Re: Subject vs. object vs. agent
- On Mittwoch, April 16, 2003, at 08:50 Uhr, Ivan A Derzhanski wrote:
> case marking, which is drivenYes and no. I would distinguish 3 layers:
> by grammatical (formal) functions in those languages that have
> them, not by thematic (semantic) roles.
1) Formal layer. Case markings.
2) Syntactical layer. The syntax of the language (subject, object
etc.). Logical structure above formal encoding.
3) Thematic roles ((Proto-)agent and the like). Semantical aspect that
gives rise to certain syntactical structures.
It must have seemed that I was suggesting the use of agent = subject.
That would not make sense, I agree. However, I think that the subject
(or object) is not identical with any *one* case. Neither in e.g. an accusative
language or an ergative language. Distinction of layer 1 and 2 are
fundamental, I think. Mixing them obscures the nature of language
and leads to a lot of confusion. I think you'd agree with me there.
[Indeed. And although students of grammar do sometimes make that mistake,
I don't think that anyone involved in this discussion has done so. CFH]
Indeed, I now think you meant to caution me to not mix the terminology
of layer 2 and 3. I agree with you. The first sentence of my first post
had an unfortunate formulation in it. While I still think layer 3 is
essential for understanding case markings, the terminology must be
clear, of course.
Remains the fact that, a prepostion + (pro)noun may function as subject
or object, irrespective of its formal encoding.
[That depends on what you mean by "function": i.e., what might be called
"sentential" or (in my terminology) grammatical function, vs. your "logical"
The 3. layer would consider different aspects such as: is "Mary" in
_Mary has a child_ an agent ?
Lastly, to differentiate between the term "direct object" and
"accusative" makes sense (and that's why there *are* terms for it)
because they refer to different aspects ("layers") of a given language.
It does not obscure the relation at all, rather, it enables us to see
the relation at all. Just as differentiating between electrons and
protons doesn't obscure the relations to the atom. Indeed, only by
differentiating we can understand how they work.
- David Kiltz wrote:
>>> languages may vary as to how they encode a prepositionalIn Czech, I can think of one preposition that requires nominative (others
>>> participant of a noun phrase (Semitic would use the genitive)
>> Can you name any language(s) that would use the nominative (distinct
>> from accusative)? (Don't say `Esperanto': that would be cheating.)
> As soon as languages use prepositions like Indo-European and Semitic,
> no. Of course not, that's just the syntax of those languages. However,
> there is e.g. Korean: _tongmu-wa cengkwu-lul hata_ "Friend-with (subj.,
> nom.!) tennis-obj. marker plays" = He plays tennis with a friend
> (=subj., nom. case).
require various other cases) -- the Czech equivalent of "like" -- in
"he is like Peter", "Peter" would be in nominative.
But it perhaps may be argued that it is not a preposition, that
"Peter" is in fact a subject of an implicit subordinated clause: "He is
like Peter is". At least this seems to be the origin of this
- --- In firstname.lastname@example.org, David Kiltz <dkiltz@g...> wrote:
> Otherwise, you becomeIn general, maybe. But we are speaking of Quenya at a certain
> completely circular: Direct Object = Accusative, Accusative =
> Direct Object. This is false.
conceptual stage. I mentioned the Plotz letter already, let's quote
the relevant sentence: "The difference between nom. and acc. was
abandoned - it was adequately expressed by word order." (VT6:14).
Clearly, JRRT is speaking of the difference between (grammatical)
subject and direct object here, because that's what is expressed by
word order. In relation to a verb, only one noun or pronoun can be
the subject, all other are objects (is that really just my
definition?!). If you can't always distinguish between subject and
direct object by their agreement with the inflected verb, you have to
mark one of them: the subject (ergative languages) or the direct
object (accusative languages). It seems obvious that JRRT used "acc."
in the latter sense: a case, earlier marking a direct object.
Both case markings and prepositions can identify the role of indirect
objects. In the languages I mentioned, case marking is mostly
rudimentary (ok, there are remnants of accusative markers in some
German nouns... yet, and so?). You don't need prepositions for a
(grammatical) subject, unless it's an ergative language, and Quenya
This was not necessarily so in earlier conceptual phases of Quenya or
Qenya: the Secret Vice poems and the Bodleian Declension hint at a
special subjective case, cf. the thorough discussion in VT28 (showing
some ergative character of the language, transferred later to
Adunaic, as it seems).
- We have
1 the _logical_ opposition of _subject_ and _predicate_, the underlying concept and the statement made about it.
In "Eric kills Edgar" EITHER noun could be the subject, according to what is required by the syllogism:
(Eric as subject:)
"Eric kills Edgar"
"Whoever kills Edgar or captures him deserves a reward"
"Eric deserves a reward"
(Edgar as subject:)
"Eric kills Edgar"
"Whomever Eric kills, dies quickly and gets a decent burial"
"Edgar dies quickly and gets a decent burial"
2 the _grammatical_ opppositions:
main noun - attribute (the genitive or adjective or apposited noun constructed in congruence with noun)
_The car_ - _of Eric_
_The blue_ - _car_
_The Rolls Royce_ - _car_
(_Car_ main noun in all phrases)
subject - predicate (the verb constructed in congruence with the subject)
_Eric_ - _kills [Edgar]_
_Edgar_ - _is killed [by Eric]_
older terminology called both main noun and subject "suppositum"
both adjective, apposition and predicate "appositum"
and genitive "appositum ex obliquo", which term might make some sense for objects, adverbials of circumstance and agents too
subject - object (the opposite of the subject in an active construction, the noun that would have been subject of the opposite verb, same concept with opposite construction)
_Eric_ - _[kills] Edgar_
subject - agent (the opposite of the subject in a passive construction)
_Edgar_ - _[is killed] by Eric_
subject/predicate - adverbial of some circumstance
_Eric killed Edgar_ - _yesterday_
3 the _thematic_ or _semantic_ oppositions:
agent - patient (whatever thing is brought to being or modified in being by agent)
_Eric_ - _Edgar_
sense object - sensor (whoever or whatever animal senses the sense object)
advantage - the one to whose advantage it is - [benefactor]
disadvantage - etc
thing/action - circumstance thereof (time, place, et c...)
_Eric killed Edgar_ - _yesterday_
These distinctions and oppositions do NOT coincide perfectly, but rather overlap, and they overlap differently in different languages.
If you regard the ergative case in ergative languages as a special kind of nominative or if you regard it as an instrumental of agent of essentially passive constructed verbs, though logically the subject, is beside the point of this general outline of necessary grammatical distinctions and their partial overlapping with logic and semantics.
A consistent international teminology does not really exist.
In English a verb or a preposition takes an object. In other languaeges only verbs do. In German a _Präpositionsobjekt_ is the object of a verb, marked by a preposition rather than an oblique case in its pure form:
_Er wurde WEGEN DES TODSCHLAGS ZUR TODESSTRAFE verurtheilt_ (_He was sentenced FOR MANSLAUGHTER TO DEATH PENALTY_) is a forensic sentence with two _Präpositionsobjekte_ in a row.
Ordinarily the noun after a preposition is not called _Präpositionsobjekt_ but just a _Hauptwort mit einer Präposition_, _noun with a preposition_. Introducing the term _Objekt eiener Präposition_, _object of a preposition_ would cause confusion in German grammatical terminology, since it would make people call things _Präpositionsobjekt_ when all they meant was _object of a preposition_.
A subject however is in origin a logical term, meaning the underlying: what underlies any statement is what the statement is about. In logic, as distinct from grammar, there is no special formal requirement on cases or moods involved.
And in many languages that _usually_ comes out by what the verb conjugates in congruence with (finite moods with nominative, some infinite moods with some oblique cases). That is where the concept of grammatical subject comes in.
If you have an impersonal verb, you do not have any real subject, only objects. But you may have logical subjects: what would have been logically the subject of the finite verb if it had had a meaning requiring another thematic role for subject.
Lith. (I use sh as transcription of s-hachek):
_man reishkia atsakyti_ (_for-me is-necessary to-answer_) has no noun or pronoun in the nominative, only one in the dative, marking the one (in this case _me_) who has the disadvantage to need answering or to whom it is a necessary advantage to answer.
If one had had another verb, like simple _answer_, or modal _can answer_ one would logically have said:
_ash atsakau_ (_I answer_)
_ash galiu atsakyti_ (_I am able to answer_)
which makes _man_ the logical, but not grammatical, subject of
_man reishkia atsakyti_ (_for-me is-necessary to-answer_==_I must answer_)
Grammatically _man_ remains the dative object of _reishkia_.
And if you say:
_ash megstu kâva_ (_I like coffee_)
or (somewhat unidiomatically for such a material thing as drink)
_man patinka kavà_ (_me pleases coffee_)
the first person has the same thematic role which would logically have made him the subject in the second sentence if _patikti_ had meant something about the person feeling liking.
In Russian you tend to avoid passive verbs. Instead of saying the Russian equivalent of
_This gazette is much read_
(passive construction, where _this gazette_ is subject of _is much read_)
you would say
_This gazette much [they] read_
(active construction, where _this gazette_ is object of _[they] read much_)
But as in good logic, independently of the construction, _this gazette_ is what you are talking about and _[they] read it_ is what you are saying about it, you can call _this gazette_ - thematically the patient or sense object of the act of reading, grammatically the accusative object of the verb _[they] read_ - the logical subject of the sentence, since it would logically have been the subject if Russian had here used a verb with passive construction. Of course there would be this possibility of making grammatical subject coincide with logical subject to use a verb with passive construction, like a reflexive verb:
_This gazette reads-itself much_
but I think that might be somewhat less idiomatic. Not that I really know Russian, but I gathered so from reading a Russian grammar a few hours ago.
And when I said above that adverbials of circumstance - and concomitance with a certain other person is such a thing - would have been _apposita ex obliquo_ in the older grammatic terminology (say 14th C), I took my stand also against regarding them as expressable by a nominative as such, i.e. a noun which is the main noun of the noun phrase or the noun without a phrase which is the _suppositum_ of a verb, the subject of a sentence.
Even if the prepositions of Quenya were conjugated exactly as verbs, that would not be a clear case for regarding the pronominal endings as nominatives:
- A, because they do not correspond in sense to nominatives but rather to some general oblique case
- B, because we cannot infer from the fact that finite verbs are constructed in congruence with nominative nouns or separate pronouns that pronominal endings stand for pronouns in the nominative, etymologically, so the coincidence of (even active) conjugation of verbs with conjugation of prepositions would not mean the latter were prepositions with nominatives. In Turkish the conjugation of verbs coincides with the possessive pronominal endings, as far as I can recall my language typology textbooks. Anyone fluent in Turkish please correct me if I am wrong!
Hans Georg Lundahl
Gå före i kön och få din sajt värderad på nolltid med Yahoo! Express
[Non-text portions of this message have been removed]