Discourse Processing

We have focused mainly on the processes involved in understanding individual sentences. However, in real life we are generally presented with connected discourse (written text or speech). According to Graesser, Millis, and Zwaan (1997, p. 164), there are important differences between the processing of sentences and discourse:

Connected discourse is. much more than a sequence of individual sentences. a sentence out of context is nearly always ambiguous, whereas a sentence in a discourse context is rarely ambiguous. Both stories and everyday experiences include people performing actions in pursuit of goals, events that present obstacles to these goals, conflicts between people, and emotional reactions.

Most research on discourse comprehension has been based on written texts. Some researchers have used published texts (e.g., articles; books) written by professional writers, whereas others have used specially constructed texts. The former approach has the advantage of ecological validity (applicability to real life), but poor control over many variables affecting comprehension. In contrast, the latter approach has the advantage that textual variables can be manipulated in a systematic way, but the resulting texts tend to be artificial and uninteresting. How should researchers proceed? The best solution was proposed by Graesser et al. (1997, p. 166): "Discourse psychologists are on solid footing when a hypothesis is confirmed in a sample of naturalistic texts in addition to properly controlled textoids [experimenter-generated texts]."

Inference drawing

Comprehension of discourse would be impossible without access to stored knowledge. A simple illustration of the crucial role played by such knowledge is the process of inference or filling-in of gaps. Schank (1976, p. 168) described it as "the core of the understanding process".

Some idea of how readily we make inferences can be formed if you read the following story taken from Rumelhart and Ortony (1977):

1. Mary heard the ice-cream van coming.

2. She remembered the pocket money.

3. She rushed into the house.

You probably made various assumptions or inferences while reading the story. Possible inferences include the following: Mary wanted to buy some ice-cream; buying ice-cream costs money; Mary had some pocket money in the house; and Mary had only a limited amount of time to get hold of some money before the icecream van arrived. None of these assumptions is explicitly stated in the three sentences that were presented. It is so natural for us to draw inferences to help understanding that we are often unaware that we are doing

A distinction can be drawn between bridging inferences and elaborative inferences. Bridging inferences need to be made to establish coherence between the current part of the text and the preceding text, whereas elaborative inferences serve to embellish or add details to the text. Most theorists accept that readers generally draw bridging inferences, which are essential for understanding. What is more controversial is the extent to which non-essential or elaborative inferences are drawn.


Perhaps the simplest form of bridging inference is involved in anaphora, in which a pronoun or noun has to be identified with a previously mentioned noun or noun phrase (e.g., "Fred sold John his lawn mower, and then he sold him his garden hose". It requires a bridging inference to realise that "he" refers to Fred rather than to John. How do people make the appropriate anaphoric inference? Sometimes gender makes the task very easy (e.g., "Juliet sold John her lawn mower, and then she sold him her garden hose"), and sometimes the number of the noun provides a useful cue (e.g., "Juliet and her friends sold John their lawn mower, and then they sold him their garden hose").

It seems reasonable that the ease of establishing the appropriate anaphoric inference should depend on the distance between the pronoun and the noun to which it refers: this is the distance effect. However, Clifton and Ferreira (1987) showed that distance is not always important. Their participants presented themselves with passages one phrase at a time, and the reading time for the phrase containing the pronoun was measured. The reading time was fast if the relevant noun was still the topic of discourse, but it was slow otherwise. Distance as such had no effect on reading time. Thus, the distance effect is found normally because greater distance reduces the probability that the noun to which it refers is still the topic of discourse when the pronoun is presented.

When are inferences drawn?

Consider the following passage from a study by O'Brien, Shank, Myers, and Rayner (1988):

All the mugger wanted was to steal the woman's money. But when she screamed, he stabbed her with his weapon in an attempt to quiet her. He looked to see if anyone had seen him. He threw the knife into the bushes, took her money, and ran away.

O'Brien et al. (1988) were interested in seeing when readers drew the inference that the "weapon" referred to in the second sentence was in fact a knife. They compared reading time on the last sentence in the passage quoted here, and in an almost identical passage in which the word "weapon" was replaced by "knife". There was no difference in the reading time, suggesting that the inference that the weapon was a knife had been drawn immediately by readers.

O'Brien et al. (1988) also considered reading time for the last sentence when the second sentence was altered so that the inference that the weapon was a knife was less clear ("But when she screamed, he assaulted her with his weapon in an attempt to quiet her"). This time, the last sentence took longer to read, presumably because the inference that the weapon was a knife was drawn only while the last sentence was being read.

Other evidence confirms the notion that it is only strong and obvious inferences that are drawn immediately. Singer (1979) asked participants to read pairs of sentences. In some cases, the subject noun of the second sentence had been explicitly mentioned in the first sentence (e.g., "The boy cleared the snow with a shovel. The shovel was heavy"). In other cases, the subject noun had not been specifically referred to before (e.g., "The boy cleared the snow from the stairs. The shovel was heavy."). Singer (1979) found that the time taken to read the second sentence in the pair was greater when the subject noun of the sentence had not been explicitly mentioned before. This suggests that the inference that a shovel was used to clear the snow was not drawn while the first sentence was being read, but was drawn subsequently.

Which inferences are drawn?

Everyone agrees that various inferences are made while people are reading text or listening to speech. What is of interest theoretically is to understand why inferences are made, and to be able to predict which inferences are likely to be made. The constructionist approach originally proposed by Bransford (e.g., Bransford, Barclay, & Franks, 1972) and later developed by others (e.g., Johnson-Laird, 1980; van Dijk & Kintsch, 1983) represents one very influential theoretical position. Bransford argued that comprehension typically requires our active involvement to supply information that is not explicitly contained in the text. Johnson-Laird (1980) argued that readers typically construct a relatively complete "mental model" of the situation and events referred to in the text (see Chapter 9). A key implication of the constructionist approach is that numerous elaborative inferences are typically drawn while reading a text.

Most early research supporting the constructionist position involved using memory tests to assess inference drawing. For example, Bransford et al. (1972) presented their participants with sentences such as, "Three turtles rested on a floating log, and a fish swam beneath them." They argued that the inference would be drawn that the fish swam under the log. To test this, some participants on a subsequent recognition-memory test were given the sentence, "Three turtles rested on a floating log, and a fish swam beneath it." Most participants were confident that this inference was the original sentence. Indeed, the level of confidence was as high as it was when the original sentence was re-presented on the memory test! Bransford et al. (1972) concluded that inferences from text were typically stored in memory in the same way as information directly presented in the text.

Memory tests provide a rather indirect measure of inferential processes, and there is the ever-present danger that any inferences that are found on a memory test were made at the time of test rather than during reading. This issue has been considered in detail in a number of studies. The findings indicate that many (or most) inferences found on memory tests reflect reconstructive processes occurring during retrieval. As a result, there has been a marked reduction in the use of memory tasks in inference research.

Minimalist hypothesis

Problems of interpretation with the memory studies (e.g., Bransford et al., 1972) suggested to some theorists that the evidence for the constructionist position is relatively weak. For example, McKoon and Ratcliff (1992, p. 442) reached the following conclusion: "The widely accepted constructionist view of text processing has almost no unassailable empirical support.it is difficult to point to a single, unequival piece of evidence in favour of the automatic generation of constructionist inferences." McKoon and Ratcliff (1992, p. 440) proposed an alternative view which they referred to as the minimalist hypothesis:

In the absence of specific, goal-directed strategic processes, inferences of only two kinds are constructed: those that establish locally coherent representations of the parts of a text that are processed concurrently and those that rely on information that is quickly and easily available.

It is important to clarify some of the notions contained in the minimalist hypothesis. In sum, here are the main assumptions made by McKoon and Ratcliff (1992):

• Inferences are either automatic or strategic (goal-directed).

• Some automatic inferences establish local coherence (two or three sentences making sense on their own or in combination with easily available general knowledge); these inferences involve parts of the text that are in working memory at the same time (this is working memory in the sense of a general-purpose capacity rather than the Baddeley multiple-component working memory system discussed in Chapter 6).

• Other automatic inferences rely on information that is readily available either because it forms part of general knowledge or because it is explicitly stated in the text.

• Strategic inferences are formed in pursuit of the reader's goals; they sometimes serve to produce local coherence.

The greatest difference between the minimalist hypothesis and the constructionist position concerns the number of automatic inferences that are formed. Those who support the constructionist view claim that numerous automatic inferences are drawn in reading, whereas those who favour the minimalist hypothesis argue that there are very definite constraints on the number of inferences that are generated automatically.

McKoon and Ratcliff (1986) tested this discrepancy between the two theories. They argued that a sentence such as "The actress fell from the fourteenth storey" would automatically lead to the inference that she died from the constructionist viewpoint but not from the minimalist hypothesis. Participants read several short texts containing such sentences, followed by a recognition memory test on which they had to decide very rapidly whether or not certain words had been presented in any of the texts. There were critical test words that represented inferences from a presented sentence but which had not actually been presented (e.g., "dead" for the sentence about the actress). The correct response to these critical test words was "No". However, if participants had formed the inference, then this would presumably lead to errors.

Colour-naming time in the Stroop task as a function of whether or not participants had been asked to guess the instrument in preceding sentences, and as a function of whether the words on the Stroop test were in or out of context with the preceding sentences. Based on data in Dosher and Corbett (1982).

Business Correspondence

Business Correspondence

24 chapters on preparing to write the letter and finding the proper viewpoint how to open the letter, present the proposition convincingly, make an effective close how to acquire a forceful style and inject originality how to adapt selling appeal to different prospects and get orders by letter proved principles and practical schemes illustrated by extracts from 217 actual letter.

Get My Free Ebook

Post a comment