Sentence complexity
المؤلف:
Paul Warren
المصدر:
Introducing Psycholinguistics
الجزء والصفحة:
P27
2025-10-29
34
Sentence complexity
The previous section has looked at how the abstract word entities lemmas that are selected at the functional level are inserted into positional-level sentence frames. It was mentioned earlier that one of the determinants of the sentence frame might be the perspective being taken by the speaker, such as which elements in the message are being placed in focus. So the speaker might choose the passive form the mouse was hunted by the cat to place the mouse in focus. However, there may be a cost associated with using the passive form, because it has been argued that the passive form is syntactically more complex than the active.
Indeed, for some time work in psycholinguistics responded to a suggestion from linguistic theory that passives are derived from actives. This syntactic account makes a distinction between underlying or deep structure representations of sentences and the surface structure representations that we actually see and hear Chomsky, 1965. It was argued that a surface passive form is obtained from an underlying active-like representation by means of a transformation, expressed in a rule such as that in 2.14, where NP stands for Noun Phrase e.g. the determiner +noun sequence e a , BE stands for the appropriate form of the verb o e , and the arrow shows the operation of the rule, changing the string on the left to that on the right.
The implication is that it is more complicated to produce a passive sentence than an active one, because the passive is further removed from the basic underlying concept. This is encapsulated in the Derivational Theory of complexity e.g. (Miller & Chomsky, 1963).
However, while results of some early studies suggested indeed that passives took longer to produce than actives, subsequent research showed that the speaker’s choice of a passive rather than an active form is based on what is being placed in focus Tannenbaum Williams, 1968. Imagine for example that participants are asked to use a passive sentence to describe the cat-chasing-mouse event shown earlier in Figure 2.3. If the picture is preceded by a paragraph that has the mouse as its topic, then it takes no more time to produce a passive than an active sentence to describe the picture. In terms of the model of language production sketched earlier, the pre-verbal message in such a situation includes the intention to present the event from the perspective of the mouse, which combines easily with the instruction to use a passive sentence, such as e o se Throughout this book standard abbreviations for word classes and phrase types are used, such as S for sentence or clause, N for noun, V for verb, P for preposition, Adj for adjective, Adv for adverb, as well as NP, VP, etc. for phrases in which a noun, verb, etc. is the head, or the main element. In fact, both the inherent salience of an entity which relates to concreteness, animacy, etc. and its salience in the dis course context whether it has recently been talked about, has been the focus of discussion, etc. influence the likelihood that the words expressing this entity will be in syntactically prominent positions, i.e. early in the sentence and/or in subject position. This in turn can affect the likelihood of the speaker producing an active or a passive sentence (Prat-Sala & Branigan, 2000).
It has been argued that more specific aspects of theories of sentence structure can be tested by looking at patterns of pausing and other durational patterns in speech production. For instance, the sentences in 2.15 and 2.16 appear to have the same structure, differing only in the verb. Indeed one grammatical analysis Postal, 1974 argued that ae is the object of the first verb, i.e. of the verb in the main clause, in both sentences. However, Chomsky 1973 argued that in 2.15 ae is the subject of the second verb, i.e. of the verb in the subordinate clause. Intuitively it seems reasonable that the sentences in 2.15 and 2.16 are somehow different, even though they are superficially similar. For example, although it is true that if you persuade someone to do something then you are also persuading that person, it is not the case that if you expect someone to do something then you are also expecting that person. What you are expecting is the complete situation, as reflected in the fact that the sentence in 2.17 is an acceptable alternative to that in 2.15 while the corresponding sentence for e s a e in 2.18 is simply ungrammatical.
Chomsky’s account for this difference was that the sentences in 2.15 and 2.16 have a different underlying or deep’ structure, something like 2.19 and 2.20. Crucially, the structure in 2.19 has a boundary marked by’ between two underlying clauses after expected (i.e. before the subordinate clause), but in 2.20 there is no corresponding boundary after persuaded.
Constituent boundaries in sentences are often marked by lengthening of words before those boundaries as well as by pausing, and these effects are more likely at major syntactic boundaries, such as before clauses, including subordinate clauses. With this in mind, Cooper 1976 examined recordings of critical sentences for evidence of such lengthening. Specifically, the duration of expected was measured for sentences like 2.15 and 2.17, and compared with its duration in 2.21, where there is no clause boundary immediately after expected. Similarly, the duration of persuaded in 2.16 was compared with that in 2.22.
Cooper found that expected was longer in both 2.15 and 2.17 than in 2.21, while e s a e was not any longer in 2.16 than in 2.22. This supports the idea that expected precedes a clause boundary in 2.15, just as it does in 2.17, but the same is not true of persuaded in 2.16. This result is claimed as support for the distinction in the grammatical analysis of the sentences that is shown in 2.19 and 2.20. It is interesting as an illustration of how psycholinguistics has often been used as a test-bed for theoretical notions in linguistics.
Another notion of complexity takes into account how many additional clauses are used to express notions related to the main idea in the sentence. Multi-clause sentences consist of a main clause and a number of subordinate clauses. For example, sentence 2.23 has a main clause The cat hunted the mouse, and one subordinate clause which ran away. The sentence in 2.24 has two subordinate clauses for the same main clause.
To determine the level of complexity, we can calculate a subordination index first suggested by Goldman-Eisler, Skarbek & Henderson, 1965, based on the ratio of subordinate clauses to total clauses so the sentence in 2.24 has a higher level of complexity, at 2/3 or 0.67, than that in 2.23, at 1/2 or 0.5. If greater complexity requires more planning, then this might be one factor determining the amount of pausing in different tasks. Earlier in this chapter we saw that describing a cartoon story resulted in less pausing than interpreting the cartoon. It turns out that the subordination index calculated over the cartoon descriptions is lower at 0.19 than that calculated over the cartoon interpretations 0.50 Goldman-Eisler et al., 1965. This is understandable, given that an interpretation of a cartoon story is likely to involve structures that explain cause and effect, i.e. is likely to require multi-clausal structures such as the man was shouting because the boy broke the window when he kicked the ball).
However, it turns out that structural complexity as shown by a subordination index is not the only factor that determines the amount of pausing. Consider a comparison of subordination indices and pause data from cartoon descriptions and interpretations with those from interviews. Interviews had nearly the same degree of structural complexity as cartoon interpretations 0.49, but a lower pause rate than both cartoon tasks Goldman-Eisler e al., 1965. The crucial difference is that the interviews included a second speaker, the interviewer, who was likely to interrupt the participant. A pause in a dialogue can indicate a transition point between one speaker and the other. When interviewees want to keep the floor, then they need to reduce the opportunities for the interviewer to take a turn, and they can do this by lowering the pause rate.
الاكثر قراءة في Linguistics fields
اخر الاخبار
اخبار العتبة العباسية المقدسة
