New CBB

Is anyone familiar with the morphosyntactic alignment presented in http://www.eskimo.com/~ram/lexical_semantics.html?

Basically, there are three "true" verb arguments which mirror the semantic roles: agent, patient, and theme (or "focus", but I think that is misleading). This system wasn't very obvious to me, but it seems very easy to understand and apply.

For example, in "I like pizza", I am the patient (not agent) and pizza is the theme. In "I showed him the bird", I am again the agent, he is the patient, and the bird is the theme. In addition, this would be the same verb as to see and to look at (the first being agentless and the second having an identical agent and patient).

Are there any natural languages that have or come close to having an alignment like this? I suppose an ergative, secundative language would get you halfway there. What are your thoughts on including this in an otherwise naturalistic conlang?

The question seems a little misleading; in some sense, all morphosyntactic alignments are based on semantic roles. They vary in which semantic roles they bucket together.

It may help to consider that semantic roles are really infinite. We group them together under conventional terms for sake of convenience, but each term we coin actually represents an infinite spectrum of distinct semantic roles. However, if we don't have to deal with languages that distinguish those roles, we can get away with throwing them all in a box and calling it "theme." Just don't forget that this sleight-of-hand has occurred. For example, the semantic-role frameworks I'm familiar with would consider the "patients" in your examples to be "experiencers."

In other words, be wary of thinking about any particular semantic role as a stable, objective thing. They're actually imaginary boxes that we've drawn in particular ways for convenience, but which could easily be drawn differently.

The system you describe sounds to me most like a fluid-S morphosyntactic alignment, which could give you behavior like this bit:



FWIW, my Feayran acts somewhat close to this, except that it typically treats semantic themes as morphosyntactic patients, and semantic experiencers as morphosyntactic locatives. So you get things like:

Lhujukústuholukzeì.
visible-1.AGT-CAUS-bird.PAT-male.LOC
I showed him the bird.*

Which has the same morphosyntactic structure as:

Lujukústudokkokw.
position-1.AGT-CAUS-bird.PAT-rock.LOC
I put the bird on a rock.


* Fun fact: you would only say this if the bird was concealed somewhere, and then you revealed it to someone. If the bird was up in a tree somewhere and you pointed it out to someone, you would instead say:

Kujukúlkushokùoqalhèe.
looking-1.AGT-CAUS-male.PAT-bird.LAT
I made him look towards the bird.

Now, the other person is the patient, and the bird is a lative argument.

_________________
任何事物的发展都是物极必反，否极泰来。

Yeah, I meant fluid-S and secundative.


That's very cool. What about a normal "A sees B" sentence? Would this still use the locative, or would A be marked as an agent? Also, is there any natlang that works like this?

I'm familiar with Latejami. I analyze the alignment to be plain nominative-accusative-dechticaetiative. But then I don't consider the derivational argument structure morphology to be part of what determines the alignment.

If instead of the argument structure morphology, you had case markings on the arguments, one for A(gent), one for P(atient), one for AP, one for F(ocus), and perhaps a bunch of other cases, or perhaps you had the verb agree with P regardless of whether P was the subject or object, then that wouldn't be nominative-accusative. But the way Latejami works, how it passivizes, adjectivizes, etc, I don't consider it fluid-S. Another thing which makes me see it as nominative-accusative rather than fluid-S is that the argument structure morphology in Latejami is derivational rather than inflectional.