Referential form as instructions to the comprehender

gender and number. The claim is that the more specific a referential form is, the greater will its positive effect be on the activation of its referent and the greater its negative effect be on all other referents.

3.4.1 The environmental view versus speaker control

Three of the current models of referential management treat the choice of referential form as one solely based on the preceding environment. This environmentally conditioned choice of referential form is deterministic. It searches for necessary and sufficient conditions that require a given form. It asks the question, “What has occurred so far in the discourse or in the real world that causes a particu- lar choice of reference form?” The asking of this question is particularly problematic because of the as- sumptions it makes. It assumes that only what has occurred so far in the discourse or within the knowledge of the speech act participants has any bearing on the referential choice. It ignores any view of a discourse as an unfolding mutually negotiated dialogue. Rather than asking what has happened that forces this form, it would be better to ask “What has occurred so far in the discourse or in the real world that allows a particular choice of reference form?” Communicators have certain goals and ideas, and they structure their communication to reach them. The structures that they build are not arbitrary nor are they required by what has previously oc- curred in the communication. In Clancy’s analysis of the Pear Films 1980, she found that individual speakers displayed a wide range of latitude in their choice of referential form. In one episode one es- tablished character meets three others. The single character is in his last episode, and the other charac- ters will continue on until the end of the film. At this point different speakers made different choices about using pronouns to mark prominence and point of view. Some opted for the single character and others opted for the new characters. Those opting to adopt the viewpoint of the new characters could not base their decision on past importance, but had to be building for future importance. This same ef- fect is reported by Perrin 1978 in Mambila. There three referents had been introduced as secondary participants, and then one was chosen as the primary participant and referred to with a zero. This was not because of past relevance, but rather showed the importance the participant was to have in the rest of the story. Payne 1993 has shown that in Yagua, references that do not include a verbal affix persist half as long as those where the equivalent device is used in conjunction with a verbal affix. The choice then becomes based not on activation to this point, but rather intended persistence from this point on- ward in the texts.

3.4.2 Referential form as instructions to the comprehender

Rather than taking a view of reference as a search for necessary and sufficient conditions that need to be fulfilled so an automatic choice can be made, I believe reference form can best be analyzed as in- structions to the comprehender. The idea that referential form is an instruction is not new. Various au- thors have looked at the use of reference form as an instruction to a comprehender to delimit their search for a referent Clark and Sengul 1979, Du Bois 1980a, Givón 1990, Hinds 1977. Referential form as a search instruction cannot be the whole story. It does not account for the prominence effects that Clancy 1980 found in both English and Japanese nor what Perrin 1978 found in Mambila. Nor does it answer the basic question of why we would develop such a complicated system of search instructions. Given that language producers have goals and motivation, and that they want to communicate with their audience, it would not be surprising for them to help comprehenders to keep track of the various referents. The choice of referential form is used not only in allowing identification to be made of a given object in the real world or world of discourse, but also to tell how important each participant is in the story, and what role they will play. This identification of the importance of referents could be done on an individual episode by episode basis, as well as over a whole narrative. The precise definition of the signals have not, as yet, been totally formulated. We can use Givón’s topic continuity hierarchy 1983b to help establish what is signaled, as in table 3.1. 3.4 The cognitive effects of reference forms 57 Table 3.1. Topic continuity hierarchy Most continuousaccessible least surprising zero anaphor unstressedclitic pronouns stressedindependent pronouns unmodified definite NPs restrictively modified definite NPs referential indefinite NPs Least continuousaccessible most surprising If this is combined with Gernsbacher 1989, 1990 and Chang’s 1980 work, a likely pattern emerges. Zero pronouns show little ability to enhance activation of the referent or suppress activation of a nonreferent Corbett and Chang 1983. Pronouns do not increase the activation of their referent, but do suppress the activation of nonreferents by the end of the sentence. Names immediately increase the activation of their referent and suppress the activation of all other participants Gernsbacher 1990. Just looking at Gernsbacher’s activation data we can conclude that speakers can vary the level of activation of participants by the choice of referential forms. His data consisted primarily of isolated sentences with little or no discourse context. It is unclear at this stage what effect a surrounding con- text would have on actual activation levels, but it is clear that the choice of form resulted in a change of activation. By sentence end a participant which was referred to by name had the highest level of ac- tivation. If the participant was referred to only by a pronoun it will have the second highest activation level. When one participant is referenced by a pronoun any participant who was not referenced will have a lower activation level than the participant which was referenced. Finally, when a participant was referenced by a name, any participant which was not referenced is strongly suppressed, giving them the lowest activation. This is summarized in table 3.2. Table 3.2. Effects on activation of participants by different referential devices Referential device Effect on the referenced participant Effect on all other participants name maximum enhancement maximum suppression common noun medium enhancement medium suppression pronoun little enhancement medium and slower suppression zero no enhancement little suppression

3.4.3 Activation and search instructions