Logical polysemy, event and argument structure of some German shooting-verbs1

base verb SCHIESSEN and a basic configuration, these rules profile further segments, until the entire causal chain (14) is profiled. Not only prefix derivation, but also zero-derivation occurs in this process (cf. Copestake / Briscoe, 1996:16, 18). In (24) the verb profile of configuration 3 is shifted to the right: (24) Das Gewehr schießt. [The rifle shoots.] We have seen that profiling bears directly on the argument structure of a verb, since, according to (13), the head participant of the profiled action chain surfaces as subject and the tail participant as direct object, while any intermediate participant is realized as an antecedent oblique. What is then the semantic import of profiling? We claim that only profiled segments of a verb's base are semantically necessary, in the sense of motivating implications from sentences containing the verb. On the contrary, unprofiled segments are not implied by the verb, but just expected. Consequently, there is no implication from (18) that a projectile was fired. The but-test (cf. Cruse 2000:56) shows that this is only an expected feature of the verb's meaning: (25) (a) Hans hat die Pistole abgeschossen, aber kein Projektil ist abgefeuert worden. [Hans shot off the pistol, but no projectile was fired.]5 (b) ?? Hans hat die Pistole abgeschossen, aber ein Projektil ist abgefeuert worden. [?? Hans shot off the pistol, but a projectile was fired.] According to Pustejovsky's typology of arguments of lexical items (1995:63-64), the arguments u (bullet) and w (head) of (26a) are default arguments of the verb schießen, since they do not need to occur necessarily in the syntax structure: 5 We assume for (25a) a situation, in which the pistol was loaded with a blank cartridge, but the speaker is unaware of this fact. (26) (a) Hans hat dem Politiker eine Kugel in den Kopf geschossen. [Hans shot a bullet into the politician's head.] (b) Hans hat dem Politiker in den Kopf geschossen. [Hans shot into the politician's head.] (c) Hans hat geschossen. [Hans shot.] (d) * Hans hat eine Kugel geschossen. [Hans shot a bullet.] How can one explain, however, that (26d) is not grammatical, while (26b) and (26c) are acceptable? We believe that one can overcome this paradox, if we do not classify the arguments of a verb according to their possibility of being omitted, but rather list the conditions on which they can be omitted. For example, the argument w of schießen (i.e. the target) may be only erased, if u (i.e. the projectile) has already been erased. Unfortunately, not all aspects of the analyzed shooting-verbs fit into the sketched model. We will refer only to three difficulties. First of all, rules (13) (i) (v) cannot predict sentence (27), because the tail of the verb profile (i.e. the arrow) is realized as an oblique instead of a direct object: (27) Er hat mit einem Pfeil auf den Polizisten geschossen. [He shot at the policeman with an arrow.] Secondly, it seems that sentence (28) cannot be analyzed according to the causal structure model, since the target (i.e. the wall), although it occurs in the action chain before the participant that surfaces as a direct object (i.e. the hole), is realized as a subsequent oblique, instead of an antecedent oblique as predicted by rule (13) (iv): (28) Er hat ein Loch in die Wand geschossen. [He shot a hole into the wall.] In connection with the difficulties of (27) and (28), it should be stressed, though, that these sentences probably instantiate very marked argument realizations, which apparently do not occur, for example, in French, Italian or Portuguese. Lastly, unaccusative schießen places a further problem. At first sight one can assume the configuration (29) for this reading: