CHAPTER 3.

A PILOT GROUP DESIGN STUDY: THE USE OF UNESTABLISHED REFERENT BY FIVE COMMUNICATIVELY IMPAIRED CHILDREN.

3.1 INTRODUCTION

3.2 THE RECORDING OF UNESTABLISHED REFERENTS

3.3 THE IDENTIFICATION OF UNESTABLISHED REFERENTS

3.4 METHOD

3.5 RESULTS

3.6 DISCUSSION

3.1 INTRODUCTION

The literature survey in Chapter 2 has drawn attention to a number of studies in which it has been suggested that a high frequency of instances of failure to establish referent is characteristic of the conversations of children in whom the communicative aspects of language are impaired. (Bishop & Adams, 1989; Damico, 1985; Haines, 19a4; Johnston, 1985; Johnson, Johnston & Weinrich, 1984; Jones, Smedley & Jennings, 1986; McTear, 1985b; Rapin, 1987; Smedley, 1989). Among these children are those to whom the term semantic-pragmatic disorder' has been applied. Major characteristics attributed to this group are inappropriate or 'odd' conversational interaction, fluent and mainly intelligible speech, scores within the normal range on non-verbal tests of intelligence and delayed verbal comprehension (Adams & Bishop, 1987; Bishop & Rosenbloom, 1987; Rapin, 1987; Rapin & Allen, 1983). In the current study the term communicatively impaired is used to refer to this group of children.

As already discussed (2.2.), the establishment of referent corresponds closely to the parameter of Specificity/Accuracy on Prutting's Pragmatic Protocol (Prutting and Kirchner, 1987).

Using the Pragmatic Protocol as a framework, Prutting and Kirchner found that the conversations of a heterogeneous group of language-impaired children were more frequently marked by inappropriateness on the parameter Specificity/Accuracy than on any other parameter of the Protocol (1.7.3). A large range of variability was, however, found within the language-impaired group. Prutting and Kirchner suggest that this variability could reflect the lack of homogeneity of these subjects due to the general nature of the diagnostic classification used in their selection. They speculate that, if sub-groups of language-disordered children were studied (for example those distinguished from each other by their level of verbal comprehension) particular pragmatic deficits might be found to be associated with particular combinations of linguistic and cognitive deficits (McTear & Conti-Ramsden, 1989; Prutting & Kirchner, 1987).

An investigation, along the lines proposed by Prutting and Kirchner, was carried out with communicatively impaired children as the subjects. A pilot study, using a group design approach, compared the frequency with which five communicatively impaired children used unestablished referents with the frequency of its use by children matched for age and by children matched for verbal comprehension. As discussed in 2.6, the aim was to determine whether unestablished referent was used more frequently by communicatively impaired children than by their age matched peers, but was used with equal frequency by the communicatively impaired children and younger children at a similar level of verbal comprehension. Such a finding would support the hypothesis that failure to establish referent is a feature of the language of communicatively impaired children and that it is associated with a delay in verbal comprehension.

3.2 THE RECORDING OF UNESTABLISHED REFERENTS

Two approaches to recording the frequency with which instances of unestablished referent occur in conversation have been reported in the literature. One approach is to note every referring expression where the listener has no way of knowing what is being referred to (Adams & Bishop, 1989; Damico, 1985; Liles, 1985).

An alternative approach notes only those instances of unestablished referent which the observer judges to "penalize the interaction" (Prutting & Kirchner, 1987, p. 109) or to produce "a sense of oddness and interruption of the normal conversational flow" (Bishop & Adams, 1989, p. 242). Both Prutting and Kirchner and Bishop and Adams demonstrate reliability between trained observers for these judgments. The position of Prutting and Kirchner is that if an aspect of the Pragmatic Protocol "is utilized incorrectly but does not seem to penalize the interaction" (p. 109) it should not be judged inappropriate. Bishop and Adams take the following view:

"Not every non-established referent automatically confers inappropriacy onto the entire utterance: there appears to be a degree of 'non-establishment' which the listener cannot tolerate. Our impression was that in general a non-established referent must represent a key part of the sentence to convey a sense of inappropriacy to the listener" (p. 251).

Other investigators have also presented evidence that unestablished referents can occur without apparently disrupting the flow of conversation. Van Hekken, Vergeer and Harris (1980) looked at the ambiguous use of pronouns in conversation between eight pairs of pre-schoolers and at the effects of this ambiguous use on social responsiveness. Social unresponsiveness was defined as the listener either making a verbal statement not related to what the previous speaker had said, or failing to respond at all (whether verbally, non-verbally or by turning towards or watching the speaker). They found that social unresponsiveness was no more likely to occur following the ambiguous use of a pronoun than following its unambiguous use. This finding may be partly explained by the difficulty (see 2.3.5) which young children appear to have in detecting ambiguity (Bowman. 1984), and also by their lack of awareness of the use of cues to indicate communicative breakdown (McTear, 1985b). More requests for clarification can thus be expected as children grow older.

It is not only in the conversations of young children, however, that instances of unestablished referent may occur without apparent reaction from the listener. It has been observed that conversational partners will let problems in understanding pass, either in the expectation that the meaning will become clear later (Garvey, 1984; McTear, 1984. 1987) or because they feel there is little to be gained by interrupting the conversation (McTear, 1984). In addition, factors such as the personality of the listener, the nature of the conversation and the relationship between the participants may influence whether or not a failure to understand is signalled. Where an individual has communicative difficulties the degree to which a conversational partner uses compensatory strategies may affect how apparent these difficulties are (McTear & Conti-Ramsden, 1989). It appears that the co-operative principle (Grice, 1975) predisposes the participants in the majority of conversations to try to avoid allowing errors to disrupt the interaction and to assume that sense can be made of what is said. It follows that social responsiveness on the part of the addressee does not necessarily indicate that there has been no ambiguity in the speaker's message. Ambiguity of the message is listed in the Pragmatic Protocol as an example of inappropriateness on the parameter Accuracy/Specificity. The absence of a request for clarification, or of some other evident reaction, following an unestablished referent is thus not in itself an indication that there has been no inappropriate effect on the communicative exchange.

It seems likely that when a conversational partner cannot work out what is indicated by a referring expression, there will be some negative effect on the communicative interaction. To what degree the conversation is disrupted may, however, range from a very mild effect, which is easily corrected or assimilated, to a complete conversational breakdown. The nature and degree of disruption brought about in a conversation by any given failure to establish referent appears to be influenced by the context in which the referring expression occurs. It is therefore appropriately investigated by means of a qualitative analysis. the other hand, a quantitative analysis is required to investigate the frequency of use of unestablished referent by the subjects.

On the basis of these considerations, a decision was taken that the recording of unestablished referents in the current study would follow the first of the two approaches outlined above. A count was made of all instances of the use by subjects of a referring expression which did not enable the addressee(s) to identify the referent.

3.3 THE IDENTIFICATION OF UNESTABLISHED REFERENTS

In analysing the transcripts recorded for the current study a substitution test was applied to determine whether referring expressions had adequately established referent.

Substitution Test: Can you identify and/or does the data give evidence that the addressee(s) could identity the entity, or set of entities, indicated by each referring expression? Is the entity identified with appropriate specificity for the context?

Prutting and Kirchner's (1987) concern with "the overuse of non-specific terms" (p.114) has here been extended to cover the concept of an appropriate degree of specificity. In 1.4.4. above, it was pointed out that failure to establish referent may result not only from the use of over-general semantic terms but also from the use of terms which are over-specific for the context. The hypothetical example was given of a child saying his sister had started 'walking' to school (rather than using the less specific term 'going') when the message he intended to convey was simply that she had started school. In this context the use of the more specific term 'walking' results in a failure to convey the intended meaning. The concept of an appropriate degree of specificity can be used by the coder to identify referring expressions which, like this one, are inappropriately situated on the continuum from general to particular.

A number of other types of referring expression can also be assessed in relation to their degree of specificity. For example, a demonstrative, such as 'here', or a locating device, such as 'round the corner', may fail to give a listener sufficiently specific information about a location for the demands of a particular context.

In addition, the concept of appropriate specificity can be applied where the speaker uses a term which the addressee can identify in a general way, although requiring further information. Perera (1984) points out that in this regard there is a difference between written and spoken language. For example, in writing it would be normal to state, "The director is visiting Senegal, on the West Coast of Africa" (p. 199). However. in conversation one would say simply, "The director is visiting Senegal", providing the additional information if the listener requested it. This is done partly to avoid density of content but also because it is not socially acceptable to assume that addressees are ignorant or to treat them like an audience rather than as conversational partners. Several exchanges of this type occur in the data recorded for the current study. In one of these a child asks a puppet, who has been refusing to put on a hat before going out in the sun, "Do you want to get sun cancer?" and the puppet replies. "What's sun cancer?" (Transcript EY, Turn 064-065). The referring term 'sun cancer was not here coded as failing to establish referent since, within the context of the conversation, it could immediately be deduced that it referred to an unpleasant effect of going unprotected into the sun. In other words, the general nature of the referent was established, with the opportunity for detail to be added if the listener required it. It should be emphasised that whether such terms are judged to be appropriately specific is dependent on the context in which they occur. For example, if the addressee had never heard of Senegal, and the speaker were to open the conversation by asking, "Do you like Senegal?", the addressee would have no way of knowing whether the reference was to a place, a person, a song title or any other of a number of types of entity. The referent would therefore be coded as unestablished.

The selection of an appropriately specific referring expression is closely associated with the need to establish and maintain topic. This association has been discussed by Schegloff (1972) who makes use of the terms 'focussing off' and 'focussing on'. Focussing off involves using a general term to indicate a referent when the details of its identity are not relevant to the current topic. Schegloff gives the example of a conversation where the speakers are discussing a friend who has the habit of borrowing money. One of them says:

"Cal Major came in 'n delivered something and she w- said she didn't have the change. Would I loan 'er the money to pay im." (p. 126).

Since 'what was delivered' is not relevant to the topic, it is focused off by the use of the general term something', which in this context is appropriately specific. On the other hand, a referent which is relevant to the current topic can be focussed on. This is achieved by the use of a referring term which is specific and which establishes the relationship of the referent to the topic. In the exchange used by Schegloff (1972) to illustrate this point, the topic of conversation is speculation about why armed police have surrounded a particular department store. One of the speakers suggests there may have been a bank hold-up and points out there is a bank "right on the corner". Because of its relevance to the topic, the location of the bank is identified precisely (rather than through the use of a general expression, such as there ). In addition, since the department store is part of the topic, the bank's location is described in terms of its relationship to the store, i.e. 'on the corner' (of that block) rather than 'in such-and-such a street'.

It can be seen from this discussion that when judging whether a referring expression is of an appropriate degree of specificity to establish referent, one is likely to need to consider the relationship of that referent to the current topic of conversation.

The aim of the study was thus to identify and record all instances of the use by subjects of referring expressions which did not establish referent, and to compare the frequency of their use by the communicatively impaired children with the frequency of their use by these children's age and comprehension matched controls.

3.4 METHOD

3.4.1 SUBJECTS

There were a total of fifteen subjects; five of whom were communicatively impaired children, aged 4;9, 5;6, 5;6, 6;9, and 7;10 respectively. Each of the communicatively impaired children was matched to a child with normal language on the basis of chronological age and to another younger child, also with normal language, on the basis of verbal comprehension level. The communicatively impaired children are identified as A, B, C, D, and E; their age matched controls as AO, BO etc.; and their comprehension matched controls as AY, BY etc. (see Table 4 below).

TABLE 4 LIST OF SUBJECTS

All the communicatively impaired subjects, except one, and all controls lived within a single school district of North Brisbane consisting of households of similar socio-economic composition. The fifth subject (B) lived in a similar, neighbouring school district. All subjects and controls, with the exception of the two youngest controls who had not yet started formal education, attended local State primary or pre-schools.

The communicatively impaired subjects were found through the local Speech Therapy Service. having been referred as inappropriate in their conversational interaction, but having, at the time of the study, adequately fluent and intelligible speech.

In all cases the 'oddness which was at times apparent when talking with the children was emphasised by those involved with them. As discussed in 2.6. above, the additional selection criteria for these children were that they had been assessed as within the normal range of non-verbal intelligence but as below the normal range in verbal comprehension. Each of the five children was tested on the Reynell Developmental Language Scales - Revised (Reynell. 1979). As shown in Table 1, subjects A, B, C and D scored at least 1.5 standard deviations below the mean on this test. Subject E scored at 10 to 14 months below his age level but was above the upper age limit for calculating a standard score. All the children had received psychological evaluation by a qualified person in the Guidance and Special Education or Child Guidance Services of their district. Four had produced scores on the WPPSI (Wechsler Preschool and Primary Scale of Intelligence) or WISC-R (Wechsler Intelligence Scale for Children - Revised) indicating that non-verbal intelligence levels were within the normal range. The guidance officer responsible for psychological evaluation of the fifth subject (Subject B) took the view that the administration of a formal test would be of doubtful validity, but was satisfied from the observation of her functioning in pre-school that her non-verbal intelligence was within the normal range.

None of the communicatively impaired subjects nor of the control subjects matched to them for chronological age or verbal comprehension level had a history of hearing loss. mental retardation, emotional disorder or neurological impairment (including oral, motor or sensory deficits). Subject B at the time of the investigation was temporarily wearing an eye patch for 'lazy-eye'.

In addition none of the ten children used as controls had a history of speech or language impairment or of academic difficulties requiring remedial services.

3.4.2 ADDITIONAL PARTICIPANTS

Each subject chose a friend to participate in a videoed conversation, the aim being to facilitate initiation and spontaneity. Bearing in mind Garvey's (1984) description of children's peer relationships as identified by the negotiated quality, constructed by their members, of equivalent status, a chosen peer can be regarded as a facilitative partner. Prutting and Kirchner (1987) also take this view. They comment that their choice of a familiar adult as conversational partner for the children in their study may have allowed fewer opportunities for initiation by the disordered communicators, since the adults were in the more dominant position in the relationship.

The parents of the three youngest language-impaired subjects expressed concern that the children had no close friends whom they could ask to take part in the videoed conversation. In one of these instances the child's parents arranged for his cousin to be involved and in the other two the respective pre-school teachers selected class-mates whom they felt were suitable. One of these sessions took place in the children's pre-school where it was felt they would be more relaxed. For this the physical lay-out of the other sessions, all of which took place in the speech therapy centre where the investigator was employed, was copied as closely as possible.

The decision to record a triadic rather than a dyadic conversation was taken during the pilot sessions leading up to the study. Videoed studies of dyadic conversation involve either a concealed camera and operator (an option not available in this case) or the presence in the room of the camera operator. In the pilot sessions it became clear that the children found it natural to involve the investigator/camera operator in their conversations and became inhibited if she avoided responding. Therefore a decision was made to record triadic rather than dyadic conversations.

An additional reason for the decision to record conversations between three people was the potential within these for a number of features which do not occur when two speakers only are involved. Some of these features have direct bearing on the establishment of referent. They include:

1. Instances of a referent directed at two addressees being established for one addressee but not for the other.

2. Instances of collaboration between two of the participants over the establishment of a referent for a third. 3. Instances of collaboration between two of the participants to jokingly deceive the third about the identity of a referent.

4. Instances of one participant asking a second to establish certain referents for a third ('Tell x who we met, what we saw' etc.).

Dyadic conversations. particularly conversations between mother and child, have been given considerable emphasis in the literature on child language development. Furthermore many of those clinical assessment procedures which require the recording and transcription of a conversation are designed solely for dyadic interactions (Blank & Franklin, 1980; Crystal, Fletcher & Garman, 1976; Prutting & Kirchner, 1987). However, many of the conversations in which young children participate are non-dyadic. Non-dyadic interactions were included in McTear's pioneering study of the pragmatic aspects of children's conversation (McTear 1985b), in which reference was made to exchanges between an adult and two children. In undertaking the study reported here, such exchanges were assumed to be a common conversational situation (albeit one whose role in early language development has been relatively little explored).

An additional feature of the conversations recorded for the current study resulted from the presence of a puppet. The puppet was an active participant in the situation presented to the children. The opportunity thus existed for the children (and for the investigator herself) to talk to the puppet, and also for them to make the puppet itself 'speak . This opportunity for role play was taken up and very much enjoyed by many of the participants in the study. On two occasions a child increased the number of characters involved by spontaneously fetching an additional puppet to join in the conversation.

3.4.3 MATERIALS

The puppet, a dog named Louie, was a 21cm. high hand puppet of the soft toy type.

The hat was a straw sun hat, slightly too large for the puppet.

The swing was of a type used for infants and had a green sling seat with a small plastic tray in front of it. It could be set in motion either by being pushed, or by turning a handle mounted on the triangular frame from which the seat was suspended.

Neither the hat nor the swing were visible from where the children were seated at the beginning of the videoed conversation (see Figure 1). This situation was maintained for the adaptation of the lay-out made for the session with subject A, which took place at his pre-school. All the other sessions took place in the Speech Therapy Centre at which the investigator was employed.

The camera used to videotape the sessions was a Sharp VHS Movie, model VC-C19PXE, with built-in microphone.

3.4.4 COLLECTION OF DATA

A videotape was made of each subject in a three-way conversation with a friend of their choice and the investigator.

The children were presented with a goal-directed situation, the same introduction being used for all participants in the study. The aim was not to produce a standardised setting but to motivate the participants to refer to entities removed from the here and now. Since the children were asked to explain a problem to a friend and to discuss a solution, the situation was a conversationally demanding one.

FIGURE 1 LAYOUT OF ROOM

(not to scale)

Each subject accompanied the investigator into a room laid out as shown in Figure 1. The investigator showed the subject a hat behind a screen and a swing down some outside steps and then played the child a video of a short conversation between 'Louie', the puppet, and herself. In this conversation Louie refused to wear a hat and was therefore refused permission to go outside And play on the swing. The 'Slip, Slap, Slop' skin cancer prevention campaign is given considerable emphasis in Brisbane schools and pre-schools, and it therefore seemed probable that the children would be able to relate this situation to their own experience. The subjects were then asked A number of questions About the situation and if necessary provided with the answers. (These questions are listed in Appendix D.) Following this, the subject's friend joined him or her in the room. The subject was asked to tell the friend what Louie wanted to do and why the adult would not let him do it, and then to decide how the two children could help him. The subsequent conversation was videoed up to the point where the children took the puppet to the swing. In three instances where the conversation to this point produced less than 20 analysable turns by the subject, transcription was continued to include the conversation which took place while the puppet was given a swing and returned to the room.

3.4.5 TRANSCRIPTION AND CODING

3.4.5.a CHAT transcription coding system

The video taped conversations were transcribed according to the CHAT Transcription Coding System of the Child Language Data Exchange System (MacWhinney; 1988) developed at Carnegie-Mellon University.

The CHAT transcription coding system (MacWhinney; 1988) is designed for transcribing naturalistic conversational data for automatic analysis by the computer. However, the comprehensive set of coding symbols available in the CHAT system also make it appropriate for the manual analysis required when examining pragmatic functioning.

The system offers a large array of coding options with the intention that researchers will chose those appropriate to their particular project. A key to the main coding conventions used for the transcriptions of the data recorded for the current study is given in Appendix B and the fifteen transcripts are shown in Appendix C. The number of analysed turns and the number of unestablished referents, with turn numbers, are given at the end of each transcript. Coding includes over-lapping, interrupting. retracing, pausing, and a code for indicating that the speaker is performing the role of a character. The last of these was particularly useful in making it clear when one of the participants was speaking for the puppet. Non-verbal aspects of conversation relevant to the establishment of referent (such as gesture. action. tone of voice, and identity of addressee) were also coded. Using the CHAT system, it was possible to indicate the precise point in the conversation at which these occurred.

A further feature of the CHAT system which makes it suitable for the current study is that it is designed for coding conversations with any number of participants.

3.4.5.b Method of calculation

The incidence of unestablished referents in the conversations recorded for the study was calculated in relation to the number of analysable turns in each transcript by dividing the number of unestablished referents by the number of turns.

3.4.5.c Definition of a turn

McTear has described turn-taking as switching between the roles of speaker and listener (McTear, 1985b). An approach which defines the boundaries between turns from the point of view of their acoustic dimensions has come to be seen as failing to take into account the interactional significance of gaps and overlaps. (See Inglis (1987) for a review of definitions of turn in the literature.) Conti-Ramsden, for example, in work carried out with Friel-Patti in the early 1980's used a definition of a turn involving measurement of the pause between strings of utterances (Conti-Ramsden & Friel-Patti, 1984) but subsequently dropped this in favour of a definition based on the change of speaker (private communication, March, 1989). The co-operative nature of turn-taking, and the fact that it is Governed by rules which determine at which points in the conversation participants may choose that a change of speaker takes place, require the definition of turn to be concerned with the conversational function of all features of the interaction (Goodwin, 1981; Sacks. Schegloff & Jefferson, 1974; West & Zimmerman, 1983).

In the current study a turn is defined as "a string of one or more utterances by an individual with or without accompanying gesture, or of one or more conversationally meaningful non-verbal acts.

Where speakers overlap because one interrupts another or because one chimes in with another in unison or attempted unison (which occurs several times in the data when two children are haranguing the puppet) change of speaker continues to be taken to indicate change of turn and the adjacent turns are simply taken to be overlapping.

To determine which non-verbal acts are conversationally meaningful a participant based criterion is used. If a non-verbal act is noted and acted upon by a conversational partner, or if it is in response to an initiation by a partner, it is regarded as conversationally meaningful. For example, the non-verbal turn in the following exchange is conversationally meaningful, since it is in response to the oblige in the preceding turn.

077*ANN: are you going on the swing now, Louie?
078*AO-: O.
      %gpx: makes Louie nod

In contrast, the non-verbal turn by the same child in the following quotation is not conversationally meaningful since it is not related to what is said by the conversational partner. Turns which are not coded as conversationally meaningful are not numbered.

070*ANN: oh, do you know what Louie said? he's unhappy because he still hasn't been on the swing.
   *AO-: www.
   %act: carries Louie back to the blackboard

'No response' is also regarded as conversationally meaningful and included in the data as a turn. A 'no response' turn occurs when a participant in a conversation fails to reply to an oblige by a conversational partner. The definition of an oblige, which is taken from Blank and Milewski's Cognitively-Based System of Assessment of Dialogue with Pre-Schoolers (1980), is as follows:

"Obliges are those utterances which through words, tone or gesture explicitly convey the expectation that a reply be offered by the listener. Commonly they are expressed as questions or commands although they need not take this form" (Blank & Milewski, 1980. p. 1; see also Blank & Franklin, 1980).

Conversationally meaningful non-verbal acts include deictic gestures (e.g. pointing, touching), symbolic gestures (e.g. shaking the head, nodding) and descriptive gestures (e.g. indicating the proportions of a referent with one's hands). The role of non-verbal acts in establishing referent is further discussed in Chapter 4 with reference to the data gathered in this study, and is illustrated with a number of still frames from the transcribed videotapes (Figure 2 a-j).

In transcribing conversationally meaningful non-verbal acts from the video tape a particular problem was presented by eye gaze. Eye gaze. in the sense of one participant looking at another's face, is one of the thirty parameters of Prutting's Pragmatic Protocol, one of its functions being described as supplementing or supporting linguistic aspects of the message. An accurate recording of eye gaze requires a complex methodology involving a camera focussed on the face of each participant which was not available for this study. Eye gaze as defined in the Pragmatic Protocol has therefore been omitted from the transcription. However eye gaze in the sense of turning the eyes to regard some aspect of the physical setting of the conversation, has been included where it constitutes an essential supplement to linguistic deixis in a referring expression. The transcription also notes instances of the addressee's head being turned away from the speaker when the speaker is making a gesture.

For ease of reference the conversationally meaningful turns in the transcribed conversations were numbered. Each of the turns contributed by the subjects was analysed, using the Substitution Test described above (3.3), to ascertain whether or not referring expressions established referent for the addressee(s).

Turns which could not be analysed because they were totally inaudible or unintelligible to the transcriber were numbered but not included in the turn count used in calculations. In a very small number of instances. intelligible referring expressions appeared to rely for their interpretation on an unintelligible or inaudible part of an utterance. In such instances the referring expression was judged to be unanalysable.

3.4.6 RELIABILITY

Three of the fifteen videoed and transcribed conversations were randomly selected and scored by an independent judge, a speech and language therapist who had previously been trained by the investigator in the scoring procedure. The number of analysable turns in the three transcripts was 197 which was 31.47% of the 626 analysable turns in the total data. Each turn was scored for the establishment or non-establishment of referents and reliability was calculated using the formula

This produced a percentage of agreement of 97.50%. The five instances of disagreement between the two judges were resolved through discussion.

3.5 RESULTS

Table 5 presents the mean number of unestablished referents per turn in each of the three groups in the transcribed conversations.

TABLE 5

Separate pair-wise comparisons between the communicatively impaired children and the children who acted as their controls were conducted, using t-tests for independent samples.

Table 6 shows the outcome of a t-test between the communicatively impaired children and their age-matched controls. This indicates that the communicatively impaired sample used significantly more unestablished referents ( x = .071 unestablished referents/turn) than did their age-matched peers (x = .012 unestablished referents/turn, t = 2.236, p lies between .025 and .05).

TABLE 6

Table 7 shows the outcome of a similar test between the communicatively impaired children and their verbal comprehension matched controls which showed that although the communicatively impaired sample made more errors (.071 unestablished referents/turn) this difference was not to a level of significance (t = 1.661. p lies between .05 and < .1).

TABLE 7

3.6 DISCUSSION

3.6.1 RELATIVE FREQUENCY OF THE OCCURRENCE OF UNESTABLISHED REFERENT

The findings reported here are consistent with those of a number of the studies discussed in 2.5. above (Bishop & Adams, 1989; Jones, Smedley & Jennings, 1986; Rapin, 1987). These studies found a high frequency of failure to establish referent in the conversations of communicatively impaired children, who had verbal comprehension deficits and a quality of 'oddness' in their conversational interaction.

In the current study it was found that the contributions to conversation made by five communicatively impaired children were characterised by a significantly higher frequency of unestablished referent than were those of the peers with normally developing language who were matched to each of these children on the basis of chronological age.

This finding is consistent with the hypothesis that communicatively impaired children fail more frequently to establish referent than do normally developing children of the same age. Individual variation was noted, however, particularly in relation to subject B who, while meeting the selection criteria for the communicatively impaired children, did not once in her conversation fail to establish referent. A possible explanation for this variation is discussed in 4.3.2 below.

The use of unestablished referent by the communicatively impaired children was also compared with that found in the conversations of younger children, each of whom was matched to a communicatively impaired child on the basis of verbal comprehension age. The language impaired children were again found to use unestablished referents more frequently than the controls, but this difference was not to a level of significance.

This finding is inconclusive in relation to the important issue of whether language impaired children are experiencing difficulties in establishing referent which fall into a pattern of developmental delay, or whether some of their difficulties fall outside this pattern and are specifically associated with their communicative impairment.

Before the findings of this study can be confirmed, it is necessary for it to be replicated with a larger sample. In particular, further investigation is needed of the relative frequency with which unestablished referent is used by children with the profile of communicative impairment in comparison with children who are not language-impaired but are at a similar stage in the development of verbal comprehension.

In addition, as discussed below (3.6.3), it can be argued that frequency of occurrence is not in itself a sufficient basis on which to compare the causes and effects of the failure to establish referent in different clinically identified groups.

3.6.2 POTENTIAL CLINICAL RELEVANCE

Comparison between children whose conversations are marked by inappropriateness and younger children at a similar level of verbal comprehension have the potential to increase our understanding of the functioning of the communicatively impaired children in a way which is directly relevant to assessment and remediation. A number of commentators (Bishop & Adams, 1989; Neville & Gunn. 1987) have made the point that an uneven profile of development (for example, one in which facility in using long and complex sentences is combined with difficulty in conveying precise meaning) can in itself create an impression of oddness for the addressee and thus affect how others interact with the child. In such a situation the modification of the communicative partner's input into conversations with the child is clearly of prime importance. If this modification is to be well-principled and effective, it must be based on a clear understanding of the nature of the child's pragmatic difficulties. It is necessary to ask the following questions. Should the input of the care-givers be based on the assumption that the child is displaying a straightforwardly immature pattern in areas such as the establishment of referent? Or are there also specific difficulties which are characteristic of the communicatively impaired children but not of younger children with a similar level of verbal comprehension? If so, it is important that the care-giver's input. and other aspects of the remedial program, are informed by an understanding of these.

3.6.3 THE LIMITATIONS OF INVESTIGATING FREQUENCY

Prutting and Kirchner (1987) emphasise that "frequency alone cannot be considered an index of severity" when using the Pragmatic Protocol (p. 114). They point out that certain combinations of pragmatic deficits may be more penalizing to the communicative exchange than others. It could possibly be the case, for example, that a failure to establish referent is more penalizing to the communicative exchange when it occurs in combination with a failure to mark a change in topic. McTear and Conti-Ramsden (1989) also emphasise that the frequency of pragmatic deficit may not be the most important factor determining communicative effect. They point out that a 50% error score could actually be less damaging to a conversational exchange than a 20% error score, if "the 20% that is deficient has a devastating effect on communication" (p. 159).

It therefore appears that there are limitations to what the investigation of comparative frequency can contribute to our understanding of pragmatic impairment. The severity of the communicative effect of any given instance of unestablished referent may vary as the result of contextual factors, such as the other pragmatic deficits with which it is combined (Prutting & Kirchner, 1987) or the point in the conversation at which it occurs (Bishop & Adams, 1989; McTear & Conti-Ramsden, 1989). For this reason it is essential to carry out a detailed qualitative analysis of how pragmatic deficits (such as the failure to establish referent) arise within a conversational context, and of the nature of their communicative effect. The following chapter aims to undertake a qualitative analysis of this kind.

A protocol, the Protocol for the Establishment of Referent, was developed to act as a framework for the analysis in Chapter 4 of the instances of failure to establish referent in the data recorded for the present study. On the basis of this analysis two questions are addressed. Do the unestablished referents in the conversations of the communicatively impaired children arise from the same causes as those in the conversations of younger children at a similar level of verbal comprehension? And what are the communicative effects of these unestablished referents?

Go to Chapter 4

Return to Contents