Genitive arguments and semantic roles

Iver Larsen iver_larsen at
Sun Oct 28 17:41:57 EST 2001

> Perhaps (Iver?) one thing we need here is a definition of
> semantic roles and how they differ from semantic functions.

Well, I can at least try to clarify how I use the terms, although I cannot
speak for all schools of semantics and linguistics.

Let me take a clause and describe it in *grammatical terms* first:

In "Peter gave John the book"
the head of the clause or the predicate is "gave"
"Peter" functions as subject in the clause
"John" functions as object
"the book" functions as indirect object
In languages with grammatical case, "Peter" would be in the nominative, "the
book" in the accusative and "John" in the dative.

It is possible to make a passive transformation in two steps:
First, the agent can be cut off, and the indirect object is then made the
subject as in "John was given the book (by Peter)".
Second, this can be further "passivized" by cutting away the beneficiary,
and the object is then made the subject as in "The book was given (to John
by Peter)"
Notice that if any of the participants are cut off they can be put back by
use of a preposition.
Now, in Greek, it is apparently not possible to do the first kind of passive
transformation. An indirect object cannot be made subject. It remains in the
dative case as indirect object. However, the direct object can be made
subject when the agent is cut off.

In semantic terms the clause above represent an underlying proposition with
a trivalent semantic verb "give" as nucleus. When all the participants are
expressed, there are three roles involved. "Peter" has the role of agent,
"John" has the role of beneficiary and "the book" has the role of patient. I
would not use the term semantic function in any technical sense.
Nor would I use the term semantic passive. I see the active to passive
transformation as a grammatical rule that changes the verb and makes the
former object the new subject. In most languages that have passives, there
is an affix in the verb, but some languages also employ an auxiliary verb
like "be" in English.
What corresponds to the grammatical passive transformation in semantics is
that the agent role is drawn into the verb nucleus and made implicit. When
that happens, the verb is reduced in valency by one. If the verb has 3
valencies like "give" "place" etc. then only the patient and beneficiary
roles are kept. In English, either the patient or beneficiary may become the
subject, but in Greek only the patient may become subject. If the verb has
two valencies and the agent is drawn into the nucleus, only one valency
remains, usually the patient, and it is expressed as subject. If a verb has
only one valency to start off with - corresponding to an intransitive verb -
a passive transformation is not possible.
In semantics we also talk about lexical features of nouns. For instance,
"John" and "Peter" are +human while "the book" is +human. It is part of the
semantic property of the verb what kind of lexical items may fill the
participant slots. The semantic verb "give" above is different from the
semantic verb underlying a clause like "It gave him a headache" because the
subject "it" does not represent a +human agent. Rather, it represents
a -human cause. So, in semantics terms this sentence has a different
underlying structure: "it caused him to get a headache."

I hope I haven't caused you a headache,

Iver Larsen

More information about the B-Greek mailing list