Conditions of specifier omission

There are certain circumstances, fairly well defined at this point, in which predicates* may or must appear without specifiers. These include the following:

• when used as a modifier of a preceding predicate

talo iso "big house"

• vocatives & imperatives

tule, seneki! "come, kitty!" (is this too ambiguous? e.g. polo! could be either "hey, runner!" or "run!")

• with pronouns as an alternative possession structure

ni mama "my mother"

• with the quantifiers po and hu in a pragmatic rather than logical context

po mehe i pua "men are bad"; ana hu sai la ni! "give me some tea!"

• with the negative operator na in the same way

na sene i ne tia "there are no cats in there"

• with the adverbializer mo when the predicate is non-referential

ta lalu mo sene "she sings like a cat"

It's this last usage in particular that I'd like to explore today, as there are other particles that exert what might be called an "adverbializing" force, in which context it's often the case that none of the specifiers appears to fit the bill completely.

This came up for me recently around locator predicates. Suppose *sala means "inside." Predicates like this let us refine our expression of position: ne ka talo can mean "in, on, at, around the house," but something like ne ka sala ka talo ni can mean only "inside my house."

The question is, what particle should precede sala in the above example? It's true that my house does, indeed, have a real, specific, individual inside space that you can point at. This would seem to necessitate ka as used above. The thing is, though, that it's not really being used that way. We're not saying ka sala ka talo se i *kali "the inside of your house is pretty," or something like that, where the existence and character of that "inside space" are being commented on.

Clearly, none of the other specifiers are more appropriate (a, ko, po, hu, etc.), but my question is whether we can do away with the particle altogether in this kind of construction and just say ne sala ka talo ni. In this way, ne sala becomes something of a derived particle of its own, as well as a set adverbial phrase: Ne kea sa ta? Ne sala. This would be similar to Spanish debajo de, or Polish wewnątrz, etc.

I don't see any particular reason not to do this, though it does raise the question of where to draw the line. Can I say, for instance, Ni si ana ka sahi la mama le Mia? for "I gave Mia's mother the wine?" This is approaching the Welsh rule that the possessed noun in a possessive construction is definite by default, and therefore no article is needed.

The thing is that ne sala as a fossilized adverbial phrase is much more relevant, much less marked than la mama. I think for this reason, it feels okay to me to do this in the former case and not in the latter, but I'm bothered by the fact that there doesn't seem to be an absolute criterion to determine where the division should fall.

If frequency/utility/markedness is the issue at hand, though, why do ne talo and la talo sound so awful to me? They seem like ideal candidates for this kind of construction: Ni halu ko mene la talo "I want to go home." Ne kea sa le Iuli? Ne talo le Lopeto. "Where's Julie? At Robert's." Actually, I have no problem at all with these when used in a possessive construction, but on their own...Nea kea sa le Iuli? Ne talo. Hm. I don't know.

It may end up hinging on the semantics of the specifierless noun: ne talo and ne sala are cultural/spatial concepts that have their own general, abstract existence. La mama isn't really in the same boat, at least here in the United States; plus there's something that feels very unique and definite about a mother that makes this difficult for me as well.

Well then. I was hoping to find a definitive answer to this question here, but I don't think it's going to happen just yet. Let's percolate on it.

*One of two ways of referring to "content words" in Koa as of today, the other being "lexemes." I'm not sure which is preferable at this point. These stand in contrast, of course, to the other main Koa word class, particles. A subset of predicates known as "names" constitute something of a third class in that their behavior differs slightly from ordinary predicates.

Documentation!

Looking over the previous entry, I'm drawn to the inevitable conclusion that Koa has now reached a level of complexity at which we can't continue to function without a cumulative reference grammar. I guess I've got myself a new project.

The sweeping generalization in Koa

As I was giving Amelia a brief crash course in Koa the other night, trying to come up with unproblematic demonstration sentences (not always an easy enterprise in this language), I came up against an unexpected stumbling block in the Koa translation of the sentence "I eat squash."

What specifier do you use with squash here? It's not ka, since we're not talking about any real, specific squash sitting there on the discourse stage, unless of course we introduce a new rule saying that ka is used in a general situation like this on the logic that it's always on the stage because it's all the squash there is. I don't know about that one.

It's not a for the same reasons and even more so: we're not talking about specific squashes, not raising anything to the stage.

In Koa writings of long ago, I once used ko here, as in ...suo ko sihi..., but this also makes me nervous. Obviously no one is actually eating "squashness" or "squashity" or whatever, so the sentence wouldn't be using the particle with its usual meaning. Given that there's no actual possibility of misunderstanding here, though, my "naturalness over logic" approach would tend to say that this might be a possibility.

Could we come up with a context in which this would be a problem? How about "I like cats." Using ko, this would be Ni hopi ko sene. This would also translate as "I like felinity." I don't know if this is going to work.

The particle that I threw in off the top of my head when talking to Amelia was hu. Indeed, this makes total sense when interrogative or negative: ni na suo hu sumo "I don't eat any squash," equivalent to "I don't eat squash, period." The positive gets a bit more Loglanish, though: ni suo hu sumo, "I eat some amount/set of squash." This does, of course, logically mean that the person in question is a squash-eater, since "some" here means any amount greater than 0 ("there is some squash that I eat"), but the unnaturalness kind of chafes.

Plus, does it make any sense with a count noun? Could Ni hopi hu sene mean "I like cats?" Well, using the logical paraphrase, this would mean "there are some cats that I like," which is not what this sentence generally means in English. I think a better translation might be Ni hopi po sene: "I like [all] cats," in the same sense as Po mehe i pua "[all] men are bad."

Then again, is the assertion "I like cats" equivalent to "I like all cats?" I'm not sure that it is. In a way, I feel like "I like felinity" is more on the mark.

A crazy idea that occurred to me in the shower this mornin, centered around the fact that we're not talking about either any real cats or felinity as an abstract concept. As such, none of our particles are 100% appropriate, which leads me to recall that, in some languages, objects in this situation would be incorporated.

Could this be a solution in Koa? Could "I eat squash" become Ni suo sumo ("I squash-eat"), and "I like cats" be Ni hopi sene ("I cat-like")?

Well, what would this mean? In a structure like this, the second word should be modifying the first adjectivally. To figure out if using words in this way would cause a problem, we're going to need a sentence in which the object could also be a description of the agent, something that's proving more difficult than we might have expected.

I think the most fruitful area will be if the object is an adjective. For example, forgiving the subject matter, take ta [murder] [evil]. Using "object incorporation," this could mean either "he is an evil murderer" or "he murders evil ones," obviously a completely unacceptable ambiguity even if we have loosened up in recent years.

So no object incorporation, then. Luckily, as it turns out, we already have a solution to this problem! Exactly one year ago (eerily), I wrote: "the predicate logic design isn't necessarily all that relevant to human linguistic needs; what I've done is to give that meaning to these particles in conjuction with an article, but to give them a more pragmatic/specifier-type when immediately preposed to a noun." Among the following examples are these:

po *neko = cats in general

poa *neko = every cat, period

poka *neko = all of the cats onstage/in the given set

So there we have it. In that case, "I eat squash" becomes ni suo po sumo, and "I like cats" is ni hopi po sene. And let's try not to forget it again a year from now...

How's your day going?

Mo kea sa ka pai se?

Written to Mia via chat on 11/1/10

This is an interesting one. Since ideally the question word is of the same form as what's going to replace it ("koa," "pua," etc. in this case), it looks on the face of it like the question should really be Kea sa ka pai se? I was going to say that this is obviously stupid, but now I start to think about it...

Isn't the "what kind" question necessarily in reference to a definite NP, where as the "what" question refers to an indefinite one? So we could have kea sa a talo be "what is the house," and kea sa ka talo be "what kind of house is it."

The logically possible combinations, then, and their translations would be as follows:

kea sa a talo? < a talo i kea? "what is a house?"

kea sa ka talo? < ka talo i kea? "what is the house like?" "what kind of house is it?"

keka sa a talo? < a talo i keka? "which of them is a house?" or similar

keka sa ka talo? < ka talo i keka? "which one is the house (you were talking about?)"

If this is the case, a question like "how is your day" would be translated as Kea sa ka pai se? Well, at least that part. As to aspect, though...

...should it be aorist as above? Or imperfect because we're talking about the internal structure of a bounded event? Or perfective because we're talking about the show so far? Maybe it's the difference between "how has your day been?" and "how is your day going," which pragmatically is not very important.

Now, if it is some aspect other than aorist, how on earth do we apply topicalization?!

Ka pai se i si kea?

Si kea [sa [ka pai se]]?

Kea [si sa [ka pai se]]?

Kea [sa [ka pai se] si]?

Kea [sa [ka pai se] si DUMMY VERB?

I think the only reasonable possibility would be to front the entire verb complex avec aspectual particle, or the dummy verb strategy that we're going to have to figure out later.

Really the whole topicalization-with-verbs issue is a problematic one. I don't know if we ever really thought about what the hell we would do if, for example, we're topicalizing a transitive verb:

ta si suo a nuhu "[and then] he ate a beetle."

ei si suo sa ta a nuhu? "[wait, what?] he ATE a beetle?"

Obviously that doesn't work. What if we passivize the verb so we can have an oblique argument?

a nuhu si pa suo o ta > ei si pa suo sa a nuhu o ta?

Yeah, I don't know what to say about that, really. Anyway, why use the passive when it's the verb we want to be emphasizing in the first place? A couple options:

• Use a dummy verb: ei suo sa ta si *teke a nuhu? lit. "Is it eating that he did to the beetle?"

• Use some kind of cleft construction, topicalizing the whole sentence instead of just the verb. We don't have a way of doing this preconfigured; how to translate "Is it that X..." into Koa? Ei tia sa, ko ta si suo a nuhu? You don't have to analyse this into an IE cleft construction at all, actually, which is nice -- it just means "is that it, is that right?" In fact, we could even use the "true" root, which doesn't exist yet, to do this!

So yes, actually, I'm potentially happy with either of those.

What about "how" in the genuine adverbial sense? I came up with mo kea on the analogy of Bislama olsem wanem, literally "like what?" Does it make sense for me to use this? We need some examples.

Okay, here we go: "How do I find a frog?"

Um...you know, even before I get to the "how," this is anything but straightforward. We haven't given any thought to the semantics of questions, but what exactly is being asked here? It's something like "what are the steps by which I might find a frog?" My temptation is to render this with "can" ("how can I...") but that's just Polish thinking, I bet. Let's leave that aside and just use the root verb for now.

HOW sa ni luta a iki?

Well, what are some potential answers to this question?

* "With a froggy divining rod."

* "Look in a pond," or "do the following things..."

* "Very cautiously."

What are these answers, then? One is an adverb of manner, another an instrumental noun, another an imperative verb. Clearly the question can't possibly anticipate all of these syntactically. Well...actually, if the question were more pointedly "what instrument do I use to find a frog," it would be appropriate to begin with me kea. But supposing that we have no knowledge to start with...

1) maybe mo kea makes sense, but

2) maybe the whole structure of the question should be different: essentially "I do what so I find frog?"

Kea sa ni teke la ko ni luta a iki?

I mean, that's really the question here, once we take away conventionalized IE ways of saying it.

Se luta a iki mo ko se teke tika...

Of course, clausal connectors are an area I've almost entirely neglected so far, so I have no idea how this is going to work. I'm sure some morphologically simple languages would say something more like "se lu luta a iki, se teke tika..." etc. Lots of stuff to figure out.

Back to musical chairs with the particles

So after a mere five minutes or so with the letter reproduced in the previous post, Allison pointed out that a phrase like ti kani was ambiguous: it could mean either "she sings" or "that singer."

Damn damn damn damn.

I mean, thank heavens she caught it. But I'm embarrassed that I could have let this happen, particularly with the enormous quantity of thought I gave to the decision to let the demonstrative subsume the role of the 3rd-person pronoun. But it was not to be; and really, I didn't want it to be that much anyway.

This means that ti is back to being a demonstrative and only a demonstrative, and we have a slot open for 3rd singular.

I was tempted at first to let hi return to its own, but really I think my original reasons for jettisoning it are still valid: (e.g. ni loha hi, pretty close to unpronounceable for me).

My solution for the present, that I'm feeling pretty pleased with overall, is this: the 3rd singular pronoun becomes ta, a nice gift for Mandarin speakers (which hopefully compensates for the irritation of ni), and the topicalizer reverts to sa to free up the position.

Now, yes, I know that we got rid of sa in that role to avoid sequences like ei se sa si sano, but I think Koa speakers can figure it out, and that particular particle has always felt so right to me in that position. And I think ta has a very nice pronouny feel to it.

Take note, then: all those ti kani pua, &c., from the letter to Allison should now be reforged as ta kani pua.

Nominal vs adjectival predicates in Koa, OR attribution vs identity

I wrote Allison an e-mail last week asking for her help in clearing up some long-standing confusion/debate in the Koa universe. We haven't solved the big problems yet, but the process of writing and initially talking about the e-mail has provoked some changes already, so I want to document it here.

Hey, Allison,

Pursuant to our eventual conversation (or possibly even instead, depending on how easy this ends up being to resolve), I wanted to give you the scoop on the Koa issue I'm struggling with along with some examples. You may want to print this, as it is several thousand pages long. :)

What causes the issue is that Koa doesn't have IE-style parts of speech at all. There are basically three categories: content words, particles, and names. Particles are a closed class of monosyllabic words with primarily inflectional/syntactic and some derivational functions; names are predictably an open set not required to conform as strictly to Koa phonology; and the bisyllabic base content words which can be combined to render new meanings.

What English POS a Koa content word belongs to is determined entirely by syntactic position: any word can fall into any class depending on how it's used. For example, using koa "good":

eme koa

magic good

"good magic" (adjectival)

ti eme i koa

that magic 3P good

"this magic is good" (verbal/stative)

ka koa ta ni si ana la ka sene

the good TOP 1SG PERF give ALL the cat

"as for the good one(s), I gave it/them to the cat" (nominal)

ei se si suo koa ne tipai?

QU 2SG PERF eat good LOC this.day

"have you eaten well today?" (adverbial)

The logic behind all this goes something like the following. Starting with kani "sing" and pua "bad," we can progress like this:

ka kani

the sing

"the one singing, the one who sings, the singer"

ti kani

3SG sing

"she sings"

ka kani pua

the sing bad

"the one singing badly/who sings badly, the bad singer"

ti kani pua

3SG sing bad

"she sings badly / she is a bad singer"

The problem comes up with the gloss of that last item. Are "she sings badly" and "she is a bad singer" really identical in meaning outside of lojban? I got here because I was trying to translate the sentence "she's a good singer, but she's singing badly today." I was going along these lines:

ti kani koa, no ti ma kani pua ne tipai

3SG sing good but 3SG IMPF sing bad LOC this.day

"she's a good singer, but she's singing badly today"

See, verbs without an aspect particle are automatically aorist, so ti kani koa has a timeless, general, non-referential meaning, whereas ti ma kani pua is referring to a real instantiation of bad singing occurring at some point in time.

Another option would be to use some derivational morphology to help accent the distinction being made. The suffix -ma indicates frequent/serious involvement with the root in question, either professionally or otherwise, so we could make kánima, "singer":

ti kani-ma koa, no ti ma kani pua ne tipai

3SG sing-AG good but 3SG IMPF sing bad LOC this.day

"she's a good singer, but she's singing badly today"

But what if she's not someone who identifies as a "singer?" What if she's actually an accountant who just sings at family gatherings sometimes and is actually pretty insecure about her voice, despite the fact that she sings well (except for tonight). I feel like using -ma here dilutes whatever utility that suffix would ordinarily have.

Or again, we could use the putative particle mo. I've been going back and forth for at least 8 years on whether this should actually exist; it would be an adverbializing particle, meaning "in the manner of" or "as/like," etc. Literally, then, we could do something like

ti ma kani mo pua

3SG IMPF sing as bad

"she is singing in the manner of a bad one = she is singing badly"

I feel like this starts to get at the disconnection I want to see between her current performance and her actual identity. At the same time, though, I wonder whether I'm being picky about having access to exactly the same distinctions I use in English, despite the fact that this is an IAL and there's going to HAVE to be a lot of context informing anything anyone says.

In other words, is the difference between English "she sings badly" and "she is singing badly" (and the Koa analog, the difference between an aorist and imperfect verb) sufficient to be the ONLY structural difference between these two concepts in a language?

English tends to get at this sort of thing with parts of speech, too, which is something that makes me feel uncomfortable about Koa. What if I want to say something totally straightforward, like "John is a man?" Note that "masculine" in the glosses below doesn't mean "characterized by some qualities associated with men" like in English, but rather a pure adjectival form of "man" which English has trouble forming.

a mehe

a man

"a man; a male/masculine one"

a keli mehe

a dog man

"a male/masculine/man dog"

le Keoni i mehe

NAME Adam 3SG man

"John is a man OR John is male/masculine"

Is this a problem? Is there a difference between "John is male" and "John is a man," leaving aside questions of age?

Going back to mo, I originally conceived it with the aim of using it to set off all adverbial phrases. I realized later that it's not actually necessary for that purpose, and abandoned it. Later I started to think it might not be superfluous after all and relegated it to limbo. What about a concept like "she sings like a dog?" Can I do this with that V X structure where X is automatically adverbial?

ti kani keli

3SG sing dog

"she is a canine singer = she dogsings = she sings like a dog?"

I was thinking that maybe "dogsinging" isn't quite the same concept as singing like a dog, in which case it would be preferable to say

ti kani mo keli

3SG sing as dog

"she sings like a dog"

Although once again I'm wondering if I'm just trying to parallel English structure here. A similar example:

ti kani 1642

3SG sing 1642

"she is a 1642 singer = she sings like it's 1642?"

or

ti kani mo ne 1642

3SG sing as LOC 1642

"she sings in the manner of in 1642 = she sings like she/they did in 1642"

I guess the semantic I'm trying to put my finger on is that of "like" above: resembling something, but not actually being it. She's not really singing back in time, so the sentence shouldn't be saying she is. But, but...in the sentence that started all this off, she really WAS singing badly! So if that's what mo means, I can't use it anyway. But there is this situation, which may really and finally necessitate it:

ti kani ka keli ni

3SG sing the dog 1SG

"she sings my dog. what the hell does that even mean."

ti kani mo ka keli ni

3SG sing as the dog 1SG

"she sings like my dog"

...in which case the adverbial construction V X would be interchangeable with V mo X in carrying implied indefiniteness: essentially V mo a X. Mo could be used, or not, depending on speaker preference and whether it seems to add clarity. It would be required, though, when X is definite, modified, etc. I think.

Anyway, this is the patch of thorns that I've been unable wholly to extricate myself from for the last ten years or so, and I'm hoping you have some amazing crystal-clear thought that forever eliminates all doubt. Or something. What do you say?

-Julie

The line between demonstratives and personal pronouns

Vis-à-vis the previous post, if ti can be both "he" and "that," what does this mean for the other personal pronouns?

This is to say: what, actually, should ni talo and se talo mean?

One option is that we've created a three-way deictic system along Romance lines. I have to say, though, that using the personal pronouns in this way feels intensely counterintuitive to me. I need some feedback here from other humans.

Another option would be to treat these structures as appositive. In this way, we'd have

ti neko = him, the cat = "that cat"

se neko = you, the cat = "you cat," as in "you cat, you!" or "you, cat that you are..."

ni neko = me, the cat = "me, being the cat that I am..." etc.

Maybe that's the most sensible thing. Note that this means ni neko does NOT mean "my cat" anymore.

I find myself wishing that the 3rd-person pronoun could be to instead of ti. On the other hand, I like the kea/tia opposition several thousand times more than this putative kea/toa idea, so I'm stymied again.