Archive for October, 2021


Friday, October 29th, 2021

A language with a small number of verbs that combine with nouns to make actions, with the verbs in question being HTTP request methods.

Detail #421: A Quirky Numeral Structure

Friday, October 29th, 2021

Consider a language with singulars, duals and plurals. The language has a rather strict distinction between mass and count nouns, and explicitly marks different types of individuated, specific, indefinite, etc references.

Now, this entirely eliminates the need for the numerals one and two, as you would never say 'two bikes', you'd say bike-dual. You would never say 'there are two of them', you'd say 'they-plur are they-dual'.

This of course leads to problems when counting higher numbers. You have nothing to put after 'twenty' or 'thirty' when you want to form 21 and 22.

Twenty bike-SINGULAR = 21 bikes.
Twenty bike-DUAL = 22 bikes.

Probably unlikely.

Detail #420: Ambiguous Reference with Possessive Pronouns

Tuesday, October 26th, 2021

English has some ambiguity with its possessive pronouns, but the level of ambiguity could be taken to a weirder level in this way:

reflexive third person ownership or other third person referent: singular possessive pronoun

Thus "he sees his house" can have 'his' either be reflexive or not, but "they see his house" also can signify reflexive possession.

reciprocal ownership uses a third person plural possessive pronoun

Thus "they see their house" can mean 'they see some other persons' houses' or 'they see each other's houses'.

I don't think this kind of idea is entirely unrealistic.

Individual incidences

Tuesday, October 26th, 2021

Warning: this is a long one.

In Maltese, there's a systematic distinction between the abstract action of a verb and a single discrete occasion of it. From Teach Yourself Maltese by J. Aquilina (Hodder & Stoughton 1965), 149:

There are two kinds of verbal nouns. One which (i) denotes the action or state indicated by the meaning of the verb, (ii) is of masculine gender singular in number, and, (iii) like any other noun, can be preceded by the definite article, but has no plural (Exx. dfin 'burial' from difen/jidfen 'he buried/buries', id-dfin 'the burial'), and another which (i) expresses a single occurrence of action or state indicated by the verb (ii) is of feminine gender singular in number; (iii) forms its plural by suffix iet and (iv) can be preceded by the definite article (Exx. difna 'a burial'; id-difna 'the burial'; difniet or id-difniet 'burials, the burials').

More examples of these kinds of pairs:

daħk "laughter" vs daħka "a laugh"
taħwid "confusion" vs taħwida "a mess"
ġbir "gathering" vs ġabra "a collection"
żfin "dancing" vs żifna "a dance"
bligħ "swallowing" vs belgħa "a gulp"
xorb "drinking" vs xarba "a drink"
ferħ "joy" vs ferħa "a joy"

I think if it hadn't been for studying Maltese in around 2005 I might not have thought that hard about this, since the languages I know best make the distinction haphazardly (if at all): Polish śmiech "laughter, laughing, a laugh"; radość "joy, a joy"; pochowanie "burial, burying," pochówek "a burial"; picie "drinking," łyk "a drink."

In Koa, the abstract action or state is very straightforward, since there's a specific article (ko) set aside just for this: iolo "happy," ko iolo "happiness"; ipo "(to) drink," ko ipo "drinking. The question is how to get at the single occasion/occurrence/incidence/action/etc.

Well, up to this point, we've been doing something that's possibly a little weird: we've been using a double article. I don't know that I ever thought this through thoroughly in the past, but I believe what I was trying to get at was that, despite its appearance as an article, the use of ko is really more of a derivational operation. This would certainly be true in English: good -> goodness, drink -> drinking, etc. If we look at it this way, the change in meaning is something like this:

koa "a good one, good, be good"
ko koa "goodness, of goodness, be goodness"

In other words, ko changes the meaning from that of the root itself, into that of the state, condition, or action of doing or being the root. What I'm realizing, though, is that that doesn't say anything about specification or referentiality. I had been thinking that in a phrase like ko koa the ko would count as a specifier because "goodness" is a kind of universal and as such is always on the discourse stage (or, let's say, on the shelf immediately adjacent to it). But in all other cases, that kind of concept would require the particle po! If ko is derivational, we'd expect something more like this:

a kokoa "an example/occasion of goodness not yet present on the discourse stage"
ka kokoa "the example/occasion of goodness we're already aware of in this discourse"
hu kokoa "there is an example/occasion of goodness out there such that..."
po kokoa "goodness in general"

What we've been doing so far is allowing simply ko koa, rather than po kokoa, for the final example. I guess the rule we'd have to make explicit -- if that's really how it is -- would be something like "ko behaves as a specifier when used alone, and as derivational morphology when preceded by another specifier." After almost 20 years, I'm not at all sure that this makes any sense at all!

In fact, looking over our particles, ko is the only one that could be said to be derivational in nature. There are pronominals (ni se ta nu so tu), locators/adverbials/adjuncts (he la lo me mo ne no o pe), specifiers (a u hu ka ke le pi po ti to), tense/aspect markers (cu, io, ma, mi, si, su, va, vu), modals, evidentials/viridicals/miratives (ho ki ku li lu pu te vi ia), valence operators (hi mu pa), clause-level operators/conjunctions (e ha na ve), syntactic markers (i, sa, vo), and qualifiers (ce, ie, iu). I think that ko, incredibly enough, may have been mis-assigned!

What this ought to be is a suffix, almost all of which are derivational at this point (in fact I think all except for the pronouns indicating possession). Some options: -ko (just moving it to the back instead of the front), -te, -ti, -mi, -i, -pe, -vi. To be fully transparent, the one I was going to suggest at the start of all this was -te, but I'm curious to see how the others feel as well.

With -ko:
húlako "dance"
púhuko "talk"
súoko "meal"
súsoko "kiss"
élako "life"
cíniko "kindness"
lóeko "coldness"

This is pretty okay, though the diminutive suffix gets repetitive with /k/: húlakoki "a little dance," púhukoki "a little talk." Assuming ko is going to maintain the other half of its dual role as a complementizer, this also potentially risks confusion...although Polish seems to have no problem with że as complementizer and -że as an emphatic suffix (the latter not particularly common, to be fair). Ta sano ko ta no móeko "she said she had no dreams" -> Ka sánoko ko ta no móeko... "the statement that she had no dreams..." A little awkward.

With -te, -ti:
húlate, húlati "dance"
púhute, púhuti "talk"
súote, súoti "meal"
súsote, súsoti "kiss"
élate, élati "life"
cínite, cíniti "kindness"
lóete, lóeti "coldness"

These seem pretty solid, and here the diminutive works better: húlateki "a little dance," púhuteki "a little talk." Ka sánote ko ta no móete also has improved flow...though I'm sort of surprised to say I like the aesthetics of forms with -te less than those with -ko! I like how neutral these feel, though. I think I prefer -te to -ti? Or do I?

With -mi, me:
húlami, húlame "dance"
púhumi, púhume "talk"
súomi, súome "meal"
súsomi, súsome "kiss"
élami, élame "life"
cínimi, cínime "kindness"
lóemi, lóeme "coldness"

I think I could be okay with -me, though it's awfully close to -ma in a way that could actually be confusing: élame "life" vs élama "a being," for example. So probably no on this. Incidentally, I'm finding it surprisingly difficult to be okay with with the word for "kiss." Really the only one I've liked so far is súsoko, with súsoti in second place. Ana súsokoki la ni "give me a little kiss" -- ugh, -koki just is not very good. And I just realized that "height" would be kókeko, yikes. Ana súsoteki la ni? Ana súsotiki la ni? I think -te might feel the most Koa.

With -i:
húlai "dance"
púhui "talk"
súoi "meal"
súsoi "kiss"
élai "life"
cínii "kindness"
lóei "coldness"

Oof, I think this is just weird. Possible confusion with the VP marker i as well.

With -pe, vi:
húlape, húlavi "dance"
púhupe, púhuvi "talk"
súope, súovi "meal"
súsope, súsovi "kiss"
élape, élavi "life"
cínipe, cínivi "kindness"
lóepe, lóevi "coldness"

Hm. -pe doesn't feel neutral enough. I like the -vi forms but I'm not sure they should actually mean what these will mean. Ana súsoviki la ni...Ka sánove ko ta no móevi...No, I don't think so.

Okay, Olga likes -te, so I think that's what we'll call it for now. That said, I want to revisit our previous conclusion that ko's assignment had been a complete mistake. I had said that po kóate would mean the same thing as ko koa, and I'm not actually sure that's true. Let's go back over the determiners.

a X = a referential instantiation of X not yet raised to the discourse stage
ka X = a referential instantiation of X already on the discourse stage
hu X = a non-referential instantiation of X
po X = the non-referential set of all instantiations of X
ko X = X itself, uninstantiated

It looks like we need to be clear on the difference between an instantiation and an occasion/occurrence/incidence:

Instantiations: a kind person, a dog, a runner
Incidence: an act of kindness, a moment of caninity, a run

Clearly the semantics of an incidence of a strongly nominal root are a little...weird. You can imagine fantasy stories, maybe, featuring interludes of being a dog? But it's very clear and useful in roots that are more stative or verbal. Anyway, here's what the determiners would look like with the -te suffix:

a Xte = a referential incidence of X not yet raised to the discourse stage
ka Xte = a referential incidence of X already on the discourse stage
hu Xte = a non-referential incidence of X
po Xte = the non-referential set of all incidences of X
ko Xte = the concept of incidence X itself, uninstantiated

Clearly ko X and po Xte are not the same thing, even if the semantic load between them is slight to nonexistent. For example in actual usage,

po súsote i mu iolo ni "kisses make me happy"
ko suso i mu iolo ni "kissing makes me happy"

These are similar but not identical. There are different sentences that could better clarify the distinction; the first seems to be saying that the act of kissing, or fact of kissing, is what makes them happy, whereas the second refers to the kisses themselves. Here's another shot:

po súsote se i mu iolo ni "your kisses make me happy"
ko suso se i mu iolo ni "kissing you makes me happy"

Here we also see the fact that ko can introduce a clause with a clear pronominal object. I think, then, that ko in its original meaning isn't going away: we're just adding -te for a separate meaning that previously wasn't being ideally encoded. Whew! That was a lot.

On another topic, here's an interesting bit of emergent subtlety with our new suffix:

súsote "a kiss, a single instance of kissing"
pasúsote "a kiss, a single instance of being kissed"

So a súsote potentially means "a kiss X gave (someone)," and a pasúsote is "a kiss someone gave X" -- a shift of perspective from the giver to the receiver. We don't have a "massage" root yet, but if we did, would the default way of talking about one you received be the pa- form? I don't know if this should be a prescriptive distinction, but certainly a nicety of expression available to a skillful speaker.

Summing all of this up, before I finally post this behemoth and catch my breath: for those who don't speak Maltese, there's an important systematic difference to learn to make between e.g. ko cini "kindness, being kind" on the one hand, and cínite "a kindness, act of kindness" on the other. And I think élateni for "vida mía," while not quite as evocative as koélani, is still quite nice.


Sunday, October 24th, 2021

allocutive agreement that marks whether the listener is a man or a woman: binarist, overdone, old information

allocutive agreement that marks whether the speaker has a crush on the listener: inclusive, novel, transformative

Detail #419: Direct and Indirect Object Case Marking: a Different Approach

Thursday, October 21st, 2021

Let us imagine a language that is similar to standard average European. Let's further imagine that word order informs us which NP is the direct object and which is the indirect one. Let's also imagine that there are two cases, emphatic oblique (-n in the examples below) and oblique (-m in the examples below).

I give you-m a-n book-n

"It is a book I give you / a book I give you"

I give you-n a-m book-m

"It is you I give a book" / "you I give a book"

When word order operations occur, however, some restrictions appear: in the initial position, recipients are -m, accusatives -n - but this does not prevent the other NP from taking the same marker after the verb.

Detail #418: Subtle Clusivity

Tuesday, October 19th, 2021

One could imagine a language where certain constructions signal inclusivity, while others signal exclusivity, without there being any dedicated morphemes for clusivity.

1. Reflexives in two ways

In some languages, there is a reflexive pronoun that can be used for any person (see the Russian себя), whereas in others, reflexives are person-specific (myself, yourself, etc).

In Russian, in some circumstances you can use the person-specific possessive or accusative, but this is unusual. However, we could imagine a language in which first person plural uses the third-person reflexive whenever the listener is not included. This of course limits the clusivity to reflexive constructions, unless the clusivity-signaling reflexive is intentionally overused, maybe as a dative or something else like that, or just as a dummy object with intransitives.

2. Gender (dis)congruence

In a language where plural first person pronouns encode gender, in a system where, e.g. the masculine pronoun can refer to a mixed group (but feminine pronouns cannot), feminine first person plurals when speaking to males can signal clusivity. This is a pretty restrictive situation in which clusivity emerges, but maybe it could be taken one step further, such that gender congruence with a singular listener (or uniform group of listeners) becomes a way of signalling clusivity, rather than signalling the gender of the group.

3. Differential object or subject case on the first person pronoun

For some reason, I imagine a vocative case could actually double as an inclusive subject or object marker.

4. The selection of auxiliaries, especially ones that denote evidential information?

One could imagine a couple of near-synonymous auxiliaries, where one is just for whatever reason associated with the inclusive or the exclusive second person plural.

5. Differential object case on a noun phrase object

Perchance deriving from a historical "our", where the language normally would prefer reflexives possessive pronouns. However, this might disable the marking for clusivity if the subject is not also the first person pronoun, and it disables mixed clusivity in a clause (e.g. "we-excl sold our-incl harvest in town").

6. Word order

"Our house" = inclusive, "house of ours" = exclusive. "They us saw" = inclusive, "they saw us" = exclusive. This could very well be a statistical rule rather than a strict one, such that if the context leads to parsing it differently, such different parsing is permissible - but 90% of the time, this will hold.

For subjects, I imagine this might be less common, although I can also imagine that a SVO language could have VSO as an exclusive structure, since putting the verb first feels like a more "pressing" narrative, where the listener might be unaware of what happened.

7. Selecting between different semantically similar structures

E.g. something like the English perfect and the English past tense. I imagine a language could start associating such a pair with a distinction such as this, due to the situations in which one is likely to use one or the other: 'have done' seems slightly more likely to be used when telling someone who did not participate, "did" slightly more likely when talking to someone who did participate.

8. Dual or trial

One can also imagine that the dual / (trial /) plural distinction might, for second person plural under some circumstances become an inclusive/exclusive distinction instead. However, I want to keep the ability to use the dual/plural distinction itself, so - how about discongruence conveying clusivity. Dual + singular verb = exclusive, plural + dual verb = exclusive? This of course requires an unusually rich verbal morphology with regards to number, and we're also restricting it to elements that have congruence on the verb. Maybe the clusivity distinction becomes so important that in all other positions, the distinction is clearly one of clusivity, or maybe both distinctions are important enough that they're simply thoroughly ambiguous and only context serves to disambiguate between "we two" versus "we, but not you" versus "we several" versus "we and you".

Big business in progress, pandemic edition

Tuesday, October 19th, 2021

I'm sort of surprised and pleased to say there's been an awful lot going on recently, after quite a few years of...not an awful lot. There hasn't been time for exhaustive articles about everything, but I wanted to jot down the main points at the very least. (Incidentally, this post is a bit of a reference to this one from just about 11 years ago.)

1) I appear to be bringing back the phoneme /c/ after 13 years! I say "appear" because I've been mulling it over for about the past year, and it seems that I've decided on something because I've changed polo and sumo back to colo and cumo in my lexicon. My rationale for losing it back in that original post was pretty sound, but I've been drawn back to it because:

(A) I kind of need it for some additional particles that I could but would prefer not to live without
(B) There are enough possible realizations of this phoneme -- [S] and [tS] as standards, but also potentially [c], [C], even [ts] -- that I can still claim adherence to its founding IAL charter
(C) After this post I feel entitled to exercise a little creative license, and /c/ has always felt like Koa -- that is to say, Koa has felt very slightly empty without it -- and it makes me happy.

2) This one feels a little risqué, but I'm going to give it a shot: u having been freed up from its erstwhile role heading an adjectival clause, or marking a dependent clause for a minute, I've decided to try it out as a plural definite article. This means we can have an unambiguous e.g. ka sona "the duck" versus u sona "the ducks." This is the very first example of plurality we've had in Koa outside of pronouns, and in the end it may get vetoed, but I think it has the potential to be useful! It also gives us some neat compound forms: tiu "these ones," tou "those ones," keu "which ones?" pou "all of them," and so on. (Note! In case you missed it, shout out to Lapine: hoi, hoi, u embleer hrair...)

3) After several teacup storms in recent years, I've decided I do in fact want dedicated pronominal forms that can be used as predicates, as opposed to niími or similar. Using our newly minted plural article, this gives us:

nika "I"
seka "you"
taka "he/she/it"
nuu "we"
sou "all of you"
tuu "they"

Their exact use still needs to be determined, but at the very least we know we can now say things like tika i nika "this one's me."

4) So important that it gets its own number: now that we've got taka for the 3rd singular predicative pronoun, it's finally, at last, almost unbelievably, possible to definitively claim tata for "dad." Silly though it may be, this one word choice has caused me as much loss of sleep as an entire category like irrealis marking, and left me with serious rancor towards papa after feeling like it was foisted upon me nonconsensually. So now I can finally heal and move on.

5) There's been quite a bit of uncertainty about the meaning of the forms aha, aka as against hua, huka. Theoretically the corresponding members of each set both mean "something" and "someone," so what's the difference? Which should be used? It turns out that the answer is in the prefix: a denotes something real in the world not yet on the discourse stage, whereas hu refers only to theoretical existence. As such, aha means "something," yes, but specifically "something in particular." It would need to be used of a referent not yet raised to the stage, but definitely existing somewhere. Hua, on the other hand, would mean "something" in its more usual generic sense of "something unknown or unspecified." So ni halu ko sano aha "I want to say something [and I know what that is]," vs ni halu hua ala na ilo ka mea "I want something but I don't know what."

6) Similarly, we have both naha, naka and nahua, nahuka for "nothing" and "no one." This is less clear to me. The latter set is more correct in terms of formal logic, in the sense that ni na me hua = ni me nahua "I don't have anything, I have nothing." But we've never really gotten into it with negation across a clause, and natural languages are frequently anything but logical in this regard, e.g. Polish nic nigdy nie powidziaÅ‚am nikomu lit. "I never didn't tell nothing to no one," in fact meaning "I never told anyone anything." I can't really see a problem with ni me naha for "I have nothing"; I don't think I'd even really go to bat enthusiastically against ni na me naha for "I don't have anything." So this one remains unsolved, but I just wanted to mention it out loud as something that will require real attention one of these days. (I've also played around with naa instead of naha. I kind of like it despite its extreme similarity to na "not." Just a thought.)

7) I've been omitting stress marking when penultimate and unambiguous. So for example natepanae "invisible" (penultimate), aika "time" (considering ai a diphthong here so still penultimate), but naíka "unacceptable" (clarifying the separation between the vowels, because potentially this could be nai-ka rather than na-ika). Stress is still written any time it's not some variation of penultimate: haná "unless," mílani "dear one."

8) It occurred to me that since alienable possessives are really bound morphemes that take the place of suffixes, they should maybe be written that way as well to avoid confusion: átoni rather than ato ni "my father." It ultimately doesn't matter enormously because the pronunciation is identical either way, so in the end it's probably an aesthetic decision. I certainly greatly prefer palóhani for "my love."

9) Verbal complements of halu "want" clearly need to be introduced by ko and I'm sort of embarrassed that I allowed them to appear without it for so long: ni halu ko lahe rather than ni halu lahe "I want to leave." It's very unclear what the latter could even mean. HOWEVER, with the return of /c/, we can make lu volitive again and assign the irrealis to cu from Lithuanian -Å¡u! So: ni lu lahe "I want to leave," ni cu lahe "I will/would leave." This is so wonderfully practical, and delightfully clearly related to halu, I'm really pretty chuffed about it. It's also going to be useful in word formation, I think: like maybe luláevama "aspiring player."

10) "Whatever." For one dimension of this, I realized it's clearly ka vi tai, or more compactly kavitái. I think there might need to be another form for other uses, but we'll get there later.

I'm 100% positive that there are other things that I'm forgetting, but this is a pretty solid start. More coming soon!

Making sense of interrogatives

Monday, October 18th, 2021

So here's a surprisingly counterintuitive translation question: What's your name?

I had previously told Marisa Kea sa se noa? for this bit of extremely simple first-day-of-language-class material, but that's not right at all: that would mean "what's your name like?" Similarly, kea sa ta? looks like "what is he?" but really means "what's he like?" You could rephrase it as "what's his set membership?" or "what indefinite predicate can describe him?" as opposed to keka sa ta? which means "what is his identity?" or "what definite predicate is he a match with?" We're talking about identity with this name question -- matching sets 1 to 1 -- so oddly enough I think it should be Keka sa se noa?

Wait, here's a crazy thought: should a question intended to elicit a name actually be headed by kele??? That's never existed before but maybe it should: Kele sa se noa? So not "indicate to me which one out of a given set is your name," like maybe names on a list, but basically "tell me the name of your name." If that's possible, this could even just be Kele sa se?

I think that's actually exactly right, and finally begins to address some of my earliest questions from the end of this post back in 2002. What about the other specifiers? Does keko mean anything? What about kehu and kepo? I'm thinking maybe the last two don't since they're technically quantifiers rather than determiners, but I'm not sure about keko. "What...noninstantiated quality?" When would that be used? I think we'll have to come back to that one later.

Thinking more about kele, I think maybe it could elicit more than just le phrases. For example, kele sa to puu? "what's the name of that tree? what kind of tree is that?" It certainly wouldn't be kea, because kea sa to puu? would be "what is that tree like?" Keka sa to puu would be "which tree is that?" I suppose that could be answered with "pine" or "spruce"...well, maybe there's some gray area:

Kea sa to puu? = What's that tree like? What is that tree?
Keka sa to puu? = What tree is that? Which tree is that?
Kele sa to puu? = What is that tree called?

The second one could give an identity answer of "the first one on the list," "the one we were talking about," or a general answer of "spruce." The third one could give a specific answer of "Arthur," or a general category answer of "spruce." Even the first one could have an answer starting "Well, it's called a spruce..." I think the important thing isn't picking the precise right one for the application, it's the ability to use the difference between them to potentially resolve the nature of the inquiry in a more granular way.

Next up: asking about meaning, the thing that had me really confused back in 2002. It's one thing to ask what a specific tree is, but it's quite another to ask what a tree is in general: in other words, what "tree" means. As far as I'm aware, there's no prescribed way to do this in Koa at all at this point. Kea sa le puu? And/or do we need a verb of meaning, Kea sa le puu i X? Let's think about it.

Koa Day

Monday, October 18th, 2021

I hereby formally announce that the birthdate of Koa has, at last, been discovered! Actually, it wasn't all that hard to find once I thought to look for it...the files containing the initial phonological thoughts from my dorm room in 1999 still had metadata with creation dates, after passing through five computers and a variety of operating systems.

And so, henceforth, the august date of September 13th shall be known as Koa Day! I'm really quite pleased; it's not like it particularly matters, I guess, but it's nice for the creative project that you might call my life's work to have a special day of its own to celebrate.

Also: Koa is 22! That's a pretty decent age for a conlang in (more or less) continuous development.

Also also: for strictest accuracy, I should give honorable mention to 1997 in which I first came up with a general phonology and concept which was definitely the precursor to Koa (and in fact koa "good" and pua "bad" date to that era). I'm not considering the birthday to belong to those days, though, because I didn't have any principles figured out other than a vague idea of "universality," and no development followed. From 9/13/1999 onwards, though, this language has been recognizably Koa, with the same bones and guiding philosophy.