Establishing and Accessing Referents in Discourse

Establishing and Accessing Referents in Discourse

Memory & Cognition 1984, 12 (5), 489-497 Establishing and accessing referents in discourse GREGORY L. MURPHY Brown University, Providence, Rhode Island Four experiments explored the processes by which people understand references. Subjects read naturalistic stories one sentence at a time, and their reading-comprehension times were record­ ed. The first two experiments investigated differences between definite and indefinite reference. Their results suggest that establishing a new discourse referent (as indicated by an indefinite article or pronoun) produces longer comprehension times than referring to an already existing referent. The last two experiments investigated reference to one versus two discourse referents. They showed that accessing more referents led to longer reading times even when the referring sentences were identical. The results are explained in terms of a discourse model approach to comprehension. According to this view, readers construct a mental model of the discourse do­ main, and the ease of comprehension is related (in part) to the number of elements in the model that are accessed and the operations performed on them. Reference is one of the primary functions of commu­ and successfully refer to the desired person, even when nication. From the prelinguistic pointing of young chil­ the man is not wearing a cheap suit (perhaps it just looks dren to the use of elaborate descriptions in fluent adults, cheap, or perhaps it is not really a suit). Furthermore, people find it necessary to pick out individual entities or the reference may be successful even if the speaker and groups of entities. It is no surprise, then, that linguists, the hearer both know that the description is false (Clark philosophers, and psychologists have devoted much ef­ & Marshall, 1981). These cases can be handled by the fort to devising an account of reference. This task is a simple fact that people's discourse models can include ob­ difficult one because it requires (eventually) an explana­ jects and information that do not correspond to objects tion of how language corresponds to the world, and that and facts of the real world. Ifthe speaker and listener have correspondence is anything but straightforward. For psy­ similar models, then an incorrect description may still pick chologists, the problem ofunderstanding reference raises out one element sufficiently for comprehension (Clark & issues of memory representations, linguistic processes, Carlson, 1981; Clark & Marshall, 1981; Johnson-Laird perception, and reasoning. & Garnham, 1980). One recent approach to explaining reference has been Another useful result of considering understanding to the use of mental "discourse models." As a listener un­ involve such discourse models is that this approach may derstands some discourse, he or she builds a mental world help explain how people understand potentially ambigu­ that represents the information in the discourse, and refer­ ous, vague, or difficult texts (see Collins, Brown, & Lar­ ence is made by picking out objects in this model of the kin, 1980, and Johnson-Laird, 1981). Because the world. A reference to something that is not in the model discourse model is a sort of mental world, people can use will not be understood, or else will require work on the their real-world knowledge to organize and supplement part ofthe listener to alter the model. Johnson-Laird and it. The use of mental models in problem solving and Garnham (1980) demonstrated that this approach can ac­ reasoning was reviewed by Gentner and Stevens (1983). count not only for "normal" reference, but also for refer­ Much of the support for this approach to language com­ ence to nonexistent objects or to objects that are prehension comes from studies of ambiguous or misdescribed (cases that have posed serious problems for anomalous reference. The experiments described in this some past theories). article focus on simpler cases of "normal" reference. For example, someone might say at a crowded party, In particular, this approach suggests that an important factor in understanding is whether the referent is al­ (l) The man in the cheap suit is a gangster. ready included in the discourse model. In most cases, a definite reference will in fact pick out something already in the model-that is, something that has been mentioned I would like to gratefully acknowledge the advice of Herbert Clark in the discourse or that is present in the context (Clark on these experiments and the helpful comments of Richard Gerrig, Ray­ & Marshall, 1981). Usually, indefinite references in­ mond Gibbs, and Ed Wisniewski on an earlier draft of this paper. This troduce a new entity into the conversation. When the research was supported by a BioMedical Research Grant from Brown listener hears an indefinite article (or no article for plural University and by Grant MH-20021 to Herbert Clark. Requests for reprints may be addressed to Gregory L. Murphy, Hunter Laboratory nouns), he or she can guess that a new entity is being men­ of Psychology, Brown University, Providence, Rl 02912. tioned and that a new element in the discourse model must 489 Copyright 1984 Psychonomic Society, Inc. 490 MURPHY be created. All other things being equal, it should there­ easier and faster to do. In order to test this prediction, fore take longer for listeners to understand indefinite refer­ it is necessary to construct passages that either invoke a ences than to understand definite references. Experiments new element or refer to an old element, with the refer­ 1 and 2 investigated this claim. ring expressions carrying the same amount of informa­ Another implication of this approach is that reference tion. That is, it is important not to confound the to more than one object in the discourse model should complexity ofthe expression or its meaning with the fac­ take longer than reference to a single object. Since the tor of interest. Experiment 1 therefore compared noun entities in the discourse model are in fact separate "ob­ phrases in identical contexts that differed only in the use jects," picking out two or more of them should be more of the definite or indefinite article. time consuming than picking out one, even when the mul­ Experiment 1 used the reading-time paradigm, in which tiple objects are described by a single noun. This hypothe­ subjects read stories, a sentence at a time, and pressed sis was elaborated and investigated in Experiments 3 a button as they understood each sentence. One story con­ and 4. tained the following sequence in which the article was Although the experiments described here were inspired varied for different subjects: by the discourse-model approach to comprehension, their primary goal was not to evaluate that approach as a whole, (2) Though driving 55, Steve was passed by a truck. but to investigate specific hypotheses about reference. Later, George was passed by a/the truck too. There is no single theory of the use of discourse models in understanding (which is why I have been calling it an Notice that the same actions and properties are predicated "approach," rather than a "theory"), and different in­ in both versions. Furthermore, both versions repeat the stantiations of the approach could be devised to make noun in the second sentence (i.e., although the indefinite different predictions than those I give. These experiments version introduces a new object, it does not introduce a can be thought of as providing data on which more detailed totally new concept). The only difference between the two accounts of understanding can be based in any theoreti­ is whether the actions and properties are predicated of a cal framework, although they are particularly relevant to new object or of one already mentioned, as signaled by the discourse-model approach. the article. There are a number ofalternative theories of reference An alternative hypothesis is that the definite article that may suggest other psychological explanations of how prompts the reader to search memory for the object be­ reference is understood. However, like the discourse­ ing referred to, whereas the indefinite article indicates that model approach, these theories usually do not provide the a new object is being mentioned and thus no such search necessary detail to make clear predictions for on-line is necessary (Irwin, Bock, & Stanovich, 1982). There­ measures of comprehension. Rather than describing al­ fore, indefinite reference should be easier to comprehend. ternative approaches here, I consider more specific al­ This hypothesis has some plausibility, and it may be true ternatives for the individual reference processes of the when the definite reference is only obscurely related to separate experiments. its referent. However, the main hypothesis assumes that fmding the antecedent of a simple definite reference should be easier than establishing a whole new referent in the EXPERIMENT 1 discourse model. Other investigators have looked at various effects of As Clark and Marshall (1981) showed in detail, under­ definite and indefinite articles, but apparently never in standing definite reference can often involve complex order to investigate instantiation of referents. A number reasoning about possible referents based on the hearer's of experiments have suggested that memory for text is knowledge ofthe domain of discourse and of the speaker organized around discourse referents, rather than (see also Clark & Murphy, 1982). Definite reference can­ individual sentence meanings (Bransford, Barclay, & not be interpreted through syntactic rules, although there Franks, 1972; de Villiers, 1974; Ehrlich & Johnson­ are some syntactic constraints on reference (see, e.g., Las­ Laird, 1982; Garnham, 1981), but they do not address nik, 1976). Furthermore, it cannot be interpreted solely the questions of understanding processes raised here. through semantic analysis of the referring expression. As More recently, Irwin et al. (1982) showed that the ar­ mentioned above, the noun phrases used to refer are often ticle preceding a noun influenced the time it took to per­ vague, ambiguous, or even incorrect from an "objective" form a lexical decision on the noun, in particular, that viewpoint [as in (1)].

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    9 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us