Questioning the unquestionable: Koa phonology

A crazy idea hit me out of the blue a couple days ago. Or actually, as it turns out, maybe not so crazy, but it certainly seemed that way at the time. By way of background, let me quickly sketch out the (simplified) consonant phoneme inventories of a bunch of languages with small phonologies...

Pirahã: p t ' m n s h

Hawaiian: p k ' m n l h w

Samoan: p t ' m n ŋ l f s v

Blackfoot: p t k m n s h w y

Plains Cree: p t c k m n s w y

Aita Rotokas: p t k m n ŋ r β ɣ

Maori: p t k m n ŋ r f h w

Kiribati: p pɣ t k mɣ n ŋ r β βɣ

Tongan: p t k ' m n ŋ l f s h v

Arapaho: b t č k ' n θ s x h w y

Ainu: p t č k ' m n r s h w y

Cheyenne: p t k ' m n s š x h v

Comanche: p t k ' m n r s h β w y

Finnish: p t d k ' m n ŋ l r s h v y

Bemba: p t č k m n ŋ l f β s w y

Ojibwe: p t č k m n l r f s š ð y w

Inuktitut: p t k q m n ŋ l r s ł v ɣ y

Cherokee: t d ts tl k g kw m n l s h w y

Bislama: p t č k m n ŋ l r f v s h w y

Cebuano: p b t d k g ' m n ŋ r l s h w y

Tagalog: p b t d k g ' m n ñ ŋ l r s h w y

Japanese: p b t d č j k g m n r s z h (w) y

Hittite: p b t d ts k g kw gw m n r l s x h w y

And now, Koa: p t k ' m n l s h

I was really surprised to realize that Koa is conspicuous in the company of these other languages. Exotic, even: along with Pirahã, it is the only language to lack any kind of highly sonorant voiced labial consonant. (It's also somewhat unusual in lacking /y/, but I'm leaving that aside for the moment because (1) Koa does sort of have it in words like ia, and (2) it's not quite as universal.)

So what to do about this? Obviously /w/ has tried to make it into Koa on numerous occasions in the past, and was finally permanently rejected a couple years ago. The thing is, though, that it was always in the company of /y/, and I always conceptualized it as an invariant [w], both of which doomed it for various reasons.

Picturing a putative incoming phoneme as /v/ is really different. It wouldn't need to have all those distribution restrictions: it could pattern just like any other consonant. It could have acceptable variant pronunciations from [v] to [β] to [w], making it accessible to speakers of nearly every language on Earth.* And because its ideal form is still consonantal rather than vocalic, it preserves the overall "feel" of Koa, rather than making it feel all sloppy and amorphous like it was with /y/ and /w/.

It seems, then, like it wouldn't be at all unrealistic to add /v/ to Koa. I'd finally recoup the loss of forms from the departure of /c/ -- 5 particles and 500 roots. I'm kind of terrified of the idea, as it would be a tremendous change, and with so much being created right now, if I work /v/ into things it might be difficult to extricate it if I change my mind later. On the other hand, looking at all those inventories up there, it seems more like the question should be why on Earth not to add it.

I have two possible reasons. One is that, for speakers that pronounce /v/ as [w], we get into some of the same problems of ambiguity that we were grappling with way back when with the other semivowels: mova and moa both end up as [moa], etc. The thing is that (1) like Amelia points out, languages cope with way more homophones than this with absolutely no trouble, and (2) Allison is totally right that the advantages are likely to massively outweigh the disadvantages.

More seriously, I'm really happy with my usitative particle, ua, and I would have to give that up if I were bringing va on board. The grief might be too great to bear. Of course, I could always be idiosyncratic and just use ua instead of va. Hm.

Anyway, this decision is far from being made at this point, but it's certainly looking increasingly likely.

*There are exceptions. Bengali seems to have nothing in this neighborhood and would probably have to resort to [b]. Some other languages like Basque, Catalan and Japanese are kind of marginal, but have similar enough sounds that they could probably pull it off. In any event, it's no worse than /h/ in terms of cross-linguistic universality.

Clarification of pronominal possession strategies

Reading Describing Morphosyntax on the train this morning, I realized what's been going on in my brain all along with the two possible structures for pronominal possession (talking about ni mama vs ka mama ni for "my mother"): I'm pretty sure it's a question of inalienable vs alienable possession.

I don't think I would necessarily prescribe this usage, but it makes sense typologically (alienable possession typically takes more morphology), and perhaps is even a little intuitive. I might explain it in terms of "pronoun + noun" being more inherent, more basic possession: family, body parts, home, language, etc.

And of course, as always, there's no clear line dividing the alienable and the inalienable. It would never be wrong to use either of them in any situation. But I think it's a useful shading to have, and might make speakers coming from that kind of linguistic background feel more comfortable.

Inchoative and cessative

Okay, that's it. Can we please stop beating around the bush and choose these? I've had them on my to-do list since at least 2002.

Here are the available particles, excluding those with no high vowels on the grounds of this post:

hi ku li mi no pu su tu u

ua ue ui uo ie io iu

Wow -- we're really down to the bottom of the barrel. Note that u is back after 10 years or so; I've decided it's a nice enough particle that it deserves a role more important than inclusive "or," and I've moved that over to au in homage to Esperanto.

Inchoative first. Let's translate the phrase "...and then, he started to speak":

e he toa, ta X puhu

Actually, maybe it won't be that hard. I'm thinking su for cessative (sort of like a special kind of si), and mi for inchoative.

e he toa, ta mi puhu.

Likewise, we can do ni su suo ka lono "I ate up my soup," "I finished eating my soup," etc.

Let's let them stand for the moment and see how we like them.

Question: are these necessarily perfective because of their meaning? I'm thinking probably yes. Can I say ta ma mi puhu [when...] "he was starting to talk when..."? Not sure. There isn't a logical reason why not, but there might be a typological one.

Adieu to the shortest-lived particle

Tonight's walk with Sadie gave me some good Koa processing time, which has resulted in the following decision: I really don't need ki as an indefinite pronoun. I can get exactly the same pragmatic effect by just using the passive in presentation form:

(i) si pa ipo ka sahi ni!

ai (i) pa puhu le Koa ne le Niu Ioliku?

You may notice I'm debating the necessity of the initial i in this construction.

Now that ki is freed up, I can use it for what I discovered on my walk to be its true purpose, at least for today: the suite of obligation, necessity, and desire.

I had trouble visualizing it before, I think, because I was thinking of this having verbal force. If I think of it as adjectival, it becomes much easier: ki suo could be glossed as something like "in need of eating," for example; ki pa suo as "needing to be eaten." In answer to the question "What should we have for dinner?" one could answer: [well,] ka sihi i ki pa suo "the vegetables need eating" (or "want eating" if we're British). In the same way, I can look at my watch and exclaim, ni ma ki lahe! "I've gotta get going!", literally something like "I'm in need of leaving."

Of course, we'll still be able to use full verbs or adjectives to express specific concepts: things like ni ma tau ko lahe or (i) tau ko ni lahe "I need to go." But in many cases the force of these auxiliaries in English is more modal than lexical, and I think having this particle will make a lot of much more elegant translations possible.

By the way, a thought from Whole Foods today: one can translate Latin omnia quae fieri possunt "the gamut of existential possibilities" as Koa po a te tai. I think that's pretty damn awesome.

Sweet Jesus, VS syntax?!

Spurred on by the aside about obligation in my previous post, and already bouncing ideas about "subjectless" clauses around in my head, I made a discovery on the BART this morning that could be profoundly important if I decide to stick with it.

To start with, let's look at some problematic sentences, using only syntactic structures we've had up to this point.

1. (ko) ne tia i *kuma "it's hot in here"

2. ko ni lahe i tau "I have to get going"

3. ko ni lu mene i te tai "I might go"

4. a ame i (tai) ne ka talo "there's a bird in the house"

5. ka *uli i ma *uli "it's raining" (cf. Turkish yağmur yağıyor)

There's a problem with all of these, which is that (maybe with the exception of 5) they're all stupid. In theory they could be ways of expressing these concepts, but as the basic unmarked way, they fail. "That I leave is necessary?" I really don't think any natural language is going to structure this kind of statement this way, because it completely ignores pragmatics in striving to retain a syntax identical to that found in unproblematic transitive clauses.

I started to recognize this in an earlier post:

In the main, my concerns [...] pertain to the existential construct in general -- I keep wanting to say something like na tai neko me ni, which is clearly in violation of everything everywhere. I think what's happening is the collision of the logical design of the language with human language intuition; hopefully they won't end up being too difficult to reconcile.

Indeed: so how to resolve this?

One option would be to use the topicalizer to rearrange the sentences into something that feels less silly:

1. *kuma sa ne tia

2. tau sa ko ni lahe

3. te tai sa ko ni lu mene

4. tai sa a ame ne ka talo / a ame sa tai ne ka talo

5. ma *uli sa ka *uli

Some of these feel significantly better at first glance, but only at a rapid first glance, because this doesn't solve anything: these sentences are even more pragmatically anomalous than they were before! Because it's the topicalizer that's allowing me to move the verb up to the front this way, I'm ending up saying things like

1. as for being hot, now, that's what it is in here

2. as for necessary things, now, that's what my leaving is

3. as for possibly being the case, that's the deal with my going

4. existing, now, that's what a bird is doing in the house

5. currently raining, now, that's what the rain is doing

I mean, there may be situations where you'd want to slant things this way. But not for the unmarked sentence frame! Furthermore, this completely fails to address what on Earth we'd do if there were no additional arguments/adjuncts to the clause: "it's hot!", for example. Just kuma? You can imagine a language saying it this way, and I'd like it to be an option for Koa in a very informal sense, but once again not as the basic means of communicating this.

Tok Pisin was, of course, my original inspiration for Koa's basic design, and the way Bislama manages this sort of thing eventually wandered into my mind. They just do away with the subject NP altogether: i gat wan pijin long haos "there's a bird in the house," etc. Is there any reason we can't do this in Koa as well?

1. i *kuma ne tia

2. i tau ko ni lahe†

3. i te tai ko ni lu mene†

4. i me ame ne ka talo OR i tai hu ame ne ka talo

5. i ma *uli

And wow, does that ever work better than anything we've seen before. I can't see any particular problems it would cause at this point, and it gives us a great way of saying i kuma!

2 and 3, though, raise an interesting question: should there be an intonation break after the VP, since the following NP is actually the subject? I.e., i tau, ko ni lahe? I don't want to say yes, because that's not what I'm trying to say: "it's necessary, you know: me going." Once again, it's easy to think of a situation where that would be appropriate, but not as the basic structure.

This is where it hit me: all of the situations where this comes up are intransitive verbs...and I seem to remember from both typology class and Describing Morphosyntax that there's a tendency for SV languages to switch to VS in intransitive clauses (cf. Polish, szła dzieweczka...). What if this is just another allowable syntactic structure in Koa for intransitive verbs?

I think this is sort of like what's called "presentation form" in English: you can say "there is X," "there sits X," etc. In English it starts to feel marginal the further you move from statives: "there sleeps X" "there lives X" are okay but a little weird, but "there eats X" or "there died X" are definitely unacceptable except with humor.

In Koa, though, my philosophy has always been that what you can do to one predicate, you can do to any other. I don't feel comfortable drawing a prescriptive line between which verbs can do this and which can't; theoretically, then, this opens up clauses like i musa po oto "crows are black."

Maybe this could just be limited automatically by speakers' pragmatic instincts. My analysis skills are not sophisticated enough at this point (or maybe I just don't want to think about it that hard right now) to say exactly why, but I feel like po oto i musa is much more appropriate for this statement. And I really hope it's not just because that's how it's done in English.

How much Polynesian is too much Polynesian?

There's no denying that Koa has a particularly Polynesian-flavored phonology. In fact, I'm pretty sure that this was one of my original influences when I was first working on the language. As we know, though, the same theoretical phonology can produce very different results depending on frequencies, distributional constraints, etc.

Comparing an average body of Koa text to that of any Eastern Polynesian language, for instance, it's immediately obvious that Koa has fewer vowels, and fewer VV sequences. This has a huge effect on the way the language feels in the mouth and, more importantly, how easy it is for someone from a non-Oceanic part of the world to pronounce it.

In fact, I've striven to avoid an excessive pileup of vowels wherever I've been able. As things have been, the maximum number of sequential vowels has probably been limited structurally to 4, in phrases like nae a olu "see a flounder," and this is not particularly frequent. By far CV reigns, with just enough VV action to make things nice and fluid.

The reason I bring this up now is that, as I discussed earlier, I'm coming near the end of my particles, and finally needing to think seriously about where the VV contenders are going to fit in. For instance, I need a "must" particle; it occurred to me that it might be rather nice to use oe, which I glossed as "should" in the first edition of Ea Opi le Koa. The problem with this is that it immediately produces tokens that look way more Polynesian than anything prior. Let's check out some examples.

ni ma oe mene la talo "I've gotta go home now"

ai ni oe na te pa nae? "do I have to be invisible?"

se na oe ipo hu sahi "you don't have to drink wine"

ta oe luke tika "he's gotta read this"

(Honestly, quite apart from phonological considerations, I'm really wondering whether this is the way I want to do this. I conceived it in order to render the E-o -enda kind of semantic -- sahi oe pa ipo "wine that must be drunk -- but I'm not so sure anymore. Anyway, that's not really the point of this post.)

Allison says, quite rightly, "I think there's only so long you can side-step this kind of problem with the phonology you're working with. Sometimes those vowels are just going to want to have a party together in the middle of your sentence and there won't be a damn thing you can do about it." Nevertheless, there are definitely steps I can take to reduce the partying, and keeping forms like oe out of mid-phrase position is one of them.

So, then, a resolution: we'll just have to deal with the nae a olu kind of vowel clusters, and should try not to worry about them overmuch; but the VV particles need to be used for clause-scope functions where they'll find themselves in the company of other vowels as rarely as possible.

Indefinite pronouns and obscuring the agent

I've tentatively created a new particle, ki, equivalent to "oni" in Esperanto or "me" in Yiddish. The idea is that it stands in for a participant without revealing its identity, either because this is unknown or irrelevant, or because its referent is general.

For example, a sign in an occult shop window might say Ki luke po lisu ne tia "Omens read here." The English translation raises the question, though, of how this putative form differs from a straight passive clause like Po lisu i pa luke ne tia.

I feel like there's a definite pragmatic distinction. Let's look at a different situation: suppose it's our Christmas party and everyone is drinking mulled wine. After playing the guitar for a bit, I go back to my glass and discover with surprise that it's empty, as it had been nearly full when I left it. When Amelia asks what's wrong, I can answer in a number of ways:

1. Ka sahi ni i si pa ipo!

2. Ki si ipo ka sahi ni!

3. Huka i si ipo ka sahi ni!

4. Keka sa si ipo ka sahi ni?

1. Straight passive. The arguments are right, but the focus is wrong: here we're focused on the wine, and the current state in which it finds itself (i.e., drunk).

2. Indefinite pronoun:. I feel like this gets at what I'm trying to say better. I'm focusing on the drinking, and the agency/volition whereby it was accomplished, while allowing the actual agent to remain unnamed (in this case, because I genuinely don't know who it is) and unemphasized.

3. Existential complex. Both this and the previous sentence could be glossed by "someone drank my wine!" in English, but there's a very important semantic/pragmatic difference. Huka can mean "someone" -- literally "some of those currently on the discourse stage" -- so it could mean "some person," but is more likely to have the force of "one of them." As such, it's far too accusatory for the intended meaning, and focuses too much on the agent herself rather than on the action.

4. Identity question. I put this in just for fun as an example of how we could get at a very similar meaning in a totally different way. This suffers from the same problems as 3 above, but with the right intonation and humorous atmosphere could still be an appropriate response.

Based on the above, it seems to me that ki is justified, but we'll leave the matter open for now. I'm making a lot of assumptions in my analysis that are based on English focus, and I'd better be sure that this is what's happening in Koa as well: otherwise, if the straight passive is more neutral than in English, for example, there would really be no need for ki.

"Do they speak Koa in New York?"

Ai ki puhu le Koa ne le Niu Ioliku?

Ai le Koa i pa puhu ne le Niu Ioliku?

And so on.

Specifier flowchart

1. Is the predicate referential? That is to say, does it refer to something real and specific in the universe?

Yes → 2

No → 5

2. Is the referent already on the discourse stage, either because it has already been raised or because it lives there by default? Or, is the predicate specific and identifiable in the given context?

Yes → 3

No → 4

3. Does the referent need to be pointed out to specify it? Answer "no" if the referent is identifiable without pointing.

Yes → 8

No → 7

4. Is the amount/quantity of the referent indefinite, unknown, or irrelevant to the discourse? Or, is the referent a mass noun that is not specifically bounded in some way? Or, is the referent being mentioned without the intention of raising it to the discourse stage?

Yes → 10

No → 9

5. Is the referent an abstraction of the quality described by the predicate, whether theoretical or actual?

Yes → 6

No → 11

6. Is the referent a specific instantiation of the abstraction in question?

Yes → 13

No → 12

7. Use ka.

8. Use ti/to depending on deictic distance.

9. Use a.

10. Use hu.

11. Use po.

12. Use ko.

13. Use ko preceded by a specifier. Go back to question 1 and choose "Yes" to determine the correct specifier.

The Future

27) I've decided that lu, instead of being a desiderative particle, should instead mark the future tense. The range of usefulness is just so much greater, and otherwise there are some future semantics I'm not sure how to get at. This means that Koa is now one of these languages that draws a future/nonfuture distinction, which is kind of awesome.