Proto-Ronquian

From AkanaWiki
Jump to: navigation, search

Proto-Rompian (a.k.a. Proto-Ronquian) as reconstructed by T2. Rompian comes from the reconstructed word for 'language':

  • roxʷ < *roŋxʷ < *romp ARN
  • ɻɔ̃ʔ < *rɔ̃p < *romp RT
  • rǒp < *rɔ̃p < *romp RTJ
  • rof < *romf < *romp RNW
  • rɔm < *romp MT
  • raj (< rV(V)(C) < *romp) RTI
  • reː < *raj (< rV(V)(C) < *romp) RRK

Initially the name Proto-Ronquian was used, as the word for language was thought to be *roŋkʷ. However as serious reconstructing began, it quickly became obvious that labiovelars were an innovation of A-Rox Ŋʷoskʷuɣmʲa from labials under some conditions and consequently the word for language was rather *romp. Therefore the term Proto-Ronquian is obsolete and Proto-Rompian is more accurate. Nevertheless, the former is more widely used.

Proto-Northwest-Tuysáfa was also sometimes used.

List of used abbreviations:

Phonology

Sound inventory

Zju has suggested the following consonant inventory.

Consonants Labial Coronal Dorsal Uvular Glottal
Fortis stop p t k q ʔ
Lenis stop type b d g ɢ
Fricative f s ɬ x χ h
Sonorant w r l j ʀ
Nasal m n ŋ ɴ
  • R denotes a consonant phoneme of unknown quality that appears after a coronal onset and which is reflexed as w in RT and MT, but as retroflexing of the onset in RTJ. It was not deleted by the CRV > CVː sound change in MT, unlike most other liquids, but lengthened the following vowel nevertheless.

KathAveara has suggested the following consonant inventory.

Consonants Labial Coronal Palatal Velar Labio-velar Uvular Glottal
V.less Stop p t k q ʔ
Voiced Stop b d g ɢ
Fricative f s, ɬ x χ
Approximant l, r j w
Nasal m n ŋ ŋʷ ɴ


Vowel inventory:

Vowels Front Central Back
High i u
Mid e ə o
Low a

ə stands for a vowel that shows up variously as e or a in RT and RTJ, with or without palatalising effect, and only as a in MT.

Vowels could be preceeded and followed by any of w, j.

Allophony

One allophonic process can be safely reconstructed for PR.

Lenition

Onsets of strong syllables lenited intervocally when being a single consonant and in some consonant clusters:

  • hC > C
  • p t k q > b d g ɢ
  • f s ɬ x χ > w r l j ʀ (<β z ɮ ɣ ʁ at pre PR stage)
  • b d g ɢ > f s x χ
  • m n ŋ ɴ w r l j ʀ > m n ŋ ɴ w r l j ʀ (sonorants don't lenite)

Consonant clusters containing plain stops and non-stop consonant lenited only the stop:

  • *trip 'offense, insult'
  • *sədrip < *sətrip 'offenses, insults'

Clusters of a plain stop and a glottal stop also behaved in this way:

  • *əbʔik 'once' < *pʔik 'one'

Plosive series

One of the two plosive series could have had allophones under some circumstances. The possible scenarios are:

  • T, D
  • T ~ Tʰ, D
  • Tʰ, D ~ T
  • Tʰ, T

With the first three being most likely, since fricatives also come in voiceless and voiced series and the two sets change their series under the same condition - namely intervocalic lenition.

Phonotactics

At least the following sets of onsets are likely to have existed:

  • (h)(C)C(r l)(ʔ h j w)
  • NC

Where C doesn't include glottal consonants and NC is a cluster of a homorgranic nasal and plain stop. It's quite possible that not all combinations of (h)(C)C(r l ʔ)(j w) were permitted.

When -j-, -ʔ- or -w- was infixed, any of preceding j, w, ʔ was deleted.

A liquid and ʔ were metathesized for the purpose of forming an ejective at a later stage.

The following codas seem to have been permitted:

  • T
  • NT
  • N
  • ɬ
  • r

Where T is a plain stop and NT is a homorganic cluster. Additionally, l, s, h and ʔ are also somewhat likely to have been permitted as coda consonants, but currently no reflex of any of them is known.

Strong syllables

Strong syllables probably carried some kind of tone or stress. Words had one and only one strong syllable.

Weak syllables

Weak (reduced) syllables had the same set of onsets as strong syllables. However they had no coda and a limited set of vowels, or only one vowel - ə. Additionally they were atonic or unstressed if there was phonemic tone or stress in PR.

It's possible that PR words had no weak syllables at all and only one syllable instead.

Grammar

PR grammar was largely analitical, however with some traces of inflectional morphology.

Nouns

Nouns likely had multiple ways of forming plural number, or it was that plurality was not yet a well defined category in PR and there were several concurrent ways of expressing plural meaning.
Many, if not all, of the following affixes were likely derivational rather than inflectional, which were later grammaticalised differently.

Prefixes

  • *ə- - forming construct state in some cases.
  • *h- - forming plural number in some cases.
  • *t- - paucal or collective prefix.
  • *m- - deverbal prefix forming abstract nouns.
  • *s(ə)- - deverbative prefix, perhaps for resultative nouns. A vowel was inserted before consonants.
  • *r- - augmentative prefix.
  • *Vr- - agent prefix.
  • *Vk- - instrumental prefix.

Infixes

  • *-j- - singulative infix, placed between onset and nucleus of the main syllable.
  • *-w- - plurative infix, placed between onset and nucleus of the main syllable, forming plural number in some cases.
  • A paucal infix of the shape *-nt- could also have existed.
  • *-ʔ- - of unknown exact function. An example is *pʔom 'hat' < *pom 'head'.
  • *-r- - diminutive infix.

The usage of singulative and plurative infix was lexicalised in some cases, if they were not originally a part of the derivational morphology in the first place.

Suffixes

  • *-t - dual or plural suffix
  • *-Ri - diminutive suffix.
  • *-nu - agent suffix.
  • *-(h)it - resultative suffix.
  • *-Ra - instrumental suffix.

Construct state

Construct state was formed by prefixing a lone vowel of unknown quality (probably already reduced in PR), which on its part triggered consonant lenition. That vowel was subsequently lost in most cases, leaving consonant lenition as the marker of the construct state.
There might have been some other changes involved in forming the construct state, such as deleting glottalic consonants in the onset.

The construct state had various functions, such as marking the possessee. It contrasted with absolute state, which was the unmarked form of the noun.

Verbs

A verb in PR was either transitive or intransitive - transitive verbs couldn't be used in intransitive constructions and vice versa. Instead, derivational morphology existed.


Attributive forms

Verbs had an attributive form, which essentially turned the verb into an adjective. It was formed by the prefix g-.

Affixes

Prefixes

  • *g- - turning verbs into adjectives(called attributive verbs). It had the allomorph *k- before voiced obstruents probably already in PR. Additionally the allomorphs *gə-, *kə- before certain consonant clusters existed either in PR or at a later stage.
  • *n- - antipassive prefix; reduces the valency of the verb, e.g. pʔi 'grow' > mpʔi 'grow up, become an adult'
  • *nr- - the state of resembling a noun.

Infixes

  • *-Ert- - reversing the meaning of a verb.

Suffixes

Particles

PR had at least the particle *ni for indicating plural number in nouns. It was placed before its head noun.
Two particles *t and *n also existed, marking the noun for paucal and/or plural number.

Pronouns & Determiners

Personal Pronouns

Proto-Ronquian had 3 persons in at least 3 numbers, singular, dual and plural. It also had a 3rd person collective pronoun. There may have been an animacy distinction in the 3rd person too.

sing. dual plur. coll.
1st na nat haj
2nd muŋ mint < miŋt wor
3rd Anim. gi gut -
3rd Inan. mdo do
  • mdo seems to be formed from *do with prefixation of *m-; its connection with the other two *m- prefixes is unclear.

Indefinite Determiners

Demonstratives

Interrogative Determiners

Indefinite Quantifiers

Numerals

Proto-Ronquian uses a mixed base-6/24 number system. The basic numerals are reconstructed as:

KathAveara Zju
cardinal ordinal cardinal multiplier ordinal
1 pʔik mʔək *pʔik *əbʔik *hmək
2 ki: kso *qis *əɢis *qzo
3 tʔuq ntoq *tʔuq *ədʔuq *ntoq
4 paɬ *pjaɬ, *paɬ *əbaɬ
5 fəɴ *fjaŋ, *faŋ *əwjaŋ, *əwaŋ
6 ɬja ~ ɬa *ɬja, *ɬa *ɐla
12 ɬat *ɬat
18 qoɬ *quɬ
24 tram *tram


Numbers 4, 5, 6 had variants with and without singulative infix. Numbers 6 and 12 shared the same root, with 12 having dual suffix, and 6 having either the bare root or singulative infix.

The construct state of a numeral was used as a multiplier.

Syntax

Prepositions

There were two sets of prepositions - noun-like (nouns in construct state) and primary.

Sound changes

Sound changes preceeded by an interrogative are entirely speculative (or nearly so) and rather have the status of suggestions.

Conditions preceeded by an interrogative mark are unknown and/or speculative.

To A-Rox Ŋʷoskʷuɣmʲa

  • Tʔ > Tʼ
  • Tʼ > T / _V[-stress]
  • NT > NÞ > Þ / _#
  • V > ɐ / _[-full syllable]
  • ɬ > ʃ / _#
  • C > Cʲ / _i, ?
  • C > Cʷ / _u, ?
  • Cj Cw > Cʲ Cʷ / _V
  • P > Kʷ / ?_{a o u}
  • P > Kʷ / {o u}_#
  • C > Cʲ / {e i}_#, _Cʲ
  • C > Cʷ / {o u}_#, _Cʷ
  • ɬj ɬ > ʃʲ ʂ
  • q > χ / #_
  • ɐ > ɪ / _Cʲ; ʊ / _Cʷ
  • ʀ > ʁ
  • ɢʲ ɢ ɢʷ > ʁʲ ʁ ʁʷ
  • ɲʲ ŋʲ > nʲ
  • ɳʷ > nʷ
  • si > ʃi
  • tj > tʃ
  • N > Þ / _# NB: Only two cases found, neither from an original NT coda, details about this SC are unclear. It may or may not be connected with NT > NÞ > Þ / _#.
    It may also have not occured at all, but rather such a correspondance being raised via a single consonant suffix being attached to those two words('thou' and 'five') before NT > NÞ > Þ / _#.

To Mhakh Thandim

  • T[-voice]ʔ > Tʼ
  • b > w / V_V
  • tr > dr / V_V
  • drV dlV > rV[+length] lV[+length]
  • CC{ʔ r l w j s}V > C{ʔ r l w j s}V[+length]
  • C{ʔ r l w j s}V > CV[+length]
  • md sd ng rg > mr sr nl rl
  • rl > ral
  • ŋ > ʔ / #_VN
  • D T > T Tʰ
  •  ? ʀ > ∅ / _VC[+uvular]
  • ʀV, Vʀ > V[+breathy]
  • hN > N[-voice]
  • hV hCV > V[+breathy] CV[+breathy]
  •  ? Vh > V[+breathy]
  • q > h / #_
  •  ? g > h / #_, ?
  • q > ʔ / _#
  • ɬ > u / _#
  • ɬ > s / #_
  •  ? V > a/ə / ?
  • ek > ak / _#, ?
  • ij uw ow > ii uu uu
  • aL > aj
  • a o u e i > ə ɔ ʊ ɛ ɪ
  • jɛ > e
  • aa oo uu ee ii > a o u e i
  • CCC Cj > CCəC Cɪj / ?
  • Q > K
  • NT > N / _#
  •  ? NT > N[-voice] / #_
  • hV > jV[+breathy]
  •  ? k kʼ > ʔ / #_[+ ?]
  • NT > ND / V_V
  • x > χ
  • pʰ tʰ kʰ > ɸ θ x / _#, #_
  • p t k > pʰ tʰ kʰ / when not in CCs; however #_TV is a valid condition
  • m > ∅ / #_u
  • ɸ > w / V_V

At some point there was a reduplication process word initially that affected some of the words, which took place before changing NT clusters. The reduplicated vowel was a, which agreed in length with the main vowel.

Metathesis of onset consonants also took place, e.g. in klu /kʰlʊ/ destroy, but salcu /səlʔʊ/ destruction, ruin < *lku, *s(a)-lku

To Ronc Tyu

  •  ? u > o / [+uvular]_
  • Q > K
  • V > ∅ / [-full syllable] (almost unconditional - some RT words are not monosyllabic)
  • regressive voicing assimilation (e.g. sd > zd)
  • D > N / _[+stop]
  • C[-stop +voice]V > C[-stop +voice]V[+mid tone] / ?
  •  ? CwV > CwV[+high tone]
  • ʔV > V[+low tone]
  • C[+stop +voice] > C[+nasal] / _V[+low tone]
  • C[+stop +voice]V > C[+stop -voice]V[+mid tone]
  • hNV > NV[+high tone] / #_
  • md > mr > nr > ndr > nɖʐ
  • zt > zd > d
  • sm > sb > zb > b
  • tr > r / #_
  • q χ ɴ > g x ŋ
  • s > ʣ / #_, ? (or was it tz/dz > ʣ with initial d being lost elsewhere?)
  • j > dʑ / #_V[+full syllable], ?
  • ∅ > j / #_i
  • Simplification of consonant clusters in the onset leads to new voiced consonants.
  • hV > hV[+mid tone]
  • h > ∅ / N_V
  • hC[-voiced obstruent]{w y ∅}V[+mid tone] > C[-voiced obstruent]{w y ∅}V[+high tone]
  • t d k g > ts dz tɕ dʑ / _{V[+front] j}
  • s > r / #_V, ?
  • r > ∅ / #_u, ?
  • ɬ > u / _#
  • ɬ > ∅ / #_j
  • ɬ > l / #_, ?
  • ɬ > hl / #_, ?
  • ij uw aː> iə uə aə / ?
  • ja wa > iə uə
  • ij uw > ɛɪ ɔʊ / ?
  • aj aw > ɛɪ ɔʊ / ?
  • aL > a
  • a > o / _{p m}
  • VN(T) > V[+nasal](T) / _#
  • p t k > ʔ / _#
  • tj dj kj gj sj zj ŋj lj rj wj > ts dz tɕ dʑ ɕ ʑ nj j j j
  • x > ɕ / _V[+front]
  • x > h
  • kwu > ku

Evidence from the multiplier numbers (mìc nòc < *bʔik *dʔuq < *əbʔik *ədʔuk < *pʔik *tʔuk) suggests that all instances of a nasal followed by low tone vowel could go back to clusters of a voiced stop and glottal stop.

Construct state development

Construct state fell out of RT when unstressed vowels were deleted and initial consonant clusters were simplified. This completely distorted lenition patterns, they become unproductive and soon ceased to be used. Marking possession by simple juxtapostion without any morphological change is the sole reflex of the construct state.

To Rrób Tè Jĕhnò

  • Tʔ > Tʼ
  • Vɬ > V[+length] / _#
  • tr > dr / V_V
  • VN > V[+nasal +low tone] /_#
  • VNT > V[+nasal +what tone?]T / _#
  •  ? VNT > V[+nasal +?]T / V_V
  • C[+stop]N[+voice] > N[-voice]
  • hN > N[-voice]
  • NT > əT ~ əD / #_, ?
  •  ? ja aj > e / ?, eː /?
  •  ? wa aw > o / ?, oː /?
  •  ? ẽ õ → ɛ̃ ɔ̃
  • Vʀ > Vː
  • oɬ > oː
  • wi > uː
  • i u > e o / _#
  • V[+nasal] > V[-nasal +length] / lengthens only in open syllables
  • V[+?] > V[+low tone] / ?
  • V[-low tone] > V[+high tone]
  • V > V[+low tone] / N_C
  • D > N / _V[+low tone]
  • D T > T Tʰ
  • V > ə / _[-full syllable]
  •  ? l N > ə / #_C, ?
  • rk > rək
  • q > k / _i
  •  ? u > o / [+uvular]_
  •  ? səf > sf > fʰ / _[+full syllable] (? sC > Cʰ)
  •  ? sʔ > sʰ
  • kC[+obstruent -aspiration -ejectiveness] > Cʰ
  • kN[+voice] > N[-voice]
  • regressive PoA assimilation for stops
  • C[+stop]C[+stop] > C[+stop]əC[+stop]
  • C[+stop -aspiration]h > C[+stop +aspiration]
  • kw kʰ kʼw > p pʰ pʼ
  • ə > e / [+full syllable]

Retroflection and r loss

r stands for any liquid other than lʲ

  • tʰr tr tʼr > ʈʰ ʈ ʈʼ
  • tsʰr tsr tsʼr > ʈʰ ʈ ʈʼ
  • tɕʰr tɕr tɕʼr > ʈʰ ʈ ʈʼ
  • kʰr kr kʼr > ʈʰ ʈ ʈʼ
  • qr qʼr qʰr > ʀ
  • hr > r[-voice]
  • sr sʰr ɕr ɕʰr xr xʰr χr χʰr > ʂ ʂʰ ʂ ʂʰ ʂ ʂʰ ʂ ʂʰ
  • ɬr ɬʰr ɬʲr ɬʲʰr > ꞎ ꞎʰ ꞎ ꞎʰ (ɬ[+retroflex])
  • lr ʎr > ər
  • rt rtʼ rtʰ > ʈ ʈʼ ʈʰ
  • rts rtsʼ rtsʰ > ʈ ʈʼ ʈʰ
  • rtɕ rtɕʼ rtɕʰ > ʈ ʈʼ ʈʰ
  • rk rkʼ rkʰ > ʈ ʈʼ ʈʰ
  • rs rsʰ rɕ rɕʰ > ʂ ʂʰ ʂ ʂʰ
  • nr rn > ɳ
  • r > ∅ / {V C}_C

Palatalisation

  • s sʰ tʰ t tʼ ɫ > ɕ ɕʰ tsʰ ts tsʼ ɬʲ / _V[+front]
  • ts > ʦ
  • sj sʰj tʰj tj tʼj ɫj > ɕ ɕʰ tsʰ ts tsʼ ɬʲ
  • k kʰ kʼ > tɕ tɕʰ tɕʼ / _V[+front]
  • kj kʰj kʼj > tɕ tɕʰ tɕʼ
  • klʲ kʰlʲ kʼlʲ > tɕ tɕʰ tɕʼ
  • nj ŋj > ɲ
  • n ŋ > ɲ / _V[+front]

χ loss

  • χ > ɕ / _V[+front]
  • χ > h / ?
  • χ > f / ?

To Rof Njiŋe Wek

  • CʔV > CV[+creaky]
  •  ? i > oː / q_
  •  ? pa > pu
  •  ? V[-creaky] > V[+breathy] / [+stop]_
  • NT > NÞ > Þ / _#
  • (ʔ/h)V > V[+breathy] / #_
  • rV > V[+creaky] / #_

To Ree Rɛɛ Kıbyaa

  • sn sm > st sp
  • m > v / _VN
  • m > ∅ / #_u
  • D > T
  • Tʔ > D
  • Q > K
  • i > ɨ / K_
  • i > ɨ / ?
  • tr > tw
  • s > ʃ
  • ɬ > s / #_
  • ɬ > l / _#
  • VN > V[+length]
  • p’ t’ k’ b d > b d g v r / V_V
  • n > r / ? #_V
  •  ?i ?ɨ ?e > ı u a
  •  ?iː ?ɨː ?ɛː > ɛː aː ːɨ

Important correspondances

The correspondance
RT táen pass htáen, attr gláen, RTJ tràì, MT thang 'doubt'
conflicts with some of the posited sound changes, so they'll need to be changed to say the least. Currently, each word requires a slightly different reconstruction: tajN, trajN, taŋ

  • RTJ zr - MT r + tense vowel < *dr
  • RTJ cr - MT l
  • RTJ length - MT breathiness < *X
  • RTJ ê - MT i < *E
  • RT a - RTJ á - MT aj < *aL
  • RT ou - RTJ ó - MT ú < *oɬ
  • RT kw - RTJ p - MT h < ? *kp
  • RT kw + low tone - RTJ pʼ - MT ʔ < ? *kpʔ
  • RTJ í - MT ú
  • RT s - RTJ h - MT h


RT - pwanc, pl dànc- RTJ blád, dlád - MT pán /pʰan/, all meaning child. Note the parallel irregularity in RT and RTJ. PR was probably *bwan~blan, with the sonorant changing in either RT or RTJ and lengthening the vowel in MT.

RT pienc - RTJ bád, dlyád - MT púyan /pʰujən/: orphan - a derivative from *bl/wan using a infix?

RTJ àì# - MT əN

NB There is an element which made a vowel breathy in MT and diphthongised it in RT in most cases. Could it be a coda /h/?


  • RT dz - RTJ x - MT s
  • RT dź - RTJ hx - MT ds

In what way is breathiness connected with the coda?

  • ang /əŋ/ rightness, righteousness - grágh < *qraɴq
  • āng /ə̤ŋ/ hurt, wound, injure, harm - grág < *qraŋk, *?aŋk, *qhraŋk

What are the conditions for prothetic vowels and why does CrV → CV[+length] not apply? Are RT fricatives rather voiced because of VsV → VzV → zV?

  • asakh /əsəx/ wolf - zac, hsac - srég < *(a)srVk
  • asac /əsəʔ/ eagle - zoc, hsoc - ságh < *(a)sVq


kr → ʈ in RTJ?? There are a few other examples besides this one of a k(ʼ) ~ tr(ʼ) correspondance.

  • gak’ókh /kək’ox/ hunt - getrʼóg < *kakʔrok

How to interpret the onset:

  • gen /kɛn/ be different, be unusual, be another - ngàì


  • MT yakh /jəx/ throat - RTJ yág /ják/ id. < *jak
  • MT yaŋ /jəŋ/ be happy - RTJ yágh /jáq/ id. < *jaɴq
  • MT yaŋ /jəŋ/ happiness - RTJ grágh /ʀáq/ id. < *?aɴq

however:

  • MT tokh /tʰɔx/ wood - RTJ dóg /tók/ id < *dok
  • MT tong /tʰɔŋ/ ant - RTJ dóg /tók/ id < *doŋk
  • MT tung /tʰʊŋ/ vomit - RTJ dúgh /túq/ id < *duɴq
  • MT sung /sʊŋ/ pool, pond - RTJ súg /súk/ < *suŋk
  • MT súkh /sux/ be slow - RT sruc /ʂuʔ/ - RTJ srúg /ʂúk/ < *sruk NB! In mT ru → ū → u due to CrV → CV[+length]
  • MT sung /sʊŋ/ pool, pond - RTJ súg /súk/ < *suŋk
  • MT safai /səfaɪ̯/ source, origin - RTJ hfáí /fʰɛ́/ id < *səfaj
  • MT safeng /səfɛŋ/ thanks - RTJ hfég /fʰék/ id < *səfeŋk
  • MT rúmbú /rumbu/ wrist - RTJ rrĕbú /rəpú/ id < *rumpu

Tracing the origin of RTJ initial ĕwĕ-:

  • MT rómbáth /rombaθ/ palace - RTJ ĕwĕbád /əwəpát/ id
  • MT róla /rolə/ raven - RT ogá id - RTJ ĕwĕqa /əwətɕʰa/ id
  • MT pgo /pʰkɔ/ stick - RTJ ko /kʰo/ id

It's interesting that no word in MT begins in p, only ɸ, pʰ and pʼ.


RTJ has retained a distinction that MT and RT have merged:

  • MT nath /nəθ/ teeth - RT nac /naʔ/ id - RTJ nèd /nèt/ id < *net
  • MT nath /nəθ/ both - RT nac /naʔ/ id - RTJ nàd /nàt/ id < *nat

Evidence of reduplication in MT:

  • MT mummung /mʊmmʊŋ/ squirrel - RT mwinc /mwĩʔ/ id - RTJ mòg /mòk/ id < *muŋk
  • MT nonnong /nɔnnɔŋ/ girl - RTJ nògh / nòq/ id < *noɴq

Vowel deletion leading to consonant clusters, also a r - ʈʂ correspondance and unstressed vowel reduction:

  • MT khirúm /xɪrum/ fast (abstain from eating) - RT grunc /grũʔ/ id - RTJ kĕzrúb /kʰəʈʂúp/ id < *khirump

One possible origin of voiceless nasals?

  • MT dandí /təndi/ summer - RT ndàe id - RTJ hnò /n̥ò/ id < *(ta)ntV
  • MT dandu /təndʊ/ admit, confess - RT ndù id - RTJ hnô /n̥ô/ id < *(ta)ntu
  • MT ē /ɛ̤/ feeling, sense, intuition - RTJ ĕháí /əhɛ́/ id < *Vhɛ

Numerals comparison

RTJ RT MT ARN pre ARN RNW RRK ƏT RTI PR
1 p’íg pìc p’ikh pʲik *pjik² pęq bik bek piš *pʔik
2 káe x̠i *qi qó: kıı kuś čey *qis
3 t’úgh tòc t’uc tʷ'uq *tʔuq tįq duk dok truk *tʔuq
4 pyáo phau ʃʷaʃ *xwaʃ³ púl pal pel pal *pjaɬ, *paɬ
5 fàì fyein fang ʃʷax̠ *fjax̠ fu huu fən huu *fjaŋ, *faŋ
6 dlyă sa ŋʷoʃkʲe *ʃe ɬu sa ɬa sa *ɬja, *ɬa
12 dlád lác sath ʂat sat ɬat šat *ɬat
18 khó kóu (24: x̠oʃ) kol qul kol *quɬ
24 crà tróun thám (36:sram) twaa trən tree *tram

¹ In RTJ none of qʰ q qʼ appears before /i/.
² The singulative infix -j- was inserted, which deleted the glottal stop.
³ The plurative infix -w- was likely inserted, which caused the dissimilation pw > xw to occur, which was subsequently assimilated to ʃw, due to the influence of coda ʃ. This onset ʃʷ later replaced the original onset in the word for 5.

4, 5 and 6 all seem to have had variants with and without the singulative infix.

Pronouns comparison

Onset correspondances

Only some basic types are shown here, without, for example, including the correspondances that are given rise to by palatalisation. Uvulars and velars merged in MT and RT, so the uvular series is not presented either, being completely analogous to the velar one.

MT RT RTJ PR Comment
ɸ θ x p t k +high tone pʰ tʰ kʰ p t k
pʰ tʰ kʰ p t k +mid tone p t k b d g
pʼ tʼ kʼ p t k +low tone pʼ tʼ kʼ pʔ tʔ kʔ sometimes MT has ʔ instead
m n ŋ m n ŋ +mid tone m n ŋ m n ŋ
m̥ n̥ ŋ̊ m n ŋ +high tone m̥ n̥ ŋ̊ hm hn hŋ
pʰ tʰ kʰ m n ŋ +low tone p t k bʔ dʔ gʔ merged with b d g in almost all branches

Intervocalic correspondances

MT RT RTJ Comment
r ʈʂ
r ʈ
mb p
nd t
jV[+breathy] h h
l g tɕʰ
r ʈʂ

Rhyme correspondances

Coda correspondances

MT RT RTJ PR Comment
m n ŋ ŋ ~ m n ŋ ɴ There are some hints of former nasalisation in RTJ.
m n ŋ ŋ p t k q mp nt ŋk ɴq
ɸ θ x ʔ ʔ p t k q p t k q

Lexicon

The partially reconstructed lexicon is available here: Rompian Etymological Dictionary

Internal reconstruction

In all likelihood the liquid series comes from voiced fricative series:

w r l j ʀ < β z ɮ ɣ ʁ

Firstly, this makes the reconstruction of two stop series - voiceless and voiced - all the more more likely. And secondly, it raises the question was there ɦ, and if it was present, what are its reflexes? Some or all of w, r, l, j could already have been present before this shift, however.