Re: Trigger language?
|From:||Christophe Grandsire <christophe.grandsire@...>|
|Date:||Thursday, January 23, 2003, 22:23|
En réponse à vaksje <vaksje@...>:
> Perhaps I was thinking of agent and patient instead.
Those are semantic roles, and as such exist in any language :) .
> object just a way to make these clear?
Nope, they are just a way to make the whole thing confusing! :)) Subject and
object are entities defined and working properly only with nominative-
accusative languages, i.e. the majority of European languages. They don't work
anywhere else that well (even in a nominative accusative language like
Japanese, they don't work that well. Look at a sentence like "watashi wa anata
ga suki da", meaning "I like you". Well, what is translated as a subject is
actually a topic - particle "wa" - while what is translated as an object is
actually a subject in the original sentence!!! - particle "ga" -). And they
don't always mark agent and patient. In the sentence "I'm given the book", the
object is indeed the patient but the subject is no agent at all!!! And a
sentence like "I see her" contains neither an agent nor a patient, but an
experiencer and an experiencee.
How are they marked in trigger
> (isn't this simply marking the topic's function on the verb?)
Yep, but that's all there is. The topic function is a semantic
feature. "Subject" and "object" are syntactic ones, not to be confused. My
Itakian doesn't have subjects (it only has triggers) but in imperfective
sentences it always has a direct object, and that's only a definition of
syntax, telling about how it looks but not what its *function* is.
> languages then?
Strict ergative languages like Euskara can accept a definition of subject and
object, but those are pretty useless since if we are to accept them the subject
of an intransitive sentence is marked as the *object* of a transitive one.
Subject and object are then useful to explain people knowing only accusative
languages how an ergative language works, but that's where their use stops. For
anything else, an analysis directly using the syntactic cases, the semantic
roles or the division into topic and comment are the best things to do.
> I see. :)
Well, let me define topic and comment: the topic is "what we are talking about"
(and thus can be omitted, but is always present at least in context) and the
comment is "what we're saying about the topic" (and is thus mandatory in a
sentence because it's always the reason why we are uttering this sentence, to
comment on something ;)) ). Any utterance is analysable into topics and
comments, and as such are universal (what's interesting is to see how languages
treat those, grammatically like in Japanese with the topic marker "wa", or
contextually like European languages, which resort to optional or indirect
things like reordering the sentence, using passive voice, or a construction
like "as for").
Take your life as a movie: do not let anybody else play the leading role.