Languages of Teppala

Humans on the planet Teppala are confined to a single continent, Rilola, and its offshore islands. Thus all human languages can be traced back to a single starting point, and have many traits in common.

Historical distribution of Teppalan languages

Human civilization on planet Teppala peaked twice: first around 1700-2600AD, and then again around 3500-4200AD. During the first peak, the human population was concentrated on the south coast of the continent of Rilola, from 10°N to about 28°N. During the second peak, humans were more spread out, but the areas with the most political and military power were in the interior of the continent, though always along major rivers and lakes, ranging from about 30°N to 37°N in the eastern (older) areas of settlement and upwards to about 45°N in the far west.

After the second peak, human civilization entered a long decline, the population decreased, and languages with small populations went extinct.

During the peaks of human civilization, it was common for every religion to have its own language. Since religions were tied to political parties, almost every political party also had its own language. This is why many Teppalan languages have unusual names, such as the Gold language and the Moonshine language; these were named after political parties rather than ethnic groups.

Since political parties (and religions) coexisted with rival parties within the same ethnic group's nation, the boundaries of a given language often hinged on tiny differences such as speech registers or a set of unique vocabulary words used only members of a particular political party. When languages such as this coexisted, they tended to evolve in parallel directions, so that mutual intelligibility might persist among a pair of languages for hundreds of years.

However, democratic governments were very rare throughout Teppalan history, and many political parties, once in power, banned and persecuted members of rival parties. In these nations, the population really did consist of just a single ethnicity since anyone not claiming membership would be arrested or killed.

Traits common to all Teppalan languages

Phonology

1. Bilabial consonants are very common, especially the stops p b and the nasal m.
2. The commonest syllable shape is always CV, even in languages that allow dense consonant clusters and/or diphthongs and triphthongs.
3. If a language has only one series of coarticulated consonants, they are labialized consonants.
4. No language has more than six vowels. All vowels are one of these inventories:

   /a i u/

   /a ə i u/

   /a e i o u/

   /a e i o u ɨ/

   /a ə ɨ/

   Some differences in Romanization may appear such as writing a schwa as e or the high central vowel as schwa.

5. Any consonant that can occur in syllable-final position can also occur in syllable-initial position.
6. Languages are consonant-strong: consonants have greater effect on surrounding vowels than vowels have on consonants. For example, if all vowels are inherently unrounded, rounded allophones appear adjacent to labialized consonants.

Grammar

Grammatical traits common to all of the languages on the entire planet throughout all of recorded history are confined to negatives.

• For example, no Teppalan language has or has ever had a definite or indefinite article.

Traits common to most Teppalan languages

Phonology

1. The commonest stop in a language is usually /p/. However, in some languages, this is because /p/ stands alone whereas stops further back in the mouth are divided into several articulation types. For example, a language may have /p/ as its only bilabial stop but also have a distinction between a plain velar stop /k/ and an ejective /ḳ/; or there may be co-articulations such as palatalization or labialization associated with dorsal stops but not with bilabials.
2. Languages with five or more phonemic vowels often do not permit diphthongs; those that do have a very small set. Thaoa is an outlier in that it has six phonemic vowels and several diphthongs. Note that rising diphthongs are generally parsed as a sequence of consonant + vowel since, in almost all Teppalan languages, there is no restriction on which vowels can follow an onset of [j] or [w].
3. Dense consonant clusters do not appear, except in some languages where a certain vowel, usually /a/ or /ə/, is not distinguished at the phonemic level from silence. That is, some languages may always pronounce /tk/ as [tək], without the schwa actually being present as a phoneme.
4. Syllables are commonly front-loaded, such that a sequence like /papsa/ is more likely to be pronounced [pa.psa] rather than *[pap.sa]. This happens most often when the first element of a cluster is lower on the sonority hierarchy than any following consonants, but in many languages, nasal-stop clusters such as /mp nt ŋk/ will also be front-loaded.
5. On the continent of Rilola, the voiced velar stop /ġ/ (IPA /g/) is rarely used even in languages with a robust series of voiced stops. This is a long-standing trait and crosses language family boundaries. On the islands of Laba, however, /ġ/ is common.
6. There are no minimal pairs between a diphthong and a sequence of the same two vowels. Thus diphthongs can be analyzed as allophones of vowel sequences.
7. Voiceless obstruents occur more frequently than voiced ones. In some languages, /b/ or /d/ is the only voiced stop. In others, there are no voiced stops at all but the voiced velar fricative /g/ (IPA /ɣ/) takes on a stop allophone after a nasal or a high tone.
8. There are often marginal consonant phonemes. These mostly arise from previously existing consonant clusters that were worn down. However, some marginal phonemes arise from sound changes affecting consonants that previously were more common, which survived in only a few phonemic environments. For example, in Khulls voiced stops survived a lenition shift only after a nasal. Later, the nasal sometimes disappeared, meaning that the voiced stops could no longer be analyzed as allophones of voiced fricatives. But they remained rare.
9. It is common to have restrictions forbidding certain consonants to appear in certain parts of a word; for example, in Khulls /r/ cannot begin a word. Most languages allow only a small subset of their consonants to appear at the end of a word

Grammar

• Polysynthesis is common, and it is nearly always fusional. It is common to find words with more morphemes than phonemes due to the prevalence of single-phoneme grammatical suffixes that represent two individual morphemes that at an earlier stage of the language were pronounced individually but combined into a new single sound after a sound change.
• Pronouns are of limited usage, as verbs usually carry the relevant information about the person, number, and gender of the participants. In some languages, pronouns are entirely absent.

Gender

• If there is any grammatical gender, feminine words in most semantic fields outnumber masculine ones.
• Grammatical gender, if present, classifies people based on age and sex rather than just sex. The age categories are not firmly defined and can be used metaphorically. There are often several age categories for children, but all adults share just a single age category.
• Many languages have a consonant-based gender system; if not grammaticalized, associated more weakly due to persistence of previously grammaticalized morphemes and contact with nearby languages.
• Mixed gender categories are often present; a man and a woman, referred to as a unit, will take an epicene gender rather than having one gender overrule the other. If there is no epicene, a group containing both males and females will be described with words in one of the feminine genders.
• Gender and animacy, if present, can be inherited by nouns describing syntactically inanimate objects, by borrowing from a parent object. That is, a man's arm will be animate (and masculine), and so will his books or any other possessions.

Structure of words

• Most languages have a fixed word order, with SOV word order by far the most common. Anomalous constructions within SOV languages may require additional marking on certain words, usually those found near the beginning of the sentence.
• Suffixes are nearly universal; infixes are common; prefixes are rare and usually confined to a single part of speech (for example, in Andanese, the only prefixes are the noun classifiers).
• Verbs are generally the longest words in a sentence. Verbs are heavily inflected even in languages where noun inflection is absent or relatively skimpy.

Parts of speech

• There are no adjectives or adverbs. Verbs are used instead of these. In some languages, even the nouns can be analyzed as a subset of the verbs.
• Pronouns play only a minor role in the language, and some languages lack pronouns altogether, instead using nouns and verbs with person markers.
• Person markers on nouns, denoting their possessor, are common. Languages that lose this system often redevelop it from grammatically unrelated words later on.

List of very early sound changes

Many minor languages have been wiped from this list; some are recorded on their own articles, but others are only in the edit history.

Pre-Mapi changes

Proto-Greater-Laban (18343 BC) to Mapi (year 14000 BC)

The Pre-Proto-Macro-Hyper-Greater-Laban (PGL) language had a consonant inventory of /pʷ p mʷ m hʷ w t c s č š j k kʷ ŋ ŋʷ h l r ř/ and a vowel inventory of /a i u ə/. Schwa is usually spelled "e". /ŋ/ was allophonically [g] and the other nasals could become stops occasionally in word-initial position. At first, "c" was homophonous with the cluster /ts/, but later /ts/ was dragged down to dental and /c/ remained alveolar.

See Primordial scratchpad for details on the language.

1. The primordial final nasal -/n/ disappeared, but lengthened any preceding vowel (including schwa).
2. The schwa vowel ə disappeared in all positions, creating new consonant clusters.
3. Consonant clusters like hp ht hk flipped to put the stop first, as in ph th kh.
4. Possibly also /ki/ > /ć/ or even just /k/.
5. Word-final h disappeared, and also lengthened any preceding vowel. Thus, the final syllable /-hə/ had vanished completely. However, note that word-internal /h/ that was at the end of a syllable had survived this change.
6. Syllable-final -p became the glottal stop ʔ, except before a "weak" sound such as /h/ or before a vowel-initial syllable.
7. The consonant clusters ʔs ʔš ʔts ʔč became c č c č.
8. The consonant clusters ks kš kts kč kh sk šk čk tsk hk all become kh.
9. The consonant clusters ts th st ht all became th.
10. The consonant clusters ph sp šp čp tsp hp' all become ph.
11. Clusters of two voiceless stops were resolved in favor of the second stop, except for /pk/ and its kin.
12. The aspirated nasal mh became h.
13. The labialized stops kʷ kʷh changed to pʷ pʷh.
14. The aspirated stops pʷh ph th kh changed to hʷ h h x respectively. (possibly use f f f x)
15. The affricates c č became s š.
16. The glottal fricative h disappeared unconditionally, leaving vowel hiatus. This included the deaspiration of hʷ to w along with any other remaining sequences such as /nh/. VOWEL HIATUS STAGE 1!!!!
17. The vowel sequences ai au contracted into new vowels ē ō, but retained the glide if another vowel followed.
18. The identical vowel sequences aa ii uu became ā ī ū. The vowel system was thus /a i u ā ē ī ō ū/ and a tiny bit of remaining schwa.
19. The vowel ō was shortened to o, but ē remained long.
20. Before another vowel, the sequences aw iw ow uw āw ēw īw ūw became ō ū ō ū ō ē ī ū.
21. Before another vowel, the vowel u changed to w.
22. Voiced stops disappeared unconditionally, leaving vowel hiatus. VOWEL HIATUS STAGE 2!!!!! Note that sequewnces like /sua/ now contrasted with /swa/ from the earlier change.
23. The labialized fricative sequences sw šw xw hw changed to h. Note that there was still an /x/, from earlier /kh/. THus there was a contrast between /h/ vs /x/, but neither of them had labialized forms.
24. The short vowels e o merged as ə (usually spelled "e"). The long vowels ē ō merged as ə̄ (usually spelled "ē").

Pre-Tapilula changes

NOTE

Some languages in this section are reconstructed internally, meaning that they have the smallest possible phonology capable of developing into the phonologies of its immediate daughter languages. Marginal phonemes can thus be inserted arbitrarily into any language provided that they merge into one of the phonemes of the daughter language.

NOTE

Many ideas deleted. See history for info.

There were 3 central vowels: /a ɜ ɨ/. Further back in time, each of these three could be preceded by the palatal glide /y/, even after a consonant, but later the palatal glide disappeared after labials & labialized consonants, and merged with the other consonants to form palataloids. This happened without creating any new vowels.

The daughter languages are Silatibarra (8000 BC; sometimes just called "Southeast Laban"), proto-paleo-Andanese (PPAnd), and proto-macro-Haswarabic (PMH).

All of the labialized consonants were "stops" (nasal or oral). The labialized nasals /mʷ ŋʷ/ may have been pronounced as voiced stops (as in Khulls). The lack of the expected */xʷ hʷ/ was due to a late change in the parent language, and in many ways the /h/ patterned as if it were the labialized counterpart of /x/.

Syllables were CVC at maximum, and there were limits on the coda ... a preference for nasal codas if there is one. PoA was not contrastive before another consonant. The final stops allowed were /t, kw/ and possibly /ć/. There were thus no syllables like /kʷya/ or even /pya/.

Note that Mapi is the parent language of the Paleo-Andanese languages, so called because they are the ancestral languages of the people who, 14000 years later, began to speak Tapilula and later Andanese.

Stops & frics early on came to be voiced when occurring after a nasal.

Mapi (14000 BC) to Primordial MRCA (11000 BC)

The grammar of this language may have had infixes, because they are present in both Owl and Tapilula. The MRCA of those is in the middle of this chart, ~11000 BC. Nonetheless, Tapilula's infixes can be "recovered" (biological sense) from suffixes and therefore Mapi can have essentially any grammar.

init:

Rounded labials:      pʷ mʷ    w
Plain bilabials:      p  m
Alveolars:            t  n  s  l  r
Postalveolars:        č  ň  š  ł  ř
Palatals:             ć  ń  ś  
Velars:               k  ŋ  x
Labiovelars:          kʷ ŋʷ 
Glottals:                   h

1. The fricatives š ś h shifted to y y Ø unconditionally . This created vowel sequences of /aa aɜ aɨ ɜa ɜɜ ɜɨ ɨa ɨɜ ɨɨ/ most of which occurred as the only vocoid in a word since most roots had been bisyllabic.
2. The sequences ɨa ɨɜ ɨɨ shifted to ʕa ʕɜ ʕɨ, where /ʕ/ is a voiced stop after a nasal and otherwise a fricative(?). Thus prenasalized stops were created. rule3 might not be needed if we keep h
3. The sequences ɨwa ɨwɜ ɨwɨ shifted to ʷa ʷɜ ʷɨ.
4. The sequences ɨya ɨyɜ ɨyɨ shifted to ʲa ʲɜ ʲɨ . Thus new "mismatched" palatalized (pʲ mʲ ) and labialized (tʷ nʷ sʷ lʷ rʷ xʷ ) consonants were created, and they were a full set.
5. The sequences aa aɜ ɜa ɜɜ shifted to a: a: ɜ: ɜ:. aɨ ɜɨ shifted to ɜ: ɜ: as well.
6. mʷ nʷ ŋʷ > mbʷ nd ŋġʷ. These may have also been the reflexes of /mw nw ŋw/, but the syllable boundaries would have kept them apart even so. That is, a syllable-straddling /m-w/ would turn into /m-mbʷ/.
7. Fricatives became weakened between vowels, and in syllable-final position: s x sʷ xʷ changed to h h hʷ hʷ.
8. The cluster ts shifted to s.
9. The labial consonants p pʷ pʲ became B w y.

And the vowel inventory was probably

SHORT VOWELS

a   ɜ   ɨ

LONG VOWELS AND FALLING DIPHTHONGS

ā  ai  au
ɜ̄  ɜi  ɜu
ɨ̄   ī   ū

See Repilian languages.

Primordial (11000 BC) to Southeast Laban (8000 BC)

1. The fricatives h hʷ (including the newly generated ones) in syllable-final position voiced to Ø w. ś had been shifted to /y/ earlier. Thuis frics no longer occured at end of tillable.
2. rh > h.
3. mʲ > mb.
4. Possibly, also hʷ h ś > vless stops when after a nasal, but they later become voiced in Tapilula.
5. The voiceless stops č ć shifted to s₂ s₃ (cover symbol: $) between vowels.
6. Most postalveolar consonants trapped in final positions become plain alveolars: č ñ ł become t n l. Note that there were no longer any palatals in the language.
7. ć ń ś > s₄ ň s₅.
8. Probably lʷ > w; maybe ł > y.
9. s₅ s > s₁.
10. Probably tʷ > kʷ.
11. The sequences ɜ̄ ɜi ɨ̄ ī shifted to ē ē ī ī unconditionally.
12. The high vowel ɨ shifted to ū before a labialized consonant in a closed syllable. (This is a dummy shift to make it easier to understand changes that happen in Tapilula, but it may be of use in other daughter languages as well.)
13. Final -ɨ after a labialized consonant shifted to u. Final -ɨ after a palatalized consonant shifted to i. These are both dummy shifts as well.

And the vowel inventory may have been /a ɜ ɨ/ for short vowels and /ā ē ī ō ū/ for long vowels.

The prenasals contrast with syllable-straddling sequences of nasal + *voiceless* stop, and with nasal + prenasal (that is, /mmb/ etc).

Southeast Laban (8000 BC) to Tapilula (500 BC)

Rounded labials:                  w      mbʷ
Plain bilabials:          m              mb  B                        
Alveolars:            t   n       l   r  nd  $
Rounded alveolars:                   (rʷ)
Postalveolars:        č   ň       λ   ř  
Palatals:                         y  
Velars:               k   ŋ   h              G 
Labiovelars:          kʷ      hʷ         ŋġʷ

Tentative consonants are in parens. Note that /nʷ/ is not inherited from MRCA, so it's possible that /tʷ/ > /kʷ/ > /p/ and that all /tʷ/ in Tapilula is secondary just as all nʷ must be.

There may be a few other consonants not listed here resulting from the creation of new labialized and palatalized consonants in the parent language.

The cover symbol B can be used for "a consonant that needs to disappear" to Ø since there is so much vowel hiatus required to get things the way they are in Tapilula. The cover symbol $ can likewise be used for any consonant that ultimately ends up as /h/, since there is so much /h/ in Tapilula as well. Lastly, G indicates any consonant that ends upo as /g/, though this might be adqeuately covered by grammatical reanalysis from hiatus and not need a new consonant.

The B consonant was most likely just /b/, but the capital letter spelling leaves open the possibility that it traces back to more than one phoneme and that these were still distinct at a fairly late stage of development, just as $ corresponds to a set of phonemes that merged only very late.

The vowel inventory was

a ɜ ɨ
ā ɜ̄   ē ī ō ū

These are spelled inconsistently however, and there may have also been a fourth pair of long vowels for primordial /ai au/.

1. Final short ɜ ɨ shifted to ɨ Ø unconditionally. Here, "final" refers to polysyllabic words only.

   It is possible that at this time, a plural -ɨw (later /u/) became generalized to all consonant stems, from original ɨ-stems. It is not clear what the original plural would have been from the pre-existing consonant stems.

2. In at least some combinations, word-internal ɨ also disappeared to Ø, thus allowing the later creation of word-initial /ḳ/ and, through contraction of -ɨB-, a source of new labialized consonants. Alternatively, this shift could be united with the shift below in which /ɜy ɨy ɜw ɨw/ become /i i u u/.

   This created new clusters, which in some cases collapsed to new single consonants. All were rare:

   1. bh shifted to p. (If B represents more than one phoneme, it will need to be separated from the true /b/ by this shift.)
3. mB > mb.
4. Word-final clusters became simple, losing their second element.

   Note that this implies there will be little or no final /ə/ in Tapilula.

5. Probably all $G > $.
6. Intervocalically, the geminate nasals mm nn ňň ŋŋ shifted to the prenasalized voiced stops mb nd ňǯ ŋġ .
7. An h or hʷ after a nasal also switched to a voiced stop, thus creating a prenasalized voiced (not voiceless) consonant.
8. The cluster mB now also shifted to mb.

   This is important much later on, as it allows epenthetic /b/ for vowel-initial words when padded by a 1st person marker (later extended to places where it would not be etymologically sound).

   This also means that /mB/ and /mh/ have identical reflexes and that analogy may have begun very early. Also, note that /mbʲ/ later just becomes /mb/, so even the new /h/ that arises later on can sometimes be treated as B since this /h/ can come from /gi/ before a vowel. Importantly, the 1P>2P marker in Tapilula is /nambə/, just as if it were from /nam/ + /gə/ instead of /nam/ + /hə/. Essentially, the passive morpheme vanishes.

9. The sequences pm tn čň kŋ shifted to mm nn ňň ŋŋ.
10. A prenasalized voiced stop following a closed syllable became denasalized; any such preceding coda became voiced. This was allophonic, so that for example [bb] was still underlyingly /pmb/. Or, maybe they were devoiced.
11. Word-initial approximant w became hʷ .

    NOTE ON POLITICS: This is 3770 BC, a point of no political significance, but where the language suddenly began changing very rapidly instead of very slowly.

12. Before a vowel, the sequences ɜy ɨy ɜw ɨw changed into i i u u.

    NOTE, another way of saying this (more in line with how i write now) is that /ē ī/ merged into a new short /i/ before hiatus; the /u/ shift may have been separate.

    the resulting sequences are treated as new coarticulates... also, the high vowel probably disappears before some consonants, creating clusters such as /kl/, and these can contrast labialization.

13. The vowel ɨ became u before a labial in a closed syllable, or after a labialized consonant. That is, ʷɨ > ʷu.
14. The vowels a ɜ before a labial in a closed syllable very likely became shifted to o o around this time, because it cannot be accounted for by the later shift that does almost the same thing. By contrast, /ʷa ʷɜ/ did NOT shift to /o/.
15. The vowel ɨ became i when adjacent of a palatal in either direction (though in compounds, this did not always apply progressively).
16. Possibly at this time kʷ shifted to p.
17. Remaining B shifted to Ø (not /g/).
18. The labialized consonants w mbʷ kʷ hʷ ŋġʷ delabialized to Ø mb k h ŋġ before /i/.

    try saying püa etc and remmeber syllable integrity ... earlier, there was a rule a bit below this one that would have these going to /w/, essentially deleting the /i/.

19. The clusters tl tr shifted to kl kr. Other similar shifts of clusters almost certainly followed. The cluster kt shifted to k (not /ḳ/).
20. Before a vowel, the sequences ti ki shifted to č.
21. The clusters kl kr merged as kʕ. Thus, in handheld object prefixes, /ətr/ > /oḳ/.

    The Play word /see/ requires that ml also shift to mʕ (and later to /mb/). If this shift does not happen, the Play word would end up as /seu/ instead.

22. Syllabic nasals were created: im ɨm um > ṁ, ɨn > ṅ, and ɨŋ > ŋ̇.

    Possibly the sequences mbṁ ndṅ ŋġŋ̇ and others like them shift to simple syllabics ṁ ṅ ŋ̇.

23. The palatal approximant y was changed to ʕ in all positions. Between vowels, this became silent (Ø) but used to show diphthongs' separation. Thus the palatalized consonants pʲ mbʲ mʲ became the pharyngealized clusters pʕ mbʕ mʕ .
24. The postalveolar nasal ň changed to ŋ. However, ndʲ here became nd, not something such as /nġ/.
25. The voiced approximants λ ř r changed to y.

    It is possible that there was a shift of /iy/ > /ig/, which would account for the name of the religion beginning with yìga- in Tapilula. Alternatively this could be due to some sort of syllable split, such that /yiya/ came to be seen as /yii/ + /a/, and all double vowels collapsed. The /g/ would be a regular insertion to break up hiatus in such a case.

26. The vowel sequences aɨ ɨa changed to ɜu ʷɜ. Note that the /ɨa/ covers all such sequences that did not previously have an intervening /y/ sound ... therefore, this was probably allophonically a back vowel even if it were not always rounded.
27. The sequences ʷa ʷɜ changed to ʷo.

    It needs to be that /kʷ/ > /pʷ/ > /p/ had taken place by this time, since otherwise there would be no contrast of /p/ vs /pʷ/.

    This also implies pk (since it was really kʷk) shifting to pp. This does not rule out a new /pk/ arising from new compounds.

28. yʷ čʷ > y č, thus the word for hand is just "yò". This did not affect any $ that was just /s/ by this time.
29. n$ʷ (when not /nčʷ/ as above) shifted to ntʷ (not to /ndʷ/). Likewise the sequence m$ shifted to mp.
30. The sequences au ɜu now both become o. ai ɜi changed to ɛ unconditionally.
31. The true mid vowel ɜ, in a syllable adjacent to an /ɛ/ or /o/, changed to match that vowel. Some double schwa words also changed, e.g. mɜčɜ > močo "fire".
32. The "clear" labialized consonants kʷ ŋʷ became the rounded bilabials pʷ mʷ. (There was little or no /kʕʷ/.)
33. The pharyngealized nasals mʕ mʕʷ shifted to mb mbʷ.

    This may incorporate early shifts like /mh/ at morpheme boundaries. Thus for example, hàga "fairy" could actually be /hàmba/.

34. The voiceless fricatives h hʷ became voiced to g w unconditionally.
35. Remaining ɜ ɨ in open syllables shift to ə i, respectively. This new schwa vowel is IPA /ɨ/but is spelled "ə" or "y", since it is never ambiguous with IPA /j/.
36. The voiceless affricate č shifted to h. (This probably had shifted to /s/ during depalatalization. if not, /hy/ etc also switched.)

    Gold /hìga ~ hìa/ showing up as /hà/, even if only in one word, relies on this assumption, and there are no words that rely on the sequence /hia/ existing in pre-Gold as it has no specific reflexes.

37. Syllables *preceding* heavy syllables became LOW tone.

Thus the final phonology was:

Rounded bilabials:     pʷ      mʷ  mbʷ mhʷ~hʷ  w
Spread bilabials:      p   pʕ  m   mb  mf ~f  (Ø)         
Alveolars:             t       n   nd          l
Rounded alveolars:     tʷ      nʷ  ndʷ     
Velars:                k   ḳ   ŋ   ŋġ      h   g

The (f) was pronounced [mʰ] in some dialects. Pharyngealization consonants can be spelled with voiced stop letters.

Dreamlandic changes

This change occurred in the Lenian languages but is not on that page because the dictionary is not synced properly with the MRCA.

1. The nasals mʷ m n nʷ ŋ become prenasalized voiceless stops mpʷ mp nt ntʷ ŋk when facing a labial.

Proto-Dreamlandic retains the final primordial nasals even before voiceless consonants, and also gains new voiceless prenasal clusters from plain nasals before labials. As of 07:07, 1 May 2022 (PDT)~ this error is fixed in the dictionary for word-medial clusters (that is, some /p t k/ has been replaced with /mp nt ŋk/), but not for word-final codas. This is because the codas are not indexed in the MRCA dictionary either; they are only derived when a word needs to appear in Dreamlandic.

MRCA to Nuclear Tapilula (700 BC to 0 AD)

1. The nasals m mʷ became f hʷ before a vowel facing another labial.
2. Possibly, g shifted to Ø at the head of a closed syllable, at least when in initial position. This could explain the otherwise unpreditcable g~Ø variation in Gold and Andanese, but not why they sometimes disagree with each other, or why it also drops out in medial position in a few words (unless these are all compounds). Therefore, even if this rule is assumed, analogy is required. In Dreamlandic, the reflexes of /g/ and /Ø/ are identical, so no explanation is needed.
3. All syllable codas were deleted. The resulting ephemeral voiced stop ġ shifted to the fricative g.
4. The glottalized stops pʕ pʕʷ kʕ changed to b bʷ ḳ unconditionally.
5. The rounded labials pʷ bʷ mʷ became plain labials p b m .

The consonant inventory was:

Rounded bilabials:                     hʷ  w
Spread bilabials:      p       m   b   f  (Ø)
Alveolars:             t       n   d       l
Rounded alveolars:     tʷ      nʷ  dʷ         
Velars:                k   ḳ   ŋ       h   g

Post-Tapilula changes

Tapilula to Gold (year 1900)

Alternate names: Medium, Walking Girls, Wolf in Wool, Soft Hands, Slingshot, Broken Shields, Ukieipi, G̣ʷidiʕìləs

Although Gold is the proper name of the language, the Gold party survives for thousands of years, and therefore these names can be used to give a more precise definition of the stage of the language being referred to.

Note on politics: It is not clear whether the Tapilula language split apart in 0 AD or in 500 AD. If 500, the various Subumpamese languages are almost independent branches since they begin diverging not long after 600 AD. However, Sub and Gold share much more in common than either of them does with Andanese, despite being supposedly only 100 years closer together. So perhaps the true date is 0 after all.

Tapilula to Gold (proper)

1. The aspirated velar stop k became č before the vowel /i/. If another vowel followed, the /i/ disappeared. This happened even if the /i/ was accented.
2. When a "velaroid" consonant (/k ḳ ŋ h g l/) followed an accented high tone vowel, the vowel metathesized, leaving a closed syllable. Thus, for example, /àli/ > /ail/. These closed syllables were all high-toned, and are thus written without tone marks. Thus, for example, aa implies àa. Later, daughter languages introduced tone contrasts and independent sequences.
3. A schwa before another vowel in any syllable disappeared. Thus əa əe əi əo əu əə shifted to a e i o u ə. This happened in both open and closed syllables.
4. The sequences iu and ui shifted to ə̄.
5. The double-vowel sequences aa ee ii oo uu əə shifted to the single vowels a e i o u ə in closed syllables only.
6. The sequences ii uu əə (which now occurred only in open syllables) shifted to əi əu ə. This means that the /ə/ and "same vowel" infixes had switched places in some constructions, but this did not lead to confusion, at least in the four-vowel languages such as Play, because the infixes typically did not occur in their free forms.
7. The sequences ie uo shifted to i u in open syllables only.
8. The sequences ai ei oi merged as ei; the sequences au eu ou merged as ou.

   NOTE ON POLITICS: SUBUMPAMESE CAN BE CONSIDERED TO BREAK OFF HERE, AS THIS IS THE LAST OF THE SHARED CHANGES. (that is, there were only A FEW of them).

9. The sequences ea eə shifted to ee; meanwhile, oa oə became oo. Then, aə shifted to aa. Thus, the sequences /ee aa oo/ once again appeared in both open and closed syllables. Note, however, that much inherited /ea oa/ had participated in grammatical alternations with /əa/, which had become a simple /a/ by this time, and this is the form that was usually generalized.
10. The sequences ia ie io iə shifted to ī . Then ua ue uo uə shifted to ū.
11. In absolute final position, syllable-final ŋ changed to n. (But see below.)
12. Accented vowel-initial syllables gained a pharyngeal ʕ as an onset. Then the clusters nʕ kʕ shifted to g ḳ.
13. After long vowels, all consonants became voiced. Also, consonants occurring after initial vowels also became voiced. This created the new consonants v ǯ . Thus, final -h in words like hʷīh became -g. However, analogy made it so that the change was confined to open syllables in most words. This sound change did not affect diphthongs. There was no voiced velar stop, as all four velars simply shifted to fricatives.
14. After initial unstressed /u/, all consonants other than palatals became labialized. This change extended even to clusters. Because of the voicing rule, however, all of these consonants were voiced. bʷ vʷ merged as w.
15. In word-initial position, the six vowels a e i o u ə were deleted to Ø unconditionally.

    A few words began with double vowels, but these were grammatical alternants of V-/g/-V, V: (that is, inherited long vowels), and /g/-V-(g)-V pairlets, meaning that in these few words, the vowel deletion could affect both vowels, only the outermost vowel, or neither of the two vowels. Then, because of this same analogy, the words that now began with g came to be pronounced with Ø onset when following a word that ended with a consonant, and for some speakers also in isolation, but never when following a word ending in a vowel. In both Leaper and Play, this shift was quickly undone, as it had never been phonemic. In Thaoa, it may have become firm, but with the expense of now having no words beginning in /g/.

    Sometimes root-initial vowels were retained due to classifier prefixes; here, too, analogy played an important role. For example, the food item prefix mi- cleaved onto the root, and ceased to be seen as a classifier prefix, meaning that these new roots began with consonants after all. This was helped by the fact that only a few such roots began with vowels, even including those that had come to be pronounced with /g/.

16. All schwas and diphthongs became low tone.
17. Labialized consonants lost their labialization when occuring after another labial or labialized consonant.
18. After a stressed syllable, intervocalic ʕ ʕʷ became g gʷ. This is due to reanalysis, not a true sound change.
19. The glottal fricatives h hʷ became velar; there was no spelling change.

NOTE ON POLITICS: THIS IS WHERE THAOA & TARYTE SPLIT OFF (YR 1085 AD)

The consonant inventory was:

                       BASIC                         LABIALIZED

Bilabials:             p   b   m   f   v                     mʷ      w  
Alveolars:             t   d   n       l             tʷ  dʷ  nʷ            
Postalveolars:         č   ǯ           y                       
Velars:                k       ŋ   h   g   ḳ                 ŋʷ  hʷ  gʷ

The vowel inventory was

Short vowels:          a  e  i  o  u  ə
Long vowels:           ā  ē  ī  ō  ū 
Falling diphthongs:      ae ei ao ou
                            əi    əu

The long vowels /ā ē ō/ can be spelled aa ee oo, but the high vowels /ī ū/ are usually not, because /i u/ before another vowel would indicate a glide.

There were also eo oe in historical compounds. These would not have occurred in individual roots because of ancient vowel harmony rules.

Nuclear Gold

Gold's unconditional shift of /f v b/ > /s d d/ sets it apart from all related languages. Note that it requires /mb/ > /nd/ as well but yet /ṁb/ (syllabic) remained.

1. All remaining e o shifted to a. This included elements of diphthongs. Resulting /aa/ was spelled as /ā/; thus, ae ao eo oe ee oo all shifted to ā. Thus there were only four vowels. In some cases, a short vowel was formed instead due to grammatical analogy, though unlike the /ea oa/ > /əa/ shift above, there was no etymological basis for this shift.
2. The velar fricatives h g were rounded to hʷ gʷ before /u/. The labiodental /f/ did not shift, and by this time was likely already [fʲ] to maximize distinction. If /v/ existed, it did not shift either.
3. The plain labials f v b shifted to s d d. This happened by intermediate palatalization, and therefore the sequences /fj vj bj/, which were very rare, also shifted to /s d d/. The sequence /fʷ/ remained, often spelled ħʷ, but this quickly merged with the existing /hʷ/.

   This apparently includes a shift of bh to dh, meaning that the devoicing of aspirates did not happen until the Gold language split apart.

4. In absolute final position, syllable-final h changed to s.
5. The postalveolar affricates č ǯ shifted to š ž.

The consonant inventory was:

Bilabials:             p           m      (ʕ)  w   mʷ            
Alveolars:             t       d   n   s       l   nʷ      tʷ  dʷ
Postalveolars:                         š   ž   y                                  
Velars:                k   ḳ       ŋ   h   g   gʷ  ŋʷ  hʷ  kʷ      ḳʷ

The vowel inventory was

Short vowels:      a  i  u  ə
Long vowels:       ā  ī  ū
Diphthongs:          ai au
                     əi əu

The consonants kʷ ḳʷ only occur as the realizations of clusters like /k/ + /gʷ/, and thus never occurred word-initially, whereas the labialized nasals occurred only word-initially (or morpheme-initially).

Notes on Play and Leaper

Note that both Play and Leaper often show reflexes implying a shift of /hi gi/ > /s d/ between vowels, but in different places; this change was grammatically conditioned and did not occur within the history of the Gold parent language. This shift was later analogized with the preexisting true sound change of f v > s d to replace intervocalic /s si/ of any origin with /d d/ when the preceding vowel was originally long. When a vowel did not follow, there was no shift. The shift also did not occur after a vowel that became long in Leaper but was a diphthong in Gold.

Note that this implies that there was no shift of /sʲ/ to /s/, and therefore likely no shift of /lʲ/ > /l/.

Play's reflex of /ll/ is /ww/ (the document is not on this wiki, unlike almost all others, so I have to put the notes here).

Gold (1900) to Khulls (4700)

Alternate names: Kuroras, Leaper

Alternate names: Khulls

(Dummy edit link)

This language was originally spoken in AlphaLeap.

1. The velar fricatives h hʷ came to be spelled x xʷ.

   As /tanči/ "wine" demonstrates, a preceding coda /n/ did *not* assimilate to the /x/. This also implies nx shifted to nt in Leaper but not in Play; it is possible that the shift was pre-Gold, but was then undone in Play due to its close attention to morphology, but it would make more sense to have the shift appear after the separation.

2. tʷ dʷ nʷ shifted to tl dl nl.
3. When not occurring after a labialized consonant, the vowels ŭ ù ū shifted to ɜ̆ ɜ̀ ɜ̄. (This is spelled differently from schwa to ease confusion.) This was a low back vowel comparable to IPA [ɤ].
4. The sequences ə əi əu (all syllables with inherited /ə/ were toneless) shifted to ʉ ɜi ɜu.
5. In a closed syllable, the new ʉ vowel disappeared and created a syllabic consonant. In an open syllable, ʉ changed to ʷ, thus labializing the preceding consonant and then disappearing. Where /ʉ/ collapsed, stress shifted syllables to the nearest adjacent one. This tone was mid-tone (ă), which was sometimes called the low tone since it behaved as such when joined to any other morpheme that carried stress. Thus all morphemes that had once contained a schwa came to be pronounced entirely with low tones.

   Note that any syllabic ṅ formed here always assimilated to a following consonant because they arose from a non-syllabic nasal, which had already been assimilated to a following consonant. By contrast, the primordial syllabic nasals /ṁ ṅ ŋ̇/ still did not assimilate, and thus words like /mṅpà/ "to ask" still existed.

6. Sequences like aʕa became pharyngealized vowels; these could still have tones, but later all pharyngealized tones merged with each other except for sandhi effects. Pharyngealized vowels are spelled â ; though there is only one pharyngealized surface tone, pharyngealized vowels exhibited different sandhi effects depending on their origin, and this is not reflected in the Romanization. Note that /iʕV/ did not create pharyngealization, but /uʕV əʕV/ did, and both caused labialization.
7. The cluster sg shifted to x. This is actually a tone shift, since it had been [x] all along but was previously abound to a high tone.
8. The clusters pʷn kʷn shifted to pʷt kʷt.

   Note that this shift, needed to process the placename Gatupəna, is unsatisfactory, and would not help explain what would happen to the ejective cluster /ḳʷn/. Syllabic nasals are a possible answer, since the only clusters considered unwieldy are stop+nasal (so /pʷl/ etc were fine). This would produce the forbidden clusters /pṁ tṅ kŋ̇/ and these would probably lose the stop and become high tones.

9. After a high tone, the voiced stop d shifted to ṭ. This includes a shift of dl to ṭl after a high tone as well.
10. The clusters mh nh ŋh dh became mp nt ŋk t, except that nč replaces /nt/ before any /i/. These clusters were often morphologically /s/ + a voiced consonant, but the [h] pronunciation is actually the more archaic one.
11. After the vowel [u] (any tone, any length), k ḳ in a syllable coda became coarticulated labial-velar stops kp ḳṗ . This change also took place after the /ʷ/ that had replaced earlier schwa, since this was behaving as an allophone of /u/. Likewise, it took place after /au/, but not /ɜ̄/, even though both ended up as /ō/ later on.
12. After a syllabic nasal, the final stops k ḳ (which was the only ones that did occur) changed to match the position of the nasal. However, these were written with the letters for "kp ḳṗ".
13. The voiced coronal stops d dʲ dʷ became r ž gʷ. The sequence rl (always from Gold /dʷ/) became ll, which in word-initial position then changed to a simple l.
14. In word-initial position, r shifted to l.
15. Labialized coronals became velar.
16. Labialization disappeared before any syllabic consonant.

    This shift makes it possible to interpret the syllabic consonants as sequences of short low-tone /u/ + C. Previously, this would have failed because there was a contrast between /ʷC/ and /ʷuC/.

17. The sequences mmṡ ŋŋṡ shifted to mpṡ ŋkṡ.
18. The sequences gp gṗ shifted to kp ḳṗ.
19. In unstressed position after a vowel, the syllabic consonants ṁ ṅ ŋ̇ ḷ ṡ shifted to plain consonants m n ŋ l s. Thus unstressed closed syllables were created.
20. Final ḳ raised the preceding vowel to a high tone à (á if it was long) and then disappeared, though it left an allophonic glottal stop in some positions. Then, final k disappeared and changed the preceding vowel to the long high tone á. The surface tone change did not apply to pharyngealized vowels, but the sandhi effects did. Thus there were two pharyngealized tones .... both pronounced the same, but with different effects on surrounding unstressed syllables.
21. The labialized fricative šʷ became ħʷ . The ħ is a spelling convention to distinguish it from /x/. The cluster gʷš (which had always been [gʷšʷ] because of syllable metrics) most likely became ħʷ as well, not /xʷ/. The /sg/ > /x/ shift above is different from this because /sg/ had never been phonetically realized as such.

    NOTE ON POLITICS: The Proto-Moonshine language breaks off here. (Year 3958) The tropical survivor language must have also broken off shortly afterward; what remains is for the language of the ruling class of Baeba Swamp.

Post-Moonshine changes

All of these changes take place in just 800 years, despite the list being nearly as long as that for the preceding 2,000 years.

1. The diphthongs ai au shifted to ē ō unconditionally. Then aiʕ auʕ became ê ô. Note that the sequence /uʕ/ was distinct from a coda /ʕʷ/.
2. The diphthongs ɜi ɜu shifted to ĕ ū unconditionally. If pharyngealized forms existed, they followed the rule above.
3. The sequences ya yɜ (on all tones) shifted to ye of the same tone. Later, all /ye/ became /e/, but this was not phonemic because of intervening consonant changes. Note that this does NOT include /yau yɜu/, which had escaped the change by shifting to /yō yū/. This /ō/ is distinct from the one that forms below.
4. The mid vowels ɜ̆ ɜ̀ ɜ̄ shifted to ŏ ò ō unconditionally. Likewise, any remaining unstressed ɜ became o.
5. The velar-palatal sequences ky ḳy ŋy hy xy gy shifted to č č n̆ š š ž.
6. Nasal consonants followed by /y/ hardened to prenasals: my n̆y became mby n̆ǯy . (This includes the reflexes of /ny/ and /ŋy/.) These later became stops.
7. After a high tone, the fricatives x xʷ ħ ħʷ shifted to k kʷ q qʷ.
8. sl>q, which is an allophone of /h/. This shift is essentially a restating of a consequence of the above, since there never was an /sl/, but only a coarticulated /lh/ cluster. Likewise, posttonic ħ ħʷ (spelling here used for emphasis) shifted to q qʷ.
9. The voiceless bilabial stops p pʷ shifted to h hʷ except after a high tone. The plain /p/ had a brief intermediate of /ɸ/ but this stage lasted mere years before shifting to /h/. The labialized stop shifted directly. This shift excludes /kp/ and any other context in which the stops were part of a cluster; note that since the high tone always ended in a glottal stop, this environment can be considered to be a cluster as well.

   NOTE THAT THE SPELLING OF /h/ as ħ is for clarity only, because in many names, /x/ is spelled with the plain "h".

10. The sequences ly hy (the latter only from /py/) shifted to λ š.
11. The clusters ml nl changed to mbl ndl, thus restoring voiced stops to a marginal phonemic position.
12. The labialized consonants mʷ ŋʷ changed to mbʷ ŋġʷ .
13. The remaining clusters tl ṭl shifted to `l . This is a plain /l/ but makes the preceding vowel high-toned.
14. The sound /l/ disappeared after any stop, even over a morpheme boundary.
15. The voiced prenasals mbʷ mb nd nǯ ŋġʷ shifted to plain voiced stops bʷ b d ǯ ġʷ. There may have been an extremely rare plain voiced velar stop, arising only from the sequence /ŋ̇l/, which would have changed to /ŋġl/, then to /ŋġ/, and finally to /ġ/. Note that the original velar nasal must be syllabic for it to occur before another consonant.
16. The coarticulated stops kp ḳṗ shifted to p ṗ. (If there ever was a voiced /ġb/, it too would shift.)
17. The voiced velar fricative g came to be pronounced as a voiced stop ġ asfter a high tone. This was allophonic, and occurred at least a thousand years after the shift of /d/ > /ṭ/. Thus the two are not connected and this newer shift is not represented in the script.
18. Probably the sequences ăʕʷ ĕʕʷ ĭʕʷ ŏʕʷ ŭʕʷ shifted to ô ô û ô û, thus finally eliminating diphthongs from the language, even those that had arisen from VCV sequences. But it is possible (remember the ʕʕ>ʔ rule) that pharyngealization was not pronounced in this position even in a closed syllable (it had been eliminated for sure in an open syllable). It is also possible that the diphthongs simply remained and that /ʕʷ/ was seen as a consonant.

Note that the only /y/ is before /i/ and unlabialized /u/, the latter of which was rare. The only other clusters in the language had initial elements unmarked for place of articulation, possibly aside from a few marginal holdovers across morpheme boundaries involving inherited syllabic nasals.

Thus the final consonant inventory was

Rounded bilabials:       pʷ  ṗʷ  bʷ      hʷ          w
Spread bilabials:        p   ṗ   b   m   
Alveolars:               t   ṭ   d   n   s   r   l
Postalveolars:           č       ǯ       š   ž  (λ)  y
Velars:                  k   ḳ       ŋ   x   g
Labiovelars:             kʷ  ḳʷ  ġʷ      xʷ  gʷ
Postvelars:              q               h       ʕ
Labialized postvelars:   qʷ

And the vowels /a e i o u/ on six tones: à ă ā á â a͆, where the last two differ in sandhi effects only.

All five vowels are unrounded except when following a labialized consonant. Because /u/ almost always follows a labialized consonant, its unrounded form is very rare unless analyzed as /Ø/. This can be spelled /ʉ/.

A rare palatal lateral λ (IPA /ʎ/) can be added, which occurs only in environments where /y/ can also occur. Unlike the other five palatal consonants /č ǯ š ž y/, however, it is entirely of secondary origin, arising entirely from the sequence /ly/, and it cannot contrast with the sequence /ly/, even over a morpheme boundary.

The w is always phrayngealized as the onset of a stressed syllable.

The sounds q̇ q̇ʷ z are not part of Leaper's phonology, but have distinct letters in the script because they occur in loans from Qaš and different speakers replace them with different Leaper phonemes. Specifically, q̇ q̇ʷ can be replaced either with /ḳ ḳʷ/ or with /q qʷ/, and some speakers even pronounce the original phonemes after all; likewise z can be replaced either with /ž/ or with /s/, and some speakers pronounce the /z/ as in Qaš as well.

Late dialectal changes

One late change might be the deletions of all word-initial g, thus creating many vowel-initial words. Then, old word-initial e- deletes, but the new e- is unaffected. This change is difficult to explain, even assuming the two /e/'s had different sounds (because the old one was always preceded by /j/ and the new one never was).

Macro-Pabap languages

See Macro-Pabap languages for all non-Babakiam stem groups.

Gold (1900) to Play (4100)

The Play language evolved from the Soft Hands dialect of Gold, also known as Wolf in Wool, Broken Shields, and perhaps at least one other name. It drove out the Lazy Palms language and took relatively few loanwords. There were also several other languages spoken in this territory, including one language spoken by Star immigrants, probably a branch of Amade.

Wolf in Wool had not yet evolved its characteristic sound, so the relative scarcity of loanwords was not due to the acoustics of the language, but rather a cultural identification with the new language being imported from overseas. Any loans that were taken in had /e o/ shifting to /ə/ for the entire time period of this language, though /ē ō/ may have been borrowed as /əi əu/ or /ai au/ or either.

1. At the end of a syllable, the pharyngeal fricative ʕ disappeared and changed the previous vowel to a high tone. It also voiced the following consonant.
2. Syllable-final k ḳ ŋ changed to kʷ ḳʷ ŋʷ.
3. Feeding on the above change, in compounds, if the final consonant was one of /kʷ ḳʷ/ and the first consonant of the next morpheme was one of the velars k ḳ h ŋ, it also became labiovelar. Thus for example /kk/ > /kʷkʷ/ or /kʷ:/. It did not happen for other consonants. Prenasals did not shift; later, the cluster /ŋʷk/ becomes /mk/, which is pronounced as spelled but later becomes [ŋk], [mpt], etc depending on dialect.
4. In initial position, the labialized coronals tʷ dʷ nʷ shifted to t d n. Elsewhere, even in clusters, they decoupled to the sequences tu du nu.
5. The bilabial approximant w changed to v (in internal reconstructions, also spelled "β") before a vowel.
6. Then l lʷ both became w (not */v/) in all positions although it retained a rhotic allophone. The distinction between this new /w/ sound and the one that had just changed to /v/ is important later on, as it keeps sequences like /ʕl/ from being corrupted to /ʕʷ~gʷ/ and then on to /v/, /b/, and /p/. Rather, /l/ stays as /w/.

   Notably, the sequence sl (which was pronounced as IPA [hl] or for some speakers [ɬ]) shifted here to sw, and did not become */hʷ/ or */f/. That is, it behaved as the sequence that it was morphologically, instead of sliding with the phonetics into a new single consonant.

   NOTE ON POLITICS: Proto-Highland Poswa breaks off here.

7. The labiovelar consonants kʷ ḳʷ hʷ gʷ became p ṗ f v unconditionally. This includes sequences like /kʷl/, despite the precedent set by /sl/ above, because in this case, /kʷl/ was already [kʷ] at the surface level in the proto-language.
8. Sequences of two vowels in which the first vowel was i or u became rising diphthongs. Then all clusters of a consonant followed by a semivowel came to be pronounced as coarticulated single consonants. Thus pua became pʷa, pia became pʲa, and so on.
9. Stressed syllabic nasals were opened to sequences containing a schwa.
10. The voiced fricative g assimilated to a neighboring glide /j/ or /w/, thus creating sequences of /jj/ and /ww/. The shift thus was gj jg gw wg > jj jj ww ww. This includes g after /ī/ and /ū/.
11. The voiced fricatives d dh g became silent between vowels and occasionally in initial position (due to compounding).

    When I wrote this, there was no /ž/ in the language at this stage, and so it is possible that ž also shifts to Ø.

    NOTE ON POLITICS: This time period is around 3100 AD, near the beginning of the "Time of Happiness" (Yeisu Kasu: 3138 - 3302 AD). The branches of the language that fork off from mainline Bābākiam in 3138 all die out, and therefore all of their names in the history are written in Babakiam, but they could be revived as minor local languages, and there would be quite a lot of them.

12. A voiced consonant in a cluster after /p/ or /s/ changed briefly to ʕ and then disappeared.

    This shift is responsible for important consequences in verb morphology in Poswa more than 5000 years later. Note that the inherited clusters gh hg had been merged as h already in Gold; /hg/ was morphologically equivalent to /sg/, which explains why /sg/ shows up in Play as š instead of s like the others. Lastly, this shift explains why the Play toponym Fanašasa corresponds to Leaper Xʷanaxanta.

13. The voiced fricatives v z ž g changed to b d ǯ ġ before a high tone. Unlike other languages, Play considered the long vowels to be high tones here.

    This is how Play does /g/ > /k/ even though /g/ was a fricative. Note however that in hypothetical words like /vuau/, where a /d/ dropped out, the initial /v/ was part of a separate syllable, not stressed, and so did not shift to /b/.

14. The post-velar fricative consonants ħ ʕ, which had been developing labial compression, changed unconditionally to f v.
15. The velar fricatives h g were fronted to š ž unconditionally. šʲ žʲ became š ž. This includes the /čʲ/ sequence, which had long ago become [šʲ] but was maintained in spelling because of its importantly distinct grammatical behavior.

    Importantly, this shift included conditions in hiatus ("holes" in Play terminology), so that čiva became čua.

16. The labialized voiced stops bʷ dʷ ǯʷ ġʷ changed to b.
17. The palatalized voiced stops bʲ dʲ ǯʲ ġʲ changed to ǯ.
18. Any remaining voiced stops b d ǯ ġ changed unconditionally to p t č k (except when in clusters).
19. The voiced fricative žʷ changed to v.
20. Tones were eliminated. However the stress accent (nouns on the penultimate syllable, verbs on the ultimate) remained and became regularized.
21. The voiced stops d ǯ ġ (now found only in clusters) changed to n nʲ ŋ unconditionally.
22. Remaining v changed to b.
23. Remaining z changed to s.
24. Newly created vowel sequences beginning with i or u collapsed into rising diphthongs, thus creating a new series of palatalized and labialized consonants.

    This same shift happened twice but many words missed by the first change were captured by this change. Note, however, that the reflex of /buya/ is still /buya/; it did not become /bʷia/ and then /bia/.

25. The labialized consonants bʷ žʷ changed to b unconditionally. (Despite the fact that a nearly identical sound change had occurred only shortly before this one, this rule was very common in verb forms that were created by the shift of /bua/ > /bʷa/ > /ba/, and likewise for other vowels.)
26. The palatalized consonants bʲ žʲ changed to ž unconditionally. (The above shift also applies here; many verbs underwent a shift of /bia/ > /bʲa/ > /ža/.) This shift did not apply to words such as bivu, from earlier /buivu/, because the /i/ in this word was not [ʲ] but still a true /i/.
27. A schwa ə in a word in which the following syllable had /a/ changed also to a. Note that this is the only vowel change in the entire history of the language going back 3500 years, even before the Gold language, except for a few diphthongizations such as /ua/ > /wa/. However, the vowel system became very unstable in the succeeding period as the language developed into Poswa and Pabappa.
28. The stress was shifted to the first syllable in all words.

Play (4100) to Poswa (8700)

Alternate names: Cherry, Blossom, Pavopa, Wupupa

NOTE that this section is outdented for prominence. Poswa is a daughter language of Play, not of Gold.

1. Long vowels in initial syllables became double: ā ī ū became aa ii uu. This shift did not happen if the long vowel was supported by another following vowel. Meanwhile double vowels in final syllables became long: aa ii uu shifted to ā ī ū. Medial syllables followed morpheme boundaries.
2. The double vowels ii uu became ʲi ʷu in all positions.
3. Any bʷ žʷ created by the previous shift changed to b. Likewise any bʲ žʲ changed to ž.

   NOTE THAT THIS IS AN EXACT REPETITION of a shift that occurred late in the history of Play. It is repeated here because new /ʷ ʲ/ had just been created. It could be said that the shift was in continuous operation for several hundred years. It may have applied twice in a few words, such as buivu > bivu > žuu. It did not generally occur over morpheme boundaries because the rebracketing that created vowel-initial morphemes had not yet taken place.

4. Between two consonants in a single syllable, the diphthongs au ai əu əi changed to o ae u e. iu ui ii uu > y y i u. The change was bypassed whenever a consonant cluster was frontloaded onto the next syllable, however.

   Although /ae/ behaved like a regular vowel in this shift, it was still not possible to produce syllables like */paer/, as it would be later. This is because the shift of /u i/ > /r l/ had not yet happened.

5. ā aa changed to aba in all positions.
6. At the beginning of a syllable and after /p m/, the semivowels w j shifted to r l. Syllable-finally, the shift also occurred in some positions but was allophonic and remained so for thousands of years.

   This is the point at which the /ae/ diphthong starts behaving differently than the others; in particular, there was no *syllable-final [al], even allophonically before another /l/. All instances of the [al] sequences were open syllables followed by a singleton /l/ onset.

7. The medial clusters pt mt mn shifted to tt nt nn. Then pk mk mŋ became pt mpt mn . ms mš became mps mpš.

   The shift of /pt mt mn/ > /tt nt nn/ might need to be back-dated because it affects the way vowels were compressed. For example, Play /pk/ front-loaded onto a following syllable, but Play /pt/ did not. This would make more sense if the contrast was either /pk/ vs /tt/ or /pt/ vs /tt/.

8. The cluster sf changed to ff.

   In most words where an /sf/ cluster might be expected, only a single /f/ was found due to a much earlier shift that occurred in the Gold language. This had been maintained through analogy. Likewise, where one might expect /pf/, there was often just a /p/. The words where these clusters did occur were newly coined compounds.

9. The labialized alveolar stop tʷ shifted to pʷ in word-initial position or after one of /r l s/, and to pt between vowels.
10. The labialized consonants šʷ sʷ nʷ shifted to pš ps bʷ. There was also a rare word-initial /bʷ/, from earlier /bū/ > /bu/ > /bʷ/. This shift also includes sšʷ ssʷ shifting to ppš pps.
11. The consonant clusters mr ml shifted to br bl unconditionally.
12. In initial position before a vowel, the voiceless labial fricatives f fʷ changed to w. It also happened often to a word-internal /f fʷ/ preceded by a /w/ of any origin, but note that the sequence /fVf/ only appeared in words that were originally compounds. This shift did not affect fʲ. NOTE: Shift affecting internal /fʷ/, not just /f/, added late to cover for situations in Pabappa.
13. After a labialized consonant (except /w/), the schwa vowel y changed to u.
14. Labialized consonants lost their labialization when they occurred before /u/.
15. The vowels i u changed to e o in closed syllables.
16. The vowel y in closed syllables changed to either i or u depending on the other vowel in the root. The default choice was /u/, unless it followed a /w/. /i/ appeared only when it followed a /w/ or was in a word in which an /i/ or /e/ was in an adjacent syllable and that was the only other vowel in the word.
17. The sequences wu wo changed to wi we.
18. Then, p s disappeared before nasals and sporadically in stem-final position due to back-formation from plurals.

    IMPORTANTLY, it also disappears before /b/ in many, if not all, words. Apparent exceptions can be explained by later compounding; note that they mostly involve /p bl/ --> /bbl/ where /bl/ would be expected, but seldom /p b/ > /bb/ where /b/ would be expected.

19. Unaccented i changed to e unless the accented syllable of the word contained an /i/.
20. Then unaccented y changed to i unless the syllable ended in a labial or the accented syllable of the word contained a /y/.

    NOTE ON POLITICS: Politically, the proto-Poswobs became independent here (5547), but the language remained unified through physical contact for another 450 years or so.

21. The consonant cluster ŋʲ assimilated to nʲ.
22. Then unaccented u changed to y except when the syllable ended in a labial or the accented syllable also contained a /u/.
23. Before a vowel, unaccented y yb changed to u.
24. Before a vowel, unaccented a ab changed to i.
25. Next, i changed to y if the next syllable had /u/.
26. Any l r became devoiced after a voiceless stop (probably only /p/ exists; so the rule would be pl pl shifting to prʰ plʰ).
27. The palatalized consonants pʲ mʲ sʲ lʲ rʲ changed to f v š ž b before a vowel. šʲ žʲ bʲ also became š ž b. The rare sequence rrʲ most likely became bb, not *rb. However, mmʲ shifted to mv, not *vv.
28. The voiceless allophones of /lʲ rʲ/ shifted to the voiceless counterparts of the plain forms; that is, plʲ prʲ shifted to pš pp, matching the /ž b/ of the plain forms. This explains why the rare sequences /pž pb/ (from compounds; never inherited from Play) did not also become voiceless.
29. The sequence ppp (the reflex of earlier /pprʲ/) shifted to pp. But /ppš/ remained.

    NOTE ON POLITICS: (year 6000; THIS IS WHERE PABAPPA BREAKS OFF)

30. In stressed syllables, in only a few words, ol ul or ur changed to we wi wa wa. The conditioning environment was that the syllables had to be unstressed and have only one consonant before them; in other words, they occurred in compounds only, in a syllable which would be stressed if it weren't a compound.
31. In an accented syllable, ar shifted to o, except after /w/.
32. Simultaneously, syllable-final r in most words changed to bʷ. Aquatically, it was f/v before coronals, b in absolute final position, stays r before šž + labials + velars (but really pronounced as /w/). rl > vl. rr > rw. Sometimes au+labial > o even so, no particular rule.
33. The surviving final r changed to vʷ fʷ before labials.
34. fʷ (and probably ffʷ) became w before a vowel (that is, everywhere except before a labial).

    Note that /ffʷ/ defies the pattern set by /ppf/ because in the latter case there was no assimilation, but /ffʷ/ had both members as bilabial, meaning it was really /fʷ:/ all along.

35. In words not affected by the previous shift (mostly due to grammatical analogy), syllable-final ar yr shifted to o, and er ir shifted to u.
36. In unstressed syllables, the sequences el il merged as i and or ur merged as u.
37. Unstressed ol ul became e before a consonant or at the end of a word.
38. i y shifted to u before a labial in a closed syllable, or before a syllable beginning with a labial cluster (in this case, even /pʷ/ etc counted as "clusters"). Thus for example /tipwu/ > /tupwu/.

    However, other labialized consonants such /tʷ/ did not trigger the shift, so /titwu/ stays /titwu/. If there was a new /fʷ vʷ/ series, it most likely did not trigger the shift either. But note that the "tʲw" cluster had not yet become /fʷ/.

    NOTE ON POLITICS ... THIS IS WHERE TUPPY BREAKS OFF (year 6843)

39. ŋ ŋʷ were denasalized to g gʷ in all positions.
40. The palatalized velar consonants kʲ gʲ became the postalveolar affricates tš dž.

    Note: these may have still been palatals such as ć ǵ, because /gʲ/ behaves differently than the rare cluster /dž/ that came from collapse of /tVž/ until the later shift of /g/ > /dž/ merges /g gʲ dž/ all together.

41. The labialized velar consonants kʷ gʷ pʷ bʷ were decomposed to the clusters kw gw pw bw.
42. Final y in trisyllabic words disappeared. Due to analogy, it disappeared in some shorter words as well. However, consos that now occurred at the end of a word because of the dropped y became labialized, though this is not shown in Romanizaiton.
43. The remaining palatalized consonants became labiodental fricatives: fʲ tʲ merged as f, and nʲ changed to v. The sequence nnʲ shifted to nv and later /mv/ (still later /mb/ in most environments).
44. The cluster pf, when straddling a syllable boundary, became ff. It is likely that /mpf/ and /ppf/ did not exist at this time, therefore making this shift unconditional.

    NOTE, actually /pptʲ/ did exist, and this changes to /ppf/ in the shift above.

45. Medial vowels in trisyllabic words disappeared if the resulting consonant cluster was acceppable ("the Debra shift"). wr > rw (distinct from earlier shift). Here again, labialization hung around if the deleted vowel was o u or y. Thus there were minimal pairs such as puppa "salamander" vs pupʷpa "kind, humanitarian". This period (around the year 7300) is the beginning of what is often considered "Classical Poswa".

    Poswobs invaded Pabappa territory beginning around 7414, and the language was essentially unchanged at that time. The next few changes never occur in native words because they would have appeared in compounds only, and grammatical reanalysis eliminated the clusters that fed the shift; however, they occur in opaque loanwords and names.

46. The clusters tm km fm vm shifted to vb. šm žm sm šb sb shifted to žb.
47. The clusters pn bn shifted to vž. šn sn became žv.
48. The sequences pm tn shifted to mm nn. tp became pp. Importantly, however, the nasal clusters /mn nm/ both remained intact.
49. The clusters pk and bg became pw and bi respectively. mk became mw (not mpw). Similar changes applied to other clusters ending in velars. The clusters /bk pg/ did not exist at this time because regressive voicing assimilation had automatically shifted them to /pk bg/. This voicing assimilation only happened when the Debra shift brought the consonants into direct contact.
50. All posttonic k g shifted to w i, taking any intervening vowels with them; this is the same as the shift above except that there was no voicing assimilation. This also included the gʲ spoken of above which was listed as /dž/ but still behaves distinctly from preexisting /dž/ from /tž/. Note also that this shift did not affect /ž/.

    NOTE: This shift is important and may be handled incorrectly in some dictionary entries. It seems to imply that *ANY* posttonic /w/ also consumes preceding vowels, which often happens when coming from earlier /fʷ/. On the other hand, there needs to be a path that generates posttonic /uw/, as distinct from /w/, and this might need to come from sequences like /wiw/.

51. Posttonic sequences like igigʷ and agik shifted to uw. That is, wherever a sequence like /iw/ arose, this became /uw/. But double /k/ sequences still ended up as a simple /w/.
52. The voiced velar stop g was fronted to dž unless it occurred in a cluster after another consonant and before /a o u/.

    This shift is not being followed exactly; it is behaving mostly like /k/ except in the onset where it does indeed shift to /dž/.

53. The clusters šb bš were devoiced to šp pš respectively. Then sb bs shifted to sp ps. That is, the fricative dominated the stop, unlike most such changes.

    These clusters existed despite a previous shift of the same clusters because in both cases they appeared only in foreign loans and in some ad-hoc compounds. This means that an original sb could variably appear in Poswa as any of s b žb sp, depending on when the two consonants came together.

54. The clusters žp pž became žb bž respectively.
55. The clusters tv ttv shifted to v vv; this was actually through a [d] allophone, which explains why they did not both become geminates.
56. Likewise, tf ttf shifted to t tt. This may seem a contradiction but it is actually the same shift as the above, explained by the path of /t/ > /d/ > /ð/ > /v/ which is compressed into one step above. (Note that in one word, /pts/ shifts to /pps/ as though /ts/ cannot start a syllable.)
57. The clusters nr nl shifted to ng (IPA [nd]).
58. The clusters pl bl became p b respectively when overlaying two unaccented syllables. The same also happened for other stop + nonstop clusters such as pr br pš ps bž pt pf bv, althoiugh pš, pf, and ps survived as aspirates pʰ for long enough to survive a particular subsequent shift. Clusters like mž mdž shifted to mb.
59. The sequences yw ww wy changed to ʷu in all positions.
60. When unaccented, the sequence wi shifted to i before another vowel.
61. iy ii > ia ie. This also includes /ir/ > /ia/ (when not labialized).
62. In wholly unstressed syllables, except when preceded by /a/, the clusters rl lr changed to al ar respectively, and raised the preceding vowel.
63. The cluster rgw shifted to vbw, with both consonants labialized.
64. A velar-onset syllable preceded by another changed to alveolar if the vowel was a back vowel, but postalveolar if it was a front vowel. The affected consonants were k g kw gw r.
65. The diphthongs al yl shifted to ae e. These did not affect /lʷ/.
66. Geminate stops overlaying two unaccented syllables reduced to single if there was another geminate earlier in the word.
67. The cluster mp shifted to mb when occuring after a voiceless stop.
68. Initial ps pš shifted to p, also sometimes if overlaying two unaccented slabs just as /pl bl/ had.

Baba (~6000) to Old Pabappa (~6500) ... too early?

Alternate names: Pespimbesa

1. Debra shift I. (But rV and wV didnt compress). ("ae" and "al" remained distinct here). If the deleted vowel was y, o or u, the preceding consonant became a labial. If a consonant was already labialized or palatalized, it did not contract.

   This shift must be broken into steps because it does not turn /tut/ > /pt/. It may be that the /tʷ/ > /p/ actually happened later in a second shift where only /ə/ was deleted.

2. In the syllable coda, lʷ rʷ shifted to u. Also, in clusters like /kʷr/, this became [kʷrʷ], and then the resulting lʷ rʷ shifted to w. This can be identified as part of the same shift above.
3. Homorganic stop-fricative sequences like pʷf resulting from the vowel elision came to stand at the head of a syllable, and therefore both consonants became labialized, resulting in sequences that were spelled pfʷ.
4. After a stop, fʷ shifted to a fully rounded voiceless bilabial fricative, here spelled h because there was no other /h/.

   This also extended to shifts like tʷl shifting to tʷ, since there was no /lʷ/ or /rʷ/.

5. fʲ shifted to f. This means that it cannot participate in the shift below where palatalized consonants stain the following vowel to /i/.
6. ŋ was denasalized to g in all positions, except bound clusters such as /ŋp ŋt ŋr ŋk/ which behaved as prenasals.

   NEED to check dictionary to make sure this rule has been followed properly, including subsequent changes. For example, if kg shifts to gg later, it ends up as bb ultimately, the only voiced geminate in the language. But perhaps they should be devoiced when they were still stops.

7. Clusters like np nb shifted to mp mb (homorganic). This does NOT shift /mpt/ etc intp /nt/. which happens later and is important to happe nalter.
8. Following a consonant, diphthongs of all types were resolved in favor of the first vowel. Thus, all palatalized consonants shifted to the corresponding plain form (unlike in Poswa, where most had distinct reflexes). That is to say, ʲa ʲe ʲi ʲo ʲu ʲy all became a simple i and ʷa ʷe ʷi ʷo ʷu ʷy became a simple u.

   Importantly, this did not come from a bare /w/, even in word-initial position; /wes/ remained /wes/ instead of shifting to /us/.

   Note that this means syllable-final lʷ rʷ, which earlier became /u/, now disappear to Ø except in the rare cases mentioned below. And əu shifts to ə (usually later to /o/).

9. Triphthongs with a high middle element were collapsed to two-vowel sequences along the opposite pattern; thus, for example, iwa> ua.

   It is possible that iai shifted to just an i. If not, it would be /ia/, not /ai/.

10. Stops occurring before a nasal assimilated fully.
11. Clusters of two fricatives of differing points of articulation were assimilated in favor of the second consonant.
12. The clusters žbž špš sps shifted to žž šš ss, and likewise for mismatched pairs.
13. Clusters of fricatives and stops of dissimilar voicing were resolved in favor of the second consonant. This includes fž sž shifting to vž žž (later to /d Ø/).
14. The clusters gv kf became bv pf.
15. Word-initial pš tš shifted to š while word-initial ps ts shifted to s.

    Note that initial /tš/ occurs only in the copula; it is not /kʲ/, which had remained as such and then dropped its palatalization in a shift just shortly before this one.

16. The fricatives šʷ žʷ (possibly only found in the coda) shifted to f v.

    NOTE that this is an important and newly introduced shift, as before they behaved as normal /š ž/ and therefore mostly shifted to /s Ø/.

17. The fricative š changed to h in initial position and to s elsewhere.
18. The voiced sequences ž dž came to be pronounced z.
19. The cluster rr was reduced to r.

Old Pabappa (6500?) to Pabappa (8700)

1. Double nasals were reduced to singles.
2. The vowel ə either disappeared or became i or o (governed by the surrounding vowels' HEIGHT (not backness)). If it disappeared, it also labialized the new final consonant.

   NOTE THAT THIS INCLUDES MEDIAL SYLLABLES, since /rə/ did not delete earlier. However it is not clear what /rʷ/ will turn into, and /b/ is unlikely.

   This also includes WORD-FINAL SYLLABLES that did not delete during the Debra shift because there was no following syllable.

   Consider also forcing /kʷ/ > /p/ at least in final position, since it will otherwise disappear. Additionally, earlier an implied shift of tʷ > p took place word-finally, which would imply /kʷ/ > /p/ too if kept.

3. The fricatives v z shifted to d Ø in all positions. This included doubled forms.

   It is possible that this causes the relatively common cluster bv to become dd (not /bd/), as per the shift involving /ps/ below. This would be an affricate at first.

   Another possible exception is that vz (common in a suffix, from earlier -fž-) might also become dd because the two consonants were shifting at about the same time, and the v > d shift passed through /ð/. Still, if /z/ disappeared even slightly before /v/ became /ð/, the reflex would be a simple d after all. One reason to expect that /z/ > /Ø/ was slightly earlier is that the shift of /f/ > /s/ is displaced below.

   Note also that the shift of /zz/ > /Ø/ means that the reflexes of the palatalized forms of sl šl žl are all Ø; this is important because many words for handheld objects end in /-sla -šla -žla/, which therefore all become /-a/ when possessed.

4. The velar stops k g changed to the fricatives š ž in all positions.

   These were weakly labialized, but it is important not to confuse them with the true labialized fricatives below. This is complicated, because the reflexes essentially switch places: /š ž/ mostly become /p b/, but /šʷ/ becomes /s/ except in initial position (and it may never have occurred initially), while /žʷ/ might not exist at all.

5. The labialized consonants sʷ šʷ žʷ shifted to f f b. (It is not clear that a distinct /žʷ/ still existed, but if it did, it too would have been carried away.) Then nʷ shifted to m at least in final position.
6. The clusters šs sš became ss šš. (Unless /š/ had already shifted to /f/.)
7. The fricative f changed to p in initial position and s elsewhere. This also triggers a shift of all pf to ts, even though /ps/ remained as such.
8. This was around the year 7000. Note that the Play 1st and 2nd person past passive participle endings /-su -si/ had become /-s -se/ by this stage; it is unlikely that Pabappa would preserve them since the 1st person would have become silent, but other daughter languages may preserve both.
9. The voiceless stops p t became the geminates pp tt when following a voiceless stop plus a vowel.
10. The postalveolars š ž dž became f v bv in all positions.

    The shift of /dž/ > /bv/ is listed here because if /d/ remained unassimilated it would devoice and then fall into the /tp/ > /tt/ rule below. But a shift of /ž/ > /v/ without affecting an adjoining /d/ is highly unlikely.

11. The cluster ph (phonetically something like /pɸ/) shifted to pp.

    Note that this is actually a dummy shift to prevent it from being voiced to /b/. In fact, it remained as roughly /pɸ/, still distinct from the rarer true /pf/, which shifts to /pp/ later on, and also distinct from a simple /p/, which in some cases shifts to /b/ later on.

12. The glottal fricative h shifted to f before /u/.

    It is possible that this shift can be moved further down so as not to interfere with the shift above.

13. The voiced stops b d became the voiceless stops p t in all positions. Adjacent fricatives also devoiced.

    It is possible that the cluster vv should become ff here, even though there was no stop. It is the only surviving voiced fricative, and therefore the only surviving voiced fricative cluster.

14. Word-final s disappeared to Ø.

    The word-final /-s/ was soon restored in just one instance, the Play word pais and anything rhyming with it; this is because all of the inflected forms of /pais/ had become identical to those of Play /pasi/, which had not lost its /s/ because it was still padded by /e/. The time interval between these two shifts was very short, and effectively by merging the two words the shift was undone in just one environment.

    Note that there is probably no word-final /f/, but if it does somehow appear, it would also shift to Ø.

15. Word-final e disappeared, except after a consonant cluster; in this case it changed to i.

    Many longstanding irregulars were regularized here. For example, nisol ~ nipse- became nipsi ~ nipse- by making what had been the genitive (ending in -s) the new nominative form.

16. Before a nasal, p s t n assimilated completely.
17. Clusters of a nonlabial stop followed by a labial stop were resolved in favor of the nonlabial one. It seems most likely that this is just tp shifting to tt.
18. Final o was were lowered to a except if the accented vowel was mid-height (e or o).

    NOTE ON POLITICS: This is considered to be the classical stage of Middle Pabappa.

19. The clusters tl ttl both changed to ll.
20. The sequences mr sr lr became mpr spr rr. Any other nonlabial consonant before /r/ became labial.
21. Clusters of a nasal followed by any other consonant of differing point of articulation were assimilated in favor of the point of articulation of the second consonant. Thus surviving mn shifted to nn.
22. ml became mpl.

    NOTE ON POLITICS: This occurred around 7414 AD.

23. The voiceless labiodental fricative f changed to w in initial position.
24. The voiceless glottal fricative h shifted to Ø.
25. Unstressed syllables of the form CVCC where the two latter C's were a geminate or one of a few other types of consonants changed the vowel to a very short schwa /ə/.
26. Intervocalic voiceless stops became voiced.
27. The cluster pl became bl in all positions.
28. The cluster pr became b in word-initial position or after /m/, p after /s/, and br elsewhere.

    Consider lbr rbr > lb rb as well, and perhaps extension to all clusters (though other clusters would be rare, since pVrr was rare or perhaps nonexistent).

29. The geminates pp tt ss became p t s in all positions.
30. The cluster sp became ss.
31. The rare vowel sequences ei ou (found almost entirely in verb conjugations) shifted to e o.

    NOTE ON POLITICS: (ANDANESE PABAPPA SPLITS OFF HERE)

32. Initial v became f.
33. The labiodental fricatives f v became the bilabial stops p b in all positions.
34. The marginal schwa phoneme disappeared, creating some new clusters and geminates. (Debra shift II.)
35. The bilabial stop b, between two identical unaccented vowels (e.g. -aba, -obo), turned to m sporadically as the result of analogy from various noun declensions.

Sister languages of Pabappa

For sound changes, see Macro-Pabap_languages#East_of_Paba.

Sister languages split off from Pabappa in several waves, and could arguably include tiny, relic branches that are older than the Poswa-Pabappa split as well as those that are younger. The arbitrary breaks at around 6500, 7000, 7414, and the unlabeled Andanese Pabappa split do not correspond to wars, as the Pabaps had long since ceased fighting wars.

See also

• Teppala
• Minor languages of Teppala