Anaphora
المؤلف:
Paul Warren
المصدر:
Introducing Psycholinguistics
الجزء والصفحة:
P203
2025-11-12
59
Anaphora
It is clear that an important aspect of language comprehension is the making of connections between the different parts of a discourse. In order for a larger stretch of speech or text to make sense, it has to show both coherence and cohesion. Coherence requires that there is consistency between the events or states in series of sentences. We have seen in the previous section that this often involves making the appropriate inferences. To take another example, the sentences in (12.20) work as a connected pair as long as we make the interpretative connections that allow them to make sense together e.g. that the man and Jane are strangers, that the man asked Jane for directions, and so on.
Lack of coherence can cause delays in processing. It takes longer to decide that he refers to Henry in (12.22) than it does in (12.21), even though he can only refer to Henry and not to Jane in both cases (Stevenson & Vitkovitch, 1986). This is because picking up money is not clearly associated with jumping over a ravine, whereas falling into the river is.
Cohesion on the other hand involves making the appropriate links between the words and phrases in a text. One of the most extensively studied ways in which this happens is anaphora, which involves making connections between textual elements such as pronouns and elements earlier in the discourse which the pronouns are standing for, such as nouns or full noun phrases. For instance, the two sentences in (12.23) are interpretable as a pair because we link some anaphors (the pronouns she and him) to their antecedents (Jane and a man respectively).
Successful interpretation of discourse therefore often requires that we can identify anaphors and link them back to their antecedents. But even when we can achieve this successfully, lack of coherence can still cause difficulty, as (12.22) has shown.
The forms of anaphors and of their antecedents play an important role in our ability to recognise them and to make the anaphoric connection. Occasionally an anaphor and its antecedent will be exactly the same phrase, as would be the case if (12.17) is used as the continuation of (12.16) above, or if the second sentence in (12.23) had started with Jane and not she. However, this is rather unusual, and can in fact seem marked if the anaphor and its antecedent are close together in the text, as they would be in (12.23).
Anaphors tend to be either pronouns, as in (12.23), or definite noun phrases e.g. phrases introduced with the rather than a, such the man rather than a man. Antecedents, on the other hand, will often involve indefinite phrases, particularly when they refer to entities that are being introduced into the discourse for the first time. So, in (12.20) and (12.23) a man introduces a new entity. Given this difference between anaphors and their antecedents, if we encounter a pronoun or a definite noun phrase, then there is a good chance that this is in an anaphoric relationship to something earlier in the discourse, or to something presupposed by the speaker or writer, as was pointed out in Chapter 11 in relation to the famous garden path sentence that involves The horse . . .. If we have identified an anaphor, then we need to find the antecedent in order to fully interpret the text.
Bridging inferences
Relating an anaphor to its antecedent is something readers and listeners can do with great efficiency. Often, though, the antecedent may not be made explicit, and so anaphor resolution requires some additional inferences to be made, referred to as bridging inferences. Making these inferences takes extra time and can result in delays in sentence processing. For instance, Haviland and Clark (1974) found that readers show faster com prehension times for the second sentence in (12.24) than for the same sentence in (12.25). The difference is that the anaphor e ee has an explicit antecedent in a preceding sentence in the first example, but not in the second. It is argued that additional interpretative work is required before the reader can connect the two sentences in 12.25, resulting in delays in processing.
Garrod and Sanford (1982) suggest that Haviland and Clark’s effect depends on how readily the coherence can be established between sentences. Garrod and Sanford found that the second sentences in pairs like (12.26) and (12.27) were processed equally rapidly, even though in (12.27) there is no explicit antecedent for the car. They argue that this ease of processing of (12.27) compared with (12.25) is because a is implicit in the role of vehicle used to drive to London in the first sentence in (12.27), whereas ee in (12.25) may not be an obvious component of picnic supplies.
If the anaphor is a pronoun, rather than a full noun phrase, then it will be less readily interpreted through bridging inferences, i.e. the antecedent for a pronominal anaphor needs to be mentioned explicitly. So the sentence pair in (12.28) is anomalous, but that in (12.29) is not. The pro nominal form in (12.28) does not work without an explicit antecedent. As a result, there is a temptation – initially at least – to assume that it in (12.28) refers to London.
Instances and categories
Other investigations using techniques such as the measurement of eye movements or reading times have shown that the degree of specificity of the noun phrases involved can affect the processing of anaphoric relations. For instance, although each of examples (12.30) and (12.31) has identical nouns in its antecedent and anaphor phrases, Garrod and Sanford (1982) demonstrated through a self-paced reading task that the processing of the sentences with more specific nouns, or instances, as with shark in (12.30), is faster than that of the sentences with more general nouns, or categories, as with fish in (12.31).
Moreover, if the antecedent is an instance (shark), and the anaphor is the category name (fish), as in (12.32), then processing is faster than if the opposite is the case, as in (12.33).
In addition, Garnham (1989) showed in a clause-by-clause self-paced reading task that if a category name like fish is used with a type of verb that is more likely to be associated with the instance (shark), such as attack, then the interpretation of fish is that it must be a type of fish, such as a shark, that is likely to attack, and the difference between category-instance and instance-category orders is removed, so that (12.35) is now not slower than (12.34).
الاكثر قراءة في Linguistics fields
اخر الاخبار
اخبار العتبة العباسية المقدسة
