Porphyrogenitos' scratchpad and intermittent glossolalia

[Porphyrogenitos]

Inspired by the "Germanic Spanish" thread - now I'm wondering about a Germanic language (Gothic, presumably) that stayed in Greece and developed like late Koine and medieval Greek.

I'm thinking it would undergo umlaut, unlike historical Gothic, to provide front rounded vowels to "feed into" the Koine sound changes. It would have a Greek-based orthography (which, in the modern-day, would be a "deep" orthography). The name, I'm thinking, would be Raza Gytiska Ραζα Γυτισκα. (Or Gytska Γυτσκα?) The vowels:

Orthographic / Old Gytiska / Modern Gytiska

ι / i / i
ει / iː / i
υ / y(ː) / i
η / e(ː) / i
οι / ø(ː) / i
ɛ / ɛ / e
αι / ɛː / e
α / a / a
ου / u(ː) / u
ω / oː / o
ῳ / ɔː / o
ο / ɔ or o / o

Not all long/short vowel pairs were explicitly distinguished in the orthography, though they may have occasionally been distinguished by a macron or acute. In any case, length is not distinctive in Modern Gytiska.

Consonant changes:

Gothic /b d g/ lenite to fricatives everywhere except after nasals
Although their voiceless allophones remain as such and are written in the orthography
Gothic /w/ merges with /b/ as /v/
/ʍ/ merges with /f/
/h/ is dropped
<ddj> (whatever its exact value) becomes <γι> /gj/ [ʝ]
The clusters /zd/ and /dz/ merge with /z/

Changes then progress similarly to medieval Greek.

A small preview:

Αττα ουσαρ, θου ιν ιμνα, [ata usar θu in imna]
Βειναι ναμω θειν [vine namo θin]

Of course, there are many grammatical and inflectional changes not seen here.

[Porphyrogenitos]

Revising the vowels to something simpler. No umlaut. Length distinctions merge before beginning of orthographic tradition, along with back vowel shift of ɔ > o > u > y. As it happens, this still gives the name Γυτσκα Gytska.

Although...hmm. I'm making it so that Bible Gothic long /a/ splits off in quality before neutralization of length, becoming [ɒ] (or perhaps [ɔ], after the back chain shift) - later merging into /o/. I just don't want to let either omicron or omega go to waste!

Orthographic / Bible Gothic / Old Gytska / Modern Gytska

ι / i(ː) / i / i
η / eː / e / i
ε / ɛ(ː) / ɛ / i
υ / u(ː), iu̯ / y / i
ου / oː / u / u
ω / ɔ(ː) / o / o
ο / aː / ɔ / o
α / a / a / a

[Porphyrogenitos]

Romlang that goes through the High German Consonant Shift and has some other German-like features. I'm sure this has been done before, but I wanted to provide my own take on it.

Rough list of sound changes

Shared with Vulgar Latin/Proto-(Western)-Romance

• Neutralization of intervocalic /b/ and /w/ to [β]

• ns → s

• tj, kj → ts

• kw → k / _[+front]

• Development of Western Romance 7-vowel system (5 vowels in unstressed syllables)

• Allophonic vowel length in open syllables

• Note: No palatalization of /k/ before [+front]

Independent development

• Intervocalic d g → ð ɣ

• s ts → s̠ ts̠

• Voicing of s̠ → z̠ except in consonant clusters

• Intervocalic and word final p t k → ff ss xx (→ f s x finally)

• Word-initial, geminate, and post-sonorant p t k → pf ts kx

• Word-initial and geminate b d g → p t k

• β ð ɣ → b d g (word-initial b d g reintroduced through loans)

• Vowels shortened before new geminates

• Voicing of initial f → v (new initial /f/ introduced through loans)

• a-umlaut: before /a/, i e u o → e ɛ o ɔ

• i-umlaut: before /i/, e ɛ u o ɔ a → i e y ø œ æ

• Unstressed word-internal vowels reduced to [ə] and [ɪ]

• Final unstressed /i e u o/ reduced to [ə] and [ɪ]

• Final [ə] and [ɪ] deleted; prop or epenthetic [ə] inserted for phonotactics

• Final unstressed /a/ reduced to [ə]

• Vowel length retained and phonemicized after loss of final vowels

• Final /Cj/ clusters (palatalized consonants) become /jC/, projecting an i-offglide onto the preceding vowel, forming new diphthongs

• sk → ʃ

• ts̠ → tʃ

• Remaining s̠ (i.e. in consonant clusters) → ʃ

• z̠ → z

• Gemination lost

• Long i y u → ai øi au

• Long ɛ ɔ œ → ie uo yœ → i u y, also new i u y from other sources

• æ → ɛ

• Merger of mid-low and mid-high vowels

Hmm, okay, I think that's about it.

Nominal declension

• Like Romanian, maintains a distinction between a nominative-accusative and a genitive-dative case

• Masc-neut-fem gender system

• Nom-acc forms descended from the Latin accusative; except nom-acc plural, which is from the Latin nom plural

• Gen-dat forms generally from Latin dative in the singular and (like the Romanian definite article) from the genitive in the plural

• I will provide more detail later

Some very quick samples:

es pfater bekkel - the old father (nom/acc)
sos pfäter bekkel - the old fathers (nom/acc)
si pfater bekkel - the old father (gen/dat)
sor pfäter bekkler - the old fathers (gen/dat)

sa kchechene bekkle - the old kitchen (nom/acc)
sas kchechenes bekkles - the old kitchens (nom/acc)
si kchechen bekkel - the old kitchen (gen/dat)
sar kchechener bekkler - the old kitchens (gen/dat)

so pflatsch bekkel - the old palace (nom/acc)
sa pflatsch bekkle - the old palaces (nom/acc)
si pflatsche bekkle - the old palace (gen/dat)
sor pflatscher bekkler - the old palaces (gen/dat)

Now, where will it be set? Well, if I really want it to have the New High German diphthongization of the high vowels, it will have to be located somewhere even further north or east of Romansch. Maybe near the headwaters of the Lech in Vorarlberg? I would also make it participate in some early changes with Romansch, but I'm not sure where to find info on Romansch historical phonology.

[Porphyrogenitos]

Was toying with the idea of "a language where every word is a morphological verb" yet again. And again with the stipulation that it be morphologically minimalistic, unlike the Iroquoian languages which inspired the concept. This time the sesquisyllabic structure of Mon-Khmer languages was providing the inspiration, and the idea is that the sesquisyllable itself would function as the pronominal prefix, encoding person, number, gender, and agent/patient, like Iroquoian, but perhaps a bit richer than Iroquoian, with the prefix taking on reflexivity/reciprocity and perhaps an applicative function as well. Meanwhile, the full syllable would be the verb root.

How to handle, well, everything else the verb and verb phrase would need to do, without resorting to affixation like Iroquoian? (Since I want to keep words fairly minimal and stick to the typical form of a Mon-Khmer word.) Well, I'm thinking there could be some kind of ablaut or apophony in the root that could handle a limited number of those functions. And I would allow for compounding, with two full syllables/roots preceded by a single pronominal prefix, although it wouldn't necessarily be preferred. There could also be serial verbs, where both are fully inflected, or where one (despite being formally identical to a full lexical verb) has an auxiliary usage (like in Iroquoian).

And actually, the morphological verbiness of words wouldn't be absolute - there would be interjections, ideophones, and certain proper names or loanwords could be morphologically simple.

Some possible words:

kebay [kəˈbaj]
ke-bay
3s.ᴍ.ᴀɢᴛ-burn
'he burns it', '(of a person) burner'

As in Iroquoian, agent pronominal prefixes without a specified patient have an implicit third-person inanimate object (uh, I think it works something like that, iirc).

bay [baj]
∅-bay
3s.ɪɴᴀɴ.ᴘᴀᴛ-burn
'it is burnt', 'something which is burnt'

I am thinking that the third person inanimate patient prefix will be a null morpheme.

prelakbay [prəˌlakˈbaj]
pre-lak-bay
3s.ᴍ.ᴀɢᴛ/3s.ᴍ.ᴘᴀᴛ-hit-burn
'he brands him'

A possible example of a compound ("brand" = "hit-burn"?) and a transitive pronominal prefix.

nemyu [nəˈmju]
ne-mju
3s.ɴʜ.ᴀɢᴛ-be.cat
'it is a cat', 'cat'

I am thinking that for "nouny" statives, ones that refer to animate entities may take agent prefixes. Note the non-human gender marker.

pok [pok]
∅-pok
3s.ɪɴᴀɴ.ᴘᴀᴛ-be.tree
'it is a tree', 'tree'

As an inanimate, 'tree' gets the phonologically null inanimate patient prefix.

Maybe there could be some kind of alternation where, say, definite inanimate arguments of stative verbs take agent prefixes instead? Like:

empok [əmˈpok]
em-pok
3s.ɪɴᴀɴ.ᴀɢᴛ-be.tree
'it is the tree', 'the tree'


Landan [ˌlanˈdan]
London
'London'

An example of a loanword and a proper name, which takes the two-syllable form of a compound, while being monomorphemic.

Also, a big block of possible sesquisyllables:

Code: Select all

me	mi	mu	mer	mel	mem	men	meng
ne	ni	nu	ner	nel	nem	nen	neng
pe	pi	pu	per	pel	pem	pen	peng
te	ti	tu	ter	tel	tem	ten	teng
ke	ki	ku	ker	kel	kem	ken	keng
che	chi	chu	cher	chel	chem	chen	cheng
be	bi	bu	ber	bel	bem	ben	beng
de	di	du	der	del	dem	den	deng
ge	gi	gu	ger	gel	gem	gen	geng
je	ji	ju	jer	jel	jem	jen	jeng
se	si	su	ser	sel	sem	sen	seng
he	hi	hu	her	hel	hem	hen	heng
le	li	lu	ler	lel	lem	len	leng
re	ri	ru	rer	rel	rem	ren	reng
we	wi	wu	ur	ul	um	un	ung
ye	yi	yu	ir	il	im	in	ing
pre	pri	pru	prer	prel	prem	pren	preng
tre	tri	tru	trer	trel	trem	tren	treng
kre	kri	kru	krer	krel	krem	kren	kreng
kle	kli	klu	kler	klel	klem	klen	kleng
kwe	kwi	kwu	kur	kul	kum	kun	kung
kye	kyi	kyu	kir	kil	kim	kin	king
pye	pyi	pyu	pir	pil	pim	pin	ping
sye	syi	syu	sir	sil	sim	sin	sing
sre	sri	sru	srer	srel	srem	sren	sreng
e	i	u	er	el	em	en	eng

<e> represents schwa. Everything else is basically what you'd expect. Roughly this number of sesquisyllables could give me over three times as many pronominal prefixes as Mohawk, allowing for a wider range of grammatical categories to be encoded.

I'm not intending to make it very phonologically exotic. I might even be lazy and just make it have a voicing contrast in the stops, although if I wanted to go with a SEA vibe I'd make it a three-way distinction between plain, aspirated, and voiced. No tones!

[Porphyrogenitos]

Okay, yet again again revising a particular concept of mine I've posted about before.

On numerous occasions I've toyed with a language where all syntactic relations are encoded through four basic constructions - modified-modifier, subject-predicate, action-goal, and apposition. In basically all of my previous drafts I had tried to make it have exclusively monosyllabic, morphophonologically self-segregating roots that head-marked via internal modification for each of these relationships. The results were always... workable, but kind of damn ugly.

I have found this concept interesting because I have never been a big fan of syntax, thus the minimalistic strucure, but it is also a way to allow me explore syntax by building up more elaborate constructions from these basic ones. I have also realized that what I want to do with it is eventually actually use it, and build the language up through use - creating a corpus that I can go back and refer to. This minimalistic structure allows me to do that with a pretty low bar for entry - I don't have to think of declension tables and so on before diving in.

But I have decided to give it another go, this time giving up the monosyllabic concept and adding just a touch more naturalism, giving it a more relaxed and natural feel when spoken.

Phonology

I still want it to be, at some level, morphophonologically self-segregating - it should be obvious where the word boundaries are just from seeing a raw transcription of the phonetic data, or seeing it written without spaces. Why do I want this? It's just kind of a leftover aesthetic requirement from some of the loglangs that inspired this concept.

In keeping with the minimal syntax, I am giving it a rather minimal, slightly Polynesian-inspired phonological structure. This is also to keep me from being tempted to add morphological/phonological alternations - I want pretty much all the lexical action to be via the syntax.

/m n/ ⟨m n⟩
/p t k ʔ/ ⟨p t k '⟩
/v s h/ ⟨v s h⟩
/ɾ l j/ ⟨r l y⟩

/i e a o u/ ⟨i e a o u⟩

Syllable structure is CV(V), although it is not so much the syllable that matters as the mora. That second, optional vowel is an additional mora.

Glottal stop does not contrast with zero at the beginning of a word and is not written there.

All vowels are neutralized to schwa in unstressed syllables. A vowel in this position is written ⟨e⟩. Although as we will see, this context only occurs in a limited number of items.

Roots

All lexical roots must be bimoraic.

Thus we can have disyllabic roots like kata, mina, po'a, ha'a, ipa, etc.

We can have roots with a sequence of two identical vowels, which is realized as a long vowel, like kaa, mii, haa, sii, vee, etc. In principle these could just be written as ka, mi, ha, si, ve, etc, but the double vowel prevents any ambiguity in the case that words are run together without spaces, and reminds the reader that these vowels are long. It also distinguishes long /e/ from schwa.

And we can also have a sequence of two different vowels, which may be realized as a hiatus or a diphthong - it doesn't really matter, the syllabicity isn't contrastive, just the presence of two morae: kua, mia, poa, eo, au, sei, etc.

There can never be a root that deviates from this structure - loanwords are hammered into it, whatever distortions it takes. France becomes rasa, Britain becomes rita, Deutsch becomes toi, America becomes meri, Constantine becomes kota. There are 3900 possible root words, so there should be enough to go around.

I don't intend to use all 3900 possible roots, in any case. I want to keep the lexical reasonably minimalistic - not Toki Pona minimalistic, but enough that a lot of lifting is done by syntactic "compounding".

Unstressed syllables

Unstressed syllables do not occur in roots. In fact, it would be better to call these reduced syllables, since stress is not really an operative feature at the word level.

The only words with reduced syllables are particles, and there is a very limited set of these: me, pe, re, se, ke. That's it. In principle I could decide to add more particles in the future, but I think I will stick with these. Unlike lexical words, these words are not even considered to have a mora. Their vowel may as well be epenthetic. I selected them to be maximally distinct and reasonably unobtrusive in pronunciation.

Morphology

I do intend for there to be almost no morphology, but I will make a couple concessions within the constraints of the phonology I have outlined. I will allow for blends, probably composed of the first morae of two words: vita + maka = vima.

I will also allow for reduplication, probably of initial morae once again, I assume mainly for hypocoristic purposes: toma 'Thomas' becomes toto.

Syntax

I have actually brought back a syntactic relationship from a very early draft of this language that I later abandoned: topic-comment. But since I am now going in a slightly more naturalistic direction, I thought it would be appropriate to bring it back. More on that below.

There are no rigidly defined word classes - all lexical words can, in principle, fill any slot in these constructions, though some will fill certain ones more often than others.

Apposition

Apposition is somewhat semantically underspecified, as are all these relationships, really. It encompasses conjunction as well as what might be termed 'apposition of identity', which of course sometimes overlap. You can think of it as creating dvandva compounds. This also results in it handling some kind of noun-adjective-like relationships.

mama papa 'mother and father', or if taken as a lexicalized compound-like pair, 'parents'.

kesa kota 'Caesar Constantine' i.e. 'the Caesar, Constantine'

pili site 'bird little' i.e. 'bird, the little one', or 'a bird and a little one'

mama pili 'a mother and a bird' or 'a mother bird' i.e. 'the mother, the bird'

Other syntactic constructions can be used, if too much ambiguity arises with apposition.

Modified-modifier

This is marked with re. This is also a very semantically underspecified and polysemous construction.

It may be used for adjective-like modification: pili re mama 'a motherly bird, a mother bird, a bird that is a mother' (but also 'bird of someone's mother', see below!)

It may be used to form relative clauses: tata re temo 'a father of drinking, a father who drinks'

It may be used for possession: mama re pili 'mother of a bird' (but also 'a bird-like mother', see above!)

Also in these cases, further syntactic specification can be used to reduce ambiguity.

Subject-predicate

Marked with se. This can encompass copula-like and subject-verb like relationships:

pili se hoa 'a/the bird is big'

mama se temo 'mother drinks/is drinking'

I have also introduced the particle pe, which marks negation of predicates and predicates only. It follows the predicate:

pili se hoa pe 'a/the bird is not big'

mama se temo pe 'mother is not drinking'

Action-goal

Marked with ke. Encompasses both verb-object and movement-location relations. It is here where we start to see syntactic nesting: A phrase bound in one construction can participate in another construction.

temo ke lia 'to drink water, drinking water, [someone] drinks water'

mama se pana ke lia 'mother goes/went to the water'

This also raises the issue of syntactic ambiguity. This language is not intended to be Lojban, despite some of my inspiration coming from loglangs. mama se pana ke lia could in principle be parsed as:

[[mama se pana] ke lia]

with mama se pana 'mother goes' being an "action" unto itself, affecting the target of lia 'water'. Maybe this kind of makes sense, in the case of a subject-predicate pair. Here's a tougher example:

mama re site se temo ke lia 'little mother drinks water', or:

[mama re [site se temo ke lia]] 'mother of the little one's drinking of water' or even funnier sounding, 'mother of the fact that the little one is drinking water'

Note that 'mother of [the little one who is drinking water]' is precluded; that would have to be mama re site re temo ke lia, 'mother of the little one of drinking water'.

Okay. So, as we can see, this whole thing relies on some parsings being much more likely than others. We are probably going to read and hear mama re site se temo ke lia as 'little mother drinks water', or maybe 'mother of the little one drinks water'. But again, there would still be ways to disambiguate these. Perhaps mama re tola site se temo ke lia - 'mother of the one the little (i.e. the little one) drinks water'.

We also face the question of where to put pe. I think it is actually a matter of scope and there could be multiple valid ways of doing it. It depends on how closely linked a modifying phrase is to the predicate.

We could have:

pili se temo ke lia re site 'a bird drinks water a little bit' or 'a bird drinks a little water' (more syntactic ambiguity)

and as its negation

pili se temo ke lia re site pe 'a bird does not drink water a little bit' or 'a bird drinks water a little bit NOT'

or

pili se temo ke lia pe re site 'a bird drinks water NOT, a little bit'

So pe can actually help reduce syntactic ambiguity - we can see above that it forces the interpretation of re site as modifying the phrase [temo ke lia], since lia 'water' is "enclosed" within the phrase by pe. Although this then introduces a semantic question - is this saying that the bird does not drink water, but only to a small degree? That is, its not-drinking of water is rather small? I suppose that must be what it means. Gosh, this is already getting to be too much syntax for me.

Topic-comment

Not much to say here for the moment. Marked with me. Works similar to topic-comment in Korean and Japanese.

pili me temo 'as for the bird, it's drinking'

It can also topicalize entire phrases:

pili se temo ke lia me site 'as for the bird's drinking of water, it's [only] a little'

Okay, so this is my toolbox I hope to work with in developing this language further. I may or may not end up making more tweaks. All vocabulary here is provisional, but I may end up keeping some of it.

[Porphyrogenitos]

A bit more regarding pe

Allowing an unstressed neutral particle that did not strictly define a unique syntactic relationship was a concession I made to naturalism. In another concession to naturalism, I have given it a diachronic origin - a "virtual" one, anyway, since I am not placing this still-unnamed conlang in any fictional or constructed historical setting, at least not for the foreseeable future.

pe, an unstressed particle that negates a preceding predicate, originates from a lexical word, poa, which has the meaning 'not, does not, is not'. Like all roots it is principle polyfunctional, though I expect it to most commonly occur as either a predicate or a modifier.

Some examples of its use:

pili se temo ke lia?
'Is the bird drinking water?'
poa.
'It is not.'

pili se temo?
'Is the bird drinking?'
poa ke lia.
'Not water.' or 'It is not doing it [drinking] to water.'

This language thus has a kind of echo response system for yes/no questions - or rather like Latin, it has a variety of adverbs and particles that express attitudes toward the truth value of a statement, one of which is poa.

mama se pana ke lia re poa.
'Mother doesn't/didn't go to the water.'

It's weird to translate it like this but we could also think of it as literally meaning 'Mother went to the water not-ly' or 'Mother went to the water, in fact not.'

In principle, I think the re poa construction should be synonymous with a simple postponed pe:

mama se pana ke lia pe.
'Mother doesn't/didn't go to the water.'

However, I can imagine poa being used in emphatic contexts or to contradict another speaker:

mama se temo ke lia re poa!
'Mother did not drink the water!"

The two might be used in combination in the manner of the English didn't not:

temo ke lia?
'[did you] drink water?'
temo ke lia re poa pe...
'I didn't not drink water...'

So, diachronically, unstressed pe would have originated from a reduced form of the re poa construction, where re was eroded down to nothing and poa came to have a reduced vowel. This is rather like many instances of grammaticalization where a form is reduced and becomes a grammatical element, while the original form still remains available for use, like English I'm gonna leave (exclusively grammatical interpretation) vs. I'm going to leave (can be interpreted lexically or grammatically).

As it happens, I am also thinking of a diachronic origin for the topic marker me.

And, unrelatedly, I am thinking there might actually be a single instance of sandhi in the language: The particle re might become le after words with a second syllable beginning in /ɾ/: tasi re keo vs. aru le kasu. Just to make it come off the tongue a bit easier.

[Porphyrogenitos]

Toying with phoneme frequency and its effects on phonoaesthetics using the Lexifier app. Here's a sample wordlist and the (naturally very simple!) code I used for it:

ate
ase
ehe
era
mana
mare
naa
nai
nao
nata
neo
nere
nito
noo
numo
pala
pona
tama
tano
tata
tasa
tahe
tene
tesi
tehe
tisa
kee
voe
sano
sake
sia
hoka
lao
lovo
rana
repu
rero
ritu
roti
roke

Code: Select all

letters: a e i o u m n p t k ' v s h l r y

C = t n s ' r h k m p l v y
G = t n s r h k ' m p l v y
V = a e o i u

words: CVGV CVV

filter: ^' > 

[Porphyrogenitos]

Okay, the first translation of an actual (very brief!) text. I've also begun compiling vocab, and tweaked the frequencies of the consonants to create a slightly different aesthetic in the vocab.

I am now using the following glossing abbreviations for the particles:

re - ᴍᴅ (modifier)
se - ᴘʀ (predicate)
pe - ɴɢ (negation)
ke - ɢʟ (goal)
me - ᴛᴘ (topic)

Bashō's "Frog Haiku": See here for many translations.

lia site re seo
water little ᴍᴅ year

pata se na'a
frog ᴘʀ jump

lia se lane
water ᴘʀ sound

This is, of course, tentative.

My use of the modifier re seo is an idea I had for a way to express the meaning of 'old, ancient, aged' - to essentially say that something is "of years" or "of the years". There may also be an ordinary adjective meaning 'old' but I thought this was more interesting for the purpose of the poem. Of course, in a different context it could be interpreted as "yearly", "annual", or "of a [single] year".

For 'pond' I am simply putting 'water' and 'little' in apposition - 'the water, the little one'.

'Frog jumps' is an ordinary subject-predicate construction.

Rather than lane re lia 'sound of water', I decided to go with lia se lane, with lane 'sound' serving as a predicate with the subject lia 'water'. So essentially 'water does a sound' or 'water resounds'. In principle it could be interpreted as 'water is a sound'. I thought this solution was more surprising and dynamic, and thus in the spirit of the final line of a haiku.

A more idiomatic back-translation into English:

a pond of many years
frog jumps
water resounds

A less idiomatic back-translation:

the little water of years
frog jumps
water does a sound

[Porphyrogenitos]

A sketch of an auxlang-inspired-ish scheme - I possibly posted an earlier revision of this somewhere.

Phonology

/m n/
/p t k/
/b d g/
/ts/
/v s/
/l r j/

/i e a o u/

Maybe I'll add /z/ or /h/.

The only word-initial clusters are (stop/fricative)+(l,r,v). No /j/ clusters. Coda can be /n t v s l r j/. Coda /j/ may not follow /i/.

Stress is always penultimate (on multisyllabic words).

Word classes

Lexical words are always minimally a monosyllabic root plus an inflection. That is, a word like ka cannot be lexical, but aka or kaja could. Native roots have a strong preference for a monosyllabic shape, and loanwords are adapted as monosyllables if possible.

Nouns, verbs, and pronouns all take the same set of inflectional endings, but (pro)nominal and verbal endings carry slightly different meanings.

Some roots are inherently nominal, some are inherently verbal.

Pronouns inflect like nouns but take the form of a C(C) root plus an inflection - so, like ja or tra.

Single-vowel and vowel-coda monosyllables (e, aj, an, us) are reserved for interjections and grammatical particles.

Nouns

Nouns have five cases, if we want to call them that:

Nominative: vala - land (subj.)
Accusative: valu - land (obj.)
Genitive: vale - of (a/the) land, land's
Dative: valo - to (a/the) land
Predicative: vali - (it) is (the/a) land

Nouns do not inflect for number. However, they do inflect for possession:

valar - my land (subj.)
valuv - our land (obj.)
vales - of your (sg.) land
valol - to your (pl.) land
valaj - his/her/its land
valon - to their land (subj.)
valut - one's/someone's land (obj.)

With the predicative form, these personal endings mark the person being predicated: valin - they are land(s)

In the case of the third person singular, the /j/ is absorbed into the /i/ suffix: vali - (it) is (the/a) land (as illustrated above)

Thus, as we will see for verbs, third person singular inflection on verbal and predicative forms is essentially unmarked.

Pronouns

Pronouns work basically the same way as nouns. They also take predicative forms, but these are naturally rarer, seeing as there are fewer occasions to say "you are me" and so on. But you might end up saying ri 'he/it is me' or jir 'I am him/that one' when you answer the phone.

ja - 1sg
va - 1pl
sa - 2sg
la - 2pl (this form may also be used as a polite form)
ja - 3sg
na - 3pl
ta - third person impersonal, used like English 'one' or like the Finnish 'zero-person' construction.

There are other types of pronouns, such as demonstratives and relative pronouns, but I have not decided on forms for these yet.

Nouns may be compounded, with the vowel of the first element affecting the meaning of the word. There is a strong preference for endocentric compounds (which are always head-final).

Verbs

Finite verbs have one and only one form: The one ending in -i, marked for person with the pronominal ending discussed above, with 3sg being unmarked. vrati 'he drinks, he drank, he will drink', vratir 'I drink, I drank, I will drink', and so on. All information regarding tense, aspect, etc must be conveyed through separate words, or through compounding.

Multiple verbs can be chained together into serial verbs, though I haven't decided what exactly this will look like.

All verbs have a deverbal nominal form that inflects like a noun, except for the predicative case. It is semantically to the English -ing form when used as a noun: vrata 'drinking', vrate 'of drinking', etc.

Verbs will probably also take the place of most prepositions: There will be words like nari 'go into', juki 'be on top of', and so on, that can be used alone, or in compounds, or as part of serial verbs.

Word formation

The main conceit of this language, aside from its auxlangy minimalistic inflection, is that all word formation is handled through compounding. Instead of a suffix like the English -er, you would use a compound like vratikena 'drink-man, 'drinker'.