48

YOUNG CHILDREN’S COMPREHENSION OF COMPLEX SENTENCES* MICHAEL Portsmouth

J. FLUCK Polytechnic

Studies of children’s speech have shown that sentences with subordinate clauses in the subject generally appear later than those with such clauses in the object; this fits Yngve’s model of sentence production. However, there are reasons for believing that children must learn to perceive and interpret ’subject’ sentences at an earlier age than ’object’ sentences. The present study was intended to investigate this and to test specific

predictions based on hypothesized processing strategies. The predictions were generally supported and it was shown that children do not acquire a strategy which allows them to interpret ’object’ clauses before mental age 7, whereas a strategy which can handle ’subject’ sentences is established before this.

INTRODUCTION The use of sentences containing subordinate clauses facilitates the precise and explicit communication of relationships and distinctions. The child whose competence does not include such devices may be at a disadvantage in an educational system which emphasizes the use of language for such purposes (Bernstein, 1972) A numbcr of researchers have studied the relation between language use and socio-economic background and have included children’s use of subordinate clauses among their measures (Lawton, 1968; Williams, 1971). Lawton summarizes the evidence concerning the use of these clauses as follows: middle class children use them more frequently than their lower working class counterparts, who tend to fall back on short simple sentences linked by repetitive conjunctions; among children who use subordinate clauses, object clauses (0) appear earlier than subject clauses (S). (S) The dog who followed the woman carried a basket. I(O) The dog followed the woman who carried a basket. These findings suggest that production of (S) sentences is a more complicated process than that of (0) sentences, and they support Yngve’s (1960) model of sentence production in this respect because it would assign a mean ’depth’ (Martin and Roberts, 1966) of 1.56 to l~S) but only 1.11 to (0). However, it is not certain that the same pattern would obtain for perception. The research to be reported here was designed to investigate young children’s handling of sentences like (S?. and (0) above and (&dquo;&&dquo;) below: (&dquo;& dquo;) The dog followed the woman and carried the basket.

*I would like to thank Mr. A. Ovens, Head ofSt. John’s C. Gosport, and his staff for providing subjects and facilities.

ofE. Primary School,

49 ,

It

was

some

hoped to discover which type young listeners learn light on the strategies used and their development.

to

interpret first

and

to

shed

There have been relatively few studies of how listeners cope with subordinate clauses. Clark and Clark (1968) studied children’s recall of sentences with a time adverbial clause and found that recall was best when the order of the clauses matched the order of the events described. Miller and Isard ~~1964} and Blumenthal (1966) found that adults experienced great difficulty with multiple embedding, and Blumenthal showed that this was associated with a strategy in which successive verbs were erroneously related to the sentence subject. Wright (1969) compared adults’ handling of (S) and (0) sentences using Savin and Perchonock’s (1965) ’overflow’ technique and found no difference; but work by Mathews ~( 1968) and Glucksberg and Danks (1969) has cast doubt on this technique. The author (Fluck, 1972), using Slobin’s (1966) verification time method, found that adults could process (S) sentences quicker than (0), suggesting that the latter are perceptually more complex. Consideration of the available evidence concerning sentence perception makes it possible to postulate processing strategies for children from which specific predictions may be made about their handling of sentences like those described above. Any account of sentence perception must take account of aspects of sentence structure which transformational grammar B(Chomsky, 1965) represents at the deep structure level (Blumenthal and Boakes, 1967; Mehler and Carey 1967; 1968; Bever, Lackner and Kirk, 1969; Levelt, 1971). However, transformational grammar does not describe the operations involved in perception; perceptual complexity does not increase regularly with derivational complexity (Fodor et al., 1967; 1968), and neither are the different aspects of linguistic organization separate in perception as they are in transformational grammar (Fodor et al. ’(ibid); Lieberman, 1963; Slobin, 1966). Bever (1970,1971) has introduced the notion of ’perceptual mapping operations’ whereby listeners relate surface structure features to their knowledge of possible deep structures; they include strategies for identifying the component clauses of sentences. Miller (1962) has argued that listeners must employ a ’delayed decision’ strategy, delaying their perceptual decisions while the input accumulates in a temporary store and then making global decisions about the identity of whole groups of words every 1 or 2 seconds. This would allow a more feasible decision making rate than a word or syllable based strategy, and would enable more advantage to be taken of semantic and syntactic redundancy. There is evidence that the phrase may be a natural ’decision unit’ (Fodor and Bever, 1965), but the primary unit seems to be the surface structure clause corresponding to the deep structure sentence unit (Bever, Lackner and Kirk, 1969; Jarvella, 1971). The strategies and the predictions derived from them are presented in the following section. Children aged 5 and 6 were chosen for this initial investigation for two reasons: first, other studies show that a number of Important linguistic developments occur around this time (Chomsky, 1969; Cromer, 1970; Slobin, 1966); second, on entering school, children interact with a variety of new linguistic models in the form of their teachers and classmates.

50



Fig. 1. 1. 2. 3. 4. 0.

Strategy A Read item. Is item null? Could items read Code a clause.

so

far form

a

clause?

Stop.

Strategies and Predictions. Three strategies are postulated,

The

Strategy

in order of

acquisition.

A:

This first strategy can only cope with simple one clause sentences. It codes the first clause available, e.g. &dquo;The dog followed the woman&dquo; in (&dquo;&&dquo;) above; it then breaks down with a complex sentence like those under discussion because it cannot provide a noun phrase for the actor of the second clause. It is presented in the form of an algorithm in

Fig. 1. z

~

Strategy

B:

’ .

This strategy can handle (S) and (&dquo;& dquo;) sentences but not (0) types. It develops when children come to distinguish between the ’topic’ and ’comment’ of sentences; Homby, Hass and Feldman (1970) have shown that this occurs between 5 and 7. The ’topic’ is what the sentence is about, its referent is prominent either perceptually, or because of previous sentences and it usually forms the grammatical subject. In this strategy, the listener codes a ’topic’ noun phrase as soon as possible and maintains a separate representation of this so that it is always available to interpret pronouns or complete clauses: it correctly provides an actor for the second clauses of (S) and (&dquo;&&dquo;) sentences, but (0) sentences are wrongly interpreted as i(S) types. Strategy B is summarized as an algorithm in Fig. 2 : 1. In order to interpret passive sentences this strategy requires an additional operation

51

Fig.

2: 1.

Strategy

B ’ ’

1.

Read item.

2.

Is item null?

3.

Has

4.

Could items read

5.

Code NP

6.

Could item be

a

7.

Rewrite RP

T.

8.

Could items read

9.

Code a clause with NP1 + Verb + Verb + (Object).

&dquo;

a

been coded?

’topic’ i(TD as

.

.

so

far form

T and

a noun

phrase (NP)?

store.

relative pronoun

(PR)?



as

so~

far form

Clause?

a

(NP,)

=

Actor +

-

10.

Could items read as

so

far form

clause if T

were

invoked .

~

0.

a

subject?

Stop.

.

-

~

_

52

Fig.

2: 2. 8.

8p. 9.

9p. 10.

An extension of Strategy B to cope with passives Could items read so far form a clause? Is clause passive? Code clause with NP, + Verb + c~NP2) = Actor + Verb + (Object). Code clause with NP, + Verb + (NP,) = Object + Verb + (Actor). Could items so far read form a clause if T were invoked as

subject?

which is shown in Fig. 2 : 2. It might be expected that the ability to cope with passives would develop shortly after Strategy B insofar as the ’topic’ must be identified and interpreted as the logical object of the sentence.

Strategy

C:

This strategy contains a modification which enables it to interpret (0) sentences When a clause is detected it is not coded until the next item has been examined. If this is a relative pronoun then the preceding noun phrase is located and coded so that it may be used to interpret the pronoun before the first clause is coded. Thus the final clause is assigned the correct actor. This strategy is shown algorithmically in Fig. 3: 1. To cope with passives, it would require a similar modification to B, see Fig. 3 : 2. It is suggested that children progress through these strategies in the order shown as they make the necessary inductions about their language and as they develop the relevant information processing skills. It is further suggested that most children will develop Strategy B shortly after they enter school but that Strategy C will not develop until later, possibly in their second year. Passive versions of the (S) and (0) sentences were included in the investigation to

correctly.

53

Fig.

.

-

3: 1. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

13. 0.

Strategy C Read item. It item null? Has T been coded? Could items read so far form a NP? Code NP as T and store. Could item be a RP? : Rewrite RP as T. Could items read so far form a clause? Read next item. . Could item be a RP? Code preceding NP and translate RP as NP. Could items read so far form a clause if T were invoked as subject? Code a clause with NP, + Verb + (NP,) = Actor + Verb + (Object). ’

&dquo;

.



B,

B,



’ ’









,

Stop.



54

Fig.

3: 2.

An extension of

Strategy

C

to

cope with

passives.

13p. Is clause passive? 14.

14p.

Code a clause with NP1 + Verb + + Verb + (Actor). Code a clause with NP1 + Verb + + Verb + (Actor).

(NP,) = Object

(NP,) = Object

ascertain how children coped with them. Insofar as the ability to handle passives is seen as an extension of Strategy B (see Fig. 2: 2), it might be predicted that it would appear before the ability to handle

Young children's comprehension of complex sentences.

48 YOUNG CHILDREN’S COMPREHENSION OF COMPLEX SENTENCES* MICHAEL Portsmouth J. FLUCK Polytechnic Studies of children’s speech have shown that senten...
772KB Sizes 0 Downloads 0 Views