so far, numbers are very fun in this language (still unnamed, dw)
Every number has 2 words for it, which we've termed "count" and "quantifier" words.
The count words are generally used only when referring to the number itself, as in counting or in math (think like "one plus one is two" type contexts), whereas quantifier words are used when quantifying a nominal.
The only exception for this is the number "one". Unlike other numbers, its count word is transparently an adjective, and is used as such when quantifying nouns when it's the only number being used. (The quantifier word is used in more complex numbers [e.g. 11], as well as for contrast with other numbers)
The quantifier words are also just nouns which mean "group of [#] size", so like "singleton", "pair", "trio", "quartet", etc. And as such, nouns they quantify are put in the genitive plural.
Additionally, the numbers in this language are base 20, rather than decimal. There are unique number words from 1 thru 10, 15, and 20.
11-20 are [literally translated]
11: one and ten
12: two and ten
13: three and ten
14: one against fifteen
15: fifteen
16: one and fifteen
17: two and fifteen
18: two against twenty
19: one against twenty
20: twenty
Nt- and St-stems are missing, because they are gone (RIP).
Yes, Dt-stem is identical to Gtn. Since Dt is the reflexive / reciprocal versions of Gtn, those forms will probably be accompanied by some version of the word "self".
Vowel deletion after personal prefixes and before case / number suffixes is not shown.
Declension will probably be as regular as possible for all forms (-as for singular oblique forms and -in in all plural forms)? If there are exceptions, they will be in the nominal versions of G-stem.
The forms can easily accomodate non-triliteral stems. For example, if we want to form an agent noun from Dt (pluraction reciprocal) stem of the root tlgrm:
duplicate the penultimate consonant (D-stem): tlgrrm
insert t after the first consonant (-t-stem): ttlgrrm
apply pattern CVCCVC to the last 4 consonants: ttl + gVrrVm
apply pattern CV(C) * n to the leading consonants: tVt + lV + gVrrVm (this is where non-finite forms start to diverge with finite ones, which use C(C)V * n)
fill in all vowels except the last with i, and the last with a: titligirram
voila, we have our participle! (since its reciprocal, it will probably be more organic to use in the plural: titligirramin)
In Armenian, when we want to say “damn you” or “go to hell”, we use the expressions "գրողը քեզ տանի" [groxy qez tani] or "գնա գրողի ծոցը" [gna (kori) groxi tsocy], which translate to “may the writer take you away” or “go and get lost in the writer’s embrace” in English. You might wonder, “Who is this writer-person?” and “Why is it considered a curse?”
According to traditional Armenian belief, Grox (the writer) is a spirit who records a person's deeds during their lifetime, determining the purity of their soul. This concept may be linked to Tir, the god of writing and literature in Armenian mythology. In some interpretations, it was believed that anyone whose name Tir wrote in his notebook would die. This is where the curse "may the writer take you" originates.
During the Christian era, Grox was mistakenly represented as a Christian spirit who no longer recorded human deeds but instead determined each person's fate, inscribing it on their foreheads. Over time, Grox came to be depicted as an evil spirit, sometimes identified with Satan. Thus, the curse "get lost in Grox’s embrace," which originally signified death, took on a more negative connotation. However, this was not originally characteristic of Grox in Armenian traditional beliefs.
So, if you want to get creative with your curses, instead of saying “go to hell,” you can use the phrase “get lost in the writer’s embrace”.
I've been meaning to showcase my conlang for SO LONG. lol. it's really mostly just a script with a few dozen nouns but I think it sounds cool and cozy and I write it everywhere when I get bored.
the blue scroll on the side says "Sho'eyic" (/ʃoejix/ in IPA phonemes). then is shown the 19 glyphs. tried to make it look pretty-- medieval manuscript style 🔥🪻
here's the IPA for further reference, cause my handwriting is a bit messy:
Dass Blaubeeren und Heidelbeeren nicht dasselbe sind.
Blaubeeren wachsen in Deutschland wild und wurden aufgrund ihrer anspruchsvollen Bedürfnissen noch nie kultiviert. Sie sind kleiner, aromatischer und innen blau.
Heidelbeeren sind eine gezielte Züchtung aus einer amerikanischen Verwandten der Heidelbeere. Sie wachsen höher, die Früchte sind größer aber nicht so aromatisch.
Im Alltag sind die Begriffe regional verwurzelt, im Süden sagt man eher Heidelbeere, im Norden Blaubeere.
apparently, you need at least 6 (six!) degrees of phonetic length in order to describe how the 3 degrees of contrastive phonemic length in estonian are actually pronounced
breeding, (animal) husbandry, the management and care of farm animals by which quailties advantageous to humans are developed and perpetuated;
curation, selection, the act of carefully choosing a particular subset of candidates for some purpose;
nurture, upbringing, the environmental influences that contribute to a person's development
Etymology: noun derivative of focar "breed", a Middle Borlish borrowing from contemporary Willem focken "breed (animals)". This descends, possibly via Norse, from Classical German fuccō "I strike, hit; I have sex with". Extension to other winnowing processes is seen from the sixteenth century; in the last senses it is contrasted with engin "nature, innate character".
Y focment d'ollom ny relicquat m'ajoutaç.
/i fɔkˈmɛnt dɔˈlɔm ni ˌre.lɪˈkwat ˌma.ʒuˈtats/
[i fʊkˈmɛnt dʊˈlɔm ni ˌʀe.lɪˈkwat ˌma.ʝʊˈtats]
df breeding of=relic in-df museum 1s.obl=please
I appreciate the museum's relic curation.
Why has the English language turned to gardening instruments for referring to sexual promiscuity? Hos and rakes. Silly ass language. I propose we bring this term back only for word play reasons.
“This place is full of so many rakes and hos, you’d think it was a farmhand’s shed”
Of course ounce is a terrible unit of measurement, it's hard to measure things in snow leopards and - what's that? wait a minute post canceled what do you mean ounce (meaning "snow leopard") comes from once which comes from lonce (Old French) and lonza (Italian) which mean lynx (meaning "lynx") and the L was dropped because lonce was interpreted as l'once (meaning "the once")
In another post I made on the verbal conjugation in my conlang I still grouped together the first 2 layers into “imperfective <> perfective” opposition, but the more I thought about it, the less clear it became, what exactly that opposition meant and how should it be deduced from already developed morphology, so in the end perfectivity has undergone mitosis.
Atelic vs telic
This is very similar to how Slavic languages structure their aspects: we have base verbs - which are atelic, or in Slavic terminology “imperfective” - and then we slap prefixes on them - which makes them telic, or “perfective”. (The conlang itself doesn’t have any prefixes yet, that’s something I’ll need to work on later. But before I do that, I want to finish Zaliznyak’s “Лекции по русской аспектологии” and find more info about verbal prefixes in Georgian). Imperfective verbs describe an unfolding event - perfective verbs indicate a single point in time, when a certain result was achieved. An action as it is described by an imperfective verb has duration and can be interrupted, while still having taken place, - an action described by a perfective verbs is an instantaneous flipping of the switch between “the result hasn’t been achieved yet” to “the result has been achieved”. Different prefixes are used to form different words expressing differnt kinds of results an action can settle into: рисовать (“to draw”, imperfective, can last, doesn’t have a clear goal), нарисовать (“to draw to completion”, perfective, only true, when there’s a finished picture), дорисовать (“to finish drawing an already half-drawn picture”, perfective, again, finished, when there’s a picture), срисовать (“to copy a picture”, perfective, only true, when we have a finished copy) etc.
Singulative vs pluractional
Unlike the previous opposition, this one is finished in terms of morphology: singulative is the unmarked form, pluractionality is expressed through changing a verb’s stem:
G -> Gtn (for processes) or D (for actions)
Gt -> Dt
D (transitive/causative counterpart of G-stems expressing processes) -> Dtn
S -> Stn
N -> Ntn
In terms of its meaning, pluractional verbs express some form of event plurality:
there are many subjects doing the action
there are many objects being affected by the action
the action is done repeatedly
the action is done carefully, or with intensity
the action lasts unusually long
While some of those things can be expressed in Russian (like, вырисовывать - “to draw slowly and carefully”, or танцевать “to dance” vs вытанцовывать - “to dance with abandon”), this is where it definitely diverges from from the aspectual system of the conlang, because in Russian repeated actions are always imperfect, and here event plurality and telicity are decoupled from each other, and each of the 4 combinations is possible. (Important to note, however, that telic verbs are instantaneous changes; the preparatory stage leading to them is implied, but is not what is being talked about - which means that while a pluractional atelic verb can describe a single action that just lasts very long, pluractional telic verbs describe a plurality of those moments of achieving the result, and thus a plurality of actions).
Aorist vs imperfect vs perfect
This last layer is, probably, somewhat similar to what Romance languages have??? This last section is going to be somehow even more rambly then the rest of this psto, I’m terribly sorry…
aorist hides the internal structure of an event (the specific combination of 2 previous aspectual layers) presenting it as an indivisible point; because of that the aorist forms cannot be used to describe a background action (no “while A was happening”) or talk about the present, because that would put another action or the present moment in the foreground of an unfolding action, separating it into before and after, which a single indivisible point can’t have; the primary use of this aspect is to narrate a story (“A happened, and then B happened, and then C happened”); the internal structure is not thrown out, though, aorist can freely combine with atelic and pluractional forms, which express duration and repetition, it’s just that that structure cannot be acknowldged and engaged with by other verbs
imperfect aspect presents the 2 previous layers of aspects as they are, unbounded by any time frames; this is the aspect that should be used to describe background actions and to talk about the present, it can be used to describe regular occurences and habits. It’s easy to see, how imperfect can be atelic (have a duration) or pluractional (repeat multiple times), it may not be entirely clear how imperfect can be singulative and telic: my interpretation is that the boundedness of the imperfect is applied to the preparatory stage of a telic verb, in other words it places the reference point expressed by the tense (past or non-past) into the preparatory stage and states that the action is going to reach the result at some point in the future from the reference point (essentially, this is conditional and simple future, “would” and “will”)
perfect aspect describes a result of an action completed prior to a reference point. Important to note, that atelic verbs do not resolve into any result! But they can combine with the perfect aspect, which in this case takes on evidential meaning: we don’t know if “someone has finally done something”, but we know that they “have been doing” it, we can see the evidence of the past action - this is the atelic perfect
These 3 aspects also have a developed morphology: imperfect and perfect add consonants of the verb root, and aorist and imperfect put certain restrictions on the vowels immediately before and after the last root radical.
I think, that’s it. Looks… relatively coherent, probably functional. Practice could reveal some additional creases in need of being ironed out, but I’m not at the stage, where I feel ready to proceed to generate heaps of words and example sentences
I did not want to plan any kind of evolution for my conlang, but the choice of Semitic languages as the main inspiration for the verbal morphology brought with it a vaguely Canaanite Semitic sound system at the initial stages of thinking about the language - and I don’t like it. So, the whole point of writing down the wall of text below is to drag the language towards “more Icelandic/Faroese-sounding with vowel harmony” direction, without breaking the grammar in the process.
The Starting point
In terms of prosody, the only permitted syllable structures are CV (light syllables) and CVV or CVC (heavy syllables). The word is then divided into feet, starting from the beginning. Each foot is formed by 2 consecutive syllables - unless the syllable is heavy and the next syllable is either light or last, in which case the foot is formed by just 1 heavy syllable:
(CV.CV), (CV.CVV), (CVC.CVC) - 1 foot out of 2 syllables, each
(CVV)CVC - 1 full foot plus 1 extra-metric (and word-final) syllable
(CVC)(CV.CV) - 2 feet, the first is 1 syllable long, because the next syllable is light
The primary stress falls on the last syllable of the last non-word-final foot; the secondary stress falls on the last syllable of each foot preceeding the primary stress. Some examples (primary stress is red, secondary stress is blue):
(CV.CV)CV
(CVC)(CV.CVC)(CVV.CVC)
(CVC)(CV.CVC)(CVV)CV
Already at this stage (and while the strict rhythm of the prosody hasn’t collapsed yet) some weak consonants disappear in certain positions, causing compensatory lengthening:
y and w at the end of a syllable lengthen preceding closed vowels: iy, iw -> ii; uy, uw -> uu
h and 2 at the end of a syllable lengthen any vowel: ah/a2, ih/i2, uh/u2 -> aa, ii, uu
2 also lengthens preceding consonant: C2 -> CC
some n (first root radical or infix) assimilate with the following consonant: nC -> CC
In reaction to blurring the lines between geminated consonants and sequences of /n/ + consonant, all morphologically long voiced plosives are realized as “prenasalized” voiced stop, and remain such throughout all further changes: //bː, dː, dʒː, ɡː// -> /mb, nd, ndʒ, ŋɡ/.
…and then the following processes happen, somewhat overlapping with each other (sorry for a lot of words, there are a couple other tables to look at):
Emphatic consonants disappear
The presence of emphatic consonants in a word, first, spreads to vowels in the form of retracted tongue root (with accompanying quality differences); emphatic consonants, then, merge with regular ones; and, finally, tongue root position distinctions disappear, leaving 2 intersecting sets of vowels (and in effect vowel harmony)
Pharyngeal consonants (which are part of the emphatic group) change quality in the following ways: /ʕ/ -> /ŋ/, /ħ/ -> /h̃/ -> /ŋ/ before voiced plosives, /h/ elsewhere. …this is a weird choice, I agree, but it, probably, can be justified by rhinoglottophilia? Out of the rest of the emphatic consonants, T, C and q simply merge with plain t, c, k, and only r (which also triggers retracted tongue root initially) remains as is
Glottal stop (2, /ʔ/), technically, does not belong to this group, but I should mention it somewhere, and it also disappears, so let’s do it here: word-initial glottal stop is lost completely; intervocal glottal stop is replaced with /h/. In all other positions it has already been replaced by compensatory lengthening
Unstressed light syllables lose their vowels
Unstressed light sillables (and that means only fully unstressed) lose their vowels in 2 contexts:
after a vowel (and aw and ay are treated as vowels for the purposes of this rule), or
if the consonant of the unstressed light sillable is a voiceless obstruent and either (a) the preceding consonant is a continuant (not identical to the voiceless obstruent of the light sillable), or (b) the following consonant is a liquid, glide, or non-homorganic nasal:
/ta.ka.ˈra/ -> /ta.ˈkra/ (after a vowel)
/taw.ka.ˈra/ -> /taw.ˈkra/ (after a diphtong)
/-l.ka.ˈta/ -> /-lk.ˈta/ (after a sonorant)
/-s.ta.ˈka/ -> /-st.ˈka/ (/s/ is also a continuant!)
/-s.sa.ˈta/ (doesn’t change, because the voiceless obstruent and the preceding continuant are identical)
/-t.ka.ˈra/ -> /-t.ˈkra/ (before a liquid)
/-t.ka.ˈma/ -> /-t.ˈkma/ (before a non-homorganic nasal)
Important to note, that compensatory lengthening in place of 2 and h is no longer active at this point, meaning, starting with original forms:
(ta.mah)ra -> tmaara (h disappears during compensatory lengthening)
(ta.ma)(ha.ra) -> tmahra (h is preserved intervocally, and then doesn’t disappear (!) as the vowel from its unstressed light syllable is removed)
The rules are slightly different for light syllables with /h/, y and w:
the vowel after /h/ (which may be a reflex of any of h, H or 2), when /h/ is preceded by a consonant, does not disappear before a voiceless consonant and causes devoicing of the following voiced consonant
h after a vowel becomes /ŋ/ before a voiced plosive (in other words, it behaves like H, regardless of whether it has been h, H or 2 historically) and causes devoicing of the following sonorant
the vowel after y and w only disappears if the previous vowel is short; sequences Vy, Vw become ii and uu, if V is either i or u
Some examples:
/as.ha.ˈfa/ - the syllable /ha/ does not lose vowel, because it’s after a consonant and f is voiceless
/as.ha.ˈba/ -> /as.ˈhba/ -> /as.ˈpa/ - devoicing of b
/as.ha.ˈla/ -> /as.ˈhla/ -> /as.ˈɫa/ - devoicing of l
/as.ha.ˈda/ -> /as.ha.ˈða/ -> /as.ˈhða/ -> /as.ˈθa/ - see the next section for spirantization of voiced plosives
/a.ha.ˈba/ -> /ah.ˈba/ -> /aŋ.ˈba/ - voicing to /ŋ/, just like H, before a voiced plosive
/a.ha.ˈda/ -> /ah.ˈda/ -> /aŋ.ˈda/ - spirantization of plosives happenes only after light syllables lose their vowels in the 1st context (after a vowel)
/a.ha.ˈma/ -> /a.ˈhma/ -> /a.ˈm̥a/
/a.ha.ˈra/ -> /a.ˈhra/ -> /a.ˈʂa/ - ʂ is the voiceless counterpart to r
/a.ha.ˈfa/ -> /ah.ˈfa/ - normal scenario
/i.ja.ˈba/ -> /ij.ˈba/ -> /iː.ˈba/ - that /iː/ later becomes /ʊi/ or /ɔi/ after the Great Vowel Shift
/u.ja.ˈba/ -> /uj.ˈba/ -> /iː.ˈba/
/uː.ja.ˈba/ - /ja/ doesn’t lose vowel, because /uː/ is long
/a.ja.ˈba/ -> /aj.ˈba/ - normal scenario again
Spirantization of voiced plosives
Non-labial voiced plosives become fricative post-vocally:
d -> ð
dʒ -> ʒ
ɡ -> ɣ
That happens after light syllables lose their vowels in the 1st context, but before they lose them in the 2nd context (see the section above). The effective result is that spirantization happens either after a vowel, or after a consonant cluster: kd -> [ɡd], but skd -> [zɡð] (in consonant clusters voiceness spreads backwards)
Prosody is restructured
As light syllables lose their vowels, new syllable shapes become possible and the old prosody gives way to a new system:
if the last syllable of the word is super-heavy, it has the primary stress
otherwise, if the word has 2 syllables, and the first is light, the primary stress falls on the 2nd syllable
otherwise, if the word has 2 syllables, the stress falls on the first syllable
otherwise, if the penultimate syllable is heavy, it has the primary stress
otherwise, the primary stress falls on the antepenultimate syllable
the first secondary stress falls on the first heavy or super-heavy syllable that is not immediately preceding the primary stressed one
each further secondary stress falls on the nearest following heavy or super-heavy syllable that is not immediately after the secondary stressed one or immediately before the primary stressed one, unless that (super-)heavy syllable is further than 2nd syllable away from the previous secondary stressed syllable, in which case the next secondary syllable is the 2nd syllable from the previous secondary syllable, not immediately preceding the primary stressed syllable (…that’s a lot of words, but the rule itself is probably simpler then it sounds)
Some nonce words as examples:
ta.ˈkawr, taam.ˈkawr - super-heavy last
ta.ˈkar, ta.ˈka, ˈtaa.ka - 2-syllable words
ˈta.ma.kar, ta.ˈmaa.kar - more then 2 syllables
ta.ˌnus.tal.ˌna.gar.ˈram.tin - the 1st secondary stress falls on the first heavy syllable, the 2nd secondary stress falls on a light syllable, because the nearest heavy syllable (gar) is further then 2nd syllable away from the first secondary stressed syllable (nus)
Great Vowel Shift
The vowel shift happens in parallel to other changes to vowels described above. The tables above summarize the process:
“advanced” and “retracted” refer to tongue root positions, and effectively mean “with no emphatic consonants” and “in the presence of at least one emphatic consonant”
short vowels are lengthened under the primary stress AND before a single consonant (“AND” here means “both conditions have to be true”)
long vowels are shortened if they are unstressed OR if they are before a consonant cluster
shortened vowels marked gray and with asterisk only get shortened before a consonant cluster
for the purposes of previous rules a consonant cluster that can start a syllable (i.e. a voiceless obstruent + a liquid, glide or non-homorganic nasal) is considered a single consonant and not a consonant cluster
The End result
consonants marked with asterisk are the results of spirantization of voiced non-labial plosives
consonants marked gray are the results of either voiceness assimilation, or, in case of voiceless consonant, realisation of a sequence //h// + a voiced obstruent after a consonant
consonants with blue consonant cluster in parentheses are marginal, they are realizations of those clusters of root radicals, where the first one is //h// (a historic h, H, or 2)
Phonemic status of those 3 groups of consonants is not entirely clear to me. [v] and [z] are, probably, just allophones - but with others you can have things like: (maa)(si.ra) -> maasra -> [ˈmɔa.ʂa] contrasts with maara -> [ˈmɔa.ɹa]; masba -> [ˈmɛz.bɛ] contrasts with (mas)(ha.ba) -> mashba -> [ˈmɛs.pɛ]; (mas)(ha.da) -> mashda -> [ˈmɛs.θɛ] contrasts with masda -> [ˈmɛz.dɛ] etc.
Yellow vowels are “advanced”, blue vowels are “retracted” and greens are neutral.
As I’ve said before, I’m still thinking about the grammar of the conlang, which means all of the words in this post will be nonce words. No real words were harmed in making of the examples.
There are 2 numbers, singular and plural, and 3 cases, which we will provisionally call nominative, dative and instrumental. Provisionally, because the exact role of those cases is not fully determined, it’s just known that there are 3 of them, with the 2 oblique cases, dative and instrumental, splitting from a single oblique case relatively recently.
The 6 possible form do not form a neat table, with certain markers associated with specific columns, rows or cells of that table (with the exception of the two suffixes) - rather there is an array of possible transformations that can derive one form from another and forms are connected to each other by this relation of “base form to derived form”. There are certain restictions on which transformations can mark transition from one form to another, but no transformation has a strictly defined semantics: the result is that any combination of the forms that is combinatorically possible is a legitimate paradigm.
So, in order to describe the system, I’ll need to answer the following questions:
what are the transformations marking a derived form?
which graphs connecting the six forms are allowed?
what are the restrictions limiting the choice of a transformation when constructing a form?
Transformations
There are 4 possible ways to mark a derived form:
To add a suffix to a form that doesn’t have a suffix. The shape of that suffix depends on the number, is identical for all cases and is *as in the singular and *in in the plural. The final vowel of the base form is dropped.
To make a short stem into a long stem. That transformation is done by applying the pattern CVCVV to the beginning of a stem: the Vs are filled with the first vowel of the stem, the Cs with the 1st and the 2nd consonants. If the first vowel of the stem is long, it is reanalyzed as a sequence *a2, *iy, *uw with *2, *y, *w filling the position of the 2nd consonant. If the original stem started with a light syllable (in other words, CV), a “homorganic” weak consonant added to the end of the pattern in order to avoid hiatus.
To change the first vowel of the stem (or the first 2 identical vowels of a lengthened stem). The 3 vowels form a hierarchy with *a being the least marked and *u being the most marked; derivation is done by increasing the markedness by 1: *a becomes *i, *i becomes *u and *u, not having a more marked vowel to choose from, drops back to *a.
To do nothing. The derived form is identical to the base form. This transformation isn’t used as widely within a paradigm as the other three, because it is the main victim of restrictions.
Connections
There are 2 main ways to be a noun and 1 slightly irregular: you are base singular, base plural or numerically neutral.
Base singular and base plural noun have nominative singular or plural, respectively, as their base form.
From it they form nominative of the secondary number and dative/instrumental of the primary number.
The dative/instrumental form of the secondary number is then formed from the the nominative of the secondary number.
The forms of dative and instrumental of the primary number can be identical or can be split, the forms of dative and instrumental of the secondary number are always identical.
In the paradigm of numerically neutral nouns both nominatives, singular and plural, combine the traits of primary and secondary forms.
Restrictions
The base form of a paradigm, which is either of the nominatives, can have any of the 3 vowels and have a short or long stem, but never has a suffix
Nominative singular never has a suffix
The 2 nominatives and dative/instrumental singular must be 3 (or 4) distinct forms, which means that the 4th transformation, the “do nothing” one, can only form dative/instrumental plural from nominative plural
Stem vowel change has to be applied to the whole paradigm and marks all forms directly derived from the base nominative: either alone, or in combination with a suffix or stem lengthening
Numerically neutral nouns are all stem vowel changing, and both singular and plural oblique cases are formed with a vowel change
The nominatives of a numerically neutral must have different vowels, neither nominative can have a suffix, and if only one nominative has a long stem, the direction of lengthening of the stem between them must be opposite to the direction of increase of markedness of the vowel
Numerically neutral nouns always have the same form for dative and instrumental cases
Examples
The forms are going to be organized in this way:
nom.sg. nom.pl
dat.sg. dat.pl
instr.sg. instr.pl.
the simplest paradigm (base singular, no stem vowel change, no stem lengthening, no oblique case split)
kalm kalmin
kalmas kalmin
kalmas kalmin
base plural with vowel change and oblique case split (*u is the most marked vowel, so it changes to *a; note, also, that instrumental pl. is identical to nom. sg., which is allowed by the rules!)
kalm kulm
kalaam kalmin
kalaam kalm
vowel change with base singular (the last vowel of the stem is dropped before a suffix)
kalmu kilmin
kilmas kiliimin
kilmas kiliimin
numerically neutral noun (only numerically neutral noun can have all 3 vowels in the 1st position; between the nominatives the direction of increased markedness of the vowel (*u -> *a) is opposite to the direction of lengthening of the stem)
kalm kuluum
kilmas kalaam
kilmas kalaam
long stem in base singular
kuluum kalaam
kalaamas kalaamin
kalaamas kalaamin
applying stem lengthening to a long vowel (*kiilim -> (*kiylim -> *kiyiilim) -> *kuyuulim)
kiilim kuyuulim
kuulimas kuyuulimin
kuulim kuyuulimin
applying stem lengthening to a short syllable (*kilim -> (*kilii_im -> *kalaa_im) -> *kalaa2im)
kalaa2im kilim
kiliiyim kulumin
kiliiyim kulumin
Many, many more are possible
Conclusions
the main sources of inspiration for all of this were analyses of noun declension in Nuer (Paradigmatic Chaos in Nuer by M. Baerman and Paradigmatic Saturation in Nuer by M. Baerman and I. Monich) + to a lesser extent one analysis of Arabic broken plurals (Foot and Word in Prosodic Morphology: The Arabic Broken Plural by J.J. McCarthy and A.S. Prince)
given how varied and not easy to generalize the paradigms are on the surface level, the language has to have 1) some exceptions that completely break even the rules described above and 2) words that have more then one possible paradigm
given that the language as a whole is inspired by the Semitic “root-and-pattern” morphology, different choices of transformations making up a particular paradigm will probably at some point acquire some semantics, will be associated with some meaning
After weeks of small tweaks the conjugation of weak verbs has achieved a form that I don’t actively hate.
Some examples:
The main cause of my dissatisfaction with the original approach was mainly “procedural”: bits and pieces of code that applied different deviations from the standard way a form was meant to be constructed were thinly spread over at least half a dozen of different files and functions, which was a mess. The second point I didn’t like, was the fact that however I tried to formulate the rules, when consistently applied they would change the structure of at least some forms to the point of being completely indecipherable. Just ditching the irregularity of weak verbs was, unfortunately, also not acceptable, because I had already settled both on a very extensive system of vowel phonemes (around 20), and a conjugation system that only allow 3 vowels (*a, *i and *u) - even taking into account vowel harmony, caused by emphasis spreading, that would have meant only 5 vowels (plain *a and emphatic *i would both result in [ɛ]) out of 20 would have been allowed in verb forms, which would’ve been strange. Interaction with weak root radicals is the primary way of injecting all the vowels into the verb paradigm.
The new version keeps the step of constructing a new form perfectly regular and predictable in (almost) all cases. Weird things happen after you’ve gotten to this stage, and they are applied all together in one swoop:
*h and *2 in a syllable coda lengthen preceding vowel (which then becomes a diphthong) and drop
*y and *w in a syllable coda lengthen preceding *i and *u and then drop (so, *uy -> *uu, *iy -> *ii, etc.), assimilation to a preceding vowel helps differentiate aorist and optative forms; *ay and *aw are treated as diphthongs
*2 is thoroughly removed from every remaining position: it disappears word-initially; becomes *h intervocally and lengthens preceding consonant (the last one is… weird, I agree)
*n̆ marks a weak *n, which is the *n of prefixes, infixes and first consonant of a root; it fully assimilates to the following consonant, except for *nh, *n2, *nw -> *nn, *nn, *mm (it then partially reappears, because morphologically long voiced stops are pronounced as sequences of a nasal + a voiced stops: *bb -> [mb], *jj -> [ndʒ], etc.)
doubled verbs (verbs with 2 identical consonants in their root) are reanalysed as having in place of their second consonant some sort of virtual fragment that fully assimilates to either following (like *n̆) or preceding (like *2) consonant; that achieves a similar effect to what Semitic languages are doing - they try to avoid separating identical consonants by vowels - without altering the pattern of a form, which is what they actually do in order to achieve that effect
“pharyngeal” consonants aren’t weak; the first version took Hebrew-ish approach to them, meaning *3 and *H couldn’t be geminated or stand immediately before a consonant; ultimately I decided, that treating them irregularly is unnecessary, especially considering that neither of the two is actually pharyngeal; *3 is [ŋ], a velar nasal (!), putting those restrictions on it is bizarre
I don’t think that those rules are ideal, but I also don’t think I can come up with anything better at balancing all the different requirements (relative procedural simplicity, relative transparency of the resulting forms, as few ambiguous forms as possible, full range of vowel phonemes in verb forms…)
When I was little the one thing that baffled me the most was foreign languages and people being able to speak and understand them. It was a very profound feeling of not fully understanding how is that even possible and what even is that thing that I don’t understand how it is possible. Nothing else caused as much utter confusion.
After, I guess, only slightly over a decade of having gotten my English to the point where I can say that I know it, I think I get now how the trick is done: the language is made up of layers upon layers of repeating patterns and as a human being, a member of a species that is, apparently, biologically wired to be able interpret and reproduce those patterns, you can develop a skill of, first, internalising them and, then, letting them express through you your feelings, interests and intentions and be meaningfully and productively involved into the lives of other members of your species, by letting the patterns inform you of their feelings, interests and intentions.
Cool. Great, even. But I guess that realisation has sated my interest in languages in general to such a degree that I find it hard to motivate myself to dedicate more time and effort to actually learn another one. I was, and still is, interested in the root-and-pattern morphology of Semitic languages and the logographic writing system of Japanese/Chinese. And I do feel like the time and effort already spent on German, and to somewhat lesser extent Spanish, Finnish and Hungarian put some kind of obligation on me to continue (content warning: “sunk cost fallacy”). But, I guess, neither of those 2 feeling is strong enough to move me out of a metaphorical ditch my psyche has been stuck into for some years now, so… That’s it. End of obsevation.
After the positions of vowels within a stem are known, there are 4 remaining steps:
add subject prefix and suffix consonants and mark the positions of their vowels with placeholder vowels
replace the placeholder vowels depending on the stem and tense-aspect-mood form of the verb
add the locative and object prefixes, which have their own vowels, as is
pronounce the resulting form (saved for a future post)
Adding affixes
Subject affixes are as follows:
The suffix of the past tense / subjunctive mood of the imperfect and perfect aspects is:
*t after a single consonant and before any suffix
*it otherwise
Replacing placeholder vowels
The first vowel of a D[t[n]] or S[t[n]] stem is always *u
The first vowel that is the vowel immediately following the *n-prefix of N[t[n]]-stem verbs or immediately preceding the first *t-infix is *i
The first vowels before and after the suffix *t that are also after the penultimate consonant of the stem are *i in the past tenses and *u in the subjunctive mood
The vowels immediately before or after the last consonant of the stem are *u in the optative tense and *i in the aorist tense
In imperfect aspects the vowel immediately after the doubled consonant is always *a and doesn’t get overwritten by *i or *u in the past tenses or subjunctive mood, respectively
Vowels between the root consonants of aorist and optative tenses of the G-stem are spefic for each root. The only restriction is that if the vowel of the optative (the vowel between first two radicals) is *i or *u, then the vowel of the aorist (the vowel between the last two consonants) is the same. If the vowel of the optative is *a, the vowel of aorist can be any of the three.
Object and locative prefixes
The object prefixes are as follows:
Locative prefixes aren’t ready yet. 🙃
Examples
We are going to use the nonce stem *klm again
Each example has 4 or 5 stages:
the stem with additional consonants
the stem with vowels
the form with subject affixes
the form with the vowel replacement
the form with object prefix (if there is any)
3m.sg subject, 3n.sg object, perfect indicative of G-stem
ktlm <- because of the perfect aspect
ktalm
giktalmas
giktalmas
gigiktalmas
3m.sg subject, aorist of Gt-stem
ktlm <- because of a Gt-stem
ktalm
giktalmas
giktalmis <- because of the aorist tense
3f.sg subject, 2m.sg object, aorist of D-stem
kllm <- because of a D-stem
kallam
takallamas
tukallimis <- *u because of a D-stem; *i because of the aorist tense
mitukallimis
2m.sg subject, 1.sg object present subjunctive of G-stem
kllm <- because of imperfect aspect
kallam
mikallamat <- suffix caused by the present tense of subjunctive mood
mikallamut <- because of subjunctive mood; *u doesn’t replace the preciding *a because of the imperfect aspect
namikallamut
2pl subject, 1pl object pluperfect of Stn-stem
sttnklm
statnaklam
fistatnaklamtan <- *t of the suffix triggered by the past tense of the perfect aspect
fustatnaklimtin <- *u because of an Stn-stem; *i - because of *t-suffix
gifustatnaklimtin
3f.pl subject, 2pl object present subjunctive of Stn-stem
stknllm <- *n of an Stn-stem cannot be closer to the beginning of the stem then before the 3rd last consonant; the doubling of *l triggered by imperfect aspect of the present subjunctive tense
staknallam
tastaknallamtan <- *t of the suffix caused by the present subjunctive tense
tustaknallamtun <- *u of the prefix caused by an Stn-stem; *u of the suffix - by the subjunctive mood
fitustaknallamtun
3m.pl subject, 1sg object perfect subjunctive of Dtn-stem
kttnllm <- the 1st *t is from the aspect (perfect), the 2nd from the stem (Dtn)
ktatnallam
giktatnallamtan
guktatnallumtun <- *u after the doubled consonant, because the doubling is triggered by an Dtn-stem and not by the imperfect aspect
naguktatnallumtun
connegative aorist of N-stem
nklm
naklam <- the first 2 consonants are separated, because there’s no prefix
naklam <- no subject prefix in connegative
niklim <- the first *i from an N-stem; the 2nd *i - from the aorist tense
2m.sg optative of G-stem
klm
kalma <- optative of a G-stem with no suffix
mikalma
mikilmu <- own vowel of *klm root, could be any other; N.B.: the vowel won’t necessarily be *u, because in subjunctive mood only the vowels immediately before and after the last consonant of the stem are turned into *u
3f.sg pluperfect of G-stem
ktlm
ktalm <- 4-consonant stem with *t infix and both a prefix and a suffix
N.B.: since I’m rewriting my verb conjugator, there might be differences between what I’m describing below, and what forms the conjugator produces now. The vowel positions of prefixes and suffixes, the vowel quality of the stem and surrounding suffixes and the pronunciation of the resulting form will be laid out in the next post. Also, bear in mind that the language doesn’t actually have any 🤗 verbs yet: at the moment I’m still trying to develop the grammar
Overview
The structure of a finite verb form is as follows:
each verb either has or doesn’t have a locative prefix modifying its meaning; its role is similar to postpositions of English phrasal verbs (“make up”, “give in”, “get along” etc.) or Slavic, German, Hungarian, Georgian verbal prefixes
each transitive verb has a direct object prefix
each verb has a subject prefix in all its forms, except the connegative (a form used for negation together with a special negative verb); both object and subject markers express a combination of person (1st, 2nd or 3rd), number (singular or plural) and gender (feminine, masculine and neuter)
each verb belongs to one of 12 derived stems, expressing the verb’s valency (intransitive, transitive, reflexive/reciprocal, passive, causative, …) and lexical aspect (perfective and imperfective); the stem encodes both the derived stem a verb belongs to, as well as its tense-aspect-mood form: there are 8 tense-aspect-mood forms lying at the intersections of 3 tense-moods (past, present and subjunctive) and 3 grammatical aspects (aorist, imperfect and perfect) - with the present aorist being the one missing
there’s only one tense-aspect-mood suffix, marking past and subjunctive tense-mood of the imperfect and perfect grammatical aspects; there is technically also a vowel suffix in the optative (subjunctive aorist) of G-stem, but it can be thought of as part of the vowel pattern
there are only 2 subject suffixes: one for 3rd person singular and one for all persons plural
Only locative and object prefixes have their own fixed vowels, the rest can have the vowels *a and *i, but those can be overwritten by the vowels *u and *i associated with a particular derived stem, grammatical aspect or tense-mood
Derived stems and grammatical aspects are formed from a root, consisting of 3 or more consonants, by adding consonants as prefixes or infixes or doubling existing consonants
Consonants of derived stems and grammatical aspects
As examples we are going to use 2 nonsensical verb roots: 3-consonant klm and 4-consonant bcdf
G-stem is the base stem, it has no markings; aorist is the base aspect, it has no additional consonants
all D-stems and imperfect aspect forms double their penultimate consonants; an imperfect aspect of D-stems doubles a consonant only once, there is no stacking; examples:
klm -> kllm
bcdf -> bcddf
all t- and tn-stem, as well as perfect aspect forms have a *t infix after their first radical; tn-stems also have an *n infix after their last *t infix, but no sooner than before 3rd last consonant (in 3-consonant roots *n is inserted immediately after last *t infix, but in longer roots they may get separated by root radicals); in perfect aspect forms of t- and tn-stem themselves *t is added twice; for example:
Gt-stem or perfect of G-stem:
klm -> ktlm
bcdf -> btcdf
perfect of Gt-stem:
klm -> kttlm
bcdf -> bttcdf
Gtn-stem:
klm -> ktnlm
bcdf -> btncdf
imperfect of Gtn-stem:
klm -> ktnllm
bcdf -> btcnddf
N-stems and S-stems have *n and *s prefixes respectively; *t infix is added after that prefix; for example:
simple S- and N- stems:
klm -> sklm
bcdf -> nbcdf
simple Stn- and Ntn-stems
klm -> stnklm
bcdf -> ntbncdf
perfect of St-stem:
klm -> sttklm
bcdf -> sttbcdf
imperfect of Ntn-stem:
klm -> ntknllm
bcdf -> ntbcnddf
Vowel positions of a stem
Positions of vowels in a stem depend on its length; there may be some variations caused primarily by the presence of subject prefixes and suffixes. We are going to use the vowel *a as a dummy vowel in our examples, and mark the position of subject prefix or suffix with ~
All stems with more then 4 consonants use the structure C¹aC²C³aC⁴ for its last 4 consonants; a beginning consonant have a vowel inserted after it if it’s in an even position or if it’s a 1st consonant in a stem not following a subject prefix (the dot is used to separate the two parts):
kttlm -> ka.tatlam / ~k.tatlam
bttcdf -> bata.tacdaf / ~bta.tacdaf
ktnlm -> ka.tanlam / ~k.tanlam
ktnllm -> kata.nallam / ~kta.nallam
btcnddf -> batac.naddaf / ~btac.naddaf
stnklm -> sata.naklam / ~sta.naklam
stbcddf -> satab.caddaf / ~stab.caddaf
ntbcnddf -> natabca.naddaf / ~ntabca.naddaf
Stems with 4 consonats follow the same structure C¹aC²C³aC⁴ - unless it has an *n prefix or *t infix with a subject prefix, in which case the first two consonants stick together (C¹C²aC³aC⁴), and additionally the last vowel gets deleted before a suffix.
Examples:
kllm -> kallam
bcdf -> bacdaf
nklm -> naklam / ~nkalam / ~nkalm~
ktlm -> katlam / ~ktalam / ~ktalm~
3-consonant stems (only possible for aorist aspect of G-stems) have the structure C¹VC²VC³ or C¹C²VC³ (after a subject prefix) in aorist tense and C¹VC²C³ or C¹VC²C³V (without a suffix) in optative
I made a thing! It’s a verb conjugator for my conlang. Just to be clear, there aren’t any verbs or any other words in the language yet, but when there are some, this is how they will conjugate.
Some additional notes under the cut:
The whole conjugation system is largely based on Akkadian. I wanted to have a root-and-pattern system similar to Semitic languages and Akkadian, firstly, had more derived stems than any other semitic languages, and secondly, used D-stems and t-stems both for derivation (intensity/transitivity and reflexivity/reciprocity, respectively), and conjugation (forming imperfect and t-perfect tenses), changing the stem much more drastically in the process and making the whole system feel even more root-and-pattern-y, than other Semitic languages.
The conjugation system theoretically can work with roots with any number of consonants larger then or equal to 3. I put the upper limit at 6, but be warned that long roots generate forms that get very unwieldy very fast. That might be useful, because it allows easier extraction of a consonantal root from an arbitrary noun, either as is, or with some modification, akin to Arabic forms XI-XV.
The system of tenses is taken from Georgian. Unlike Georgian I don’t want my conlang to have a complicated system of morphosyntactic alignment: it will be a nominative-accusative language. There is a certain rhyming, though, between the fact that Georgian changes alignment between different tense series (“screeves”) and that Akkadian imperfect and t-perfect taken as the basis for imperfect and perfect series, respectively, use valency changing derived stems as described above.
The language has vowel harmony based on emphatic spreading, similar to Moroccan and Egyptian Arabic, but more extensive: in the presence of “etymologically” emphatic/pharyngealized consonants all vowels move down and back in terms of their place of articulation. “Etymologically” long vowels and dyphthongs are realized as diphthongs, which are all borrowed from Faroese, so *iː -> /ɔj, ʊj/, *aː -> /ɛa̯, ɔa̯/, *uː -> /œw, ʏw/ etc.
There is spirantization of plosives similar to Hebrew/Aramaic, but it is applied differently to different stops: *p ang *g turn into /f/ and /ɣ/ almost everywhere, *d and *dʒ become /ð/ and /ʒ/ only after a vowel, everything else doesn’t change. Apparently in Akkadian itself long voiced plosives started to turn into sequences of a nasal + a plosive, that process is applied to the conlang, too: *bː -> /mb/, *dː -> /nd/ etc.
Personal prefixes are neither Akkadian, nor Georgian. Initially I found the fact that Georgian uses g- prefix with pluralizing -t and -en suffixes in its verbal paradigm very amusing, because it makes some of its verb forms superficially similar in structure to German past participles (especially combined with the fact that both languages have verbal prefixes that stand before g-). To replicate that effect early versions of the personal prefixes were essentially Semitic, but with g- insted of t- in th 2nd person. The problem is that conjugation must somehow echo personal pronouns, by choosing one form of prefixes over another I also put restrictions on how pronouns will look in the future, and I simply don’t like the look of Semitic personal pronouns. In the end, personal prefixes are taken from a reconstruction of Proto-Khasian of all places, by moving every [non-labial] consonant’s place of articulation forward, so *ʔ -> /g/, *k -> /t/ (and when I get to pronouns: *ŋ -> /n/). Nice thing about this choice is that now g- prefix is even more common (it is used not only in the 1st, but also the 3rd person), and t- still marks 3sg.f, just like in Semitic languages.
There’re still gaps in the verbal paradigm: I don’t include 1) nominal forms (infinitive and participles), 2) object prefixes, 3) lexical prefixes, similar to those Georgian has, and 4) negative forms. Of those 3 I know how object prefixes should look, and the next update to the conjugator will probably reflect that. I include connegative form, but I still don’t know how the actual negative verb(s) meant to be used with it will look like. The remaining points (1) and (3), I’m also still unsure about.
