Word order when words change category

[zaba]

So, consider a sentence with a complement clause like the one below:

1SG-know Tjina-GEN ajawaska-ACC drink-NMZ-3POSS
'I know (of) Tina's yage drinking.' (i.e. 'I know that Tina drinks yage)

How would you determine the word order
looking only at the roots you get [[V ] + [S+ O + V]] (with the SOV being the complement clause)

BUT

the final V of the CC is nominalised.
So when I describe the order, can I still really say that the CC is SOV?

[MalFet]

I'm not recognizing those two morpheme glosses on "drink", but if Tjina is in a genitive case and the "verb" of the lower clause is nominalized, "drink" is probably not a functional verb in this sentence. Likewise, it might not really make sense to talk about Ss, Os, or Vs here. That's an architecture for finite clauses, and this doesn't really look like one.

[zaba]

I think that [Tina's yage drinking] is a complement clause -- is this naive?

1SG-know Tjina-GEN ajawaska-ACCUSATIVE drink-NOMINALIZER-3POSSESSIVE

[MalFet]

It's not naive, but there are as many clause classifications as there are languages. In other words, the real question here is what are you trying to accomplish such that the difference between a complement clause and a nominalization is important?

Generally speaking, a nominalized verb is just that: a noun.

[Daniel]

"3POSS" means "possessed-by-3rd", right?
("Possessive" is ambiguous, either meaning genitive or its inverse.)

I'm wondering about the lexical semantics of "know"-- could it mean roughly "be aware of" or "be familiar with"? It may not be a complement clause at all but rather a true nominal. "Her drinking is too frequent" is probably also possible. That's not a clause then.

In English, there are some big differences between true verbs and nominalized verbs:
"The playing of football was fun." (Less verby)
"Playing football was fun." (More verby)

Sometimes it is argued that nominalized verbs still have internal verbal syntax. So I'd expect (from just that sentence) that this is an OV language (if not, it might be weird like the example above in English). But then the possessor is not a subject. It's instead in the Possesor-Noun relationship, which is probably the same word order for normal nominals ("her food", not "food her").

[zaba]

"3POSS" means "possessed-by-3rd", right?
("Possessive" is ambiguous, either meaning genitive or its inverse.)

Yes.

I'm wondering about the lexical semantics of "know"-- could it mean roughly "be aware of" or "be familiar with"?

Not really. It is in the sense of "know a fact"

this is an OV language

Mainly, yes.

But then the possessor is not a subject. It's instead in the Possesor-Noun relationship, which is probably the same word order for normal nominals ("her food", not "food her").

Yes, but the hyphen are morpheme boundaries, not word boundaries.

1SG-know Tjina-GEN ajawaska-ACC drink-NMZ-3POSS
'I know (of) Tina's yage drinking.' (i.e. 'I know that Tina drinks yage)

Literally: I know Tina's drinking (of) yage.
There are three components here, (I) sentence; (II) VP, and (III) NP (or is this a VP?).

The sentence is the entire thing, above.

II. VP
VERB = 1SG-know

III. NP (or VP?) [which I claim to be a complement clause]
SUBJECT  = Tjina-GEN
OBJECT = yage-ACC
VERBALIZED-NOUN = her drinking (of it)

How can I analyse this?

[Daniel]

Yes, but the hyphen are morpheme boundaries, not word boundaries.

I was referring to the full NP, Tjina.

The sentence is the entire thing, above.

II. VP
VERB = 1SG-know

III. NP (or VP?) [which I claim to be a complement clause]
SUBJECT  = Tjina-GEN
OBJECT = yage-ACC
VERBALIZED-NOUN = her drinking (of it)

How can I analyse this?

More data! It's hard to generalize from this.

I'd still question whether Tjina is a "subject". I'd call it a possessor. It's more like "Tina's book" than "Tina read".

We've said this is a nominalized predicate-- a noun, not a clause/verb. But with some internal structure. It's acting like a clause in a way, but it doesn't appear to be one, at least not a normal one.

Not really. It is in the sense of "know a fact"

I meant it more in terms of argument structure. Consider:
"I saw Tina's drinking"
vs.
"I know that Tina drank."

It's possible this really isn't clausal at all, even semantically. It's possible this is just a verb "know" that means "experience as fact" or "observe" rather than "know that X". That's all in the very subtle details of translation (and never try to understand the exact meaning via translation), but I see it as a real possibility that the language has no exact equivalent to English "know that", but rather that it uses a structure like "observe Xing".

[zaba]

We've said this is a nominalized predicate-- a noun, not a clause/verb. But with some internal structure. It's acting like a clause in a way, but it doesn't appear to be one, at least not a normal one.

Okay, so maybe it shouldn't be analyzed as a complement clause. What do you think it should be best analyzed as, then? Maybe just a nominalization construction? This construction is rather rare in the corpus so I don't have many other examples.

[Daniel]

It's a predicate/clause acting as a noun. So, yes, a nominalized construction.

You should definitely look for examples of this in subject position: "Eating is good" etc.

[zaba]

"Smoking is good" would be:

Smoke-NOMINALIZATION good
(there is no need for a copula in this construction)

[Daniel]

That's why this does not seem to be specifically a complement clause.

Instead, it's just a noun (which happens to have a verb inside it) that is sometimes used as a complement. The same is true of "apple":

I like the apple.
The apples are good.

[zaba]

But we can still agree that this is a complement clause, right?

the ghost anger-self-subordinator really eat-NMZ-ACC 3-want sausage-ACC
'the ghost, having angered himself, really wanted to eat sausage'

[Daniel]

Not really. It appears to function as one practically, but it's not clear that it is distinct from a nominal complement:

I want apples.
I want apple-eating.

You'd have to find some evidence that it is more "verby" than that.

In terms of "How does this language do complement clauses?" that may be the answer, but that's just how it would translate the English equivalent. It doesn't mean there's a separate grammatical process/form to describe.

[zaba]

I guess you're concerned because the compliment classes have no verbs on the surface. Fair enough. This looks like an example of English influencing the classification. Do you agree?

[Daniel]

This looks like an example of English influencing the classification. Do you agree?

Yes. This is also why I often wonder about the accuracy of the glosses. It seems consistent in the examples you post, but it's unclear what, for example, "nominalizer" means so a lot of this argumentation is circular, though it seems to be developing toward a good analysis.


There are three ways we could analyze such a structure:
1. By translation/function. It's not totally inappropriate to call this a complement clause in the sense that it seems to function as one from our English perspective.

2. Language internally: what distributional evidence is there for a relevant category of "complement clause"? None. There are nominalized predictes and unsurprisingly they can be used where nouns are used, even with case endings. The fact that a noun can be a complement is very uninteresting.

3. Universally/theoretically: if you subscribe to the idea that all languages share an underlying structure, we might wonder how these develop. Possibly these are "complement clauses" (or whatever all languages have along those lines) being realized through the specific tools available in this language.

So the type of analysis really does determine the answer.

As far as I'd be concerned, for writing a descriptive grammar, I'd just call these nominalized predicates and note that they can appear wherever normal nouns can, with the same case endings. That's clear.
Are there any exceptions to that? If so, you may find a reason to discuss whether there is some kind of special "complement clause" construction. If not, there's no point-- you can leave the theoretical reasoning (#3 above) to someone else.



Edit:
I happen to be looking at a book by Aikhenvald and Dixon (2006; Oxford) called Complementation so for a typological perspective that could be helpful for you.