Tewanian languages - a subfamily of Indo-European

[Zekoslav]

1. General introduction

There seems to be a growing interest in Indo-European on the board, conlangs or natlangs, so I think it's the perfect time to post my own - a small language family of consisting of a proto-language (Proto-Tewanian) and it's three descendants (ATM known as language A, B and C).

There's already some snippets of information about them posted in a scratchpad thread, but I hope it will be OK to make a new one since the information there is outdated and not well presented.

I've been working on this language family for years, making bigger or smaller revisions - but only recently has the proto-language become stable enough so I can start working on it's descendants properly. Since the descendants are still in a state of flux, there will be no complete description of them here, but there will be an almost complete description of the proto-language.

I'm completely open to suggestions, in particular I would like to ask for resources to help me develop the languages' vocabulary.


2. Introduction to Proto-Tewanian

Proto-Tewanian, PTw. tewaniˑyakaˑ dujuˑš "Tewanian language", dujuˑš naˑs "our language", is a rather typical ancient Indo-European language, archaic in phonology but innovative in morphology. It's earliest developments show it to be an intermediate between Indo-Iranian and Hellenic, with some links to Albanian as well.

However, it was never meant to be spoken on Earth - it's origins lie in a comedy I wrote when I was a teenager, where uttering magic words can teleport you to a random place and time in the habitable universe (and that obviously happened to a whole Indo-European tribe). I decided to keep it that way to allow it to have a completely different environment to develop in.

It's most characteristic innovation is agglutinative morphology of verbs: while person-number (and case number) endings are still fusional, the rest of the verbal morphology is largely agglutinative. The main reason for this is that PTw. allows derived verb stems to be derived from other derived verbs stems, which other old IE. languages do not.

Another innovation of PTw. is the loss of distinction between imperfective and perfective aspects (first, they were reinterpreted as remote and recent past, the recent past later disappearing), with perfect aspect being preserved. As a consequence, PTw. lost nearly all traces of PIE. aorist. Proto-Tewanian has also innovated a future tense, as well as new perfective past and future tenses and perfect past and future tenses.

It has preserved the PIE. optative and innovated a conditional.


3. Phonology

3.1 Consonants

Code: Select all

CONSONANT INVENTORY

-                   lab.      den.                pal.      vel.
                              -sib.     +sib.               –lab.     +lab.
stops     -vcd.     p /p/     t /d/                         k /k/     kʷ /kʷ/
          +vcd.     b /b/     d /d/                         g /g/     gʷ /gʷ/
frics.    -vcd.                         s /s/     š /ʃ/   
          +vcd.                        (z /z/)   (ž /ʒ/)
africs.   -vcd.                         c /ʦ/     č /ʧ/
          +vcd.                         j /ʣ/     ǰ /ʤ/
nasals              m /m/     n /n/     
liquids                       l /l/
                              r /r/
semivow.                                       y, i /j/             w, u /w/

Sonorants, but not obstruents, can be geminated. This is represented by doubling the letter: <mm>, <nn>, <ll>, <rr>. Semivowels are represented as <y>, <w> when in onsent, as <i>, <u> when in coda, and as <iy>, <uw> when geminated.

Voiceless and voiced fricatives contrast with each other only in clusters with a stop, but not otherwise. That is /ps/, /pʃ/ and /kʃ/ contrast with /bz/, /bʒ/ and /gʒ/, but single /s/ and /ʃ/ don't contrast with single /z/ and /ʒ/.

Several synchronic and diachronic phenomena suggest that these clusters should be analysed together with the affricates /ʦ/, /ʧ/, /ʣ/ and /ʤ/, either all as clusters or all as affricates - they will be dealt with later.


3.1.1 Consonant allophones

When they don't participate in affricate-like clusters, the distribution between /s/, /ʃ/ and /z/, /ʒ/ is as follows: voiceless before a voiceless consonant, voiced before a voiced consonant (obstruent or sonorant). Word initially and after /n/ and /r/, voicing before a sonorant is inhibited

Code: Select all

s, ʃ > z, ʒ / _C[+vcd], !{#, n, r}_

PTw. esmi "COP-1.sg.Pres." > [ˈɛzmɪ]

PTw. esti "COP-3.sg.Pres." > [ˈɛstɪ]


Intervocalically, voiced stops, but not voiced affricates, become voiced fricatives. This is one of the phenomena that suggest treating affricates together with clusters of stop and fricatives.

Code: Select all

b, d, g, gʷ > β, ð, ɣ, ɣʷ / V_V

PTw. tagaibas "house-Dat.pl." > [ˈtaɣajβas]

PTw. sedaibas "chair-Dat.pl." > [ˈsɛðajβas]


Before a front vowels and /j/, velars (both plain and labialised), become palatals. Closer the vowel, stronger the palatalisation, but for simplicity, all these allophones will be represented as one.

Code: Select all

k, g, ɣ, kʷ, gʷ, ɣʷ > c, ɟ, ʝ, c[super]ɥ[/super], ɟ[super]ɥ[/super], ʝ[super]ɥ[/super] / _V[+front], j

PTw. kʷiš "who" > [ˈcɥɪʃ]


3.2 Vowels

Code: Select all

VOWEL INVENTORY

-         front     central   back
high      iˑ/iː/              uˑ/uː/
          i /ɪ/               u /ʊ/
mid       ê /eː/             (ô /oː/)
          e /ɛ/  
low       eˑ/æː/    a /a/     aˑ/ɑː/ 

overlong vowels: iː, êː, eː, aː, ôː, uː

In the last syllable of the word, Proto-Tewanian distinguishes three degrees of vowel length: short, long and overlong, represented as <V>, <Vˑ> and <Vː> respectively (I usually represent them by macrons and double macrons, Don Ringe's style, but the results of attempting that atrocity on this board are disastrous.).

Overlong vowels are usually the result of vowel contraction, so contracted vowels will be represented as overlong even where they don't contrast with long vowels, to simplify morphology.


3.2.1 Vowel allophones

Short front vowels [ɪ] and [ɛ] are raised when folowed by a /j/, with the resulting [ij] and [ej] becoming [iː] and [eː] when followed by a consonant.

Code: Select all

ɪ, ɛ > i, e / _j

ij, ej > iː, eː / _C

PTw. pateres "father-Nom.pl." > [ˈpatɛrɛs]

PTw. pateyes "master-Nom.pl." > [ˈpatejɛs]


Before a tautosyllabic nasal, lax high vowels become tense.

Code: Select all

ɪ, ʊ > i, u / _N$

PTw. sunuš "son-Nom.sg." > [ˈsʊnʊʃ]

PTw. sunun "son-Acc.sg." > [ˈsʊnun]


3.3 Timing and stress

Proto-Tewanian is mora-timed, and the rhythm of short, long and overlong syllables dominates it's prosody - primary stress is on the first syllable, with very light secondary stress on every heavy or superheavy syllable.

When counting morae, the following rules apply: a) short vowels count as one mora, b) each additional degree of length counts as one mora, c) coda sonorants, but not obstruents, count as one mora, d) syllables of more than three morae are prohibited, which is resolved by removing one degree of length from the syllable's nucleus vowel (Osthoff's rule). In practice, this means that in syllables closed by a sonorant, long vowels become short, and overlong vowels become long.

Code: Select all

V - 1 mora

Vː, VR - 2 morae

Vːː, VːR - 3 morae

3.4 Sandhi

Only external sandhi will be described here, since internal sandhi has been largely morphologises.

1. When two vowels meet at the edge of a word, the following rules apply:

a) After a high vowel or a semivowel, insert the corresponding semivowel

Code: Select all

0 > j / {j, ɪ, iː, iːː}_V

0 > w / {w, ʊ, uː, uːː}_V

b) Insert a /j/ before an initial front vowel, and a /w/ before an initial back vowel

Code: Select all

0 > j / V_V[+front]

0 > w / V_V[+back]

When the first word ends in a consonant, following rules apply:

a) Devoice all word-final consonants

Code: Select all

C[+voiced] > C[-voiced] / _#

b) Voice all word-final consonants and consonant clusters when followed by a voiced sound

Code: Select all

C[-voiced] > C [+voiced] / _#{V, C[+voiced]}

Note that this rule applies to fricatives as well, making the rules of their voicing word-finally, different from the rules of their voicing word-internally - word, finally, they also voice intervocalically and when preceded by /n/ or /r/.

c) Assimilate voiced fricatives to the preceding sonorant

Code: Select all

mz, nz, lz, rʒ > mː, nː, lː, rː / _#C

d) shorten coda geminates and give the preceeding vowel one additional mora

Code: Select all

V(ː)Cː > V(ː)ːC / _#C

When this happens, /ɪ/ and /ʊ/ become /iː/ and /uː/ before a nasal, and /eː/ and /oː/ before a liquid - this is the only situation where the vowel /oː/ appears outside of interjections.

Code: Select all

ɪNː, ʊNː > iːN, uːN / _#C

ɪRː, ʊRː> eːR, oːR / _#C

While these rules may seem complicated at first, they are actually quite simple, as this example will demonstrate: word final [ins] is preserved before a voiceless consonant (and pause), becomes [inː] before a vowel and becomes [iːn] before a voiced consonant.


4. But what does the language really look like?!

To demonstrate all that has been said in action (especially sandhi), here's Schleicher's fable in three different versions. The first two versions are in Proto-Tewanian's usual orthography, one without sandhi, and one with sandhi - words that change between these versions have been bolded. The third version is in IPA, showing not just sandhi but all allophony.


Without sandhi:

Buwaˑt awiš, yasmi smi kismaˑna ne buwaˑnt. Awiš kʷe ecwans enakʷyaˑt : yan gʷeriˑyasun wajan wujaˑt, yan kʷe mejiˑyasun baran biraˑt, yan kʷe ǰemanun tekanti biraˑt.

Awiš kʷe ecwaibas mluwaˑt : "Ceˑr mai biˑdaˑsetai in enakʷyunnaˑn, yun ǰemuː ecwans ajeti."

Ecwans kʷe mluwaˑnt: "Taušiye, awê! Ceˑr naˑs biˑdaˑsetai in enakʷyunnaˑma, yun sai ǰemuː, patiš, awyêyaˑ wlanêyaˑ tepantaˑ westakaˑ webeti, awibas kʷe kismaˑna ne santi."

Awiš in taušiːnnaˑt, piltawiˑn kʷe bugaˑt.


With sandhi:

Buwaˑd awiš, yasmi smi kismaˑna ne buwaˑnt. Awiš kʷe yecwann enakʷyaˑt : yan gʷeriˑyasun wajan wujaˑt, yan kʷe mejiˑyasun baran biraˑt, yan kʷe ǰemanun tekanti biraˑt.

Awiš kʷe yecwaibaz mluwaˑt : "Ceˑr mai biˑdaˑsetaiy in enakʷyunnaˑn, yun ǰemuː wecwann ajeti."

Ecwans kʷe mluwaˑnt: "Taušiye, awê! Ceˑr naˑz biˑdaˑsetaiy in enakʷyunnaˑma, yun sai ǰemuː, patiš, awyêyaˑ wlanêyaˑ tepantaˑ westakaˑ webeti, awibas kʷe kismaˑna ne santi."

Awiž in taušiːnnaˑt, piltawiˑn kʷe bugaˑt.


In IPA:

ˈbʊwɑːd ˈawɪʃ |ˈjazmɪ zmɪ ˈcɪzmɑːna nɛ ˈbʊwɑːnt ‖ ˈawɪʃ cɥe ˈjɛʦwanː ˈɛnacɥjɑːt | ˈjan ˈɟɥɛriːjasun ˈwaʣan ˈwʊʣɑːt |ˈjan kɥɛ ˈmɛʣiːjasun ˈbaran ˈbɪrɑːt | ˈjan cɥɛ ˈʤɛmanun ˈtɛkantɪ ˈbɪrɑːt ‖

ˈawɪʃ cɥe ˈjɛʦwajβaz ˈmlʊwɑːt |ˈʦæːr maj ˈbiːðɑːsɛtaj ˈin ˈɛnacɥjunːɑːn |ˈjun ˈʤɛmuːː ˈwɛʦwanː ˈaʣɛtɪ ‖

ˈɛʦwans cɥɛ ˈmlʊwɑːnt | ˈtawʃijɛ | ˈaweː | ˈʦæːr nɑːz ˈbiːðɑːsɛtaj ˈin ˈɛnacɥjunːɑːma |ˈjun saj ˈʤɛmuːː | ˈpatɪʃ | ˈawjeːjɑː ˈwlaneːjɑː ˈtɛpantɑː ˈwɛstakɑː ˈwɛβɛtɪ |ˈawɪβas cɥɛ ˈcɪzmɑːna nɛ ˈsantɪ ‖

ˈawɪʒ ˈin ˈtawʃiːnːɑːt | ˈpɪltawiːn cɥɛ ˈbʊɣɑːt ‖


So, what can a reader expect if they continue following this thread? More complicated rules without sufficient examples? Of course! But they will also find out which name in Proto-Tewanian becomes Āegon in one descendant and Óin in another, as well as just how different from it's parent a descendant language can become!

That is, if their head doesn't explode first...

Edit: I've revised everything with examples and (unsuccessful?) attempts at humour.

[Zekoslav]

• A revision of Proto-Tewanian accent system

While writing the Beginner's Guide to Slavic Accentuation, I researched the PIE. accent system and decided to make it's development in Proto-Tewanian more nuanced than simply "it's too complex so it's replaced by a fixed stress system".

So, everything that was said about timing, stress and vowel length in PTw. is now outdated. Well, not everything - it's still mora-timed.


1. The new Proto-Tewanian accent system

a. The basics

Due to it's close link with Greek and Indo-Iranian, Proto-Tewanian inherited an accent system very closely related to both of these languages (more to Greek than to Indo-Iranian), but has nevertheless gone through some unique developments.

With Proto-Tewanian being mora-timed, the phonological unit that bears the accent is the mora. An accented mora has a high tone (like Vedic udātta), the following mora has a falling tone (like Vedic svarita), and the rest of the morae have a low tone (like Vedic anudātta).

Short vowels, which contain one mora, can only have the high tone on that mora, while long vowels, which contain two morae, can have the high tone either on the first or on the second mora, resulting in the opposition of rising and falling pitch accents (like Greek acute and circumflex accents ).

Because of this similarity, It would be aesthetically pleasing to romanize Proto-Tewanian vowels and accent like that of Greek - if only the combinations of diacritics required to do it looked good in all fonts...


b. Rules restricting the placement of accent

A short accented vowel cannot precede a long vowel - historically, in such combinations the accent moved to the first mora of the long vowel. In Vedic terms, a long svarita takes the accent from a short udātta, because it's tonal contour is similar or identical to the tonal contour of a falling pitch accent. This rule is called the Long svarita rule.

Code: Select all

ˈá ... âː > a ... ˈâː / _

Diachronically speaking, long svarita rule has already become morphologized by the time of Proto-Tewanian, and the secondary oxytones created by it have copied the accent pattern of primary oxytones inherited from PIE. Speaking of diachrony...


c. The development of Proto-Tewanian accent from PIE.

The origin of the opposition of rising and falling pitch accents exactly parallels their development in Greek - in case of vowel contraction following the loss of laryngeals, a rising tone resulted if the first vowel was accented, and a falling tone resulted if the second one was accented.

So, in the a-stem nouns, for example, the nominative singular develops a rising pitch, PIE. *-éh2 > PTw. /-áː/, but the nominative plural develops a falling pitch, PIE. *-éh2es > PTw. /-âːs/.

Long vowels that were not the result of contraction could also receive a falling pitch in special cases. Two of the PIE. accentual rules governing underlyingly unaccented word forms (explained here), the Oxytone rule (the final syllable of a morphologically complex stem receives the accent) and the Basic accentual principle (the first syllable of a morphologically simple stem receives the accent) were redefined so as to apply to morae - so, if a word whose first syllable contained a long vowel was accented by the application of the Basic accentual principle, it received a falling pitch because it's first mora received the accent.

This was the general rule in vocatives and imperatives, but it also occurred in many consonant stems and even o- and a-stems which were no longer analyzable as derived, paralleling Greek (although the individual lexemes which went through the process were different in Proto-Tewanian and Greek). The exact development will be explained together with the relevant morphology.

Also, when word-final /n/ and /r/ were lost following a long high vowel, the vowel's pitch changed from rising to falling, so, PIE. *ph2tḗr "father" > PTw. /paˈtîː/.

In Sanskrit and probably also PIE., finite verbs were unaccented when in an independent sentence. This was generalized in Proto-Tewanian to finite verbs in all types of sentences - the verb received an accent on the last syllable of the stem according to the Oxytone rule, which could move onto the first syllable of the suffix according to the Long svarita rule. Again, the details will be explained together with morphology.

That's it for now. In the next post I should explain the rest of the sound changes from PIE. to PTw., together with more examples of the accentual changes.

[Zekoslav]

• Declensions!

It's been a long time since I've last posted something, but recent IE. conlang activity has reminded me that I had a thread about this language, so let's get back in action!

• neuter en-stem declension:

bãmun "word"


First, a note on transcription. Consonant and vowel transcription has remained the same, while accent transcription is based on Ancient Greek: mācrōns for long vowels, ácútés for rising pitch and tĩldẽs for falling pitch. Since falling pitch can only appear on long vowels, there's no need to mark vowels with both macrons and tildes, although I've toyed with such an idea...

You'll see that there are multiple forms for some cases. I like to put some morphological synonymy as a seed for future divergences (and to allow me not to commit to one sole option ). In general, when two forms are listed in the same row, they're not completely synonymous, but when they're listed one below the other, they are. Forms in parentheses are rare past or future forms and not really a part of the language at the stage presented.


Discussion:

Most forms derive directly from PIE, but there's been some morphological change, most notably regularization of ablaut. These changes are the same for all regular consonant stem nouns:

• root: same ablaut grade, o- e- or zero, throughout the paradigm

• suffixes: one ablaut grade (synchronically unpredictable) in the nom./voc. cases for animates and nom./acc./voc. for inanimates, and another ablaut grade in other cases, usually the e-grade.

In addition, the ablative case has been replaced by the instrumental, and the dative case with the locative (with locative endings winning in the singular and dative endings winning in the dual and plural). However, there's a separate locative case in singular only derived from the PIE. postposition *en "in" which can only serve the locative function. This prevents the language from being a Sanskrit clone too kitchen sinky.

As for accent, most cases are accented on the first syllable of the suffix. There are two exceptions:

• if the first syllable of the suffix is short, and the second is long, then the second syllable is accented instead (e.g. in the ins. sg. bāmenḗ and dat./loc. du. bāmembā́n)

• certain cases, most commonly the vocative, have recessive accent: the first mora of the first moraic trochee in the word is accented, in practice, it's usually the first mora of the word (resulting in falling pitch on long vowels, e.g. nom./acc./voc. sg. bãmun, voc. du. bãmānī, voc. pl. bãmāna)

However, if an initial short vowel is followed by a long vowel, then the initial vowel is left extrametric and the first mora of the second vowel is accented (e.g. késmun "hair" has nom./acc. pl. kesmā́na but voc. pl. kesmãna). This is a general phonotactic constraint preventing short accented vowels from preceding long unaccented vowels.


Singular:

The nominative singular form, if going by regular sound changes, should have actually been *bãmin. It's been changed to the attested form through the spread of pronoun/adjective ending, e.g. *tún bãmin "that word" > tún bãmun . This change is crucial for the development of the language as a whole, and I haven't decided to post neuter n-stems first for no reason (the other being that they're the ones I've most recently revised)!


Dual and plural:

The nom./acc./voc. plural ending -āna derives from PIE. *-ōnh2, which is semi-regular (the final laryngeal, lost after resonants, has been added by analogy with other stems).

There's a tendency to reanalyse this ending as -ān-a, with a special lengthened grade suffix followed by the regular consonant stem neuter nom./acc./voc. pl. suffix -a, and to spread it's ablaut to other case forms. This has already been completed in the nom./acc./voc. du. (*bāmnī́ > bāmā́nī) and the gen. du./pl. (*bāmnáuš > bāmā́nauš; *bāmnũ > bāmānũ, bāmā́nū), and in some daughter languages it will spread to the entire dual and plural paradigm, which is shown in parentheses.

The two accent variants in gen. pl. have arisen because one has introduced the nom./acc./voc. ablaut while keeping the original accent (i.e. *bām-n-ũ > bām-ān-ũ), whereas the other has introduced both ablaut and accent from said case forms (i.e. *bām-n-ũ > bām-ā́n-ū). As for why this double treatment exists only in gen. pl., it has to do with the accent pattern of other en-stems, which will be shown later.

As always, I invite you to tell me your opinions

[elemtilas]

Awesome! This is very nicely presented and I look forward to more!

[Zekoslav]

Awesome! This is very nicely presented and I look forward to more!

Thanks! And it looks like I've helped to spread this PIElang contagion... I'm interested to see how distinct we all manage to make our conlangs.

I hope my presentation is understandable, since it presupposes some familiarity with PIE. when I discuss the origin of certain forms, and I didn't want to post every historic detail yet (I might, though: would you be interested in seeing various phases of the conlang's development, as in, a table for each phase to be able to compare?).

[elemtilas]

Thanks! And it looks like I've helped to spread this PIElang contagion... I'm interested to see how distinct we all manage to make our conlangs.

A very fine contagion to suffer from! Yes, your presentation is very tidy and understandable. Especially when it comes to texts, though, it would be nice to see, in addition to the plain / unsandhied version the same but fully broken down phrase by phrase with grammatical bits and parts of speech.

As for comparison, I have the Talarian translation here:

Tewanian
Buwaˑt awiš, yasmi smi kismaˑna ne buwaˑnt. Awiš kʷe ecwans enakʷyaˑt : yan gʷeriˑyasun wajan wujaˑt, yan kʷe mejiˑyasun baran biraˑt, yan kʷe ǰemanun tekanti biraˑt. Awiš kʷe ecwaibas mluwaˑt : "Ceˑr mai biˑdaˑsetai in enakʷyunnaˑn, yun ǰemuː ecwans ajeti." Ecwans kʷe mluwaˑnt: "Taušiye, awê! Ceˑr naˑs biˑdaˑsetai in enakʷyunnaˑma, yun sai ǰemuː, patiš, awyêyaˑ wlanêyaˑ tepantaˑ westakaˑ webeti, awibas kʷe kismaˑna ne santi." Awiš in taušiːnnaˑt, piltawiˑn kʷe bugaˑt.

Talarian
Wellan-cohes, weweyssi walmanuça-ne xowios-ca hahâms, içatla maxuça waconar-can rómati, iriloç-he wiram çerewana ffárati. Xowios-coç hahames feffâti: Cartay-ca-he mamaç haxanatar, wirahaharomomtoç. Hahas-toy xowiay-ca feffâti: Harcato! Xowie, cartay-ca-he wosaç haxanatar, wirawalmanffartaromtos, xowiay-he walnar-ça-ne! Tlewehetasa, xowios-cas sexoman-sa-han xaxâtenti.

I.E. of 1868
Avis, jasmin varnā na ā ast, dadarka akvams, tam, vāgham garum vaghantam, tam, bhāram magham, tam, manum āku bharantam. Avis akvabhjams ā vavakat: kard aghnutai mai vidanti manum akvams agantam. Akvāsas ā vavakant: krudhi avai, kard aghnutai vividvant-svas: manus patis varnām avisāms karnauti svabhjam gharmam vastram avibhjams ka varnā na asti. Tat kukruvants avis agram ā bhugat.

English
Upon the hill, a sheep with no wool saw some horses, one of which drew a large waggon, the other of which swiftly bore a man. The sheep said to the horses: “It pains my heart, to see the man leading horses.” The horses replied: “It pains our heart, to see the man wearing wool; and the sheep has no wool!” Hearing this, the sheep fled into the plain.

Clearly, a number of roots are recognisably related: xowio- / awi- ; haxa- / ecwa- ; walman- / wlane- ; carta- / cer- ; -he / -cwe ; wira- / gweri- ; ffara- / bara-. Or at least, "recognisably" because I expect them to be and am looking for correspondence!

[Zekoslav]

I have to admit, I improvised that Schleicher's fable: I've dabbled with sound changes and morphology a lot, but have barely touched the vocabulary (enough to allow me to work with morphology, but no more). Also, it's been ages since I've written that introductory post! Since not even phonology and morphology has escaped the urge to revise, it's a good opportunity to rewrite that fable (and thank you for bringing that up)!

Your language also looks very interesting, it has a very different feel from my own while conserving a lot of PIE. stuff. It looks like some dependent clauses are expressed by some sort of compound word?

[elemtilas]

I have to admit, I improvised that Schleicher's fable: I've dabbled with sound changes and morphology a lot, but have barely touched the vocabulary (enough to allow me to work with morphology, but no more). Also, it's been ages since I've written that introductory post! Since not even phonology and morphology has escaped the urge to revise, it's a good opportunity to rewrite that fable (and thank you for bringing that up)!

That's the language inventor's greatest foil: the desire to rewrite!

Your language also looks very interesting, it has a very different feel from my own while conserving a lot of PIE. stuff. It looks like some dependent clauses are expressed by some sort of compound word?

As I recall, yes! It's been ages since I worked with that language, and probably couldn't write anything in it now without a lot of painful work.

[Zekoslav]

• Schleicher's fable 2.0

As requested, here's the new version of Schleicher's fable with an interlinear gloss. It's a nice way to get a feel for the language while I'm working on posting more bits of grammar.

• The fable

• The gloss

Code: Select all

Áw-iš          écw-ai=kʷe
sheep-NOM.SG   horse-NOM.PL=and

"The sheep and the horses."

Code: Select all

Y.a-sy.ãi        aw-ễi          kesm.́ān-a     né    buw-́ā.t, 
REL-FEM.DAT.SG   sheep-DAT.SG   hair-NOM.PL   NEG   be-3SG.PST.ACT

s-́ā              gur-én        écw-ans        dirc-in-́ā.t.
DEM-FEM.NOM.SG   hill-LOC.SG   horse-ACC.PL   see-INCH-3SG.PST.ACT

"A sheep who had no wool saw horses on a hill."

Code: Select all

Ain-án       dirc-in-́ā.t            gúr-un              wáj-an        wéj-en-êi,
one-ACC.SG   see-INCH-3SG.PST.ACT   heavy-MASC.ACC.SG   cart-ACC.SG   pull-INF-LOC.SG

ain-án=kʷe       míc.en-un         bár-an, 
one-ACC.SG=and   big-MASC.ACC.SG   load-ACC.SG   

ain-án=kʷe       tším.n-un   ãc-u        bér-en-êi.
one-ACC.SG=and   man-ACC.SG  quick-ADV   carry-INF-LOC.SG

"She saw one pulling a heavy cart, one (carrying) a big load, and one quickly carrying a man."

Code: Select all

Áw-iš          écw-aibas      mluw-́ā.t:
sheep-NOM.SG   horse-DAT.PL   say-3SG.PST.ACT

"Cẽr-Ø=m.ai            b́īd-ās-e.tai              t-ún               dêr-ta-went-yãi,
heart-NOM.SG=1SG.DAT   break-DESID-3SG.PRS.MED   DEM-NEUT.ACC.SG.   see-PART-having-FEM.DAT.SG

yún    tšim.ũ-Ø     écw-ans        áj-e.ti."
COMP   man-NOM.SG   horse-NOM.PL   drive-3SG.PRS.ACT

"The sheep said to the horses: "My heart wants to break when I see that man is driving horses.""

Code: Select all

Écw-ai         aw-ễi          mluw-́ā.nt:         "Táuš.iy-e, aw-ễ!
horse-NOM.PL   sheep-DAT.SG   speak-3PL.PST.ACT   hear-IMP   sheep-VOC.SG


Cérd-a=nās             b́īd-ās-a.ntai             t-ún              dêr-ta-wénd-bas,
heart-NOM.PL=1PL.DAT   break-DESID-3PL.PRS.MED   DEM-NEUT.ACC.SG   see-PART-having-MASC.DAT.SG


yún=s.ai           tšim.ũ-Ø,    pát-iš,
COMP=REFL.DAT.SG   man-NOM.SG   master-NOM.SG   


aw-yũ          kesm.ém-biš    gʷérm-un           wésm.un-Ø        wéb-e.tai,
sheep-GEN.PL   hair-INST.PL   warm-ACC.SG.NEUT   clothes-ACC.SG   weave-3SG.PRS.MED


áw-ibas=kʷe        kesm.́ān-a     né    és-ti."
sheep-DAT.PL=and   hair-NOM.PL   NEG   be-3PL.PRS.ACT

"The horses said to the sheep: "Listen, o sheep! Our hearts want to break when we see that man, the master,
makes himself warm clothes out of sheep's wool, and the sheep no longer have wool.""

Code: Select all

T.á-sm.i          tauš-t-ễi,             áw-iš          ajr-ễi         úd      bug-́ā.t.
DEM-NEUT.LOC.SG   hear-PART-NEUT.LOC.SG  sheep-NOM.SG   field-LOC.SG   out of  flee-3SG.PST.ACT

"Having heard that, the sheep fled the field."

• The commentary

Vocabulary

The vocabulary I used is intentionally conservative: nearly all words are reconstructable for PIE. and most can be found in some of the fable's versions listed on Wikipedia. for "quickly", I even reverted to a descendant of PIE. *h1oh1ḱú, because I started to like how it looked. The only notable innovation is that, of all words, *h2wĺ̥h2neh2 "wool" was replaced by *késmn̥ "that which is combed" .


Syntax

The language's syntax is also standard ancient IE. However, ancient IE. syntax was very interesting: it had internally-headed relative clauses! In such clauses the "antecedent" is nested within the dependent clause, and it's place in the main clause is left empty or, in the case of ancient IE., taken by a demonstrative pronoun.

So, the first sentence is literally "To which sheep there was no wool, that (one) saw horses on a hill.".

Complement clauses have the same syntax as relative clauses, with a relative pronoun in the dependent and a demonstrative pronoun in the main clause (I've glossed the relative pronoun as a complementizer in such cases, but it's possible to analyze this differently), but most other complex sentences are handled with nominalization and participles. I've used a "dative absolute" in the last sentence as well.

If you learned Ancient Greek, you may know that neuter plural subjects take singular verbal agreement. This is also the case here.


Next time: probably more declensions (sound changes are currently in flux and will be adjusted as to produce good looking morphology).

[Zekoslav]

• More declensions!

This time, let's see what an irregular neuter en-stem looks like

nãmun, "name"


Unlike last one, this word has kept root and suffix ablaut, even though sound changes have transformed it into a simple alternation of a long and a short vowel.

• Root ablaut:

This word is notoriously difficult to reconstruct (which and how many laryngeals there were), but N/A/V sg. can be derived from PIE. *HnóHmn̥, while the plural can be derived from PIE. *Hn̥Hmṓnh2. Even though ablaut has been preserved, it's distribution has changed: singular now has a long vowel and dual and plural have a short vowel.

• Suffix ablaut:

While regular neuter en-stems have the e-grade throughout (with the exception of N/A/V), this one has kept a more archaic pattern: zero grade if the case ending begins with a vowel, and e-grade if it begins with a consonant. This pattern was usually replaced by the e-grade throughout because the zero grade usually messed up root-final consonants (i.e. it was a phonotactics-based adjustment), but in this word it created a rather mundane cluster /mn/ so it was kept.

This is responsible for a somewhat different accent pattern: usually it's the last vowel of the stem which is accented*, but this can't happen when it's, well... gone due to zero grade. When this happens, the either the case ending or the first mora of the word is accented, depending on the exact case form. As it happens, in neuter nouns it's always the case ending.

*There's three accent patterns in Proto-Tewanian: fixed accent on the first vowel of the stem, fixed accent on the last vowel of the stem, and mobile accent. While most languages eventually got rid of mobile accent in polysyllabic words, in Proto-Tewanian it was kept because the first two accent patterns became mobile through sound changes (such as accent moving from a short accented vowel to a following long unaccented vowel) and instead the three patterns tend to influence each other in complex ways.

• accent patterns:

In the D/L/I pl., which I aptly call the "b-cases", there's multiple accent patterns possible: one where the final vowel of the stem is accented, and one where the case ending is accented. In these cases the fixed and mobile accent patterns got confused early, so basically every noun has both options.

You may be asking, but why didn't the word bãmun "word" also have these doublets? Well, it's because it has a long root vowel in the relevant case forms. I've decided to include another accent constraint, inspired by Greek. I'm not yet sure about the details (how phonetically regular will it be), but there will be several changes to produce well-formed moraic trochees.

e.g. if the word was accented as bāmembás, it would have the moraic structure of (μ μ) μ ˈμ with an extrametric mora, but if it was accented as bāmémbas it would have the moraic structure of (μ μ) ˈμ μ with two well-formed moraic trochees. So, in this case the original accent was kept while in namémbas/namembás it was optionally influenced by the mobile accent pattern.

Another interesting consequence of these constraints is that, in dual and plural, N/A is only distinguished from V by pitch (see table). As has been said, vocative always has recessive accent, so as to produce a moraic trochee at the beginning of the word. If the initial vowel is long, this results in falling pitch on that vowel, e.g. (ˈμ μ) μ μ. If two initial vowels are short, this results in high pitch on the first vowel, e.g. ˈμ μ μ μ. But if the initial vowel is short and second vowel is long, then the second vowel forms the first moraic trochee, and this results in falling pitch on the second vowel, e.g. μ (ˈμ μ) μ μ. This is the case in all cases (pun semi-intended) which have recessive accent.

• bits of morphology:

Any further regularization of this noun's paradigm (in the table, shown in parentheses) takes place withing each number: singular has already been remodeled on the N/A/V sg., dual will be remodelled on the N/A/V du., and plural will be remodelled on the N/A/V pl.

Such a remodelling, with case endings added not to the stem, but to the entire N/A/V du. form reinterpreted as a stem, has actually already happened for "eye" and "ear" in PIE. times, as is shown by some remarkable correspondences between Indo-Iranian and Balto-Slavic. I'm actually undecided if I want to keep it a feature of irregular nouns only or to make it a feature of the dual number as a whole (there's patterns in the pronoun declension which allow this to happen). It's probably best to let each option happen in a descendant language .

[Zekoslav]

Sound changes from PIE

1. Consonants

1. PIE voiced aspirates become unaspirated and merge with voiced unaspirates.

Code: Select all

bʰ dʰ ǵʰ gʰ gʷʰ > b d ǵ g gʷ / _

• This happens after Lachmann's law, by which voiced unaspirates, but not voiced aspirates, lengthen the preceding vowel when they undergo devoicing.

2. PIE labiovelars are unrounded before rounded vowels but preserved before unrounded vowels. They are also unrounded before a consonant.

Code: Select all

kʷ gʷ(ʰ) > k g / _V[+round]
kʷ gʷ(ʰ) > k g / _C

3. PIE palatovelars usually become dental affricates. Before dentals stops and *n, they become postalveolar fricatives instead (voiced *ǵ(ʰ) then devoices). After dental stops, they also become postalveolar fricatives (voiced *ǵ(ʰ) and the preceding voice dental stop then devoice).

Code: Select all

ḱ, ǵ(ʰ) > c /t͡s/, j /d͡z/ / _
ḱt, ǵ(ʰ)d(ʰ) > št /ʃt/, žd /ʒd/ / _
ḱn, ǵ(ʰ)n > šn / _
tḱ, d(ʰ)ǵ(ʰ) > tš / _

• This happens after original syllabic *n̥H has become nā or na.

• The affricate /t͡s/ is distinguished from the clusters /ts/ and /tʃ/. The affricate /d͡z/, however, doesn't contrast with clusters because there is no /z/ and /ʒ/ and no /dz/ and /dʒ/ to contrast with.

4. PIE *s becomes š /ʃ/ according to the RUKI rule

Code: Select all

s > š / {r, k(ʷ), u(H), w, i(H), y}_

5. PIE *m becomes n word-finally.

Code: Select all

m > n / _#

6. Original clusters *ls and *rš assimilate to ll and rr. This happens unconditionally, and the resulting geminate remains before a vowel but is shortened, with the compensatory lengthening of the preceding vowel, before a consonant. Lengthened i > ê, u > ô.

Code: Select all

els als erš arš > ell all err arr/ _V 
els als erš arš > ēl āl ēr ār / _C
ils uls irš urš > ill ull irr urr / _V 
ils uls irš urš > êl ôl êr ôr / _C

• This is a very late change, and it reintroduces long vowels before RC clusters, which had been shortened by Osthoff's law.

2. Vowels

1. PIE *ē and *ō become ī and ū before a word-final resonant. This resonant is then lost and the accent changes from acute to circumflex.

Code: Select all

ḗR ṓR > ĩ ũ / _#

2. Changes related to the loss of laryngeals. They are numerous...



2.a.

Before a consonant, there is loss of laryngeals followed by compensatory lengthening.

True vowels *eH, *aH and *oH become long ē, ā and ō in all positions.

Semivowels *iH and *uH become long ī and ū if they are stressed regardless of the following consonant, also long ī and ū if they are unstressed and followed by an obstruent, but short i and u if they are unstressed and followed by a resonant.

Syllabic consonants m̥H, n̥H, l̥H and r̥H behave similar to semivowels. They become ama, ana, ala and ara if they are stressed regardless of the following consonant, mā, nā, lā and rā if they are unstressed and followed by an obstruent, and ma, na, la, ra if they are unstressed and followed by a sonorant.

Code: Select all

eH aH oH > ē, ā, ō / _C
íH úH > ī́ ū́ / _C
iH uH > ī ū / _O
iH uH > i u / _R
ḿ̥H ń̥H ĺ̥H ŕ̥H > áma ána ála ára / _C
m̥H n̥H l̥H r̥H > mā nā lā rā / _O
m̥H n̥H l̥H r̥H > ma na la ra / _R

2.b.

Before a vowel, there is loss of laryngeals followed by vowel contraction.

Semivowels *i and *u insert a *y and *w between themselves and the following vowels, while other vowels contract with the following vowel. If the first of the two vowels was accented, the result is the circumflex accent, if the second was accented the result is the acute accent.

Syllabic resonants before a laryngeal and a vowel behave like syllabic consonants before a normal consonant (see below), not like syllabic resonants before a laryngeal and a consonant (see above).

Vowels also contract with a following syllabic resonant, resulting in the circumflex or the acute accent according to usual rules. Word-internally this long vowel is then shortened by Osthoff's law, but it remains long word-finally, where analogical levelling may change the accent from circumflex to acute.

Code: Select all

i u > iy uw / _V
ée éa éo áe áa áo óe óa óo > ẽ ã õ ã ã õ õ õ õ / _
eé eá eó aé aá aó oé oá oó > ḗ ā́ ṓ ā́ ā́ ṓ ṓ ṓ ṓ / _
íR̥ úR̥ éR̥ áR̥ óR̥ > íyR̥ úwR̥ ẽR ãR õR / _
iŔ̥ uŔ̥ eŔ̥ aŔ̥ oŔ̥ > iyŔ̥ uwŔ̥ ḗR ā́R ṓR / _

2.c.

PIE "syllabic laryngeals" insert an epenthetic schwa before or after themselves, depending on the syllable structure and the surrounding consonants (exact conditions are a work in progress).

Schwa then becomes *i before *y regardless of length, which is subject to laryngeal metathesis, *Hi > *iH / C_C. Schwa is then subject to compensatory lengthening, *Hə, *əH > ə, ə̄. Long schwa then becomes ā, and short schwa becomes e, possibly also i under some conditions (again, exact conditions are a work in progress).

Code: Select all

ə > i / _y
əH > ə̄ / _
ə̄ > ā / _
ə > e / _
? ə > i / ?

3. PIE syllabic resonants which didn't undergo special laryngeal-related changes insert an epenthetic *ɨ before them. An epenthetic *ɨ which behaves the same way is also inserted in intolerable zero grades.

This *ɨ then becomes u after labials and labiovelars and i after coronals and plain velars. After this change, labiovelars are delabialized by the u they conditined. Epenthetic *ɨ also become u before a word-final *m.

Code: Select all

0 > ɨ / between too many consonants
m̥ n̥ l̥ r̥ > ɨm ɨn ɨl ɨr / _
ɨ > u / _m#
ɨ > u / {P, Kʷ}_
ɨ > i / _

4. PIE *o becomes u before a word-final *m.

Code: Select all

o ō > u ū / _m#

5. PIE *o merges with PIE *a.

Code: Select all

o ō > a ā / _

• Since this happens after labiovelars getting delabialized by following rounded vowels, PIE *Kʷo and *Kʷa should remain distinct as Ka and Kʷa. However, there wouldn't be many cases of PIE *Kʷa, especially of root-initial PIE *Kʷa, unless supplied by loanwords (i.e. words invented a priori and not descended from reconstructed PIE words.).

6. Disyllabic sequences *eyi and *ayi undergo umlaut to become *iyi and *eyi. These then contract to become ī and êi. The sequence *uwu also contracts to ū at the same time, and the sequence *eyey (limited to i-stem dative singular endings and probably disappeared from them by haplology like in most IE languages) contracts to êy.

Code: Select all

eyi ayi > iyi eyi / _
iyi uwu > ī ū / _
eyi eyey > êy / _

7. The dipthong *ey becomes ê.

Code: Select all

ey > ê / _

3. ????



4. Profit!

*ptnéh2ti "spread-3.sg.present" > putnā́ti

*ptnh2énti "spread-3.pl.present" > ptínenti

*pth2éh1s "spread-2.sg.imperfect"> pútās

*pth2h1té "spread-2.pl.imperfect" > pútāte


*ǵnh1yétoy "be born-3.sg.present" > šnā́yetai

*ǵnh1h1tó "be born-3.sg.imperfect" > jínāta


*dʰǵʰmṓn "human-N.sg." > *jmũ > jimũ

*dʰǵʰménm "human-A.sg." > *jménun > jímnun

*dʰǵʰmnés "human-G.sg." > jimnés


*dʰǵʰomíHos "earthling-N.sg." > tšamíyas