Chinuk Wawa isn’t “Nominative-Accusative” nor “Ergative-Absolutive”

This is either a trivial point or a huge one.

In linguistic typology, there’s general agreement that our data on the 7,000-ish known languages shows they fall into just 3 broad sorts, based on how they mark:

• The 2 core roles in a Transitive clause, which are always differentiated from each other:
  • Who or what does an action (Agent).
  • What or who the action is done to (Object).
• And, the only role in an Intransitive clause (Subject), which is sometimes differentiated into sub-types.

The types of syntactic alignment

In one common type of languages, like English and French, Agents and Subjects are encoded in one way (with what are traditionally called “subject pronouns” by teachers of those languages), while Objects are encoded in a separate way (with “object pronouns”). In these NOMINATIVE-ACCUSATIVE languages [a label we get from traditional Greco-Latin grammar], you find (A+S) versus (O).

In a second type of language, also common worldwide, Agents are encoded in one way, while Objects & Subjects are shown differently — so in these ABSOLUTIVE-ERGATIVE languages [a needlessly fancy term!] it’s (A) versus (S+O).

In the interesting third type we’re concerned with, sub-types of intransitive Subjects are distinguished. You have Agent-like ones (S.a, as in ‘I’m watching’, where ‘I’ am in control of that action) (“unergatives” for some linguists), and Object-like ones (S.o, as in ‘I tripped’, where ‘I’ have no control over the action) (“unaccusatives” for some linguists). In these STATIVE-ACTIVE languages, you’d have Agent marking on the ‘watch’ verb, and Object marking on the ‘trip’ verb. So it’s (A+S.a) versus (O+S.o). Because they’re diagnosed by the division within their Subject system, these are also called SPLIT-S languages. The literature of linguistic research seems to suggest that SPLIT-S languages are pretty rare, but we have plenty of them in North America’s large Siouan family of languages, such as Lakota.

Where does Chinuk Wawa fit into this?

Now then — in my PhD dissertation, I characterized Kamloops (BC) Chinuk Wawa as a Nominative-Accusative language.

I’m having second thoughts! Yup, linguists can be wrong.

Based on a good deal of work I’m doing lately with Grand Ronde (Oregon) Chinuk Wawa, I’m seeing the Jargon more as a SPLIT-S (Stative-Active) language. To my knowledge this analysis hasn’t been put forth before, so I’ll try to explain why I think this way.

For starters, it’s easy to see that Transitive clauses in Jargon do distinguish do-ers from do-ees. Do-ers (Agents) come before the verb, and do-ees (Objects) follow. Their position in the clause — their “word order” — is their role-marker.

(Shop talk: This is where I tell “generative” linguists that they’re out of touch, if they think a relatively isolating language like Chinuk Wawa “has no marking” of arguments. It’s primarily because of syntax that CW speakers understand without fail what roles are being expressed in an utterance.)

And with Intransitive clauses, I had originally concluded that their only role (“argument”) is their Subject, which is obviously not an Object.

So I figured Chinook Jargon must be a (A+S) versus (O) language. That is, an ACCUSATIVE-NOMINATIVE language, much like the French and English components of its heritage.

But a major feature of Intransitive clauses in Jargon that I took little account of in that analysis is that word order matters within the Intransitive class too.

Briefly, let’s recall that Intransitive clauses in Chinook Jargon aren’t just the ones with obviously verbal predicates in them, like these have (Subjects will underlined):

(1) Wík-qʰə́nchi kʰəpít snás… (S.o) (read on to understand these abbreviations)
not-ever stop rain
‘It never stops raining…’ (‘The rain never stops…’)

(2) T’álap’as pi lílu łaska míłayt... (S.a)
coyote and wolf they live…
‘Coyote and Wolf lived (there)…’

[I analyze the łaska here as an “agreement marker” on the verb, not as a statement of the Subject.]

In both (1) & (2), the subjects of the verbs ‘stop’ and ‘live’ are easily seen to be marked, by virtue of their position before those verbs, as Agent-like.

Now, there are also lots & lots of Intransitive clauses where the predicate doesn’t look like a verb of action:

(3) Wík (t)k’úp-tílixam nayka álta. (S.o)
not white-person I now
‘I‘m not a White person now.’

(4) x̣ə́nłq’i ya pʰík’w. (S.o)
crooked his back
‘His back is crooked.’

In cases like these last two, you have the Subject being either identified (as a non-White person) or described (as being crooked).

Now, I’ve often worked with an idea that such clauses are using a “null copula”, a silent be-verb that I’ve symbolized with Ø.

But I’m moving steadily in the direction of feeling that that view, too, is unconvincing. (If you can’t call yourself into question, are you really alive?) 🙂

Let me explain that.

I do believe in “nulls”, but only when they’re in a “paradigm” — a complementary distribution with some provable non-null form having comparable function, as in the Jargon 3rd person pronoun system (where yaka covers the animates and Ø is the norm for inanimates).

The more I think about it, the more I realize that the supposed null be-verb is not in any such relation. There just isn’t any non-null form that also means ‘be [identified as / equated with]’ or ‘be [a certain quality]’!

This has consequences.

If there’s no “be-verb” for identifying & describing functions, one implied view is that the non-Subject portions of example sentences (3) & (4) are “stative verbs”.

That is, the nominal (t)k’úp-tílixam, by virtue of not having its other major role as an argument in a Transitive clause (–and there’s a complementary distribution, by the way–), is a verb stem meaning ‘to be a White person’.

And the adjectival x̣ə́nłq’i, due to its not being in its other major role as a dependent member in a noun phrase (–again there’s a statement of a complementary distribution–), is a verb root meaning ‘to be crooked’.

Following through with the theme of a paradigm of be-verb (“copular”) expressions, there indeed are special verbs in Jargon (just not nulls) for other senses of ‘existing’, for instance míłayt ‘be somewhere; exist’ and t’úʔan ‘have’. (Yes, ‘have’ is considered a copular verb by linguists.) Thus, a simple sketch of the copular system in this language would be:

• • míłayt ‘be somewhere; exist’
  • t’úʔan ‘have’
  • (stative verbs) ‘be identified as’; ‘be described as’

(There are linguists who would instead analyze (3) & (4) as “verbless copular constructions” but I’ll pursue that no farther for now.)

If we entertain this idea of stative verbs being common in Chinuk Wawa (because if there are any, there are tons of them, as there are endless ways to identify or describe a Subject), a couple of questions emerge straightaway.

One question is, what else is there besides stative verbs? For a linguist, everything that’s not what we’ve called a stative verb is an “active verb”, and whether it’s transitive or intransitive, its Subject will be marked as S.a.

The second question is, how are stative-verb Subjects marked? Like Agents, or like Objects? Or…a logical possibility…both ways??

In both (3) & (4), the Subjects occur after the stative verbs, that is, in the same position as Objects, so we’d describe them as S.o. Compare this with the non-statives in (1) & (2), where we found S.o with ‘stop’ but S.a with ‘live’.

Many of my readers will be aware that subjects of statives don’t always have to pattern this way in Chinook Jargon. It’s totally okay to alter our examples & put the Subject first:

(3′) Nayka wík (t)k’úp-tílixam álta. (S.a)
I not white-person now
‘I‘m not a White person now.’

(4′) ya pʰík’w x̣ə́nłq’i. (S.a)
his back crooked 
‘His back is crooked.’

But a number of known factors indicate that this alternative ordering of Subject first in a stative clause (thus S.a) is not considered the default / norm way to talk.

• I’ve heard it expressed a number of times that it sounds more like “White” talk, as it sticks to an English and French pattern.
• An evaluation by Grand Ronde’s language program has stated that elder speakers “often” used this S.a stative ordering, but that at least as often, they used the more distinctive Subject-last S.o ordering.
• And I hypothesize that a detailed examination of discourse and textual usage will show some significant portion of seeming S.a statives to in fact be “topicalizations” — moving the Subject away from a post-stative-verb position in order to highlight its relevance.

So we can argue that Subjects normally come after stative verbs, and thus are marked S.o.

And, so far, it sure looks to me like Chinuk Wawa could be a SPLIT-S language. That’s to say, it would seem as if certain verbs are active, and the rest are stative.

Branching off & going with the flow of this…

But now I’m going to repeat example clauses (1) & (2), so you don’t have so scroll so much while we refer back to them:

(1) Wík-qʰə́nchi kʰəpít snás… (S.o)
not-ever stop rain
‘It never stops raining…’ (‘The rain never stops…’)

(2) T’álap’as pi lílu łaska míłayt... (S.a)
coyote and wolf they live…
‘Coyote and Wolf lived (there)…’

Neither of the verbs here requires you to put the Subject just where we happen to find it in these examples. Take a look at (5) & (6):

(5) Álta ya kʰəpít. (S.a)
then she stop
‘Then she just quit.’

(6) …ánqati kwánsəm yaka míłayt sáx̣ali-táyí. (S.o)
…Distant.Past always he exist above-chief
‘…(since) long ago God has existed.’

[Again, I see an “agreement marker” yaka on the verb of (6), separate from the Subject sáx̣ali-táyí.]

The difference between the Subject placements in (1) & (5) strikes me as reflecting how the S.o rain has no control over stopping, whereas a human S.a “she” does control her up-and-quitting.

In a similar way, (2) & (6) differ from each other in that S.a Coyote & Wolf have control over where they reside, whereas anything that exists — even including the S.o Creator!! — can’t have chosen to come into being. The difference is “Agent” versus “non-Agent” status.

So, did you notice? — Examples (1-2) & (5-6) show us a further interesting wrinkle. These are pairs of clauses built on the same exact predicate, showing that it can have either an Agent-like or an Object-like Subject.

Here’s where I note that a subtype of SPLIT-S (and therefore rarer still?) is the “FLUID-S” language, where at least some of the intransitive predicates allow you this choice of S.a or S.o marking. In other words, unlike a SPLIT-S style of strict division, where some Intransitives mark their Subjects like Agents, and the other Intransitives mark Subjects like Objects, there’s an overlap zone. In such a “FLUID-S” language, you find the same verb able to function both ways. So e.g. the root/stem for ‘fall’, marked with A, will mean ‘throwing yourself down on the ground’, and marked with O, it’ll mean ‘accidentally fall’ or ‘be dropped’ by someone.

So I’m thinking, unexpectedly, that Chinook Jargon is a FLUID-S language.

Thanks to some work by Henry Zenk and Tom Larsen that has come my way and inspired today’s post, I further speculate that there are additional grammatical structures in the Jargon that back up this view. I’ll have to write about those soon.

Origins?

As a practitioner of “linguistic archaeology”, I always have to ask how Chinuk Wawa came to have a given feature.

It’s unclear to me whether your average FLUID-S language lets few or many verbs take advantage of this flexible Subject marking. Is there such a thing as an average FLUID-S language? In Chinuk Wawa, I’m not quickly finding many verbs that allow this kind of choice, although more such may turn up. The first place that occurs to me to go looking is at the set of verbs that involve potentially gravity-assisted motion. Huh? My reasoning is that these denote actions that easily could either occur by human choice, or by the laws of physics just doing their thing, e.g.:

• láx̣ ‘to lean, tip’
• yíx-yix ‘to be disoriented, reeling’
• x̣ə́ləl(-x̣ələl) ‘to move, shake, quiver’
• t’łə́x̣ ‘to tear, rip’
• t’sə́x̣ ‘to split’
• t’łə́p ‘to sink’
• t’ík-t’ik ‘to drip’

It remains to be seen whether we have sufficient documented examples of each of these to judge whether any of them allow FLUID-S marking choices.

By contrast, in a language of the Caucasus Mountains, Tsova-Tush a.k.a. Bats, a check of 303 intransitive verbs with fluent speakers showed that about two-thirds easily allow this fluidity!

My hunch, awaiting further research, is that Salish languages such as the Southwest Washington ones that helped form Chinuk Wawa, are comparable with Tsova-Tush. It’s pretty easy to use a given Salish base form with either Transitive or Intransitive inflection (in this case, shown by affixes). I know that certain linguists have been investigating questions of “unergativity” and “unaccusativity” in sister languages for several years, mainly in Central Coast Salish. So I suspect that SW WA Salish may turn out to be an excellent historical model (source) for Chinuk Wawa’s FLUID-S syntactic alignment.

As with so many of these “historical source” questions, it will be harder work to determine whether the less well-described Chinookan languages can be seen as FLUID-S. That family certainly has been described as “ergative” languages instead, for many years, but as pioneering researcher of ABSOLUTIVE-ERGATIVE systems RMW Dixon takes care to point out (as in §2.5 “Avoid Sloppy Terminology” of his 2010 “Basic Linguistic Theory” textbook), the word “ergative” is very often misused.

What do you think?

PS — some of my logic-oriented readers may be thinking, there are other possible combinations of the role-marking we’ve been talking about.

• Yes, sure enough, there are indeed languages where S, A, and O are all clearly distinguished from each other by unique grammatical markings (but this is rare, and I understand it’s usually limited to some small sub-area in the grammar of one of the above language types — e.g. it’s confined to 3rd persons in the Peruvian language Cashinawa).
• And yes, some languages are said to not differentiate the marking of S, A, and O from each other at all, e.g. Thai, which leaves it to your grasp of context to figure out those roles; this too seems rare.
• Finally, I should specify that even though Subject can be subdivided into “S.a” and “S.o” subtypes, no known language arranges (S.a+O) versus (A+S.o) marking, just as no language marks (A+O) identically. It’s utterly fundamental to all human speech, and by implication thought, that we distinguish “do-ers” from “do-ees”.