Sandulescu Online - Displacement Constraints on Discourse

1975.Oslo: "Displacement Constraints on Discourse."

by C. George SANDULESCU, Stockholm.

(Paper given by C. G. Sandulescu at the Second Scandinavian Conference of Linguistics held at Lysebu (outside Oslo) between 19 and 20 April 1975; and published in: Papers from the Second Scandinavian Conference of Linguistics, Lysebu, April 19-20, 1975, edited by Even Hovdhougen; issued by the Department of Linguistics of the University of Oslo, Norway, 411 pages.)

1.0 ABSTRACT: The paper attempts to provide a tentative answer to the question: To what extent is a text with sentences (or rather, clauses-as-sentences) in another order the same text ( i. e. evinces an identical semantic interpretation) ? To begin with, points of similarity and disparity are established between sentential and textual paraphrasing. Preliminary data tends to show that the category of Displacement is closely associated with paraphrase via "alternative surface realizations in discourse". Juncture typology in discourse is to all likelihood one fundamental criterion for handling all typologies situated at a level above and beyond the sentence. Temporal Juncture, Participant Boundary, Discourse Head, Degree of Pairability, Degree of Clausal Restriction along a clausal dimension etc -- all prove to be essential constructs in obtaining a clear picture of what the semantic interpretation of a text/discourse might look like.

1.1 It is being assumed from the outset of the present discussion that the following theoretical issues are not only being given favourable consideration by current linguistic science, but also that they have received -- or are in the process of receiving -- a positive solution within mainstream linguistics on the basis of a consistent and integrated theoretical treatment of both spoken and written data:

(a) the scientific status of discourse in language studies;

(b) the description of discourse structure and typology;

1.2 It is further assumed that the direction in which the above questions are being given a solution is somehow summarized by the following cursory remarks:

(a) a discourse is considered to be a non-conjoined multi-sentence (cf. John Carroll 1972); the minimal language segment within which it is theoretically profitable to speak of "discourse" is the two-sentence sequence;

(b) the sentence -- or more accurately, the clause-as-sentence -- is, in contradistinction to the approach adopted by conventional linguistics, considered to be the minimal unit of investigation;

(c) multi-sentence research is implicitly put on a par with sentence research as regards theoretical acceptability: it is a fallacy to consider the sentence as the highest linguistic unit (cf. Sections 2.1 and 2.2). One of the central topics of the present study is the relation between Clause, Sentence, and Multi-Sentence (or Discourse).

1.3 Some of the major theoretical premises on which the present research is based are derived from the following sources: I am indebted to Roman Jakobson (1973) for a discussion of the scope of linguistics in relation to other branches of science, to Charles Fillmore (1973) and Ray Jackendoff (1972: chapter 6) for hints as to the status of discourse within the frame of reference of linguistics as a whole, and finally, to Ragnar Rommetveit (1974) for renewed interest in the Communication Act with due attention paid to sociolinguistics and cultural information.

1.4 Given the ambiguous theoretical status of discourse in relation to text, the methodological procedures resorted to as part of the present research spring from the conviction that a data-oriented approach of the kind adopted by, say, Labov & Waletzky (1967) proves in the long run to lead to more profitable strategies as regards the consistent interpretation of both type- and token- data. The reason why an exclusively model-centred approach to discourse of the kind adopted by, say, J. Petöfi (1973) is judged as less adequate is simply its inability to cope with reliable token-data on discourse at the present time. It is here considered premature, in other words, to overgeneralize on insufficient data.

1.5 Our purpose in the present study is twofold: first, to outline a way in which textual principles, viz, the fundamental principles structuring discourse as a distinct and specific level of linguistic inquiry can begin to be understood, as incorporated into an integrated theory of linguistic description, and moderately formalized. By an integrated theory we clearly mean a theory which pays adequate attention to all COMMUNICATIVE aspects of language in its use. Renewed interest in such aspects has in recent years taken the shape of research into pragmatics, semiotics, not only of the restricted kind provided by the Prague School (cf. Danes 1974), and by the London School (cf. Halliday 1973), but also of the more comprehensive type provided by [ + INITIAL ! ]Grice (1975). Secondly, our purpose is to show that there are constraints on discourse, quite distinct from those of grammar, that are ultimately derived from specific patterning and semantic linkage, as embodied in sets of abstract connectors. The nature of such inter-sentential connectors has so far been very little investigated and is still, from the purely theoretical point of view very unclear indeed. There may be rules as well as constraints at the level of discourse which are really "level-specific".

Furthermore, initial investigation of displacement constraints will hopefully throw some light on (a) textual competence, and (b) text typology.

(VAN DIJK 1972a : 3) If native speakers have the ability to distinguish between (linguistically) coherent and less coherent discourses, their competence should contain a device for distinguishing between grammatical and less grammatical texts. In that case, a grammar will have to formulate rules for the derivational description of textual structures. Only then may it serve as an adequate formal model for a postulated, idealized psychical system of linguistic knowledge like competence. Our competence is not sentential but textual.

Hypothetically, we might really begin at the beginning and suppose, for instance, -- within the stream of linguistic convention -- that there exists the following typology (partly overlapping) of competence:

(a) linguistic competence (cf. Chomsky 1965 : 4ff)

(b) communicative competence (cf. Hymes 1972 : 269ff)

Let us further suppose -- against convention this time -- that linguistic competence is subsumed to (and part of) communicative competence, and that textual competence is subsumed to (and thus part of) linguistic competence. Postulating finally that communicative competence -- the most comprehensive of all -- covers the whole range of semiotic systems accessible to man, we obtain the following hierarchy:

{A Level} Communicative Competence INCORPORATES:

{B Level} Linguistic Competence INCORPORATES:

{C Level} Textual Competence.

Chomsky's "linguistic" competence is here defined as that part of linguistic competence proper which excludes and discards (for mere methodological purposes) both the A Level, and more important, the C Level too.

1.6 The study of displacement constraints is seemingly profitable only in the case of text characterized by the feature [ - PARTICIPANT BOUNDARY ]. Given the strict sequential rules (cf. Labov 1970, and Mohan 1974), and the conversational postulates (cf. Gordon & Lakoff 1971) specific to discourse evincing [ + Participant Boundary ], the analysis of potential ranges of displacement is full of difficulties, and somewhat redundant in the latter case.

1.7 It is a commonplace to say (cf. Halliday 1973 : 107) that text can be spoken or written, long or short. But as regards text length it should be borne in mind from the start that the present study has concentrated on short texts only -- operationally defined --, i. e. on what we call partial sequence pQ, which are taken, for the sake of oversimplification, to stand for total sequences tQ. This assumption should be understood to be semiotically similar to Labov's (1967 : 12) own assumption that the narrative structure of oral versions of personal experience may in the long run prove to be more relevant to the overall analysis of discourse than the minute investigation of, say, Icelandic sagas.

2.0 FALLACIES OF CONVENTION. Conventional linguistics has for a number of years been subjected to two fallacies of judgment here labelled (a) the "anonymous" letter fallacy, and (b) the "sententialization" fallacy, which are best expressed and summarized by Katz & Fodor (1963/1964 : 479-518).

2.1 The "anonymous" letter fallacy. This issue is closely related to the whole discussion about context as summarized by Enkvist (1974b), but I here wish to connect the question of context with a wider issue (basically handled by the philosophy of linguistics), namely that of the meta-boundaries imposed upon the material under scrutiny, or, in other words, the issue of structural spans. The essence of the controversy boils down to the complex relationship between (a) the status of the sentence, and (b) what we propose to call here "contextualizability". "Sentences can be analysed outside contexts" and "Sentences cannot be analysed outside contexts" are, are, according to all principles of logic, two mutually exclusive propositions. But it is the paradox of a branch of humanistic science that both these propositions are almost equally valid. to start the modern discussion, here is what Katz & Fodor had to say, very much in line with what was said before them:

(KATZ & FODOR 1963/1964 : 484) ... let us consider a communication situation so constructed that no information about setting can contribute to a speaker's understanding of a sentence encountered in that situation. [...] The type of communication we shall consider is the following: a number of English speakers receive an anonymous letter containing the English sentence S. We are interested in the difference between this type of situation and one in which the same anonymous letter is received by persons who do not speak English, but are equipped with a completely adequate grammar of English. [...] Suppose S is the sentence "The bill is large". [...]

It has been found out of late, however, that "anonymous letter" procedures could only tell things about a very small area of language studies, and in addition, that such statements were only valid for a very restricted type of sentences, i. e. sentences with a very high quotient of "decontextualizability" (for a discussion of the "ideology" load of the selection of demonstration data, cf. Fillmore 1973 : passim). This issue has quite a lot to do with the issue of sentence typology which need not concern us here. But it is worth noting that such sentences as the ones below, selected for ample discussion by the researchers named next to them, fully enjoy the following important characteristic: they all evince very wide displacement ranges when inserted into ?? a given ?? discourse.

(1) The farmer killed the duckling. (Sapir 1921 : 86-98)

(2) The bill is large. (Katz & Fodor 1963 : 174ff)

(3) The box is in the pen. (Bar-Hillel 1960 : 158-63)

(4) Sincerity may frighten the boy. (Chomsky 1965 : 63-111)

2.2 The "sententialization" fallacy. It has been stated that text linguistics is a non-problem in linguistics by virtue of the fact that all sentences can be conjoined by means of connectives:

KATZ & FODOR !(1963/1964 : 490) [...] discourse can be treated as a single sentence in isolation by regarding sentence boundaries as sentential connectives. As a matter of fact, this is the natural treatment (sic !). In the great majority of cases, the sentence break in discourse is simply and-conjunction. (In others, it is but, for, or, and so on.) Hence, for every discourse, there is a single sentence which consists of the sequence of n-sentences that comprises the discourse connected by the appropriate sentential connectives and which exhibits the same semantic relations exhibited in the discourse.

According to the above quoted view, DISCOURSE IS PARAPHRASED INTO SENTENCE BY POTENTIAL AND-INSERTION, by virtue of what we prefer to call here The Conjoiner Insertability Principle. But instead of starting from the meta-assumption that all sentences are morphemically connected by means of a hypothetical and potential conjoiner, one may equally well start from the premise that sentences, at least most of them, are morphemically disconnected, i. e. that (co-ordinating) conjunction insertion is a fairly superficial phenomenon; one might suppose in exchange that sentences are only associated by means of an abstract connector which has its proper place within the overall semantic interpretation of the discourse. In certain instances, this abstract connector is actualised in the shape of a conjoiner, which is always and invariably only a partial expression, never a complete coverage, of the semantic range of the abstract connector. But it must be strongly emphasized that in most cases the abstract connector is not actualised in any way, its one and only materialization being the sentence boundary. Furthermore, inappropriate actualisation of abstract connectors may often lead to inappropriate surface structure of discourse.

The Katz & Fodor solution is clearly unsatisfactory. It is therefore advanced here that conjunction deletion, or in other words, conjoiner deletability may, within the theoretical model, function as a valid principle in the analysis of discourse -- The Conjoiner Deletability Principle --, in much the same way in which Katz & Fodor's proposed potential conjunction insertion is meant to turn discourse into sentence. This latter principle is, on the contrary, meant to turn sentence into discourse. Thus, the strategy we are proposing in the investigation of discourse is the exact converse of the proposal made by the two semanticists in 1963: namely, we propose potential conjoiner deletion not only as a means of turning sentence into discourse, but also as a means of keeping discourse as discourse. All this is closely related to the fundamental postulation of the nature of the structural span in linguistic description: assuming that discourse is the structural span that linguistics is ultimately concerned with, we are ascribing discourse status to language units in much the same way in which Katz & Fodor ascribed sentence status to discourse units in 1963. Any discourse remains more acceptable and appropriate, particularly on sociolinguistic grounds, after a process of actual conjunction deletion than it remains after a hypothetical and cumbersome process of and-insertion, to say nothing of the fact that the latter process cannot cross participant boundaries, as will be more clearly seen in the next section.

2.3 On the notion of Participant Boundary. In two distinct footnotes (sic !), Katz & Fodor dismiss (a) the study of discourse, and (b) the study of everyday conversation and dialogue as being irrelevant to theoretical linguistics, or, at least, lying clearly outside the central area of "mainstream" linguistics. Their first dismissive footnote provides linguistic evidence in support of the statement "discourse can be treated as a single sentence in isolation by regarding sentence boundaries as sentence connectives"; appended to this statement we find the following footnote:

(KATZ & FODOR 1963/1964 : 490) To illustrate this, let us consider the two-sentence discourse: "I shot the man with a gun," "If the man had had a gun too, he would have shot me first." The first sentence of this discourse is ambiguous in isolation, but not in this setting. But the problem of explaining this disambiguation is the same as the problem of explaining why the single sentence "I shot the man with a gun, but if the man had had a gun too he would have shot me first," does not have an ambiguous first clause. Likewise, consider the discourse, "I heard the noise," "The noise was completely inaudible" and its single sentence equivalent "I heard the noise, and the noise was completely inaudible". In showing why the single sentence is anomalous, a theory of semantic interpretation exhibits precisely those semantic relations in which the anomaly of discourse resides. This technique of replacing discourses or stretches in discourses with single compound sentences, by using sentential connectives in place of sentence boundaries, clearly has a very extensive application in reducing problems of setting selection to problems of semantic interpretation of sentences in isolation.

In addition to the fact which comes out quite clearly in the last lines of the quotation that the Conjoiner Insertability Principle is a mere tactical move deliberately resorted to in order to do away with language data that cannot possibly be accommodated by the theoretical model, it is also worth emphasizing that this kind of argumentation cannot handle the sociolinguistic category of Participant Boundary, e. g. in dialogue:

Speaker A: I shot the man with a gun.

Speaker B: Just because he had no gun himself !

Speaker A: I heard the noise !

Speaker B: But how could you ? The noise was inaudible !

This important linguistic phenomenon which is at present in the very focus of research attention (cf. conversational implicatures etc) was totally dismissed in the very next footnote (sic !), which reads:

(KATZ & FODOR 1963/1964 : 490) Sometimes a discourse cannot be directly converted into a compound sentence in this way. For example, The discourse "How are you feeling today ?" "I am fine, thanks" does not convert to "*How are you feeling today and I'm fine thanks" because the compound sentence is ungrammatical. But the fact that sentences of different types cannot be run together in the obvious way may not pose a serious problem because it is not at all clear that [...]

The argument that such sentences "cannot be run together" because "they are of different types" is totally unfounded not only because the issue of sentence typology is left unclarified, but also because not even "I am fine" and "thanks" could be run together as "I am fine and thanks", though they ultimately belong to the same type of utterance, with "thanks" standing as a sentence equivalent, the semantic interpretation of which is "I thank you" (which indeed becomes a sentence of exactly the same type as "I am fine").

In a word, Katz & Fodor are aware that their Conjoiner Insertability Principle is completely blocked by the occurrence of what we have called here Participant Boundary. This brief discussion proves at lest that Participant Boundary is an important linguistic phenomenon which is worth a lot of attention, precisely because it is worth "dismissing". It is indeed an extremely troublesome factor in text analysis: Enkvist (1973b : 131) dismisses it too in the process of macrosyntagm selection as part of his experiment in theme dynamics (as dialogue seems to evince idiosyncratic discourse-head progression). Cassirer (1972 : passim) in his analysis of Hjalmar Söderberg's "Pälsen" has to cope with the even more compex issue of Participant Boundary detaching segments of "interior monologue" from the overall structure of the text, and his disregard of the different "possible worlds" thus generated seems, in our opinion, to lead, partially at least, to slightly erroneous conclusions as regards both hierarchy of motifs and motif priority. The very fact that dialogue segments are deliberately avoided by researchers in the process of sample selection is an ideology-loaded attitude in handling the data.

3.0 DISPLACEMENT IN TWO-SENTENCE DISCOURSE. The device I have chosen for illustrating the nature of the research task is that of examining, as thoroughly as I can, a few shot texts. This paper is my contribution, in other words, to the recent, but respectable, tradition in Scandinavia regarding the linguistic analysis of relatively short texts, and remarkably illustrated by Cassirer (1972), Enkvist (1973; 1974), and Sigurd (1974) (the last one dealing with Buffalo Bill ...). I consider all these pieces of research as instances of linguistic principles applied to close and minute textual analysis, and hence as true instances of applied linguistics. In addition, this latter, and "practical & applied" part of the present paper takes up and expands implicit points in a similar tradition made by researchers in the English-speaking countries, particularly by Labov & Waletzky (1967), Fillmore (1973), Halliday (1973), and Hymes (1973).

For ease of identification, discourse heads will in the following data be transcribed in block capitals; then, the conventional asterisk [ * ] is here meant to cover the whole area of grammaticality, well-formedness, acceptability, and appropriateness... rather than be restricted, as is usually the case in sentence linguistics, to the first one or two of these categories.

3.1 Bidirectional displacement. The interdependency between the various types of competence (discussed in Section 1.5) can best be illustrated on the basis of very simple data: ball is a lexical item become famous as part of the 1963 discussion of word ambiguity, but word semantics happens to affect discourse in a remarkable way. Enkvist (1974b : 35) in his discussion of Katz & Fodor (1963/1964 : 486ff) gives among others the following two instances of two-sentence discourse:

(5) ((a) (an object rolled in) (b) (it was a colourful ball))

(6) ((a) (we danced all night) (b) (it was a colourful ball))

As the present approach assumes the clause-as-sentence to be the minimal unit of investigation within the frame of reference of Discourse Analysis (and the sentence/clause controversy renders the terms somewhat impractical), we are going to use unit in the subsequent discussion in order to replace the rather awkward clause-as-sentence.

By displacement in two-sentence discourse, we simply mean unit reversal according to the rule ab ==> ba ; the issue is, of course, much more complicated in multi-sentence discourse. Applying this rule to (5) and (6), we obtain:

(7) * (b) (it was a colourful ball) (a) (an object rolled in)

(8) (b) (it was a colourful ball) (a) (we danced all night)

Thus, unit reversal ab ==> ba, as a particular case of displacement is possible in (8), but not in (7), and the reason lies not only in the intrinsic semantics of the initial lexical items -- it, it, an, we -- as we might be tempted to believe, but also in a complex set of abstract assumptions derived from the semantic interpretation of each of these four sentences as well as from the linkage between them. We propose to call this linkage The Abstract Connector: we suggest that it is the task of an abstract connector to make explicit the complex relationship linking ball to danced in (6) and (8) at the level of the semantic interpretation, and the at least different kind of relationship (perhaps much simpler), providing a far laxer linkage between ball and object in (5) and (7). The potential assumptions -- we could even call them discourse presuppositions -- derived from the degree of collocability of the two lexical items are further reinforced by the occurrence in (6) and (8) of the supporting lexical items all night. This kind of extra support in presuppositional patterning is missing in (5) and (7). Furthermore, it is of great interest to note that the abstract connector in (8) may be actualised graphematically as in (9), where the colon provides additional spectralization of its essence:

(9) ((b) It was a colourful ball) (a) (:) (we danced all night))

Bidirectional displacement ab <==> ba is possible in (9), with punctuation preserved intact; there are sequential constraints imposed on the linear manifestation of the micro-text by the abstract connector and the semantic interpretation. (9) is simply an instance of two-sentence discourse characterized by complete unit reversibility, in contradistinction to the situation existing in (5). This difference between the two discourses is derived from the different nature of the abstract connector. But in (9) the greater degree of fusion is also signalled graphematically by the lower-case letter in the initial grapheme of the (a) unit; thus, colon insertion and capital letter replacement are graphematic processes closely linked with the displacement operation. Unit reversal is blocked in (5) by the explicit interplay between definiteness/indefiniteness in the linear manifestation as well as by corresponding pronominalization processes.

By way of summary, one could say that displacement in two-sentence discourse is the simplest type of displacement, for the very reason that it operates with a minimal number of units at the level of the minimal unit of discourse. It concurrently provides the simplest instance of textual paraphrasing in the cases in which inter-sentential relationships are being investigated.

3.2 Juncture typology in deviant discourse. The difficult and intricate question of the features of pathological discourse has been investigated by many authors, from among whom we mention the discussions of aphasia by Roman Jakobson in 1940, 1956, and 1966. Very recently, Ragnar Rometveit (1974 : 53ff) reopens the discussion of pathological discourse with an extensive analysis of the homonym symptom within a clearly micro-sociolinguistic frame of reference. Let us have a look at the following typical discourse utterable by a patient evincing the homonym symptom:

(10) I too was invited, I went to the ball ... and it rolled and rolled away.

There are three types of juncture in this text: we could best examine them if we take them one by one:

(11) (a) (I was invited) (b) (I went to the ball) (c) (it rolled away) (it rolled away)

The simplest displacement test will show that this is a temporal juncture in the Labov et al. (1967) sense:

(12) (( I went to the ball) (and) (I was invited))

It can in consequence be said that this first segment fulfils the minimal conditions of a narrative sequence by evincing at least one temporal juncture. Secondly, there is another type of juncture:

(13) The ball rolled and rolled away.

Comparing this with other data of the dame type, e. g. --

(14) He laughed and laughed.

(15) Mary looked and looked.

it becomes quite clear that there is no temporal sequence in such examples, even if they are being rewritten in various other ways by means of insertion and/or deletion processes (viz. transformations):

(16) He laughed and he laughed.

(17) He laughed. He laughed.

They are clear instances of repetition (cf. Bellert 1970), not sequentially organized in any way. Thirdly, it is this last type of juncture between ball with the semantic reading 'dancing party' and rolled that clearly violates the connectedness constraints imposed on normal discourse and thus gives the discourse its deviant (or pathological) features. It is interesting to see that of the two types of constraints discussed in the present study -- i. e. temporal constraints imposed on the partial sequence vs discourse-head constraints imposed on the same partial sequence --, the aphasic (at least in this particular type of aphasic discourse) will as a rule observe the temporal constraints imposed on the discourse head by event sequentiality, but will violate the constraints characterizing connectedness between discourse heads at the level of the DH juncture. We can thus say that aphasic discourse of the type discussed by Rommetveit (1974 : 53ff) is characterized by constraint violation leading to disconnectedness between (nominal) discourse heads: it is a violation of the constraints operating at the level of the DH juncture, with normal connectedness seemingly preserved at the level of the temporal juncture. However, such a suggestion, based on a handful of data only, needs confirmation and possible correction as a result of scanning extensive stretches of aphasic discourse (with special attention paid not only to the homonym symptom, but also to other pathologic speech phenomena as well). A diagrammatic summary of the three-way division of juncture discussed above would look as follows:

TYPE ONE: Temporal Juncture:

(18) (pQ1) ((a) (I was invited to the ball) (b) (I went to the ball))

TYPE TWO: Repetitive Juncture:

(19) (pQ2) ((a) (the ball rolled) (b) (the ball rolled))

TYPE THREE: Aphasic DH Juncture:

(20) (pQ3) ((a) (I went to the ball i ) (b) (the ball j rolled))

The partial Sequence pQ1 specifically characterizes narrative discourse; the partial sequence pQ3 provides, on the other hand, an excellent linguistic illustration of the homonym symptom. It is worth pointing out, however, by way of digression, that the study of two-sentence discourse to the complete neglect of a far wider structural span is fraught with danger. Thus (20) is ambiguous, but it might be possible to conceive of foregoing discourse, semi-science-fiction football playing perhaps, in which balli in (20a) would have an identical semantic reading with ballj in (20b),e.g.

(21) (a - 1) (I saw that the Ruritanian footballers had forgotten the ball right in the middle of the playing-ground) (a) I went to the ball (to pick it up)) (b) the ball rolled away (as if pushed by an invisible hand)))

This functions therefore as a strong warning that the investigation of two-sentence discourse is as incomplete and almost as artificial as the study of sentences in isolation, so arduously propounded by conventional sentence linguistics.

Type Three Juncture, illustrated under (10) and (20), i.e. the "aphasic" juncture, is in point of actual fact a non-juncture, an idiosyncratic juncture, or at the most a juncture evincing a violation of elementary connectedness constraints. (It must be said in passing that it is quite reminiscent of certain poetic idiosyncrasies, like Dylan Thomas's "His room so noisy to my own" etc). In this sense, DH constraints are more fundamental to normal discourse than the temporal constraints, in that it is the violation of DH constraints that, given a clear set of pragmatic conditions, causes the "pathologic-locutor" shock in the hearer. On the other hand, non-observance of temporal constraints may easily characterize the casualness of informal spontaneous conversation in face to face situations. Rommetveit (1974 : 60) comments on Moscovici & Plon (1966), who say that eye contact is largely irrelevant in structuring discourse, in the following way: "what brought about informality of style appeared to be spatially defined face-to-face orientation as such". Tentatively formalizing this statement by something like [ + SITUATIONAL PROXIMITY ], as a semiotic feature of the communication act, it is quite obvious that what is obtained is a completely different discourse structure. Informal spontaneous discourse and its characteristic patterning have so far remained almost totally uninvestigated. Allegations to the effect that "all this has been done !", and even that "this was done 25 years ago !" are the direct outcome of either ill will or sheer ignorance or both. According to tentative conclusions by Sinclair & Coulthard (1975), it is far easier to investigated institutionalised spoken discourse, rather than the non-institutionalised variety, viz. informal spoken conversation. Hence, a feature [ + INSTITUTIONALISED ], or something approximating it, may emerge as an outcome of the acute need to systematically investigate non-institutionalised as well as semi-institutionalised discourse.

4.0 DISPLACEMENT IN MULTI-SENTENCE DISCOURSE.

4.1 Paraphrasing. Passing on to paraphrasing and the notion of textual paraphrase, the following two major senses could be distinguished in discourse analysis, one more specific, the other one, more abstract:

(VAN DIJK 1972 : 139) If the current conception of paraphrase is also valid for texts, two texts can be said to be paraphrases if their semantic structures are identical and if morpho-syntactic and lexematic surface structures differ. Textual paraphrase, in this sense, characterizes the semantic representations underlying the sentences at the surface of the text. In a more loose way of definition, however, this purely "linear" conception of the SR's of a text seems to be irrelevant; texts can be paraphrased with sentences in another order, and with fewer sentences than that of the paraphrased text. In this more abstract sense, a summary may be considered a "minimal" paraphrase.

In this connection, Mohan (1974 : 90) discusses the question of alternative realizations in discourse: in point of fact, all paraphrases are alternative realizations of one and the same semantic interpretation, and as such they are to be considered equivalent. Let us further hypothesize that paraphrasing as a set of processes (both below and above sentence level) is part and parcel of textual competence (cf. Section 1.5).

4.2 Process Typology. On the strength of the tentative definition outlined above, we distinguish the following types of processes (the term is used in the Fillmore 1973 sense) performable on discourse:

(I) Below Sentence Level:

(a) Sentence Reduction, ( via Constituent Deletion )

(b) Sentence Expansion, ( via Constituent Addition )

(II) Above Sentence Level : (at Partial Sequence pQ level)

(a) Sentence Deletion,

(b) Sentence Insertion,

(III) Above Sentence Level : (at Total Sequence tQ Level)

(a) Text Reduction, ( i. e. via pQQ deletion )

(b) Text Expansion, ( i. e. via pQQ insertion / addition )

All these processual operations (viz. transformations) invariably lead to textual paraphrase: they are all rule-governed and presuppose definite sets of constraints. Whereas ( I ) implies intra-sentential operations, ( II ) postulates the acceptance of the "sentence-as-clause" as the minimal operational unit; in its turn, ( III ) is quite similar to ( I ) -- and even to ( II ) -- with the only difference that it takes pQ as its minimal operational unit, rather than either S -- like ( II ) --, or the sentential constituent -- like ( I ).

The present investigation focuses on ( II c ) exclusively, i .e. on textual displacement of clausal units which function as independent sentences (with The Conjoiner Deletability Principle leading to the formation of what we have called the "clause-as-sentence"). Given the difficulty of the matter under consideration, it dismisses for purely methodological reasons all intra-sentential reshuffling and substitutive operations, viz. the processes enumerated under ( I ); in other words, intra-sentential semantic and morpho-syntactic structure is even at the surface level taken as a constant, not as a variable. To put it very briefly, it takes into account only one variety of paraphrasing:

(VAN DIJK 1972 : 139) [...] texts can be paraphrased with sentences in another order.

In the light of what was said in the foregoing pages, it is assumed that there are clear constraints on sentential sequentiality in discourse.

4.3 Discourse-head pairability. It is often difficult to match two Noun Phrases in much the same way in which it is as often difficult to match two sentences or two clauses. Certain -- not all -- Noun Phrases of sentence linguistics become discourse heads in the analysis of discourse structure. One area of grammar most sensitive to the issue of discourse-head pairability is the category of comparison, and the whole issue of comparative clauses. The question has been very thoroughly examined, among others, by Austin Hale (1970), who point to the ill-formedness of sentences of the type --

(22) * The committee meeting was longer than the table.

But it is the question of pairing clauses within a comparative structure that is the hardest nut to crack. Thee are, of course, instances of transition in that clausal pairability is or may be blocked by lack of pairability at the level of the discourse heads. What constraints are, for instance, violated in the following sentence that makes it quite inappropriate in point of discourse structure ?

(23) * Cassius Clay eats apples, and Muhammad Ali drives a Ford.

A similar type of constraint on discourse is violated by the compound sentence

(24) * Princess Christina left very early, but Fru Magnusson talked to practically everybody in the room.

These and many others are in fact instances in which a transition is established between discourse-head pairability within the clause, on the one hand, and the completely different issue of actual inter-clausal pairability in which at least one of the discourse heads functions as a semi-clause:

(25) ** It was five dollars a dozen hotter in Chicago than eggs were expensive in China.

If one were to attempt to capitalize the clausal discourse-head equivalents for purely didactic and methodological reasons, one is bound to obtain something like --

(26) ** It was five dollars a dozen hotter in Chicago than eggs were expensive in China.

In such an instance the customary discourse heads have been replaced by full-fledged clauses, though the first clause is obviously deprived of its own discourse head which is presupposed to be provided by the latter one. But the semantic picture is further complicated by the ambiguous status of the lexical item expensive, which, in addition to belonging to the latter clause, seems to require to be granted identical status with its adjectival counterpart hotter, viz. an independent element outside the two clauses proper, building, in other words, their comparative conjoiner. But the issue of comparative and other conjoiners is best dealt with by resorting to less extreme demonstration data.

4.4 Clausal pairability. Passing from an analysis of the degree of pairability of discourse heads within the sentence to the far more complex issue of pairability beyond sentential or clausal boundaries, it is data of the following type that we have to cope with:

(27) My grandmother wrote me a letter yesterday and six men can fit in the back seat of a Ford.

(28) * The window was cleaner than Peter talks quickly.

Both examples have received extensive attention in articles by Lakoff (1971) and Hale (1970) respectively, and their apparent (and/or evident) lack of acceptability (and/or appropriateness) has been given more than ample space within the frame of reference of conventional sentence linguistics. But it is worth pointing out at this stage that the semantic interpretation of the actual compound (or complex) sentence may be further complicated by more subtle linkage devices often of an archetypal nature, as is quite the case with the following:

(29) John eats apples and I know many people who never see a doctor.

There is obviously no common discourse head to the two clauses, but if they are viewed together particularly in the light of a culture-specific archetype of the kind

(30) An apple a day keeps the doctor away

a complex paradigmatic relationship is established between the two lexical items apple / doctor, providing a deep semantic connection between the two clauses, which, in their surface structure, evince no common discourse head whatever. The nature of such a linkage procedure is so subtle and complex that it is liable to cancel, at least at first sight, the connective values of the conjoining morpheme and, achieving an instantaneous effect of an opposite nature. It is such demonstration data that Katz & Fodor were most anxious to avoid in 1963, or were, in one way or another, ignorant of: but it is up to the history of linguistics to pass judgment as to which of the two was the greater sin.

For reasons of simplicity, however, the discussion has so far been kept within the conventional boundaries of the sentence, and has discussed language material that is quite characteristic of sentence linguistics. It is easy to see, however, that passing on to the analysis of conversation, paradigmatic relationships of archetypal derivation (be they either culture-specific, or situation-idiosyncratic) become quite common. In all the above examples, clausal displacement, i. e. clausal reshuffling, would not have in any way "improved" the degree of appropriateness if the semantic interpretation. It is high time now to pass on from the discussion of intra-sentential evidence to a discussion of linguistic parameters characterizing discourse structure proper.

4.5 Concatenative discourse structure. Short texts have the great advantage over long ones that they evince a simpler and clearer discourse structure. It is necessary to emphasize again that discourse is viewed as a separate level quite distinct from the level of grammar, and that there is very little in common between sentence structure, as it is understood by sentence linguistics, and discourse structure, as it is understood by the separate and distinct discipline of discourse analysis. A text, for instance, is characterized by junctural typologies and quite specific discourse-head progression, which are quite impossible to identify adequately between the overnarrow boundaries of the sentence. To illustrate this statement, let us have a look at a simple text characterized by two negative features in point of juncture: it does not evince temporal juncture, and it does not evince any kind of Participant Boundary; but it does evince discourse-head progression all the same:

(31) We love swimming in the sea near our home. We live five miles from Chester. Chester is about twenty miles from Liverpool. Liverpool is a big industrial city. Industrial cities are important to the economy of a country.

It goes without saying that sustained and-conjoining of the type proposed by Katz & Fodor (1963/1964 : 490) would diminish the degree of appropriateness of the above text to a point very close to zero appropriateness. However, identifying discourse-head progression would be the first step towards identifying discourse structure, thus:

(32) ((a) (we love swimming in the sea near our home)

(b) (we live five miles from Chester)

(d) (Liverpool is a big industrial city)

(e) (industrial cities are important to the economy of the country)

Such foregrounding of potential discourse heads delineates inter-sentential (or inter-clausal) linkage devices, and gives an overall picture of one possible structure of this particular text, which could be formalized in the following way:

(33) ((a) ( p ==> ==> q )

(b) ( q ==> ==> r )

(d) ( s ==> ==> t )

(e) ( t ==> ==> u ))

On the basis of this textual diagram, it is easy to see that the text is characterized by such symmetry in the distribution of the discourse heads as well as in their placement in relation to actual clausal boundaries (initial / medial / final) that it is impossible to perform any clause displacement operations on the text without destroying this symmetry, which functions in much the same way in which the temporal juncture functions in the short narratives of personal experience described by Labov et al. (1967 : passim), and which led them to the important conclusion that a narrative universal can indeed be established on the basis of temporal sequentiality. In a remotely similar way, the various sentences of (31) seem to be locked in position by symmetry constraints. Thus a structure of the type

pq / qr / rs / st / tu

which lies closest to temporal juncture, will be called a concatenative structure, and generally is to be found in didactic texts, of which (31) above is a pale parody.

4.6 Radial discourse structure. But a different kind of text is very likely to pose a completely different type of problem:

(34) The human eye is one of nature's most marvellously complicated mechanisms. Whenever I think about the human eye, I am amazed at its complexity. Although it looks very simple from the outside, the intricacy of the interior is amazing. It is made up of so many parts, and yet it is so small, that it is difficult to think of it without being astonished. It is really miraculous in its complexity.

Skipping one stage of the analysis (of this text's information structure), that of DH identification, and passing on directly to a diagram of minimal and primitive formalization, we obtain the following textual structure (devised on the basis of formalization procedures derived from Mohan (1974):

(35) ((a) (p ==> ==> )

(b) ( ==> ==> p)

(d) (p ==> ==> )

(e) ( ==> p ==> )

(f) (p ==> ==> )

(g) ( ==> p ==> )

(h) ( ==> ==> p)

(i) ( ==> ==> )

(j) (p ==> ==> ))

Textual structure is in this case characterized by the occurrence of one single discourse head -- the human eye -- which lies right in the centre of discourse semantics, with all clauses staring from that very same centre and distributed in somewhat radial fashion in relation to it. Each clause is characterized by the occurrence of one discourse head only. There is only one clause (or rather clause equivalent, in the linear manifestation) characterized by complete absence of that discourse head. In consequence, what is different from one clause to another in the organization of the message is not the discourse head, which is constant and identical, but rather the distribution of it -- in initial, medial, or final position -- in the surface realization of the discourse. Once the clausal structure of a text is identified and formally specified, particularly as an outcome of the operation of the Conjoiner Deletability Principle, the clausal distribution of the discourse head in initial, medial, or final position is subjected to certain constraints, so far uninvestigated (but perhaps related to what traditional stylistics crudely calls "elegant variation"). Such constraints could never be specified by ordinary transformational-generative rules, which are more concerned to giving answers to the question WHAT ? rather than answering the questions HOW ? and WHEN ? in relation to overall discourse structure. For want of a better and more accurate label, we may provisionally refer to them as "rhetorical" constraints. Thus, concatenation and radiation are two textual principles which in the above illustrations operate separately, but which, in most "science" texts, for instance, can be found active side by side and simultaneously. Let us now increase the complexity of the discourse structure further and see what happens.

4.7 Variable degree of pairability. Turning again to the notion of pairability, it is quite easy to show that lack of pairability is as obvious within extended discourse as it is within the boundaries of one and the same sentence (cf. data under Sections 4.3 and 4.4):

(36) * John opened the door and organic chemistry was required of all students, because if seventeen were not a prime, then the dairy association would have to back down. Thus, we can conclude that television will have a beneficial effect upon the nation's young.

Such a partial sequence pQ poses special problems in that not only all clauses are well-formed, but also the conjoiners are sequentialized in accordance with definite constraints too. It is extremely easy to apply the Conjoiner Deletability Principle on such a sequence and obtain the following set of clauses:

(37) ((a) (John opened the door)

(b) (organic chemistry was required of all students)

(c) (seventeen is not a prime)

(d) (the dairy association has to back down)

(e) (we can conclude that X)

(f) (X (television will have a beneficial effect upon the young)))

The characteristic of this sequence is that there are no abstract connectors in any way linking the various clauses, one reason being that there are no jointly shared discourse heads and that there are no global sets of presuppositions either. The operation of the Conjoiner Insertability Principle does not provide connectedness to a set of completely disconnected clauses. In the same way, the operation of the Conjoiner Deletability Principle can only emphasize the totally artificial nature of conjoiner insertion. Such a sequence evinces a completely free range of displacement at the absolutely highest level, and is perhaps situated at the opposite pole in relation to the text discussed under (31), (32), (33), which, given its rigid symmetry of discourse heads, evinces the lowest possible level of displacement range.

4.8 Cumulative structures. It sometimes happen that certain types of discourse are structured on a repetitive pattern, in which part of the first clause is taken up in the second clause, repeated and slightly expanded; this expanded repetition is taken up in the next clause with a slight extra addition to its repeated segment. Here are three different illustrations of this type of discourse structure, one of which, strange as it may seem, is a transcript of actual conversation:

(38) He attacked the man. He attacked the man in the van. He attacked the man in the van with a hammer. He attacked the man in the van with a hammer which he held in both hands.

(39) Where's the watch ? Where's the watch I put in my pocket ? Where's the watch I put in my pocket to take to the shop ? Where's the watch I put in my pocket to take to the shop because it had stopped ?

(40) The thing won't grow. The thing won't grow even slowly. The thing won't grow even slowly in this cold. The thing won't grow even slowly in this cold, you know.

Concatenation, Radiation, and Cumulation are therefore three of the principles (the term is perhaps used in the Gricean sense) which represent some of the possibilities of discourse mapping in situations characterized by the two features [-TEMPORAL JUNCTURE] AND [-PARTICIPANT BOUNDARY].

4.9 Increasing structural intricacies. Let us now gradually increase the degree of textual complexity, and have a look at a text characterized by a parallel progression of two (or even three) distinct discourse heads, which are being carefully balanced against each other. It is at about the level of such texts that the accurate assessment of clausal displacement ranges becomes a useful instrument in obtaining textual paraphrases of one and the same initial discourse structure:

(41) The two brothers were quite different. Bob was tall, fair, and slim; John was short, dark, and fat. Bob was like his mother, and John was like his father. Bob was never happier than when he had something practical to do; John, on the other hand, was clumsy when using his hands. Their sister, Mary, was also clumsy with her hands. Bob rarely spoke to other people unless he was spoken to first, but John was always the centre of a group, talking and chattering as if his life depended on it. In fact, they were so different that it was hard to believe they were brothers.

The text under examination is obviously mapped on the basis of a set of definite principles that permit extensive reshuffling without substantial changes in the overall semantic interpretation of the total sequence tQ. This phenomenon is partly due to the fact that the inter-clausal linkage is based on abstract connectors evincing different degrees of semantic tension. It is quite noteworthy that the present text also allows of extensive clausal deletion, though analysis of paraphrase via clausal deletion does not form the topic of the present discussion. But in this case too, it must be again emphasized that discourse structure is characterized by a specific operation of a principle of dH symmetry, and that there may be a clear limit to displacement when certain symmetry constraints are violated (as already mentioned, "elegant variation" has much in common with the whole issue of symmetry constraints. Here by way of explicitation is a reduction of (41) to sequences of dH's:

(42) ((a) (p & q) (b) (p) (c) (q) (d) (p) (e) (q) (f) (p) (g) (p) (h) (q)

(i) (q) (j) (r) (k) (p) (l) (p) (m) (q) (n) (q) (o) (q) (p) (p & q)

(q) (p & q) (r) (zero) (s) (p & q))

Our last illustration evinces a degree of complexity a detailed analysis of which would most certainly go beyond the bounds of the present paper. It is, however, a fascinating piece of semi-organized discourse which is offered as food for thought to the enthusiastic reader:

(43) ((a) (we want to go to America, soon, on the 24th of this month, if possible)

(b) (will you come and see us one day this week, Friday evening perhaps)

(d) (so would Frieda)

(e) (we are not really enemies)

(f) (it is only a question of attitude)

(g) (I send you your book)

(h) (thank you for lending it to me)

(i) (do come and see us one day this week))

A most profitable way of formalizing the investigation of displacement ranges is to resort, for instance, to set theory (as Labov et al. 1967 have done), and try to separate the free clauses from the restricted ones.

5.0 FURTHER THEORETICAL CONSIDERATIONS.

5.1 On the nature of displacement. I suggest that the notion of displacement can be successfully used in the investigation of textual semantics in general, and the detection of patterns of assertions and presuppositions in particular. After all, a text could be defined -- at least the connectedness aspect of it -- as a network or a set of sequentialized assertions accompanied by a far more complex hierarchical network of presuppositions. Displacement can give very useful indications as to the exact possible range of surface manifestations of one and the same underlying semantic interpretation to the extent the same global set of constraints holds. I further advance that discourse mapping could profitably be investigated at both micro- and macro- levels by studying the various aspects of sentential (and/or clausal) processes of displacement, replacement, deletion, insertion, etc. One of the remote and ultimate goals of this research is to find out to what extent discourse mapping as a multifarious process could be tentatively associated with mental activity and with psychological reality (which the generative-transformational approach to language has failed to reflect).

5.2 Rudiments of text structure. There are two fundamental types of texts that may, in the last analysis, be derived on the basis of a remote analogy with the basic structure of the sentence S ==> NP VP. In the first place, there are texts patterned on the tense markers of the Verb Phrases; this is the large group discussed by Labov et al (1967), and the temporal sequentiality evinced by certain tense markers is classified within the category of Temporal Juncture. Texts evincing temporal juncture are classified as narratives; there is a t1-then-t2 type of constraint imposed on the minimal sequence of the type --

(44) He came in and hit her

but within identical syntactic and pragmatic conditions, it is cancelled in --

(45) He hit her and came in

as the temporal constraint imposed upon the juncture is clearly violated, and the semantic interpretation is decidedly changed.

Secondly, there are texts patterned on what might be called Noun Phrase sequentiality, i. e. evincing nominal dH juncture. This second fundamental type of discourse evades definition on the straightforward basis outlined in (44) and (45) for the simple reason that the discourse heads are not sequentialized along the time dimension in the same way. The picture is further complicated by the fact that such discourse heads are primarily located within the Noun Phrase, which is heavily subjected to substitution phenomena -- pronominalization, among others -- which are far more complex than the similar phenomena operating within the Verb Phrase, where substitution (and consequently, pronominalization) are very restricted indeed, and, in many cases, language specific. All this does not mean, however, that discourse patterned on the basis of temporal juncture (i. e. narrative discourse) does not evince (nominal) discourse head progression; it only means that so far temporal juncture has been more easily detectable, has received more attention, and in general evinces formal characteristics which are more obviously linguistic. Thus, the two types of discourse structure are in clear cases sufficiently easy to keep apart and single out: (i) VP sequentiality roughly corresponds to narrative, mainly on the basis of tense/time specification, and (ii) NP sequentiality approximately corresponds to non-narrative. This latter type may assume very complicated forms. Both types are heavily subjected to pragmatic constraints, in addition to the already well-known semantic and syntactic ones. The relation S (cf. Mohan 1974) in texts not characterized by the feature [ + PARTICIPANT BOUNDARY ] is further subdivided in consequence into the following two categories: the relation N (the t1-then-t2 relationship), in other words, (the action1-then-action2 relationship), and (ii) the relation NDH (the dH1-then-dH2 relationship) or to put it otherwise, the nominal discourse head relationship. Both types are of the x-then-y variety implying sequentialization in different types of time (for time/tense typology,cf. Fillmore 1973).

5.3 "Thematic entailment". By way of summary we can say that the two fundamental types of discourse structure are detectable on the basis of two quite distinct types of discourse heads. Both types of discourse are ultimately reducible to the same sequentiality formula, with one important distinction: whereas in one case we are concerned with a "primary sequence" at the level of the semantic interpretation, in the latter case, we have to do with a so-called topic sequentiality which can never in any way be linked to event sequentiality. Thus by replacing the two variables x and y, we obtain two different kinds of formulae in accordance with the syntactic location of the discourse head:

(46) the ACTIONm-then-ACTIONn semantic relationship.

(47) the TOPICp-then-TOPICq semantic relationship.

This type of "thematic entailment" has in a considerably modified version been interpreted as "theme dynamics" (cf. Enkvist 1973b); it can even be analysed in part in terms of topic and comment. However, such studies tend to underestimate the independent existence of a distinct level of discourse.

5.4 Micro vs. Macro. The present discussion has in view two types of textual structure: (a) textual macro-structures, operating at the level of the total sequence tQ, and subject to a set of constraints called global constraints: they could roughly be enumerated as (i) topic constraints, and (ii) deictic constraints (time, place, and person deixis, with the time constraints further subdivided into three, as suggested by Reichenbach in 1947); (b) textual micro-structures operating at the level of partial sequences pQ, usually two-sentence discourse (e. g. "Where is the bicycle ? I want to shave this morning !" or "Where is the soap ? I'm hungry !") where the main problem is inter-sentence linkage in situations of close proximity, and ultimately the structure of the abstract connector. The terms micro- and macro- structure have been used in a completely different sense by Van Dijk (1972). In our approach, micro- and macro- structures very roughly correspond to texture and structure respectively, as used by the New Criticism, and in particular by John Crowe Ransom in 1940.

5.5 Pragmatic constraints on discourse. In addition to the semantic-syntactic constraints so far investigated, discourses are also subjected to definite sets of pragmatic constraints. In this section we are going to very briefly enumerate the fundamental types of constraints on discourse directly derived from factors which have to do with the set of variables of the communication act. It must be pointed out from the start that such constraints are best analysed in texts characterized by the feature [ + PARTICIPANT BOUNDARY ]. Pragmatics is concerned not only with the description of the communication act, and the restrictions imposed upon it, but also with the investigation of its potential relation to message structure.

We distinguish three different types of pragmatic constraints: (a) the Felicity Conditions, covering mainly the situational factors; (b) the Belief Conditions, which is an all inclusive category, covering roughly the presuppositional patterning of discourse, but also a wide range of other relationships, which may be interpreted at least partially as conversational postulates, or as instances of entailment, expectation, etc; finally, (c) the Sincerity Conditions, already quite extensively discussed by formal logic in a different frame of reference.

All three sets, however, share certain common properties: for instance, they are each further subdivided into two. By the side of (i) a speaker-based subset, there is (ii) a hearer-based subset. Then, in addition to (i) general conditions required by all, or most, communication acts, there are (ii) certain conditions which are act-specific. Worth noting is also that both sender and receiver are functional roles of the communication act, which leads us to the question of Role Swapping and Turn Taking. These two constructs bring us back to our discussion of displacement: it is due to pragmatic constraints that no displacement can be operated on a two-sentence discourse made up of one question and one answer, and split by a participant boundary. In fact, the structure of turn taking in conversation is a potential area of investigation of displacement which looks quite promising, but given the fact that it involves the pragmatic feature [ + PARTICIPANT BOUNDARY ] clearly falls outside the bounds of the present study.

6.0 Final remarks. By way of summary, we wish to advance the following tentative conclusions:

6.1 Irrespective of whether they are conjoined or not, sentences within a text evince various degrees of displacement (i. e. have distinctly different displacement ranges), depending on the operation of global or local constraints. From among the important sets of constraints discussed, mention should be made of the pairability and symmetry constraints.

6.2 Displacement is one of the very few reliable instruments available for the investigation of what is commonly called "textual competence", and represents the first step towards the detailed study of both paraphrasability and text typology.

6.3 Text types (e. g. narrative vs. non-narrative) evince quite specific displacement ranges, with constraints imposed by different clausal components (VP in the case of narrative, and NP in the case of non-narrative).

SPECIAL NOTE: The notion of text is highly ambiguous in that it can denote either a partial sequence pQ, or a total sequence tQ, or both at the same time. (Hence, the even more ambiguous and even questionable concept of context, with some scholars going as far as narrowing down part of it to co-text ...)

6.4 One question which should be given greater attention is the relation between (i) the empirical question of (temporal) SEQUENCE (i. e. the progression in "real time" of the narrative), and (ii) the theoretical question of (hierarchical ?) ORDER (within the framework of discourse structure proper.

6.5 Displacement is also one of the most reliable procedures for the investigation of both the assertive and the presuppositional structure of a text. The interplay between textual constraints and sentential assertions as well as intra-sentential and inter-sentential presuppositions can provide valid answers which may lead to the emergence into the mainstream of research of so far unexplored fields of investigation (as it seems to have been the case with Pragmatics and Conversational Analysis).

6.6 Non-narrative run-on texts evince clear sets of constraints derivable both from successive "discourse head realization" at the level of linear manifestation in the surface structure and from specific patterns of inter-sentential presuppositions at the level of the semantic interpretation.

6.7 The rudiments of a clear discourse typology can be derived from the fundamental set of discourse features:

(i) [ + / - PARTICIPANT BOUNDARY ]

(ii) [ + / - TEMPORAL JUNCTURE ]

(iii) [ + / - DISCOURSE-HEAD JUNCTURE ]

6.8 There is clearcut sentence typology too, according to potential displacement range in that there are two fundamental types of units: (i) free units, i. e. clauses which can be freely moved up and down the sequentiality scale; (ii) restricted units, i. e. clauses which are for various reasons (time/tense, topic, symmetry etc) either completely bound, or bound to various degrees.

6.9 Clausal sequentiality isa direct reflection of speaker- and hearer- based Felicity Conditions, Belief Conditions, and Sincerity Conditions at the level of the whole text.

6.10 It has been stated that the relation between Precedent and Subsequent could be discussed in terms of stimulus and response (e. g. question and answer sequences). It is interesting to note, however, that a similar type of relationship is discussed by Mohan (1974) in terms of bleeding and feeding relations; it is true, though, that he clearly implies the existence of Participant Boundary between the two sentences under consideration. But feeding and bleeding could seemingly be applied to non-dialogued discourse.

REFERENCES to "Displacement Constraints in Discourse":

(1) Bellert, Irena.

1970 "On a Condition of the Coherence of Texts" in: Semiotica, II, 1970, 4, 335-363.

(2) Cassirer, Peter.

1972 Modell för struktur- och innehållsanalys av en text. Institutionen för nordiska språk vid Göteborgs Universitet, 1972, 74 pp. (mimeographed).

(3) Enkvist, Nils-Erik.

1973a Linguistic Stylistics, Mouton 1973. 179 pp.

(4) Enkvist, Nils-Erik.

1973b " 'Theme Dynamics' and Style: An Experiment" in: Studia Anglica Posnaniensia, 5, 1973, 127-135.

(5) Enkvist, Nils-Erik.

1974a "Några hypoteser om tema, rema och focus", Uppsala Symposium om språkfilosofi, November 1974 (mimeographed).

(6) Enkvist, Nils-Erik.

1974b Reports on Text Linguistics: Four Papers on Text, Style and Syntax, edited by Nils-Erik Enkvist, Publications of the Research Institute of the Åbo Akademi Foundation, 117 pp.

(7) Fillmore, Charles.

1970 "Subjects, Speakers and Roles" in: Synthese, vol. 21, 1970, 251-274.

(8) Fillmore, Charles.

1973 "May we come in ?" in: Semiotica, IX, 1973, 2, 97-116.

(9) Hale, Austin.

1970 "Conditions on English Comparative Clause Pairings" in: Roderick A. Jacobs & Peter S. Rosenbaum, eds, Readings in Transformational Grammar, Toronto, pp. 30-55.

(10) Kartuunen, Lauri.

1969 Discourse Referents, Preprint No. 70, International Conference on Computational Linguistics COLING, 38 pp. (cf. also "Textreferenten" (German translation of the same article) in: Ferenc Kiefer, ed., 1972, Linguistische Forschungen, 1, Semantik und generative Grammatik 1, Athenäum Verlag, pp. 175-198).

(11) Katz, Jerrold J. & Jerry A. Fodor.

1963 "The Structure of Semantic Theory" in: J.A.Fodor & J.J.Katz, eds 1964, The Structure of Language, Readings in the Philosophy of Language, Prentice Hall, pp. 479-518.

(12) Labov, William & Joshua Waletzky.

1967 "Narrative Analysis: Oral Versions of Personal Experience" in: June Helm, ed., 1967, Essays on the Verbal and Visual Arts, Proceedings of the 1966 Annual Spring Meeting of the American Ethnological Society, University of Washington Press, Seattle & London, pp. 12-45.

(13) Gordon, David & George Lakoff.

1971 "Conversational Postulates" in: Papers from the Seventh Regional Meeting, Chicago Linguistic Society, April 1971, pp. 63-84.

(14) Lakoff, Robin.

1971 "If's, And's, and But's about Conjunction" in: Charles J. Fillmore & D.Terrence Langendoen, eds, 1971, Studies in Linguistic Semantics, New York, pp.115-150.

(15) Mohan, Bernard A.

1974 "Do Sequencing Rules Exist ?" in: Semiotica, 12: 1, 1974, pp. 75-96.

(16) Rommetveit, Ragnar.

1974 On Message Structure. A Framework for the Study of Language and Communication, London, 143 pp.

(17) Saloni, Zygmunt & Andrzej Trybulec.

1974 "Coherence of a Text and its Typology", in: Semiotica, 11: 2, 1974, pp. 101-108.

(18) Sigurd, Bengt.

1974 Experiment med text, PILUS 26, University of Stockholm, (mimeographed).

(19) Smaby, Richard M.

1971 Paraphrase Grammars, D. Reidel, Dordrecht, Holland, 146 pp.

(20) Van Dijk, Teun A.

1972a Some Aspects of Text Grammar, A Study in Theoretical Linguistics and Poetics, Mouton, 1972, 375 pp.

(21) Van Dijk, Teun A.

1972b "Foundations for Typologies of Texts" in: Semiotica, VI: 4, 1972, 297-323.

ends