Hiding Waters analysis critique: another noun/verb merger

[bradrn]

Yes they do. This is a key part of all known fluid-S systems — agent and patient marking are used on intransitive verbs to specify agentivity rather than semantic role. (The classic example is Tsova-Tush: as vuižnas ‘I fell down [on purpose]’ vs so vožensO ‘I fell down [accidentally]’.) This is also part of why fluid-S systems are so remarkable; otherwise such languages would merely have widespread argument elision, which is not at all strange.

I would be very interested in a reference to any part of Dixon’s book where he describes a system remotely similar to that of Hiding Waters. (Or Old Skourene, for that matter.)

From section 3.1 of Ergativity:

This scheme, which we can call 'fluid S-marking', is found in just a few languages. It is said that in Bats, a Northeast Caucasian language, some intransitive verbs (e.g. 'go', 'play', 'look', 'speak') MUST have a 1st or 2nd person pronominal S NP in ergative case, while others can mark S function by EITHER absolutive case (implying that the action was involuntary) OR by ergative case (implying that the referent of the S NP controlled the activity, or that it was his fault). Verbs in the latter set include 'fall', 'get drunk', 'fear', and 'lie down'.26 The use of ergative or absolutive on an S NP appears to be semantically determined: instead of having to recognize classes among intransitive verbs, we can simply say that ergative case is used for a controlling S, while absolutive case is used for S NP's in other instances. The semantic nature of intransitive verbs dictates that for some the S NP is ALWAYS agentive, for some it CAN BE, and for others it NEVER is. (The ergative case in Bats is seen to be clearly 'marked'; cf. ?2.31.)

"Agentivity" in the intransitive is control/volition of their own action/situation, not a separate referent.

Yes, exactly correct. What I am saying is that in Hiding Waters, agent and patient marking on ‘intransitive’ predicates is used to distinguish separate referents rather than just marking agentivity. If it were a system purely based on agentivity, hulkuxkoụ̀n and huxslọkoụ̀n would mean something like ‘he brings [intentionally]’ and ‘the fish brings [accidentally]’; instead, they mean ‘he brings [unspecified]’ and ‘[unspecified] brings fish’.

[Vardelm]

Yes, exactly correct. What I am saying is that in Hiding Waters, agent and patient marking on ‘intransitive’ predicates is used to distinguish separate referents rather than just marking agentivity. If it were a system purely based on agentivity, hulkuxkoụ̀n and huxslọkoụ̀n would mean something like ‘he brings [intentionally]’ and ‘the fish brings [accidentally]’; instead, they mean ‘he brings [unspecified]’ and ‘[unspecified] brings fish’.

Ah, OK, now THAT makes sense. Somehow I thought you were objecting to the referent being treated as A in transitive and O (sometimes) in intransitive. That made no sense given what I understood about active-stative langs or what you had said in your ergativity thread. So yeah, now I can see the issue you're raising.

Instead of the intransitive being an anti-causative 0-derivation, couldn't the transitive just be a causative 0-derivation? I think I've seen verbs make this shift when they go from intransitive to transitive.

[bradrn]

Yes, exactly correct. What I am saying is that in Hiding Waters, agent and patient marking on ‘intransitive’ predicates is used to distinguish separate referents rather than just marking agentivity. If it were a system purely based on agentivity, hulkuxkoụ̀n and huxslọkoụ̀n would mean something like ‘he brings [intentionally]’ and ‘the fish brings [accidentally]’; instead, they mean ‘he brings [unspecified]’ and ‘[unspecified] brings fish’.

Ah, OK, now THAT makes sense. Somehow I thought you were objecting to the referent being treated as A in transitive and O (sometimes) in intransitive. That made no sense given what I understood about active-stative langs or what you had said in your ergativity thread. So yeah, now I can see the issue you're raising.

Yep, I have no objections to treating the S like A or O.

Instead of the intransitive being an anti-causative 0-derivation, couldn't the transitive just be a causative 0-derivation? I think I've seen verbs make this shift when they go from intransitive to transitive.

I already suggested this:

… Either there’s some sort of zero-derived causative construction happening here …

But I rejected this because trailsend said predicates are underlyingly transitive, and in any case it’s somewhat inconsistent with (23a–d).

EDIT: Actually on further reading it does turn out there’s ‘some sort of zero-derived causative’, see following post for details.

[bradrn]

OK, I finally got around to reading the rest of trailsend’s post:

24a) Kestulisdọq xụlqulilhq.

kestu<li>sdọq

west.IPFV<CLF(generic).AGENT>

xụlqu<li>lhq

hunt.ESS<CLF(generic).AGENT>

The hunters are making their way westward.

The hunters are marked as agents on kestulisdọq because they are actively going. (Note that I fixed an error I'd made in (10) which made its way into cedh's original version of this: the hunters need to be marked as agents, not patients, of xụlqulilhq to make them professional hunters and not professional prey.)

24b) Kestuslịdọq xụlqulilhq.

kestus<lị>dọq

west.IPFV<CLF(generic).PATIENT>

xụlqu<li>lhq

hunt.ESS<CLF(generic).AGENT>

The hunters are being driven westward.

In (24b) the hunters are marked as patients on kestuslịdọq, which means they aren't effecting the movement westward, just experiencing it—since such things don't just happen, the listener understands that something is driving them. But:

24c) Kestuslọdọq hulhlọwh.

kestus<lọ>dọq

west.IPFV<CLF(generic).PATIENT>

hulh<lọ>wh

wind.ESS<CLF(generic).AGENT>

The wind is blowing to the west.

In (24c) the wind is marked as a patient on kestuslọdọq because it is experiencing the westward motion, not actively effecting it—but no one has to make the wind go anywhere, that just happens, so no effecting agent is assumed.

This is basically just normal fluid-S active-stative alignment: all intransitive arguments have the same semantic role, in this case ‘thing which moves west’, while the agent/patient marking shows the control of this argument over the action.

There might be a way to tease apart whether ṇ~ng is an intransitive root with a zero-derived causative, or a transitive root with a zero-derived anticausative. It's not always the case that a predicate with just an agent or just a patient can be interpreted intransitively. For example, (23a) is definitely interpreted as "he is bringing something", not "he is bringing himself". So, we could group together roots like ṇ~ng and kest~dọq, which can be interpreted intransitively in the right context if only one argument is marked, and call them "intransitive" roots, while roots like h~koụ̀n, which are always interpreted transitively even when only one argument is marked, we could call "transitive."

Yes, I think this classification would work well. Then there are two types of predicates:

• ‘Intransitive’ predicates: when monovalent, A vs O marking reflects differences in agentivity; when divalent, has a causative interpretation with O being the original argument and A being the causer
• ‘Transitive’ predicates: when divalent, has both arguments expressed; when monovalent, A vs O marking reflects the suppressed argument

The only wrinkle in this is your example of monovalent ṇu~ng with a causative interpretation. As I see it, what’s happening there is that it starts off as monovalent intransitive ‘O sleeps’, then is promoted to divalent ‘A makes O sleep’ — which is then somehow being treated as a basically transitive predicate and thus is demoted to ‘A makes ___ sleep’. I’m not entirely sure what the best way is of analysing this construction.

[trailsend]

Yes, I think this classification would work well. Then there are two types of predicates:

• ‘Intransitive’ predicates: when monovalent, A vs O marking reflects differences in agentivity; when divalent, has a causative interpretation with O being the original argument and A being the causer
• ‘Transitive’ predicates: when divalent, has both arguments expressed; when monovalent, A vs O marking reflects the suppressed argument

The only wrinkle in this is your example of monovalent ṇu~ng with a causative interpretation. As I see it, what’s happening there is that it starts off as monovalent intransitive ‘O sleeps’, then is promoted to divalent ‘A makes O sleep’ — which is then somehow being treated as a basically transitive predicate and thus is demoted to ‘A makes ___ sleep’. I’m not entirely sure what the best way is of analysing this construction.

It's just behaving like a transitive predicate, no? Just like in h~koụ̀n, the patient marker can be omitted, but the transitive interpretation (which for ṇ~ng is causative) remains, just with the patient unspecified.

So we can say that a semantically causative, transitive predicate may be transparently derived from any intransitive predicate, and such a derived predicate can mark both an agent and a patient, or, just like other transitive predicates, omit an argument to leave it unspecified. Since the derivation is not overtly marked, a derived causative predicate with one unspecified argument is differentiated from an intransitive predicate just by context.

But, thinking on it, would it be simpler to describe it this way?

Predicates may mark an "agent", which is the referent effecting the state the predicate describes, and a "patient", which is the referent experiencing the state. One of these arguments may be omitted to leave them unspecified (and in fact, in some constructions, both arguments may be omitted). We can then distinguish two groups of predicate roots: transitive roots are those for which a predicate with only one marked argument is still understood to describe one party effecting a state on something else. Intransitive roots, however, are those for which a predicate with only one marked argument is understood to describe one party effecting a state on itself (or one party experiencing a state without any other party effecting it) when the semantic context permits. In such a context, a speaker may mark the predicate's single argument as either an agent or as a patient to indicate a presence or absence of agentivity.

[bradrn]

Yes, I think this classification would work well. Then there are two types of predicates:

• ‘Intransitive’ predicates: when monovalent, A vs O marking reflects differences in agentivity; when divalent, has a causative interpretation with O being the original argument and A being the causer
• ‘Transitive’ predicates: when divalent, has both arguments expressed; when monovalent, A vs O marking reflects the suppressed argument

The only wrinkle in this is your example of monovalent ṇu~ng with a causative interpretation. As I see it, what’s happening there is that it starts off as monovalent intransitive ‘O sleeps’, then is promoted to divalent ‘A makes O sleep’ — which is then somehow being treated as a basically transitive predicate and thus is demoted to ‘A makes ___ sleep’. I’m not entirely sure what the best way is of analysing this construction.

It's just behaving like a transitive predicate, no? Just like in h~koụ̀n, the patient marker can be omitted, but the transitive interpretation (which for ṇ~ng is causative) remains, just with the patient unspecified.

Isn’t this just what I said?

So we can say that a semantically causative, transitive predicate may be transparently derived from any intransitive predicate, and such a derived predicate can mark both an agent and a patient, or, just like other transitive predicates, omit an argument to leave it unspecified. Since the derivation is not overtly marked, a derived causative predicate with one unspecified argument is differentiated from an intransitive predicate just by context.

Yes, this appears to be the case.

But, thinking on it, would it be simpler to describe it this way?

Predicates may mark an "agent", which is the referent effecting the state the predicate describes, and a "patient", which is the referent experiencing the state. One of these arguments may be omitted to leave them unspecified (and in fact, in some constructions, both arguments may be omitted). We can then distinguish two groups of predicate roots: transitive roots are those for which a predicate with only one marked argument is still understood to describe one party effecting a state on something else. Intransitive roots, however, are those for which a predicate with only one marked argument is understood to describe one party effecting a state on itself (or one party experiencing a state without any other party effecting it) when the semantic context permits. In such a context, a speaker may mark the predicate's single argument as either an agent or as a patient to indicate a presence or absence of agentivity.

I dislike these sorts of purely semantic definitions because they inevitably end up as post hoc rationalisations which depend on the specific language in question. e.g. Does ‘I see you’ have one argument effecting a state on another, or one party effecting a state on itself? How about ‘The food pleases’? ‘You sneeze’? Languages differ on their opinion of all three — you can argue each either way. For this reason I prefer to look for more objective criteria such as ‘the A argument has the same semantic role as the S_A argument’. (I’m tempted to say I’m looking for ‘purely syntactic’ criteria, but obviously something involving semantic roles can’t really be ‘purely syntactic’.)