C) Generative computer pictures
5 Case Studies: Using the Data Type »Image«
5.4 Another Border Line Case: Mental Images
5.4.1 An Example Task: Understanding Reports From Absent Spatial Events A typical example from our ordinary life is the task of a radio sports reporter: apart from
emotional effects, which we shall ignore in the following, he has to give to his audience a (more or less objective) verbal description of the development of spatiotemporal configurations his audience cannot perceive by themselves. The reporter’s behavior is often explained by means of reference semantics: the meaning of the utterances forming his report is understood as being anchored in his (visual) perceptions, as introduced in section 4.3.1.
While examining from the perspective of computational pragmatics the verbal activ- ity of a radio sports reporter who describes objectively what he sees happening on, e.g., a soccer field, our focus of attention is directed essentially to the following three general problems: the speaker should be sure that any assertion of his description can be under- stood in its particular context by the listeners assumed with respect to reference, plausi- bility, and adequacy:
Reference: first, a listener should be able to correctly and uniquely identify those ob- jects in the common discourse universe that are used by the speaker to anchor con- textually the assertion. Ambiguities in the literal meaning of definite noun phrases and the reference of corresponding pro-forms must be resolvable. For example, the correct use of under-specific definite descriptions, like ‘the penalty area’, or ‘the defender’, is to be controlled by the speaker’s anticipation of the listener’s under- standing.
Plausibility: even if the listener is able to anchor the utterance correctly in the con- text, she may fail to understand the assertion since the new information communi- cated is not plausible for her in the contextual situation. Since the new information essentially transforms or further restricts the context of the assertion in question, such a rejection due to lacking plausibility may occur if the additional restrictions are incompatible with the given context. The speaker has to anticipate whether the listeners are able to integrate the meaning of a continuation of the description presently planned into the understanding assumed so far. On the verbal surface, this shows essentially in what has been recognized but is not said.
Adequacy: finally, under the assumption that the assertion communicated also is plausible for her, the listener may draw implications that the speaker does not want her to draw. In the case of an objective description, the question is whether the listener’s conclusions are adequate with respect to the events observed by the speaker. In particular, it is an interesting task for the speaker’s anticipation of the listener to initiate – under the general restriction of economy [GRICE1974] – addi-
tional information only in cases where it is necessary to keep the listener’s under- standing adequate: on the verbal surface, such additions may be found in gram- matically optional, locative expressions like ‘she receives the ball at the left pen- alty spot’; here, again, some further consequences appear in what is left out in the description actually produced.
In order to explain how a listener understands the report grounded in the visual per- ception of the speaker, the listener is usually assumed to have constructed a visual men-
tal model – a “mental image” – which substitutes percepts of scenes not perceptually present. “The radio reporter has solved his task only if he describes the reality of a sports event so vividly and obviously to the listener that the listener believes she sees that reality” a German linguist wrote [DANKERT 1969, 94]. The essential claim is such
that the spatial implications the listener is able to draw from the reporter’s descriptions can simply be “seen” in those visual pseudo-percepts: the listener would be able to “see in her mind’s eye” that a certain player stands to the left of the opponent penalty area af- ter being merely told that that player is beside the penalty area. That particular under- standing is assumed to be the only consistent way for her to continue the contextual mental image. In doing so, she resolves ambiguities both included in the meaning of the preposition used and in the reference of the noun phrase, but without using spatial rea- soning in the explicit way described in Table 2 of section 4.3.1.3.
It should be rather clear that such a conception opens a way to solve the problem of integrating the need for referentially anchoring semantics with the idea of partner mod- eling: first, the mental image would allow the listeners to anchor the speaker's utter- ances referentially in analogy to the speaker himself. Although mental images are not precisely visual percepts, they are conceived of as being very close relatives that can be used as substitutes. Thus, we could assume the very same kind of semantics to be used both by the speaker and his audience. The listener model of the speaker correspondingly has to deal with mental images, as well. The speaker, then, is thought of as taking into account the mental image his listeners are able to construct in accord to his utterances: if this mental image does not fit to his communicative intentions, he has to change his ut- terance plan accordingly (Fig. 121). Before presenting the computational example, the function of mental images has to be elaborated a bit further.