1974 - Prerequisites for an Integrated Theory of Discourse

By C. George SANDULESCU, Stockholm.

0 Epigraph:

Everything that can be thought at all can be thought clearly.

Everything that can be put into words can be put clearly.

                                                                    WITTGENSTEIN, Tractatus,4.116


A discourse is a multi-sentence.


"Sentence grammars cannot account for many relevant phenomena of natural languages (NOTE 1) in a sufficiently general and consistent way."


NOTE 1: e. g. (a) the structure of sequences of connected sentences,
                       (b) pronominalization, (c) referentialization, (d) semantic relations,
                       like (d1) synonymy, (d2) presupposition, (d3) entailment, 
                       (d4) inference etc


Discourse forms a separate LEVEL of linguistic description.


Discourse is superordinated in relation to Text: Text is subordinated in relation to Discourse.


A Text -- by assumed definition -- does not evince directly analysable suprasegmentals, e. g. Key, etc or paralinguistic (in the TRAGER, not in the ABERCROMBIE sense) features.


A discourse is a multi-sentence: in this sense, it comes nearest to (a) Texteme, and (b) Sequence, particularly in the written mode.


The category of Key has been specifically devised to cope with discourse suprasegmentals.


Suprasegmental parameters, such as Key, seem to have such significance in the analysis of spoken discourse that a discourse typology might even be set up on the strength of evidence coming exclusively from the area of suprasegmentals.


There are two fundamental approaches to language study: (a) a data-bound approach, and (b) an intuition-bound approach. 

All other approaches, and/or methods, and/or labels are ultimately reducible to either (a), or (b), or both.


STATEMENT: "No sharp separation line between theoretical text linguistics and applied text linguistics can be drawn; such a clear-cut division analogous to, say, that of theoretical linguistics and applied linguistics does not seem to be appropriate in any case."

COMMENT:    There are no clear-cut divisions in linguistics. At all.


An exclusively intuition-bound approach can never lead to conclusions directly and minutely reflecting discourse data.


Spoken Discourse postulates Participant Boundary as one of its cardinal categories.


"There is the domain of the sentence, and the domain beyond the sentence; there is no particular reason why the rules of one should duplicate or even resemble those of the other."


Discourse is structured; the sets of constraints operating in discourse are reminiscent of restrictions operating at sentence level, but neither identical, nor similar to them.


A discourse is semantically structured on the basis of (a) assertions, and (b) presuppositions. In addition to syntactic and stylistic constraints of the conventional type, a discourse is also characterized by a set of pragmatic constraints (which have so far been either neglected or insufficiently investigated).


"Syntax realises meaning, but does not organize it."


"Structure -- as we normally understand it -- is frequently a confusion between two different aspects: one being prospective (looking forwards, and influencing what happens afterwards) and the other being retrospective (looking backwards, and influencing what has gone before)."


"Any linguistic descriptive apparatus of the conventional types selects form the patterns that can bear meaning the patterns that must bear meaning, and therefore, ignores the rest."


The written mode of discourse is characterized by patterns clearly divergent from those of the spoken mode in that it focuses solely on 'patterns that must bear meaning'.


Discourse Analysis must needs be able to cope with  heterogeneous -- not merely homogeneous -- structures.


All dynamics of Topic and Comment (alias Theme and Rheme) ultimately operates at the level of discourse and is the direct outcome of discourse structure: this is in fact what Topic & Comment is about.


Topic/Comment Analysis (as well as Theme Dynamics) have to do with the surface structure of specific types of discourse.


All discussions of Style, for much the same reasons, are ultimately discussions pertaining to discourse structure.


A consistent theoretical model of discourse as a LEVEL of linguistic description might radically change our conventional notions of Style, Context, Deictics, and perhaps a few others.


The notion of context, in at least some of its more divergent meanings, could be completely incorporated into the notion of discourse. We would then be able to rephrase it either as 'foregoing discourse' or even as 'subsequent discourse' . . . Or, both simultaneously.


One of the important aspects of discourse structure which has been greatly neglected is Tense. The inability of conventional linguistics to cope adequately with, say, the Sequence of Tenses is evidence of its inability to cope with discourse structure.


The category of Participant Boundary (implicit in the definition of Move) and, deriving from it, the existence of unified structures across participants are essential theoretical constructs for an adequate description of Spoken Discourse. All conventional linguistics has consistently failed to accommodate this notion


The following micro-sociolinguistic categories, highly relevant to the analysis of spoken discourse, justify incorporation into the theoretical apparatus (quite regardless of their momentary rejection by symbolic logic): Role, Role Structure, Role Repertoire Repertoire Range; Turn, Turn Taking; Participant Boundary; Adjacency Pair, and a few others.


An Exchange X fulfils the condition of Appropriateness if, and only if, Move M2 shares the same set of (Pragmatic) Presuppositions P1, P2  . . . Pn with Move M1.


Connectedness in Discourse Analysis is a possible and probable counterpart of well-formedness in sentence linguistics.


Within the framework of the present theory, Connectedness is aimed at replacing the earlier concepts of cohesion, coherence, and congruence.


The question of degree of connectedness (or disconnectedness) of a discourse is ultimately a pragmatic value judgment, and as such, it should be simultaneously placed within the frame of reference of (a) Discourse Analysis, (b) Pragmatics, and (c) Axiology.


I distinguish between Connectedness and Appropriateness of a discourse in that the former is exclusively internal, whereas the latter is partly external to the utterance.


Research strategies ultimately mean research politics.


Research politics is the coverage of research alternatives, intimately coupled with circumstantial selection of one particular alternative or set of related alternatives. The selection process is, as a rule, idiosyncratic at individual, group, and societal level.


There is a close interrelationship between the selection of research strategies and the question of ideologies underlying a given scientific set-up.


Each of these Propositions may and should be expanded into comprehensive elaborations on the basis of circumstantial evidence provided by the data, thus leading to the realization of relatively separate and independent units; such units may be spoken, or written, or both.



1 CARROLL 1973.
1.1 LABOV 1970 and passim
  SINCLAIR 1972 and 1974.
  PETÖFI 1973.
  KIEFER 1974.
2 PIKE 1964 and 1972.
  SINCLAIR 1974.
  PETÖFI 1974.
2.1 PETÖFI 1971.
2.2 TRAGER 1958.
  PETÖFI 1971.
3 CARROLL 1973.
  STEGER 1974.
  PETÖFI 1973.
3.1 BRAZIL 1974.
  LAVER 1974.
  SINCLAIR 1974.
3.2 BRAZIL 1974.
4 LABOV passim.
4.1 PETÖFI (1973 : 13).
4.2 LABOV passim.
  SHUY 1974.
  JAKOBSON 1973.
5.1 COULTHARD 1973.
  SINCLAIR 1974.
6 CHATMAN 1974.
  JAKOBSON 1974.
6.1 TABER 1966. 
6.2 BELLERT 1973.
  KEENAN 1971. 
7 SINCLAIR 1974.
8 SINCLAIR 1974.
  HASAN 1968.
9 SINCLAIR 1974.
  FROMKIN 1971.
9.1 JAKOBSON 1973.
  PETÖFI 1974.
11 ENKVIST 1974.
11.2 HASAN 1968.
11.3 HALLIDAY 1962.
  HASAN 1968.
  ELLIS    1970.
11.4 HOENIGSWALD 1965.
  TABER (1966 : 20)
12 PETÖFI 1973.
  SINCLAIR 1974.
13 KEENAN 1971.
13.1 FISHMAN 1971.
  KEENAN 1971.
  GOFFMAN Passim.
14 BELLERT (1973 : 85).
  KEENAN 1971.
15 HALLIDAY 1962.
  ELLIS 1970.
15.1 HALLIDAY 1962.
  HASAN 1968.
15.2 PEIRCE Passim.
  KEENAN 1971.
16 HASAN 1968.
17 JAKOBSON 1960
17.1 ANDERSKI 1972.
  KUHN 1962.
17.2 ROSSI-LANDI 1974.