Re: Tagalog & trigger idea: I'd like comments. :)
|From:||H. S. Teoh <hsteoh@...>|
|Date:||Saturday, November 20, 2004, 0:33|
On Fri, Nov 19, 2004 at 06:42:10PM +0000, Ray Brown wrote:
> It is inaccurate to use the terms without defining its use. This IMO has
> been one source of confusion. In the conventional use of _topic ~ comment_
> (theme ~ rheme) and of _focus ~ presupposition_ (new info ~ given info) we
> are working at the discourse level. Using any of these terms to describe a
> syntactic feature is probably not a good idea.
> "Chang (1997:iv), on the other hand, uses the term 'voice'"
> H.S. Teoh
> >As for me, I see the choice between these two an arbitrary theoretical
> >choice. So far, they seem to be two sides of the same syntactic
> >operation to me: there is a particular NP we want to make the
> >"subject", and the way this is done is to (1) attach a particular
> >affix to the verb and (2) mark the NP with a particular marker.
> I know H.S. Teoh does not then go on to argue for 'voice' (tho I think his
> use of "subject" is significant) , but the quote seems to follow on from
> my observations in the section above, and to link to.....
After reading what you wrote subsequently, I think perhaps "voice" is
the best term for the distinguished NP in Tagalog. "Topic" or "focus"
suffers from cross-domain confusion (syntactic vs. semantic), and
"trigger" suffers from multiple interpretations (the NP triggers the
verb affix vs. the verb affix triggers the NP marking).
"Voice", OTOH, seems to parallel much more closely with what happens
in, say, English, when a particular NP is "promoted" to subjecthood.
Of course, the Tagalog situation is much more symmetrical in the sense
that no particular voice is preferred (such as the active voice being
preferred in English), and the prospective "subject" does not have
less marking than the other NPs. But they are, after all, different
languages, and perhaps we should not expect a perfect fit in all the
details of the system.
> The more I learn about this construction, the more I feel we are dealing
> with 'voice', with the one NP argument or another promoted to subject. The
> fact that English (and Latin & most european languages) have only active
> and passive voice does not, of course, mean that those are the only two.
> We know that ancient Greek had a third 'voice', known as 'middle'. Trask
> lists, besides active, middle & passive: "reflexive, causative, and
> adjutative, to name a few".[...]
I agree. If we drop the expectation that voice systems should have one
predominantly preferred voice, and that "active" and "passive" are a
necessary subset of the voices, then Tagalog fits the mold quite well.
It has been argued that Tagalog's failure to meet the first criteria
excludes the possibility of a voice system; but now I'm beginning to
question the validity of this argument. The expectation of a
predominant voice seems to me to be a matter of preference, orthogonal
to whether something is a voice system. After all, at the grammatical
level, both active and passive are equally valid voices, even if the
speakers of a language choose to use one of them more often.
If we're willing to extend the definition of 'voice' to not require
active and passive, and accept that a system with agentive,
patientive, instrumentative, and benefactive verb markings equally
qualifies as a voice system, then the behaviour of Tagalog's so-called
"trigger system" very much matches the description of a voice system.
In fact, the correspondence might be closer than we think. Let's
consider a concrete example in English:
John gives a book to Mary
John-SUBJ give-ACTV book-OBJ Mary-DAT
Now let's promote "book" to subjecthood:
A book is given to Mary by John
book-SUBJ give-PASV1 Mary-DAT John-AGT
And if we use "Mary" as subject:
Mary is given a book by John
Mary-SUBJ give-PASV2 book-OBJ John-AGT
Now, it happens that in English, PASV1 and PASV2 are the same. But it
is not hard to see that PASV2 is actually functioning as a third
voice, since the subject of PASV1 is the thing which is given, whereas
the subject of PASV2 is the person to whom it is given.
Notice how the ACTV voice on the verb corresponds with the AGT role of
"John", PASV1 corresponds with OBJ, and PASV2 corresponds with DAT.
Furthermore, notice how "Mary" is always marked with DAT until she
becomes the subject of the sentence, and "book" is always marked OBJ
until it is in subject position, and "John" likewise is always marked
AGT except when he is the subject.
If we now rename SUBJ to TRG (the conventional way of representing the
Tagalog "trigger NP") and substitute ACTV, PASV1, PASV2 for the roles
they correspond with (AGT, OBJ, DAT, respectively), we see that it is
one and the same as the Tagalog system, save for word order:
John-TRG give-AGT book-OBJ Mary-DAT
book-TRG give-OBJ Mary-DAT John-AGT
Mary-TRG give-DAT book-OBJ John-AGT
I think I've convinced myself that what we're looking at is really
just a voice system.
> It is perhaps appropriate at this point to reply to H.S. Teoh's questions:
> >Out of curiosity (and a desire to truly understand), what is the
> >precise definition of "subject"? What properties must an NP satisfy in
> >order to be correctly called a subject?
> According to Trask, "Subjects most typically exhibit a large number of
> grammatical, semantic and discourse properties". The properties Trask
> chose to list are:
Thanks for the info. This is very helpful.
Now as far as the fronted NP in Tatari Faran is concerned:
> - it represents an entity with independent existence;
I'm not sure what exactly does "independent existence" mean?
> - it controls co-reference, including reflexives, pronouns and null
Co-reference, definitely. I would expect null anaphors as well, once I
get to those parts of the grammar.
> - it it controls switch-reference systems;
Haven't worked out that part of the grammar yet.
> - it controls verb agreement;
Tatari Faran lacks verb agreement.
> - it is the topic in an unmarked sentence;
What is an "unmarked" sentence?
> - it is the target of advancement processes;
> - it can be relativized, questioned and clefted;
Interesting. In Tatari Faran, if an NP is being questioned, it must
always be the fronted NP:
diru kita bata' na tsana? - is it the girl who spoke to the chief?
bata' na ta diru kei tsana? - is it the chief to whom the girl spoke?
/ta/ here is the interrogative marker. It usually follows the verb:
diru kei bata' na tsana ta? - did the girl speak to the chief?
> - it undergoes raising;
I haven't worked out enough grammar to address this. But it is likely
to be so.
> - it receives minimal case marking;
Not really true in TF; the least marked case is the absolutive, but
the absolutive is usually not used when a verb is present.
> - it is an agent in an unmarked sentence.
> It is not expected, of course, that the subject in any language will
> exhibit all of these properties; and, as Trask observes, few of these
> properties may be unique to subjects in any given language. It would,
> however, be expected that subject exhibits more of these properties than
> other NPs in a given language.
I guess this does apply to the fronted NP in Tatari Faran, then. Among
the features that distinguish it from the other NPs are:
- it controls co-reference, and possibly null anaphors as well
- it probably will control switch reference
- it probably can undergo raising
- it can be questioned (when an NP is questioned, it must be the
There are several other candidate features, but I'm unsure about the
> >Would it be valid to say that the fronted NP in Tatari Faran is a subject?
> >Or should I perhaps coin a totally new term for it?
> I do not know enough about Tatari Faran - but I hope the above helps[...]
Indeed it does. I am now inclined to call it the "subject" unless I
discover evidence to the contrary.
And now, for Tim's post, which I found very helpful especially in
bringing out the fact that Tagalog's actor NP appears to have
subject-like properties as well. (I'm replying to both messages 'cos
I'm nearly past my daily posting limit. Sighh...)
On Fri, Nov 19, 2004 at 10:43:41PM +0000, Tim May wrote:
> I agree that these differences from the voice systems of more familiar
> languages do not necessarily mean that the "trigger" or "focus" system
> should not be described as a form of voice. Indeed, it is often
> described as "symmetrical voice" - see e.g. Foley.
I agree. "Symmetrical voice" makes sense.
> "Subject" is a little tricky. For example, Himmelmann uses the term,
> but is careful to qualify it, as in the following (a footnote from
> this paper ):
> | As is well-known, considerable controversy surrounds the question of
> | whether the grammatical relation subject exists in Tagalog. Following
> | Schachter (1976), it is generally agreed that ang-phrases in
> | post-predicate position show many but not all of the presumably
> | universal subject properties proposed by Keenan (1976). Still, as
IMHO, one should consider revising the set of "universal" subject
properties, since it is possible that when Keenan formulated them, the
languages he had in mind happened to share some subject properties
that Tagalog doesn't have. My view is that if the prospective Tagalog
"subject" satisfies a sufficient number of these properties, perhaps
it *should* be called a subject.
> | while other subject diagnostics are inapplicable or inconclusive. The
> | major point of contention pertains to the so-called agent-related
> | properties of subjects, in particular the properties of serving as the
> | antecedent in reflexive constructions, the target in Equi-NP deletions
> | and the addressee in imperatives.
Now, these "agent-related properties" I consider to be somewhat on
shaky ground. Correct me if I'm wrong, but the idea of "agent" seems
to me to be a semantic concept, whereas "subject" merely plays a
syntactical role. The fact that the two happen to coincide in the
majority of the languages that we (or the linguists) are familiar with
doesn't necessarily mean that we should conflate the two.
> | for grammatical relations. Artawa & Blake (1997:505f), among others,
> | profess serious doubts in this regard and argue for the viability of
> | the subject notion in Balinese, a language for which the basic facts
> | relevant to this issue are quite similar to the Tagalog ones. Here I
> | adopt the position that there are subjects in Tagalog, with the
> | proviso that the subject in Tagalog differs in some regards from
> | subjects in other languages such as English.
I agree that one should accept that Tagalog subjects should behave
slightly differently from English subjects.
> Manning on these supposed difficulties with identifying a subject
> in Tagalog:
> | Schachter (1977) points out that Tagalog has a split in apparent
> | `subject properties' (in roughly the sense of Keenan (1976))
> | between those borne by the ang-marked NP and those borne by what he
> | calls the Actor - the A or S NP. See (9).
> | (9) Ang-marked NP Actor
> | Obligatory element of every clause Reflexive binding
> | Launches floating quantifiers Equi target
> | Relativization Imperative addressee
Now, this is very interesting. I did not know this before. So the
actor NP in Tagalog is more distinguished than the other NPs beside
the prospective "subject" (i.e. the ang-marked NP). What does "equi
I consider that the actor NP binding reflexives and serving as the
addressee of imperatives is a result of the semantics of the actor
role, rather than pertaining to the syntactic function of subjects.
Intuitively, it makes sense for reflexives to bind to the actor - if
the subject were a locative, for example, it would be of little use
for reflexives to bind to it. It would be much more useful in the
usual case to have reflexives refer to the actor, which I presume
would take on the agentive role. Ditto for imperatives.
> | Because Keenan's criteria do not consistently pick out a notion of
> | subject in Tagalog, Schachter concludes that the ang-marked NP is the
> | Topic, but that various other properties key off the macrorole of
> | Actor, and that Subject isn't a useful notion in the description of
> | Tagalog.
I disagree. The criteria which are inconsistent with the ang-marked
NP, if I read the footnote correctly, are the very criteria that
describe the agentive (i.e. semantic) aspects of the subject in
languages like English. If we consider only the purely syntactic
criteria from Keenan's list, I suspect we'll find a much closer match
with the ang-marked NP.
> | However, he is careful to point out that these Philippinist
> | conceptions of Topic and Actor are somewhat at variance with normal
> | usage (the Topic can be what would normally be called a focus, for
> | example, and the Actor can have various thematic roles in the context
> | of Tagalog verbs meaning roughly `receive' and `endure'). But the
> | Topic has reference-related prominence while the Actor has
> | role-related prominence, serving as the protagonist.
This I find to be the key to the problem: the ang-marked NP plays the
syntactic role of subject, while the Actor plays the semantic role of
agent. In English, the subject happens to coincide with the agent; in
Tagalog they are independent. This reinforces my conclusion that the
problems arise from attempting to conflate the two.
> | Another key property of the Topic is that it is the only position
> | that can be relativized on.[...]
Again, this emphasizes the *syntactic* function of the "Topic".
> | On the other hand, Schachter shows that the Actor can always
> | control a reflexive (regardless of whether it is the Topic) - see
> | (12a-b), while it cannot itself be a reflexive (12c):
This sounds like a restriction arising from *semantic* considerations.
> | Also, in the basic pattern of control, it is always the actor that
> | is the gapped controllee, regardless of the verbal voice of the
> | complement. For example, (13a) shows a topic actor controllee,
> | while (13b) shows a non-topic actor controllee.[...]
> | Thus the kind of `subject properties' that Keenan (1976b)
> | identifies are split between two NPs in Tagalog sentences (except
> | when the Actor and the Topic coincide) and it is not immediately
> | obvious to which of these we should apply the term `subject'.
I think it becomes obvious if we refine our definition of 'subject' to
cover only syntactic functions, and not conflate it with semantic
This post mentions how some linguists try to apply traditional (i.e.
"Standard Average European", SAE) terms to Tagalog, while others go
ahead and coin new terminology (redefining 'topic', 'focus', etc.) to
describe aspects of it that does not fit into the traditional SAE
To which I would say, if we are going to apply SAE terminology to
Tagalog, we should be prepared to broaden the definitions of some of
this terminology, such as "subject" as we've seen. Since the terms
were originally defined based on the grammar of SAE languages, one can
hardly expect them to work well without revision for languages which
are not SAE. Perhaps the terminology would become more useful if
refined in this way - e.g., if the definition of 'subject' is revised
to refer only to the syntactic construct, and another term, perhaps
'agent', is used to refer only to the semantic role, then we could
describe language with more precision.
As for the other camp: if we are going to invent new terminology, we
should at least not overload existing terminology in incompatible and
contradictory ways, such as the use of "topic" or "focus" for a
syntactic construct, when the terms refer to the semantic domain when
applied to SAE languages. This has only caused unnecessary confusion,
as Ray can testify. :-)
My personal stance after going through this thread is that the Tagalog
system isn't as strange as it has been made out to be. It looks like a
voice system where the syntactic function of the "subject" has been
divorced from the semantic function of the agent/actor. To me, this
seems perfectly logical. Even though in SAE the subject is equated
with the agent, there is no reason why this continues to hold in
non-SAE languages. Why not just accept that they are different, and
that Tagalog has merely taken the liberty to treat them differently?
Computers aren't intelligent; they only think they are.