FFF CONFERENCE CTF07

Robert D. Van Valin, Jr. - Argument linking in deverbal nominals

Deverbal nominals have linking properties which appear to be similar to those of the source verbs, but there are in fact differences which pose a challenge to theories of the syntax-semantics interface. Examples are given in (1) and (2).


(1)    a.     the arrest of Bill by FBI agents in New York City
         b.     the destruction of the city by the enemy
         c.     the investigation of the murders by Sherlock Holmes

   
(2)    a.    Tepehua (Watters 1988)
                ’iš-puš-ka    kafe    
                3GEN-pick-NMZ        coffee
                ‘the picking of coffee’


         b.     Georgian (Harris 1981)
                 c’erilis         dac’era    cems    mier
                 letter-GEN        writing    me.GEN    by
                 ‘the writing of the letter by me’

         c.    Hebrew (Berman 1978)
                bniyat         ha-báyit    al ydey    ha-soxnut
                building        the-house    by    the-agency
                ‘the building of the house by the agency’      

 
Deverbal nominals inherit the semantic structure of the source verb (Nunes 1993), e.g. [do´ (x, Ø)] CAUSE [BECOME destroyed´ (y)] ? destruction (x, y). The linking principles for them are related to the linking principles for verbal arguments, but with variations which follow from the differences between nominals and verbs. While there are action nominals like running and stative verbs like know, nominals are basically static and verbs non-static. Nunes (1993) argues that English deverbal nominals are inherently intransitive; they never take more than one direct coreN argument, which is realized by the of-marked NP in the examples in (1) and the genitive case NPs in (2). Nunes (1993) shows that a deverbal nominal takes an undergoer as its single macrorole if the verb from which it is derived contains a state predicate, otherwise actor. Thus only deverbal nominals from activity verbs will take an actor as their of-marked coreN argument, as in (3) and (4). (The subscripts ‘CL-A’ and ‘CL-U’ stand for ‘clausal actor’ and ‘clausal undergoer’, respectively.)
   (3)    Deverbal nominals from achievement and accomplishment verbs (±causative)


    a.    The balloon popped.                    Achievement
    a´.    the popping of the balloonCL-U
    b.    The cat popped the balloon.       Causative achievement
    b´.    the popping of the balloonCL-U by the catCL-A
    b´´.    *the popping of the catCL-A    [must be interpreted as undergoer]
    c.    Chris died.    Accomplishment
    c´.    the death of ChrisCL-U    
    d.    The enemy destroyed the city.    Causative accomplishment
    d´.    the destruction of the cityCL-U by the enemyCL-A
    d´´.    *the destruction of the enemyCL-A     [must be interpreted as undergoer]
   
   (4)    Deverbal nominals from activity verbs
    a.     The dog barked.
    b.    the barking of the dogCL-A

These differences will be explored for a wider variety of languages, including German, Georgian, and Hebrew, in order to investigate this aspect of the syntax-semantics interface. Central questions will be, do the same constraints on transitivity and macrorole selection occur in other languages? How much language-specific variation is there in the linking principles across languages?