La ventana se abrió (the window opened)

grammars SRG
#1

The Spanish se will undoubtedly reveal a host of complex problems in the SRG. It is a clitic with many, many functions.

Let’s start with La ventana se abrió which is the Spanish version of `The window opened’.

I noticed that the item that is marked as correct in the Spanish MRS testsuite (which is where the example comes from) is clearly not correct, and in fact I think I can find somewhat better readings among the ones available. But I am not sure the ones I find are good enough still.

I filed an issue on SRG’s github: Middle voice "se" not working properly ("The window opened") · Issue #5 · delph-in/srg · GitHub
Would someone have an opinion on this? Is it OK to have an ARG2 in the MRS for La ventana se abrió or should I work towards not having it there at all? The mrs is given below:

Screen Shot 2023-03-07 at 1.13.15 PM
Screen Shot 2023-03-07 at 1.13.15 PM1388×596 99.1 KB

I think maybe it is fine to have it there so long as it is linked to the ref of window (which it is not, currently). But the English version doesn’t have an ARG2 there at all.

Sorry for cross-posting. I think we can discuss it here and I will just link to this discussion in the GitHub issue.

#2

I think that depends on whether we think mediopassive is a voice alternation (like passive), where the semantic roles don’t change, or something more like the causative/inchoative alternation, where they do.

#3

Alright!

I have found this comment in the grammar:

; — transitive verbs that participate in ‘causative alternation’
; May take adjunts like "él solo, por sí mismo, por sí solo (by itself,…) (e.g. las puertas se cerraron
; ellas solas), but can’t be modified by agenitive adverbs/complements (*la puerta se abrió
; intencionadamente/por el portero) nor by final clauses (*la puerta se abrió para airear la habitación))

This indicates the ARG1-only analysis, do I understand right?

(By the way, the documentation in this grammar is beautiful; I hope I don’t ruin it too much.)

#4

Then the interesting constraint is that we can’t have processes that remove semantic information. So either you have two entries for the verb, or you have a lexical process that adds causative when se isn’t there.

1 Like