The main aim of the current study has been to develop a theoretical framework, the P.E.R., within which instances of unestablished referent in children's conversations can be analysed, and which has clinical potential for use in assessment and therapy with communicatively impaired children. For this potential to be realised, further studies, particularly in relation to the reliability and the validity of the procedure are required.
The inter-rater reliability demonstrated in the study related only to the actual identification of instances of unestablished referent. Not only does this reliability need to be replicated with further trained observers, but it needs to be extended to cover the categorisation of the identified referents. Categorisation involves placing the referring expression on the appropriate parameter, from A to J, of the P.E.R. and specifying what major processes have been responsible for the failure to establish referent. The specification of these processes is carried out within the broader framework provided by Prutting's Pragmatic Protocol. The findings of the current study suggest that in categorising unestablished referents it may be possible to differentiate pragmatic delay from pragmatic disorder. It has been postulated that those unestablished referents which involve the losing of the thread of the conversation are distinct from those which simply involve ambiguity. It appears that referring expressions with an inappropriate degree of specificity or with a lack of relevance to the conversational topic tend to have this confusing communicative effect, and to be associated with organisational difficulties. The demonstration of the validity of the P.E.R. as an assessment procedure, would also demonstrate that the analysis of these organisational difficulties is essential for effective therapy.
To establish the validity of a language assessment procedure it is necessary to demonstrate that it differentiates in a meaningful way between language impaired and normally developing children, and that it reflects systematic developmental changes in normal children (Adams & Bishop; 1989). Further research on unestablished referent, within the framework provided by the P.E.R., would seek to demonstrate this validity, and to define more rigourously and in more detail the categories of the P.E.R. profile. Further studies are needed on the development of the establishment of referent in normal children's conversations, and on the successful and unsuccessful use of referring expressions in the conversations of those who have been described as "communicatively impaired" in the current study, as well as other groups of language disordered children.
Among the many areas requiring investigation in relation to the establishment of referent in normal language development, three which have been identified in the course of the current study will be mentioned. It was suggested that the use of playfully deceptive referring expressions (4.4.5) should be investigated as a possible link between the very early use of deliberately false labels (Garvey. 1984) and the much later development of the use and understanding of irony (Winner, 1988). Another little explored type of referring expression is the colloquial use of 'this' and 'these' to indicate that a noun is specific for the speaker but not for the addressee (Perera, 1986; see 2.4.4). Studies of the acquisition of this usage could be informative in relation to the acquisition of the distinction between definite and indefinite reference. A third potential area of investigation, discourse deixis, is of considerable relevance to the organisation of conversation. Little appears to be known at present of the role of discourse deixis in early language development or in the conversations of different groups of language impaired children.
In seeking to establish the validity of the P.E.R. it is necessary to demonstrate, not only that it can distinguish between language impaired and non-language impaired children but also that it can distinguish between those children who are "communicatively impaired" and those who have other types of language disorder. In particular, a comparison is needed between the occurrence of unestablished referent in the conversations of communicatively impaired children (that is to say, those with the 'semantic-pragmatic profile' of impaired verbal comprehension and inappropriate conversational interaction) and its occurrence in the conversations of those with impaired expressive language, but with normal verbal comprehension. On the basis of the findings of the current study, one would predict that in the conversations of children with expressive, but not receptive, language difficulties there would be a low incidence of the type of unestablished referent which is associated with difficulty in following the line of thinking expressed by the speaker.
Considerable further research is needed to determine the precise role of the establishment of referent, and of the other parameters of Prutting's Pragmatic Protocol, in the conversations of communicatively impaired children, and in early language development as a whole. It is hoped that the results reported, and the analytical approach developed, in this pilot study will contribute to the formulation of a theoretical perspective within which such investigations can constructively be pursued.