Rain (2600 AD) to Yellow Shoes (Nannapànnu)

The original consonant inventory was

Rounded bilabials:     pʷ  mʷ  w    
Spread bilabials:      p   m   b       
Alveolars:             t   n   l    
Palatals:                      y 
Velars:                k   ŋ  (Ø)  h
Labiovelars:           kʷ               

The vowel inventory was /a i u ə/, with labialized consonants appearing before all four vowels. There were syllabic nasals /ṁ ṅ ŋ̇/.

1. The alveolars t n l shifted to f m w.
2. The velars k ŋ h shifted to t n s.
3. Labialization was lost.

At this point the phoneme inventory was

Labials:           p   m   f   b
Alveolars:         t   n   s
Palatals:                      y
Velars:            k

It is not clear if the delabialization shift implies /w/ > /b/ or not, so tentatively there could be a separate /w/ sound alongside the inherited /b/.

Rain (2600 AD) to Hunting Vest (Lākaha)

End time is 4400 AD. These people might settle Dreamland after the collapse of civilization on the mainland.

The original consonant inventory was

Rounded bilabials:     pʷ  mʷ  w    
Spread bilabials:      p   m   b       
Alveolars:             t   n   l    
Palatals:                      y 
Velars:                k   ŋ  (Ø)  h
Labiovelars:           kʷ              

The vowel inventory was /a i u ə/, with labialized consonants appearing before all four vowels. There were syllabic nasals /ṁ ṅ ŋ̇/.

1. The velars k ŋ h shifted to č ň š unconditionally.
2. Labialization was lost.
3. The sequences àa àə ə̀a shifted to ā. Then ăa ăə ə̆a shifted to â, and ə̀ə ə̆ə shifted to ə̄ ə̂.
4. The postalveolar fricative š shifted to s.
5. The sequences ìa ìə ùa ùə (where the first vowel has a high tone) shifted to èa ìe òa ùo.
6. The vowel sequences ìi ùu became ī ū.
7. The vowel sequences ĭə ŭə (equivalent to /iə̀ uə̀/) shifted to yè wò.
8. Before a vowel, remaining i u shifted to y w.
9. The vowel sequences ài ăi àu ău shifted to ē ê ō ô.
10. The vowel sequences ə̀i ə̆i ə̀u ə̆u shifted to ī î ū û.
11. The sequences ky ŋy ty ny sy ly shifted to č ň č ň š y.
12. The sequences tw nw sw lw shifted to p m f w. All other consonants preceding /w/ shifted to labials.
13. Remaining post-consonantal /w/ and /y/ were deleted.
14. The sequences èa ìe òa ùo shifted to ya ye wa wo. (Tone may have been influenced by surrounding syllables.) Meanwhile ùi ìu shifted to wi yu.
15. All sounds preceding a /w/ again became labials.
16. All post-consonantal /w/ and /y/ were deleted.

The circumflex is an ad-hoc symbol for a long low tone. However, it may make sense to retain the circumflex vowels as sequences, as there are other sequences that would arise at morpheme boundaries.

Thus the consonant inventory was

Bilabials:          p  m  b  f  w
Alveolars:          t  n     s  l
Palataloids:        č  ň     š  y
Velars:             k

And there were six vowels, on two tones, and could be short or long.

Rain (2600 AD) to Baseball Cap (Hahakànna)

The original consonant inventory was

Rounded bilabials:     pʷ  mʷ  w    
Spread bilabials:      p   m   b       
Alveolars:             t   n   l    
Palatals:                      y 
Velars:                k   ŋ  (Ø)  h
Labiovelars:           kʷ            

The vowel inventory was /a i u ə/, with labialized consonants appearing before all four vowels. There were syllabic nasals /ṁ ṅ ŋ̇/.

1. Before a vowel, i shifted to y.
2. Labialized consonants defeated any following /y/.
3. The velar sequences ky ŋy hy shifted to č ň š. The alveolar sequences ty ny ly also shifted to č ň ł.
4. Labialization was lost.
5. The postalveolar fricative š shifted to s.
6. The sequences py my by shifted to č ň y.

Thus the consonant inventory was

Labials:              p   m   b   w
Alveolars:            t   n   l       s
Palataloids:          č   ň   ł   y   
Velars:               k   ŋ           h

Rain (2600 AD) to Snow Pants (Tākapi)

The original consonant inventory was

Rounded bilabials:     pʷ  mʷ  w    
Spread bilabials:      p   m   b       
Alveolars:             t   n   l    
Palatals:                      y 
Velars:                k   ŋ  (Ø)  h
Labiovelars:           kʷ                

The vowel inventory was /a i u ə/, with labialized consonants appearing before all four vowels. There were syllabic nasals /ṁ ṅ ŋ̇/.

1. Single nasals metathesized across a vowel to form clusters with the next consonant. These all became homorganic; here, a /w/ behaved as a labiovelar, thus the resulting cluster was /ŋw/.
2. The clusters ŋw ŋh nl shifted to ŋʷ h l. /mb/ remained.
3. Double nasals shifted to singles.
4. The syllabic nasals ṁ ṅ ŋ̇ shifted to um un uŋ unconditionally.
5. The sequences aa aə əa , on all tones, merged as ā. əə shifted to ə̄.
6. Any h bordering an /i/ in either direction shifted to s.
7. The sequences ii uu shifted to ī ū if the first tone was high; otherwise they shifted to yi ʷu.
8. Any low-tone i before a vowel became a palatal approximant y.
9. The sequences ty ky ny ŋy sy ly shifted to č č ň ň š ł. (/hy/ > /sy/ earlier.)
10. Labial and labialized consonants swallowed a following y.
11. The approximant lʷ shifted to w. The alveolars tʷ sʷ nʷ changed in a split shift to kʷ hʷ mʷ.

Thus the consonant inventory was

Rounded bilabials:     pʷ  mʷ  bʷ  w    
Spread bilabials:      p   m   b  (Ø)   
Alveolars:             t   n   l       s
Palatals:              č   ň   ł   y   š   
Velars:                k   ŋ      (Ø)  h
Labiovelars:           kʷ  ŋʷ          hʷ   

The four-vowel inventory remained, but there were many more long vowels and vowel sequences than there had been before. The falling diphthongs were /ai au əi əu/. These could be analyzed as /ē ō e o/, giving the language a six-vowel inventory with no (falling) diphthongs and a two-way length contrast on all six vowels, but the orthography nevertheless used four vowels.

Sun languages

These languages are derived from HP-4 but shifted /h/ into a voiced fricative. The first Sun people were a dark-skinned tribe who either rebelled against or invaded the Lenians who had recently conquered the sea. These conquerors came to be called the Rain people; their languages were closely related, and the hostility may not have lasted long.

Early shifts

This language assumes Sun branches earlier than the six Rain languages.

The original consonant inventory was

Labials:     p   m   b
Alveolars:   t   n   l
Velars:      k   ŋ   g

The vowel inventory was /a i u ə/, with /w/ permitted after any of /p m k/ before all four vowels. There were also standalone approximants /w y/ which could be analyzed as independent consonants or as allophones of the vowels (there was hiatus, but this too could be analyzed as /uwa/, etc).

1. When not labialized, the obstruents p m k shifted to f m₃ k₂.

   This is a dummy shift and it is likely that /ṃ ḳ/ will just shift back to the regular forms, or to /m č/.

2. The nasal sequences mm nn shifted to m₂ n₂.

   This is another dummy shift ... see below.

3. The labial approximant w shifted to g in at least some environments (perhaps after a high tone).

The phonology at this stage was

Labials:        p       m₁  m₂  m₃  b   f
Alveolars:      t       n₁  n₂      l
Palatals:                           y
Velars:         k₁  k₂  ŋ           g

If double nasals were inherited from the parent language (unlike Rain), then it is possible that they could split into duplicate sounds as well, so that there would be, e.g. /m₁ m₂ m₃ n₁ n₂ ŋ/. There would be only one /ŋ/ because it occurred *only* when doubled.

It is almost certain that /mʷ/ would merge with /mm/, however. It is possible that the fortis nasals turn back into prenasalized voiced stops, and if the shift happens early enough, it could be that this is a dialect where they never really changed in the first place, even if the other Sun languages do not do that.

This branch then might go two ways: one path shifts /f/ > /b/, possibly thus ending up with the same phonology it began with. The other branch shifts /f/ > /h/, while keeping /g/, thus having two velar fricatives and no other fricatives.

It is not clear whether m₃ or m₁ is the more marked phoneme. Theoretically there could have even been a fourth one, from /mbʷ/, but this would have been so rare as to possibly never arise in the parent language.

Note that /k₁/ is marked, not k₂. If /k₂/ is /č/, it could probably best be represented as /t/. It could also be such as

/k₂/ shifts to /t/ before /i/, and remains as /k/ elsewhere.

This could be problematic, though, since there was little /k₁1/ in the parent language.

Sun to Sparc

The original consonant inventory was

Labials:        p       m₁  m₂  m₃  b   f
Alveolars:      t       n₁  n₂      l
Palatals:                           y
Velars:         k₁  k₂  ŋ           g

Sun to TRU64

The original consonant inventory was

Labials:        p       m₁  m₂  m₃  b   f
Alveolars:      t       n₁  n₂      l
Palatals:                           y
Velars:         k₁  k₂  ŋ           g

Sun to Deskjet

The original consonant inventory was

Labials:        p       m₁  m₂  m₃  b   f
Alveolars:      t       n₁  n₂      l
Palatals:                           y
Velars:         k₁  k₂  ŋ           g

Sun to Impresario

The phonology at this stage was

Labials:        p       m₁  m₂  m₃  b   f
Alveolars:      t       n₁  n₂      l
Palatals:                           y
Velars:         k₁  k₂  ŋ           g

1. The labial fricative f shifted to b unconditionally.
2. Before any /i/, the velar stop k₂ shifted to t.

   The rationale for this is that this branch had more /ʷi/ than the parent language did. However, there was still markedly less of it than before other vowels, and so this shift would make /ki/ quite rare.

3. The nasals m₂ n₂ (the ones that arose from geminates) shifted to mb nd.
4. The labial nasal m₃ also shifted to mb.
5. The sequence nd came to be spelled r.

The phonology at this stage was

Labials:     p   m   mb  b
Alveolars:   t   n       l   r
Palatals:                y
Velars:      k   ŋ       g

Sun culture

It is possible that these are the languages spoken by dark-skinned people, in which case the presence of /s/ came to be seen as a stereotypically Lenian trait. However, it is also possible that there is no correlation between tribal identity and position in the linguistic tree, and even that there is no genetic grouping based on the /h/ > /g/ sound change at all, since the Wet languages begin diverging from each other almost immediately.

Consider also a satem/centum like split, where one side of the divide really is genetic (marked by a single shared change), while the other side is not (marked by preserving the old state and then dividing only later on). Since the Rain languages all split at the exact same time, they share no changes (though sometimes the same shift occurs at different times in two Rain languages), which would mean that the Sun branch is the one that is truly genetic. It may have split off earlier than the Rain languages did and therefore have a slightly different starting phonology, but the shift of /bʷ/ > /w/ is likely to have occurred independently, and even if the double nasals are retained, the phoneme inventory will still be the same.

Alternatively, Sun splits off very early from within one of the Rain branches, or is a seventh branch of the tree that came to be considered different only for superficial reasons.

Becuase of the recent loss of the primordial velars, /g/ is likely to remain as a stable sound rather than following the others into hiatus. There is also a possibility that it could merge with /ŋ/, in either direction. The voiced labial /b/ was an approximant except after a syllabic nasal. Syllabic nasals themselves may well disappear.

This branch is likely to be more conservative than the Rainy branch, which itself is more conservative than most other Tapilula languages. Tone is likely to remain, as with Rain.

Shared cultural history

A short timeline:

1. Both light and dark skinned people set out from the city of Saffaslujje (also Tapaluka, etc) around 600 AD and sail eastward through the vast Hipatal ocean. They mostly settle large islands and leave smaller islands alone. There were two light-skinned groups, here HP-1 and HP-2. The dark-skinned group was HP-3. Culturally, HP-1 and HP-3 had the most in common despite their appearance, because they were the stronger ones, who were able to take the best islands for themselves. The two groups were initially separate, but likely had many intimate contacts throughout the years. The distinction thus is linguistic. Nevertheless, clear racial differences persisted from one island to the next, and for the larger islands even between different shorelines, because of the different paths the two groups had initially set out on.
2. HP-2 was forced to migrate longer distances and settle islands in poorer natural habitats, but their ultimate goal was not the islands but Dreamland, which they reached around 1700 AD.
3. After the discovery of Dreamland, HP-2 rises and becomes numerically superior and vastly stronger and richer than the other two groups, but only a few of these people, who by then called themselves Dreamers, were interested in returning to the islands. Additionally, a small number of HP-2 people had remained on the islands all along and did not move to Dreamland. These people lost their cultural identity and came to associate with the blonde HP-1 settlers.
4. Around 2600 AD, a new group of blonde Lenian settlers, here known as HP-4, migrates eastward from Saffaslujje and swarms over HP-1 and HP-3 territory, subjugating both equally. (They also conquer the few remaining HP-2 islanders to better control the seas.) Some HP-1 and HP-3 languages survive, but HP-4 becomes numerically dominant even during the early period in which there was only one HP-4 language.
5. The HP-4 conquerors were stronger than the islanders, but much weaker than the Dreamers, so they never manage to invade Dreamland. Additionally, at least one group of their subjects either rebels or actually invades them, and these people become the Sun people. The Sun people were of the dark-skinned group, but friendly to the Lenians who had been on the islands all along (the Rain people), and hostile to the Lenian conquerors newly arriving from the west. Importantly, the Sun people's language was HP-4, not HP-3.
6. The Sun people become powerful enough to start building colonies of their own. It is not clear yet what their political allegiances were; some may have defected to Laba, and it is possible that nearly all of the islanders eventually defect to Laba even as Laba's government enforces a rule denying Laban citizenship and basic civil rights not only to dark-skinned settlers, but to the mixed-race people who had developed from prolonged contacts between the descendants of the original migrants.
7. Eventually Laba wins control of parts of the mainland, but the language likely does not change; this is a political revolution, not an outright war.
8. By the time Dreamland starts fighting large wars, Laba has become a firm ally of Dreamland and holds firm control over the entire Hipatal sea. However, they are a very weak ally and Dreamland is instead run by the Baywatch party, located in the extreme eastern edge of Dreamland, far from Laba's territory on the western tip.