User:Soap/MRCA

This is a scratchpad for the MRCA. The orthography is different here for ease of searching. Most notably, tone markers are dropped because there are no monosyllabic low tones at this stage and the tone of longer words can be predicted from its consonant layout.

PRONUNCIATION NOTES

Most coda consonants are silent:

All non-nasals are silent except before a vowel (liaison);
All nasals are silent before non-voiceless stops except in the DRM dialect;
All consonants are silent in the Tapilula branch except /mb nd mf/ etc which always load onto the trailing syllable.

Phonology

15:14, 8 November 2023 (PST)

Consonants

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

Vowels

The vowel inventory was /a e i o u ə/, where /e o/ were alternants of /ə/ in most cases, and tone was still analyzable as a coda if an unpronounced /ʔ/ is used to explain all unexpected highs. (Later, even this analysis failed due to grammatical renalaysis.) Below, the words are written etymologically, e.g. pal for what was always just /pa/ except with liasion before a vowel. THis is because in some daughters, the final consonants might return through reanlysis and dropping of vowel-initial suffixes.

Grammatical morphemes

09:20, 29 September 2023 (PDT)

Undifferentiated core

pum "if ....."

man not; negative (used as suffix in Lava Bed, but have been prefixed earlier)

The final /n/ suggests a suffix position by MRCA times, but a cognat epreifx could have existed

pal XOR; exclusive OR

(g)at possibly means "and, in addition" as it appears at the end of several words with similar meanings. Not inherited as such into any daughter language (it would just be /a/). For example mfà-t- mʷò incorporates this (Play loses this because it merges with mfà-n- mʷò of almost opposite meaning).

nə ".... then ....." (never occurs by itself)

From ther verb for walk. Can be seen as "therefore"
nə gal [a derived copula appearing in DRM]

-n oblique suffix (genitive)

It is possible this was originally syllabic, even though another syllabic /-n/ suffix existed marking the imperative mood. It could be that these were one morpheme far back in the past, and the imperative developed from a phrase in which the verb was placed in the genitive (or perhaps subjunctive, assuming they were united), and then a helper word was used that was later dropped.
The alternative view is that the oblique was originally /-ni/, still distinct from the locative /-ni/ up above because the /i/ pronunciation was from two different time periods.

-i accusative marker

Exists in proto-Dreamlandic. Possibly not a distinct suffix in MRCA, since final /-i/ could come from so many origins. This was almost certainly earlier /ɨ/.
Possibly for patients only (secundative alignment), as in Lava Bed languages. This would have consequences for word order.
This is used even after vowels, and probably does not require an epenthetic /g/. If it did, it was analyzed away in DRM.

-ni locative suffix

This can also be analyzed as oblique -n followed by a standalone verb /i/, assuming other inherited traits hold. Despite ending in /i/ it is not cognate with accusative /i/, and was a true [i] even thousands of years back, unlike the more common /ɨ/.

-i₂ a hypothetical second suffix set up to divide the locative /-ni/ into genitive /-n/ and a suffix that was separate from the /-i/ accusative but ultimately merged with it. If -i₂ exists, there could be some syntactically unexpected functions folded into the accusative. For this to work, there would need to be a sound change like /ɨɨj/ > /īj/ > /i/ without insertion of /g/.

ta genitive standalone word (appears before noun)

yom hand. Evolves to an instrumental in DRM and to a noun classifier in Gold, but probably had no special grammatical function in MRCA

ŋa verb of preparation (know ---> ask; eat ---> cook). THIS CANNOT EXIST IF /ŋa/ "involuntary" also exists. The lower one is most likely a mistake.
- ŋa lo "if ....."

-i- past tense infix
nda past tense marker (coexisted with /i/)

This may have been a carrier morpheme for person marker infixes, creating constructions like I-did-you a favor.

gu plural (goes before noun)

possibly originally /gup/, but this idea conflicts with its presence as a simple /w/ before some vowel-initial nouns, and it is unlikely that these were all analogical.
This most likely will go back to a primordial form /w/ or /hw/, which leads to a suffix /-u/ after consonants and u-stems and the replacement of the final vowels /a e i o @/ with /o/. Because this is destructive, it disappears early on as a noun suffix except in a few common nouns and counter words, perhaps also pronouns, and then comes to be used as a prefix through analogy from the counter words.

hṅ during; while; when

wi self (agent or used in isolation); possibly only an emphatic, used in forms such as wi nam "I myself". This is probably a primordial CV word and might simply be the same word as the word for thumb.
mfup reflexive verb marker (behaves as if the accusative of "self")

mbat possibly a noun SUFFIX from an older stage of the language, marking abstract nouns

Bound verbs

pə fail to
- possibly earlier /pən/; if so, this consonant would remain in DRM
- pə pal try to; attempt (/pal/ = OR)
gəpa inchoative (never contained /l/)
hu should, must (expanded to /mbihu/ with /mbi/ as a carrier verb); evolves as a suffix
mu ~ mbu capable of; able to; habilitative; prone to [used for the AGENT]; evolves as a suffix
twomo in the state resulting from; [used for the PATIENT; pairs with /ŋa/-type verbs]
pa valency raising operator
no desiderative (possibly a vowel-harmony variant of /nə/ "(and) then; to walk". Always goes before verb. This verb never had a trailing consonant

to becomes the copula in at least some daughters

i and o may be passivizers of some sort; /ondat/ means "blush" (o + undat "embarrass") and /iKa/ was used early on as a word for meat (from Ka "eat").

the /o/ is a mistake, as it was actually just a very short content word relating to blushing. it is likely that only one morpheme is needed here.

Occasionally bound nouns

ən a child. Repurposed as a suffix in Andanese to form diminutives, but in MRCA it would have been required to place the child word in front because of animacy issues. Perhaps it was used suffixally even in MRCA as though it were the genitive of a word that was just a bare /ə/. In Andanese, the /-n/ was lost early on, and then the schwa, which meant that the primordial final consonant was once again at the end of the word, just as it has been *before* the MRCA broke up, meaning that it appeared as though Andanese had evolved backward.
gək a place to be. MRCA was head-initial so this would be of litttle use in compounds.

Other verbal suffixes

-tə irrealis (forms most moods in Play)
-ŋa involuntary action (assuming Gold/Play shifted it to a desiderative meaning). THIS MAY BE A MISTAKE
-gə generic verb suffix (possibly constructed from nulls)

-gəniŋ (used to construct passives)

-tu probably of weak meaning; occurs in /məntu/ "put to sleep" even as /mən/ by itself also means this (because it is not reflexive by default). It could perhaps have to do with forced transitivity as opposed to voluntary activity.

Primordial person markers

These may be something other than person markers. For example, they evolve into meaning self/non-self, but only at a late stage, possibly post-MRCA.

-m possibly a very archaic 1st person agent suffix; earlier more likely identity/possessor
-t possibly a very archaic 2nd person agent suffix; earlier more likely identity/possessor

y- (earlier /r/) possibly a very old 1st person patient prefix or standalone word (but note that it comes from /r-/, not some type of /i/, and that this /r/ is all that is left of the stem by the time of the /r/ > /y/ shift. Thus, if it were a word by itself, it would need to be /rɨ/ at some stage)
l- possibly a very old 2nd person patient prefix or standalone word. See above as to whether it contained a vowel; it is not required to also evolve to /i/.

See here for how /r/ could have evolved from /l/, thus uniting the two. The /a/ in the 1st person pronoun words is still unexplained, but perhaps /n ~ na/ was a variation at one point (the plain /n/ is definitely needed though).

Try to find the tornado dream verbs.

Topic markers

These were likely prefixes, but are written as separate words for simplicity. All descendant languages used them as fully bound forms except for those that detached into pronouns.

SAPs

Note that verbs using these still require distinct person markers.

nam 1st person agent
ŋam 1st person patient

gət 2nd person topic (either agent or patient)

the /hə/ for the 2nd person patient form may therefore be an analogy confined to just Lava Beds.

nambə 1st person agent, 2nd person patient [NOT LISTED AS CONTAINING /t/]
ŋambə 1st person patient, 2nd person agent [NOT LISTED AS CONTAINING /t/]

Third persons

ka epicene human agent; tree

Possession markers

These may appear only before the oblique suffix, since it would not be possible to put the accusative suffix in (unless it was once syllabic). Likewise, because the deletion of final consonants came early, they would make little sense as standalone forms.

-m- 1st person
-t- 2nd person

Verbs

See User:Soap/scratchpad.

The verbal morphology was much more complex than the noun morphology, already containing a nearly complete Lava Bed system. The non-Lava Bed languages moved this to the suffixes, to the copula, or dispensed with it entirely, while Play and Andanese only made the verb conjugations far more difficult (each independently).

Gluons

The so-called gluons mark the roles of the 1P and 2P (speech act participants) in each sentence. That is, they are not giving the person numbers of the agent and patient, as in most languages, but rather giving the roles of the 1st and 2nd person, and these appear on all verbs, even if there is a 3rd person agent or patient or both. The Andanic clade preserved this system but transformed the surface realizations with schwa deletion, while Play preserves the system fairly well phonetically, but has shifted the meanings to egophoricity. In the MRCA, they were still 1st and 2nd person role markers, as they remain in Andanese.

The gluon matrix was such as

         AGT     PAT     OBS
AGT      ---      əl      ək
PAT       ey      ---     əŋ
OBS       əḳ      əh       Ø

These are always infixed into the final syllable of the verb. Thus from latan "smile", one says latəlan "I smile at you". Forms such as *latanəl, where a final vowel is assumed to have been lost, were not used in the MRCA, but may have appeared analogically in some languages that underwent a second round of vowel loss.

The past tense infix went inside this infix, and was something like -ig-. There may have been irregular strong verbs preserving a bare /-i-/ infix that triggered consonant mutation and then disappeared. For example, a word such as lahəlan "I smiled at you" might exist, where /ti/ > /h/ following a rule from thousands of years earlier.

The matrix may have matured to

         AGT     PAT     IDT     OBS
AGT      ---      Ø      əl      ək
PAT      əg      ---     ---     əŋ
IDT      ey      ---     ---     ---
OBS      əḳ      əh      ---     əh

even in MRCA, but this is best kept for the clade just outside Lava Bed. This expanded form was found in the Andanic clade, and possibly in the early Gold/Play clade, but Gold/Play later reduced it. Note the duplicate /əh/, a defect in the system, and that Gold might not have enforced 1P and 2P values for the rows and columns (if it were not pronounless).

3rd and 4th person markers

Reflexes of the 3rd and 4th person markers appear in both Lava Bed clades, so must have appeared in proto-Lava Bed, and separate forms for the different roles must also be constructed, and these would need to be projected back to MRCA, but they need not have been used in the same way in MRCA since they were plainly derived from freestanding words rather than infixes. For example, what became the generic human agent marker in Play is a content word in MRCA, ṗo "teenager, adolescent". This means that they could have appeared before the verb, and could even have behaved as ordinary nouns. However, they would have irregular inflections, and so could be considered a subset of pronouns.

Note that the separate forms are for the roles of the 3rd and 4th persons, not for marking whether each participant is 3rd or 4th person. This is for the same reason as above with the gluons. Here, the concept of 4th person merely means the second non-SAP mentioned in the sentence, regardless of its role.

It is most likely that the observer role was Ø for all non-SAP's, as it would be meaningless to include in a sentence, unless it can be unified with the genitive at a distant pre-MRCA stage. But then there would still need to be a separate zero morph, at least if keeping the analysis in line with the Lava Bed system.

Notes

The "tornado verbs" may have been to-n to-go to.