dhoklang scratchpad (NP Cappadocian)

[dhok]

Another shot at Satemophrygian, which I've been working on on and off.

Sound Changes

1) General core IE sound changes: *h₁e *h₂e *h₃e > ĕ ă ŏ; *eh₁ *eh₂ *eh₃ > ē ā ō; Szereményi's, Stang's, Osthoff's, Pinault's, etc. The usual, basically. Interconsonantal *H of any sort becomes *ă (no triple-reflex).

2) Lex Rix: #HC > HeC. Shared with Greek, Armenian and Phrygian. Laryngeals then drop.

3) Accented short *ó becomes lengthened ó:.

4) Satemization. *ḱ *ǵ *ǵʰ > intermediate affricates which we will represent as *ĉ *ĵ *ĵʰ because only Esperanto provides a decent diacritic for <j>. *kʷ *gʷ *gʷʰ merge with their plain counterparts as /k g gʰ/, but only after *o(:) is raised to *u(:) after a labiovelar. (Both Armenian and Baltic have traces of roundedness in both the labiovelars and *o, so despite the fact that Cappadocian is satem I'm going to go for this being legal.) RUKI also applies: *s > š / {R u K i}_.

5) Winter's Law: before plain voiced stops vowels lengthen.

6) *o(:) merges to /a(:)/.

7) Cappadocian consonant shift (shared with Phrygian): *T *D *Dʰ > T T D.

8) *t (whether from *t or *d) becomes /s/ before /i(:) y/. *d (from *dʰ) becomes /z/. Original *s becomes /š/ in this position.

9) *ĉ *ĵ become /š ž/. At some point /ž/ then becomes /z/, merging with *z from assibilated *dʰ.

10) *l becomes /n/; *r becomes /š/ word-finally and /l/ otherwise. (cf. Massachusett)

11) *m becomes /n/ word-finally. Word-final *ns > s.

12) *NC > C; if C was voiceless it voices (except for *š).

13) Unaccented *ĕ > ĭ. *ăi *ĕi > e i; *āi > ai.

14) Word-final long vowels shorten.

probably a bunch of other stuff I didn't think about. *w *y should drop after consonants although this is likely to fuck up the verb system (how do you distinguish *bhéreti types from *gʷʰédʰyeti other than after coronals?)--maybe they should create geminates, which then go *VC₁C₁ > V:C₁, so that you get lengthened-grades. syllabic consonants go to /aR/

nouns are mostly boring so we're going to start with verbs (also fuck pronominals). paradigm dump



(edit: man it's 2018 and the board still can't do tables?)

as in Latin, the "preterite" is either the perfect or the aorist stem depending on verb. even if it's from the perfect stem, it gets the augment. probably there should be subjunctive or optative formations coming off the preterite stem to make aspectual distinctions; as in Greek these would just go bibāla bibālēs bibāle bibālāmas bibālēti bibālāzi. note the unusually archaic optative and the 2sg and 3sg endings shared with Greek and Armenian.

there are also mi-verbs in the present system. these are fairly normal except that they go -mi -ši -si -mas -ti -zi. not sure how to do their subjunctive or optative. probably get a root extension out of nowhere--maybe y or w? e.g. sītāmi 'I give' > sītāwa 'that I give'. this could be by analogy from roots ending in *w, particularly būmi

real passives are formed peraphrastically, by using the middle participle along with the auxiliary verb būmi, which has no other function. this goes būmi būši būsi būmas būti būzi. as noted, there is a subj. būwa būwis etc. the perfect stem on that should probably be an aorist ibūša.

[missals]

pres/ind and pres/subj looking very ripe for merger with just a small vowel change - Western Romance style /a/ with /a:/, /i/ with /e:/, and it would be an almost total merger

is ibi- the "augment"? what's the diachronic source of that?

[Frislander]

I like this, particularly rule 2, I hadn't thought about that change that way before. I can't wait to see what happens in another few thousand years if you bring it up to the present day.

is ibi- the "augment"? what's the diachronic source of that?

Looks like the augment was added onto the reduplicated perfect (see ibelam in the imperfect) which then became the preterite as in Latin.

[dhok]

Yes, the augment gets added onto all preterites regardless of whether they come from old aorists or old perfects. This means that perfect-derived preterites like that of bela look like they're supposed to be pluperfects...but they're not; pluperfects don't exist. (At least I don't think so. Maybe there's a periphrastic pluperfect...I'm still not sure what I want to do with the future, if I even have it. What are ways of getting futures? Latin uses either the old subjunctive or a suffixed form of *bʰuH-...)

Presumably, as in Greek, you should be able to use the perfect stem to form subjunctives, optatives, etc. which wouldn't take the augment. The question is whether the aspect implied would be perfect (like perfect subjunctives/participles/infinitives in Latin) or perfective (like aorists in Greek). Armenian seems to have an aorist, which forms a subjunctive, but it's usually formed with the suffix -tsʰ-. There don't seem to be any relics of the perfect in Armenian...

Not many mergers for the indicative and subjunctive, I'm afraid, except for the 1sg, but that merger is quite old--look at Greek, where λύω can be either subjunctive or indicative. Sanskrit apparently has -āni in the 1sg subjunctive...no idea where that one's from. I did consider scrapping the subjunctive and just using the optative for all irrealis constructions.

I suppose I could switch from Greekish endings -a -is -i to Sanskritty/Hittitesque -ami -iši -isi; that'd give you subjunctive -āmi -ēši -ēsi. The "thematic++ subjunctive" is best-known to me from Greek and Latin (where it has gotten mixed up and reshuffled a bit); where else does it show up? Anatolian doesn't have it, Gothic's "subjunctive" appears to derive from the optative, and...Slavic? Baltic? Tocharian? It's 9:30 here so I'm going to press submit on this and hit the hay.

[dhok]

burp

We'll make a new rule that word-final long vowels do not shorten if they were accented. This, plus the "unaccented *ĕ goes to /i/" rule and "*ó goes to /ā/ rule" splits ā- and o-stems into two different classes each. Declensions a1 and o1 were stem-accented, a2 and o2 ending-accented.

In the a-stem feminines, a new nominative plural -ai develops, as in Greek and Latin, replacing earlier *-ās. This is then used as the basis for the acc. and gen. plural, which tack their old endings onto the nominative plural ending. (Hey, it more-or-less worked for Sanskrit.)



Note that the two classes are the same except in the nom. and voc. sg. (I don't know whether Cappadocian still has real accentuation by the time it's attested.) In the verbal system the assumption I've made is that the accentuation is pretty consistently on the root. I suppose if I wanted to upfuck things I could do a Sanskrit and have thematic verbs have both theme-vowel-accented and root-accented classes (cf. bhárati vs. tudáti)--because of the differing reflexes of accented and unaccented *ĕ and *ŏ this would give you, e.g., bela vs. tūtā (*bʰéroh₂, *tudóh₂); the 1sg, 1pl and 3pl of tudáti-type verbs would be identical in the subjunctive and indicative.

O-stems:



Note that the stem-accents and the ending-accents are very different. The expected outcome of the nom/gen/acc in ending-accents goes -ās -āša -ān -ai -ān -ās, where the gen. pl. is identical to the acc. sg. and the acc. pl. to the nom. sg., which would be a problem, so the gen. and acc. plural were reshaped along the lines of the ending-accented feminines, which they resemble.

ainas sā klēs kītālis pegi seš siptan aštā newan tešan īšazi šadān. šnewas agditān

esmi eši ēsi/ismās iste isezi

[Xwtek]

I am confused with your conjugation. Can you give the verbal template?

[Frislander]

I am confused with your conjugation. Can you give the verbal template?

Well that's the problem - in IE there is really no verbal template per se outside of root + derivation + inflection, since all the inflectional categories of verbs are marked in a single inflectional suffix. However I will elucidate:

The table above is set up as per traditional IE studies: the 6 forms given per tense/aspect/mood/voice combination are person/number forms, with person down the side as 1st, 2nd 3rd and number across the top as singular plural. So the present indicative active is like so:

Code: Select all

  SG    PL
1 bela  belamas
2 belis beliti
3 beli  belazi

As for the actual TAM/voice categories themselves, they are fairly standard IE ones: For tense aspect, PRES = present, IMPF = imperfect and PRET = preterite. For moods IND = indicative, SUBJ = subjunctive, OPT = optative and IMP = imperative (the last one only shows 2nd and 3rd person forms). Finally for voices ACT = active and MID = middle. There are two participles (PART), an active and a middle, which inflect like nouns.

Hope that sheds som light.

[Kuchigakatai]

(edit: man it's 2018 and the board still can't do tables?)

You can use the [ code ] tag and take advantage of the monospaced font it provides...

Yes, I know I should figure out the code to add working tables as tags, and also glosses as in the old Zeeb.

[Xwtek]

I am confused with your conjugation. Can you give the verbal template?

Well that's the problem - in IE there is really no verbal template per se outside of root + derivation + inflection, since all the inflectional categories of verbs are marked in a single inflectional suffix. However I will elucidate:

The table above is set up as per traditional IE studies: the 6 forms given per tense/aspect/mood/voice combination are person/number forms, with person down the side as 1st, 2nd 3rd and number across the top as singular plural. So the present indicative active is like so:

Code: Select all

  SG    PL
1 bela  belamas
2 belis beliti
3 beli  belazi

As for the actual TAM/voice categories themselves, they are fairly standard IE ones: For tense aspect, PRES = present, IMPF = imperfect and PRET = preterite. For moods IND = indicative, SUBJ = subjunctive, OPT = optative and IMP = imperative (the last one only shows 2nd and 3rd person forms). Finally for voices ACT = active and MID = middle. There are two participles (PART), an active and a middle, which inflect like nouns.

Hope that sheds som light.

Well, it explains why the pronomial ending is so inconsistent. However, I saw in Wikipedia that subject agreement in Latin doesn't fuse with aspect, only tense. Also, I usually segment verb conjugation because verb normally has hundreds or thousands of form, and memorizing it one by one will be impractical. For example, what is the 2ND > 3SG reprocial inchoative causative subjunctive form of to take a bath. If the language lacks conjugation this complex, in natlang, there will be hardly any fusion.

Language like this normally simply simplify the suffix, for example

Code: Select all

PRES.IND
ACT SG  ACT PL    PASS SG   PASS PL
bel-a-0 bel-a-mas bel-a-š   bel-a-mida
bel-i-s bel-i-ti  bel-i-saš bel-i-zdi
bel-i-0 bel-a-zi  bel-i-taš bel-a-daš

PRES.SUBJ
ACT SG  ACT PL    PASS SG   PASS PL
bel-ā-0 bel-ā-mas bel-ā-š   bel-ā-mida
bel-ē-s bel-ē-ti  bel-ē-saš bel-ē-zdi
bel-ē-0 bel-ā-zi  bel-ē-taš bel-ā-daš

IMPF
ACT SG    ACT PL      PASS SG     PASS PL
i-bel-a-0 i-bel-a-mas i-bel-a-š   i-bel-a-mida
i-bel-i-s i-bel-i-ti  i-bel-i-saš i-bel-i-zdi
i-bel-i-0 i-bel-a-zi  i-bel-i-taš i-bel-a-daš

PFV
ACT SG       ACT PL         PASS SG      PASS PL
i-bi-bal-0-a i-bi-bal-0-mas i-bi-bal-š   i-bi-bal-mida
i-bi-bal-0-s i-bi-bal-0-to  i-bi-bal-saš i-bi-bal-zdi
i-bi-bal-0-i i-bi-bal-0-zi  i-bi-bal-taš i-bi-bal-daš

So we can conclude that the marker for Present Indicative is -a- or -i-, Present Subjunctive is -a- or -e-, Imperfect is i-ROOT-a- or i-ROOT-i-, and Perfect is i-Ci-Root-0-. Passives and subject marker is still fused, though.

[dhok]

I am confused with your conjugation. Can you give the verbal template?

Well that's the problem - in IE there is really no verbal template per se outside of root + derivation + inflection, since all the inflectional categories of verbs are marked in a single inflectional suffix. However I will elucidate:

The table above is set up as per traditional IE studies: the 6 forms given per tense/aspect/mood/voice combination are person/number forms, with person down the side as 1st, 2nd 3rd and number across the top as singular plural. So the present indicative active is like so:

Code: Select all

  SG    PL
1 bela  belamas
2 belis beliti
3 beli  belazi

As for the actual TAM/voice categories themselves, they are fairly standard IE ones: For tense aspect, PRES = present, IMPF = imperfect and PRET = preterite. For moods IND = indicative, SUBJ = subjunctive, OPT = optative and IMP = imperative (the last one only shows 2nd and 3rd person forms). Finally for voices ACT = active and MID = middle. There are two participles (PART), an active and a middle, which inflect like nouns.

Hope that sheds som light.

Well, it explains why the pronomial ending is so inconsistent. However, I saw in Wikipedia that subject agreement in Latin doesn't fuse with aspect, only tense. Also, I usually segment verb conjugation because verb normally has hundreds or thousands of form, and memorizing it one by one will be impractical. For example, what is the 2ND > 3SG reprocial inchoative causative subjunctive form of to take a bath. If the language lacks conjugation this complex, in natlang, there will be hardly any fusion.

Language like this normally simply simplify the suffix, for example

<snip>

So we can conclude that the marker for Present Indicative is -a- or -i-, Present Subjunctive is -a- or -e-, Imperfect is i-ROOT-a- or i-ROOT-i-, and Perfect is i-Ci-Root-0-. Passives and subject marker is still fused, though.

Not everything with verbal inflection is Inuktitut.

Plenty of languages with relatively small paradigms (forms in the dozens rather than hundreds or thousands) have fusional endings. Since Cappadocian is derived from (late-stage) PIE, the inflectional template is basically that of any other older IE language, though the details are closer to Greek or Sanskrit than to Latin.

The basic template (we'll ignore, for the moment, forms like participles and infinitives) is:

[augment]-[stem]-[thematic vowel]-[personal ending]

The augment is a prefix that was originally *h₁e or *e (well, originally it was a particle, but it eventually becomes a prefix) and marks past time. It's only found in Greek, Phrygian and Indo-Iranian (did Classical Armenian have it? I can't remember). In Cappadocian it becomes i-.

The stem is formed, in some way, from the root. It gives you the meaning of the verb, obviously, but it also often marks aspect (perfective or imperfective). Imperfective or present stems in PIE were often formed with a nasal infix (e.g. *seykʷ- > *sinékʷ-/*sinkʷ-), by reduplicating the root with an initial *i (*deh₃- > *dídeh₃-), by adding the suffix *-sḱ- (*mleh₃- > *mleh₃sḱ-), doing nothing to the root at all (*bʰer-), or some combination of the above, or something else.

A different stem in PIE, and in Cappadocian, marks aspect. For example, the perfect was formed by reduplicating a root with an initial *e and then sometimes, but not always, changing the vowel in the root to *o (e.g. *ǵenh₁- > *ǵeǵónh₁-). In Cappadocian the perfect has fallen together with the "aorist", which was formed in various ways, often by suffixing an *-s-. A Cappadocian verb's preterite stem will derive from either the perfect or the aorist, but there is no longer any difference in meaning between them.

This was followed by the thematic vowel, which marked mood and sometimes tense, but which was often dictated by the following personal ending. For example, the indicative of 1sg forms took a thematic vowel *-o- in PIE, but that of 2sg forms took *-e-. A lengthened thematic vowel usually marked the subjunctive; a suffixed *-eyh₁- or *-ih₁- (the former in the singular, the latter in the plural) marked the optative. Many of the daughters reshuffled the thematic-vowel system (like Latin, where we see -imus and not the expected **-umus), but Cappadocian is fairly conservative.

Finally we get the personal ending, which marked person and number, as well as voice. So Cappadocian -mas (from *-mos) is the 1pl active ending, and -mida (from *-medʰh₂) is the middle ending.

Cappadocian is derived from PIE and is a fictional relative of Greek, Latin, Sanskrit, etc. Aside from the choice of sound changes, most of it is already set in stone, or at least constrained: the PIE 1pl middle ending was *-medʰh₂, so Cappadocian's 1pl middle ending is going to be whatever happens to *-medʰh₂ when put through Cappadocian sound changes--that ending is -mida. (Of course some IE languages, like Slavic and Baltic, lost the middle, and others, like Italic and Tocharian, reshaped it--but these changes were not arbitrary, and neither should similar changes in an IE-derived conlang).

[Xwtek]

Cappadocian is derived from PIE and is a fictional relative of Greek, Latin, Sanskrit, etc. Aside from the choice of sound changes, most of it is already set in stone, or at least constrained: the PIE 1pl middle ending was *-medʰh₂, so Cappadocian's 1pl middle ending is going to be whatever happens to *-medʰh₂ when put through Cappadocian sound changes--that ending is -mida. (Of course some IE languages, like Slavic and Baltic, lost the middle, and others, like Italic and Tocharian, reshaped it--but these changes were not arbitrary, and neither should similar changes in an IE-derived conlang).

Indeed, the change should not be arbitrary. My proposed change is analogical. For example, the change is:

1. 1SG and 3SG Subjunctive Active is lengthened by analogy of other verb forms (Compare belēs, belāmas, belēti, belāzi.
2. 1SG and 3PL Imperfect is changed to -a and -azi by analogy of other verb form (Compare belis, ibelis; beli, ibeli; belamas, ibelamas; beliti, ibeliti)
3. 1SG Middle Imperfect gets -š to avoid collapse from 1SG Active Imperfect and in analogy of 1SG Middle Present Indicative.
4. That suffix is extended to 2SG, 3SG dan 3PL as generalization of change #3 and in analogy of 2SG, 3SG, and 3PL Present Indicative.

However, on the second thought, the perfect is so aberrant, it is quite reasonable that it retained fusional. However, it can also be changed by analogy of Imperfect. Maybe the thematic vowel be dropped in analogy of 2SG, 1PL, 2PL active and middle perfect, but 1SG and 3SG's thematic vowel cannot be dropped because the form will merge.

[Frislander]

Well, you could enact some such similar changes in your own IE-conlang, and indeed some of those may occur in a daughter language of this, however there is no reason these changes must or should take place at this stage at this time and place, and as this is an older IE language like Greek or Latin these kinds of irregularities are to be expected.