From AkanaWiki
Jump to: navigation, search

Proposed cognate sets & discussion



  • EMiw = Eastern Miwan
  • EV = Eigə Valley
  • Far = Faraghin
  • FMiw = Forest Miwan
  • Ng = Ngauro
  • NT = Ndak Ta
  • OEMiw = Old Eastern Miwan
  • PEI = Proto-Eigə-Isthmus
  • PEV = Proto-Eigə Valley
  • PI = Proto-Isthmus
  • PM = Proto-Miwan
  • PNg = Proto-Ngauro

Possible Miwan-Faraghin cognates

  • I've found what's almost certainly a Miw/Isthmus cognate: [Eastern] Miwan tun "red" and Faraghin rušnen "paint, color" (probably from Proto-Isthmus *dzusn- or *ɖusn-).

  • I've just had a startling realization: given some of the historical changes I'm considering for the Miwan languages, Miw and Ferogh may actually be cognates! In Miwan, I'm thinking proto-vowels *e *o would become *i *u; if we posit a proto-Isthmus form for Ferogh like *fejog-s or *s-pejog-s, we can imagine an early Miwan cognate *m-fijug or *m-pijug (positing a nasal prefix of some kind)... with lenition of the final *-g and deletion of the consonant after *m-, voilà, Miju...!

The root *pejog/*fejog might just mean "people" or "tribe", or be an unanalyzable name (like Ndak).

Proto-Isthmus *(V)s- (which must have been fairly frequent, given the amount of *s-induced lenition of initial stops in Faraghin) is probably related to (or identical with) the prefix *as- that marks the genitive/accusative case on pronouns. I think of it as a kind of relational marker: if the (pro)noun it's attached to stands next to a verb, it's the verb's object; if next to a noun, it's a genitive possessor. Attached to a verb, it might be used as a kind of subjunctive/relative marker, or maybe form causatives or something.... But anyway, it's easy to imagine an ethnonym being used in genitive constructions so much that that form became generalized: "to be (one) of the Pejog" could be the normal way to phrase ethnic identity.

Basilius responds:
What if the PI form of the ethnonym was *s-mejog-s? This would eliminate the need for obscure prefix in Miwan (perhaps, in PI too - if *s- is allowed to be part of root sometimes?). I couldn't find any precedents for initial *s+nasal.
Also, the genitive/accusative marker could form adjective of even substantive ("property of..."), with appositive use; its use as acc. can be due to contamination of original finite forms with analytical ones, based on some verbal noun. --Basilius 16:21, 10 March 2009 (UTC)

Here's a few more, not previously posted:

  • FMiw za:ska: "war" could be cognate with either Far daghan "stay, camp" (< PI *dasg-), or Far nirgha "war" (if < PI *njalsga)-- this last is why I tentatively proposed PEI *nj- > Miwan z-.

(It's now proposed that nirgha is from PI *njadsga or *njadzga, and that za:ska is its cognate in Ngauro which was loaned into Miwan.)

  • FMiw pasta "cloth" as a cognate with Far baš "flat" (< PI *bas, *bes, *bos) looks pretty straightforward.
  • EMiw timpi:za "necklace" might possibly begin with the root seen in Far demen "give" (< PI *dim-); so might FMiw dimbal "wife", but if so the voicing is irregular.
  • FMiw kwintas "bird" might share a root with Far ghantač "dragon" (< PI *s-gant-, *s-gent, *s-gont-), maybe PEI *kwen-t(a)-.
  • More speculatively, FMiw mul "oblivious" might be somehow related to Far muns "soft, weak" (< PI *mun-ts).
  • Similarly, maybe FMiw ni:glu "feeble" and Far ner "bad, weak" (< PI *nidz, *niɖ) are related.

Possible Ngauro loans in Ndak Ta with Miwan or Faraghin cognates

  • NT kaime "study", and maybe also kenla "teacher, scribe", seems to reflect the same root as FMiw ki:mat "comprehension" and probably also kim "recipe, spell, algorithm, prescription".
  • NT mos "city, town" seems plausibly related to Far mašt "house, palace" (< PI *mas, *mes, or *mos, with a mystery suffix *-t added sometime after Western Isthmus).
  • NT naka "god" could possibly be cognate with Far nagat "lord, chief".
  • NT diàka "king, ruler" I've already noted as possibly cognate with Far čark "king" (probably < PI *ʈalk).
  • NT ntindo "metal" seems similar to FMiw di:ndi:n "metal", but the borrowing, if it is one, could be in either direction.

(Zhen Lin comments: "/di:ndi:n/ sounds ideophonic, perhaps, representing the "sound" of metal.")

  • NT alpau "make, create" might just possibly be somehow related to Far foghan "make, craft" (if the latter < PI *s-pug-s-).
  • NT sai "female" is almost certainly cognate with Far šoi "woman" (< PI *saj, *sej, *soj)... which suggests that NT dado "male" probably reflects the Ngauro word for "man".

(However, Radius Solis comments: "...šoi's etymon is sai as recently as pre-proto-Ferogh - no need to go all the way back to PI to find that exact form, let alone PEI. And PPF was roughly contemporary with early Ndak Ta. So I think it more likely that PPF borrowed it from NT rather than it being native to the EI family.")

  • NT eplain "ball" seems similar to EMiw plaj "bead"; but again the loan could be in either direction. (If it's a Ngauro loan, the initial vowel in eplain might be epenthetic: Ndak Ta doesn't allow initial clusters like /pl-/.)
  • Finally, NT santa "legal", santau "illegal", and sanlena "public" seem to share a root *san-, but the derivational morphology doesn't appear to be native.... Just possibly, the root is the same as Far šan "fire" (< PI *san, *sen, *son); the idea might related to some kind of metaphorical connection between fire/cooking/sacrifice and public/legal actions...?

An alternative (maybe better) Faraghin cognate might be rantan "test (by experience), prove" (if < PI *dzant- or *s-dant-); it's just possible that FMiw zinti: "taboo" is also related...

Proposed etymologies

cedh audmanh proposes:

Drawing on the established changes, I'd like to posit the following etymologies:

  • PEI *njadz-ga "war" > PI *njadsga > Far. nirgha; PNg. **nzatska > Ng. *zaska. PEI *-ga is a suffix of some sort, maybe a verbal noun, in which case the root *njadz- might mean "to fight".
  • PEI *dzus-n "red" > PI *dzusn- > Far. rušn-en "color"; PEV **tsun > EMiw tun (possibly with a lengthened vowel due to the loss of coda s). PEI *-n might be an adjectivizing suffix.
  • PEI *dim- "to give" > PI *dim > Far. dem-en; PEV **tim- > EMiw timpi:za "necklace"
  • PEI *gwent-a-ts "bird" > PI *s-gont-a-ts > Far. ghantač "dragon"; PEV **kwentats > FMiw kwintas. From this we can surmise that the voicing change for EV plosives is blocked medially after nasals. The change of coda ts > s might be a common EV phenomenon; here it is attested for Miwan, but the same thing happens in PEI *njadzga > Ngauro *zaska. If *-a is an agentive suffix, the root might mean something like "to fly". I don't know how to interpret *-ts.
  • PEI *nag-a > PI *naga-t > Far. nagat "lord, chief"; Ng. **naka, hence NT naka "god". The *-a suffix might be a honorific, or maybe a case ending, possibly agentive (see above). The meaning of the PEI word could be something like "father", but it appears to have had a fairly wide semantic range.
  • PEI *trelk-a "leader, chief" > PI *ʈelka > Far. čark "king"; PEV **djelka > Ng. **diaka, hence NT diàka "king".
  • PEI *dzant- "to decide, to judge" > PI *dzant > Far. rant-an "to test; to verify; to prove"; PEV **tsant > Ng. **santa "legal", **santau "illegal", **san-lena "public" (all borrowed into NT).
  • PEI *gejm- "to know, to understand" > PEV **kejm- > FMiw ki:mat "comprehension"; Ng. **kajme "to study" (hence NT kaime). Here we have a new change for Ngauro: ej > aj.

From these words it appears the root structure of PEI should be extended to C(r,l,j,w)V(n,r,l,j,w)C.

Radius Solis responds:

That looks excellent and I see no problems with any of those etymologies.

For *gwent-a-ts, perhaps the -ts is an augmentative suffix in Isthmus that is responsible for turning "bird" into "dragon". Then the EV version might have lost its meaning over the years, or augmented forms came to be seen as default, or the -s could even be another morpheme entirely.

Corumayas responds:

I agree, it looks great overall. Thanks! Some detaily comments:

  • PEI *njadz-ga "war" > PI *njadsga > Far. nirgha; PNg. **nzatska > Ng. *zaska. PEI *-ga is a suffix of some sort, maybe a verbal noun, in which case the root *njadz- might mean "to fight".
  • PEI *gwent-a-ts "bird" > PI *s-gont-a-ts > Far. ghantač "dragon"; PEV **kwentats > FMiw kwintas. From this we can surmise that the voicing change for EV plosives is blocked medially after nasals.

I'd been thinking that the EV voicing change only occurred in onsets, so the PNg. form ancestral to FMiw za:ska: might be *nza(d)zka, with *z assimilating to the voicelessness of the *k. Similarly, if my version is true the *t in FMiw kwintas isn't voiced because it's originally the coda of PEI *gwent-, and the *t in Far nagat might go back to PEI and be preserved as a final glottal stop in Ngauro (which wouldn't be reflected in the Ndak Ta loan).

New note: In case I might want to preserve *nj- clusters in Ngauro, I'm considering calling the initial *n- in PI *njadsga a prefix of some kind, so that the Ngauro cognate can be derived from PEV *jadz-ka.

  • The change of coda ts > s might be a common EV phenomenon; here it is attested for Miwan, but the same thing happens in PEI *njadzga > Ngauro *zaska. If *-a is an agentive suffix, the root might mean something like "to fly". I don't know how to interpret *-ts.

Radius Solis wrote: For *gwent-a-ts, perhaps the -ts is an augmentative suffix in Isthmus that is responsible for turning "bird" into "dragon". Then the EV version might have lost its meaning over the years, or augmented forms came to be seen as default, or the -s could even be another morpheme entirely.

I like that idea, Radius. *-TS is an attested nominalizer in Isthmus, which was still productive in Faraghin (where it's glossed as "quality"), so it might well be the same suffix as that. It's also possible that the suffixes were added independently in both daughters rather than going all the way back to PEI, and are either different morphemes or acquired different meanings in each branch.

New note: However, looking at it again I've noticed that Far ghantač can't come from *gwent-a-ts at all; it has to come from PI *s-gonta-ʈ, *s-gonta-tj[a,e,o,u], or *s-gonta-tuj. So we are certainly dealing with two different suffixes here.

  • PEI *dzus-n "red" > PI *dzusn- > Far. rušn-en "color"; PEV **tsun > EMiw tun (possibly with a lengthened vowel due to the loss of coda s). PEI *-n might be an adjectivizing suffix.

EMiw tun doesn't have a long vowel though. I'm skeptical of explaining Miwan long vowels as due to compensatory lengthening anyway... the correspondences are complex and difficult to predict, and I think the reason is that length is actually part of the tonal system (which we should probably work out at some point). (Of course, deleted coda consonants are highly likely to affect tone, so they are relevant; it's just more complicated than *deleted coda > long vowel.)

  • PEI *nag-a > PI *naga-t > Far. nagat "lord, chief"; Ng. **naka, hence NT naka "god". The *-a suffix might be a honorific, or maybe a case ending, possibly agentive (see above). The meaning of the PEI word could be something like "father", but it appears to have had a fairly wide semantic range.

I don't think "lord" and "god" are actually too distant semantically, and neither need invoke the concept "father" (for which we have a perfectly good PI root already in *BAdAu). (Also see above about the final *t: the Ngauro word might be *nakaʔ).

  • PEI *trelk-a "leader, chief" > PI *ʈelka > Far. čark "king"; PEV **djelka > Ng. **diaka, hence NT diàka "king".

This has become the most troubling correspondence for me. Firstly, I'm not sure if Faraghin /r/ should come from *l at all (though if Radius approves of it I'll leave it alone). More bothersome, the *k looks like it should become /g/ in Ngauro. Maybe the final /a/ was added in the NT loan by analogy with naka, which would explain both its absence in Faraghin and the non-voicing of the /k/.

The PEV form should be *drelk(a), and the Ngauro word djak(a); the change *Cr > Cj occurs in Ngauro but not Miwan. I've been a little worried about what happens to the *l in the Ngauro word, but *elk > /ak/ is not unreasonable. The Proto-Miwan cognate might be **drilk(a).

  • PEI *gejm- "to know, to understand" > PEV **kejm- > FMiw ki:mat "comprehension"; Ng. **kajme "to study" (hence NT kaime). Here we have a new change for Ngauro: ej > aj.

I was thinking PEI *gajm-, with *aj > PM *e > FM /i/ (FM lacks /e/). In fact, I've been thinking about reducing the PEI diphthongs to just *aj *aw, since I think the others attested in PI could all potentially be derived from those plus the various vowel changes induced by prevocallic *w. On the other hand, if *j *w can occur as coda consonants then there seems to be no good reason not to allow them to follow any vowel. (Incidentally, note that the ken- in NT kenla is the same Ngauro morpheme as the kaim- in kaime.)

  • From these words it appears the root structure of PEI should be extended to C(r,l,j,w)V(n,r,l,j,w)C.

I'm amenable to that, except that the coda consonant is definitely optional (cf. PI pronouns *da "I", *fe "you[PL]", *njo "he/she"). Also, it seems like *s can show up in several places in the syllable... so maybe something like *(s)C(r,l,j,w)V(n,s,r,l,j,w)(C)...?

More possible cognates

TzirTzi writes:

@Corumayas (and anyone else working on the EI family) - I've been comparing the Miwan and Proto-Isthmus lexicons, and I was wondering if the following might possibly be cognates.

The following are patchy:

  • PI *DZJUF-en, 'scare', and OEMiw *traf, 'fire' (thus from PEI *dr(j)af)
  • PI *Vs-gun, 'old(er)', and FMiw asku, 'polite'
  • PI *mAGZ(-)Aʈ, *maGZ(-)atJU, 'loot, take as booty', and OEMiw *me:g, 'be loud, be roudy' (thus from PEI *mesg-)
  • an alternative etymology to the current one for EMiw timpi:za - PI *TSImp-en, 'twist' = stimp-en, < PEI stimp- > PM timp- + -i:za = nominaliser

And the following are very patchy:

  • PI *Vs-gIʈ, *Vs-gItJU, 'mighty, powerful', and FMiw aska:r, 'haunted' (thus from PEI something like *as-gar(dV))
  • PI *kIL(-)t, *kIDZ, *kIjV, 'anger', and FMiw guljad, 'defeat' (thus from PEI gVlj(V)d))
  • PI *Vs-gAF-en, 'steal', and OEMiw iskefli, 'lice' ('stealers (of blood)'?)
  • PI *(Vs-)gJU(-)TS, 'sharp', and FMiw askini:, 'painful' (thus from PEI *as-guj-, as there are no attested Eige reflexes of PEI *uj)
  • PI *tLJUB-en, 'marry', and FMiw dimbal, 'wife (vulgar) (thus from PEI *tluj(m)b, and somewhere tl -> t and uj -> u)

I realise that none of them are great, but looking for them passed the time - no worries if none of them are usable.

Basilius writes:

Another couple of crazy comparisons...

  • PI *balon 'iron' (if a compound based on 'glitter', -> 'metal') ~ *bu:r OEMiw 'to shine'
  • PI *dADZAd-a- dance ~ varja EMiw 'to dance' (the straightforward semantics does tempt one to add a correspondence or derivation pattern...)

--Basilius 17:32, 10 March 2009 (UTC)

Based on *dADZAd-a ~ varja I just had a discussion with Basilius how this could come about. Because initial v z are both very common in Miwan but have no established source other than loans from Ngauro based on PEI *w *j, we agreed that the most sensible solution for the above correspondence might be to posit PEI instead of *f, and also to add two voiced fricatives *ð *z. In Isthmus these would be eliminated by fortition to PI *d *dz, and a second shift of > PI *f would remove the orphan POA. In Eigə Valley both *θ *ð would shift to labials, with voicing not being reversed.
Such would be supported by the correspondence PI *dzant-a "to test, prove, try" ~ Ng. zanta "legal", even though this cognate set would also work with initial *dz ~ *s or *j ~ *z. The evidence must therefore come from Miwan - and we've found a promising candidate: FMiw zinti: "taboo". The Faraghin ~ FMiw correspondence would work out if the PEI form was *zent-e-, though Ngauro would preserve the [e] in both syllables rather than lowering it to [a]. The semantic connection looks tempting nonetheless...

--Cedh 00:31, 11 March 2009 (UTC)

Hmm. I'm hesitant to change the phoneme inventory of PEI that much. For what it's worth, /v z ʀ/ in late EMiw may also come from earlier /b d g/ (which the EMiw wordlist lacks), so Ngauro loans aren't necessarily the only source. As an alternative explanation, maybe the first syllables of varja and *dADZAd-a are different prefixes, and the cognate root is *jad-.... I'm not sure about zinti:, either, since the vowels don't match; but maybe the vowel correspondences should be more complicated than they are right now.

-- Corumayas 22:06, 14 March 2009 (UTC)

As for /v z ʀ/, it seems that it cannot be so simple... FMiw and Old EMiw appear to have both /v z/ and /b d g/ word-initially; zinti: 'taboo' comes from FMiw for which 'dance' is attested with initial v- (as varja:). Also, it seems to me that PM -rj- corresponds nicely with PI -dz-, so there's rather a suffix added in PI? Or maybe some contraction in PM? --Basilius 03:22, 15 March 2009 (UTC)

Ngauro approximant codas

With your current revisions, the reconstruction of PEI becomes more and more solid. Cool!
It just occurred to me that a pattern for Ngauro approximant codas appears to be emerging. Some food for thought:

  • PEI diphthongs.
    • *aj *aw > ai~e au~o
    • *ej *ow > i u
    • *uj *uw > əu
    • *ij *iw > no Ngauro reflex specified so far, except for a tentative *ij > ?e (where did that idea come from?), so what about both > əi in parallel to the above?
    • *oj *ew > what about əi əu?
  • PEI coda *r *l.
    • No reflex for these has been specified so far, but as *r becomes j in onsets, what about having the same change in coda position?
    • Coda *l could become w in parallel. All other early EI langs seem to preserve it as /l/.
    • I'd suggest the following:
      • *ar *er > *aj > ai~e
      • *ir *ur > *ij *uj > i
      • *or > *oj > əi
      • *al *ol > *aw > au~o
      • *il *ul > *iw *uw > u
      • *el > *ew > əu
      • The outcomes of *ir *ur *il *ul presume that liquid vocalisation takes place later than the diphthong changes listed above.
    • We have to account for *trelk > djak, so I'd suggest that coda vocalisation does not happen if the onset already contains a C(j,w) cluster. Instead, the liquid would just be deleted, with preceding *e *o lowering to a.
    • There is also Ng. al-pəu with a coda lateral. Either this one comes from a different source, or morpheme-final *l generally remains. Maybe also *r > l / _#.

-- Cedh 09:06, 6 November 2009 (UTC)

Meshi words and correspondences

I've had some thoughts on making Meshi sound changes more interesting, while keeping them consistent with what already exists.

The native Meshi words we have so far are

  • aysu "speak, communicate"
  • aku "base (of mountain), source (of river)"
  • buru "fortification; settlement"
  • chima "large quantity"
  • gise "defecate"
  • ini "child"
  • kug "jar"
  • met "woman"
  • sanu "village"
  • unu "shaman"
  • Meshi

Of these the only feasible cognates I can think of are...

  • *Meshi, of course, which derives from *ʔum-pew plus a suffix. I think this is most likely PEI *siŋ "holy" (I prefer this version to *sij because of the sound changes I propose).
  • chima ~ OEMiw -ten from a PEI root *d(l)en or *d(l)ajn, which works if we posit a suffix like *-baC or *-paC to account for the change in place of articulation - possibly PEI *-pat "one", so the compount would literally mean something like "a lot, a load".
  • unu might have have a first element derived from the genitive PEI *ʔum plus another element - perhaps *tow "sand" in the sense of "desert" or "wilderness".
  • buru could conceivably be a cognate of OEMiw bur "shine".
  • It's just about possible that kug is somehow related to *gad "flow, float", though the change in final consonant needs to be accounted for.

Some are problematic:

  • gise has no obvious cognates. The FMiw word for "shit", uf, looks completely different.
  • Similarly, aysu doesn't look much like the OEMiw word (re:) or the PEI (*twik, *ts(w)ik, *θwik).
  • It's also hard to find a cognate for aku, sanu, ini, or met.

So what about the sound changes? One of the key problems we have is that a lot of odd vowel-glides whose reflexes in Meshi are not established (*ij *oj *uj *uw and perhaps *ew), and no source of /aj aw/ in Meshi, even though these appear in aysu "to speak" and Tawfwe "Thabīa". I propose the Meshi sound changes, from the PEV stage onward, be modified to the following:

  • ts dz dz > t d z / (I don't mind much about the conditions for *dz changing)
  • w > v / initially and intervocalically
  • a e o > Ø / finally
  • nasals assimilate to a following obstruent
  • a number of consonants are lost, including in codas (in some but not all instances - conditions need to be worked out, although I think a lot of clusters should go) and probably after nasals (this makes the etymologies of chima and unu much easier).
  • kw gw > fw zw
  • e o > i u
  • t d s > ʧ ʤ ʃ / _i
  • ew iw oj uj ij uw > e e e e aj aw / (this change is important as it gives us phonemic /aj aw/, explains Meshi, and makes palatalised consonants phonemic)

Then some other changes to satisfy my aesthetic preferences:

  • θ ð > s z / (also makes palatalisation contrast in more environments)
  • ŋw > gw or w / (the latter makes /v/ phonemic)
  • ŋ ŋ > n Ø / (I'd like it to vanish finally so *siŋ > shi - since *sij > shay because of dissimilation - but this might be taken care of by the previous loss of final consonants. Don't really mind what happens elsewhere but it seems odd to retain it initially.)
  • r l > j w / C_ / (This gives a nice (C)(w/j)V(C) syllable structure.)
  • Possibly more alveolars palatalise before /j/, but we could leave it as contrastive.

This leaves us with consonants /p t ʧ k b d ʤ g f s ʃ m n r l j w~v/ and vowels /a e i u aj aw/.

thedukeofnuke (talk) 11:40, 13 March 2014 (PDT)

These are great. I like the first three of your proposed cognates. If the words don't correspond to any PEI roots, might as well make up some new roots, it's time we advanced our understanding of that language. I am not gonna do anything further with Meshi, please take it over if you want to develop it. (later, I may toss you a couple grammar ideas if I am so inspired.)
FMiw "uf" looks to me like an onomatopoeia, the grunt you might make while shitting. Maybe gise represents the original EI root. --Dunomapuka (talk) 23:56, 14 March 2014 (PDT)
Thank you - glad you like them; my little diversion of Ayčasamo has got me thinking hard about the languages of the Xōron-Meshi area, and I felt it'd be good to flesh Meshi out further. I'll add the changes to the main page, and add/edit some roots on the PEI lexicon page. Right now I haven't had many thoughts on grammar, but I'd welcome any ideas you have. thedukeofnuke (talk) 11:30, 16 March 2014 (PDT)

Other discussion


Corumayas wrote: "(Again, the velar nasal *ŋ is not currently attested, but we might want it anyway.)"

Do nasals in root-final position assimilate in POA too, as nasals in pre-final position do? Because if not, I just had the idea that clusters of *ŋ plus coronal *t *d *ʦ *ʣ could be another source for the Isthmus retroflexes... Cedh 08:38, 17 July 2008 (UTC)
You mean in compounds? That could work. I think the assimilation should probably only happen within morphemes. I was thinking maybe final *ŋ could become *j in PI, to be another source of diphthongs... Corumayas 22:17, 17 July 2008 (UTC)

Verbal morphology

Just putting up my current thoughts in case anyone has comments or suggestions.


This is established: PEI has a three-way distinction between active, causative, and some kind of passive/reflexive/anticausative, marked by the "stem vowel" suffixes *-a *-i *-u (respectively). It works with the little that we know of the daughter langs, and (oddly) bears a tantalizing resemblance to the "phase" markers in Proto-Peninsular (also *-a *-i *-u). It might be almost too neat a system, though; should it be a little messier?


I decided Faraghin has subject agreement, which probably goes back at least to Proto-Isthmus; but only the 3SG suffix -n is known so far. If we have -n < the PI pronoun *njo, what would the other agreement suffixes look like? Maybe they're all derived from pronouns. Let's stick with the PI level for now, since we don't have definitive reconstructions of the PEI pronouns yet. Here are the pronouns:

  SG   PL
1   *da   *guʈ
2   *tujn   *fe
3   *njo   *ludz

Now, we could just suffix these to the verb stem; the result would look like this after the Faraghin sound changes (using the verb katan):

  SG   PL
1   katad   kataguč
2   katačin   kataf
3   katan   katalur

But that's transparent and pretty boring. We could maybe make it more interesting by having the first and second person forms grammaticalized earlier, and thus be more worn down: if they somehow lost their final consonants before the PI stage, the Faraghin system might look like this:

  SG   PL
1   katad   katag
2   katat   kataf
3   katan   katalur

That seems a little too tidy though. Would more change help? If the voiced stops in the first person forms were somehow lenited away, and the plurals were leveled, we could have something like this in Faraghin:

  SG   PL
1   kata   kateu
2   katat   kateu
3   katan   kateu

That's starting to look more realistic to me. I'm not sure how justifiable it is, though, since it involves changes (lenition of voiced stops, dropping of final consonants) that don't seem to happen anywhere else in the language.

Another possibility is that first and second person agreement are much older than third person, and are not visibly related to the pronouns at all, or are merged somehow with the "stem vowel" suffix, or something. (That last version is already starting to do both of those, actually...)

Another option would be to contract the endings before the final consonants are lost. In this case the 1 and 3 pl endings at the PI stage might be something like *-uʈ and *-ldz, which would make the Faraghin reflexes of these forms of the sample verb kateuč and katar respectively.
Looking at it closely, even though the -euč suffix is cool, I think I still like your third table with leveled plurals best.
- Cedh 20:37, 10 March 2009 (UTC)
What about kateuč for all the plurals? (A parallel contraction would make the 2SG katoin... that's identical to the 3SG causative, though.)
- Corumayas 23:45, 10 March 2009 (UTC)

Let's see what happens in the other voices...

*-i (basic causative)

  SG   PL
1   kate(d)   kači(č)
2   katet/kačin   kači
3   katen   kači(r)

*-aj (extended causative)

  SG   PL
1   katoi(d)   katoi(č)
2   katoit/katoin   katoi
3   katoin   katoi(r)

*-u (middle)

  SG   PL
1   kato(d)   kato(č)
2   katot/kačin   kato
3   katun   kato(r)

*-iw (middle of causative)

  SG   PL
1   kači(d)   kači(č)
2   kačit/kačin   kači
3   kačin   kači(r)

The outcomes of *-aw *-ij *-uw would be identical to those of plain *-a *-i *-u, and the reflex of *-uj (causative of reflexive) would be identical to that of *-iw.

With this table, I'd be in favor of adding in the 1PL, but leaving off the -d and -r in the 1SG/3PL. As for the 2SG, I think the -t versions are more interesting than the -n versions. Cedh 22:16, 14 August 2010 (UTC)


Having subject agreement seems to mean that there should be some kind of tense/aspect marked on the verb somehow. I wanted something fairly fusional, and I decided to look at Muskogean's verbal ablaut/infix system for inspiration.

The idea I think I'm going to go with is to have infixed consonants before the final vowel of the verb root, which will probably mark aspect (not only is that what they're used for in Muskogean, it's also a basic category in most of the old langs of eastern Peilaš-- including Talo-Edastean languages and Proto-Peninsular).

There are six consonants that can cluster inside the syllable like that in PEI (though I guess technically it doesn't need to be syllable-internal if it's followed by a stem vowel, does it?...): *s *n *r *l *j *w. Of these, *n *l don't produce particularly interesting reflexes (they're unchanged in most of the descendents; although in Feråjin they're both deleted with compensatory lengthening of the preceding vowel). But *j *w produce diphthongs which give some nice vowel alternations, and *s *r produce consonant alternations in the Isthmus langs, while *s gives tonal contrasts in Miwan.

Just as a demonstration, here's what this might look like in Faraghin:

Unmarked   *S-Grade   *N-Grade   *R-Grade   *L-Grade   *J-Grade   *W-Grade
katan kasan kantan kačan kartan koitan keutan
letan lesan lentan lečan lertan letan litan
lupan lufan lumpan ločan lurpan lipan lupan
rifan rifan rimfan rifan rirfan rifan rifan

I'm inclined to use at least *s *j *w, and maybe also *n and/or *r... not sure how I'd assign aspects with that many slots in the paradigm though.

- Corumayas 08:05, 10 March 2009 (UTC)

We'll almost have to include the R-grade because Faraghin already has repetitive/iterative/habitual forms in -čan. I also like the L-grade, maybe more so than the N-grade. What we should keep in mind is that there are several verbs which phonetically fit the outcome of some grades, e.g. sempan "twist" ~ N-grade and krertan "seize, capture" ~ L-grade. If these are derived from roots with a simple coda such as *tsip- or *klit-, we might mess up some cognates. On the other hand, if the infixed sonorant is part of the PEI root, we will have to find a way to affix the aspect consonant to roots with complex codas. *tsimp-n-an? *klilt-il-an? *klilt-a-l-n? Other ideas?
- Cedh 20:37, 10 March 2009 (UTC)
I think everything depends on the way that infixation emerged. If e. g. through metathesis, then some pre-PEI had **kat-S-, **kat-N-, **kat-R-; then *tsimp- and *klilt- could be < **tsip-N-, **klit-l-, with aspect forms like **tsip-əN-S-, **klit-əl-S; **tsip-əN-N-, **klit-əl-N; **tsip-əN-R-, **klit-əl-R- etc.; and these could be partly subject to the same metathesis to produce *tsip-əN-S-, *klit-əl-S-; *tsip-əNN-, **klit-əl-N-; but **tsip-əRN-, **klit-əRl-.--Basilius 21:56, 10 March 2009 (UTC)
I was thinking the Faraghin forms in -čan were a later development, but I suppose they could be descended from the R-grade. One problem I see with R-grade is that (with the current set of sound changes, at least) stops adjacent to *r at all three POAs merge in Proto-Isthmus; you couldn't tell, say, whether a Faraghin form ločan was a form of lupan or of some other verb lotan or lokan. (Another problem is that the outcomes of *r in some positions in Isthmus langs are still up in the air...)
For any roots that already have a coda cluster, maybe the infix could go before it, with a reduplicated root vowel: *tsi-ni-mp-an, *kli-li-lt-an, etc.
Of course, the full system wouldn't be preserved in each daughter lang; just like with the voice system, only bits of it might survive, and it would probably tend to become derivational. The descendants probably use auxiliary constructions for a lot of their TAM stuff.
- Corumayas 22:19, 10 March 2009 (UTC)

Here's an analysis of the regular reflexes of the various grades in Faraghin. - Cedh 10:40, 15 August 2010 (UTC)

The W-grade doesn't survive into Proto-Isthmus except:

  1. when the root vowel is *i (giving e.g. Far. letan (< PEI *lit-a-) vs. litan (< PEI *liwt-a-))
  2. when the *w ends up intervocalic (giving e.g. Far. rantan (< *zant-a-) vs. rafantan (< *za<wa>nt-a-))

The J-grade normally simply turns PEI root vowels *e *a *o into Far. oi, and PEI *u into Far. i. However, when the aspect marker ends up in intervocalic position, the vowels remain unchanged and the marker is reflected as Far. r.

The R-grade turns root-final obstruents into PI retroflexes, and merges with the W-grade before other consonants. If the marker ends up as PEI *r or PI in intervocalic environments, it is generally elided - the former in PI, the latter in Ferogh only. As a result, the root vowel merges with the stem vowel, or with the echo vowel in roots with coda clusters, as described in the following table (top row: PEI, left column: PI, resulting forms: Faraghin; note that a few combinations have the fortified reflex of an intervocalic glide instead):

*-ra(w)   *-raj   *-ri(j)   *-riw   *-ruj   -ru(w)   *-rda(w)   *-rdaj   *-rdi(j)   *-rdiw/*-rduj   -rdu(w)
*a- a oi oi aro afe a a oi oi oi eu
*e- i i oi aro afe i a oi oi oi eu
*i- i i e i i i a oi e e i
*o- a oi oi aro afe a a oi oi oi o
*u- a oi i i i o a oi i i o
*jV- i i i iro ife i a oi e i o

The L-grade simply adds *l; if this is clustered with another consonant in Proto-Ferogh, it becomes r. Root-final *s is deleted in WI.

The N-grade simply adds *n; this assimilates in POA to a following obstruent. Root-final *s is deleted in WI.

The S-grade lenites *p *t *k *d *g into Far. f s kh r gh, and adds s in intervocalic environments. Before other root-final consonants it is not distinct from the zero-grade.

The L-, N-, and S-grades of PEI roots ending in regularly merge all stem vowels except *i into Far. i; for instance, a root *guŋ- (zero-grade: Far. gin) would form the L-grade golin, the N-grade gunin, and the S-grade gosin, with no inherited distinction between regular forms and extended *-aj-causative; the simple PEI causatives in *-i- might remain distinct as golen gunen gosen though.


Okay, I just had to fix that pa:ntun issue, I just realized it was still listed the wrong way around with my correction still given as parenthetical. The word for "guitar" was specifically intended to be a metonymy taken from the music a guitar makes, which the Miw found to be colorful. The music was supposed to be the main definition of the Miw word, with only the metonymic use getting borrowed. Radius 16:31, 6 November 2009 (UTC)